fix(bypass): stop handing solvers DDoS-Guard's ?check=1 probe URL (#1300 )

html_get_page follows Anna's Archive redirects by hand, and DDoS-Guard's gate answers /search with a 302 to the same path plus `check=1`. The follower walks that handshake by reassigning `current_url`, so every downstream handoff - the 403 branch, the 503-challenge branch, both redirect-loop rescues - passed the *probe* URL to the bypasser rather than the page we actually wanted. A solver opens that in a fresh browser holding none of the cookies the probe exists to collect, so DDoS-Guard cannot verify it automatically and serves the manual CAPTCHA page that nothing can solve. The #1292 log is exactly that: a 403 handed off on `&check=1`, FlareSolverr answering "Challenge solved!", and a 4721-byte DDOS-GUARD captcha page coming back. - `_solvable_url()` strips the probe parameter, applied at the single choke point in `_run_bypasser` so all four handoffs are covered. Scoped to the hosts whose redirects we follow manually; a URL without the parameter is returned by identity, so nothing else is re-encoded. The same reports showed three further defects, all of which stand whatever the host was reacting to: - The external bypasser logged that the solve had not cleared the protection and then returned the challenge page as a success. That skipped the one recovery left - get_bypassed_page's retry-and-rotate loop, where the next mirror is a different DDoS-Guard host - and filed the captcha page's own __ddg cookies as that host's clearance, to be replayed on every later request. It now raises ChallengeNotSolvedError before storing anything. - "Check that the bypasser is reachable and working" was the one piece of advice guaranteed to waste the reporter's time: it was reachable, it ran a full solve, and it returned a captcha. ChallengeNotSolvedError carries the marker so the search layer can name the host as the cause instead of the bypasser. - The untabled-page fingerprint logged `attempt_url`, which html_get_page has since rotated past. The #1298 bundle reported the page against annas-archive.gl when the body had come from .pk - the triage cost #1289 added the line to remove. The search now asks for the response URL and logs that. Its give-up shape is the tuple ("", url), which is truthy, so the exhaustion check reads the body rather than the response. Regression fixtures are built from the pages in the reports. The two behavioural handoff tests were checked against the unfixed code: both fail there, reproducing the reporter's log line verbatim. Refs #1292 Refs #1298
seleniumbase==4.53.5 (#1299 )
2026-09-24 22:00:21 +01:00 · 2026-09-02 16:12:52 -04:00 · 2026-09-02 15:48:48 -04:00 · 2026-09-01 12:59:33 -04:00 · 2026-09-01 12:38:57 -04:00 · 2026-09-01 11:19:01 -04:00
138 changed files with 9880 additions and 1428 deletions
@@ -25,14 +25,14 @@ jobs:
        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0

      - name: Initialize CodeQL
-        uses: github/codeql-action/init@ff2f1c621b7f889edc0d3c761ac2e6a3f8cdb0dd # v3
+        uses: github/codeql-action/init@db488ddef3bf6cb639b32c2e9a7c0a7ea8271d28 # v3
        with:
          languages: ${{ matrix.language }}

      - name: Autobuild
-        uses: github/codeql-action/autobuild@ff2f1c621b7f889edc0d3c761ac2e6a3f8cdb0dd # v3
+        uses: github/codeql-action/autobuild@db488ddef3bf6cb639b32c2e9a7c0a7ea8271d28 # v3

      - name: Perform CodeQL Analysis
-        uses: github/codeql-action/analyze@ff2f1c621b7f889edc0d3c761ac2e6a3f8cdb0dd # v3
+        uses: github/codeql-action/analyze@db488ddef3bf6cb639b32c2e9a7c0a7ea8271d28 # v3
        with:
          category: "/language:${{ matrix.language }}"
@@ -236,6 +236,7 @@ pyrightconfig.json
 *.local.*
 AGENTS.md
 .claude/
+CLAUDE.md
 .nvmrc
 .playwright-mcp/
 frontend-dist/
@@ -31,7 +31,7 @@ RUN npm run build
 FROM ghcr.io/astral-sh/uv:0.12.5@sha256:e85be844203885286c60ffad8a858d48afb6c5a5c237ca0e67f12e74b8f174b1 AS uv

 # Use python-slim as the base image
-FROM python:3.14.7-slim@sha256:ce40764625a4ff50df3548277632e7f96c4e77fe75fa848aae9885476e7df5a4 AS base
+FROM python:3.14.7-slim@sha256:cae66f2ef0ec51a9891263eeee7f987dacf0a9879e8aa9353d5606e0530619a5 AS base

 # Add build argument for version
 ARG BUILD_VERSION
@@ -276,6 +276,27 @@ class DownloadHandler(ABC):
        pass
 ```

+### Optional: Listing Files Before Download
+
+Some releases bundle several books (a whole-series torrent). Shelfmark inspects a
+release before queueing it so the user can review how it will be split into books.
+Override `list_files` when your source can enumerate a release's files without
+downloading it; the default returns `None`, which the UI reports as "can't inspect":
+
+```python
+from shelfmark.download.postprocess.packs import PackFile
+
+def list_files(self, release_data: dict[str, Any]) -> list[PackFile] | None:
+    """Return the release's files (release-relative paths + sizes), or None."""
+    torrent_bytes = ...  # e.g. fetch the .torrent, or scrape the indexer's detail page
+    return extract_file_list_from_torrent(torrent_bytes)  # from download.clients.torrent_utils
+```
+
+`release_data` is the same payload the frontend sends to `/api/releases/download`
+(`source_id`, `download_url`, `content_type`, `series_name`, ...). Built-in examples:
+Prowlarr parses the `.torrent` it already fetches (magnet-only releases return
+`None`), and AudiobookBay reads the file table off its detail page.
+
 ### Download Method Parameters

 | Parameter | Type | Description |
@@ -247,7 +247,7 @@ Seconds since the last WireGuard handshake before the healthcheck bounces the tu
 | `CALIBRE_WEB_URL` | Adds a navigation button to your book library (Calibre-Web Automated, Grimmory, etc). | string | _none_ |
 | `AUDIOBOOK_LIBRARY_URL` | Adds a separate navigation button for your audiobook library (Audiobookshelf, Plex, etc). When both URLs are set, icons are shown instead of text. | string | _none_ |
 | `SUPPORTED_FORMATS` | Book formats to include in search results. ZIP/RAR archives are extracted automatically and book files are used if found. | string (comma-separated) | `epub,mobi,azw3,fb2,djvu,cbz,cbr` |
-| `SUPPORTED_AUDIOBOOK_FORMATS` | Audiobook formats to include in search results. ZIP/RAR archives are extracted automatically and audiobook files are used if found. | string (comma-separated) | `m4b,mp3,m4a,flac,ogg,wma,aac,wav,opus,zip,rar` |
+| `SUPPORTED_AUDIOBOOK_FORMATS` | Audiobook formats to include in search results. ZIP/RAR archives are extracted automatically and audiobook files are used if found. | string (comma-separated) | `m4b,mp3,m4a,mp4,flac,ogg,wma,aac,wav,opus,zip,rar` |
 | `BOOK_LANGUAGE` | Default language filter for searches. | string (comma-separated) | `en` |

 <details>
@@ -296,16 +296,7 @@ Book formats to include in search results. ZIP/RAR archives are extracted automa
 Audiobook formats to include in search results. ZIP/RAR archives are extracted automatically and audiobook files are used if found.

 - **Type:** string (comma-separated)
- **Default:** `m4b,mp3,m4a,flac,ogg,wma,aac,wav,opus,zip,rar`
-
-#### `BOOK_LANGUAGE`
-
-**Default Book Languages**
-
-Default language filter for searches.
-
- **Type:** string (comma-separated)
- **Default:** `en`
+- **Default:** `m4b,mp3,m4a,mp4,flac,ogg,wma,aac,wav,opus,zip,rar`

 </details>

@@ -314,6 +305,7 @@ Default language filter for searches.
 | Variable | Description | Type | Default |
 |----------|-------------|------|---------|
 | `SEARCH_MODE` | How you want to search for and download books. | string (choice) | `universal` |
+| `BOOK_LANGUAGE` | Default language filter for searches. Users can override this for their own account. | string (comma-separated) | `en` |
 | `AA_DEFAULT_SORT` | Default sort order for search results. | string (choice) | `relevance` |
 | `SHOW_RELEASE_SOURCE_LINKS` | Show clickable release-source links in release and details modals. Metadata provider links stay enabled. | boolean | `true` |
 | `SHOW_COMBINED_SELECTOR` | Show the option to search for and download both a book and audiobook together. | boolean | `true` |
@@ -337,6 +329,15 @@ How you want to search for and download books.
 - **Default:** `universal`
 - **Options:** `direct` (Direct), `universal` (Universal)

+#### `BOOK_LANGUAGE`
+
+**Default Book Languages**
+
+Default language filter for searches. Users can override this for their own account.
+
+- **Type:** string (comma-separated)
+- **Default:** `en`
+
 #### `AA_DEFAULT_SORT`

 **Default Sort Order**
@@ -749,6 +750,7 @@ Automatically open the downloads sidebar when a new download is queued.
 Automatically download completed files to your browser for the selected content types.

 - **Type:** string (comma-separated)
+  
 - **Default:** _empty list_

 #### `MAX_CONCURRENT_DOWNLOADS`
@@ -1313,8 +1315,9 @@ Apply per-indexer seed time and ratio preferences from Prowlarr when sending tor
 | Variable | Description | Type | Default |
 |----------|-------------|------|---------|
 | `NEWZNAB_ENABLED` | Enable searching for books via a Newznab-compatible indexer | boolean | `false` |
-| `NEWZNAB_URL` | Base URL of your Newznab indexer or aggregator | string | _none_ |
-| `NEWZNAB_API_KEY` | Your Newznab API key (leave blank if not required) | string (secret) | _none_ |
+| `NEWZNAB_INDEXERS` | Named Newznab connections. Each row accepts `name`, `url`, and `api_key`. | JSON array | `[]` |
+| `NEWZNAB_URL` | Legacy single-indexer URL, used when `NEWZNAB_INDEXERS` is empty | string | _none_ |
+| `NEWZNAB_API_KEY` | Legacy single-indexer API key | string (secret) | _none_ |
 | `NEWZNAB_EBOOK_CATEGORIES` | Newznab category IDs searched for ebooks. Most indexers use the standard 7000, but some use custom IDs. Leave empty to use 7000. | string (comma-separated) | `7000` |
 | `NEWZNAB_AUDIOBOOK_CATEGORIES` | Newznab category IDs searched for audiobooks. Most indexers use the standard 3030, but some use custom IDs. Leave empty to use 3030. | string (comma-separated) | `3030` |
 | `NEWZNAB_AUTO_EXPAND` | Automatically retry search without category filtering if no results are found | boolean | `false` |
@@ -1331,21 +1334,36 @@ Enable searching for books via a Newznab-compatible indexer
 - **Type:** boolean
 - **Default:** `false`

+#### `NEWZNAB_INDEXERS`
+
+**Named Indexers**
+
+Configure multiple named Newznab-compatible indexers. The name is shown beside each search result. For environment-based configuration, provide a JSON array:
+
+```json
+[
+  {"name":"NZBGeek","url":"https://api.nzbgeek.info","api_key":"..."},
+  {"name":"DrunkenSlug","url":"https://drunkenslug.com","api_key":"..."}
+]
+```
+
+- **Type:** JSON array
+- **Default:** `[]`
+
 #### `NEWZNAB_URL`

-**Newznab URL**
+**Legacy Newznab URL**

-Base URL of your Newznab indexer or aggregator
+Single-indexer fallback used only when `NEWZNAB_INDEXERS` is empty.

 - **Type:** string
 - **Default:** _none_
- **Required:** Yes

 #### `NEWZNAB_API_KEY`

-**API Key**
+**Legacy API Key**

-Your Newznab API key (leave blank if not required)
+API key for the legacy Newznab URL.

 - **Type:** string (secret)
 - **Default:** _none_
@@ -2161,6 +2179,7 @@ Enable Moly.hu as a metadata provider for book searches
 | `SOURCE_PRIORITY` | Fallback sources, may have waiting. Requires bypasser. Drag to reorder. | JSON array | _see UI for defaults_ |
 | `MAX_RETRY` | Maximum retry attempts for failed downloads. | number | `10` |
 | `DEFAULT_SLEEP` | Wait time between download retry attempts. | number | `5` |
+| `RELEASE_SEARCH_TIMEOUT` | How long one release search may run before it gives up and reports why. A first search on a cold start pays for a browser solve, so leave room for one. If you use a reverse proxy, its read timeout should be at least this high or it will cut the search off with a 504 first. | number | `300` |
 | `AA_CONTENT_TYPE_ROUTING` | Override destination based on content type metadata. | boolean | `false` |
 | `AA_CONTENT_TYPE_DIR_FICTION` | Fiction Books | string | _none_ |
 | `AA_CONTENT_TYPE_DIR_NON_FICTION` | Non-Fiction Books | string | _none_ |
@@ -2239,6 +2258,16 @@ Wait time between download retry attempts.
 - **Default:** `5`
 - **Constraints:** min: 1, max: 60

+#### `RELEASE_SEARCH_TIMEOUT`
+
+**Release Search Timeout (seconds)**
+
+How long one release search may run before it gives up and reports why. A first search on a cold start pays for a browser solve, so leave room for one. If you use a reverse proxy, its read timeout should be at least this high or it will cut the search off with a 504 first.
+
+- **Type:** number
+- **Default:** `300`
+- **Constraints:** min: 30, max: 1800
+
 #### `AA_CONTENT_TYPE_ROUTING`

 **Enable Content-Type Routing**
@@ -2315,6 +2344,7 @@ Override destination based on content type metadata.
 | `EXT_BYPASSER_URL` | URL of the external bypasser service (e.g., FlareSolverr). | string | `http://flaresolverr:8191` |
 | `EXT_BYPASSER_PATH` | API path for the external bypasser. | string | `/v1` |
 | `EXT_BYPASSER_TIMEOUT` | Timeout for external bypasser requests in milliseconds. | number | `60000` |
+| `BYPASS_PAGE_SOURCE_TIMEOUT` | How long to wait for a solved page to produce its content before the bypass is retried. Raise it if solves succeed but searches still fail. | number | `20` |
 | `BYPASS_BROWSER_IDLE_TIMEOUT` | How long the bypass helper process may sit unused before it is shut down. Higher keeps more searches fast, lower frees memory sooner. | number | `180` |

 <details>
@@ -2371,6 +2401,16 @@ Timeout for external bypasser requests in milliseconds.
 - **Requires restart:** Yes
 - **Constraints:** min: 10000, max: 300000

+#### `BYPASS_PAGE_SOURCE_TIMEOUT`
+
+**Page Read Timeout (seconds)**
+
+How long to wait for a solved page to produce its content before the bypass is retried. Raise it if solves succeed but searches still fail.
+
+- **Type:** number
+- **Default:** `20`
+- **Constraints:** min: 1, max: 120
+
 #### `BYPASS_BROWSER_IDLE_TIMEOUT`

 **Bypasser Idle Timeout (seconds)**
@@ -30,7 +30,7 @@ Requires mounting your Calibre-Web `app.db` to `/auth/app.db`.

 Admins can configure per-user settings by editing a user in the user management panel. Non-admin users can also edit their own settings through **My Account** (accessible from the user menu). Admins control which sections are visible in My Account via the **Visible Self-Settings Sections** option.

-There are three categories of per-user settings:
+There are four categories of per-user settings:

 ### Delivery Preferences

@@ -42,6 +42,15 @@ Override where a user's downloads are sent. Options depend on the global output
 - **BookLore library/path** — Per-user BookLore target (when using BookLore output mode)
 - **Email recipient** — Per-user email address (when using Email output mode)

+### Search Preferences
+
+Override how a user searches, on top of the global search defaults:
+
+- **Search mode** — Direct or Universal for this user
+- **Default book languages** — The languages a user's searches fall back to when they don't pick one themselves. Useful for a shared instance where readers want different languages.
+- **Metadata providers** — Book, audiobook, and combined-mode provider for this user
+- **Default release sources** — The release tab opened first for books and audiobooks
+
 ### Notifications

 Users can configure personal notification routes, separate from the global notification settings. Each route targets a URL (e.g. an Apprise-compatible endpoint) and can be scoped to specific event types or all events.
@@ -32,7 +32,7 @@ dependencies = [
 browser = [
    "pyvirtualdisplay",
    "pyautogui",
-    "seleniumbase==4.52.1",
+    "seleniumbase==4.53.5",
    "python-xlib",
 ]

@@ -43,7 +43,7 @@ dev = [
    "pytest",
    "pytest-cov",
    "pytest-xdist>=3.8.0",
-    "ruff==0.16.4",
+    "ruff==0.16.5",
    "vulture>=2.14",
 ]

@@ -147,6 +147,7 @@ See the full [Environment Variables Reference](docs/environment-variables.md) fo
 Some of the additional options available in Settings:
 - **Prowlarr** - Configure indexers and download clients to download books and audiobooks
 - **Additional audiobook sources** - Configure additional sources for audiobook discovery
+- **Direct Download mirrors** - Supply your own Anna's Archive mirror URLs; Auto mode tries them in the order listed. The `annas-archive.is` domain does not currently work as a source — use `annas-archive.gl` instead (checked August 2026; mirror availability changes)
 - **IRC** - Add details for IRC book sources and download directly from the UI. Most networks serve audiobooks from the same channel as ebooks (on `irc.irchighway.net` that's `#ebooks`, while `#bookz` is effectively inactive), so leave the separate audiobook channel blank unless your network actually indexes one. IRC audiobooks usually arrive as ZIP/RAR archives — keep those enabled under Supported Audiobook Formats or the releases are filtered out of results
 - **Library Link** - Add a link to your Calibre-Web or Grimmory instance in the UI header
 - **File processing** - Customiseable download paths, file renaming and directory creation with template-based renaming
@@ -3,3 +3,14 @@

 class BypassCancelledError(Exception):
    """Raised when a bypass operation is cancelled."""
+
+
+class ChallengeNotSolvedError(Exception):
+    """Raised when a bypasser ran but the site still answered with a challenge.
+
+    Distinct from a bypasser that is broken or unreachable, which is what every
+    "the bypass failed" message used to say. A solver can do its job perfectly and
+    still be handed something it cannot clear - DDoS-Guard's manual CAPTCHA page is
+    the case from #1292 - and telling the user to go check that FlareSolverr is
+    reachable sends them to fix a service that is working.
+    """
@@ -71,11 +71,18 @@ def _get_full_cookie_domains() -> set[str]:
    return {_get_base_domain(domain) for domain in get_zlib_cookie_domains()}


+def _replay_per_check_cookies() -> bool:
+    """Whether the per-check trio is kept rather than dropped (see env.py)."""
+    from shelfmark.config import env
+
+    return env.DDG_REPLAY_PER_CHECK_COOKIES
+
+
 def _should_extract_cookie(name: str, *, extract_all: bool) -> bool:
    """Determine if a cookie should be extracted based on its name."""
    # Checked before extract_all: a per-check token is wrong to replay for every
    # domain, including the full-session ones.
-    if name in DDG_EPHEMERAL_COOKIE_NAMES:
+    if name in DDG_EPHEMERAL_COOKIE_NAMES and not _replay_per_check_cookies():
        return False
    if extract_all:
        return True
@@ -138,9 +145,11 @@ def store_extracted_cookies(
    extract_all = base_domain in _get_full_cookie_domains()

    cookies_found: dict[str, dict[str, Any]] = {}
+    dropped: list[str] = []
    for cookie in cookies:
        name = _cookie_field(cookie, "name") or ""
        if not _should_extract_cookie(name, extract_all=extract_all):
+            dropped.append(name)
            continue
        secure = _cookie_field(cookie, "secure")
        cookies_found[name] = {
@@ -152,6 +161,18 @@ def store_extracted_cookies(
            "httpOnly": True,
        }

+    # Names only, never values. Which cookies a solve won, and which of them were held
+    # back, is the evidence needed to settle what DDoS-Guard actually treats as clearance
+    # (issue #1276) - and without it a debug log shows a solve succeeding and the next
+    # request being challenged with nothing in between to explain why.
+    logger.debug(
+        "Solve on %s won %s; keeping %s; dropping %s",
+        base_domain,
+        sorted({_cookie_field(c, "name") or "" for c in cookies}),
+        sorted(cookies_found),
+        sorted(set(dropped)) or "nothing",
+    )
+
    if not cookies_found:
        return

@@ -6,7 +6,8 @@ from typing import TYPE_CHECKING, Any

 import requests

-from shelfmark.bypass import BypassCancelledError
+from shelfmark.bypass import BypassCancelledError, ChallengeNotSolvedError
+from shelfmark.bypass.challenge import challenge_marker
 from shelfmark.bypass.cookie_store import store_extracted_cookies
 from shelfmark.core.config import config
 from shelfmark.core.logger import setup_logger
@@ -91,7 +92,13 @@ def _store_solution_clearance(target_url: str, solution: Mapping[str, Any]) -> N


 def _fetch_via_bypasser(target_url: str) -> str | None:
-    """Make a single request to the external bypasser service. Returns HTML or None."""
+    """Make a single request to the external bypasser service. Returns HTML or None.
+
+    Raises:
+        ChallengeNotSolvedError: the service answered with a page that is still a
+            challenge, whatever verdict it reported on itself.
+
+    """
    raw_bypasser_url = _coerce_config_str(
        config.get("EXT_BYPASSER_URL", "http://flaresolverr:8191"),
        "http://flaresolverr:8191",
@@ -143,6 +150,32 @@ def _fetch_via_bypasser(target_url: str) -> str | None:
            logger.warning("External bypasser returned empty response for '%s'", target_url)
            return None

+        # "Challenge solved!" is the solver's verdict on its own work, and #1289 showed
+        # it can be reported alongside a page the caller then rejects. Say what actually
+        # came back, so a later report does not have to infer it from downstream errors.
+        marker = challenge_marker(html)
+        logger.debug(
+            "External bypasser page for '%s': %d bytes, challenge_marker=%r",
+            target_url,
+            len(html),
+            marker,
+        )
+        if marker:
+            # The solver's verdict is not evidence; the page is. Returning this one as a
+            # success is what made #1292 unrecoverable: the retry-and-rotate loop that
+            # could still have saved the search - the next mirror is a different
+            # DDoS-Guard host, in its own state - was never entered, and the challenge
+            # page's own __ddg cookies were filed as this host's clearance and replayed
+            # on every later request.
+            logger.warning(
+                "External bypasser reported success but returned a challenge page for "
+                "'%s' (%d bytes, marker=%r) - the solve did not clear the protection",
+                target_url,
+                len(html),
+                marker,
+            )
+            raise ChallengeNotSolvedError(marker)
+
        try:
            _store_solution_clearance(target_url, solution)
        except AttributeError, KeyError, TypeError, ValueError:
@@ -192,16 +225,33 @@ def get_bypassed_page(
    selector: network.AAMirrorSelector | None = None,
    cancel_flag: Event | None = None,
 ) -> str | None:
-    """Fetch HTML via external bypasser with retries and mirror rotation."""
+    """Fetch HTML via external bypasser with retries and mirror rotation.
+
+    Raises:
+        ChallengeNotSolvedError: every attempt came back still carrying a challenge.
+            Reported apart from returning None because the two ask the user for
+            opposite things: None means go and check the bypasser, this means the
+            bypasser is fine and the host is the one refusing.
+        BypassCancelledError: the caller's cancel flag was set.
+
+    """
    from shelfmark.download import network as network_module

    sel = selector or network_module.AAMirrorSelector()
+    unsolved_marker: str | None = None

    for attempt in range(1, MAX_RETRY + 1):
        _check_cancelled(cancel_flag, "by user")

        attempt_url = sel.rewrite(url)
-        result = _fetch_via_bypasser(attempt_url)
+        try:
+            result = _fetch_via_bypasser(attempt_url)
+        except ChallengeNotSolvedError as e:
+            # Worth the remaining attempts rather than an immediate give-up: the retry
+            # rotates onto the next mirror, and that is a different DDoS-Guard host with
+            # its own idea of whether this caller needs a CAPTCHA.
+            unsolved_marker = str(e) or unsolved_marker
+            result = None
        if result:
            return result

@@ -222,4 +272,11 @@ def get_bypassed_page(
        if action in ("mirror", "dns") and new_base:
            logger.info("Rotated %s for retry", action)

+    if unsolved_marker:
+        msg = (
+            "The bypasser ran, but the site kept answering with a protection challenge "
+            f"(marker={unsolved_marker!r}). That is usually a manual CAPTCHA, which no "
+            "bypasser can answer - the bypasser itself is working. Try again shortly."
+        )
+        raise ChallengeNotSolvedError(msg)
    return None
@@ -82,12 +82,18 @@ _HELPER_RESULT_POLL_SECONDS = 0.05
 # what it is doing and exit before its session is killed instead.
 _HELPER_SHUTDOWN_GRACE_SECONDS = 15.0
 _HELPER_IDLE_TIMEOUT_DEFAULT = 180.0
+# How long to wait for a solved page to produce its document before the attempt is
+# abandoned. SeleniumBase's own get_page_source() allows one second; see _read_page_source.
+_PAGE_SOURCE_TIMEOUT_DEFAULT = 20.0
 _PARENT_WATCHDOG_INTERVAL_SECONDS = 5.0
+# How much of ffmpeg's stderr to quote when reporting that it died.
+_FFMPEG_ERROR_TAIL_CHARS = 500


 class _DisplayState(TypedDict):
    ffmpeg: subprocess.Popen[bytes] | None
    ffmpeg_output: Path | None
+    ffmpeg_error_log: Path | None


 class _PageWithWindowRect(Protocol):
@@ -101,6 +107,7 @@ class _BrowserWithWindowRectPage(Protocol):
 DISPLAY: _DisplayState = {
    "ffmpeg": None,
    "ffmpeg_output": None,
+    "ffmpeg_error_log": None,
 }
 LOCKED = threading.Lock()
 _PROC_ROOT = Path("/proc")
@@ -518,18 +525,6 @@ async def _bypass_method_humanlike(page: Any) -> bool:
        return False


-async def _bypass_method_cdp_solve(page: Any) -> bool:
-    """CDP Mode with solve_captcha() - auto-detects challenge type."""
-    try:
-        logger.debug("Attempting bypass: CDP solve_captcha")
-        await page.solve_captcha()
-        await asyncio.sleep(_RNG.uniform(3, 5))
-        return await _is_bypassed(page)
-    except _CDP_OPERATION_ERRORS as e:
-        logger.debug("CDP solve_captcha failed: %s", e)
-        return False
-
-
 CDP_CLICK_SELECTORS = [
    "#turnstile-widget div",  # Cloudflare Turnstile
    "#cf-turnstile div",  # Alternative CF Turnstile
@@ -609,8 +604,13 @@ async def _bypass_method_cdp_gui_click(page: Any) -> bool:
        return False


+# Ordered cheapest-first, and deliberately without a bare `solve_captcha()` entry:
+# _bypass_method_cdp_gui_click opens by doing exactly that and returns the moment it
+# works, so a separate method ahead of it could only ever repeat the half that had
+# already failed - one wasted round trip plus the backoff before the next attempt, on
+# every solve that gets this far. Measured at ~5.5s of the ~26s each solve cost, and
+# 0/19 successes for the standalone method against DDoS-Guard. See issue #1285.
 BYPASS_METHODS = [
-    _bypass_method_cdp_solve,
    _bypass_method_cdp_gui_click,
    _bypass_method_cdp_click,
    _bypass_method_humanlike,
@@ -618,6 +618,25 @@ BYPASS_METHODS = [

 MAX_CONSECUTIVE_SAME_CHALLENGE = 3

+# How many method attempts one _bypass() pass may make. Deliberately *not* MAX_RETRY:
+# that value is already the outer page-load retry in _run_bypass_in_current_process, and
+# reading it here too squared the budget - the default 10 meant 10 page loads x 4 methods
+# = 40 solve attempts on one browser, which overruns the worker deadline and reports
+# `TimeoutError` instead of a plain "bypass failed". One full pass through the methods
+# plus a spare is all this loop can use anyway: the stuck-challenge guard below aborts at
+# len(BYPASS_METHODS) + 1, so a larger number here only ever showed up in the logs.
+_BYPASS_METHOD_ATTEMPTS = len(BYPASS_METHODS) + 1
+
+# The undisturbed window a passive challenge gets before any method runs. Sized off the
+# real thing: a desktop browser clears Anna's Archive's DDoS-Guard JS check in under 10s.
+_PASSIVE_SOLVE_SECONDS = 15.0
+_PASSIVE_SOLVE_POLL_SECONDS = 1.0
+
+# Head-room the retry loop leaves itself so it can return a real failure rather than be
+# cancelled at the worker deadline. Enough for the pass in flight to unwind and the
+# browser to close.
+_RESERVE_FOR_CLEAN_FAILURE_SECONDS = 60.0
+

 def _check_cancellation(cancel_flag: Event | None, message: str) -> None:
    """Check if cancellation was requested and raise if so."""
@@ -627,13 +646,26 @@ def _check_cancellation(cancel_flag: Event | None, message: str) -> None:
        raise BypassCancelledError(msg)


+async def _wait_for_passive_solve(page: Any, cancel_flag: Event | None = None) -> bool:
+    """Poll for a challenge that clears itself, without touching the page.
+
+    Returns True as soon as the page looks bypassed, False once the window is spent.
+    """
+    logger.info("Waiting up to %.0fs for the challenge to clear itself...", _PASSIVE_SOLVE_SECONDS)
+    deadline = time.monotonic() + _PASSIVE_SOLVE_SECONDS
+    while time.monotonic() < deadline:
+        _check_cancellation(cancel_flag, "Bypass cancelled while waiting for a passive solve")
+        await asyncio.sleep(_PASSIVE_SOLVE_POLL_SECONDS)
+        if await _is_bypassed(page):
+            return True
+    return False
+
+
 async def _bypass(
    page: Any, max_retries: int | None = None, cancel_flag: Event | None = None
 ) -> bool:
    """Attempt to bypass Cloudflare/DDOS-Guard protection using multiple methods."""
-    max_retries = (
-        max_retries if max_retries is not None else _coerce_positive_int(app_config.MAX_RETRY, 10)
-    )
+    max_retries = max_retries if max_retries is not None else _BYPASS_METHOD_ATTEMPTS

    last_challenge_type = None
    consecutive_same_challenge = 0
@@ -651,6 +683,20 @@ async def _bypass(
        challenge_type = await _detect_challenge_type(page)
        logger.debug("Challenge detected: %s", challenge_type)

+        # Give a passive check the undisturbed window it needs before touching the page.
+        # DDoS-Guard's JS check on Anna's Archive has no click target: it runs, then
+        # navigates on its own - a desktop browser clears it in well under 15s. Every
+        # method below either clicks a selector that is not there or reloads, and a reload
+        # restarts an in-flight check (which DDoS-Guard also throttles), so going straight
+        # to them meant the one thing that actually solves this challenge was the one
+        # thing never tried. Costs one 15s window per solve against a minutes-long budget,
+        # and a challenge that needs interaction simply falls through to the methods.
+        if try_count == 0 and challenge_type != "none":
+            if await _wait_for_passive_solve(page, cancel_flag):
+                logger.info("Bypass successful: %s challenge cleared itself", challenge_type)
+                return True
+            logger.debug("Challenge did not clear on its own; trying bypass methods")
+
        # No challenge detected but page doesn't look bypassed - wait and retry
        if challenge_type == "none":
            logger.info("No challenge detected, waiting for page to settle...")
@@ -769,6 +815,22 @@ def _build_host_resolver_rules() -> list[str]:
 DRIVER_RESET_ERRORS = {"ProtocolException", "RuntimeError", "TimeoutError"}


+async def _read_page_source(page: Any) -> str:
+    """Read a solved page's HTML, waiting for the document to arrive.
+
+    `get_page_source()` waits one second for the `html` element. A page released from a
+    challenge is often still navigating to the real content, so the read times out even
+    though the solve succeeded: the whole attempt is retried, and the repeated requests
+    are what earn a 429 from a host that was about to serve us.
+    """
+    timeout = _coerce_non_negative_float(
+        app_config.get("BYPASS_PAGE_SOURCE_TIMEOUT", _PAGE_SOURCE_TIMEOUT_DEFAULT),
+        _PAGE_SOURCE_TIMEOUT_DEFAULT,
+    )
+    element = await page.find("html", timeout=timeout)
+    return await element.get_html_async()
+
+
 async def _get(url: str, driver: Any, cancel_flag: Event | None = None) -> str:
    """Fetch URL with Cloudflare bypass using a CDP browser."""
    _check_cancellation(cancel_flag, "Bypass cancelled before starting")
@@ -792,7 +854,7 @@ async def _get(url: str, driver: Any, cancel_flag: Event | None = None) -> str:
    logger.debug("Starting bypass process...")
    if await _bypass(page, cancel_flag=cancel_flag):
        await _extract_cookies_from_cdp(driver, page, url)
-        return await page.get_page_source()
+        return await _read_page_source(page)

    logger.warning("Bypass completed but page still shows protection")
    try:
@@ -810,14 +872,33 @@ async def _get(url: str, driver: Any, cancel_flag: Event | None = None) -> str:

 def _run_bypass_in_current_process(url: str, retry: int, cancel_flag: Event | None = None) -> str:
    """Run the CDP bypass in the current process."""
+    timeout = (
+        _CHILD_BYPASS_TIMEOUT_SECONDS
+        if os.environ.get(_BYPASS_CHILD_ENV) == "1"
+        else _IN_PROCESS_BYPASS_TIMEOUT_SECONDS
+    )

    async def _run_bypass() -> str:
        driver = None
+        # Stop retrying while there is still time to say so. A challenge nothing can solve
+        # would otherwise spend every one of `retry` passes and be cut off mid-pass by the
+        # worker deadline, which surfaces to the caller as `RuntimeError: TimeoutError` -
+        # a message that says nothing about protection and sent users looking at their
+        # reverse proxy. Giving up a pass early returns the real "bypass failed" instead.
+        deadline = time.monotonic() + timeout - _RESERVE_FOR_CLEAN_FAILURE_SECONDS
        try:
            driver = await _create_cdp_browser(url)

            for attempt in range(retry):
                _check_cancellation(cancel_flag, "Bypass cancelled before attempt")
+                if attempt > 0 and time.monotonic() >= deadline:
+                    logger.warning(
+                        "Bypass budget spent after %s/%s attempts; giving up on %s",
+                        attempt,
+                        retry,
+                        url,
+                    )
+                    break

                try:
                    result = await _get(url, driver, cancel_flag)
@@ -838,7 +919,7 @@ def _run_bypass_in_current_process(url: str, retry: int, cancel_flag: Event | No
                        await _close_cdp_driver(driver)
                        driver = await _create_cdp_browser(url)

-            logger.error("Bypass failed after %s attempts", retry)
+            logger.error("Bypass failed for %s", url)
            return ""
        finally:
            if driver:
@@ -852,12 +933,9 @@ def _run_bypass_in_current_process(url: str, retry: int, cancel_flag: Event | No
    # one call and closes it on the way out, so a helper serving many requests would build
    # and tear down a loop per bypass and would carry no deadline of its own. The worker's
    # loop lives in a thread, outlives any single bypass, and cancels the coroutine when the
-    # deadline passes.
-    timeout = (
-        _CHILD_BYPASS_TIMEOUT_SECONDS
-        if os.environ.get(_BYPASS_CHILD_ENV) == "1"
-        else _IN_PROCESS_BYPASS_TIMEOUT_SECONDS
-    )
+    # deadline passes. `_run_bypass` aims to finish inside this same budget of its own
+    # accord, so reaching this deadline now means a wedged session rather than a stubborn
+    # challenge - which is the only case worth reporting as a timeout.
    return _CDP_WORKER.run(_run_bypass(), timeout=timeout)


@@ -1155,6 +1233,20 @@ def get(url: str, retry: int | None = None, cancel_flag: Event | None = None) ->
        if cached_result:
            return cached_result

+        # Re-checked after the cached attempt, not just in get_bypassed_page: that check
+        # ran before the queue, and this call may have spent minutes holding for LOCKED
+        # while another request collected a 429 (or collected one itself, just above).
+        # A solve cannot clear a throttle - the challenge renders, the solve "succeeds",
+        # and the cleared request is refused again while the backoff is renewed.
+        remaining = network.host_cooldown_remaining(url)
+        if remaining > 0:
+            hostname = urlparse(url).hostname or url
+            msg = (
+                f"{hostname} is rate-limited (429); skipping bypass for ~{remaining:.0f}s "
+                "until the cooldown clears."
+            )
+            raise network.RateLimitedError(msg)
+
        if env.DOCKERMODE and os.environ.get(_BYPASS_CHILD_ENV) != "1":
            return _get_via_subprocess(url, retry, cancel_flag)
        return _run_bypass_in_current_process(url, retry, cancel_flag)
@@ -1339,13 +1431,48 @@ def _start_ffmpeg_recording(display: str) -> None:
        "-an",
        output_file.as_posix(),
        "-nostats",
+        # Was "0", which discards everything including the reason it could not start.
+        # Recordings have been arriving empty with no explanation anywhere: on issue
+        # #1276 all three of a session's recordings were gone and the log said only
+        # "FFmpeg already stopped", because ffmpeg exits before creating the file when
+        # it cannot open the X display. Errors only - this is a debug-mode recorder, not
+        # something to make chatty.
        "-loglevel",
-        "0",
+        "error",
    ]
    logger.debug("Starting FFmpeg recording to %s", output_file)
    logger.debug_trace(f"FFmpeg command: {' '.join(ffmpeg_cmd)}")
-    DISPLAY["ffmpeg"] = subprocess.Popen(ffmpeg_cmd)
+    # Kept beside the recording so it travels in the debug bundle, which is the only
+    # place anyone will look for it. A file rather than a pipe: nothing here would drain
+    # a pipe, and a full one would wedge ffmpeg partway through a capture.
+    error_log = output_file.with_suffix(".ffmpeg.log")
+    try:
+        stderr_handle = error_log.open("wb")
+    except OSError as exc:
+        logger.debug("Could not open FFmpeg error log %s: %s", error_log, exc)
+        stderr_handle = None
+    DISPLAY["ffmpeg"] = subprocess.Popen(
+        ffmpeg_cmd, stderr=stderr_handle, stdout=subprocess.DEVNULL
+    )
+    if stderr_handle is not None:
+        # The child holds its own descriptor; this one has done its job.
+        stderr_handle.close()
    DISPLAY["ffmpeg_output"] = output_file
+    DISPLAY["ffmpeg_error_log"] = error_log
+
+
+def _ffmpeg_error_summary() -> str:
+    """What ffmpeg wrote to stderr, for the log line that reports it died."""
+    error_log = DISPLAY.get("ffmpeg_error_log")
+    if not error_log:
+        return "No FFmpeg error log was captured."
+    try:
+        text = Path(error_log).read_text(encoding="utf-8", errors="replace").strip()
+    except OSError as exc:
+        return f"FFmpeg error log unreadable ({exc})."
+    if not text:
+        return f"FFmpeg logged nothing to {error_log}."
+    return f"FFmpeg said: {text[-_FFMPEG_ERROR_TAIL_CHARS:]}"


 def _stop_ffmpeg_recording() -> None:
@@ -1357,9 +1484,17 @@ def _stop_ffmpeg_recording() -> None:
    if not proc:
        return
    if proc.poll() is not None:
-        logger.debug("FFmpeg already stopped")
+        # Not "already stopped" - ffmpeg was asked to record until now and is gone, so
+        # the recording for this bypass does not exist. Say so, with the reason, rather
+        # than leaving an empty recording/ directory to be discovered later.
+        logger.warning(
+            "FFmpeg exited early (code %s); no recording for this bypass. %s",
+            proc.returncode,
+            _ffmpeg_error_summary(),
+        )
        DISPLAY["ffmpeg"] = None
        DISPLAY["ffmpeg_output"] = None
+        DISPLAY["ffmpeg_error_log"] = None
        return
    try:
        proc.send_signal(signal.SIGINT)
@@ -1374,6 +1509,7 @@ def _stop_ffmpeg_recording() -> None:
            proc.kill()
    DISPLAY["ffmpeg"] = None
    DISPLAY["ffmpeg_output"] = None
+    DISPLAY["ffmpeg_error_log"] = None


 def _try_with_cached_cookies(url: str, hostname: str) -> str | None:
@@ -1400,6 +1536,20 @@ def _try_with_cached_cookies(url: str, hostname: str) -> str | None:
        if response.status_code == HTTPStatus.OK:
            logger.debug("Cached cookies worked, skipped Chrome bypass")
            return response.text
+        if response.status_code == HTTPStatus.TOO_MANY_REQUESTS:
+            # Throttled, not challenged. The clearance is still good - the origin is
+            # rate-limiting this IP and would answer 429 to a browser holding the very
+            # same cookies. Discarding it here (as every other rejection does) meant a
+            # solve won seconds earlier was thrown away and the next query bought its
+            # own 20-60s browser solve, which is itself more traffic at a host that has
+            # just asked for less. Keep it, arm the backoff, and let the caller wait.
+            wait = network.note_rate_limited(url)
+            logger.debug(
+                "Cached cookies hit a 429 for %s; keeping them and backing off ~%.0fs",
+                url,
+                wait,
+            )
+            return None
        logger.debug(
            "Cached cookies rejected (%s) for %s; discarding them",
            response.status_code,
@@ -1438,6 +1588,18 @@ def get_bypassed_page(
    attempt_url = sel.rewrite(url)
    hostname = urlparse(attempt_url).hostname or ""

+    # A 429 means the origin is throttling this IP; the challenge still renders, so a
+    # solve "succeeds" but the cleared request is rejected again and the throttle is only
+    # renewed. Never spend a minutes-long Chrome solve on a cooling-down host - fail fast
+    # so the caller waits the backoff out instead of looping the solve.
+    remaining = network.host_cooldown_remaining(attempt_url)
+    if remaining > 0:
+        msg = (
+            f"{hostname} is rate-limited (429); skipping bypass for ~{remaining:.0f}s "
+            "until the cooldown clears."
+        )
+        raise network.RateLimitedError(msg)
+
    cached_result = _try_with_cached_cookies(attempt_url, hostname)
    if cached_result:
        return cached_result
@@ -203,6 +203,21 @@ ONBOARDING = string_to_bool(os.getenv("ONBOARDING", "true"))
 _DEBUG_SKIP_SOURCES_RAW = os.getenv("DEBUG_SKIP_SOURCES", "").strip().lower()
 DEBUG_SKIP_SOURCES = {s.strip() for s in _DEBUG_SKIP_SOURCES_RAW.split(",") if s.strip()}

+# Debug: keep DDoS-Guard's __ddg8_/__ddg9_/__ddg10_ in the clearance store instead of
+# dropping them after a solve.
+#
+# Which of DDoS-Guard's cookies actually *are* clearance is not settled. The store treats
+# the trio as describing one check (client IP, timestamp, token) and drops them, on the
+# reasoning that replaying a stale IP/timestamp is what re-arms the ?check=1 loop - see
+# shelfmark.bypass.cookie_store. Field reports on issue #1276 point the other way: every
+# request after a successful solve was challenged again, which is only consistent with
+# what the store keeps not being sufficient clearance on its own.
+#
+# Deliberately env-only and off by default: this is a knob for reproducing the question
+# against a live host, not a setting to offer users. Set it to true, solve once, and watch
+# whether the next search still logs "Redirect loop detected".
+DDG_REPLAY_PER_CHECK_COOKIES = string_to_bool(os.getenv("DDG_REPLAY_PER_CHECK_COOKIES", "false"))
+

 # =============================================================================
 # Legacy migration support - will be removed in future version
@@ -430,13 +430,6 @@ def general_settings() -> list[SettingsField]:
            options=_AUDIOBOOK_FORMAT_OPTIONS,
            default=[*AUDIOBOOK_FORMATS, *ARCHIVE_FORMATS],
        ),
-        MultiSelectField(
-            key="BOOK_LANGUAGE",
-            label="Default Book Languages",
-            description="Default language filter for searches.",
-            options=_LANGUAGE_OPTIONS,
-            default=["en"],
-        ),
    ]


@@ -474,6 +467,17 @@ def search_mode_settings() -> list[SettingsField]:
            default="universal",
            user_overridable=True,
        ),
+        MultiSelectField(
+            key="BOOK_LANGUAGE",
+            label="Default Book Languages",
+            description=(
+                "Default language filter for searches. Users can override this for their "
+                "own account."
+            ),
+            options=_LANGUAGE_OPTIONS,
+            default=["en"],
+            user_overridable=True,
+        ),
        SelectField(
            key="AA_DEFAULT_SORT",
            label="Default Sort Order",
@@ -1556,6 +1560,19 @@ def download_source_settings() -> list[SettingsField]:
            min_value=1,
            max_value=60,
        ),
+        NumberField(
+            key="RELEASE_SEARCH_TIMEOUT",
+            label="Release Search Timeout (seconds)",
+            description=(
+                "How long one release search may run before it gives up and reports why. "
+                "A first search on a cold start pays for a browser solve, so leave room "
+                "for one. If you use a reverse proxy, its read timeout should be at least "
+                "this high or it will cut the search off with a 504 first."
+            ),
+            default=300,
+            min_value=30,
+            max_value=1800,
+        ),
        HeadingField(
            key="content_type_routing_heading",
            title="Content-Type Routing",
@@ -1666,6 +1683,18 @@ def cloudflare_bypass_settings() -> list[SettingsField]:
            requires_restart=True,
            show_when={"field": "USING_EXTERNAL_BYPASSER", "value": True},
        ),
+        NumberField(
+            key="BYPASS_PAGE_SOURCE_TIMEOUT",
+            label="Page Read Timeout (seconds)",
+            description=(
+                "How long to wait for a solved page to produce its content before the "
+                "bypass is retried. Raise it if solves succeed but searches still fail."
+            ),
+            default=20,
+            min_value=1,
+            max_value=120,
+            show_when={"field": "USING_EXTERNAL_BYPASSER", "value": False},
+        ),
        NumberField(
            key="BYPASS_BROWSER_IDLE_TIMEOUT",
            label="Bypasser Idle Timeout (seconds)",
@@ -7,6 +7,7 @@ that talks to /api/admin/users endpoints.

 from typing import Any

+from shelfmark.core.languages import normalize_language
 from shelfmark.core.request_policy import (
    get_source_content_type_capabilities,
    parse_policy_mode,
@@ -61,7 +62,7 @@ _SELF_SETTINGS_SECTION_OPTIONS = [
    {
        "value": "search",
        "label": "Search Preferences",
-        "description": "Show personal search mode and provider settings.",
+        "description": "Show personal search mode, language, and provider settings.",
    },
    {
        "value": "notifications",
@@ -77,8 +78,9 @@ _SEARCH_PREFERENCE_PROVIDER_KEYS = {
    "METADATA_PROVIDER_AUDIOBOOK",
    "METADATA_PROVIDER_COMBINED",
 }
-_SEARCH_PREFERENCE_VALIDATABLE_KEYS = {
+SEARCH_PREFERENCE_VALIDATABLE_KEYS = {
    "SEARCH_MODE",
+    "BOOK_LANGUAGE",
    "DEFAULT_RELEASE_SOURCE",
    "DEFAULT_RELEASE_SOURCE_AUDIOBOOK",
    "SHOW_COMBINED_SELECTOR",
@@ -178,14 +180,43 @@ def _get_request_policy_rule_columns() -> list[dict[str, object]]:
    ]


+def _validate_book_languages(value: Any) -> tuple[Any, str | None]:
+    """Validate a per-user default language list against the known languages.
+
+    Accepts the list the settings UI sends as well as a comma-separated string, so an
+    API client can spell the value the way the env var does. Blank entries are skipped
+    rather than rejected, which makes "" and "en," mean the same as [] and ["en"]. An
+    empty result is a deliberate override meaning "no default language filter", so it
+    is kept as-is; ``None`` clears the override further up the chain.
+    """
+    entries = value.split(",") if isinstance(value, str) else value
+    if not isinstance(entries, (list, tuple)):
+        return value, "BOOK_LANGUAGE must be a list of language codes"
+
+    normalized: list[str] = []
+    for entry in entries:
+        if entry is None or (isinstance(entry, str) and not entry.strip()):
+            continue
+        code = normalize_language(entry)
+        if code is None:
+            return value, f"BOOK_LANGUAGE contains an unsupported language: {entry}"
+        if code not in normalized:
+            normalized.append(code)
+
+    return normalized, None
+
+
 def validate_search_preference_value(key: str, value: Any) -> tuple[Any, str | None]:
    """Validate and normalize a search preference value for user overrides."""
-    if key not in _SEARCH_PREFERENCE_VALIDATABLE_KEYS:
+    if key not in SEARCH_PREFERENCE_VALIDATABLE_KEYS:
        return value, None

    if value is None:
        return None, None

+    if key == "BOOK_LANGUAGE":
+        return _validate_book_languages(value)
+
    normalized_value = str(value).strip()

    if key == "SEARCH_MODE":
@@ -298,7 +329,7 @@ def _on_save_users(values: dict[str, object]) -> dict[str, object]:
            }
        values["REQUEST_POLICY_RULES"] = normalized_rules

-    for key in _SEARCH_PREFERENCE_VALIDATABLE_KEYS:
+    for key in SEARCH_PREFERENCE_VALIDATABLE_KEYS:
        if key not in values:
            continue
        normalized_value, validation_error = validate_search_preference_value(key, values[key])
@@ -11,7 +11,10 @@ from shelfmark.config.notifications_settings import (
    is_valid_notification_url,
    normalize_notification_routes,
 )
-from shelfmark.config.users_settings import validate_search_preference_value
+from shelfmark.config.users_settings import (
+    SEARCH_PREFERENCE_VALIDATABLE_KEYS,
+    validate_search_preference_value,
+)
 from shelfmark.core.config import config as app_config
 from shelfmark.core.request_policy import parse_policy_mode, validate_policy_rules
 from shelfmark.core.settings_registry import load_config_file
@@ -91,13 +94,9 @@ def validate_user_settings(
            if search_validation_error:
                errors.append(search_validation_error)
                continue
-            if key in {
-                "SEARCH_MODE",
-                "METADATA_PROVIDER",
-                "METADATA_PROVIDER_AUDIOBOOK",
-                "DEFAULT_RELEASE_SOURCE",
-                "DEFAULT_RELEASE_SOURCE_AUDIOBOOK",
-            }:
+            # Every key the search validator recognises keeps its normalized value;
+            # a hand-maintained subset here silently dropped normalization for the rest.
+            if key in SEARCH_PREFERENCE_VALIDATABLE_KEYS:
                valid[key] = normalized_search_value
                continue

@@ -136,6 +136,12 @@ class DownloadTask:
        default_factory=dict
    )  # Per-output parameters (e.g. email recipient)

+    # Multi-book packs: one release holding several books. `book_plan` is the split the
+    # user approved before download (list of {title, series_position, year, files});
+    # `multi_book` asks post-processing to split heuristically when no plan exists.
+    multi_book: bool = False
+    book_plan: list[dict[str, Any]] | None = None
+
    # User association (multi-user support)
    user_id: int | None = None  # DB user ID who queued this download
    username: str | None = None  # Username for {User} template variable
@@ -0,0 +1,102 @@
+"""Pre-download release inspection: list a release's files and plan a multi-book split."""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Any
+
+from flask import jsonify, request
+
+from shelfmark.core.logger import setup_logger
+from shelfmark.core.utils import is_audiobook
+from shelfmark.download.postprocess.packs import PackFile, PackPlan, plan_pack
+from shelfmark.download.postprocess.policy import (
+    get_supported_audiobook_formats,
+    get_supported_formats,
+)
+from shelfmark.release_sources import get_handler
+
+if TYPE_CHECKING:
+    from collections.abc import Callable
+
+    from flask import Flask, Response
+
+logger = setup_logger(__name__)
+
+_INSPECT_ERRORS = (OSError, RuntimeError, ValueError, TypeError, KeyError, AttributeError)
+NOT_INSPECTABLE_REASON = "This source cannot list the release's files before downloading"
+
+
+def _serialize_plan(plan: PackPlan) -> dict[str, Any]:
+    return {
+        "is_pack": plan.is_pack,
+        "ignored": plan.ignored,
+        "books": [
+            {
+                "title": book.title,
+                "series_position": book.series_position,
+                "year": book.year,
+                "files": book.files,
+            }
+            for book in plan.books
+        ],
+    }
+
+
+def inspect_release(data: dict[str, Any]) -> dict[str, Any]:
+    """Build the inspect response for a release payload (same shape as a download)."""
+    source = str(data["source"])
+    handler = get_handler(source)
+    try:
+        files: list[PackFile] | None = handler.list_files(data)
+    except _INSPECT_ERRORS as exc:
+        logger.warning(
+            "Could not list files for %s release %s: %s", source, data.get("source_id"), exc
+        )
+        return {"inspected": False, "reason": str(exc), "files": [], "plan": None}
+
+    if files is None:
+        return {"inspected": False, "reason": NOT_INSPECTABLE_REASON, "files": [], "plan": None}
+
+    content_type = data.get("content_type")
+    supported = (
+        get_supported_audiobook_formats()
+        if is_audiobook(content_type if isinstance(content_type, str) else None)
+        else get_supported_formats()
+    )
+    series_name = data.get("series_name")
+    author_name = data.get("author")
+    plan = plan_pack(
+        files,
+        supported_extensions=set(supported),
+        series_name=series_name if isinstance(series_name, str) else None,
+        author_name=author_name if isinstance(author_name, str) else None,
+    )
+    return {
+        "inspected": True,
+        "reason": None,
+        "files": [{"path": f.path, "size": f.size} for f in files],
+        "plan": _serialize_plan(plan),
+    }
+
+
+def register_release_inspect_routes(
+    app: Flask,
+    login_required: Callable[..., Any],
+) -> None:
+    """Register POST /api/releases/inspect."""
+
+    @app.route("/api/releases/inspect", methods=["POST"])
+    @login_required
+    def api_inspect_release() -> Response | tuple[Response, int]:
+        data = request.get_json(silent=True)
+        if not isinstance(data, dict):
+            return jsonify({"error": "No data provided"}), 400
+        if not data.get("source_id"):
+            return jsonify({"error": "source_id is required"}), 400
+        if not data.get("source"):
+            return jsonify({"error": "source is required"}), 400
+        try:
+            get_handler(str(data["source"]))
+        except ValueError as exc:
+            return jsonify({"error": str(exc)}), 400
+        return jsonify(inspect_release(data))
@@ -0,0 +1,134 @@
+"""A wall-clock budget for one release search, enforced through the existing cancel flag.
+
+`/api/releases` is synchronous: the browser waits on it while the search runs. Nothing
+bounded that wait, and the bypasser's own worst case is minutes long
+(`internal_bypasser.max_duration_seconds()`), so a search that ran into an unsolvable
+protection challenge outlived every reverse proxy in front of it. The user then saw
+"Server unavailable (504)" - a gateway timeout that says nothing about what went wrong
+and points the blame at their proxy config. See issue #1276.
+
+The budget is expressed as the cancel flag the download path already understands: an
+Event armed by a timer. `html_get_page`, the bypassers and the helper subprocess all poll
+it, so an expired budget stops a solve already in flight rather than only refusing the
+next one. When it trips, the search fails with a message that names the real cause.
+
+Scoped to a context variable so it applies to the request that set it and to nothing else
+- a queued download must keep its own, much longer, budget.
+"""
+
+from __future__ import annotations
+
+import threading
+import time
+from contextlib import contextmanager
+from contextvars import ContextVar
+from typing import TYPE_CHECKING
+
+from shelfmark.core.logger import setup_logger
+
+if TYPE_CHECKING:
+    from collections.abc import Iterator
+
+logger = setup_logger(__name__)
+
+# What one search may spend. A first search on a cold start legitimately pays for a
+# browser solve - jfmlima measured 60-120s for a successful one on Anna's Archive - so
+# this cannot be as tight as a proxy's default read timeout without breaking working
+# setups. It is instead well below the ~840s the bypass path could previously reach,
+# which is what turned a failing challenge into a gateway timeout.
+DEFAULT_SEARCH_BUDGET_SECONDS = 300.0
+
+_MIN_SEARCH_BUDGET_SECONDS = 30.0
+_MAX_SEARCH_BUDGET_SECONDS = 1800.0
+
+# Raised to the caller when the budget runs out, so the API can say so plainly.
+SEARCH_DEADLINE_MESSAGE = (
+    "The release search ran out of time (%.0fs). Anna's Archive is behind a protection "
+    "challenge the bypasser could not solve in that window. Raise the release search "
+    "timeout if your setup is simply slow."
+)
+
+
+class SearchDeadline:
+    """A budget with an Event that trips when it expires."""
+
+    def __init__(self, budget_seconds: float) -> None:
+        self.budget_seconds = budget_seconds
+        self.expires_at = time.monotonic() + budget_seconds
+        # A plain threading.Event on purpose: this is handed on as a cancel flag, and
+        # that is the type the download path, the CDP worker thread and the bypass helper
+        # already poll.
+        self.event = threading.Event()
+        self._timer = threading.Timer(budget_seconds, self.event.set)
+        self._timer.daemon = True
+
+    def start(self) -> None:
+        self._timer.start()
+
+    def cancel(self) -> None:
+        self._timer.cancel()
+
+    @property
+    def remaining(self) -> float:
+        return max(0.0, self.expires_at - time.monotonic())
+
+    @property
+    def expired(self) -> bool:
+        return self.event.is_set() or self.remaining <= 0
+
+
+_current: ContextVar[SearchDeadline | None] = ContextVar("search_deadline", default=None)
+
+
+def budget_seconds() -> float:
+    """The configured budget for one release search."""
+    from shelfmark.core.config import config as app_config
+
+    raw = app_config.get("RELEASE_SEARCH_TIMEOUT", DEFAULT_SEARCH_BUDGET_SECONDS)
+    if isinstance(raw, bool) or not isinstance(raw, int | float | str):
+        return DEFAULT_SEARCH_BUDGET_SECONDS
+    try:
+        value = float(raw)
+    except TypeError, ValueError:
+        return DEFAULT_SEARCH_BUDGET_SECONDS
+    if value <= 0:
+        return DEFAULT_SEARCH_BUDGET_SECONDS
+    return min(max(value, _MIN_SEARCH_BUDGET_SECONDS), _MAX_SEARCH_BUDGET_SECONDS)
+
+
+@contextmanager
+def search_deadline(budget: float | None = None) -> Iterator[SearchDeadline]:
+    """Apply a budget to everything the calling context does."""
+    deadline = SearchDeadline(budget if budget is not None else budget_seconds())
+    token = _current.set(deadline)
+    deadline.start()
+    logger.debug("Release search budget: %.0fs", deadline.budget_seconds)
+    try:
+        yield deadline
+    finally:
+        deadline.cancel()
+        _current.reset(token)
+
+
+def current() -> SearchDeadline | None:
+    """The budget in force, or None outside a search."""
+    return _current.get()
+
+
+def expired() -> bool:
+    """Whether the budget in force has run out. False when there is no budget."""
+    deadline = _current.get()
+    return deadline is not None and deadline.expired
+
+
+def cancel_event() -> threading.Event | None:
+    """The Event that trips when the budget runs out, for use as a cancel flag."""
+    deadline = _current.get()
+    return deadline.event if deadline is not None else None
+
+
+def deadline_message() -> str:
+    """The failure to report when the budget has run out."""
+    deadline = _current.get()
+    budget = deadline.budget_seconds if deadline else DEFAULT_SEARCH_BUDGET_SECONDS
+    return SEARCH_DEADLINE_MESSAGE % budget
@@ -7,6 +7,7 @@ from dataclasses import dataclass
 from typing import TYPE_CHECKING

 from shelfmark.core.config import config
+from shelfmark.core.logger import setup_logger
 from shelfmark.metadata_providers import (
    BookMetadata,
    build_localized_search_titles,
@@ -16,6 +17,8 @@ from shelfmark.metadata_providers import (
 if TYPE_CHECKING:
    from shelfmark.core.models import SearchFilters

+logger = setup_logger(__name__)
+
 MANUAL_QUERY_MAX_LEN = 256


@@ -52,44 +55,110 @@ class ReleaseSearchPlan:
        return self.title_variants[0].query if self.title_variants else ""


-def _normalize_languages(languages: list[str] | None) -> list[str] | None:
+def _to_language_codes(values: Iterable[object], *, source: str) -> list[str] | None:
+    """Resolve any spelling of a language to the ISO code the sources expect.
+
+    Anna's Archive matches `lang=` against ISO codes: `lang=english` is not a loose
+    spelling of `lang=en`, it is a facet value AA does not have, and it filters every
+    search down to nothing. Only the *per-user* override was normalised
+    (config.users_settings.validate), so a global BOOK_LANGUAGE=english - the spelling
+    the old docs used - reached the query verbatim and silently emptied every search
+    with no error anywhere. See issue #1276.
+
+    An entry that resolves to nothing is dropped with a warning rather than passed
+    through: searching unfiltered and saying so beats reporting "no results" for a book
+    the source is full of.
+    """
+    from shelfmark.core.languages import normalize_language
+
+    codes: list[str] = []
+    unresolved: list[str] = []
+    for value in values:
+        text = str(value).strip() if value is not None else ""
+        if not text:
+            continue
+        if text.lower() == "all":
+            # An explicit "search every language", not a language.
+            return None
+        code = normalize_language(text)
+        if code is None:
+            unresolved.append(text)
+            continue
+        if code not in codes:
+            codes.append(code)
+
+    if unresolved:
+        logger.warning(
+            "Ignoring unrecognised language(s) in %s: %s. Use an ISO code such as 'en', "
+            "a three-letter code, or an English name like 'English'.",
+            source,
+            ", ".join(unresolved),
+        )
+
+    return codes or None
+
+
+def _normalize_languages(languages: list[str] | None, user_id: int | None) -> list[str] | None:
    if not languages:
-        default = getattr(config, "BOOK_LANGUAGE", None)
+        default = config.get("BOOK_LANGUAGE", None, user_id=user_id)
        if isinstance(default, str):
            default_values: list[object] = [default]
        elif isinstance(default, Iterable) and not isinstance(default, (bytes, bytearray, dict)):
            default_values = list(default)
        else:
            return None
-        return [str(lang).strip() for lang in default_values if str(lang).strip()]
+        return _to_language_codes(default_values, source="BOOK_LANGUAGE")

-    normalized: list[str] = []
-    for lang in languages:
-        if not lang:
-            continue
-        s = str(lang).strip()
-        if not s:
-            continue
-        normalized.append(s)
-
-    if any(lang.lower() == "all" for lang in normalized):
-        return None
-
-    return normalized or None
+    return _to_language_codes(languages, source="the search request")


-def _pick_search_author(book: BookMetadata) -> str:
+def first_author(value: str) -> str:
+    """The first name in a possibly comma-joined author string.
+
+    Both ends of the app hand us every contributor in one string. The frontend joins
+    `authors` with ", " for display (`bookTransformers.ts`) and that display string comes
+    straight back as the `author` request parameter, while several providers set
+    `search_author` from the same joined text. Searching a release source for
+    "Blindness Jose Saramago, Giovanni Pontiero, ..." - the author plus two translators -
+    matches nothing, and the user is told the book has no releases at all.
+
+    A "Last, First" author collapses to the surname, which is still a usable search term
+    and is what the authors[] fallback has always done with the same input. See #1252.
+    """
+    first, _, _ = value.partition(",")
+    return first.strip()
+
+
+def pick_search_author(book: BookMetadata) -> str:
+    """The one author a release query should carry, from whichever field holds one.
+
+    Every release source that builds its own query wants exactly this, so it lives here
+    rather than being re-derived per source - the two branches below drifted apart once
+    already (#1252) and the IRC source carried a third copy of the same preference.
+
+    #1290 fixed the same report by merging the two branches and trimming whichever one
+    won; this keeps that outcome ("Blindness Jose Saramago" from either field, measured
+    there at 0 releases before and 49 after) and adds the empty-narrowing fallback, so a
+    credit list that merely starts with a blank entry does not fall out to title-only.
+    """
+    # Narrowing can come back empty - the joined string starts with a comma because the
+    # first contributor was blank, and `authors.join(', ')` does not drop the empty entry.
+    # Falling through to authors[] then still finds a usable name; returning "" would
+    # search by title alone and lose the author we were holding all along.
    if book.search_author:
-        return book.search_author
+        narrowed = first_author(book.search_author)
+        if narrowed:
+            return narrowed

-    if not book.authors:
-        return ""
+    # A bare string here would otherwise be iterated one character at a time; the IRC
+    # source guarded against exactly that before it shared this helper.
+    authors = book.authors if isinstance(book.authors, list) else [book.authors or ""]
+    for author in authors:
+        narrowed = first_author(author or "")
+        if narrowed:
+            return narrowed

-    first = book.authors[0]
-    if "," in first:
-        first = first.split(",")[0].strip()
-
-    return first
+    return ""


 def _pick_search_title(book: BookMetadata) -> str:
@@ -102,15 +171,21 @@ def build_release_search_plan(
    manual_query: str | None = None,
    indexers: list[str] | None = None,
    source_filters: SearchFilters | None = None,
+    user_id: int | None = None,
 ) -> ReleaseSearchPlan:
-    """Build normalized search variants shared across release sources."""
-    resolved_languages = _normalize_languages(languages)
+    """Build normalized search variants shared across release sources.
+
+    ``user_id`` picks up that user's default languages when the caller does not
+    filter explicitly, so a search started without a language filter uses the
+    reader's own default rather than the instance-wide one.
+    """
+    resolved_languages = _normalize_languages(languages, user_id)

    resolved_manual_query = None
    if manual_query:
        resolved_manual_query = manual_query.strip()[:MANUAL_QUERY_MAX_LEN] or None

-    author = _pick_search_author(book)
+    author = pick_search_author(book)
    base_title = _pick_search_title(book)

    if resolved_manual_query:
@@ -122,7 +122,12 @@ def is_audiobook(content_type: str | None) -> bool:
 # had drifted apart: the settings UI only offered m4b/mp3/m4a, which meant a FLAC
 # audiobook could never be enabled, was silently dropped from every search result, and
 # was rejected after download as "format not supported".
-AUDIOBOOK_FORMATS = ("m4b", "mp3", "m4a", "flac", "ogg", "wma", "aac", "wav", "opus")
+#
+# "mp4" is here because some trackers (MyAnonamouse in particular) ship AAC audiobooks
+# as per-chapter .mp4 files - the same ISO-BMFF container as .m4a/.m4b, just with the
+# generic extension. Without it those releases downloaded fine and then failed
+# post-processing with "No book files found in download".
+AUDIOBOOK_FORMATS = ("m4b", "mp3", "m4a", "mp4", "flac", "ogg", "wma", "aac", "wav", "opus")

 # Multi-file audiobooks are almost always distributed as an archive. These are containers
 # rather than formats: they are what a *release* looks like, and the formats above are
@@ -76,6 +76,7 @@ _BOOK_EXTENSIONS = (
    ".m4b",
    ".mobi",
    ".mp3",
+    ".mp4",
    ".ogg",
    ".opus",
    ".pdf",
@@ -45,6 +45,10 @@ _HASH_LENGTH_ED2K = 32
 _HTTP_STATUS_FORBIDDEN = HTTPStatus.FORBIDDEN
 _HTTP_STATUS_NOT_FOUND = HTTPStatus.NOT_FOUND
 _METADATA_DOWNLOAD_STATES = {"forcedMetaDL", "metaDL"}
+# How long add_download waits for magnet metadata before falling back to the info
+# hash it already knows, rather than holding the download queue on a thin swarm.
+_METADATA_WAIT_POLLS = 20
+_METADATA_WAIT_INTERVAL_SECONDS = 0.5
 _ONE_WEEK_IN_SECONDS = 604800


@@ -221,6 +225,9 @@ class QBittorrentClient(DownloadClient):
        self._category = config_text(config.get("QBITTORRENT_CATEGORY", "books"))
        self._download_dir = config_text(config.get("QBITTORRENT_DOWNLOAD_DIR", ""))
        self._tags = _normalize_tags(config.get("QBITTORRENT_TAG", []))
+        # download_id -> qBittorrent's current primary hash, for identities that no
+        # longer match it directly. See _resolve_torrent().
+        self._primary_hashes: dict[str, str] = {}

    @property
    def _can_reauthenticate(self) -> bool:
@@ -307,13 +314,31 @@ class QBittorrentClient(DownloadClient):
        params = {"category": category} if category else {}
        return self._request_torrent_info_records(params)

+    def _remember_primary_hash(self, download_id: str, torrent: SimpleNamespace) -> None:
+        """Note the primary hash a listing scan found, so later lookups skip the scan."""
+        torrent_hash = getattr(torrent, "hash", None)
+        if isinstance(torrent_hash, str) and torrent_hash:
+            self._primary_hashes[download_id.lower()] = torrent_hash.lower()
+
    def _resolve_torrent(
        self, download_id: str, category: str | None = None
    ) -> tuple[SimpleNamespace | None, str | None]:
-        """Resolve any known torrent identity to its current qBittorrent record."""
-        torrent, error = self._get_torrent_info(download_id)
-        if error or torrent:
-            return torrent, error
+        """Resolve any known torrent identity to its current qBittorrent record.
+
+        A hybrid torrent's primary hash switches from the v1 hash to the truncated v2
+        hash once metadata resolves, so a download tracked by its v1 hash misses the
+        `hashes=` lookup and falls through to a full listing. Since `get_status()`
+        polls every couple of seconds for the life of the download, remember the
+        primary hash a scan finds and try it first.
+        """
+        cached = self._primary_hashes.get(download_id.lower())
+        for candidate in (item for item in dict.fromkeys((cached, download_id)) if item):
+            torrent, error = self._get_torrent_info(candidate)
+            if error:
+                return None, error
+            if torrent:
+                self._remember_primary_hash(download_id, torrent)
+                return torrent, None

        categories = [candidate for candidate in (category, self._category) if candidate]
        for candidate in dict.fromkeys(categories):
@@ -325,18 +350,41 @@ class QBittorrentClient(DownloadClient):
                None,
            )
            if torrent:
+                self._remember_primary_hash(download_id, torrent)
                return torrent, None

        torrents, error = self._list_torrents_by_category(None)
        if error:
            return None, error
-        return (
-            next(
-                (item for item in torrents if _torrent_matches_download_id(item, download_id)),
-                None,
-            ),
+        torrent = next(
+            (item for item in torrents if _torrent_matches_download_id(item, download_id)),
            None,
        )
+        if torrent:
+            self._remember_primary_hash(download_id, torrent)
+        else:
+            # The torrent is gone; drop the note so a re-add is not looked up by a
+            # hash that no longer exists.
+            self._primary_hashes.pop(download_id.lower(), None)
+        return torrent, None
+
+    def _current_hash(self, download_id: str) -> str:
+        """qBittorrent's current primary hash for any identity we know the torrent by.
+
+        Falls back to the given ID when the torrent cannot be found, so callers
+        still address the hash they were handed and surface the client's error.
+        """
+        try:
+            torrent, error = self._resolve_torrent(download_id)
+        except _QBITTORRENT_CLIENT_ERRORS as e:
+            logger.debug("Could not resolve current hash for %s: %s", download_id, e)
+            return download_id
+        if error or not torrent:
+            return download_id
+        torrent_hash = getattr(torrent, "hash", None)
+        if isinstance(torrent_hash, str) and torrent_hash:
+            return torrent_hash
+        return download_id

    def _list_category_hashes(self, category: str | None) -> set[str] | None:
        """Snapshot the hashes qBittorrent currently reports for a category."""
@@ -495,9 +543,13 @@ class QBittorrentClient(DownloadClient):
                    message = f"{message} (torrent file fetch failed: {torrent_info.fetch_error})"
                _raise_runtime_error(message)

-            # Wait until qBittorrent has resolved magnet metadata so the returned
-            # hash is its stable primary torrent ID, which may differ from the v1 hash.
-            for _ in range(20):
+            # Prefer qBittorrent's primary torrent ID, which for hybrid torrents
+            # switches from the v1 hash to the truncated v2 hash once metadata
+            # resolves. A magnet with few peers can take minutes to fetch metadata,
+            # and the torrent is worth keeping in the meantime: every lookup goes
+            # through `_resolve_torrent`, which still matches the v1 hash against
+            # `infohash_v1` after the primary ID has changed.
+            for _ in range(_METADATA_WAIT_POLLS):
                torrent, error = self._resolve_torrent(expected_hash, category)
                if error:
                    logger.debug("qBittorrent add_download: %s", error)
@@ -506,17 +558,18 @@ class QBittorrentClient(DownloadClient):
                    if isinstance(torrent_hash, str) and torrent_hash:
                        logger.info("Added torrent: %s", torrent_hash)
                        return torrent_hash.lower()
-                time.sleep(0.5)
+                time.sleep(_METADATA_WAIT_INTERVAL_SECONDS)

-            _raise_runtime_error(
-                "Torrent metadata resolution was not confirmed within the visibility grace period "
-                f"(response={result_text})"
+            logger.info(
+                "Added torrent %s; metadata still pending after %.0fs, tracking it by info hash",
+                expected_hash,
+                _METADATA_WAIT_POLLS * _METADATA_WAIT_INTERVAL_SECONDS,
            )
        except _QBITTORRENT_CLIENT_ERRORS:
            logger.exception("qBittorrent add failed")
            raise
        else:
-            return expected_hash
+            return expected_hash.lower()

    def get_status(self, download_id: str) -> DownloadStatus:
        """Get torrent status by hash.
@@ -529,7 +582,7 @@ class QBittorrentClient(DownloadClient):

        """
        try:
-            torrent, error = self._get_torrent_info(download_id)
+            torrent, error = self._resolve_torrent(download_id)
            if error:
                return DownloadStatus.error(error)
            if not torrent:
@@ -613,7 +666,9 @@ class QBittorrentClient(DownloadClient):

        """
        try:
-            self._client.torrents_delete(torrent_hashes=download_id, delete_files=delete_files)
+            torrent_hash = self._current_hash(download_id)
+            self._client.torrents_delete(torrent_hashes=torrent_hash, delete_files=delete_files)
+            self._primary_hashes.pop(download_id.lower(), None)
            logger.info(
                "Removed torrent from qBittorrent: %s%s",
                download_id,
@@ -635,7 +690,7 @@ class QBittorrentClient(DownloadClient):
                    logger.debug("Could not create category '%s': %s", category, e)

            self._client.torrents_set_category(
-                torrent_hashes=download_id,
+                torrent_hashes=self._current_hash(download_id),
                category=category,
            )
            logger.info("Set qBittorrent category for %s to '%s'", download_id, category)
@@ -657,7 +712,7 @@ class QBittorrentClient(DownloadClient):
        - join `save_path` with the torrent's top-level directory
        """
        try:
-            torrent, error = self._get_torrent_info(download_id)
+            torrent, error = self._resolve_torrent(download_id)
            if error:
                logger.debug("qBittorrent get_download_path: %s", error)
                return None
@@ -758,6 +813,33 @@ class QBittorrentClient(DownloadClient):
            )
            return None

+    def _await_existing_torrent(
+        self, info_hash: str, category: str | None
+    ) -> tuple[str, DownloadStatus] | None:
+        """Report a torrent already in qBittorrent, waiting out magnet metadata first."""
+        for _ in range(_METADATA_WAIT_POLLS):
+            torrent, error = self._resolve_torrent(info_hash, category)
+            if error:
+                logger.debug("qBittorrent find_existing: %s", error)
+                return None
+            if not torrent:
+                return None
+            if getattr(torrent, "state", None) not in _METADATA_DOWNLOAD_STATES:
+                torrent_hash = getattr(torrent, "hash", None)
+                if isinstance(torrent_hash, str) and torrent_hash:
+                    torrent_hash = torrent_hash.lower()
+                    return (torrent_hash, self.get_status(torrent_hash))
+            time.sleep(_METADATA_WAIT_INTERVAL_SECONDS)
+
+        # Metadata is still pending, but the torrent is here and `add_download` keeps
+        # one in this state rather than giving up. Report it by info hash so the
+        # caller joins the download in progress instead of adding a duplicate.
+        logger.info(
+            "Existing torrent %s is still fetching metadata; joining it by info hash",
+            info_hash,
+        )
+        return (info_hash.lower(), self.get_status(info_hash))
+
    def find_existing(
        self, url: str, category: str | None = None
    ) -> tuple[str, DownloadStatus] | None:
@@ -767,21 +849,9 @@ class QBittorrentClient(DownloadClient):
            if not torrent_info.info_hash:
                return None

-            for _ in range(20):
-                torrent, error = self._resolve_torrent(torrent_info.info_hash, category)
-                if error:
-                    logger.debug("qBittorrent find_existing: %s", error)
-                    return None
-                if not torrent:
-                    return None
-                if getattr(torrent, "state", None) not in _METADATA_DOWNLOAD_STATES:
-                    torrent_hash = getattr(torrent, "hash", None)
-                    if isinstance(torrent_hash, str) and torrent_hash:
-                        torrent_hash = torrent_hash.lower()
-                        return (torrent_hash, self.get_status(torrent_hash))
-                time.sleep(0.5)
+            existing = self._await_existing_torrent(torrent_info.info_hash, category)
        except _QBITTORRENT_CLIENT_ERRORS as e:
            logger.debug("Error checking for existing torrent: %s", e)
            return None
        else:
-            return None
+            return existing
@@ -80,6 +80,7 @@ _BOOK_EXTENSIONS = (
    ".m4b",
    ".mobi",
    ".mp3",
+    ".mp4",
    ".ogg",
    ".opus",
    ".pdf",
@@ -248,6 +248,15 @@ class SABnzbdClient(DownloadClient):
            if trusted_url and _url_origin(trusted_url) == target_origin:
                return True

+        named_indexers = config.get("NEWZNAB_INDEXERS", [])
+        if isinstance(named_indexers, list):
+            for row in named_indexers:
+                if not isinstance(row, dict):
+                    continue
+                trusted_url = normalize_http_config_url(row.get("url"))
+                if trusted_url and _url_origin(trusted_url) == target_origin:
+                    return True
+
        return False

    def _get_prowlarr_headers(self, url: str) -> dict:
@@ -17,6 +17,7 @@ from shelfmark.core.config import config
 from shelfmark.core.logger import setup_logger
 from shelfmark.core.utils import normalize_http_url
 from shelfmark.download.network import get_ssl_verify
+from shelfmark.download.postprocess.packs import PackFile

 logger = setup_logger(__name__)

@@ -433,6 +434,56 @@ def extract_info_hash_from_torrent(torrent_data: bytes) -> str | None:
        return None


+def _decode_torrent_text(value: object) -> str | None:
+    if isinstance(value, bytes):
+        return value.decode("utf-8", errors="replace")
+    if isinstance(value, str):
+        return value
+    return None
+
+
+def extract_file_list_from_torrent(torrent_data: bytes) -> list[PackFile] | None:
+    """List the files a .torrent describes, release-relative, without downloading it.
+
+    Multi-file torrents nest every path under the torrent name (which becomes the
+    client's save folder); single-file torrents are just the named file.
+    """
+    try:
+        decoded, _ = bencode_decode(torrent_data)
+    except _TORRENT_PARSE_ERRORS as e:
+        logger.debug("Failed to parse torrent file list: %s", e)
+        return None
+    if not isinstance(decoded, dict):
+        return None
+    info = decoded.get(b"info")
+    if not isinstance(info, dict):
+        return None
+
+    name = _decode_torrent_text(info.get(b"name")) or ""
+    raw_files = info.get(b"files")
+    if not isinstance(raw_files, list):
+        length = info.get(b"length")
+        if not name:
+            return None
+        return [PackFile(name, length if isinstance(length, int) else None)]
+
+    files: list[PackFile] = []
+    for entry in raw_files:
+        if not isinstance(entry, dict):
+            continue
+        raw_path = entry.get(b"path")
+        if not isinstance(raw_path, list):
+            continue
+        segments = [seg for seg in (_decode_torrent_text(part) for part in raw_path) if seg]
+        if not segments:
+            continue
+        if name:
+            segments.insert(0, name)
+        length = entry.get(b"length")
+        files.append(PackFile("/".join(segments), length if isinstance(length, int) else None))
+    return files
+
+
 def extract_hash_from_magnet(magnet_url: str) -> str | None:
    """Extract info_hash from a magnet URL."""
    if not magnet_url.startswith("magnet:"):
@@ -5,13 +5,14 @@ import time
 from http import HTTPStatus
 from io import BytesIO
 from typing import TYPE_CHECKING, NoReturn
-from urllib.parse import urljoin, urlparse
+from urllib.parse import parse_qsl, urlencode, urljoin, urlparse, urlunparse

 import requests
 from tqdm import tqdm

-from shelfmark.bypass import BypassCancelledError, cookie_store
+from shelfmark.bypass import BypassCancelledError, ChallengeNotSolvedError, cookie_store
 from shelfmark.bypass.challenge import challenge_marker
+from shelfmark.core import search_deadline
 from shelfmark.core.config import config as app_config
 from shelfmark.core.logger import setup_logger
 from shelfmark.core.request_helpers import coerce_bool, normalize_positive_int
@@ -28,6 +29,10 @@ logger = setup_logger(__name__)
 _RNG = random.SystemRandom()

 _MAX_REDIRECTS = 5
+# DDoS-Guard's re-check probe. Its 302 to `?check=1` is one hop of a handshake rather
+# than a page: the parameter asserts the caller already holds the cookies that hop
+# issued.
+_DDG_CHECK_PARAM = "check"
 # Z-Library answers the first hit with a 503 whose only real payload is a Set-Cookie; echoing
 # that cookie back returns the 302 to the real page. Two attempts cover the handshake without
 # letting a server that keeps re-issuing cookies hold us in the loop.
@@ -47,11 +52,13 @@ _BYPASS_GRACE_SLACK_SECONDS = 30.0
 _BYPASSER_ERRORS = (
    AttributeError,
    BypassCancelledError,
+    ChallengeNotSolvedError,
    KeyError,
    OSError,
    RuntimeError,
    TypeError,
    ValueError,
+    network.RateLimitedError,
    requests.exceptions.RequestException,
 )

@@ -250,6 +257,33 @@ def _response_challenge_marker(response: requests.Response) -> str | None:
        return None


+def _solvable_url(url: str) -> str:
+    """The URL a solver should open, given one we may be mid-handshake on.
+
+    The manual AA redirect follower in `html_get_page` walks DDoS-Guard's handshake by
+    reassigning `current_url`, so by the time a 403, a 503 challenge or a redirect loop
+    hands that URL to a bypasser it is often the `?check=1` probe rather than the page
+    we actually wanted. A solver opens it in a fresh browser holding none of the cookies
+    the probe exists to collect, so DDoS-Guard cannot verify it automatically and answers
+    with the manual CAPTCHA page that nothing can solve - the failure in #1292, where
+    FlareSolverr reported "Challenge solved!" over a 4.7 KB DDOS-GUARD interstitial.
+
+    Handing over the pre-probe URL instead lets the solver's browser run the whole
+    handshake itself, which is what a real browser does and what the solver is for.
+
+    Scoped to the hosts whose redirects we follow manually: everywhere else `check` is
+    an ordinary query parameter and none of our business.
+    """
+    if not network.should_rotate_dns_for_url(url):
+        return url
+    parsed = urlparse(url)
+    params = parse_qsl(parsed.query, keep_blank_values=True)
+    kept = [(key, value) for key, value in params if key != _DDG_CHECK_PARAM]
+    if len(kept) == len(params):
+        return url
+    return urlunparse(parsed._replace(query=urlencode(kept)))
+
+
 def _fatal_mirror_reason(e: Exception) -> str | None:
    """Return why ``e`` proves the mirror is unusable, or None if it may recover.

@@ -337,6 +371,14 @@ def html_get_page(
    # so it must be a concrete selector, not the Optional parameter.
    selector = selector or network.AAMirrorSelector()

+    # A release search runs under a wall-clock budget (see shelfmark.core.search_deadline).
+    # Adopting it as the cancel flag is what makes the budget bite on a solve already in
+    # flight: the bypassers and the helper subprocess poll this flag but know nothing about
+    # deadlines. Only when the caller has no flag of its own - a queued download brings one
+    # and must keep it, and runs outside any search context anyway.
+    if cancel_flag is None:
+        cancel_flag = search_deadline.cancel_event()
+
    def _result(html: str, response_url: str) -> str | tuple[str, str]:
        if include_response_url:
            return html, response_url
@@ -361,6 +403,16 @@ def html_get_page(
        retry-loop branch above with `continue`, and with MAX_RETRY=1 there is no
        later attempt for that branch to run on either.
        """
+        # Every handoff reaches the solver through here, so this is the one place the
+        # mid-handshake `?check=1` URL has to be unwound. See _solvable_url.
+        bypass_url = _solvable_url(bypass_url)
+        # Never start a minutes-long browser solve on a budget that has already run out:
+        # nothing downstream would get to report the real reason before the caller's
+        # deadline (or its reverse proxy) cut the request off.
+        if search_deadline.expired():
+            logger.info("Release search budget spent; not starting a bypass for %s", bypass_url)
+            return _fail(search_deadline.deadline_message(), bypass_url)
+
        if status_callback:
            status_callback("resolving", "Bypassing protection...")
        try:
@@ -377,6 +429,29 @@ def html_get_page(
                "not solved. Check that FlareSolverr/the CF bypasser is reachable.",
                bypass_url,
            )
+        except network.RateLimitedError as e:
+            # Not a bypasser malfunction: the host is throttling this IP and a solve
+            # cannot help. Surface the wait as a plain failure so the search ends cleanly
+            # instead of looping another minutes-long solve against a 429.
+            logger.info("Skipping bypass (rate-limited): %s", e)
+            if status_callback:
+                try:
+                    status_callback("resolving", "Rate limited, try again shortly")
+                except _STATUS_CALLBACK_ERRORS:
+                    logger.debug("Rate-limit status callback failed", exc_info=True)
+            return _fail(str(e), bypass_url)
+        except ChallengeNotSolvedError as e:
+            # Not a bypasser malfunction: it ran, and the host answered with something it
+            # cannot clear - DDoS-Guard's manual CAPTCHA, typically. Must precede the
+            # generic handler below, whose "the protection bypasser failed" is what sent
+            # #1292 off to fix a FlareSolverr that was working perfectly.
+            logger.info("Bypass ran but did not clear the protection: %s", e)
+            if status_callback:
+                try:
+                    status_callback("error", str(e))
+                except _STATUS_CALLBACK_ERRORS:
+                    logger.debug("Unsolved-challenge status callback failed", exc_info=True)
+            return _fail(str(e), bypass_url)
        except _BYPASSER_ERRORS as e:
            logger.warning("Bypasser error: %s: %s", type(e).__name__, e)
            # Surface the real reason. Without this the caller only sees an empty
@@ -388,6 +463,10 @@ def html_get_page(
                except _STATUS_CALLBACK_ERRORS:
                    logger.debug("Bypass error status callback failed", exc_info=True)
            if isinstance(e, BypassCancelledError):
+                # The budget trips the same cancel flag a user's cancel does, so tell them
+                # apart here - "cancelled" is a confusing thing to read when nobody did.
+                if search_deadline.expired():
+                    return _fail(search_deadline.deadline_message(), bypass_url)
                return _fail("The protection bypass was cancelled.", bypass_url)
            return _fail(f"The protection bypasser failed: {type(e).__name__}: {e}", bypass_url)
        finally:
@@ -444,6 +523,9 @@ def html_get_page(
    for attempt in range(1, retry_limit + 1):
        # Check for cancellation before each attempt
        if cancel_flag and cancel_flag.is_set():
+            if search_deadline.expired():
+                logger.info("Release search budget spent before attempt %s", attempt)
+                return _fail(search_deadline.deadline_message(), current_url)
            logger.info("html_get_page cancelled before attempt %s", attempt)
            return _fail("The request was cancelled.", current_url)

@@ -472,8 +554,15 @@ def html_get_page(
                    current_url,
                    proxies=get_proxies(current_url),
                    timeout=REQUEST_TIMEOUT,
-                    # Bypasser-derived cookies win: they came from a real solved challenge.
-                    cookies={**handshake_cookies, **cookies},
+                    # Handshake cookies win. They were issued by *this* exchange, so by
+                    # definition they are fresher than anything the store holds, and the
+                    # server is waiting to see them echoed back on the very next hop.
+                    # Letting the store overwrite them meant a stored cookie of the same
+                    # name (DDoS-Guard reuses __ddg1_/__ddg2_ for both) was replayed on
+                    # every hop and the freshly issued value never left this process - the
+                    # ?check=1 probe could then never terminate, so every request ended in
+                    # the redirect-loop handoff and paid for a full browser solve.
+                    cookies={**cookies, **handshake_cookies},
                    headers=headers,
                    allow_redirects=allow_redirects,
                    verify=get_ssl_verify(current_url),
@@ -698,6 +787,12 @@ def html_get_page(
                    f"Anna's Archive returned 404 Not Found for {current_url}.", current_url
                )

+            # 429 = origin throttling this IP. Arm the per-host backoff so selection and
+            # the bypasser stop hammering it, then fall through to normal rotation onto a
+            # mirror that is not (yet) rate-limited.
+            if status == _HTTP_STATUS_RATE_LIMITED:
+                network.note_rate_limited(current_url)
+
            # Try mirror/DNS rotation on retryable errors. A failure that proves the
            # mirror is unusable also drops it from this process's rotation, so the
            # next search does not pay for it again.
@@ -851,6 +946,7 @@ def download_url(
            # Rate limited - skip to next source immediately
            # (waiting doesn't help with concurrent downloads hitting the same server)
            if status == _HTTP_STATUS_RATE_LIMITED:
+                network.note_rate_limited(current_url)
                logger.info("Rate limited (429) - trying next source")
                if status_callback:
                    status_callback("resolving", "Server busy, trying next")
@@ -3,12 +3,13 @@
 import fnmatch
 import ipaddress
 import socket
+import time
 import urllib.parse
 import urllib.request
 from datetime import UTC, datetime, timedelta
 from http import HTTPStatus
 from socket import AddressFamily, SocketKind
-from typing import TYPE_CHECKING, Any, cast
+from typing import TYPE_CHECKING, Any, NamedTuple, cast

 import dns.resolver
 import httpx
@@ -287,6 +288,108 @@ _dead_aa_urls: set[str] = set()
 _dead_aa_urls_lock = _RLock()


+# Per-host rate-limit backoff. A 429 is the origin throttling *this IP*, not a challenge:
+# a DDoS-Guard/Cloudflare solve still renders, so the bypass "succeeds" yet the cleared
+# request is rejected again and the throttle is only renewed. The single answer is to
+# wait, so a 429 sidelines the host for a growing window - mirror selection and the
+# bypasser both skip a cooling-down host until its deadline passes. The wait escalates
+# 2 -> 5 -> 10 -> 15 -> 30 minutes each time the host throttles us again *after* we
+# already waited a full window out; a host left clear for longer than the top step
+# starts the ladder over. Keyed by host so every mirror and source shares one view;
+# in-memory only, so a restart starts clean.
+_RATE_LIMIT_COOLDOWN_LADDER_SECONDS: tuple[float, ...] = (120.0, 300.0, 600.0, 900.0, 1800.0)
+# A host that has been clear this long is treated as a fresh episode: the next 429
+# restarts the ladder at 2 minutes rather than resuming the escalation.
+_RATE_LIMIT_RESET_AFTER_SECONDS = 1800.0
+
+
+class _Cooldown(NamedTuple):
+    """One host's active rate-limit window and how far up the ladder it has climbed."""
+
+    deadline: float  # time.monotonic() value at which the wait expires
+    level: int  # index into _RATE_LIMIT_COOLDOWN_LADDER_SECONDS
+
+
+_host_cooldowns: dict[str, _Cooldown] = {}
+_host_cooldowns_lock = _RLock()
+
+
+class RateLimitedError(Exception):
+    """Raised to abandon a request whose host is in a 429 cooldown.
+
+    Not a transport failure - nothing is wrong with the network, the origin is
+    throttling this IP and only time clears it. Callers surface it as a plain failure
+    rather than retrying or handing the URL to the bypasser.
+    """
+
+
+def _cooldown_key(url: str) -> str:
+    """Host a cooldown is keyed by; '' when the URL carries none."""
+    return (urllib.parse.urlparse(url).hostname or "").lower()
+
+
+def note_rate_limited(url: str) -> float:
+    """Escalate a host's 429 backoff and (re)arm its cooldown; return the wait applied.
+
+    The step advances only when a fresh 429 arrives *after* the previous window already
+    elapsed - i.e. we waited it out and the host throttled us again. A 429 that lands
+    while the host is still cooling is the same episode: it neither escalates the level
+    nor shortens the wait. See the ladder note above.
+    """
+    host = _cooldown_key(url)
+    if not host:
+        return 0.0
+    now = time.monotonic()
+    ladder = _RATE_LIMIT_COOLDOWN_LADDER_SECONDS
+    with _host_cooldowns_lock:
+        prev = _host_cooldowns.get(host)
+        if prev is not None and now < prev.deadline:
+            # Still inside the current window - same throttling episode, leave it be.
+            return prev.deadline - now
+        if prev is None or now - prev.deadline > _RATE_LIMIT_RESET_AFTER_SECONDS:
+            level = 0
+        else:
+            level = min(prev.level + 1, len(ladder) - 1)
+        wait = ladder[level]
+        _host_cooldowns[host] = _Cooldown(deadline=now + wait, level=level)
+    logger.info(
+        "Rate limited (429): backing off %s for %.0fs (step %d/%d)",
+        host,
+        wait,
+        level + 1,
+        len(ladder),
+    )
+    return wait
+
+
+def host_cooldown_remaining(url: str) -> float:
+    """Seconds left on a host's 429 cooldown; 0.0 when clear or expired.
+
+    Leaves an expired record in place: the ladder level it carries is what a later 429
+    escalates from (or resets, once the clear gap is long enough).
+    """
+    host = _cooldown_key(url)
+    if not host:
+        return 0.0
+    now = time.monotonic()
+    with _host_cooldowns_lock:
+        rec = _host_cooldowns.get(host)
+        if rec is None or rec.deadline <= now:
+            return 0.0
+        return rec.deadline - now
+
+
+def is_host_cooling_down(url: str) -> bool:
+    """True while ``url``'s host is inside its 429 cooldown window."""
+    return host_cooldown_remaining(url) > 0.0
+
+
+def clear_host_cooldowns() -> None:
+    """Forget all rate-limit cooldowns (manual reset / tests)."""
+    with _host_cooldowns_lock:
+        _host_cooldowns.clear()
+
+
 def _ensure_initialized() -> None:
    """Lazy guard so runtime setup happens once and late calls still work."""
    global _initialized
@@ -1418,8 +1521,13 @@ def get_available_aa_urls() -> list[str]:
        if not alive and _aa_urls:
            logger.warning("All AA mirrors quarantined; retrying the full list")
            _dead_aa_urls.clear()
-            return _aa_urls.copy()
-    return alive
+            alive = _aa_urls.copy()
+    # Prefer mirrors that are not serving a 429 cooldown so rotation stops hammering a
+    # throttled host. When every live mirror is cooling, keep the full live list rather
+    # than returning nothing: selection must never be left with nowhere to point, and
+    # the bypasser's fail-fast reports the "all rate-limited" case with a clear error.
+    breathing = [url for url in alive if not is_host_cooling_down(url)]
+    return breathing or alive


 def _aa_base_for_url(url: str) -> str:
@@ -265,6 +265,8 @@ def queue_release(
        series_position = release_data.get("series_position") or extra.get("series_position")
        subtitle = release_data.get("subtitle") or extra.get("subtitle")
        language = release_data.get("language") or extra.get("language")
+        multi_book = bool(release_data.get("multi_book") or extra.get("multi_book"))
+        book_plan = _normalize_book_plan(release_data.get("book_plan") or extra.get("book_plan"))

        books_output_mode = (
            str(config.get("BOOKS_OUTPUT_MODE", "folder", user_id=user_id) or "folder")
@@ -300,6 +302,8 @@ def queue_release(
            series_position=series_position,
            subtitle=subtitle,
            language=language,
+            multi_book=multi_book or book_plan is not None,
+            book_plan=book_plan,
            search_mode=search_mode,
            output_mode=output_mode,
            output_args=output_args,
@@ -408,6 +412,33 @@ def can_retry_download_task(
    return _has_staged_retry_source(task)


+def _normalize_book_plan(value: object) -> list[dict[str, Any]] | None:
+    """Keep only well-formed pack books: a title plus a non-empty list of file paths."""
+    if not isinstance(value, list):
+        return None
+    books: list[dict[str, Any]] = []
+    for entry in value:
+        if not isinstance(entry, dict):
+            continue
+        title = normalize_optional_text(entry.get("title"))
+        raw_files = entry.get("files")
+        if title is None or not isinstance(raw_files, list):
+            continue
+        files = [f for f in raw_files if isinstance(f, str) and f.strip()]
+        if not files:
+            continue
+        year = entry.get("year")
+        books.append(
+            {
+                "title": title,
+                "series_position": _optional_number(entry.get("series_position")),
+                "year": year if isinstance(year, int) and not isinstance(year, bool) else None,
+                "files": files,
+            }
+        )
+    return books or None
+
+
 def serialize_task_for_retry(task: DownloadTask) -> dict[str, Any]:
    """Serialize the task state needed for restart-safe retries."""
    raw_search_mode = getattr(task, "search_mode", None)
@@ -437,6 +468,8 @@ def serialize_task_for_retry(task: DownloadTask) -> dict[str, Any]:
        "subtitle": getattr(task, "subtitle", None),
        "language": getattr(task, "language", None),
        "search_mode": search_mode,
+        "multi_book": bool(getattr(task, "multi_book", False)),
+        "book_plan": _normalize_book_plan(getattr(task, "book_plan", None)),
        "output_mode": getattr(task, "output_mode", None),
        "output_args": dict(raw_output_args) if isinstance(raw_output_args, dict) else {},
        "user_id": getattr(task, "user_id", None),
@@ -495,6 +528,8 @@ def _restore_task_from_retry_payload(payload: object) -> DownloadTask | None:
        subtitle=normalize_optional_text(payload.get("subtitle")),
        language=normalize_optional_text(payload.get("language")),
        search_mode=search_mode,
+        multi_book=bool(payload.get("multi_book", False)),
+        book_plan=_normalize_book_plan(payload.get("book_plan")),
        output_mode=normalize_optional_text(payload.get("output_mode")),
        output_args=dict(output_args) if isinstance(output_args, dict) else {},
        user_id=normalize_positive_int(payload.get("user_id")),
@@ -105,6 +105,7 @@ def process_folder_output(
        maybe_run_custom_script,
        prepare_output_files,
        record_step,
+        resolve_book_groups,
        transfer_book_files,
    )

@@ -260,7 +261,15 @@ def process_folder_output(
        prepared.cleanup_paths,
    )

-    message = "Complete" if len(final_paths) == 1 else f"Complete ({len(final_paths)} files)"
+    pack_groups = resolve_book_groups(
+        task, prepared.files, organization_mode=plan.organization_mode
+    )
+    if pack_groups is not None:
+        message = f"Complete ({len(pack_groups)} books, {len(final_paths)} files)"
+    elif len(final_paths) == 1:
+        message = "Complete"
+    else:
+        message = f"Complete ({len(final_paths)} files)"
    status_callback("complete", message)

    return str(final_paths[0])
@@ -0,0 +1,423 @@
+"""Multi-book ("pack") release planning.
+
+A pack is one release that contains several books: a whole-series torrent with one
+subfolder per book, or a flat folder of `Series 1.0 - Title.m4b` files. The same
+planning rules serve pre-download inspection (the file list comes from the release
+source) and post-processing (the file list comes from disk), so what the user
+approved in the modal is what gets filed.
+"""
+
+from __future__ import annotations
+
+import os
+import re
+from dataclasses import dataclass
+from pathlib import Path, PurePosixPath
+
+from shelfmark.core.utils import AUDIOBOOK_FORMATS
+
+# m4b/m4a hold a whole audiobook in one file; every other audio format (mp3, flac, ...) is
+# chaptered - many files make up one book. Ebook formats are always one file per book, so
+# only chaptered *audio* matters here. A flat folder is split one-book-per-file only when
+# none of its files are chaptered audio: a bare list of `01 - Chapter.mp3` tracks is a
+# single chaptered audiobook, not a pack of books.
+_SINGLE_FILE_AUDIO_CONTAINERS = frozenset({"m4b", "m4a"})
+_CHAPTERED_AUDIO_EXTENSIONS = frozenset(AUDIOBOOK_FORMATS) - _SINGLE_FILE_AUDIO_CONTAINERS
+
+_YEAR_SUFFIX_RE = re.compile(r"\s*\(\s*(?P<year>\d{4})\s*\)\s*$")
+_SERIES_MARKER_RE = re.compile(
+    r"""
+    ^\s*
+    (?:
+        \[\s*\#?(?P<bracket>\d+(?:\.\d+)?)\s*\]     # [03] / [#3]
+      | \#(?P<hash>\d+(?:\.\d+)?)                    # #3
+      | book\.?\s*(?P<book>\d+(?:\.\d+)?)            # Book 3 / Book. 03
+      | (?P<plain>\d+(?:\.\d+)?)(?=[\s\-:.])         # 03 - / 1.0 - / 3.
+    )
+    \s*(?:[-:.]\s*)?
+    """,
+    re.IGNORECASE | re.VERBOSE,
+)
+_SEPARATOR_CHARS = " \t-_:."
+# "Gods of Risk 2.5 - Gods of Risk": the title repeated on both sides of the position.
+_REPEATED_TITLE_RE = re.compile(
+    r"^(?P<left>.+?)\s+(?P<position>\d+(?:\.\d+)?)\s*[-:\u2013]\s*(?P<right>.+)$"
+)
+_SERIES_LABEL_WORDS = r"(?:novella|novellas|short\s+story|short|story|novel)"
+# "Uncrowned Cradle, Book 7" / "Reaper Cradle, Volume 10" / "Wintersteel (Cradle, Book 8)":
+# an explicit word marks the position at the END of the name. A bare trailing number
+# is deliberately not matched — "Title - 02" is a chapter, not a series position.
+_TRAILING_MARKER_RE = re.compile(
+    r"""
+    [\s,\-:\u2013(]*
+    (?:book|volume|vol\.?)\s*\#?(?P<position>\d+(?:\.\d+)?)
+    \s*\)?\s*$
+    """,
+    re.IGNORECASE | re.VERBOSE,
+)
+# AudiobookBay renders a file inside a folder as "<folder> <file>" with no separator,
+# so a pack row reads "Author - Title Series, Book 1 Title Series, Book 1".
+_GLUED_FOLDER_RE = re.compile(
+    r"^(?P<prefix>.+?\s[-\u2013]\s)?(?P<core>.+?)\s+(?P=core)$", re.IGNORECASE
+)
+
+
+@dataclass(frozen=True)
+class PackFile:
+    """One file inside a release, path relative to the release root."""
+
+    path: str
+    size: int | None = None
+
+
+@dataclass(frozen=True)
+class PackBook:
+    """One book split out of a pack, files as release-relative paths."""
+
+    title: str
+    series_position: float | None
+    year: int | None
+    files: list[str]
+
+
+@dataclass(frozen=True)
+class PackPlan:
+    books: list[PackBook]
+    ignored: list[str]
+
+    @property
+    def is_pack(self) -> bool:
+        return len(self.books) > 1
+
+
+@dataclass(frozen=True)
+class BookGroup:
+    """One book's on-disk files, ready for transfer."""
+
+    title: str
+    series_position: float | None
+    year: int | None
+    files: list[Path]
+
+
+def _strip_series_name(name: str, series_name: str | None) -> str:
+    if not series_name:
+        return name
+    prefix = series_name.strip()
+    if not prefix or not name.lower().startswith(prefix.lower()):
+        return name
+    remainder = name[len(prefix) :]
+    if remainder and remainder[0].isalnum():
+        return name
+    return remainder.lstrip(_SEPARATOR_CHARS)
+
+
+def _strip_series_label(work: str, series_name: str | None) -> str:
+    """Drop a leading "An <Series> Novella - " style label that some packs prepend."""
+    if not series_name:
+        return work
+    # "The Expanse" is labelled "An Expanse Novella", so match without the article.
+    core = re.sub(r"^(?:the|an?)\s+", "", series_name.strip(), flags=re.IGNORECASE)
+    if not core:
+        return work
+    pattern = re.compile(
+        rf"^(?:an?\s+|the\s+)?{re.escape(core)}\s+{_SERIES_LABEL_WORDS}\s*[-:\u2013]\s*",
+        re.IGNORECASE,
+    )
+    return pattern.sub("", work, count=1)
+
+
+def _collapse_glued_folder(name: str) -> str:
+    match = _GLUED_FOLDER_RE.match(name)
+    if not match:
+        return name
+    prefix = match.group("prefix") or ""
+    core = match.group("core")
+    # "Author - X X" → "Author - X" (the folder carried the author, the file did not).
+    return (prefix + core).strip()
+
+
+def _strip_author_name(name: str, author_name: str | None) -> str:
+    """Drop a leading "Author - " (packs are often filed as `Author - Title`)."""
+    if not author_name:
+        return name
+    prefix = author_name.strip()
+    if not prefix or not name.lower().startswith(prefix.lower()):
+        return name
+    remainder = name[len(prefix) :]
+    stripped = remainder.lstrip(_SEPARATOR_CHARS + "\u2013")
+    if stripped == remainder:  # no separator after the author: part of the title
+        return name
+    return stripped
+
+
+def _strip_trailing_series_name(work: str, series_name: str | None) -> str:
+    """Drop a trailing series name left behind by a trailing position marker."""
+    if not series_name:
+        return work
+    suffix = series_name.strip()
+    if not suffix or not work.lower().endswith(suffix.lower()):
+        return work
+    remainder = work[: -len(suffix)]
+    stripped = remainder.rstrip(_SEPARATOR_CHARS + ",(\u2013")
+    if not stripped or stripped == remainder:
+        return work
+    return stripped
+
+
+def parse_pack_book_name(
+    name: str, *, series_name: str | None, author_name: str | None = None
+) -> tuple[str, float | None, int | None]:
+    """Split a book folder/file-stem name into (title, series position, year).
+
+    Strips a leading series name, a leading position marker (`Book 3 - `, `03 - `,
+    `1.0 - `, `3. `, `[03] `, `#3 `) and a trailing `(YYYY)`. Also understands a
+    trailing marker (`Title Series, Book 3`, `Title (Series, Volume 3)`), a leading
+    `Author - `, and AudiobookBay's glued `<folder> <file>` names. Returns the name
+    unchanged with no position/year when nothing would be left of the title.
+    """
+    work = _collapse_glued_folder(name.strip())
+    work = _strip_author_name(work, author_name)
+    work = _strip_series_name(work, series_name)
+
+    year: int | None = None
+    year_match = _YEAR_SUFFIX_RE.search(work)
+    if year_match:
+        year = int(year_match.group("year"))
+        work = work[: year_match.start()]
+
+    position: float | None = None
+    repeated = _REPEATED_TITLE_RE.match(work.strip())
+    if (
+        repeated
+        and repeated.group("left").strip().lower() == repeated.group("right").strip().lower()
+    ):
+        return repeated.group("right").strip(), float(repeated.group("position")), year
+
+    marker = _SERIES_MARKER_RE.match(work)
+    if marker:
+        raw = (
+            marker.group("bracket")
+            or marker.group("hash")
+            or marker.group("book")
+            or marker.group("plain")
+        )
+        position = float(raw)
+        work = work[marker.end() :]
+    else:
+        trailing = _TRAILING_MARKER_RE.search(work)
+        if trailing and trailing.start() > 0:
+            position = float(trailing.group("position"))
+            work = _strip_trailing_series_name(work[: trailing.start()], series_name)
+
+    work = _strip_series_label(work, series_name)
+    title = work.strip().strip(_SEPARATOR_CHARS).strip()
+    if not title:
+        return name, None, None
+    return title, position, year
+
+
+def _book_from_name(
+    name: str, files: list[str], series_name: str | None, author_name: str | None = None
+) -> PackBook:
+    title, position, year = parse_pack_book_name(
+        name, series_name=series_name, author_name=author_name
+    )
+    return PackBook(title=title, series_position=position, year=year, files=files)
+
+
+def _common_root_parts(paths: list[PurePosixPath]) -> tuple[str, ...]:
+    parents = [p.parent.parts for p in paths]
+    common: list[str] = []
+    for parts in zip(*parents, strict=False):
+        if len(set(parts)) != 1:
+            break
+        common.append(parts[0])
+    return tuple(common)
+
+
+def plan_pack(
+    files: list[PackFile],
+    *,
+    supported_extensions: set[str],
+    series_name: str | None,
+    author_name: str | None = None,
+    root_depth: int | None = None,
+) -> PackPlan:
+    """Group a release's file list into books.
+
+    Files in a subfolder (relative to the common root) group by that subfolder. Files
+    directly in the root split one-book-per-file only when at least two of them carry
+    a series position in their names; otherwise they are one book (a chaptered
+    audiobook, e.g. `01.mp3`, `02.mp3`). `root_depth` fixes how many leading path
+    components form the root instead of deriving it from the files' common parent.
+    """
+    supported = {ext.lower().lstrip(".") for ext in supported_extensions}
+    book_files: list[PurePosixPath] = []
+    ignored: list[str] = []
+    for pack_file in files:
+        rel = PurePosixPath(pack_file.path.replace("\\", "/").lstrip("./"))
+        if rel.suffix.lower().lstrip(".") in supported:
+            book_files.append(rel)
+        else:
+            ignored.append(pack_file.path)
+
+    if not book_files:
+        return PackPlan(books=[], ignored=ignored)
+
+    root_parts = (
+        _common_root_parts(book_files) if root_depth is None else book_files[0].parts[:root_depth]
+    )
+    depth = len(root_parts)
+
+    root_files: list[PurePosixPath] = []
+    folders: dict[str, list[str]] = {}
+    for rel in book_files:
+        remainder = rel.parts[depth:]
+        if len(remainder) > 1:
+            folders.setdefault(remainder[0], []).append(str(rel))
+        else:
+            root_files.append(rel)
+
+    books: list[PackBook] = []
+    if root_files:
+        parsed = [
+            parse_pack_book_name(f.stem, series_name=series_name, author_name=author_name)
+            for f in root_files
+        ]
+        positions = {p[1] for p in parsed if p[1] is not None}
+        titles = {p[0].strip().lower() for p in parsed if p[0]}
+        one_book_per_file = all(
+            rel.suffix.lower().lstrip(".") not in _CHAPTERED_AUDIO_EXTENSIONS for rel in root_files
+        )
+        # Split a flat folder into a book per file only with real evidence of distinct
+        # books: two or more series positions, more than one title, and no chaptered audio
+        # (a bare list of `01 - Chapter.mp3` tracks is one book, not a pack).
+        if len(positions) >= 2 and len(titles) >= 2 and one_book_per_file:
+            books.extend(
+                PackBook(title=title, series_position=position, year=year, files=[str(f)])
+                for f, (title, position, year) in zip(root_files, parsed, strict=True)
+            )
+        elif len(root_files) == 1:
+            books.append(
+                _book_from_name(root_files[0].stem, [str(root_files[0])], series_name, author_name)
+            )
+        else:
+            group_name = root_parts[-1] if root_parts else ""
+            books.append(
+                _book_from_name(group_name, [str(f) for f in root_files], series_name, author_name)
+            )
+
+    books.extend(
+        _book_from_name(folder, paths, series_name, author_name)
+        for folder, paths in folders.items()
+    )
+    return PackPlan(books=books, ignored=ignored)
+
+
+def _relative_paths(
+    book_files: list[Path], root: Path | None = None
+) -> tuple[Path, dict[Path, str]]:
+    if root is None:
+        root = Path(os.path.commonpath([str(f.parent) for f in book_files]))
+    return root, {f: f.relative_to(root).as_posix() for f in book_files}
+
+
+def group_files_into_books(
+    book_files: list[Path],
+    *,
+    series_name: str | None,
+    author_name: str | None = None,
+    root: Path | None = None,
+) -> list[BookGroup]:
+    """Heuristically split on-disk files into books (see `plan_pack`).
+
+    `root` pins the release root when grouping a subset of a larger file set.
+    """
+    if not book_files:
+        return []
+    _root, rel_by_path = _relative_paths(book_files, root)
+    path_by_rel = {rel: path for path, rel in rel_by_path.items()}
+    extensions = {f.suffix.lower().lstrip(".") for f in book_files}
+    plan = plan_pack(
+        [PackFile(rel) for rel in rel_by_path.values()],
+        supported_extensions=extensions,
+        series_name=series_name,
+        author_name=author_name,
+        root_depth=None if root is None else 0,
+    )
+    return [
+        BookGroup(
+            title=book.title,
+            series_position=book.series_position,
+            year=book.year,
+            files=[path_by_rel[rel] for rel in book.files],
+        )
+        for book in plan.books
+    ]
+
+
+def match_plan_to_files(
+    plan: list[PackBook],
+    book_files: list[Path],
+    *,
+    series_name: str | None = None,
+    author_name: str | None = None,
+) -> list[BookGroup]:
+    """Apply an approved plan to on-disk files.
+
+    Files match by release-relative path first, then by basename (archive extraction
+    and client save paths can shift the root), then by the on-disk basename being a
+    suffix of the planned name (sources that glue folder and file names together).
+    Book files the plan does not mention fall back to heuristic grouping so nothing
+    is silently dropped.
+    """
+    if not book_files:
+        return []
+    root, rel_by_path = _relative_paths(book_files)
+    by_rel = {rel: path for path, rel in rel_by_path.items()}
+    by_name: dict[str, list[Path]] = {}
+    for path in book_files:
+        by_name.setdefault(path.name, []).append(path)
+
+    claimed: set[Path] = set()
+    groups: list[BookGroup] = []
+    for book in plan:
+        matched: list[Path] = []
+        for wanted in book.files:
+            wanted_rel = wanted.replace("\\", "/").lstrip("./")
+            candidate = by_rel.get(wanted_rel)
+            if candidate is None:
+                candidates = [
+                    p for p in by_name.get(PurePosixPath(wanted_rel).name, []) if p not in claimed
+                ]
+                candidate = candidates[0] if candidates else None
+            if candidate is None:
+                wanted_name = PurePosixPath(wanted_rel).name.lower()
+                candidates = [
+                    p
+                    for p in book_files
+                    if p not in claimed and wanted_name.endswith(p.name.lower())
+                ]
+                candidate = candidates[0] if len(candidates) == 1 else None
+            if candidate is not None and candidate not in claimed:
+                claimed.add(candidate)
+                matched.append(candidate)
+        if matched:
+            groups.append(
+                BookGroup(
+                    title=book.title,
+                    series_position=book.series_position,
+                    year=book.year,
+                    files=matched,
+                )
+            )
+
+    unmatched = [p for p in book_files if p not in claimed]
+    if unmatched:
+        groups.extend(
+            group_files_into_books(
+                unmatched, series_name=series_name, author_name=author_name, root=root
+            )
+        )
+    return groups
@@ -40,6 +40,7 @@ from .transfer import (
    build_metadata_dict,
    is_torrent_source,
    process_directory,
+    resolve_book_groups,
    resolve_hardlink_source,
    should_hardlink,
    transfer_book_files,
@@ -80,6 +81,7 @@ __all__ = [
    "process_directory",
    "record_step",
    "resolve_custom_script_target",
+    "resolve_book_groups",
    "resolve_hardlink_source",
    "run_custom_script",
    "safe_cleanup_path",
@@ -2,6 +2,7 @@

 from __future__ import annotations

+import dataclasses
 import os
 from pathlib import Path
 from typing import TYPE_CHECKING
@@ -26,6 +27,7 @@ from shelfmark.download.fs import (
 )
 from shelfmark.download.postprocess.policy import get_file_organization, get_template

+from .packs import BookGroup, PackBook, group_files_into_books, match_plan_to_files
 from .scan import collect_directory_files, scan_directory_tree
 from .types import TransferPlan
 from .workspace import safe_cleanup_path
@@ -196,6 +198,19 @@ def transfer_book_files(

    is_audiobook = check_audiobook(task.content_type)
    organization_mode = organization_mode or get_file_organization(is_audiobook=is_audiobook)
+
+    groups = resolve_book_groups(task, book_files, organization_mode=organization_mode)
+    if groups is not None:
+        return _transfer_book_groups(
+            groups,
+            destination,
+            task,
+            use_hardlink=use_hardlink,
+            is_torrent=is_torrent,
+            preserve_source=preserve_source,
+            organization_mode=organization_mode,
+        )
+
    max_attempts = _max_attempts_for_batch(len(book_files))

    final_paths: list[Path] = []
@@ -299,6 +314,101 @@ def transfer_book_files(
    return final_paths, None, op_counts


+def resolve_book_groups(
+    task: DownloadTask,
+    book_files: list[Path],
+    *,
+    organization_mode: str,
+) -> list[BookGroup] | None:
+    """Split a multi-book pack into per-book groups, or None to file as one book.
+
+    An approved `book_plan` wins; a bare `multi_book` flag falls back to heuristic
+    grouping. Organization `none` keeps files as-is, and a split that yields a single
+    group is not a pack at all.
+    """
+    if organization_mode == "none" or not (task.book_plan or task.multi_book):
+        return None
+    if task.book_plan:
+        plan = [
+            PackBook(
+                title=str(entry.get("title") or ""),
+                series_position=entry.get("series_position"),
+                year=entry.get("year"),
+                files=list(entry.get("files") or []),
+            )
+            for entry in task.book_plan
+            if isinstance(entry, dict)
+        ]
+        groups = match_plan_to_files(
+            plan, book_files, series_name=task.series_name, author_name=task.author
+        )
+    else:
+        groups = group_files_into_books(
+            book_files, series_name=task.series_name, author_name=task.author
+        )
+    return groups if len(groups) > 1 else None
+
+
+def _transfer_book_groups(
+    groups: list[BookGroup],
+    destination: Path,
+    task: DownloadTask,
+    *,
+    use_hardlink: bool,
+    is_torrent: bool,
+    preserve_source: bool,
+    organization_mode: str,
+) -> tuple[list[Path], str | None, dict[str, int]]:
+    """Transfer each book of a pack through the normal single-book path.
+
+    Each book gets an isolated task copy (the single-file path mutates `task.format`)
+    carrying its own title, position and year; the searched book's position must not
+    leak onto its siblings, while author and series name apply to all of them.
+    """
+    all_paths: list[Path] = []
+    totals: dict[str, int] = {"hardlink": 0, "copy": 0, "move": 0}
+    errors: list[str] = []
+
+    for group in groups:
+        book_task = dataclasses.replace(
+            task,
+            title=group.title or task.title,
+            year=str(group.year) if group.year is not None else None,
+            subtitle=None,
+            series_position=group.series_position,
+            multi_book=False,
+            book_plan=None,
+        )
+        paths, error, op_counts = transfer_book_files(
+            group.files,
+            destination,
+            book_task,
+            use_hardlink=use_hardlink,
+            is_torrent=is_torrent,
+            preserve_source=preserve_source,
+            organization_mode=organization_mode,
+            source_root=group.files[0].parent,
+        )
+        for op, count in op_counts.items():
+            totals[op] = totals.get(op, 0) + count
+        if error:
+            errors.append(f"{group.title}: {error}")
+            logger.warning("Task %s: pack book %r failed: %s", task.task_id, group.title, error)
+            continue
+        all_paths.extend(paths)
+
+    if not all_paths:
+        return [], "; ".join(errors) or "No book files found", totals
+    if errors:
+        logger.warning(
+            "Task %s: pack filed with %d failed book(s): %s",
+            task.task_id,
+            len(errors),
+            "; ".join(errors),
+        )
+    return all_paths, None, totals
+
+
 def process_directory(
    directory: Path,
    ingest_dir: Path,
@@ -32,6 +32,19 @@ _DEFAULT_QUERY = "The Great Gatsby"
 _warmup_thread: threading.Thread | None = None
 _warmup_lock = threading.Lock()

+# Set as soon as a real release search starts. The warm-up exists to pay the cold path
+# *before* the user does; once they have beaten it to the box there is nothing left to
+# pre-solve, and running anyway is actively harmful - the bypasser serializes on one
+# browser, so the warm-up's solve goes in front of the search the user is watching. In
+# the bundle on issue #1276 that cost a full minute of a 2m27s wait, on a container 16
+# seconds old, for a throwaway "The Great Gatsby" query nobody asked for.
+_user_search_seen = threading.Event()
+
+
+def note_user_search() -> None:
+    """Record that a real search has run, so a pending warm-up stands down."""
+    _user_search_seen.set()
+

 def _as_bool(value: object, *, default: bool) -> bool:
    """Coerce a config value that may arrive as a string, bool or None."""
@@ -85,6 +98,12 @@ def run_warmup() -> bool:
    """
    from shelfmark.core.mirrors import has_aa_mirror_configuration

+    # Checked here rather than only at schedule time: the delay is what this races with,
+    # so the user's first search usually lands *during* the wait, not before it.
+    if _user_search_seen.is_set():
+        logger.info("Search warm-up skipped: a real search got there first")
+        return False
+
    if not has_aa_mirror_configuration():
        logger.debug("Search warm-up skipped: no Anna's Archive mirrors configured")
        return False
@@ -42,6 +42,7 @@ from shelfmark.config.settings import (
    _SUPPORTED_BOOK_LANGUAGE,
    migrate_audiobook_format_settings,
 )
+from shelfmark.core import search_deadline
 from shelfmark.core.activity_view_state_service import ActivityViewStateService
 from shelfmark.core.auth_modes import (
    get_auth_check_admin_status,
@@ -62,6 +63,7 @@ from shelfmark.core.notifications import (
    notify_user,
 )
 from shelfmark.core.prefix_middleware import PrefixMiddleware
+from shelfmark.core.release_inspect_routes import register_release_inspect_routes
 from shelfmark.core.request_helpers import (
    coerce_bool,
    emit_ws_event,
@@ -1024,6 +1026,9 @@ def _serialize_release(release: Release) -> dict:
    return result


+register_release_inspect_routes(app, login_required)
+
+
@app.route("/api/releases/download", methods=["POST"])
@login_required
 def api_download_release() -> Response | tuple[Response, int]:
@@ -1150,7 +1155,7 @@ def api_config() -> Response | tuple[Response, int]:
            "build_version": BUILD_VERSION,
            "release_version": RELEASE_VERSION,
            "book_languages": _SUPPORTED_BOOK_LANGUAGE,
-            "default_language": app_config.BOOK_LANGUAGE,
+            "default_language": app_config.get("BOOK_LANGUAGE", ["en"], user_id=db_user_id),
            "supported_formats": app_config.SUPPORTED_FORMATS,
            "supported_audiobook_formats": app_config.SUPPORTED_AUDIOBOOK_FORMATS,
            "search_mode": search_mode,
@@ -1175,6 +1180,12 @@ def api_config() -> Response | tuple[Response, int]:
                [],
                user_id=db_user_id,
            ),
+            # The client must not give up before this budget does. `/api/releases`
+            # answers a spent budget with a message naming the real cause (a protection
+            # challenge nobody could solve); a browser that aborted first replaces it
+            # with a generic network/proxy error and RELEASE_SEARCH_TIMEOUT becomes a
+            # setting the user can raise with no visible effect. See issue #1285.
+            "release_search_timeout": search_deadline.budget_seconds(),
            "settings_enabled": _is_config_dir_writable(),
            "onboarding_complete": _get_onboarding_complete(),
            # Default sort orders
@@ -2840,6 +2851,7 @@ def api_releases() -> Response | tuple[Response, int]:
                    manual_query=query_text if source_query_filters is not None else manual_query,
                    indexers=indexers,
                    source_filters=source_query_filters,
+                    user_id=db_user_id,
                )

                if plan.source_filters is not None:
@@ -2892,6 +2904,8 @@ def api_releases() -> Response | tuple[Response, int]:
            if languages_param
            else None
        )
+        # Without an explicit filter the plan falls back to this user's default languages.
+        db_user_id = get_session_db_user_id(session)
        # Content type for audiobook vs ebook search
        content_type = request.args.get("content_type", "ebook").strip()

@@ -2939,6 +2953,10 @@ def api_releases() -> Response | tuple[Response, int]:
        elif provider == "manual":
            resolved_title = title_param or manual_query or "Manual Search"
            resolved_author = author_param or ""
+            # The release modal sends `authors.join(', ')` as `author`, so the commas here
+            # are joins between contributors, not part of one name. This split is the only
+            # place that knows that, so `search_author` comes from it rather than from the
+            # joined text - see issue #1252.
            authors = [a.strip() for a in resolved_author.split(",") if a.strip()]

            book = BookMetadata(
@@ -2947,7 +2965,7 @@ def api_releases() -> Response | tuple[Response, int]:
                provider_display_name="Manual Search",
                title=resolved_title,
                search_title=resolved_title,
-                search_author=resolved_author or None,
+                search_author=authors[0] if authors else None,
                authors=authors,
            )
        else:
@@ -2980,18 +2998,36 @@ def api_releases() -> Response | tuple[Response, int]:
            # Search only enabled sources
            sources_to_search = [src["name"] for src in list_available_sources() if src["enabled"]]

-        # Search each source for releases
+        # Search each source for releases.
+        #
+        # Under a wall-clock budget: this endpoint is synchronous, and the bypass path it
+        # can reach used to be allowed minutes per URL with nothing bounding the request
+        # as a whole. A search that ran into an unsolvable protection challenge therefore
+        # outlived every reverse proxy in front of it and surfaced to the user as
+        # "Server unavailable (504)" - a gateway timeout that blames their proxy for a
+        # challenge failure. The budget is shared across sources, so a stuck first source
+        # cannot spend the whole request on its own. See issue #1276.
        all_releases = []
        errors = []
        source_instances = {}  # Keep source instances for column config

-        for source_name in sources_to_search:
-            source, releases, error = _search_source_releases(source_name, book)
-            if source is not None:
-                source_instances[source_name] = source
-                all_releases.extend(releases)
-            if error is not None:
-                errors.append(error)
+        # A real search is under way, so a warm-up still sitting on its start-up delay
+        # should stand down rather than queue its throwaway solve in front of this one.
+        warmup.note_user_search()
+
+        with search_deadline.search_deadline():
+            for source_name in sources_to_search:
+                if search_deadline.expired():
+                    logger.warning("Release search budget spent; %s not searched", source_name)
+                    errors.append(f"{source_name}: {search_deadline.deadline_message()}")
+                    continue
+
+                source, releases, error = _search_source_releases(source_name, book)
+                if source is not None:
+                    source_instances[source_name] = source
+                    all_releases.extend(releases)
+                if error is not None:
+                    errors.append(error)

        # Convert Release objects to dicts
        releases_data = [_serialize_release(release) for release in all_releases]
@@ -13,6 +13,7 @@ if TYPE_CHECKING:

    from shelfmark.core.models import DownloadTask
    from shelfmark.core.search_plan import ReleaseSearchPlan
+    from shelfmark.download.postprocess.packs import PackFile

 from shelfmark.metadata_providers import BookMetadata

@@ -400,6 +401,14 @@ class DownloadHandler(ABC):
        """Return private queue-time fields needed for restart-safe retry."""
        return {}

+    def list_files(self, release_data: dict[str, Any]) -> list[PackFile] | None:
+        """Return the release's file list without downloading it.
+
+        Lets the UI review a multi-book pack before queueing. Return None when the
+        source cannot know the files ahead of time (magnet links, usenet, ...).
+        """
+        return None
+
    @abstractmethod
    def cancel(self, task_id: str) -> bool:
        """Cancel an in-progress download."""
@@ -510,6 +519,10 @@ def browse_record_to_book_metadata(
    """Convert a source-native browse record into generic book metadata."""
    resolved_title = title_override or str(record.title or "").strip() or "Unknown title"
    resolved_author = author_override or str(record.author or "").strip()
+    # `author_override` is the frontend's display string, `authors.join(', ')` - every
+    # contributor, translators included. The split below is the only place that knows the
+    # commas were joins rather than part of a name, so `search_author` is taken from it
+    # rather than from the joined text. See issue #1252.
    authors = [part.strip() for part in resolved_author.split(",") if part.strip()]
    publish_year = None

@@ -526,7 +539,7 @@ def browse_record_to_book_metadata(
        provider_display_name=get_source_display_name(record.source),
        title=resolved_title,
        search_title=resolved_title,
-        search_author=resolved_author or None,
+        search_author=authors[0] if authors else None,
        authors=authors,
        cover_url=record.preview,
        description=record.description,
@@ -1,6 +1,6 @@
 """AudiobookBay download handler - resolves magnet links and uses shared client lifecycle."""

-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Any
 from urllib.parse import urlparse

 from shelfmark.core.config import config
@@ -22,6 +22,7 @@ if TYPE_CHECKING:
    from collections.abc import Callable

    from shelfmark.core.models import DownloadTask
+    from shelfmark.download.postprocess.packs import PackFile

 logger = setup_logger(__name__)
 DEFAULT_ABB_HOSTNAME = "audiobookbay.lu"
@@ -68,6 +69,19 @@ class AudiobookBayHandler(ExternalClientHandler):
            return task_id
        return None

+    def list_files(self, release_data: dict[str, Any]) -> list[PackFile] | None:
+        """Read the torrent's file list off the detail page, without downloading."""
+        raw_url = release_data.get("download_url") or release_data.get("source_url")
+        detail_url = raw_url.strip() if isinstance(raw_url, str) else ""
+        hostname = _resolve_allowed_detail_hostname()
+        if not detail_url or not _detail_url_matches_host(detail_url, hostname):
+            logger.debug("Cannot list files for AudiobookBay release without a valid detail URL")
+            return None
+        detail_html = scraper.fetch_detail_html(detail_url, hostname)
+        if not detail_html:
+            return None
+        return scraper.extract_file_list(detail_html)
+
    def _get_client(self, protocol: str) -> DownloadClient | None:
        """Compatibility shim so module-level patching still works in tests."""
        return get_client(protocol)
@@ -2,6 +2,7 @@

 import re
 import time
+from threading import Lock
 from urllib.parse import quote, quote_plus

 import requests
@@ -10,6 +11,7 @@ from bs4 import BeautifulSoup
 from shelfmark.core.config import config
 from shelfmark.core.logger import setup_logger
 from shelfmark.download import http as downloader
+from shelfmark.download.postprocess.packs import PackFile
 from shelfmark.release_sources.audiobookbay.utils import normalize_search_punctuation

 logger = setup_logger(__name__)
@@ -32,6 +34,13 @@ FIRST_PAGE_SESSION_REFRESH_ATTEMPTS = 2
 # Legacy search parameter used by older ABB flows
 LEGACY_CATEGORY_QUERY = "undefined%2Cundefined"

+# Detail pages are fetched once and shared by inspection (file list) and download
+# (magnet link) so a "review then download" round trip costs ABB a single request.
+DETAIL_PAGE_CACHE_TTL_SECONDS = 120.0
+DETAIL_PAGE_CACHE_MAX_ENTRIES = 8
+_detail_page_cache: dict[str, tuple[float, str]] = {}
+_detail_page_cache_lock = Lock()
+
 # Precompiled patterns used while parsing result cards
 LANGUAGE_PATTERN = re.compile(r"Language:\s*([A-Za-z]+)")
 POSTED_PATTERN = re.compile(r"Posted:\s*(\d+\s+[A-Za-z]+\s+\d{4})")
@@ -39,6 +48,11 @@ FORMAT_PATTERN = re.compile(r"Format:\s*([A-Za-z0-9]+)")
 BITRATE_PATTERN = re.compile(r"Bitrate:\s*([\d]+\s*[A-Za-z/]+)")
 SIZE_PATTERN = re.compile(r"File Size:\s*([\d.]+)\s*([A-Za-z]+)")
 INFO_HASH_LABEL_PATTERN = re.compile(r"Info Hash", re.IGNORECASE)
+FILE_ROW_SIZE_PATTERN = re.compile(
+    r"^(?P<name>.+?)\s+(?P<size>\d+(?:\.\d+)?)\s*(?P<unit>Bytes?|KBs?|MBs?|GBs?|TBs?)$",
+    re.IGNORECASE,
+)
+_FILE_SIZE_MULTIPLIERS = {"b": 1, "k": 1024, "m": 1024**2, "g": 1024**3, "t": 1024**4}


 def _coerce_non_negative_float(value: object, default: float) -> float:
@@ -348,6 +362,98 @@ def search_audiobookbay(
    return results


+def _get_cached_detail_page(details_url: str) -> str | None:
+    with _detail_page_cache_lock:
+        entry = _detail_page_cache.get(details_url)
+        if entry is None:
+            return None
+        fetched_at, html = entry
+        if time.monotonic() - fetched_at > DETAIL_PAGE_CACHE_TTL_SECONDS:
+            del _detail_page_cache[details_url]
+            return None
+    return html
+
+
+def _store_cached_detail_page(details_url: str, html: str) -> None:
+    with _detail_page_cache_lock:
+        _detail_page_cache[details_url] = (time.monotonic(), html)
+        while len(_detail_page_cache) > DETAIL_PAGE_CACHE_MAX_ENTRIES:
+            oldest = min(_detail_page_cache, key=lambda key: _detail_page_cache[key][0])
+            del _detail_page_cache[oldest]
+
+
+def clear_detail_page_cache() -> None:
+    """Drop cached detail pages (used by tests)."""
+    with _detail_page_cache_lock:
+        _detail_page_cache.clear()
+
+
+def _fetch_detail_page_once(details_url: str, hostname: str) -> str:
+    session = requests.Session()
+    _bootstrap_abb_session(hostname, session, DETAIL_PAGE_RETRY_ATTEMPTS)
+    return _coerce_markup_to_html(
+        downloader.html_get_page(
+            details_url,
+            retry=DETAIL_PAGE_RETRY_ATTEMPTS,
+            use_bypasser=False,
+            allow_bypasser_fallback=False,
+            success_delay=0,
+            session=session,
+        )
+    )
+
+
+def fetch_detail_html(details_url: str, hostname: str = "audiobookbay.lu") -> str:
+    """Fetch a detail page (one retry with a fresh session), cached briefly per URL."""
+    cached = _get_cached_detail_page(details_url)
+    if cached is not None:
+        logger.debug("Reusing recently fetched detail page: %s", details_url)
+        return cached
+    detail_html = _fetch_detail_page_once(details_url, hostname)
+    if not detail_html:
+        detail_html = _fetch_detail_page_once(details_url, hostname)
+    if detail_html:
+        _store_cached_detail_page(details_url, detail_html)
+    return detail_html
+
+
+def _parse_file_row(text: str) -> PackFile | None:
+    match = FILE_ROW_SIZE_PATTERN.match(text.strip())
+    if not match:
+        return None
+    multiplier = _FILE_SIZE_MULTIPLIERS[match.group("unit")[0].lower()]
+    return PackFile(match.group("name"), int(float(match.group("size")) * multiplier))
+
+
+def extract_file_list(detail_html: str) -> list[PackFile] | None:
+    """Read the torrent file rows off a detail page.
+
+    ABB renders the torrent's file table as single-cell rows between the
+    "This is a Multifile Torrent" marker (absent for single-file torrents) and the
+    "Combined File Size" row. Returns None when the page has no such table.
+    """
+    soup = BeautifulSoup(detail_html, "html.parser")
+    rows: list[PackFile] = []
+    for row in soup.find_all("tr"):
+        cells = row.find_all("td")
+        if not cells:
+            continue
+        label = cells[0].get_text(" ", strip=True)
+        if label.lower().startswith("combined file size"):
+            return rows or None
+        if len(cells) != 1:
+            rows = []  # a two-column metadata row means we're not in the file table yet
+            continue
+        text = cells[0].get_text(" ", strip=True)
+        if "multifile torrent" in text.lower():
+            rows = []
+            continue
+        parsed = _parse_file_row(text)
+        if parsed is not None:
+            rows.append(parsed)
+    return None
+
+
 def extract_magnet_link(details_url: str, hostname: str = "audiobookbay.lu") -> str | None:
    """Extract info hash and trackers from book detail page, then construct magnet link.

@@ -360,35 +466,7 @@ def extract_magnet_link(details_url: str, hostname: str = "audiobookbay.lu") ->

    """
    try:
-        session = requests.Session()
-        _bootstrap_abb_session(hostname, session, DETAIL_PAGE_RETRY_ATTEMPTS)
-
-        # Fetch detail page
-        detail_html = _coerce_markup_to_html(
-            downloader.html_get_page(
-                details_url,
-                retry=DETAIL_PAGE_RETRY_ATTEMPTS,
-                use_bypasser=False,
-                allow_bypasser_fallback=False,
-                success_delay=0,
-                session=session,
-            )
-        )
-
-        if not detail_html:
-            session = requests.Session()
-            _bootstrap_abb_session(hostname, session, DETAIL_PAGE_RETRY_ATTEMPTS)
-            detail_html = _coerce_markup_to_html(
-                downloader.html_get_page(
-                    details_url,
-                    retry=DETAIL_PAGE_RETRY_ATTEMPTS,
-                    use_bypasser=False,
-                    allow_bypasser_fallback=False,
-                    success_delay=0,
-                    session=session,
-                )
-            )
-
+        detail_html = fetch_detail_html(details_url, hostname)
        if not detail_html:
            logger.warning("Failed to fetch details page")
            return None
@@ -6,6 +6,8 @@ import re
 import threading
 import time
 import unicodedata
+from contextlib import contextmanager
+from contextvars import ContextVar
 from dataclasses import replace
 from http import HTTPStatus
 from pathlib import Path
@@ -16,7 +18,9 @@ import requests
 from bs4 import BeautifulSoup, Tag
 from bs4.element import NavigableString

+from shelfmark.bypass.challenge import MAX_CHALLENGE_HTML_CHARS, challenge_marker
 from shelfmark.config.env import DEBUG_SKIP_SOURCES, TMP_DIR
+from shelfmark.core import search_deadline
 from shelfmark.core.config import config
 from shelfmark.core.languages import language_alias_map
 from shelfmark.core.logger import setup_logger
@@ -42,7 +46,7 @@ from shelfmark.release_sources import (
 )

 if TYPE_CHECKING:
-    from collections.abc import Callable, Iterable
+    from collections.abc import Callable, Iterable, Iterator
    from pathlib import Path
    from threading import Event

@@ -111,6 +115,16 @@ def _html_response_text(response: str | tuple[str, str]) -> str:
    return response


+def _html_response_url(response: str | tuple[str, str]) -> str | None:
+    """The URL that actually answered, when the downloader was asked to report it.
+
+    None for the plain-string shape, so a caller can fall back to what it requested.
+    """
+    if isinstance(response, tuple):
+        return response[1] or None
+    return None
+
+
 def _attr_to_str(value: object) -> str | None:
    """Convert a BeautifulSoup attribute value to a plain string."""
    if isinstance(value, str):
@@ -555,13 +569,6 @@ _AA_PAGE_MARKERS = (
    "/fast_download",
    "/slow_download",
 )
-_CHALLENGE_MARKERS = (
-    "ddos-guard",
-    "just a moment",
-    "cloudflare",
-    "checking your browser",
-    "cf-browser-verification",
-)


 def _looks_like_aa_page(html: str) -> bool:
@@ -571,12 +578,120 @@ def _looks_like_aa_page(html: str) -> bool:


 def _looks_like_challenge_page(html: str) -> bool:
-    """Whether ``html`` is a protection interstitial rather than the site behind it."""
-    lowered = html.lower()
-    return any(marker in lowered for marker in _CHALLENGE_MARKERS)
+    """Whether ``html`` is a protection interstitial rather than the site behind it.
+
+    Delegates to the shared detector rather than substring-matching here. A bare
+    "ddos-guard"/"cloudflare" scan flags the protected site's *own* pages: DDoS-Guard
+    links its endpoints on everything it fronts, and AA ships a `DDOS-GUARD` comment in
+    the inline JS on every page it serves. That misread every real AA response that was
+    not a results table as an unsolved challenge, and sent users off to fix a bypasser
+    that had just succeeded - see #1289/#1292. `challenge_marker` caps its scan at
+    64 KB, which is what separates a few-KB interstitial from the page behind it.
+    """
+    return challenge_marker(html) is not None
+
+
+# Pages already fetched during the search in flight, keyed by URL. Scoped to one
+# DirectDownload.search() so nothing is carried between requests.
+_search_page_cache: ContextVar[dict[str, tuple[str, Tag | None]] | None] = ContextVar(
+    "aa_search_page_cache", default=None
+)
+
+
+@contextmanager
+def _search_page_reuse() -> Iterator[None]:
+    """Fetch each distinct AA search URL at most once per search.
+
+    One search asks AA for the same URL more than once. The language-filter retry in
+    `search()` re-runs every title variant, and when DIRECT_DOWNLOAD_LANGUAGE_FROM_PATH
+    is on the requested language is applied locally instead of as `&lang=`, so both
+    passes build a byte-identical URL - the retry differs only in the filtering it does
+    to the response it already had. A repeat is not a cheap round trip either: AA is
+    behind DDoS-Guard, so each one is a fresh browser solve, tens of seconds that buy
+    nothing. See issue #1285.
+    """
+    token = _search_page_cache.set({})
+    try:
+        yield
+    finally:
+        _search_page_cache.reset(token)
+
+
+def _is_reusable_answer(result: tuple[str, Tag | None]) -> bool:
+    """Whether a fetched page is an answer, rather than a giving-up worth retrying.
+
+    `_fetch_search_table_uncached` exists to rotate past mirrors that are not actually AA,
+    and when it runs out of them it *returns* instead of raising: a page with no results
+    table and no marker. Storing that would hand the language-filter retry - the pass this
+    cache exists for - a mirror set that may have recovered in between (DNS rotation, a
+    mirror coming back), turning a transient outage into "this book has no releases". A
+    real "No files found." is an answer and is worth keeping.
+    """
+    html, tbody = result
+    return tbody is not None or "No files found." in html or _looks_like_aa_page(html)
+
+
+# How much of an unreadable search page to quote in the debug log. Enough to carry the
+# <head> - title, injected challenge scripts - without pasting a 180 KB page into a log
+# file that ships inside the debug bundle.
+_PAGE_FINGERPRINT_CHARS = 700
+_TITLE_RE = re.compile(r"<title[^>]*>(.*?)</title>", re.IGNORECASE | re.DOTALL)
+
+
+def _log_untabled_search_page(url: str, html: str) -> None:
+    """Record why a search page with no results table is about to be classified.
+
+    #1289 cost a full investigation because the log said only "unsolved protection
+    challenge" while FlareSolverr said "Challenge solved!", and the debug bundle carries
+    no response bodies - there was no way to tell a real AA page from an interstitial
+    after the fact. These are the facts that would have settled it in one line: the size
+    (the 64 KB cap is what separates the two), which markers matched, and the head of
+    the document.
+
+    Diagnostics must never be the reason a search fails, so this swallows its own errors.
+    """
+    try:
+        title_match = _TITLE_RE.search(html[: _PAGE_FINGERPRINT_CHARS * 4])
+        title = " ".join(title_match.group(1).split())[:120] if title_match else "<none>"
+        lowered = html.lower()
+        aa_markers = [marker for marker in _AA_PAGE_MARKERS if marker in lowered]
+        logger.info(
+            "Search page has no results table: %s (bytes=%d, title=%r, aa_markers=%s, "
+            "challenge_marker=%r, over_challenge_size_cap=%s)",
+            url,
+            len(html),
+            title,
+            aa_markers or "none",
+            challenge_marker(html),
+            len(html) > MAX_CHALLENGE_HTML_CHARS,
+        )
+        logger.debug(
+            "Untabled search page head (%d of %d bytes): %s",
+            min(len(html), _PAGE_FINGERPRINT_CHARS),
+            len(html),
+            html[:_PAGE_FINGERPRINT_CHARS],
+        )
+    except Exception:
+        logger.debug("Could not fingerprint the untabled search page", exc_info=True)


 def _fetch_search_table(url: str, selector: network.AAMirrorSelector) -> tuple[str, Tag | None]:
+    """Fetch the AA search page, reusing one already fetched during this search."""
+    cache = _search_page_cache.get()
+    if cache is not None and url in cache:
+        logger.debug("Reusing search page already fetched for this search: %s", url)
+        return cache[url]
+
+    result = _fetch_search_table_uncached(url, selector)
+
+    if cache is not None and _is_reusable_answer(result):
+        cache[url] = result
+    return result
+
+
+def _fetch_search_table_uncached(
+    url: str, selector: network.AAMirrorSelector
+) -> tuple[str, Tag | None]:
    """Fetch the AA search page, retrying past mirrors that are not actually AA.

    A parked or seized domain answers 200 with a page that has no results table and no
@@ -586,10 +701,26 @@ def _fetch_search_table(url: str, selector: network.AAMirrorSelector) -> tuple[s
    """
    attempt_url = url
    for _ in range(len(network.get_available_aa_urls()) or 1):
+        # Every mirror shares the protection, so once the search budget is gone another
+        # mirror is another full solve nobody is still waiting for.
+        if search_deadline.expired():
+            raise SearchUnavailableError(search_deadline.deadline_message())
+
+        # include_response_url is what makes the diagnostics below name the mirror that
+        # actually answered. html_get_page rotates mirrors and follows redirects on its
+        # own, so `attempt_url` is only where this iteration started: #1298's bundle
+        # reported the untabled page against annas-archive.gl when the body had come
+        # from .pk, which is precisely the triage cost #1289 added the line to remove.
        response = downloader.html_get_page(
-            attempt_url, selector=selector, allow_bypasser_fallback=True
+            attempt_url,
+            selector=selector,
+            allow_bypasser_fallback=True,
+            include_response_url=True,
        )
-        if not response:
+        html = _html_response_text(response)
+        # Checked on the body, not on `response`: with include_response_url the give-up
+        # shape is the tuple ("", url), and a tuple is truthy.
+        if not html:
            # Network/mirror exhaustion path bubbles up so API can notify clients.
            # html_get_page records the concrete give-up reason on the selector; fall
            # back to the generic line only if nothing was recorded.
@@ -598,7 +729,7 @@ def _fetch_search_table(url: str, selector: network.AAMirrorSelector) -> tuple[s
            )
            raise SearchUnavailableError(f"Unable to reach download source. {detail}")

-        html = _html_response_text(response)
+        answered_url = _html_response_url(response) or attempt_url
        soup = BeautifulSoup(html, "html.parser")
        table = soup.find("table")
        if isinstance(table, Tag):
@@ -609,20 +740,38 @@ def _fetch_search_table(url: str, selector: network.AAMirrorSelector) -> tuple[s
        if "No files found." in html:
            # A real, genuinely empty answer from a healthy mirror.
            return html, None
+
+        # A search page with no table is the one shape we cannot read off the response
+        # alone, and the response body is not in the debug bundle. Fingerprint it here
+        # so the next report says which branch fired and why, rather than costing
+        # another round of guesswork - see #1289.
+        _log_untabled_search_page(answered_url, html)
+
+        if _looks_like_aa_page(html):
+            # A real AA response in a shape the caller should report as drift. Checked
+            # ahead of the challenge branch: AA's own pages carry the protection's
+            # markers, so an interstitial is only the better explanation once the page
+            # has nothing of AA's about it. A genuine interstitial has no AA markers.
+            return html, None
        if _looks_like_challenge_page(html):
            # The bypass did not actually clear the protection - the interstitial is
            # what came back. Rotating is pointless (every mirror shares the same
            # protection) and reporting it as an empty result is worse: the user is
            # told their query found nothing when the search never ran.
+            #
+            # The wording no longer blames the bypasser outright. In #1292 it was
+            # reachable and working, and the page it was handed was DDoS-Guard's manual
+            # CAPTCHA - so "check that the bypasser is working" was the one piece of
+            # advice guaranteed to waste the reporter's time. Name the marker instead
+            # and let the two causes be told apart.
            msg = (
-                "Anna's Archive answered with an unsolved protection challenge. "
-                "Check that the bypasser is reachable and working."
+                "Anna's Archive answered with a protection challenge that was not "
+                f"cleared (marker={challenge_marker(html)!r}). If the bypasser reports "
+                "solving it, the host is serving a manual CAPTCHA that no bypasser can "
+                "answer - try again shortly. Otherwise check that the bypasser is "
+                "reachable and working."
            )
            raise SearchUnavailableError(msg)
-        if _looks_like_aa_page(html):
-            # A real AA response in a shape the caller should report as drift.
-            # Not the mirror's fault.
-            return html, None

        new_base, action = selector.next_mirror_or_rotate_dns(
            fatal=True, reason="responded without an Anna's Archive page"
@@ -1909,6 +2058,22 @@ class DirectDownloadSource(ReleaseSource):
    ) -> list[Release]:
        """Search for releases using the book's metadata.

+        The whole fan-out runs under one page cache, so a URL built twice by different
+        passes is fetched once. See `_search_page_reuse`.
+        """
+        with _search_page_reuse():
+            return self._search(book, plan, expand_search=expand_search, content_type=content_type)
+
+    def _search(
+        self,
+        book: BookMetadata,
+        plan: ReleaseSearchPlan,
+        *,
+        expand_search: bool = False,
+        content_type: str = "ebook",
+    ) -> list[Release]:
+        """Search for releases using the book's metadata.
+
        Priority: ISBN search first (most precise), then title+author fallback.
        For non-English languages, uses localized titles from book.titles_by_language.

@@ -1973,6 +2138,12 @@ class DirectDownloadSource(ReleaseSource):
            query = f"{title} {author}".strip()
            if not query:
                continue
+            # `except Exception` below keeps this loop going past a failed variant, which
+            # is right for a parse error and wrong for a spent budget: without this the
+            # variants queue up behind each other and the request outlives the caller.
+            if search_deadline.expired():
+                logger.info("Release search budget spent; skipping remaining title variants")
+                break

            logger.debug("Searching direct_download: title_author='%s', langs=%s", query, langs)
            filters = SearchFilters(lang=langs if langs is not None else [])
@@ -1986,7 +2157,11 @@ class DirectDownloadSource(ReleaseSource):
            except Exception:
                logger.exception("Search error")

-        if not all_results and any(langs for _, langs in searches):
+        if (
+            not all_results
+            and any(langs for _, langs in searches)
+            and not search_deadline.expired()
+        ):
            logger.debug(
                "No title+author results with language filter, retrying without language filter"
            )
@@ -1994,6 +2169,9 @@ class DirectDownloadSource(ReleaseSource):
                query = f"{title} {author}".strip()
                if not query:
                    continue
+                if search_deadline.expired():
+                    logger.info("Release search budget spent; skipping remaining retries")
+                    break

                logger.debug("Searching direct_download: title_author='%s', langs=[]", query)
                try:
@@ -15,6 +15,7 @@ if TYPE_CHECKING:
 from shelfmark.api.websocket import ws_manager
 from shelfmark.core.config import config
 from shelfmark.core.logger import setup_logger
+from shelfmark.core.search_plan import pick_search_author
 from shelfmark.core.utils import is_audiobook
 from shelfmark.release_sources import (
    ColumnColorHint,
@@ -394,11 +395,13 @@ class IRCReleaseSource(ReleaseSource):
        if book.search_title or book.title:
            parts.append(book.search_title or book.title)

-        if book.search_author:
-            parts.append(book.search_author)
-        elif book.authors:
-            # Use first author
-            author = book.authors[0] if isinstance(book.authors, list) else book.authors
+        # Only ever the first author: both metadata fields can arrive holding every
+        # contributor joined with ", ", and an IRC query carrying an author plus two
+        # translators matches nothing. The choice between them - and the narrowing - is
+        # `pick_search_author`, shared with the search plan so this cannot drift from it
+        # again. See issue #1252.
+        author = pick_search_author(book)
+        if author:
            parts.append(author)

        return " ".join(parts)
@@ -428,14 +431,15 @@ class IRCReleaseSource(ReleaseSource):
        "m4b": 0,
        "mp3": 1,
        "m4a": 2,
-        "flac": 3,
-        "opus": 4,
-        "ogg": 5,
-        "aac": 6,
-        "wav": 7,
-        "wma": 8,
-        "rar": 9,
-        "zip": 10,
+        "mp4": 3,
+        "flac": 4,
+        "opus": 5,
+        "ogg": 6,
+        "aac": 7,
+        "wav": 8,
+        "wma": 9,
+        "rar": 10,
+        "zip": 11,
    }

    def _convert_to_releases(
@@ -8,6 +8,7 @@ from shelfmark.core.settings_registry import (
    HeadingField,
    PasswordField,
    SettingsField,
+    TableField,
    TagListField,
    TextField,
    register_settings,
@@ -16,12 +17,36 @@ from shelfmark.core.utils import normalize_http_url


 def _test_newznab_connection(current_values: dict[str, Any] | None = None) -> dict[str, Any]:
-    """Test the Newznab connection using current form values."""
+    """Test all named Newznab connections, or the legacy connection as fallback."""
    from shelfmark.core.config import config
    from shelfmark.release_sources.newznab.api import NewznabClient
+    from shelfmark.release_sources.newznab.source import _parse_indexer_rows

    current_values = current_values or {}

+    raw_indexers = current_values.get("NEWZNAB_INDEXERS")
+    if raw_indexers is None:
+        raw_indexers = config.get("NEWZNAB_INDEXERS", [])
+    indexers = _parse_indexer_rows(raw_indexers)
+
+    if indexers:
+        details: list[str] = []
+        all_successful = True
+        for name, url, api_key in indexers:
+            try:
+                success, message = NewznabClient(url, api_key).test_connection()
+            except Exception as e:  # noqa: BLE001 — surface unexpected errors to the UI
+                success, message = False, f"Connection failed: {e!s}"
+            all_successful = all_successful and success
+            details.append(f"{name}: {message}")
+
+        summary = (
+            f"Connected to all {len(indexers)} indexers"
+            if all_successful
+            else "One or more Newznab indexers failed"
+        )
+        return {"success": all_successful, "message": summary, "details": details}
+
    raw_url = str(current_values.get("NEWZNAB_URL") or config.get("NEWZNAB_URL", "") or "")
    api_key = str(current_values.get("NEWZNAB_API_KEY") or config.get("NEWZNAB_API_KEY", "") or "")

@@ -64,25 +89,60 @@ def newznab_config_settings() -> list[SettingsField]:
            default=False,
            description="Enable searching for books via a Newznab-compatible indexer",
        ),
+        TableField(
+            key="NEWZNAB_INDEXERS",
+            label="Named Indexers",
+            description=(
+                "Add each Newznab-compatible indexer separately. The configured name is shown "
+                "beside every result from that indexer."
+            ),
+            columns=[
+                {
+                    "key": "name",
+                    "label": "Name",
+                    "type": "text",
+                    "placeholder": "NZBGeek",
+                },
+                {
+                    "key": "url",
+                    "label": "URL",
+                    "type": "text",
+                    "placeholder": "https://api.nzbgeek.info",
+                },
+                {
+                    "key": "api_key",
+                    "label": "API Key",
+                    "type": "password",
+                    "placeholder": "Optional",
+                },
+            ],
+            default=[],
+            add_label="Add Indexer",
+            empty_message=(
+                "No named indexers configured. The legacy single-indexer fields below are used "
+                "as a fallback."
+            ),
+            show_when={"field": "NEWZNAB_ENABLED", "value": True},
+        ),
        TextField(
            key="NEWZNAB_URL",
-            label="Newznab URL",
-            description="Base URL of your Newznab indexer or aggregator",
+            label="Legacy Newznab URL",
+            description="Used only when the named indexer list is empty",
            placeholder="http://nzbhydra:5076",
-            required=True,
+            required=False,
            show_when={"field": "NEWZNAB_ENABLED", "value": True},
        ),
        PasswordField(
            key="NEWZNAB_API_KEY",
-            label="API Key",
-            description="Your Newznab API key (leave blank if not required)",
+            label="Legacy API Key",
+            description="Used only with the legacy Newznab URL",
            required=False,
            show_when={"field": "NEWZNAB_ENABLED", "value": True},
        ),
        ActionButton(
            key="test_newznab",
-            label="Test Connection",
-            description="Verify your Newznab configuration",
+            label="Test Connections",
+            description="Verify every named indexer, or the legacy connection when the list is empty",
            style="primary",
            callback=_test_newznab_connection,
            show_when={"field": "NEWZNAB_ENABLED", "value": True},
@@ -4,7 +4,10 @@ from __future__ import annotations

 import re
 import time
+from dataclasses import dataclass
+from hashlib import sha256
 from typing import TYPE_CHECKING, ClassVar
+from urllib.parse import urlparse

 if TYPE_CHECKING:
    from shelfmark.core.search_plan import ReleaseSearchPlan
@@ -48,6 +51,50 @@ _DEFAULT_BOOK_CATS = [7000]
 NEWZNAB_SEARCH_TIMEOUT_SECONDS = _SEARCH_TIMEOUT


+@dataclass(frozen=True)
+class _NamedClient:
+    """A configured Newznab connection and its stable cache namespace."""
+
+    name: str
+    connection_id: str
+    client: NewznabClient
+
+
+def _parse_indexer_rows(raw: object) -> list[tuple[str, str, str]]:
+    """Normalize structured Newznab indexer settings.
+
+    Invalid/incomplete rows are ignored so one partially edited row cannot disable
+    the other configured indexers.
+    """
+    if not isinstance(raw, list):
+        return []
+
+    indexers: list[tuple[str, str, str]] = []
+    seen_connections: set[tuple[str, str]] = set()
+    for row in raw:
+        if not isinstance(row, dict):
+            continue
+        raw_url = str(row.get("url") or "").strip()
+        url = normalize_http_url(raw_url)
+        if not url:
+            if raw_url:
+                logger.warning("Newznab: ignoring indexer row with invalid URL '%s'", raw_url)
+            continue
+
+        api_key = str(row.get("api_key") or "").strip()
+        connection_key = (url, api_key)
+        if connection_key in seen_connections:
+            continue
+        seen_connections.add(connection_key)
+
+        configured_name = str(row.get("name") or "").strip()
+        hostname = urlparse(url).hostname or ""
+        name = configured_name or hostname or "Newznab"
+        indexers.append((name, url, api_key))
+
+    return indexers
+
+
 def _parse_category_ids(raw: object) -> list[int]:
    """Parse a configured category setting into Newznab category IDs.

@@ -146,8 +193,11 @@ def _newznab_result_to_release(
        else None
    )

-    # Build source_id from GUID
-    source_id = result.get("guid") or f"newznab:{hash(raw_title)}"
+    # Namespace IDs from named connections so identical GUIDs returned by two
+    # indexers cannot overwrite one another in the private release cache.
+    raw_source_id = result.get("guid") or f"newznab:{hash(raw_title)}"
+    connection_id = str(result.get("_newznab_connection_id") or "").strip()
+    source_id = f"newznab:{connection_id}:{raw_source_id}" if connection_id else raw_source_id

    # Cache the raw result for the handler
    cache_release(source_id, result)
@@ -272,6 +322,7 @@ class NewznabSource(ReleaseSource):
        )

    def _get_client(self) -> NewznabClient | None:
+        """Build the legacy single-indexer client."""
        raw_url = str(config.get("NEWZNAB_URL", "") or "")
        api_key = str(config.get("NEWZNAB_API_KEY", "") or "")

@@ -284,6 +335,28 @@ class NewznabSource(ReleaseSource):

        return NewznabClient(url, api_key or "")

+    def _get_clients(self) -> list[_NamedClient]:
+        """Build named clients, falling back to the legacy single connection."""
+        configured = _parse_indexer_rows(config.get("NEWZNAB_INDEXERS", []))
+        if configured:
+            clients: list[_NamedClient] = []
+            for name, url, api_key in configured:
+                digest = sha256(f"{name}\0{url}\0{api_key}".encode()).hexdigest()[:16]
+                clients.append(
+                    _NamedClient(
+                        name=name,
+                        connection_id=digest,
+                        client=NewznabClient(url, api_key),
+                    )
+                )
+            return clients
+
+        legacy_client = self._get_client()
+        if legacy_client is None:
+            return []
+        legacy_name = str(config.get("NEWZNAB_NAME", "") or "").strip() or "Newznab"
+        return [_NamedClient(name=legacy_name, connection_id="legacy", client=legacy_client)]
+
    def search(
        self,
        book: BookMetadata,
@@ -293,8 +366,8 @@ class NewznabSource(ReleaseSource):
        content_type: str = "ebook",
    ) -> list[Release]:
        """Search the Newznab indexer for releases matching the book."""
-        client = self._get_client()
-        if not client:
+        clients = self._get_clients()
+        if not clients:
            logger.warning("Newznab not configured - skipping search")
            return []

@@ -324,40 +397,60 @@ class NewznabSource(ReleaseSource):
        all_results: list[dict] = []

        try:
-            for idx, query in enumerate(queries, start=1):
-                _check_timeout()
-                if len(queries) > 1:
-                    logger.debug("Newznab query %d/%d: '%s'", idx, len(queries), query)
+            for connection in clients:
+                try:
+                    for idx, query in enumerate(queries, start=1):
+                        _check_timeout()
+                        if len(queries) > 1:
+                            logger.debug(
+                                "Newznab [%s] query %d/%d: '%s'",
+                                connection.name,
+                                idx,
+                                len(queries),
+                                query,
+                            )

-                raw = client.search(query=query, categories=categories)
+                        raw = connection.client.search(query=query, categories=categories)

-                # Auto-expand: retry without category filter if no results
-                if not raw and categories and auto_expand:
-                    _check_timeout()
-                    logger.info(
-                        "Newznab: no results for '%s' with category filter, auto-expanding",
-                        query,
-                    )
-                    raw = client.search(query=query, categories=None)
+                        # Auto-expand: retry without category filter if no results
+                        if not raw and categories and auto_expand:
+                            _check_timeout()
+                            logger.info(
+                                "Newznab [%s]: no results for '%s' with category filter, "
+                                "auto-expanding",
+                                connection.name,
+                                query,
+                            )
+                            raw = connection.client.search(query=query, categories=None)

-                for r in raw:
-                    key = (
-                        r.get("guid")
-                        or r.get("downloadUrl")
-                        or f"{r.get('indexer')}:{r.get('title')}"
-                    )
-                    if key in seen_keys:
-                        continue
-                    seen_keys.add(key)
-                    all_results.append(r)
+                        for raw_result in raw:
+                            r = dict(raw_result)
+                            # Aggregators can identify the underlying indexer. Plain feeds
+                            # generally cannot, so use the user-configured connection name.
+                            r["indexer"] = r.get("indexer") or connection.name
+                            r["_newznab_connection_id"] = connection.connection_id
+                            key = (
+                                connection.connection_id,
+                                r.get("guid")
+                                or r.get("downloadUrl")
+                                or f"{r.get('indexer')}:{r.get('title')}",
+                            )
+                            if key in seen_keys:
+                                continue
+                            seen_keys.add(key)
+                            all_results.append(r)
+                except TimeoutError:
+                    raise
+                except Exception:
+                    logger.exception("Newznab search failed for %s", connection.name)

        except TimeoutError as e:
            logger.warning("Newznab search timed out: %s", e)
-        except Exception:
-            logger.exception("Newznab search failed")
-            return []

        results = [_newznab_result_to_release(r, content_type, categories) for r in all_results]
+        if plan.indexers:
+            selected_indexers = set(plan.indexers)
+            results = [r for r in results if r.indexer in selected_indexers]

        if results:
            nzb_count = sum(1 for r in results if r.protocol == ReleaseProtocol.NZB)
@@ -379,5 +472,7 @@ class NewznabSource(ReleaseSource):
    def is_available(self) -> bool:
        if not config.get("NEWZNAB_ENABLED", False):
            return False
+        if _parse_indexer_rows(config.get("NEWZNAB_INDEXERS", [])):
+            return True
        url = normalize_http_url(str(config.get("NEWZNAB_URL", "") or ""))
        return bool(url)
@@ -28,6 +28,10 @@ from shelfmark.download.clients.base_handler import (
    DownloadRequest,
    ExternalClientHandler,
 )
+from shelfmark.download.clients.torrent_utils import (
+    extract_file_list_from_torrent,
+    extract_torrent_info,
+)
 from shelfmark.metadata_providers import BookMetadata
 from shelfmark.release_sources import register_handler
 from shelfmark.release_sources.prowlarr.api import IndexerSeedSettings, ProwlarrClient
@@ -38,12 +42,14 @@ from shelfmark.release_sources.prowlarr.utils import (
    coerce_int_like,
    get_preferred_download_url,
    get_protocol,
+    sanitize_download_url,
 )

 if TYPE_CHECKING:
    from collections.abc import Callable

    from shelfmark.core.models import DownloadTask
+    from shelfmark.download.postprocess.packs import PackFile

 logger = setup_logger(__name__)

@@ -127,6 +133,24 @@ class ProwlarrHandler(ExternalClientHandler):

        return settings.get(indexer_id)

+    def list_files(self, release_data: dict[str, Any]) -> list[PackFile] | None:
+        """List a cached torrent release's files from its .torrent, without downloading.
+
+        Magnet-only and usenet releases cannot be listed ahead of time.
+        """
+        source_id = str(release_data.get("source_id") or "")
+        prowlarr_result = get_release(source_id) if source_id else None
+        if not prowlarr_result or get_protocol(prowlarr_result) != "torrent":
+            return None
+        download_url = sanitize_download_url(str(prowlarr_result.get("downloadUrl") or "").strip())
+        if not download_url or download_url.startswith("magnet:"):
+            return None
+        expected_hash = str(prowlarr_result.get("infoHash") or "").strip() or None
+        info = extract_torrent_info(download_url, expected_hash=expected_hash)
+        if not info.torrent_data:
+            return None
+        return extract_file_list_from_torrent(info.torrent_data)
+
    def _get_client(self, protocol: str) -> DownloadClient | None:
        """Compatibility shim so module-level patching still works in tests."""
        return get_client(protocol)
@@ -297,6 +321,8 @@ class ProwlarrHandler(ExternalClientHandler):
            search_title=title,
            search_author=task.author,
        )
+        # No language default here on purpose: this re-finds one exact release by its
+        # guid, and Prowlarr does not filter on plan.languages anyway.
        plan = build_release_search_plan(
            book,
            indexers=[indexer] if indexer is not None else None,
@@ -41,6 +41,8 @@ from shelfmark.release_sources.prowlarr.api import (
 )
 from shelfmark.release_sources.prowlarr.cache import cache_release
 from shelfmark.release_sources.prowlarr.utils import (
+    AUTHOR_UNKNOWN,
+    author_affinity,
    build_source_id,
    coerce_float_like,
    coerce_int_like,
@@ -145,6 +147,36 @@ def _build_indexer_priority(indexers: list[dict]) -> dict[int, int]:
    return priority


+def _drop_unknown_indexer_ids(
+    selected_ids: list[int] | None, indexers: list[dict]
+) -> list[int] | None:
+    """Keep only selected indexer ids Prowlarr still serves.
+
+    An indexer removed or disabled in Prowlarr stays in the saved selection,
+    where settings can no longer show it - so it cannot be unselected, and every
+    search keeps querying an indexer that is gone (#1283). Dropping it here
+    keeps the saved selection intact for an indexer that comes back.
+    """
+    if selected_ids is None:
+        return None
+
+    live_ids = {
+        indexer_id
+        for indexer in indexers
+        if (indexer_id := _coerce_indexer_id(indexer.get("id"))) is not None
+    }
+    kept = [indexer_id for indexer_id in selected_ids if indexer_id in live_ids]
+
+    stale = [indexer_id for indexer_id in selected_ids if indexer_id not in live_ids]
+    if stale:
+        logger.warning(
+            "Skipping selected Prowlarr indexers that are no longer enabled in Prowlarr: %s",
+            stale,
+        )
+
+    return kept
+
+
 def _rank_for_indexer_id(indexer_id: object, priority: dict[int, int]) -> int:
    """Preference rank for an indexer id. Lower wins, unknown ranks last."""
    coerced = _coerce_indexer_id(indexer_id)
@@ -317,19 +349,25 @@ def _extract_mam_language(raw_title: str) -> str | None:
    return None


-def _extract_mam_formats(raw_title: str) -> list[str]:
-    """Extract a list of formats from MyAnonamouse titles.
+def _split_mam_formats(raw_title: str) -> tuple[list[str], list[str]]:
+    """Split the format tokens of a MyAnonamouse title into (recognized, unrecognized).

    Prowlarr's MAM parser appends a structured bracket segment like:
      [ENG / EPUB MOBI PDF]

    We only trust this structured segment (and do not attempt generic title
    heuristics for other indexers).
+
+    Tokens after the "/" that Shelfmark does not know as a book or audiobook format
+    (e.g. ``[ENG / AVI]``) are returned separately so the UI can warn that the release
+    will download but cannot be processed, instead of showing a bare content-type icon
+    that looks like an ordinary result.
    """
    if not raw_title:
-        return []
+        return [], []

    format_set = set(ALL_BOOK_FORMATS)
+    first_unrecognized: list[str] | None = None
    for bracket in re.findall(r"\[([^\]]+)\]", raw_title):
        if "/" not in bracket:
            continue
@@ -338,15 +376,26 @@ def _extract_mam_formats(raw_title: str) -> list[str]:
        tokens = re.findall(r"[A-Za-z0-9]+", after_slash)

        formats: list[str] = []
+        unrecognized: list[str] = []
        for token in tokens:
            fmt = token.lower()
-            if fmt in format_set and fmt not in formats:
-                formats.append(fmt)
+            if fmt in format_set:
+                if fmt not in formats:
+                    formats.append(fmt)
+            elif fmt not in unrecognized:
+                unrecognized.append(fmt)

        if formats:
-            return formats
+            return formats, unrecognized
+        if unrecognized and first_unrecognized is None:
+            first_unrecognized = unrecognized

-    return []
+    return [], first_unrecognized or []
+
+
+def _extract_mam_formats(raw_title: str) -> list[str]:
+    """Extract the recognized formats from a MyAnonamouse title (see _split_mam_formats)."""
+    return _split_mam_formats(raw_title)[0]


 def _formats_display(formats: list[str]) -> str | None:
@@ -485,6 +534,7 @@ def _prowlarr_result_to_release(

    format_detected: str | None = None
    formats: list[str] = []
+    unrecognized_formats: list[str] = []
    formats_display: str | None = None
    language_detected: str | None = None
    if enable_format_detection:
@@ -492,7 +542,7 @@ def _prowlarr_result_to_release(
        if book_title:
            title = book_title

-        formats = _extract_mam_formats(str(raw_title or ""))
+        formats, unrecognized_formats = _split_mam_formats(str(raw_title or ""))
        format_detected = formats[0] if formats else None
        formats_display = _formats_display(formats)
        language_detected = _extract_mam_language(str(raw_title or ""))
@@ -554,6 +604,9 @@ def _prowlarr_result_to_release(
            "info_hash": result.get("infoHash"),
            "formats": formats or None,
            "formats_display": formats_display,
+            # Format tokens the indexer declared but Shelfmark can't process (e.g. a MAM
+            # "[ENG / AVI]"). Lets the UI warn instead of showing a bare content icon.
+            "unrecognized_formats": unrecognized_formats or None,
            # Raw torznab attributes for rich tooltips (enriched indexers)
            "torznab_attrs": result.get("torznabAttrs"),
        },
@@ -940,6 +993,7 @@ class ProwlarrSource(ReleaseSource):
                # found for this book" - the same lie as a swallowed timeout (#1249).
                msg = f"could not reach Prowlarr: {e}"
                raise SourceUnavailableError(msg) from e
+            indexer_ids = _drop_unknown_indexer_ids(indexer_ids, enabled_indexers)
            indexer_priority = _build_indexer_priority(enabled_indexers)
            # Some indexers benefit from title+author queries and extra format detection.
            enriched_indexer_ids = client.get_enriched_indexer_ids(
@@ -956,10 +1010,17 @@ class ProwlarrSource(ReleaseSource):
                if time.monotonic() > deadline:
                    _raise_timeout_error(f"Prowlarr search timed out after {int(search_budget)}s")

-            def search_indexers(
-                query: str, cats: list[int] | None, *, enriched_query: str | None = None
-            ) -> _IndexerSearchOutcome:
-                """Search indexers with given categories via Torznab/Newznab."""
+            def search_indexers(query: str, cats: list[int] | None) -> _IndexerSearchOutcome:
+                """Search indexers with given categories via Torznab/Newznab.
+
+                Every indexer gets the same title-only query. Enriched indexers used
+                to be sent "{title} {author}", but an indexer that ANDs its search
+                terms (MyAnonamouse) returns nothing whenever the metadata provider
+                spells the author differently to the tracker - "Timothy Ferriss" vs
+                "Tim Ferriss" - and the UI reports the book as missing (#1293). The
+                author still decides ordering below, where a spelling difference
+                costs a release its position rather than its existence.
+                """
                outcome = _IndexerSearchOutcome(results=[])
                target_indexer_ids = self._get_search_indexer_ids(client, indexer_ids, cats)
                if not target_indexer_ids:
@@ -967,16 +1028,11 @@ class ProwlarrSource(ReleaseSource):

                for indexer_id in target_indexer_ids:
                    _check_timeout()
-                    indexer_query = (
-                        enriched_query
-                        if indexer_id in enriched_indexer_ids_set and enriched_query
-                        else query
-                    )
                    outcome.attempted += 1
                    try:
                        raw = client.torznab_search(
                            indexer_id=indexer_id,
-                            query=indexer_query,
+                            query=query,
                            categories=cats,
                            search_type="book",
                        )
@@ -1001,14 +1057,11 @@ class ProwlarrSource(ReleaseSource):
            for idx, variant in enumerate(variants, start=1):
                _check_timeout()
                query = variant.title
-                enriched_query = variant.query  # title + author

                if len(variants) > 1:
                    logger.debug("Prowlarr query %s/%s: '%s'", idx, len(variants), query)

-                outcome = search_indexers(
-                    query=query, cats=categories, enriched_query=enriched_query
-                )
+                outcome = search_indexers(query=query, cats=categories)

                # Auto-expand: if no results with categories and auto-expand enabled, retry without.
                # Only when every indexer actually answered: a failed search says nothing about
@@ -1025,9 +1078,7 @@ class ProwlarrSource(ReleaseSource):
                        "Prowlarr: no results for query '%s' with category filter, auto-expanding search",
                        query,
                    )
-                    expanded = search_indexers(
-                        query=query, cats=None, enriched_query=enriched_query
-                    )
+                    expanded = search_indexers(query=query, cats=None)
                    outcome.results = expanded.results
                    outcome.attempted += expanded.attempted
                    outcome.failed += expanded.failed
@@ -1065,6 +1116,10 @@ class ProwlarrSource(ReleaseSource):

            results: list[Release] = []
            enriched_source_ids: set[str] = set()
+            affinity_by_source_id: dict[str, int] = {}
+            # A manual query is the user's own words; ranking it against the
+            # metadata author would second-guess what they typed.
+            wanted_author = "" if plan.manual_query else plan.author

            for raw_result in all_results:
                result_with_seed_settings = _apply_indexer_seed_settings(
@@ -1084,13 +1139,20 @@ class ProwlarrSource(ReleaseSource):
                if idx_id_int is not None and idx_id_int in indexer_priority:
                    release.extra["indexer_priority"] = indexer_priority[idx_id_int]
                results.append(release)
+                affinity_by_source_id[release.source_id] = author_affinity(
+                    wanted_author, release.extra.get("author")
+                )

                if is_enriched:
                    enriched_source_ids.add(release.source_id)

+            # Indexer priority first: it is an explicit user preference. Author
+            # agreement then orders what one indexer returned, so the editions that
+            # match the requested author lead and the rest stay reachable below.
            results.sort(
                key=lambda r: (
                    _release_indexer_rank(r, indexer_priority),
+                    affinity_by_source_id.get(r.source_id, AUTHOR_UNKNOWN),
                    0 if r.source_id in enriched_source_ids else 1,
                )
            )
@@ -14,6 +14,20 @@ if TYPE_CHECKING:

 _INTEGER_LIKE_PATTERN = re.compile(r"^[+-]?\d+$")
 _FLOAT_LIKE_PATTERN = re.compile(r"^[+-]?(?:\d+(?:\.\d*)?|\.\d+)$")
+_AUTHOR_TOKEN_PATTERN = re.compile(r"\w+", re.UNICODE)
+_AUTHOR_NOISE_TOKENS = frozenset(
+    {"jr", "sr", "ii", "iii", "iv", "phd", "md", "dr", "mr", "mrs", "ms", "et", "al", "and", "the"}
+)
+
+# Ordering tiers for author agreement between the requested book and what an
+# indexer reported. Lower sorts first.
+AUTHOR_MATCH = 0
+AUTHOR_UNKNOWN = 1
+AUTHOR_MISMATCH = 2
+
+# A mononym ("Homer") can only ever agree on one token; a longer name needs a
+# given name and a surname to agree before it counts as the same person.
+_AUTHOR_TOKENS_REQUIRED = 2


 def coerce_int_like(value: object) -> int | None:
@@ -32,6 +46,49 @@ def coerce_int_like(value: object) -> int | None:
    return int(normalized)


+def _author_tokens(value: object) -> list[str]:
+    """Split an author string into comparable lowercase name tokens."""
+    if not isinstance(value, str):
+        return []
+    tokens = [token.lower() for token in _AUTHOR_TOKEN_PATTERN.findall(value)]
+    return [token for token in tokens if token not in _AUTHOR_NOISE_TOKENS]
+
+
+def _author_tokens_compatible(wanted: str, offered: str) -> bool:
+    """Treat an abbreviated given name as the name it abbreviates."""
+    return wanted == offered or wanted.startswith(offered) or offered.startswith(wanted)
+
+
+def author_affinity(wanted: object, offered: object) -> int:
+    """Rank how far an indexer's author field is from the requested author.
+
+    Shelfmark ranks on this rather than filtering on it, so a wrong verdict only
+    costs a release its position in the list, never its visibility. That is what
+    makes the loose token comparison safe: "Tim"/"Timothy" and "T."/"Timothy"
+    agree, while a transliteration ("Dostoevsky"/"Dostoyevsky") is merely sorted
+    last instead of being hidden.
+
+    Three-way on purpose: an indexer that reports no author at all must not sort
+    below one that reports a wrong author, so "no metadata" ranks between
+    agreement and disagreement rather than counting as either.
+    """
+    wanted_tokens = _author_tokens(wanted)
+    offered_tokens = _author_tokens(offered)
+    if not wanted_tokens or not offered_tokens:
+        return AUTHOR_UNKNOWN
+
+    matched = sum(
+        1
+        for wanted_token in wanted_tokens
+        if any(
+            _author_tokens_compatible(wanted_token, offered_token)
+            for offered_token in offered_tokens
+        )
+    )
+    required = min(_AUTHOR_TOKENS_REQUIRED, len(wanted_tokens))
+    return AUTHOR_MATCH if matched >= required else AUTHOR_MISMATCH
+
+
 def build_source_id(result: dict) -> str:
    """Build the Release.source_id for a raw Prowlarr result.

@@ -35,6 +35,15 @@
    "typescript/no-misused-promises": "error",
    "typescript/no-non-null-assertion": "error",
    "typescript/only-throw-error": "error",
+    // React Compiler advisories, enforced everywhere with no per-file exemptions.
+    // The violations inherited from the oxlint 1.70 -> 1.80 bump are all resolved:
+    // three by widening a dependency to the object the compiler infers, and seven
+    // by an `oxlint-disable-next-line` that says, at the callsite, why the flagged
+    // dependency is load-bearing - five are re-run triggers that are never read,
+    // and two are values the callback genuinely uses.
+    "react/preserve-manual-memoization": "error",
+    "react/exhaustive-effect-dependencies": "error",
+    "react/memo-dependencies": "error",
    "react/no-danger": "error",
    "react/no-clone-element": "error",
    "react/no-react-children": "error",
@@ -24,17 +24,17 @@
    "socket.io-client": "^4.7.5"
  },
  "devDependencies": {
-    "@types/node": "^26.2.0",
+    "@types/node": "^26.3.0",
    "@types/react": "^19.2.18",
-    "@types/react-dom": "^19.2.4",
-    "@vitejs/plugin-react": "^6.0.5",
+    "@types/react-dom": "^19.2.5",
+    "@vitejs/plugin-react": "^6.1.0",
    "knip": "^6.32.2",
-    "oxfmt": "^0.63.0",
-    "oxlint": "^1.78.0",
+    "oxfmt": "^0.65.0",
+    "oxlint": "^1.80.0",
    "oxlint-tsgolint": "^7.0.2001",
    "tailwindcss": "^4.2.2",
    "typescript": "^7.0.2",
-    "vite": "^8.2.1",
-    "vitest": "^4.1.10"
+    "vite": "^8.2.2",
+    "vitest": "^4.1.11"
  }
 }
@@ -32,6 +32,7 @@ import {
 import { useActivity } from './hooks/useActivity';
 import { useAuth } from './hooks/useAuth';
 import { useDownloadTracking } from './hooks/useDownloadTracking';
+import { useLatestCallback } from './hooks/useLatestCallback';
 import { useMediaQuery } from './hooks/useMediaQuery';
 import { useMountEffect } from './hooks/useMountEffect';
 import { useRealtimeStatus } from './hooks/useRealtimeStatus';
@@ -58,7 +59,6 @@ import {
  isApiResponseError,
  updateSelfUser,
  setBookTargetState,
-  type DownloadReleasePayload,
 } from './services/api';
 import type {
  Book,
@@ -74,6 +74,7 @@ import type {
  ActingAsUserSelection,
  MetadataProviderSummary,
  MetadataSearchConfig,
+  MetadataSearchField,
  QueuedDownloadResult,
  QueryTargetOption,
  SearchMode,
@@ -87,11 +88,13 @@ import { bookSupportsTargets } from './utils/bookTargetLoader';
 import { buildSearchQuery } from './utils/buildSearchQuery';
 import { wasDownloadQueuedAfterResponseError } from './utils/downloadRecovery';
 import { getDynamicOptionGroup } from './utils/dynamicFieldOptions';
+import { resolveDefaultLanguageCodes } from './utils/languageFilters';
 import { getConfiguredMetadataProviderForContentType } from './utils/metadataProviders';
 import { getEffectiveMetadataSort } from './utils/metadataSort';
 import { isRecord } from './utils/objectHelpers';
 import { policyTrace } from './utils/policyTrace';
 import { buildQueryTargets, getDefaultQueryTargetKey } from './utils/queryTargets';
+import { buildReleaseDownloadPayload, type ReleaseDownloadOptions } from './utils/releasePayload';
 import { applyRequestNoteToPayload } from './utils/requestConfirmation';
 import { bookFromRequestData } from './utils/requestFulfil';
 import {
@@ -218,6 +221,7 @@ type PendingOnBehalfDownload =
      release: Release;
      releaseContentType: ContentType;
      actingAsUser: ActingAsUserSelection;
+      options?: ReleaseDownloadOptions;
    }
  | {
      type: 'combined';
@@ -490,8 +494,6 @@ function App() {
  });

  // When a book is removed from the Hardcover list currently being browsed, remove it from results
-  const searchFieldValuesRef = useRef(searchFieldValues);
-  searchFieldValuesRef.current = searchFieldValues;
  useBookTargetDeselectSync({
    activeListValue: searchFieldValues.hardcover_list,
    setBooks,
@@ -603,24 +605,6 @@ function App() {
    };
  }, [effectiveActingAsUser, pendingOnBehalfDownload]);

-  // Wire up logout callback to clear search state
-  const handleLogoutWithCleanup = useCallback(async () => {
-    await handleLogout();
-    resetSearchResultsState();
-    setActiveQueryTarget('general');
-    setPendingRequestPayload(null);
-    setPendingRequestExtraPayloads([]);
-    setActingAsUser(null);
-    setAdminUsers([]);
-    setAdminUsersError(null);
-    setHasLoadedAdminUsers(false);
-    setPendingOnBehalfDownload(null);
-    setFulfillingRequest(null);
-    resetActivity();
-    setSettingsOpen(false);
-    setSelfSettingsOpen(false);
-  }, [handleLogout, resetActivity, resetSearchResultsState]);
-
  // Combined mode state (ebook + audiobook in one transaction)
  const [combinedState, setCombinedState] = useState<CombinedSelectionState | null>(null);

@@ -653,20 +637,6 @@ function App() {
    setDownloadsSidebarOpen(true);
    prefetchActivityHistory();
  }, [downloadsSidebarOpen, prefetchActivityHistory]);
-  const handleSettingsClick = useCallback(() => {
-    if (config?.settings_enabled) {
-      if (authIsAdmin) {
-        void primeUsersCache();
-        void primeSettingsCache();
-        setSettingsOpen(true);
-      } else {
-        setSelfSettingsOpen(true);
-      }
-      return;
-    }
-    setConfigBannerOpen(true);
-  }, [authIsAdmin, config?.settings_enabled]);
-
  const headerRef = useCallback((el: HTMLDivElement | null) => {
    if (headerObserverRef.current) {
      headerObserverRef.current.disconnect();
@@ -683,6 +653,39 @@ function App() {
  const [settingsOpen, setSettingsOpen] = useState(false);
  const [selfSettingsOpen, setSelfSettingsOpen] = useState(false);
  const [configBannerOpen, setConfigBannerOpen] = useState(false);
+
+  // Wire up logout callback to clear search state
+  const handleLogoutWithCleanup = useCallback(async () => {
+    await handleLogout();
+    resetSearchResultsState();
+    setActiveQueryTarget('general');
+    setPendingRequestPayload(null);
+    setPendingRequestExtraPayloads([]);
+    setActingAsUser(null);
+    setAdminUsers([]);
+    setAdminUsersError(null);
+    setHasLoadedAdminUsers(false);
+    setPendingOnBehalfDownload(null);
+    setFulfillingRequest(null);
+    resetActivity();
+    setSettingsOpen(false);
+    setSelfSettingsOpen(false);
+  }, [handleLogout, resetActivity, resetSearchResultsState]);
+
+  const handleSettingsClick = useCallback(() => {
+    if (config?.settings_enabled) {
+      if (authIsAdmin) {
+        void primeUsersCache();
+        void primeSettingsCache();
+        setSettingsOpen(true);
+      } else {
+        setSelfSettingsOpen(true);
+      }
+      return;
+    }
+    setConfigBannerOpen(true);
+  }, [authIsAdmin, config?.settings_enabled]);
+
  const [onboardingOpen, setOnboardingOpen] = useState(false);
  useShowOnboardingDebug({
    setOnboardingOpen,
@@ -1071,85 +1074,45 @@ function App() {
    [],
  );

-  const buildReleaseDownloadPayload = useCallback(
-    (book: Book, release: Release, releaseContentType: ContentType): DownloadReleasePayload => {
-      const isManual = book.provider === 'manual';
-      const releasePreview =
-        typeof release.extra?.preview === 'string' ? release.extra.preview : undefined;
-      const releaseAuthor =
-        typeof release.extra?.author === 'string' ? release.extra.author : undefined;
-
-      return {
-        source: release.source,
-        source_id: release.source_id,
-        title: isManual ? release.title : book.title,
-        author: isManual ? releaseAuthor || '' : book.author,
-        year: book.year,
-        format: release.format,
-        size: release.size,
-        size_bytes: release.size_bytes,
-        download_url: release.download_url,
-        protocol: release.protocol,
-        indexer: release.indexer,
-        seeders: release.seeders,
-        extra: release.extra,
-        preview: isManual ? releasePreview || undefined : book.preview,
-        content_type: releaseContentType,
-        series_name: book.series_name,
-        series_position: book.series_position,
-        subtitle: book.subtitle,
-        // From the release, never the book: book.language is the provider's
-        // canonical edition, which would mislabel a translated release.
-        language: release.language ?? undefined,
-      };
-    },
-    [],
-  );
-
  // When downloading a book while browsing a Hardcover list the user owns,
  // automatically remove it from that list (fire-and-forget).
-  const searchFieldLabelsRef = useRef(searchFieldLabels);
-  searchFieldLabelsRef.current = searchFieldLabels;
-  const metadataConfigRef = useRef(activeMetadataConfig);
-  metadataConfigRef.current = activeMetadataConfig;
+  // Stable identity for the download handlers below, while still reading the current
+  // search field values, labels and metadata config. Not an Effect Event: the callers
+  // are download handlers, not Effects. See useLatestCallback.
+  const removeBookFromActiveList = useLatestCallback((book: Book) => {
+    if (config?.hardcover_auto_remove_on_download === false) return;
+    if (!bookSupportsTargets(book)) return;
+    const activeList = searchFieldValues.hardcover_list;
+    if (!activeList) return;
+    const target = String(activeList);
+    const provider = book.provider;
+    const bookId = book.provider_id;
+    if (!provider || !bookId) return;

-  const removeBookFromActiveList = useCallback(
-    (book: Book) => {
-      if (config?.hardcover_auto_remove_on_download === false) return;
-      if (!bookSupportsTargets(book)) return;
-      const activeList = searchFieldValuesRef.current.hardcover_list;
-      if (!activeList) return;
-      const target = String(activeList);
-      const provider = book.provider;
-      const bookId = book.provider_id;
-      if (!provider || !bookId) return;
+    // Only auto-remove from lists the user owns (Reading Status / My Lists)
+    const listField = activeMetadataConfig?.search_fields.find(
+      (f) => f.key === 'hardcover_list' && f.type === 'DynamicSelectSearchField',
+    );
+    if (listField && listField.type === 'DynamicSelectSearchField') {
+      const group = getDynamicOptionGroup(listField.options_endpoint, target);
+      if (group && group !== 'Reading Status' && group !== 'My Lists') return;
+    }

-      // Only auto-remove from lists the user owns (Reading Status / My Lists)
-      const listField = metadataConfigRef.current?.search_fields.find(
-        (f) => f.key === 'hardcover_list' && f.type === 'DynamicSelectSearchField',
-      );
-      if (listField && listField.type === 'DynamicSelectSearchField') {
-        const group = getDynamicOptionGroup(listField.options_endpoint, target);
-        if (group && group !== 'Reading Status' && group !== 'My Lists') return;
-      }
-
-      void setBookTargetState(provider, bookId, target, false)
-        .then((result) => {
-          if (result.changed) {
-            emitBookTargetChange({
-              provider,
-              bookId,
-              target,
-              selected: false,
-            });
-            const listName = searchFieldLabelsRef.current['hardcover_list'];
-            showToast(`Removed from ${listName || 'list'}`, 'info');
-          }
-        })
-        .catch(() => undefined);
-    },
-    [config?.hardcover_auto_remove_on_download, showToast],
-  );
+    void setBookTargetState(provider, bookId, target, false)
+      .then((result) => {
+        if (result.changed) {
+          emitBookTargetChange({
+            provider,
+            bookId,
+            target,
+            selected: false,
+          });
+          const listName = searchFieldLabels['hardcover_list'];
+          showToast(`Removed from ${listName || 'list'}`, 'info');
+        }
+      })
+      .catch(() => undefined);
+  });

  const executeBookDownload = useCallback(
    async (book: Book, onBehalfOfUserId?: number): Promise<void> => {
@@ -1213,12 +1176,13 @@ function App() {
      release: Release,
      releaseContentType: ContentType,
      onBehalfOfUserId?: number,
+      options?: ReleaseDownloadOptions,
    ): Promise<void> => {
      const requestStartedAtSeconds = Date.now() / 1000;
      try {
        trackRelease(book.id, release.source_id);
        await downloadRelease(
-          buildReleaseDownloadPayload(book, release, releaseContentType),
+          buildReleaseDownloadPayload(book, release, releaseContentType, options),
          onBehalfOfUserId,
        );
        await fetchStatus();
@@ -1300,7 +1264,6 @@ function App() {
      }
    },
    [
-      buildReleaseDownloadPayload,
      fetchStatus,
      openRequestConfirmation,
      refreshRequestPolicy,
@@ -1415,6 +1378,7 @@ function App() {
          effectivePendingOnBehalfDownload.release,
          effectivePendingOnBehalfDownload.releaseContentType,
          onBehalfOfUserId,
+          effectivePendingOnBehalfDownload.options,
        );
      }
      setPendingOnBehalfDownload(null);
@@ -1639,6 +1603,7 @@ function App() {
    book: Book,
    release: Release,
    releaseContentType: ContentType,
+    options?: ReleaseDownloadOptions,
  ) => {
    policyTrace('release.action:start', {
      bookId: book.id,
@@ -1654,11 +1619,12 @@ function App() {
        release,
        releaseContentType,
        actingAsUser: effectiveActingAsUser,
+        options,
      });
      return;
    }

-    await executeReleaseDownload(book, release, releaseContentType);
+    await executeReleaseDownload(book, release, releaseContentType, undefined, options);
  };

  const handleReleaseRequest = useCallback(
@@ -1927,10 +1893,7 @@ function App() {
  );
  const supportedFormats = config?.supported_formats || DEFAULT_SUPPORTED_FORMATS;
  const defaultLanguageCodes = useMemo(
-    () =>
-      config?.default_language && config.default_language.length > 0
-        ? config.default_language
-        : [bookLanguages[0]?.code || 'en'],
+    () => resolveDefaultLanguageCodes(config?.default_language, bookLanguages),
    [config?.default_language, bookLanguages],
  );

@@ -1942,14 +1905,27 @@ function App() {
    effectiveSearchMode === 'universal' &&
    (universalDefaultMode === 'download' || universalDefaultMode === 'request_release');

+  // Keep the last known search fields so queryTargets doesn't collapse to
+  // [general] while the metadata config briefly reloads on content type switch.
+  // Held in state rather than a ref written during render: a ref read back in the same
+  // pass is what `react/refs` forbids, and this is the adjust-state-during-render shape
+  // React documents for exactly this - carry the previous value until a new one arrives.
+  const [stableSearchFields, setStableSearchFields] = useState<MetadataSearchField[]>(
+    () => activeMetadataConfig?.search_fields ?? [],
+  );
+  const incomingSearchFields = activeMetadataConfig?.search_fields;
+  if (incomingSearchFields && incomingSearchFields !== stableSearchFields) {
+    setStableSearchFields(incomingSearchFields);
+  }
+
  const queryTargets = useMemo<QueryTargetOption[]>(
    () =>
      buildQueryTargets({
        searchMode: effectiveSearchMode,
-        metadataSearchFields: activeMetadataConfig?.search_fields ?? [],
+        metadataSearchFields: stableSearchFields,
        manualSearchAllowed,
      }),
-    [effectiveSearchMode, activeMetadataConfig?.search_fields, manualSearchAllowed],
+    [effectiveSearchMode, stableSearchFields, manualSearchAllowed],
  );
  const effectiveActiveQueryTarget = useMemo(() => {
    if (queryTargets.some((target) => target.key === activeQueryTarget)) {
@@ -1978,27 +1954,27 @@ function App() {
        ? (queryTargets.find((target) => target.field?.key === seriesBrowseCapability.field_key) ??
          null)
        : null,
-    [queryTargets, seriesBrowseCapability?.field_key],
+    // `seriesBrowseCapability` whole: the body reads `.field_key` off it unguarded
+    // inside the ternary, so that object is the dependency the compiler infers.
+    [queryTargets, seriesBrowseCapability],
  );

  const activeQueryValue = useMemo(() => {
    if (
      !activeQueryOption ||
      activeQueryOption.source === 'general' ||
-      activeQueryOption.source === 'manual'
+      activeQueryOption.source === 'manual' ||
+      activeQueryOption.source === 'direct-field'
    ) {
      return searchInput;
    }

-    if (activeQueryOption.source === 'direct-field') {
-      if (activeQueryOption.key === 'isbn') return advancedFilters.isbn;
-      if (activeQueryOption.key === 'author') return advancedFilters.author;
-      if (activeQueryOption.key === 'title') return advancedFilters.title;
+    if (!activeQueryOption.field) {
      return '';
    }

-    if (!activeQueryOption.field) {
-      return '';
+    if (activeQueryOption.field.type === 'TextSearchField') {
+      return searchInput;
    }

    if (activeQueryOption.field.type === 'CheckboxSearchField') {
@@ -2008,7 +1984,7 @@ function App() {
    }

    return searchFieldValues[activeQueryOption.field.key] ?? '';
-  }, [activeQueryOption, searchInput, advancedFilters, searchFieldValues]);
+  }, [activeQueryOption, searchInput, searchFieldValues]);

  const activeQueryValueLabel = useMemo(() => {
    if (!activeQueryOption?.field) {
@@ -2056,29 +2032,25 @@ function App() {
      if (
        !activeQueryOption ||
        activeQueryOption.source === 'general' ||
-        activeQueryOption.source === 'manual'
+        activeQueryOption.source === 'manual' ||
+        activeQueryOption.source === 'direct-field'
      ) {
        setSearchInput(typeof value === 'string' ? value : String(value ?? ''));
        return;
      }

-      if (activeQueryOption.source === 'direct-field') {
-        const nextValue = typeof value === 'string' ? value : String(value ?? '');
-        if (activeQueryOption.key === 'isbn') {
-          updateAdvancedFilters({ isbn: nextValue });
-        } else if (activeQueryOption.key === 'author') {
-          updateAdvancedFilters({ author: nextValue });
-        } else if (activeQueryOption.key === 'title') {
-          updateAdvancedFilters({ title: nextValue });
-        }
-        return;
-      }
-
      if (activeQueryOption.field) {
+        if (activeQueryOption.field.type === 'TextSearchField') {
+          setSearchInput(typeof value === 'string' ? value : String(value ?? ''));
+          if (label !== undefined) {
+            updateSearchFieldValue(activeQueryOption.field.key, value, label);
+          }
+          return;
+        }
        updateSearchFieldValue(activeQueryOption.field.key, value, label);
      }
    },
-    [activeQueryOption, setSearchInput, updateAdvancedFilters, updateSearchFieldValue],
+    [activeQueryOption, setSearchInput, updateSearchFieldValue],
  );

  const handleSearchModeChange = useCallback(
@@ -2276,7 +2248,9 @@ function App() {

      return book.provider === activeMetadataConfig.provider;
    },
-    [activeMetadataConfig?.provider, seriesBrowseCapability?.sort, seriesBrowseTarget?.field],
+    // `activeMetadataConfig` whole: the body reads `.provider` off it unguarded on
+    // the last line, so that object is the dependency the compiler infers.
+    [activeMetadataConfig, seriesBrowseCapability?.sort, seriesBrowseTarget?.field],
  );

  const handleManualSearch = useCallback(() => {
@@ -0,0 +1,193 @@
+import { useState } from 'react';
+
+import type { PackBook, PackPlan, Release } from '../types';
+import {
+  describePackPlan,
+  parseSeriesPositionInput,
+  toBookPlanPayload,
+  updateReviewBook,
+} from '../utils/packReview';
+import { ToggleSwitch } from './shared/ToggleSwitch';
+
+interface PackReviewPanelProps {
+  release: Release;
+  plan: PackPlan;
+  books: PackBook[];
+  onChange: (books: PackBook[]) => void;
+  onBack: () => void;
+  /** `null` means "treat the whole release as one book". */
+  onConfirm: (books: PackBook[] | null) => Promise<void>;
+  isSubmitting: boolean;
+}
+
+const inputClassName =
+  'w-full rounded-md border border-(--border-muted) bg-(--bg) px-2 py-1 text-sm text-(--text) focus:border-emerald-500 focus:outline-none';
+
+export const PackReviewPanel = ({
+  release,
+  plan,
+  books,
+  onChange,
+  onBack,
+  onConfirm,
+  isSubmitting,
+}: PackReviewPanelProps) => {
+  const [singleBook, setSingleBook] = useState(false);
+  const [expandedFiles, setExpandedFiles] = useState<number | null>(null);
+  const [showIgnored, setShowIgnored] = useState(false);
+
+  const payloadBooks = toBookPlanPayload(books);
+  const canConfirm = !isSubmitting && (singleBook || payloadBooks.length > 0);
+  const confirmLabel = singleBook
+    ? 'Download as one book'
+    : `Download ${payloadBooks.length} ${payloadBooks.length === 1 ? 'book' : 'books'}`;
+
+  return (
+    <div className="flex flex-col gap-4 px-5 py-4" data-testid="pack-review-panel">
+      <div>
+        <h3 className="text-base font-semibold text-(--text)">
+          This release contains several books
+        </h3>
+        <p className="mt-1 text-sm text-zinc-500 dark:text-zinc-400">
+          <span className="font-medium text-(--text)">{release.title}</span> ·{' '}
+          {describePackPlan(books, plan.ignored)}
+        </p>
+        <p className="mt-1 text-xs text-zinc-500 dark:text-zinc-400">
+          Each book below is filed separately with its own title. Fix any titles before downloading
+          — the author and series come from the book you searched.
+        </p>
+      </div>
+
+      <div className="flex items-center justify-between rounded-lg border border-(--border-muted) px-3 py-2">
+        <div>
+          <p className="text-sm font-medium text-(--text)">Treat as a single book</p>
+          <p className="text-xs text-zinc-500 dark:text-zinc-400">
+            Use this if the split is wrong and the files are really one audiobook.
+          </p>
+        </div>
+        <ToggleSwitch
+          checked={singleBook}
+          onChange={setSingleBook}
+          color="emerald"
+          ariaLabel="Treat as a single book"
+          disabled={isSubmitting}
+        />
+      </div>
+
+      <div
+        className={`flex flex-col divide-y divide-zinc-200/60 dark:divide-zinc-800/60 ${
+          singleBook ? 'pointer-events-none opacity-40' : ''
+        }`}
+      >
+        <div className="grid grid-cols-[minmax(0,1fr)_72px_72px_80px] gap-2 pb-1 text-xs font-medium tracking-wide text-zinc-500 uppercase dark:text-zinc-400">
+          <span>Title</span>
+          <span>Series #</span>
+          <span>Year</span>
+          <span className="text-right">Files</span>
+        </div>
+        {books.map((book, index) => (
+          <div key={book.files[0] ?? index} className="py-2">
+            <div className="grid grid-cols-[minmax(0,1fr)_72px_72px_80px] items-center gap-2">
+              <input
+                type="text"
+                value={book.title}
+                onChange={(e) =>
+                  onChange(updateReviewBook(books, index, { title: e.target.value }))
+                }
+                aria-label={`Title for book ${index + 1}`}
+                className={inputClassName}
+                disabled={isSubmitting}
+              />
+              <input
+                type="text"
+                inputMode="decimal"
+                value={book.series_position ?? ''}
+                onChange={(e) =>
+                  onChange(
+                    updateReviewBook(books, index, {
+                      series_position: parseSeriesPositionInput(e.target.value),
+                    }),
+                  )
+                }
+                aria-label={`Series position for book ${index + 1}`}
+                className={inputClassName}
+                disabled={isSubmitting}
+              />
+              <input
+                type="text"
+                inputMode="numeric"
+                value={book.year ?? ''}
+                onChange={(e) => {
+                  const parsed = parseSeriesPositionInput(e.target.value);
+                  onChange(
+                    updateReviewBook(books, index, {
+                      year: parsed === null ? null : Math.trunc(parsed),
+                    }),
+                  );
+                }}
+                aria-label={`Year for book ${index + 1}`}
+                className={inputClassName}
+                disabled={isSubmitting}
+              />
+              <button
+                type="button"
+                onClick={() => setExpandedFiles(expandedFiles === index ? null : index)}
+                className="hover-surface rounded-md px-2 py-1 text-right text-sm text-zinc-500 transition-colors dark:text-zinc-400"
+                aria-expanded={expandedFiles === index}
+              >
+                {book.files.length} {book.files.length === 1 ? 'file' : 'files'}
+              </button>
+            </div>
+            {expandedFiles === index && (
+              <ul className="mt-2 max-h-40 overflow-y-auto rounded-md bg-(--bg-soft) px-3 py-2 font-mono text-xs break-all text-zinc-600 dark:text-zinc-300">
+                {book.files.map((file) => (
+                  <li key={file}>{file}</li>
+                ))}
+              </ul>
+            )}
+          </div>
+        ))}
+      </div>
+
+      {plan.ignored.length > 0 && (
+        <div>
+          <button
+            type="button"
+            onClick={() => setShowIgnored(!showIgnored)}
+            className="text-xs text-zinc-500 underline-offset-2 hover:underline dark:text-zinc-400"
+            aria-expanded={showIgnored}
+          >
+            {plan.ignored.length} {plan.ignored.length === 1 ? 'file' : 'files'} ignored (not a book
+            format)
+          </button>
+          {showIgnored && (
+            <ul className="mt-2 max-h-32 overflow-y-auto rounded-md bg-(--bg-soft) px-3 py-2 font-mono text-xs break-all text-zinc-600 dark:text-zinc-300">
+              {plan.ignored.map((file) => (
+                <li key={file}>{file}</li>
+              ))}
+            </ul>
+          )}
+        </div>
+      )}
+
+      <div className="flex items-center justify-end gap-3 border-t border-(--border-muted) pt-4">
+        <button
+          type="button"
+          onClick={onBack}
+          disabled={isSubmitting}
+          className="hover-surface rounded-lg px-3 py-1.5 text-sm font-medium text-(--text) transition-colors disabled:opacity-50"
+        >
+          &larr; Back
+        </button>
+        <button
+          type="button"
+          onClick={() => void onConfirm(singleBook ? null : payloadBooks)}
+          disabled={!canConfirm}
+          className="rounded-lg bg-emerald-600 px-4 py-1.5 text-sm font-medium text-white transition-colors hover:bg-emerald-700 disabled:cursor-not-allowed disabled:opacity-50"
+        >
+          {isSubmitting ? 'Queuing…' : confirmLabel}
+        </button>
+      </div>
+    </div>
+  );
+};
@@ -8,6 +8,7 @@ import {
  toStringArray,
  toStringValue,
 } from '../utils/objectHelpers';
+import { getUnrecognizedReleaseFormats } from '../utils/releaseFormats';
 import { Tooltip } from './shared/Tooltip';

 interface ReleaseCellProps {
@@ -424,6 +425,38 @@ export const ReleaseCell = ({
      const primaryFormat = formats?.[0] || null;
      const additionalFormats = formats?.slice(1) || [];

+      // The indexer named a format Shelfmark can't process (e.g. MAM "[ENG / AVI]").
+      // Downloading it would only fail post-processing, so warn instead of showing the
+      // bare content-type icon that makes it look like any other result.
+      const unrecognizedFormats = primaryFormat ? [] : getUnrecognizedReleaseFormats(release);
+      if (unrecognizedFormats.length > 0) {
+        const unsupportedLabel = unrecognizedFormats.map((fmt) => fmt.toUpperCase()).join(', ');
+        const unsupportedTitle = `Unsupported format (${unsupportedLabel}) - Shelfmark cannot process this release`;
+        if (compact) {
+          return (
+            <span
+              className="font-semibold text-amber-600 dark:text-amber-400"
+              title={unsupportedTitle}
+            >
+              {unrecognizedFormats[0].toUpperCase()}
+              {unrecognizedFormats.length > 1 && ` +${unrecognizedFormats.length - 1}`}
+            </span>
+          );
+        }
+        return (
+          <div className="flex items-center justify-start" title={unsupportedTitle}>
+            <span className="inline-flex items-center gap-1">
+              <span className="w-13 rounded-lg bg-amber-500/20 py-0.5 text-center text-[10px] font-semibold tracking-wide whitespace-nowrap text-amber-700 sm:text-[11px] dark:text-amber-400">
+                {unrecognizedFormats[0].toUpperCase()}
+              </span>
+              <span className="text-[10px] font-medium whitespace-nowrap text-amber-700 sm:text-[11px] dark:text-amber-400">
+                Unsupported
+              </span>
+            </span>
+          </div>
+        );
+      }
+
      // Use blue for book, violet for audiobook when no format specified
      const noFormatStyle = isAudiobook
        ? { bg: 'bg-violet-500/20', text: 'text-violet-600 dark:text-violet-400' }
@@ -7,6 +7,7 @@ import { useReleaseSearchSession } from '../hooks/releaseModal/useReleaseSearchS
 import { useTabIndicator } from '../hooks/ui/useTabIndicator';
 import { useBodyScrollLock } from '../hooks/useBodyScrollLock';
 import { useEscapeKey } from '../hooks/useEscapeKey';
+import { inspectRelease } from '../services/api';
 import type {
  Book,
  Release,
@@ -18,6 +19,8 @@ import type {
  LeadingCellConfig,
  ContentType,
  RequestPolicyMode,
+  PackBook,
+  PackPlan,
 } from '../types';
 import { isMetadataBook } from '../types';
 import { bookSupportsTargets } from '../utils/bookTargetLoader';
@@ -29,7 +32,10 @@ import {
  buildLanguageNormalizer,
 } from '../utils/languageFilters';
 import { getNestedValue, toComparableText, toStringValue } from '../utils/objectHelpers';
+import { toBookPlanPayload } from '../utils/packReview';
 import { getReleaseFormats } from '../utils/releaseFormats';
+import { INITIAL_ENTER_ANIMATION, nextEnterAnimation } from '../utils/releaseModalEnterAnimation';
+import { buildReleaseDownloadPayload, type ReleaseDownloadOptions } from '../utils/releasePayload';
 import {
  getBookTitleCandidates,
  getBookAuthorCandidates,
@@ -50,6 +56,7 @@ import { BookTargetDropdown } from './BookTargetDropdown';
 import { Dropdown } from './Dropdown';
 import { DropdownList } from './DropdownList';
 import { LanguageMultiSelect } from './LanguageMultiSelect';
+import { PackReviewPanel } from './PackReviewPanel';
 import { ReleaseCell } from './ReleaseCell';

 // Combined mode configuration for the ReleaseModal
@@ -140,7 +147,12 @@ const DEFAULT_COLUMN_CONFIG: ReleaseColumnConfig = {
 interface ReleaseModalProps {
  book: Book | null;
  onClose: () => void;
-  onDownload: (book: Book, release: Release, contentType: ContentType) => Promise<void>;
+  onDownload: (
+    book: Book,
+    release: Release,
+    contentType: ContentType,
+    options?: ReleaseDownloadOptions,
+  ) => Promise<void>;
  onRequestRelease?: (book: Book, release: Release, contentType: ContentType) => Promise<void>;
  onRequestBook?: (book: Book, contentType: ContentType) => Promise<void>;
  getPolicyModeForSource?: (source: string, contentType: ContentType) => RequestPolicyMode;
@@ -762,6 +774,15 @@ const ReleaseModalSession = ({
      : supportedFormats;
  const [isRequestingBook, setIsRequestingBook] = useState(false);
  const [selectedRelease, setSelectedRelease] = useState<Release | null>(null);
+  // Multi-book packs: `multiBook` is the manual header toggle (heuristic split for
+  // releases we can't inspect); `packReview` holds an inspected pack awaiting approval.
+  const [multiBook, setMultiBook] = useState(false);
+  const [packReview, setPackReview] = useState<{
+    release: Release;
+    plan: PackPlan;
+    books: PackBook[];
+  } | null>(null);
+  const [packSubmitting, setPackSubmitting] = useState(false);
  const isCombinedMode = combinedMode != null;
  const combinedPhase = combinedMode?.phase ?? null;
  const combinedStepLabel = combinedMode?.stepLabel ?? '';
@@ -869,6 +890,11 @@ const ReleaseModalSession = ({
    } finally {
      setIsRequestingBook(false);
    }
+    // Kept against the advisory: the body really does read both. `handleClose` is aliased
+    // from the `onClose` prop, which is why the compiler names the source instead, and
+    // dropping `contentType` would let this close over a stale one and request the wrong
+    // format. Correctness first; the cost is an extra callback identity.
+    // oxlint-disable-next-line react/memo-dependencies
  }, [book, onRequestBook, isRequestingBook, contentType, handleClose]);

  const tabRefs = useRef<Record<string, HTMLButtonElement | null>>({});
@@ -1115,7 +1141,9 @@ const ReleaseModalSession = ({
    const narratorField = book.display_fields.find((f) => f.icon === 'microphone');

    return { starField, ratingsField, usersField, pagesField, lengthField, narratorField };
-  }, [book?.display_fields]);
+    // `book`, not `book?.display_fields`: the body reads `book.display_fields`
+    // unguarded after the early return, which is the dependency the compiler infers.
+  }, [book]);

  const getReleaseActionMode = useCallback(
    (release: Release): RequestPolicyMode => {
@@ -1196,7 +1224,36 @@ const ReleaseModalSession = ({

      const mode = getReleaseActionMode(release);
      if (mode === 'download') {
-        await onDownload(book, release, contentType);
+        // Look at the release's files before queueing so a whole-series pack can be
+        // reviewed and filed as separate books instead of one mangled item.
+        let inspected = false;
+        let plan: PackPlan | null = null;
+        let reason: string | null = null;
+        try {
+          const inspection = await inspectRelease(
+            buildReleaseDownloadPayload(book, release, contentType),
+          );
+          inspected = inspection.inspected;
+          plan = inspection.plan;
+          reason = inspection.reason;
+        } catch (error) {
+          console.error('Release inspection failed:', error);
+        }
+        if (inspected && plan?.is_pack) {
+          setPackReview({ release, plan, books: plan.books });
+          return;
+        }
+        // Not a pack (or couldn't be inspected): queue exactly as before. A release we
+        // couldn't inspect might still be an unnoticed pack, so leave a console breadcrumb
+        // rather than interrupting the user; the multi-book toggle forces the split.
+        if (!inspected && !multiBook) {
+          console.warn(
+            `Could not inspect release "${release.title}" before download${
+              reason ? `: ${reason}` : ''
+            }. If it contains several books, enable the multi-book pack toggle.`,
+          );
+        }
+        await onDownload(book, release, contentType, multiBook ? { multiBook: true } : {});
        handleClose();
        return;
      }
@@ -1215,9 +1272,34 @@ const ReleaseModalSession = ({
      onRequestRelease,
      contentType,
      handleClose,
+      multiBook,
    ],
  );

+  const handlePackConfirm = useCallback(
+    async (books: PackBook[] | null): Promise<void> => {
+      if (!book || !packReview) {
+        return;
+      }
+      setPackSubmitting(true);
+      try {
+        await onDownload(
+          book,
+          packReview.release,
+          contentType,
+          books ? { multiBook: true, bookPlan: toBookPlanPayload(books) } : {},
+        );
+        handleClose();
+      } finally {
+        setPackSubmitting(false);
+      }
+    },
+    // Same as handleRequestBook above: the body reads `onDownload`, `contentType` and
+    // `handleClose`, so they stay in the list whatever the advisory infers.
+    // oxlint-disable-next-line react/memo-dependencies
+    [book, packReview, onDownload, contentType, handleClose],
+  );
+
  const titleId = `release-modal-title-${book.id}`;
  const providerDisplay =
    book.provider_display_name ||
@@ -1720,6 +1802,37 @@ const ReleaseModalSession = ({
                  </div>

                  <div className="flex items-center gap-3 pr-1 pl-2">
+                    {/* Multi-book pack toggle (fallback for releases that can't be inspected) */}
+                    {!isCombinedMode && (
+                      <button
+                        type="button"
+                        onClick={() => setMultiBook((prev) => !prev)}
+                        className={`hover-surface relative rounded-full p-2.5 text-zinc-500 transition-colors dark:text-zinc-400 ${
+                          multiBook ? 'text-emerald-600 dark:text-emerald-400' : ''
+                        }`}
+                        aria-label="Multi-book pack"
+                        aria-pressed={multiBook}
+                        title="Multi-book pack: file each subfolder (or each file) as a separate book. Only needed when a release can't be inspected before download."
+                      >
+                        <svg
+                          className="h-4 w-4"
+                          fill="none"
+                          stroke="currentColor"
+                          viewBox="0 0 24 24"
+                          strokeWidth={1.5}
+                        >
+                          <path
+                            strokeLinecap="round"
+                            strokeLinejoin="round"
+                            d="M6.429 9.75 2.25 12l4.179 2.25m0-4.5 5.571 3 5.571-3m-11.142 0L2.25 7.5 12 2.25l9.75 5.25-4.179 2.25m0 0L21.75 12l-4.179 2.25m0 0 4.179 2.25L12 21.75 2.25 16.5l4.179-2.25m11.142 0-5.571 3-5.571-3"
+                          />
+                        </svg>
+                        {multiBook && (
+                          <span className="absolute top-1 right-1 h-2 w-2 rounded-full bg-emerald-500" />
+                        )}
+                      </button>
+                    )}
+
                    {/* Manual query button */}
                    <button
                      type="button"
@@ -2120,6 +2233,19 @@ const ReleaseModalSession = ({
            {/* Release list content */}
            <div className="min-h-[200px]">
              {(() => {
+                if (packReview) {
+                  return (
+                    <PackReviewPanel
+                      release={packReview.release}
+                      plan={packReview.plan}
+                      books={packReview.books}
+                      onChange={(books) => setPackReview({ ...packReview, books })}
+                      onBack={() => setPackReview(null)}
+                      onConfirm={handlePackConfirm}
+                      isSubmitting={packSubmitting}
+                    />
+                  );
+                }
                if (sourcesLoading) {
                  return <ReleaseSkeleton />;
                }
@@ -2335,7 +2461,7 @@ const ReleaseModalSession = ({

 export const ReleaseModal = ({ book, onClose, ...rest }: ReleaseModalProps) => {
  const [isClosing, setIsClosing] = useState(false);
-  const previousSessionKeyRef = useRef<string | null>(null);
+  const [enterAnimation, setEnterAnimation] = useState(INITIAL_ENTER_ANIMATION);

  const handleClose = useCallback(() => {
    setIsClosing(true);
@@ -2359,12 +2485,13 @@ export const ReleaseModal = ({ book, onClose, ...rest }: ReleaseModalProps) => {
      ].join('|')
    : null;

-  const animateEnter =
-    !rest.combinedMode ||
-    previousSessionKeyRef.current === null ||
-    previousSessionKeyRef.current === sessionKey;
-
-  previousSessionKeyRef.current = sessionKey;
+  // Decided once per session key and held for that session's lifetime, so a
+  // re-render mid-session cannot restart the enter animation.
+  const nextAnimation = nextEnterAnimation(enterAnimation, sessionKey, rest.combinedMode != null);
+  if (nextAnimation !== enterAnimation) {
+    setEnterAnimation(nextAnimation);
+  }
+  const animateEnter = nextAnimation.animate;

  if (!book && !isClosing) return null;
  if (!book || !sessionKey) return null;
@@ -3,8 +3,8 @@ import { forwardRef, useImperativeHandle, useMemo, useRef, useState } from 'reac

 import { useSearchMode } from '../contexts/SearchModeContext';
 import { useSearchBarAutocomplete } from '../hooks/searchBar/useSearchBarAutocomplete';
-import { useSearchBarHoverTimeout } from '../hooks/searchBar/useSearchBarHoverTimeout';
 import { useDismiss } from '../hooks/useDismiss';
+import { useLatestCallback } from '../hooks/useLatestCallback';
 import type { DynamicFieldOption } from '../services/api';
 import type { ContentType, MetadataSearchField, QueryTargetOption, SortOption } from '../types';
 import { SearchBarAutocompleteSession } from './SearchBarAutocompleteSession';
@@ -51,6 +51,8 @@ const EMPTY_SORT_OPTIONS: SortOption[] = [];
 const EMPTY_AUTOCOMPLETE_OPTIONS: DynamicFieldOption[] = [];
 const EMPTY_QUERY_TARGETS: QueryTargetOption[] = [];

+const SEARCH_CONTROLS_PANEL_ID = 'search-bar-controls-panel';
+
 const BookIcon = () => (
  <svg
    className="h-5 w-5 shrink-0"
@@ -195,8 +197,9 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
    const { searchMode } = useSearchMode();
    const inputRef = useRef<HTMLInputElement>(null);
    const buttonRef = useRef<HTMLButtonElement>(null);
-    const onSubmitRef = useRef(onSubmit);
-    onSubmitRef.current = onSubmit;
+    // Deferred submits below run from a timeout, not an Effect, so this is a latest-value
+    // callback rather than an Effect Event. See useLatestCallback.
+    const submitLatest = useLatestCallback(() => onSubmit());
    const selectorRef = useRef<HTMLDivElement>(null);
    const hasSearchQuery = hasActiveValue(value);
    const [isSelectorOpen, setIsSelectorOpen] = useState(false);
@@ -205,7 +208,7 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
    const selectTriggerRef = useRef<HTMLButtonElement>(null);
    const selectPanelRef = useRef<HTMLDivElement>(null);
    const autocompletePanelRef = useRef<HTMLDivElement>(null);
-    const { hoverTimeoutRef: selectorHoverTimeout, clearHoverTimeout } = useSearchBarHoverTimeout();
+    const controlsPanelRef = useRef<HTMLDivElement>(null);

    const hasMultipleContentTypes = !allowedContentTypes || allowedContentTypes.length !== 1;
    const showContentTypeSelector =
@@ -221,9 +224,20 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
      () => queryTargets.find((target) => target.key === activeQueryTarget) ?? queryTargets[0],
      [queryTargets, activeQueryTarget],
    );
-    const showActiveTargetLabel = queryTargets.length > 0 && activeTarget.source !== 'general';
+    const showActiveTargetLabel = queryTargets.length > 0 && activeTarget?.source !== 'general';

-    useDismiss(isSelectorOpen, [selectorRef], () => setIsSelectorOpen(false));
+    // Manual search browses release sources directly, one media type at a time — the
+    // combined ("both") flow doesn't apply. Present a plain, switchable Books/Audiobooks
+    // choice for it, even when combined search is forced on for metadata targets.
+    const isManualTarget = activeTarget?.source === 'manual';
+    const combinedSelectionActive = combinedMode && !isManualTarget;
+    const combinedSelectorLocked = combinedModeLocked && !isManualTarget;
+    const combinedToggleAvailable = !!onCombinedModeChange && !isManualTarget;
+    const combinedLineColor = combinedSelectionActive
+      ? 'bg-emerald-500'
+      : 'bg-(--border-muted) group-hover:bg-zinc-400 dark:group-hover:bg-zinc-500';
+
+    useDismiss(isSelectorOpen, [selectorRef, controlsPanelRef], () => setIsSelectorOpen(false));
    useDismiss(isSelectOpen, [selectPanelRef, selectTriggerRef], () => setIsSelectOpen(false));
    useDismiss(isAutocompleteOpen, [autocompletePanelRef, inputRef], () =>
      setIsAutocompleteOpen(false),
@@ -325,18 +339,15 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
    const handleContentTypeSelect = (type: ContentType) => {
      onContentTypeChange?.(type);
      onCombinedModeChange?.(false);
-      setIsSelectorOpen(false);
    };

    const handleCombinedModeSelect = () => {
      if (combinedMode) {
-        // Toggle off — revert to ebook-only
        onCombinedModeChange?.(false);
      } else {
        onContentTypeChange?.('ebook');
        onCombinedModeChange?.(true);
      }
-      setIsSelectorOpen(false);
    };

    const handleQueryTargetSelect = (targetKey: string) => {
@@ -349,14 +360,13 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
      setIsSelectOpen(shouldOpenSelect);
      setIsAutocompleteOpen(false);
      resetAutocomplete();
-      setIsSelectorOpen(false);
    };

    const effectivePlaceholder = getDefaultPlaceholder(
      contentType,
      activeTarget,
      placeholder,
-      combinedMode,
+      combinedSelectionActive,
    );
    const effectiveInputAriaLabel = activeTarget
      ? `${inputAriaLabel}: ${activeTarget.label}`
@@ -404,7 +414,6 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
                setAutocompleteDraftValue(nextValue);
                setIsAutocompleteOpen(nextValue.trim().length >= autocompleteMinQueryLength);
                setIsSelectOpen(false);
-                setIsSelectorOpen(false);
                onChange(nextValue);
                return;
              }
@@ -416,7 +425,6 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
                textInputValue.trim().length >= autocompleteMinQueryLength
              ) {
                setIsAutocompleteOpen(true);
-                setIsSelectorOpen(false);
              }
            }}
            onKeyDown={handleKeyDown}
@@ -482,7 +490,6 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
              onClick={() => {
                if (!disabled && !isDynamicLoading) {
                  setIsSelectOpen((prev) => !prev);
-                  setIsSelectorOpen(false);
                  setIsAutocompleteOpen(false);
                }
              }}
@@ -537,7 +544,7 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(

    let selectorContentTypeLabel = 'audiobooks';
    let selectorIcon = <AudiobookIcon />;
-    if (combinedMode) {
+    if (combinedSelectionActive) {
      selectorContentTypeLabel = 'books and audiobooks';
      selectorIcon = <BothIcon />;
    } else if (contentType === 'ebook') {
@@ -593,25 +600,7 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
          }}
        >
          {showQueryTargetSelector && (
-            <div
-              className="relative flex shrink-0 self-stretch"
-              ref={selectorRef}
-              onPointerEnter={(e) => {
-                if (e.pointerType !== 'mouse') return;
-                clearHoverTimeout();
-                setIsSelectorOpen(true);
-                setIsSelectOpen(false);
-                setIsAutocompleteOpen(false);
-              }}
-              onPointerLeave={(e) => {
-                if (e.pointerType !== 'mouse') return;
-                clearHoverTimeout();
-                selectorHoverTimeout.current = setTimeout(() => {
-                  setIsSelectorOpen(false);
-                  selectorHoverTimeout.current = null;
-                }, 150);
-              }}
-            >
+            <div className="relative flex shrink-0 self-stretch" ref={selectorRef}>
              <button
                type="button"
                onClick={() => {
@@ -619,11 +608,11 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
                  setIsSelectOpen(false);
                  setIsAutocompleteOpen(false);
                }}
-                className="hover-action flex items-center gap-1.5 rounded-l-full pr-2 pl-5 transition-colors"
+                className="hover-action flex cursor-pointer items-center gap-1.5 rounded-l-full pr-2 pl-5 transition-colors"
                style={{ color: 'var(--text)' }}
                aria-label={`Searching ${selectorContentTypeLabel} by ${activeTarget?.label ?? 'general'}. Click to change.`}
                aria-expanded={isSelectorOpen}
-                aria-haspopup="dialog"
+                aria-controls={SEARCH_CONTROLS_PANEL_ID}
              >
                {selectorIcon}
                {showActiveTargetLabel && (
@@ -646,266 +635,10 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
                  />
                </svg>
              </button>
-
              <div
                className="absolute top-1/2 right-0 h-6 w-px -translate-y-1/2"
                style={{ background: 'var(--border-muted)' }}
              />
-
-              {isSelectorOpen && (
-                <div
-                  className="animate-fade-in-down absolute top-full left-0 z-50 mt-2 w-[min(20rem,calc(100vw-2rem))] overflow-hidden rounded-2xl border shadow-2xl"
-                  style={{
-                    background: 'var(--bg)',
-                    borderColor: 'var(--border-muted)',
-                  }}
-                  role="dialog"
-                  aria-label="Search context"
-                >
-                  <div className="max-h-[min(24rem,calc(100vh-8rem))] overflow-y-auto p-3">
-                    {showContentTypeSelector && (
-                      <div
-                        className={`border-b ${onCombinedModeChange ? 'pb-0' : 'pb-3'}`}
-                        style={{ borderColor: 'var(--border-muted)' }}
-                      >
-                        <div className="flex items-center justify-between px-1 pb-2">
-                          <span className="text-xs font-medium tracking-wide uppercase opacity-60">
-                            Content
-                          </span>
-                          {onAdvancedToggle && (
-                            <button
-                              type="button"
-                              onClick={() => {
-                                setIsSelectorOpen(false);
-                                onAdvancedToggle();
-                              }}
-                              className={`-mt-1.5 -mr-1 -mb-0.5 flex items-center gap-1.5 rounded-xl px-4 py-2.5 text-xs font-medium transition-colors ${
-                                isAdvancedActive ? 'bg-emerald-600 text-white' : 'hover-surface'
-                              }`}
-                              style={
-                                isAdvancedActive
-                                  ? { borderColor: 'rgb(16 185 129 / 0.7)' }
-                                  : { color: 'var(--text-muted)' }
-                              }
-                            >
-                              <svg
-                                className="h-3.5 w-3.5"
-                                xmlns="http://www.w3.org/2000/svg"
-                                fill="none"
-                                viewBox="0 0 24 24"
-                                strokeWidth="1.5"
-                                stroke="currentColor"
-                                aria-hidden="true"
-                              >
-                                <path
-                                  strokeLinecap="round"
-                                  strokeLinejoin="round"
-                                  d="M10.5 6h9.75M10.5 6a1.5 1.5 0 1 1-3 0m3 0a1.5 1.5 0 1 0-3 0M3.75 6H7.5m3 12h9.75m-9.75 0a1.5 1.5 0 0 1-3 0m3 0a1.5 1.5 0 0 0-3 0m-3.75 0H7.5m9-6h3.75m-3.75 0a1.5 1.5 0 0 1-3 0m3 0a1.5 1.5 0 0 0-3 0m-9.75 0h9.75"
-                                />
-                              </svg>
-                              Options
-                            </button>
-                          )}
-                        </div>
-                        <div className="grid grid-cols-2 gap-2">
-                          <button
-                            type="button"
-                            onClick={() => handleContentTypeSelect('ebook')}
-                            className={`flex items-center gap-2 rounded-xl border px-3 py-2 text-sm font-medium transition-colors ${
-                              contentType === 'ebook' || combinedMode
-                                ? 'bg-emerald-600 text-white'
-                                : 'hover-surface'
-                            }`}
-                            style={
-                              contentType === 'ebook' || combinedMode
-                                ? { borderColor: 'rgb(16 185 129 / 0.7)' }
-                                : { color: 'var(--text)', borderColor: 'var(--border-muted)' }
-                            }
-                          >
-                            {contentType === 'ebook' || combinedMode ? <CheckIcon /> : <BookIcon />}
-                            <span>Books</span>
-                          </button>
-                          <button
-                            type="button"
-                            onClick={() => handleContentTypeSelect('audiobook')}
-                            className={`flex items-center gap-2 rounded-xl border px-3 py-2 text-sm font-medium transition-colors ${
-                              contentType === 'audiobook' || combinedMode
-                                ? 'bg-emerald-600 text-white'
-                                : 'hover-surface'
-                            }`}
-                            style={
-                              contentType === 'audiobook' || combinedMode
-                                ? { borderColor: 'rgb(16 185 129 / 0.7)' }
-                                : { color: 'var(--text)', borderColor: 'var(--border-muted)' }
-                            }
-                          >
-                            {contentType === 'audiobook' || combinedMode ? (
-                              <CheckIcon />
-                            ) : (
-                              <AudiobookIcon />
-                            )}
-                            <span>Audiobooks</span>
-                          </button>
-                        </div>
-                        {onCombinedModeChange &&
-                          (() => {
-                            const lineColor = combinedMode
-                              ? 'bg-emerald-500'
-                              : 'bg-(--border-muted) group-hover:bg-zinc-400 dark:group-hover:bg-zinc-500';
-                            return (
-                              <Tooltip
-                                content="Combined search"
-                                position="bottom"
-                                triggerClassName="w-full"
-                              >
-                                <button
-                                  type="button"
-                                  onClick={handleCombinedModeSelect}
-                                  className="group w-full"
-                                  aria-label="Combined search"
-                                >
-                                  {/* Bracket connector: vertical drops + horizontal bar with icon */}
-                                  <div className="relative flex h-7 items-end">
-                                    {/* Left vertical */}
-                                    <div
-                                      className={`absolute top-1.5 bottom-[11px] left-[25%] w-px transition-colors ${lineColor}`}
-                                    />
-                                    {/* Right vertical */}
-                                    <div
-                                      className={`absolute top-1.5 right-[25%] bottom-[11px] w-px transition-colors ${lineColor}`}
-                                    />
-                                    {/* Horizontal bar – left segment */}
-                                    <div
-                                      className={`absolute bottom-[11px] left-[25%] h-px transition-colors ${lineColor}`}
-                                      style={{ width: 'calc(25% - 16px)' }}
-                                    />
-                                    {/* Horizontal bar – right segment */}
-                                    <div
-                                      className={`absolute right-[25%] bottom-[11px] h-px transition-colors ${lineColor}`}
-                                      style={{ width: 'calc(25% - 16px)' }}
-                                    />
-                                    {/* Chain icon centered at bottom */}
-                                    <div
-                                      className={`relative z-10 mx-auto rounded-full p-1 transition-colors ${
-                                        combinedMode
-                                          ? 'bg-emerald-600 text-white'
-                                          : 'bg-(--bg) text-zinc-400 group-hover:bg-zinc-200 group-hover:text-zinc-600 dark:text-zinc-500 dark:group-hover:bg-zinc-700 dark:group-hover:text-zinc-300'
-                                      }`}
-                                    >
-                                      <svg
-                                        className="h-3.5 w-3.5"
-                                        fill="none"
-                                        viewBox="0 0 24 24"
-                                        strokeWidth="2"
-                                        stroke="currentColor"
-                                        aria-hidden="true"
-                                      >
-                                        {combinedModeLocked ? (
-                                          <path
-                                            strokeLinecap="round"
-                                            strokeLinejoin="round"
-                                            d="M16.5 10.5V6.75a4.5 4.5 0 1 0-9 0v3.75m-.75 11.25h10.5a2.25 2.25 0 0 0 2.25-2.25v-6.75a2.25 2.25 0 0 0-2.25-2.25H6.75a2.25 2.25 0 0 0-2.25 2.25v6.75a2.25 2.25 0 0 0 2.25 2.25Z"
-                                          />
-                                        ) : (
-                                          <path
-                                            strokeLinecap="round"
-                                            strokeLinejoin="round"
-                                            d="M13.19 8.688a4.5 4.5 0 0 1 1.242 7.244l-4.5 4.5a4.5 4.5 0 0 1-6.364-6.364l1.757-1.757m13.35-.622 1.757-1.757a4.5 4.5 0 0 0-6.364-6.364l-4.5 4.5a4.5 4.5 0 0 0 1.242 7.244"
-                                          />
-                                        )}
-                                      </svg>
-                                    </div>
-                                  </div>
-                                </button>
-                              </Tooltip>
-                            );
-                          })()}
-                      </div>
-                    )}
-
-                    <div className={showContentTypeSelector ? 'pt-2' : ''}>
-                      <div className="flex items-center justify-between px-1 pb-1.5">
-                        <span className="text-xs font-medium tracking-wide uppercase opacity-60">
-                          Search By
-                        </span>
-                        {!showContentTypeSelector && onAdvancedToggle && (
-                          <button
-                            type="button"
-                            onClick={() => {
-                              setIsSelectorOpen(false);
-                              onAdvancedToggle();
-                            }}
-                            className={`-mt-1.5 -mr-1 -mb-0.5 flex items-center gap-1.5 rounded-xl px-4 py-2.5 text-xs font-medium transition-colors ${
-                              isAdvancedActive
-                                ? `${searchMode === 'direct' ? 'bg-sky-700' : 'bg-emerald-600'} text-white`
-                                : 'hover-surface'
-                            }`}
-                            style={
-                              isAdvancedActive
-                                ? {
-                                    borderColor:
-                                      searchMode === 'direct'
-                                        ? 'rgb(3 105 161 / 0.7)'
-                                        : 'rgb(16 185 129 / 0.7)',
-                                  }
-                                : { color: 'var(--text-muted)' }
-                            }
-                          >
-                            <svg
-                              className="h-3.5 w-3.5"
-                              xmlns="http://www.w3.org/2000/svg"
-                              fill="none"
-                              viewBox="0 0 24 24"
-                              strokeWidth="1.5"
-                              stroke="currentColor"
-                              aria-hidden="true"
-                            >
-                              <path
-                                strokeLinecap="round"
-                                strokeLinejoin="round"
-                                d="M10.5 6h9.75M10.5 6a1.5 1.5 0 1 1-3 0m3 0a1.5 1.5 0 1 0-3 0M3.75 6H7.5m3 12h9.75m-9.75 0a1.5 1.5 0 0 1-3 0m3 0a1.5 1.5 0 0 0-3 0m-3.75 0H7.5m9-6h3.75m-3.75 0a1.5 1.5 0 0 1-3 0m3 0a1.5 1.5 0 0 0-3 0m-9.75 0h9.75"
-                              />
-                            </svg>
-                            Options
-                          </button>
-                        )}
-                      </div>
-                      <div className="grid grid-cols-2 gap-2">
-                        {queryTargets.map((target) => {
-                          const isActive = target.key === activeTarget?.key;
-                          return (
-                            <button
-                              type="button"
-                              key={target.key}
-                              onClick={() => handleQueryTargetSelect(target.key)}
-                              title={target.description || target.label}
-                              aria-label={target.label}
-                              className={`flex min-w-0 items-center gap-2 rounded-xl border px-3 py-2 text-sm font-medium transition-colors ${
-                                isActive
-                                  ? `${searchMode === 'direct' ? 'bg-sky-700' : 'bg-emerald-600'} text-white`
-                                  : 'hover-surface'
-                              }`}
-                              style={
-                                isActive
-                                  ? {
-                                      borderColor:
-                                        searchMode === 'direct'
-                                          ? 'rgb(3 105 161 / 0.7)'
-                                          : 'rgb(16 185 129 / 0.7)',
-                                    }
-                                  : { color: 'var(--text)', borderColor: 'var(--border-muted)' }
-                              }
-                            >
-                              {isActive && <CheckIcon />}
-                              <span className="block truncate">{target.label}</span>
-                            </button>
-                          );
-                        })}
-                      </div>
-                    </div>
-                  </div>
-                </div>
-              )}
            </div>
          )}

@@ -999,7 +732,7 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
                        onClick={() => {
                          onChange(option.value, option.label);
                          setIsSelectOpen(false);
-                          setTimeout(() => onSubmitRef.current(), 0);
+                          setTimeout(() => submitLatest(), 0);
                        }}
                        className={`flex w-full items-center gap-3 px-5 py-2.5 text-left text-sm transition-colors ${
                          isSelected ? '' : 'hover-surface'
@@ -1069,7 +802,7 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
                          setAutocompleteSelection(option.value, option.label);
                          onChange(option.value, option.label);
                          setIsAutocompleteOpen(false);
-                          setTimeout(() => onSubmitRef.current(), 0);
+                          setTimeout(() => submitLatest(), 0);
                        }}
                        className="hover-surface w-full px-5 py-3 text-left text-sm transition-colors"
                        style={{ color: 'var(--text)' }}
@@ -1087,6 +820,246 @@ export const SearchBar = forwardRef<SearchBarHandle, SearchBarProps>(
            </div>
          )}
        </div>
+
+        {showQueryTargetSelector && isSelectorOpen && (
+          <div
+            id={SEARCH_CONTROLS_PANEL_ID}
+            className="animate-fade-in-down flex flex-wrap items-start gap-x-8 gap-y-2 px-1 pt-2"
+            ref={controlsPanelRef}
+          >
+            {showContentTypeSelector && (
+              <div className="shrink-0">
+                <div className="flex items-center justify-between pb-1.5">
+                  <span className="text-xs font-medium tracking-wide uppercase opacity-60">
+                    Content
+                  </span>
+                  {onAdvancedToggle && (
+                    <button
+                      type="button"
+                      onClick={onAdvancedToggle}
+                      className={`-mt-1.5 -mr-1 -mb-0.5 flex cursor-pointer items-center gap-1.5 rounded-xl px-4 py-2.5 text-xs font-medium transition-colors ${
+                        isAdvancedActive ? 'bg-emerald-600 text-white' : 'hover-surface'
+                      }`}
+                      style={
+                        isAdvancedActive
+                          ? { borderColor: 'rgb(16 185 129 / 0.7)' }
+                          : { color: 'var(--text-muted)' }
+                      }
+                    >
+                      <svg
+                        className="h-3.5 w-3.5"
+                        xmlns="http://www.w3.org/2000/svg"
+                        fill="none"
+                        viewBox="0 0 24 24"
+                        strokeWidth="1.5"
+                        stroke="currentColor"
+                        aria-hidden="true"
+                      >
+                        <path
+                          strokeLinecap="round"
+                          strokeLinejoin="round"
+                          d="M10.5 6h9.75M10.5 6a1.5 1.5 0 1 1-3 0m3 0a1.5 1.5 0 1 0-3 0M3.75 6H7.5m3 12h9.75m-9.75 0a1.5 1.5 0 0 1-3 0m3 0a1.5 1.5 0 0 0-3 0m-3.75 0H7.5m9-6h3.75m-3.75 0a1.5 1.5 0 0 1-3 0m3 0a1.5 1.5 0 0 0-3 0m-9.75 0h9.75"
+                        />
+                      </svg>
+                      Options
+                    </button>
+                  )}
+                </div>
+                <div className="grid w-fit grid-cols-2 gap-2">
+                  <button
+                    type="button"
+                    onClick={() => handleContentTypeSelect('ebook')}
+                    className={`flex cursor-pointer items-center gap-2 rounded-xl border px-3 py-2 text-sm font-medium transition-colors ${
+                      contentType === 'ebook' || combinedSelectionActive
+                        ? 'bg-emerald-600 text-white'
+                        : 'hover-surface'
+                    }`}
+                    style={
+                      contentType === 'ebook' || combinedSelectionActive
+                        ? { borderColor: 'rgb(16 185 129 / 0.7)' }
+                        : { color: 'var(--text)', borderColor: 'var(--border-muted)' }
+                    }
+                  >
+                    <span className="flex w-4 justify-center">
+                      {contentType === 'ebook' || combinedSelectionActive ? (
+                        <CheckIcon />
+                      ) : (
+                        <BookIcon />
+                      )}
+                    </span>
+                    <span>Books</span>
+                  </button>
+                  <button
+                    type="button"
+                    onClick={() => handleContentTypeSelect('audiobook')}
+                    className={`flex cursor-pointer items-center gap-2 rounded-xl border px-3 py-2 text-sm font-medium transition-colors ${
+                      contentType === 'audiobook' || combinedSelectionActive
+                        ? 'bg-emerald-600 text-white'
+                        : 'hover-surface'
+                    }`}
+                    style={
+                      contentType === 'audiobook' || combinedSelectionActive
+                        ? { borderColor: 'rgb(16 185 129 / 0.7)' }
+                        : { color: 'var(--text)', borderColor: 'var(--border-muted)' }
+                    }
+                  >
+                    <span className="flex w-4 justify-center">
+                      {contentType === 'audiobook' || combinedSelectionActive ? (
+                        <CheckIcon />
+                      ) : (
+                        <AudiobookIcon />
+                      )}
+                    </span>
+                    <span>Audiobooks</span>
+                  </button>
+                  {combinedToggleAvailable && (
+                    <div className="col-span-2">
+                      <Tooltip
+                        content="Combined search"
+                        position="bottom"
+                        triggerClassName="w-full"
+                      >
+                        <button
+                          type="button"
+                          onClick={handleCombinedModeSelect}
+                          className="group w-full cursor-pointer"
+                          aria-label="Combined search"
+                        >
+                          <div className="relative flex h-7 items-end">
+                            <div
+                              className={`absolute top-1.5 bottom-[11px] left-[25%] w-px transition-colors ${combinedLineColor}`}
+                            />
+                            <div
+                              className={`absolute top-1.5 right-[25%] bottom-[11px] w-px transition-colors ${combinedLineColor}`}
+                            />
+                            <div
+                              className={`absolute bottom-[11px] left-[25%] h-px transition-colors ${combinedLineColor}`}
+                              style={{ width: 'calc(25% - 16px)' }}
+                            />
+                            <div
+                              className={`absolute right-[25%] bottom-[11px] h-px transition-colors ${combinedLineColor}`}
+                              style={{ width: 'calc(25% - 16px)' }}
+                            />
+                            <div
+                              className={`relative z-10 mx-auto rounded-full p-1 transition-colors ${
+                                combinedSelectionActive
+                                  ? 'bg-emerald-600 text-white'
+                                  : 'bg-(--bg) text-zinc-400 group-hover:bg-zinc-200 group-hover:text-zinc-600 dark:text-zinc-500 dark:group-hover:bg-zinc-700 dark:group-hover:text-zinc-300'
+                              }`}
+                            >
+                              <svg
+                                className="h-3.5 w-3.5"
+                                fill="none"
+                                viewBox="0 0 24 24"
+                                strokeWidth="2"
+                                stroke="currentColor"
+                                aria-hidden="true"
+                              >
+                                {combinedSelectorLocked ? (
+                                  <path
+                                    strokeLinecap="round"
+                                    strokeLinejoin="round"
+                                    d="M16.5 10.5V6.75a4.5 4.5 0 1 0-9 0v3.75m-.75 11.25h10.5a2.25 2.25 0 0 0 2.25-2.25v-6.75a2.25 2.25 0 0 0-2.25-2.25H6.75a2.25 2.25 0 0 0-2.25 2.25v6.75a2.25 2.25 0 0 0 2.25 2.25Z"
+                                  />
+                                ) : (
+                                  <path
+                                    strokeLinecap="round"
+                                    strokeLinejoin="round"
+                                    d="M13.19 8.688a4.5 4.5 0 0 1 1.242 7.244l-4.5 4.5a4.5 4.5 0 0 1-6.364-6.364l1.757-1.757m13.35-.622 1.757-1.757a4.5 4.5 0 0 0-6.364-6.364l-4.5 4.5a4.5 4.5 0 0 0 1.242 7.244"
+                                  />
+                                )}
+                              </svg>
+                            </div>
+                          </div>
+                        </button>
+                      </Tooltip>
+                    </div>
+                  )}
+                </div>
+              </div>
+            )}
+
+            {queryTargets.length > 1 && (
+              <div className="shrink-0">
+                <div className="flex items-center justify-between pb-1.5">
+                  <span className="text-xs font-medium tracking-wide uppercase opacity-60">
+                    Search By
+                  </span>
+                  {!showContentTypeSelector && onAdvancedToggle && (
+                    <button
+                      type="button"
+                      onClick={onAdvancedToggle}
+                      className={`-mt-1.5 -mr-1 -mb-0.5 flex cursor-pointer items-center gap-1.5 rounded-xl px-4 py-2.5 text-xs font-medium transition-colors ${
+                        isAdvancedActive
+                          ? `${searchMode === 'direct' ? 'bg-sky-700' : 'bg-emerald-600'} text-white`
+                          : 'hover-surface'
+                      }`}
+                      style={
+                        isAdvancedActive
+                          ? {
+                              borderColor:
+                                searchMode === 'direct'
+                                  ? 'rgb(3 105 161 / 0.7)'
+                                  : 'rgb(16 185 129 / 0.7)',
+                            }
+                          : { color: 'var(--text-muted)' }
+                      }
+                    >
+                      <svg
+                        className="h-3.5 w-3.5"
+                        xmlns="http://www.w3.org/2000/svg"
+                        fill="none"
+                        viewBox="0 0 24 24"
+                        strokeWidth="1.5"
+                        stroke="currentColor"
+                        aria-hidden="true"
+                      >
+                        <path
+                          strokeLinecap="round"
+                          strokeLinejoin="round"
+                          d="M10.5 6h9.75M10.5 6a1.5 1.5 0 1 1-3 0m3 0a1.5 1.5 0 1 0-3 0M3.75 6H7.5m3 12h9.75m-9.75 0a1.5 1.5 0 0 1-3 0m3 0a1.5 1.5 0 0 0-3 0m-3.75 0H7.5m9-6h3.75m-3.75 0a1.5 1.5 0 0 1-3 0m3 0a1.5 1.5 0 0 0-3 0m-9.75 0h9.75"
+                        />
+                      </svg>
+                      Options
+                    </button>
+                  )}
+                </div>
+                <div className="flex flex-wrap gap-2">
+                  {queryTargets.map((target) => {
+                    const isActive = target.key === activeTarget?.key;
+                    return (
+                      <button
+                        type="button"
+                        key={target.key}
+                        onClick={() => handleQueryTargetSelect(target.key)}
+                        title={target.description || target.label}
+                        aria-label={target.label}
+                        className={`flex min-w-0 cursor-pointer items-center gap-2 rounded-xl border px-3 py-2 text-sm font-medium transition-colors ${
+                          isActive
+                            ? `${searchMode === 'direct' ? 'bg-sky-700' : 'bg-emerald-600'} text-white`
+                            : 'hover-surface'
+                        }`}
+                        style={
+                          isActive
+                            ? {
+                                borderColor:
+                                  searchMode === 'direct'
+                                    ? 'rgb(3 105 161 / 0.7)'
+                                    : 'rgb(16 185 129 / 0.7)',
+                              }
+                            : { color: 'var(--text)', borderColor: 'var(--border-muted)' }
+                        }
+                      >
+                        {isActive && <CheckIcon />}
+                        <span className="block truncate">{target.label}</span>
+                      </button>
+                    );
+                  })}
+                </div>
+              </div>
+            )}
+          </div>
+        )}
      </>
    );
  },
@@ -3,6 +3,7 @@ import type { Dispatch, SetStateAction } from 'react';
 import { useMountEffect } from '@/hooks/useMountEffect';
 import type { AppConfig, AdvancedFilterState, ContentType, SearchMode, SortOption } from '@/types';
 import { buildSearchQuery } from '@/utils/buildSearchQuery';
+import { resolveDefaultLanguageCodes } from '@/utils/languageFilters';
 import { getEffectiveMetadataSort } from '@/utils/metadataSort';
 import type { ParsedUrlSearch } from '@/utils/parseUrlSearchParams';

@@ -73,10 +74,10 @@ export const UrlSearchBootstrapMount = ({
    }

    const bookLanguages = config.book_languages || [];
-    const defaultLanguageCodes =
-      config.default_language && config.default_language.length > 0
-        ? config.default_language
-        : [bookLanguages[0]?.code || 'en'];
+    const defaultLanguageCodes = resolveDefaultLanguageCodes(
+      config.default_language,
+      bookLanguages,
+    );

    if (parsedParams.searchInput) {
      setSearchInput(parsedParams.searchInput);
@@ -562,6 +562,10 @@ export const ActivityCard = ({
    }

    return () => observer.disconnect();
+    // None of these are read here - they are all re-measure triggers. The title's overflow
+    // depends on its text and on the width it is laid out in, and opening either panel
+    // reflows the card. Drop them and the tooltip-on-truncation goes stale.
+    // oxlint-disable-next-line react/exhaustive-effect-dependencies
  }, [item.title, item.author, isRequestDetailsOpen, isRequestRejectOpen]);

  const reviewRecord = item.requestRecord;
@@ -347,6 +347,10 @@ function SettingsContentPanel({
    if (scrollRef.current) {
      scrollRef.current.scrollTop = 0;
    }
+    // `tab.name` is never read here - it is the trigger, and the whole point: the scroll
+    // position resets *because* the tab changed. Removing it strands the new tab at the
+    // previous one's offset.
+    // oxlint-disable-next-line react/exhaustive-effect-dependencies
  }, [embedded, tab.name]);

  const updateCustomFieldUiState = useCallback((fieldKey: string, key: string, value: unknown) => {
@@ -407,6 +411,9 @@ function SettingsContentPanel({
    if (scrollRef.current) {
      scrollRef.current.scrollTop = 0;
    }
+    // `activeTakeOverFieldKey` is never read here - it is the trigger. Entering or leaving
+    // a subpage takeover is exactly when the scroll must reset.
+    // oxlint-disable-next-line react/exhaustive-effect-dependencies
  }, [embedded, activeTakeOverFieldKey]);

  const visibleFields = useMemo(() => {
@@ -1,5 +1,6 @@
 import { useCallback, useLayoutEffect, useRef } from 'react';

+import { useLatestCallback } from '../../../hooks/useLatestCallback';
 import { useMountEffect } from '../../../hooks/useMountEffect';
 import type { AdminUser } from '../../../services/api';
 import { testAdminUserNotificationPreferences } from '../../../services/api';
@@ -169,12 +170,12 @@ export const UsersManagementField = ({
    }
  }, [backToList, onRefreshOverrideSummary, onSettingsSaved, onUiStateChange, saveEditedUser]);

-  const handleSaveUserOverridesRef = useRef(handleSaveUserOverrides);
-  handleSaveUserOverridesRef.current = handleSaveUserOverrides;
-
-  const triggerSaveUserOverrides = useCallback(async () => {
-    await handleSaveUserOverridesRef.current();
-  }, []);
+  // Stored in parent UI state, so it must keep a stable identity while still invoking the
+  // latest handler. Not an Effect Event: those must not be handed to another component.
+  // See useLatestCallback.
+  const triggerSaveUserOverrides = useLatestCallback(async () => {
+    await handleSaveUserOverrides();
+  });

  const handleOpenOverrides = () => {
    if (editingUser) {
@@ -44,162 +44,172 @@ const getOptionsIdentity = (options: MultiSelectFieldConfig['options']): string
 const getSelectionIdentity = (values: string[]): string =>
  values.toSorted((left, right) => left.localeCompare(right)).join('\u0001');

-export const MultiSelectField = ({
+interface MultiSelectVariantProps {
+  field: MultiSelectFieldConfig;
+  selected: string[];
+  onChange: (value: string[]) => void;
+  isDisabled: boolean;
+}
+
+// Dropdown variant - use DropdownList with checkboxes
+const MultiSelectDropdownField = ({
  field,
-  value: fieldValue,
+  selected,
  onChange,
-  disabled,
-}: MultiSelectFieldProps) => {
-  const selected = fieldValue ?? EMPTY_SELECTION;
-  // disabled prop is already computed by SettingsContent.getDisabledState()
-  const isDisabled = disabled ?? false;
+  isDisabled,
+}: MultiSelectVariantProps) => {
+  const optionValues = field.options.map((opt) => opt.value);
+  const optionSet = new Set(optionValues);
+  const hasAllOption = optionSet.has(ALL_OPTION_VALUE);
+  const orderedOptions = hasAllOption
+    ? [
+        ...field.options.filter((opt) => opt.value === ALL_OPTION_VALUE),
+        ...field.options.filter((opt) => opt.value !== ALL_OPTION_VALUE),
+      ]
+    : field.options;
+  const nonAllValues = orderedOptions
+    .map((opt) => opt.value)
+    .filter((optValue) => optValue !== ALL_OPTION_VALUE);

-  // Dropdown variant - use DropdownList with checkboxes
-  if (field.variant === 'dropdown') {
-    const optionValues = field.options.map((opt) => opt.value);
-    const optionSet = new Set(optionValues);
-    const hasAllOption = optionSet.has(ALL_OPTION_VALUE);
-    const orderedOptions = hasAllOption
-      ? [
-          ...field.options.filter((opt) => opt.value === ALL_OPTION_VALUE),
-          ...field.options.filter((opt) => opt.value !== ALL_OPTION_VALUE),
-        ]
-      : field.options;
-    const nonAllValues = orderedOptions
-      .map((opt) => opt.value)
-      .filter((optValue) => optValue !== ALL_OPTION_VALUE);
+  const normalizeValues = (values: string[]): string[] => {
+    const deduped = new Set(
+      values
+        .map((entry) => entry.trim())
+        .filter((entry) => entry.length > 0 && optionSet.has(entry)),
+    );
+    return orderedOptions.map((opt) => opt.value).filter((optValue) => deduped.has(optValue));
+  };

-    const normalizeValues = (values: string[]): string[] => {
-      const deduped = new Set(
-        values
-          .map((entry) => entry.trim())
-          .filter((entry) => entry.length > 0 && optionSet.has(entry)),
-      );
-      return orderedOptions.map((opt) => opt.value).filter((optValue) => deduped.has(optValue));
-    };
+  const selectedExplicit = normalizeValues(selected);
+  const allSelected =
+    hasAllOption &&
+    (selectedExplicit.includes(ALL_OPTION_VALUE) ||
+      (nonAllValues.length > 0 &&
+        nonAllValues.every((optValue) => selectedExplicit.includes(optValue))));

-    const selectedExplicit = normalizeValues(selected);
-    const allSelected =
-      hasAllOption &&
-      (selectedExplicit.includes(ALL_OPTION_VALUE) ||
-        (nonAllValues.length > 0 &&
-          nonAllValues.every((optValue) => selectedExplicit.includes(optValue))));
+  // Build parent -> children map for cascading selection
+  const parentChildMap = new Map<string, string[]>();
+  orderedOptions.forEach((opt) => {
+    if (opt.childOf) {
+      const children = parentChildMap.get(opt.childOf) || [];
+      children.push(opt.value);
+      parentChildMap.set(opt.childOf, children);
+    }
+  });

-    // Build parent -> children map for cascading selection
-    const parentChildMap = new Map<string, string[]>();
-    orderedOptions.forEach((opt) => {
-      if (opt.childOf) {
-        const children = parentChildMap.get(opt.childOf) || [];
-        children.push(opt.value);
-        parentChildMap.set(opt.childOf, children);
+  // Check which children are implicitly selected via parent
+  const selectedForCascade = allSelected
+    ? selectedExplicit.filter((optValue) => optValue !== ALL_OPTION_VALUE)
+    : selectedExplicit;
+  const implicitlySelected = new Set<string>();
+  selectedForCascade.forEach((val) => {
+    const children = parentChildMap.get(val);
+    if (children) {
+      children.forEach((child) => implicitlySelected.add(child));
+    }
+  });
+
+  // Build options with disabled state for implicitly selected children
+  const dropdownOptions = orderedOptions.map((opt) => ({
+    value: opt.value,
+    label: opt.label,
+    disabled: !allSelected && implicitlySelected.has(opt.value),
+  }));
+
+  // For display purposes:
+  // - if "all" is active, check every option
+  // - otherwise show explicit + implicit parent/child selections
+  const displayValue = allSelected
+    ? [ALL_OPTION_VALUE, ...nonAllValues]
+    : normalizeValues([...selectedExplicit, ...Array.from(implicitlySelected)]);
+
+  const handleDropdownChange = (newValue: string | string[]) => {
+    const nextValues = normalizeValues(Array.isArray(newValue) ? newValue : [newValue]);
+
+    if (hasAllOption) {
+      const includesAll = nextValues.includes(ALL_OPTION_VALUE);
+
+      // When currently "all" is active:
+      // - unticking "all" clears everything
+      // - unticking a specific option converts to explicit subset
+      if (allSelected && !includesAll && nextValues.length === nonAllValues.length) {
+        onChange([]);
+        return;
      }
-    });
-
-    // Check which children are implicitly selected via parent
-    const selectedForCascade = allSelected
-      ? selectedExplicit.filter((optValue) => optValue !== ALL_OPTION_VALUE)
-      : selectedExplicit;
-    const implicitlySelected = new Set<string>();
-    selectedForCascade.forEach((val) => {
-      const children = parentChildMap.get(val);
-      if (children) {
-        children.forEach((child) => implicitlySelected.add(child));
-      }
-    });
-
-    // Build options with disabled state for implicitly selected children
-    const dropdownOptions = orderedOptions.map((opt) => ({
-      value: opt.value,
-      label: opt.label,
-      disabled: !allSelected && implicitlySelected.has(opt.value),
-    }));
-
-    // For display purposes:
-    // - if "all" is active, check every option
-    // - otherwise show explicit + implicit parent/child selections
-    const displayValue = allSelected
-      ? [ALL_OPTION_VALUE, ...nonAllValues]
-      : normalizeValues([...selectedExplicit, ...Array.from(implicitlySelected)]);
-
-    const handleDropdownChange = (newValue: string | string[]) => {
-      const nextValues = normalizeValues(Array.isArray(newValue) ? newValue : [newValue]);
-
-      if (hasAllOption) {
-        const includesAll = nextValues.includes(ALL_OPTION_VALUE);
-
-        // When currently "all" is active:
-        // - unticking "all" clears everything
-        // - unticking a specific option converts to explicit subset
-        if (allSelected && !includesAll && nextValues.length === nonAllValues.length) {
-          onChange([]);
-          return;
-        }
-        if (allSelected && includesAll && nextValues.length < optionValues.length) {
-          onChange(nextValues.filter((entry) => entry !== ALL_OPTION_VALUE));
-          return;
-        }
-
-        if (includesAll) {
-          onChange([ALL_OPTION_VALUE]);
-          return;
-        }
-
-        // If user selects every specific option individually, collapse to "all".
-        if (
-          nonAllValues.length > 0 &&
-          nonAllValues.every((optValue) => nextValues.includes(optValue))
-        ) {
-          onChange([ALL_OPTION_VALUE]);
-          return;
-        }
+      if (allSelected && includesAll && nextValues.length < optionValues.length) {
+        onChange(nextValues.filter((entry) => entry !== ALL_OPTION_VALUE));
+        return;
      }

-      // Filter out implicitly selected values - only store explicit selections.
-      const explicitOnly = nextValues.filter((entry) => !implicitlySelected.has(entry));
-      onChange(explicitOnly);
-    };
+      if (includesAll) {
+        onChange([ALL_OPTION_VALUE]);
+        return;
+      }

-    // Custom summary formatter - only count explicit selections
-    const summaryFormatter = () => {
-      if (allSelected) {
-        return orderedOptions.find((opt) => opt.value === ALL_OPTION_VALUE)?.label || 'All';
+      // If user selects every specific option individually, collapse to "all".
+      if (
+        nonAllValues.length > 0 &&
+        nonAllValues.every((optValue) => nextValues.includes(optValue))
+      ) {
+        onChange([ALL_OPTION_VALUE]);
+        return;
      }
-      if (selectedExplicit.length === 0) {
-        return <span className="opacity-60">{field.placeholder || 'Select categories...'}</span>;
-      }
-      const selectedLabels = selectedExplicit
-        .map((v) => orderedOptions.find((o) => o.value === v)?.label)
-        .filter(Boolean);
-      if (selectedLabels.length === 1) {
-        return selectedLabels[0];
-      }
-      const [first, second, ...rest] = selectedLabels;
-      const suffix = rest.length > 0 ? ` +${rest.length}` : '';
-      return `${first}, ${second ?? ''}${suffix}`.trim();
-    };
-
-    if (isDisabled) {
-      return (
-        <div className="w-full cursor-not-allowed rounded-lg border border-(--border-muted) bg-(--bg-soft) px-3 py-2 text-sm opacity-60">
-          {summaryFormatter()}
-        </div>
-      );
    }

+    // Filter out implicitly selected values - only store explicit selections.
+    const explicitOnly = nextValues.filter((entry) => !implicitlySelected.has(entry));
+    onChange(explicitOnly);
+  };
+
+  // Custom summary formatter - only count explicit selections
+  const summaryFormatter = () => {
+    if (allSelected) {
+      return orderedOptions.find((opt) => opt.value === ALL_OPTION_VALUE)?.label || 'All';
+    }
+    if (selectedExplicit.length === 0) {
+      return <span className="opacity-60">{field.placeholder || 'Select categories...'}</span>;
+    }
+    const selectedLabels = selectedExplicit
+      .map((v) => orderedOptions.find((o) => o.value === v)?.label)
+      .filter(Boolean);
+    if (selectedLabels.length === 1) {
+      return selectedLabels[0];
+    }
+    const [first, second, ...rest] = selectedLabels;
+    const suffix = rest.length > 0 ? ` +${rest.length}` : '';
+    return `${first}, ${second ?? ''}${suffix}`.trim();
+  };
+
+  if (isDisabled) {
    return (
-      <DropdownList
-        options={dropdownOptions}
-        value={displayValue}
-        onChange={handleDropdownChange}
-        multiple
-        showCheckboxes
-        keepOpenOnSelect
-        placeholder={field.placeholder || 'Select categories...'}
-        widthClassName="w-full"
-        summaryFormatter={summaryFormatter}
-      />
+      <div className="w-full cursor-not-allowed rounded-lg border border-(--border-muted) bg-(--bg-soft) px-3 py-2 text-sm opacity-60">
+        {summaryFormatter()}
+      </div>
    );
  }
+
+  return (
+    <DropdownList
+      options={dropdownOptions}
+      value={displayValue}
+      onChange={handleDropdownChange}
+      multiple
+      showCheckboxes
+      keepOpenOnSelect
+      placeholder={field.placeholder || 'Select categories...'}
+      widthClassName="w-full"
+      summaryFormatter={summaryFormatter}
+    />
+  );
+};
+
+// Pill variant - inline toggle buttons that collapse past a threshold
+const MultiSelectPillsField = ({
+  field,
+  selected,
+  onChange,
+  isDisabled,
+}: MultiSelectVariantProps) => {
  const [isExpanded, setIsExpanded] = useState(false);
  // Initialize based on option count to avoid flash of expanded content
  const [needsCollapse, setNeedsCollapse] = useState(
@@ -353,3 +363,35 @@ export const MultiSelectField = ({
    </div>
  );
 };
+
+export const MultiSelectField = ({
+  field,
+  value: fieldValue,
+  onChange,
+  disabled,
+}: MultiSelectFieldProps) => {
+  const selected = fieldValue ?? EMPTY_SELECTION;
+  // disabled prop is already computed by SettingsContent.getDisabledState()
+  const isDisabled = disabled ?? false;
+
+  // Each variant is its own component so neither calls hooks conditionally.
+  if (field.variant === 'dropdown') {
+    return (
+      <MultiSelectDropdownField
+        field={field}
+        selected={selected}
+        onChange={onChange}
+        isDisabled={isDisabled}
+      />
+    );
+  }
+
+  return (
+    <MultiSelectPillsField
+      field={field}
+      selected={selected}
+      onChange={onChange}
+      isDisabled={isDisabled}
+    />
+  );
+};
@@ -353,12 +353,12 @@ export const TableField = ({ field, value, onChange, disabled }: TableFieldProps
                );
              }

-              // text/path
+              // text/password/path
              return (
                <div key={col.key} className="flex min-w-0 flex-col gap-1">
                  {mobileLabel}
                  <input
-                    type="text"
+                    type={col.type === 'password' ? 'password' : 'text'}
                    value={toPrimitiveString(cellValue)}
                    onChange={(e) => updateCell(rowIndex, col.key, e.target.value)}
                    placeholder={col.placeholder}
@@ -7,7 +7,12 @@ import type {
 } from '../../../types/settings';
 import { HeadingField, MultiSelectField, SelectField, TextField } from '../fields';
 import { FieldWrapper } from '../shared';
-import { getFieldByKey, toNormalizedLowercaseTextValue, toTextValue } from './fieldHelpers';
+import {
+  getFieldByKey,
+  resolveListOverride,
+  toNormalizedLowercaseTextValue,
+  toTextValue,
+} from './fieldHelpers';
 import type { PerUserSettings } from './types';

 interface UserOverridesSectionProps {
@@ -175,16 +180,6 @@ export const UserOverridesSection = ({
    label: 'Email Recipient',
    description: 'Email address used for this user in Email output mode.',
  };
-  const browserDownloadGlobalValue = Array.isArray(globalValues.DOWNLOAD_TO_BROWSER_CONTENT_TYPES)
-    ? globalValues.DOWNLOAD_TO_BROWSER_CONTENT_TYPES.map((entry) => String(entry).trim()).filter(
-        (entry) => entry.length > 0,
-      )
-    : [];
-  const browserDownloadUserValue = Array.isArray(userSettings.DOWNLOAD_TO_BROWSER_CONTENT_TYPES)
-    ? userSettings.DOWNLOAD_TO_BROWSER_CONTENT_TYPES.map((entry) => entry.trim()).filter(
-        (entry) => entry.length > 0,
-      )
-    : [];

  const isOverridden = (key: DeliverySettingKey): boolean => {
    if (
@@ -200,10 +195,12 @@ export const UserOverridesSection = ({
    return userValue !== globalValue;
  };

-  const isBrowserDownloadOverridden =
-    Object.prototype.hasOwnProperty.call(userSettings, 'DOWNLOAD_TO_BROWSER_CONTENT_TYPES') &&
-    userSettings.DOWNLOAD_TO_BROWSER_CONTENT_TYPES !== null &&
-    JSON.stringify(browserDownloadUserValue) !== JSON.stringify(browserDownloadGlobalValue);
+  const { value: browserDownloadContentTypes, isOverridden: isBrowserDownloadOverridden } =
+    resolveListOverride(
+      userSettings.DOWNLOAD_TO_BROWSER_CONTENT_TYPES,
+      globalValues.DOWNLOAD_TO_BROWSER_CONTENT_TYPES,
+      Object.prototype.hasOwnProperty.call(userSettings, 'DOWNLOAD_TO_BROWSER_CONTENT_TYPES'),
+    );

  const resetKeys = (keys: DeliverySettingKey[]) => {
    setUserSettings((prev) => {
@@ -232,9 +229,6 @@ export const UserOverridesSection = ({
  const outputModeValue = readValue('BOOKS_OUTPUT_MODE', 'folder');
  const effectiveOutputMode = normalizeMode(outputModeValue);

-  const browserDownloadContentTypes = isBrowserDownloadOverridden
-    ? browserDownloadUserValue
-    : browserDownloadGlobalValue;
  const destinationValue = readValue('DESTINATION');
  const destinationAudiobookValue = readValue('DESTINATION_AUDIOBOOK');
  const libraryValue = readValue('BOOKLORE_LIBRARY_ID');
@@ -1,8 +1,17 @@
 import type { DeliveryPreferencesResponse } from '../../../services/api';
-import type { HeadingFieldConfig, SelectFieldConfig } from '../../../types/settings';
-import { HeadingField, SelectField } from '../fields';
+import type {
+  HeadingFieldConfig,
+  MultiSelectFieldConfig,
+  SelectFieldConfig,
+} from '../../../types/settings';
+import { HeadingField, MultiSelectField, SelectField } from '../fields';
 import { FieldWrapper } from '../shared';
-import { getFieldByKey, toNormalizedLowercaseTextValue, toTextValue } from './fieldHelpers';
+import {
+  getFieldByKey,
+  resolveListOverride,
+  toNormalizedLowercaseTextValue,
+  toTextValue,
+} from './fieldHelpers';
 import type { PerUserSettings } from './types';

 interface UserSearchPreferencesSectionProps {
@@ -14,6 +23,7 @@ interface UserSearchPreferencesSectionProps {

 type SearchSettingKey =
  | 'SEARCH_MODE'
+  | 'BOOK_LANGUAGE'
  | 'METADATA_PROVIDER'
  | 'METADATA_PROVIDER_AUDIOBOOK'
  | 'DEFAULT_RELEASE_SOURCE'
@@ -68,6 +78,15 @@ const fallbackDefaultAudiobookReleaseSourceField: SelectFieldConfig = {
  options: [{ value: '', label: 'Use book release source' }],
 };

+const fallbackBookLanguageField: MultiSelectFieldConfig = {
+  type: 'MultiSelectField',
+  key: 'BOOK_LANGUAGE',
+  label: 'Default Book Languages',
+  description: 'Default language filter for searches.',
+  value: [],
+  options: [],
+};
+
 const searchHeading: HeadingFieldConfig = {
  type: 'HeadingField',
  key: 'search_preferences_heading',
@@ -120,6 +139,13 @@ export const UserSearchPreferencesSection = ({
    'DEFAULT_RELEASE_SOURCE_AUDIOBOOK',
    fallbackDefaultAudiobookReleaseSourceField,
  );
+  const bookLanguageField = getFieldByKey(fields, 'BOOK_LANGUAGE', fallbackBookLanguageField);
+
+  const { value: bookLanguageValue, isOverridden: isBookLanguageOverridden } = resolveListOverride(
+    userSettings.BOOK_LANGUAGE,
+    globalValues.BOOK_LANGUAGE,
+    Object.prototype.hasOwnProperty.call(userSettings, 'BOOK_LANGUAGE'),
+  );

  const isOverridden = (key: SearchSettingKey): boolean => {
    if (
@@ -172,9 +198,12 @@ export const UserSearchPreferencesSection = ({
  const canOverrideDefaultAudiobookReleaseSource =
    isUserOverridable('DEFAULT_RELEASE_SOURCE_AUDIOBOOK') &&
    preferenceKeySet.has('DEFAULT_RELEASE_SOURCE_AUDIOBOOK');
+  const canOverrideBookLanguage =
+    isUserOverridable('BOOK_LANGUAGE') && preferenceKeySet.has('BOOK_LANGUAGE');

  if (
    !canOverrideSearchMode &&
+    !canOverrideBookLanguage &&
    !canOverrideMetadataProvider &&
    !canOverrideAudiobookMetadataProvider &&
    !canOverrideDefaultReleaseSource &&
@@ -208,6 +237,27 @@ export const UserSearchPreferencesSection = ({
        </FieldWrapper>
      )}

+      {canOverrideBookLanguage && (
+        <FieldWrapper
+          field={bookLanguageField}
+          resetAction={
+            isBookLanguageOverridden
+              ? {
+                  disabled: Boolean(bookLanguageField.fromEnv),
+                  onClick: () => resetKeys(['BOOK_LANGUAGE']),
+                }
+              : undefined
+          }
+        >
+          <MultiSelectField
+            field={bookLanguageField}
+            value={bookLanguageValue}
+            onChange={(value) => setUserSettings((prev) => ({ ...prev, BOOK_LANGUAGE: value }))}
+            disabled={Boolean(bookLanguageField.fromEnv)}
+          />
+        </FieldWrapper>
+      )}
+
      {effectiveSearchMode === 'universal' && canOverrideMetadataProvider && (
        <FieldWrapper
          field={metadataProviderField}
@@ -31,6 +31,33 @@ export const toNormalizedLowercaseTextValue = (value: unknown): string => {
  return toTrimmedTextValue(value).toLowerCase();
 };

+const toStringListValue = (value: unknown): string[] => {
+  if (!Array.isArray(value)) {
+    return [];
+  }
+  return value.map((entry) => toTrimmedTextValue(entry)).filter((entry) => entry.length > 0);
+};
+
+/**
+ * Resolve a list-valued per-user override against its global value.
+ *
+ * A key absent from userSettings, or set to null, is not an override. A stored list
+ * that matches the global one is treated as inherited, matching how
+ * buildUserSettingsPayload clears it on save.
+ */
+export const resolveListOverride = (
+  userValue: unknown,
+  globalValue: unknown,
+  hasUserKey: boolean,
+): { value: string[]; isOverridden: boolean } => {
+  const globalList = toStringListValue(globalValue);
+  const userList = toStringListValue(userValue);
+  const isOverridden =
+    hasUserKey && userValue !== null && JSON.stringify(userList) !== JSON.stringify(globalList);
+
+  return { value: isOverridden ? userList : globalList, isOverridden };
+};
+
 export const toComparableValue = (value: unknown): string => {
  if (value === null || value === undefined) {
    return '';
@@ -10,6 +10,7 @@ export interface PerUserSettings {
  EMAIL_RECIPIENT?: string;
  DOWNLOAD_TO_BROWSER_CONTENT_TYPES?: string[];
  SEARCH_MODE?: string;
+  BOOK_LANGUAGE?: string[];
  METADATA_PROVIDER?: string;
  METADATA_PROVIDER_AUDIOBOOK?: string;
  DEFAULT_RELEASE_SOURCE?: string;
@@ -139,6 +139,10 @@ export function Tooltip({
    }

    if (deltaX !== 0 || deltaY !== 0) {
+      // Genuine measure-and-adjust: the tooltip must be laid out before we know
+      // whether it overflows the viewport. The loop converges in one pass because
+      // the corrected position yields deltaX/deltaY of 0 on the next run.
+      // oxlint-disable-next-line react/set-state-in-effect
      setCoords((current) => {
        if (!current) {
          return current;
@@ -1,8 +1,8 @@
-import { useRef } from 'react';
+import { useEffectEvent } from 'react';
 import type { Dispatch, SetStateAction } from 'react';

 import type { Book } from '../../types';
-import { onBookTargetChange } from '../../utils/bookTargetEvents';
+import { onBookTargetChange, type BookTargetChangeEvent } from '../../utils/bookTargetEvents';
 import { useMountEffect } from '../useMountEffect';

 interface UseBookTargetDeselectSyncOptions {
@@ -14,15 +14,14 @@ export const useBookTargetDeselectSync = ({
  activeListValue,
  setBooks,
 }: UseBookTargetDeselectSyncOptions): void => {
-  const activeListValueRef = useRef(activeListValue);
-  activeListValueRef.current = activeListValue;
-
-  useMountEffect(() => {
-    return onBookTargetChange((event) => {
-      if (event.selected) return;
-      const currentValue = activeListValueRef.current;
-      if (!currentValue || String(currentValue) !== event.target) return;
-      setBooks((prev) => prev.filter((book) => book.provider_id !== event.bookId));
-    });
+  const handleTargetChange = useEffectEvent((event: BookTargetChangeEvent) => {
+    if (event.selected) return;
+    if (!activeListValue || String(activeListValue) !== event.target) return;
+    setBooks((prev) => prev.filter((book) => book.provider_id !== event.bookId));
  });
+
+  // Wrapped rather than handed over directly: an Effect Event must not be given to
+  // something that stores it, and `onBookTargetChange` puts its argument in a
+  // module-level listener set. Same shape as useDismiss.
+  useMountEffect(() => onBookTargetChange((event) => handleTargetChange(event)));
 };
@@ -1,11 +1,17 @@
-import { useCallback, useRef, useState } from 'react';
+import { useCallback, useState } from 'react';
 import type { Dispatch, SetStateAction } from 'react';

 import type { ContentType } from '../../types';
+import { useDependencyEffect } from '../useMountEffect';

 const CONTENT_TYPE_STORAGE_KEY = 'preferred-content-type';

-const readInitialPreference = (): { contentType: ContentType; combinedMode: boolean } => {
+interface ContentTypePreference {
+  contentType: ContentType;
+  combinedMode: boolean;
+}
+
+const readInitialPreference = (): ContentTypePreference => {
  try {
    const saved = localStorage.getItem(CONTENT_TYPE_STORAGE_KEY);
    if (saved === 'combined') {
@@ -26,53 +32,32 @@ export const useContentTypePreferences = (): {
  combinedMode: boolean;
  setCombinedMode: Dispatch<SetStateAction<boolean>>;
 } => {
-  const initialPreference = readInitialPreference();
-  const [contentType, setContentTypeState] = useState<ContentType>(
-    () => initialPreference.contentType,
-  );
-  const [combinedMode, setCombinedModeState] = useState<boolean>(
-    () => initialPreference.combinedMode,
-  );
-  const contentTypeRef = useRef(contentType);
-  const combinedModeRef = useRef(combinedMode);
-  contentTypeRef.current = contentType;
-  combinedModeRef.current = combinedMode;
+  // Both values live in one state object so each setter can derive the other
+  // from a pure updater instead of mirroring it into a ref during render.
+  const [preference, setPreference] = useState<ContentTypePreference>(readInitialPreference);
+  const { contentType, combinedMode } = preference;

-  const persistPreference = useCallback(
-    (nextContentType: ContentType, nextCombinedMode: boolean) => {
-      try {
-        localStorage.setItem(
-          CONTENT_TYPE_STORAGE_KEY,
-          nextCombinedMode ? 'combined' : nextContentType,
-        );
-      } catch {
-        // localStorage may be unavailable in private browsing
-      }
-    },
-    [],
-  );
+  const setContentType: Dispatch<SetStateAction<ContentType>> = useCallback((value) => {
+    setPreference((current) => ({
+      ...current,
+      contentType: typeof value === 'function' ? value(current.contentType) : value,
+    }));
+  }, []);

-  const setContentType: Dispatch<SetStateAction<ContentType>> = useCallback(
-    (value) => {
-      setContentTypeState((current) => {
-        const nextContentType = typeof value === 'function' ? value(current) : value;
-        persistPreference(nextContentType, combinedModeRef.current);
-        return nextContentType;
-      });
-    },
-    [persistPreference],
-  );
+  const setCombinedMode: Dispatch<SetStateAction<boolean>> = useCallback((value) => {
+    setPreference((current) => ({
+      ...current,
+      combinedMode: typeof value === 'function' ? value(current.combinedMode) : value,
+    }));
+  }, []);

-  const setCombinedMode: Dispatch<SetStateAction<boolean>> = useCallback(
-    (value) => {
-      setCombinedModeState((current) => {
-        const nextCombinedMode = typeof value === 'function' ? value(current) : value;
-        persistPreference(contentTypeRef.current, nextCombinedMode);
-        return nextCombinedMode;
-      });
-    },
-    [persistPreference],
-  );
+  useDependencyEffect(() => {
+    try {
+      localStorage.setItem(CONTENT_TYPE_STORAGE_KEY, combinedMode ? 'combined' : contentType);
+    } catch {
+      // localStorage may be unavailable in private browsing
+    }
+  }, [contentType, combinedMode]);

  return {
    contentType,
@@ -44,6 +44,10 @@ export function useDescriptionOverflow({
    return () => {
      observer.disconnect();
    };
+    // `descriptionKey` is never read here - it is the trigger. When the modal swaps to a
+    // different release the text changes under the same element, and the overflow has to
+    // be measured again; drop it and the clamp keeps the previous release's answer.
+    // oxlint-disable-next-line react/exhaustive-effect-dependencies
  }, [descriptionExpanded, descriptionKey, descriptionRef]);

  return descriptionOverflows;
@@ -164,7 +164,6 @@ export function useReleaseSearchSession(
  const lastStatusTimeRef = useRef(0);
  const pendingStatusRef = useRef<SearchStatusData | null>(null);
  const statusTimeoutRef = useRef<NodeJS.Timeout | null>(null);
-  activeTabRef.current = activeTab;

  const allTabs = useMemo(() => {
    return buildReleaseTabs(
@@ -363,7 +362,6 @@ export function useReleaseSearchSession(
    indexerFilterInitializedRef.current = new Set<string>();
    const nextInitialActiveTab = preferredDefaultReleaseSource || '';
    initialActiveTabRef.current = nextInitialActiveTab;
-    activeTabRef.current = nextInitialActiveTab;
    pendingStatusRef.current = null;
    lastStatusTimeRef.current = 0;
    if (statusTimeoutRef.current) {
@@ -381,6 +379,7 @@ export function useReleaseSearchSession(
      ? nextInitialActiveTab
      : (tabs[0]?.name ?? '');

+    activeTabRef.current = nextActiveTab;
    setActiveTabState(nextActiveTab);
    setReleasesBySource({});
    setLoadingBySource({});
@@ -458,6 +457,7 @@ export function useReleaseSearchSession(

  const setActiveTab = useCallback(
    (tabName: string) => {
+      activeTabRef.current = tabName;
      setActiveTabState(tabName);

      if (!tabName) {
@@ -1,25 +0,0 @@
-import { useCallback, useRef } from 'react';
-
-import { useMountEffect } from '@/hooks/useMountEffect';
-
-export const useSearchBarHoverTimeout = () => {
-  const hoverTimeoutRef = useRef<ReturnType<typeof setTimeout> | null>(null);
-
-  const clearHoverTimeout = useCallback(() => {
-    if (hoverTimeoutRef.current) {
-      clearTimeout(hoverTimeoutRef.current);
-      hoverTimeoutRef.current = null;
-    }
-  }, []);
-
-  useMountEffect(() => {
-    return () => {
-      clearHoverTimeout();
-    };
-  });
-
-  return {
-    hoverTimeoutRef,
-    clearHoverTimeout,
-  };
-};
@@ -5,24 +5,30 @@ interface TabIndicatorStyle {
  width: number;
 }

+// One shared instance, so the no-active-tab path below can set it repeatedly and React
+// bails out on reference equality instead of re-rendering on every resize event.
+const HIDDEN_INDICATOR: TabIndicatorStyle = { left: 0, width: 0 };
+
 export function useTabIndicator(
  tabRefs: MutableRefObject<Record<string, HTMLButtonElement | null>>,
  activeTab: string,
  tabsDependency: unknown,
 ): TabIndicatorStyle {
-  const [tabIndicatorStyle, setTabIndicatorStyle] = useState({
-    left: 0,
-    width: 0,
-  });
+  const [tabIndicatorStyle, setTabIndicatorStyle] = useState<TabIndicatorStyle>(HIDDEN_INDICATOR);

  useLayoutEffect(() => {
-    const activeButton = tabRefs.current[activeTab];
-    if (!activeButton) {
-      setTabIndicatorStyle({ left: 0, width: 0 });
-      return undefined;
-    }
-
+    // Single measurement path, so a resize that removes the active tab also
+    // resets the indicator instead of leaving it stranded.
    const updateIndicator = () => {
+      const activeButton = tabRefs.current[activeTab];
+      if (!activeButton) {
+        // The shared constant, not a fresh literal: this path now runs on every resize
+        // event, and a new object would never be Object.is-equal to the current state,
+        // so React would re-render on every frame of a window drag for an unchanged value.
+        setTabIndicatorStyle(HIDDEN_INDICATOR);
+        return;
+      }
+
      const containerRect = activeButton.parentElement?.getBoundingClientRect();
      const buttonRect = activeButton.getBoundingClientRect();
      if (!containerRect) {
@@ -41,6 +47,10 @@ export function useTabIndicator(
    return () => {
      window.removeEventListener('resize', updateIndicator);
    };
+    // `tabsDependency` is never read here - it exists only to re-run the measurement when
+    // the tab set changes (callers pass `allTabs` / `showRequestsTab`). The buttons move
+    // when tabs are added or removed, so without it the indicator sits under the old one.
+    // oxlint-disable-next-line react/exhaustive-effect-dependencies
  }, [activeTab, tabRefs, tabsDependency]);

  return tabIndicatorStyle;
@@ -1,47 +1,43 @@
-import { useEffect, useEffectEvent, useRef, type RefObject } from 'react';
+import { useEffect, useEffectEvent, type RefObject } from 'react';

 export const useDismiss = (
  isOpen: boolean,
  refs: RefObject<HTMLElement | null>[],
  onClose: () => void,
 ) => {
-  const handleClose = useEffectEvent(() => {
+  const handlePointerDown = useEffectEvent((event: MouseEvent) => {
+    const target = event.target;
+    if (!(target instanceof Node)) {
+      return;
+    }
+
+    if (refs.some((ref) => ref.current?.contains(target))) {
+      return;
+    }
+
    onClose();
  });

-  const refsRef = useRef(refs);
-  refsRef.current = refs;
+  const handleEscape = useEffectEvent((event: KeyboardEvent) => {
+    if (event.key === 'Escape') {
+      onClose();
+    }
+  });

  useEffect(() => {
    if (!isOpen) {
      return undefined;
    }

-    const handleClickOutside = (event: MouseEvent) => {
-      const target = event.target;
-      if (!(target instanceof Node)) {
-        return;
-      }
-
-      if (refsRef.current.some((ref) => ref.current?.contains(target))) {
-        return;
-      }
-
-      handleClose();
-    };
-
-    const handleEscape = (event: KeyboardEvent) => {
-      if (event.key === 'Escape') {
-        handleClose();
-      }
-    };
+    const handleClickOutside = (event: MouseEvent) => handlePointerDown(event);
+    const handleKeyDown = (event: KeyboardEvent) => handleEscape(event);

    document.addEventListener('mousedown', handleClickOutside);
-    document.addEventListener('keydown', handleEscape);
+    document.addEventListener('keydown', handleKeyDown);

    return () => {
      document.removeEventListener('mousedown', handleClickOutside);
-      document.removeEventListener('keydown', handleEscape);
+      document.removeEventListener('keydown', handleKeyDown);
    };
  }, [isOpen]);
 };
@@ -0,0 +1,31 @@
+import { useCallback, useLayoutEffect, useRef } from 'react';
+
+/**
+ * A callback with a stable identity that always runs the latest render's implementation.
+ *
+ * `useEffectEvent` is React's answer to this shape, but its contract is narrower than it
+ * looks: an Effect Event may only be called from inside an Effect, and must not be handed
+ * to another component, stored in state, or registered with something that outlives the
+ * Effect. Handlers that run from a DOM event, from an async continuation, or from a parent
+ * holding the function in its own UI state are all outside that contract - React documents
+ * the behaviour there as undefined, and the React Compiler advisories oxlint reports
+ * ("existing memoization could not be preserved") are the same fact from the other side.
+ *
+ * So this is the supported shape for those callers. The ref is published in a layout
+ * effect - after commit, before paint - rather than assigned during render, so a render
+ * React later throws away cannot leak its closure into a handler, and no event can
+ * observe the gap.
+ *
+ * Use `useEffectEvent` when the caller really is an Effect; use this everywhere else.
+ */
+export function useLatestCallback<Args extends unknown[], Result>(
+  callback: (...args: Args) => Result,
+): (...args: Args) => Result {
+  const callbackRef = useRef(callback);
+
+  useLayoutEffect(() => {
+    callbackRef.current = callback;
+  });
+
+  return useCallback((...args: Args) => callbackRef.current(...args), []);
+}
@@ -1,16 +1,14 @@
-import { useEffect, useRef, type DependencyList, type EffectCallback } from 'react';
+import { useEffect, useEffectEvent, type DependencyList, type EffectCallback } from 'react';

 export function useMountEffect(effect: EffectCallback): void {
-  const effectRef = useRef(effect);
-  effectRef.current = effect;
+  const runEffect = useEffectEvent(effect);

-  useEffect(() => effectRef.current(), []);
+  useEffect(() => runEffect(), []);
 }

 export function useDependencyEffect(effect: EffectCallback, deps: DependencyList): void {
-  const effectRef = useRef(effect);
-  effectRef.current = effect;
+  const runEffect = useEffectEvent(effect);

  // eslint-disable-next-line react-hooks/exhaustive-deps
-  useEffect(() => effectRef.current(), deps);
+  useEffect(() => runEffect(), deps);
 }
@@ -5,6 +5,7 @@ import { DEFAULT_SUPPORTED_FORMATS } from '../data/languages';
 import { searchBooks, searchMetadata, AuthenticationError } from '../services/api';
 import type { Book, AppConfig, AdvancedFilterState, ContentType, SearchMode } from '../types';
 import { LANGUAGE_OPTION_DEFAULT } from '../utils/languageFilters';
+import { describeSearchFailure } from '../utils/searchFailureMessage';

 const DEFAULT_FORMAT_SELECTION = DEFAULT_SUPPORTED_FORMATS;

@@ -263,12 +264,7 @@ export function useSearch(options: UseSearchOptions): UseSearchReturn {
          handleSearchError(error, 'Search failed');
        } else {
          console.error('Search failed:', error);
-          const message = error instanceof Error ? error.message : 'Search failed';
-          const friendly =
-            message.includes('Network restricted') || message.includes('Unable to reach')
-              ? message
-              : 'Unable to reach download source. Network may be restricted or mirrors blocked.';
-          showToast(friendly, 'error');
+          showToast(describeSearchFailure(error), 'error');
        }
      } finally {
        setIsSearching(false);
@@ -1,4 +1,4 @@
-import { useState, useCallback, useRef } from 'react';
+import { useState, useCallback } from 'react';

 import { getSettings, updateSettings, executeSettingsAction } from '../services/api';
 import type {
@@ -23,6 +23,7 @@ import {
  setThemePreference,
  THEME_FIELD,
 } from '../utils/themePreference';
+import { useLatestCallback } from './useLatestCallback';
 import { useMountEffect } from './useMountEffect';

 interface FetchSettingsOptions {
@@ -137,13 +138,9 @@ export function useSettings(): UseSettingsReturn {
    () => initialState?.originalValues ?? {},
  );
  const [isSaving, setIsSaving] = useState(false);
-  const valuesRef = useRef<SettingsValues>({});
-  const originalValuesRef = useRef<SettingsValues>({});
-
-  valuesRef.current = values;
-  originalValuesRef.current = originalValues;
-
-  const applySettingsResponse = useCallback(
+  // Stable identity, latest `values`/`originalValues`. Not an Effect Event: it is called
+  // from async fetch and save continuations, not from an Effect. See useLatestCallback.
+  const applySettingsResponse = useLatestCallback(
    (response: SettingsResponse, options: { preserveDirtyValues?: boolean } = {}) => {
      const { preserveDirtyValues = false } = options;
      cachedSettingsResponse = response;
@@ -156,11 +153,7 @@ export function useSettings(): UseSettingsReturn {
      setError(null);

      const nextValues = preserveDirtyValues
-        ? mergeFetchedSettingsWithDirtyValues(
-            hydratedState.values,
-            valuesRef.current,
-            originalValuesRef.current,
-          )
+        ? mergeFetchedSettingsWithDirtyValues(hydratedState.values, values, originalValues)
        : hydratedState.values;

      setValues(nextValues);
@@ -170,7 +163,6 @@ export function useSettings(): UseSettingsReturn {
        setSelectedTab((current) => current ?? hydratedState.selectedTab);
      }
    },
-    [],
  );

  const fetchSettings = useCallback(
@@ -12,6 +12,8 @@ import type {
  RequestSubmissionResult,
  MetadataProvidersResponse,
  MetadataSearchConfig,
+  PackBook,
+  InspectReleaseResponse,
 } from '../types';
 import type {
  ActionResult,
@@ -82,6 +84,9 @@ type ApiResponseErrorShape = Error & {
  code?: string;
  requiredMode?: string;
  payload?: Record<string, unknown>;
+  // Set only when the server explained itself, so callers can tell a real explanation
+  // apart from the `503 SERVICE UNAVAILABLE` placeholder built from the status line.
+  serverMessage?: string;
 };

 class ApiResponseError extends Error {
@@ -89,6 +94,7 @@ class ApiResponseError extends Error {
  code?: string;
  requiredMode?: string;
  payload?: Record<string, unknown>;
+  serverMessage?: string;

  constructor(
    message: string,
@@ -97,6 +103,7 @@ class ApiResponseError extends Error {
      code?: string;
      requiredMode?: string;
      payload?: Record<string, unknown>;
+      serverMessage?: string;
    },
  ) {
    super(message);
@@ -105,6 +112,7 @@ class ApiResponseError extends Error {
    this.code = params.code;
    this.requiredMode = params.requiredMode;
    this.payload = params.payload;
+    this.serverMessage = params.serverMessage;
  }
 }

@@ -112,6 +120,13 @@ export const isApiResponseError = (error: unknown): error is ApiResponseErrorSha
  return error instanceof ApiResponseError;
 };

+// The client gave up before the server answered. Distinguishable so callers can report
+// the wait rather than guessing at a cause: a search that hits this has told us nothing
+// about the network or the mirrors, and saying it did is what issue #1285 was about.
+export const isTimeoutError = (error: unknown): error is Error => {
+  return error instanceof TimeoutError;
+};
+
 const mapApiErrorToActionResult = (error: unknown): ActionResult | null => {
  if (!isApiResponseError(error) || !error.payload) {
    return null;
@@ -147,7 +162,31 @@ const DEFAULT_TIMEOUT_MS = 30000;
 // Release searches can be long-running: a source behind Cloudflare/DDoS-Guard has
 // to spin up the bypasser and solve the challenge before any results come back,
 // which routinely takes well over the default timeout.
-const SEARCH_TIMEOUT_MS = 180000;
+//
+// The server bounds them itself (RELEASE_SEARCH_TIMEOUT, reported by /api/config) and
+// answers a spent budget with a message naming the real cause. This client abort is only
+// the backstop for a server that never answers at all, so it has to fire *after* the
+// server's own deadline - a fixed 180s here beat the 300s default, so the accurate
+// message was never reachable and raising the setting did nothing. See issue #1285.
+//
+// The margin has to cover what the server still has to do *after* its budget trips, not
+// just the budget itself. The deadline is cooperative: it is handed to the bypasser as a
+// cancel flag, and internal_bypasser._CDP_UNWIND_GRACE_SECONDS allows 15s on its own for a
+// cancelled solve to close its browser - before the handler has serialized releases, built
+// the column config and put bytes on the wire. A 15s margin is entirely spent by that
+// unwind, so give it room for the unwind plus the response.
+const SEARCH_TIMEOUT_MARGIN_MS = 45000;
+const FALLBACK_SEARCH_TIMEOUT_MS = 300000; // search_deadline.DEFAULT_SEARCH_BUDGET_SECONDS
+let searchTimeoutMs = FALLBACK_SEARCH_TIMEOUT_MS + SEARCH_TIMEOUT_MARGIN_MS;
+
+// Exported for tests; callers get this applied automatically via getConfig().
+export const setSearchTimeoutFromConfig = (budgetSeconds: unknown): void => {
+  if (typeof budgetSeconds === 'number' && Number.isFinite(budgetSeconds) && budgetSeconds > 0) {
+    searchTimeoutMs = budgetSeconds * 1000 + SEARCH_TIMEOUT_MARGIN_MS;
+  }
+};
+
+export const getSearchTimeoutMs = (): number => searchTimeoutMs;

 // Utility function for JSON fetch with credentials and timeout
 async function fetchJSON<T>(
@@ -181,12 +220,15 @@ async function fetchJSON<T>(
        if (isRecord(parsed) && !Array.isArray(parsed)) {
          errorData = parsed;
        }
-        // Prefer user-friendly 'message' field, fall back to 'error'
-        if (typeof errorData?.message === 'string') {
-          errorMessage = errorData.message;
-          hasServerMessage = true;
-        } else if (typeof errorData?.error === 'string') {
-          errorMessage = errorData.error;
+        // Prefer user-friendly 'message' field, fall back to 'error'. Both must carry
+        // actual text: an empty string is not the server explaining itself, and treating
+        // it as one suppresses the placeholder below and shows the user a blank toast.
+        const explanation = [errorData?.message, errorData?.error].find(
+          (candidate): candidate is string =>
+            typeof candidate === 'string' && candidate.trim() !== '',
+        );
+        if (explanation !== undefined) {
+          errorMessage = explanation;
          hasServerMessage = true;
        }
      } catch (e) {
@@ -211,6 +253,7 @@ async function fetchJSON<T>(

      throw new ApiResponseError(errorMessage, {
        status: res.status,
+        serverMessage: hasServerMessage ? errorMessage : undefined,
        code: typeof errorData?.code === 'string' ? errorData.code : undefined,
        requiredMode:
          typeof errorData?.required_mode === 'string' ? errorData.required_mode : undefined,
@@ -241,7 +284,7 @@ export const searchBooks = async (query: string): Promise<Book[]> => {
  const response = await fetchJSON<ReleasesResponse>(
    `${API_BASE}/releases?source=direct_download&${query}`,
    {},
-    SEARCH_TIMEOUT_MS,
+    searchTimeoutMs,
  );
  return response.releases.map(transformReleaseToDirectBook);
 };
@@ -510,6 +553,18 @@ export type DownloadReleasePayload = {
  language?: string; // Release language code, for the {Language} naming variable
  search_author?: string;
  search_mode?: 'direct' | 'universal';
+  multi_book?: boolean; // Split a multi-book pack into one book per subfolder/file
+  book_plan?: PackBook[]; // The split the user approved before download
+};
+
+/** Inspect a release's file list before download (same body as downloadRelease). */
+export const inspectRelease = async (
+  release: DownloadReleasePayload,
+): Promise<InspectReleaseResponse> => {
+  return fetchJSON<InspectReleaseResponse>(`${API_BASE}/releases/inspect`, {
+    method: 'POST',
+    body: JSON.stringify(release),
+  });
 };

 export const downloadRelease = async (
@@ -572,7 +627,9 @@ export const retryDownload = async (id: string): Promise<void> => {
 };

 export const getConfig = async (): Promise<AppConfig> => {
-  return fetchJSON<AppConfig>(API.config);
+  const config = await fetchJSON<AppConfig>(API.config);
+  setSearchTimeoutFromConfig(config.release_search_timeout);
+  return config;
 };

 interface ActivityDismissedItem {
@@ -7,6 +7,7 @@ import {
  buildLanguageNormalizer,
  getReleaseSearchLanguageParams,
  releaseLanguageMatchesFilter,
+  resolveDefaultLanguageCodes,
 } from '../utils/languageFilters';

 const supportedLanguages: Language[] = [
@@ -67,3 +68,36 @@ describe('releaseLanguageMatchesFilter', () => {
    expect(visibleLanguages).toHaveLength(48);
  });
 });
+
+describe('resolveDefaultLanguageCodes', () => {
+  it('keeps an explicitly empty default as "no default filter"', () => {
+    // The backend stores [] to mean "do not filter"; substituting the first
+    // supported language here would make the UI filter where the server does not.
+    expect(resolveDefaultLanguageCodes([], supportedLanguages)).toEqual([]);
+  });
+
+  it('leaves a configured default untouched', () => {
+    expect(resolveDefaultLanguageCodes(['de', 'hu'], supportedLanguages)).toEqual(['de', 'hu']);
+  });
+
+  it('falls back to the first supported language only when nothing is configured', () => {
+    expect(resolveDefaultLanguageCodes(undefined, supportedLanguages)).toEqual(['en']);
+    expect(resolveDefaultLanguageCodes(null, [])).toEqual(['en']);
+  });
+
+  it('sends no language filter when an empty default is the whole selection', () => {
+    const defaults = resolveDefaultLanguageCodes([], supportedLanguages);
+
+    expect(
+      getReleaseSearchLanguageParams([LANGUAGE_OPTION_DEFAULT], supportedLanguages, defaults),
+    ).toBe(undefined);
+  });
+
+  it('does not smuggle the first language into a Default+German selection', () => {
+    const defaults = resolveDefaultLanguageCodes([], supportedLanguages);
+
+    expect(
+      getReleaseSearchLanguageParams([LANGUAGE_OPTION_DEFAULT, 'de'], supportedLanguages, defaults),
+    ).toEqual(['de']);
+  });
+});
@@ -0,0 +1,53 @@
+import { describe, it, expect } from 'vitest';
+
+import type { PackBook } from '../types';
+import {
+  describePackPlan,
+  parseSeriesPositionInput,
+  toBookPlanPayload,
+  updateReviewBook,
+} from '../utils/packReview';
+
+const books: PackBook[] = [
+  { title: 'Leviathan Wakes', series_position: 1, year: 2011, files: ['a.m4b'] },
+  { title: 'Caliban’s War', series_position: 2, year: 2012, files: ['b.m4b', 'b2.m4b'] },
+];
+
+describe('packReview.updateReviewBook', () => {
+  it('replaces one book without touching the others', () => {
+    const next = updateReviewBook(books, 1, { title: 'Caliban’s War (Unabridged)' });
+    expect(next[0]).toBe(books[0]);
+    expect(next[1]).toEqual({ ...books[1], title: 'Caliban’s War (Unabridged)' });
+    expect(books[1].title).toBe('Caliban’s War');
+  });
+});
+
+describe('packReview.parseSeriesPositionInput', () => {
+  it('accepts whole and fractional positions', () => {
+    expect(parseSeriesPositionInput('3')).toBe(3);
+    expect(parseSeriesPositionInput('2.5')).toBe(2.5);
+  });
+
+  it('treats blank or junk as no position', () => {
+    expect(parseSeriesPositionInput('')).toBeNull();
+    expect(parseSeriesPositionInput('abc')).toBeNull();
+  });
+});
+
+describe('packReview.toBookPlanPayload', () => {
+  it('trims titles, drops books without a title, and keeps file lists', () => {
+    const edited = updateReviewBook(books, 0, { title: '   ' });
+    expect(toBookPlanPayload(edited)).toEqual([
+      { title: 'Caliban’s War', series_position: 2, year: 2012, files: ['b.m4b', 'b2.m4b'] },
+    ]);
+  });
+});
+
+describe('packReview.describePackPlan', () => {
+  it('summarises books, files and ignored sidecars', () => {
+    expect(describePackPlan(books, ['a.txt', 'cover.jpg'])).toBe(
+      '2 books · 3 files · 2 files ignored',
+    );
+    expect(describePackPlan([books[0]], [])).toBe('1 book · 1 file');
+  });
+});
@@ -1,7 +1,7 @@
 import { describe, it, expect } from 'vitest';

 import type { Release } from '../types';
-import { getReleaseFormats } from '../utils/releaseFormats';
+import { getReleaseFormats, getUnrecognizedReleaseFormats } from '../utils/releaseFormats';

 function buildRelease(overrides: Partial<Release>): Release {
  return {
@@ -38,3 +38,26 @@ describe('releaseFormats.getReleaseFormats', () => {
    expect(getReleaseFormats(release)).toEqual(['pdf']);
  });
 });
+
+describe('releaseFormats.getUnrecognizedReleaseFormats', () => {
+  it('returns normalized, deduplicated unrecognized formats from extra', () => {
+    const release = buildRelease({
+      extra: { unrecognized_formats: ['AVI', ' avi ', 'WEBM'] },
+    });
+
+    expect(getUnrecognizedReleaseFormats(release)).toEqual(['avi', 'webm']);
+  });
+
+  it('accepts a single string value', () => {
+    const release = buildRelease({ extra: { unrecognized_formats: 'AVI' } });
+
+    expect(getUnrecognizedReleaseFormats(release)).toEqual(['avi']);
+  });
+
+  it('returns an empty list when nothing was flagged', () => {
+    expect(getUnrecognizedReleaseFormats(buildRelease({}))).toEqual([]);
+    expect(
+      getUnrecognizedReleaseFormats(buildRelease({ extra: { unrecognized_formats: null } })),
+    ).toEqual([]);
+  });
+});
@@ -0,0 +1,62 @@
+import { describe, it, expect } from 'vitest';
+
+import {
+  INITIAL_ENTER_ANIMATION,
+  nextEnterAnimation,
+  type EnterAnimationState,
+} from '../utils/releaseModalEnterAnimation';
+
+describe('nextEnterAnimation', () => {
+  it('animates the first session', () => {
+    const next = nextEnterAnimation(INITIAL_ENTER_ANIMATION, 'book-1', false);
+    expect(next).toEqual({ key: 'book-1', animate: true });
+  });
+
+  it('animates the first session in combined mode too', () => {
+    const next = nextEnterAnimation(INITIAL_ENTER_ANIMATION, 'book-1', true);
+    expect(next).toEqual({ key: 'book-1', animate: true });
+  });
+
+  it('does not animate a step transition between combined-mode sessions', () => {
+    const current: EnterAnimationState = { key: 'book-1', animate: true };
+    expect(nextEnterAnimation(current, 'book-2', true)).toEqual({
+      key: 'book-2',
+      animate: false,
+    });
+  });
+
+  it('animates a session swap outside combined mode', () => {
+    const current: EnterAnimationState = { key: 'book-1', animate: true };
+    expect(nextEnterAnimation(current, 'book-2', false)).toEqual({
+      key: 'book-2',
+      animate: true,
+    });
+  });
+
+  it('holds the decision across re-renders of the same session', () => {
+    // Regression: the decision used to flip back to `true` on the next render,
+    // replaying the enter animation mid-session.
+    const stepped = nextEnterAnimation({ key: 'book-1', animate: true }, 'book-2', true);
+    expect(stepped.animate).toBe(false);
+
+    let state = stepped;
+    for (let i = 0; i < 5; i++) {
+      state = nextEnterAnimation(state, 'book-2', true);
+      expect(state.animate).toBe(false);
+    }
+  });
+
+  it('returns the same reference when the session is unchanged', () => {
+    const current: EnterAnimationState = { key: 'book-1', animate: false };
+    expect(nextEnterAnimation(current, 'book-1', true)).toBe(current);
+  });
+
+  it('animates again after the modal closes and reopens', () => {
+    const open = nextEnterAnimation(INITIAL_ENTER_ANIMATION, 'book-1', true);
+    const closed = nextEnterAnimation(open, null, true);
+    expect(closed.key).toBeNull();
+
+    const reopened = nextEnterAnimation(closed, 'book-1', true);
+    expect(reopened.animate).toBe(true);
+  });
+});
@@ -0,0 +1,69 @@
+import { describe, it, expect } from 'vitest';
+
+import type { Book, Release } from '../types';
+import { buildReleaseDownloadPayload } from '../utils/releasePayload';
+
+const book: Book = {
+  id: 'hc-1',
+  title: 'Drive',
+  author: 'James S. A. Corey',
+  year: '2012',
+  preview: 'https://img/drive.jpg',
+  series_name: 'The Expanse',
+  series_position: 2.6,
+  subtitle: 'An Expanse Short Story',
+  provider: 'hardcover',
+  provider_id: 'hc-1',
+  source: 'direct_download',
+};
+
+const release: Release = {
+  source: 'audiobookbay',
+  source_id: 'abb-1',
+  title: 'James S. A. Corey - The Expanse Complete 2.0',
+  format: 'm4b',
+  language: 'en',
+  download_url: 'https://audiobookbay.lu/abss/expanse/',
+};
+
+describe('buildReleaseDownloadPayload', () => {
+  it('describes the searched book and the chosen release', () => {
+    const payload = buildReleaseDownloadPayload(book, release, 'audiobook');
+    expect(payload).toMatchObject({
+      source: 'audiobookbay',
+      source_id: 'abb-1',
+      title: 'Drive',
+      author: 'James S. A. Corey',
+      series_name: 'The Expanse',
+      series_position: 2.6,
+      language: 'en',
+      content_type: 'audiobook',
+    });
+    expect(payload.multi_book).toBeUndefined();
+    expect(payload.book_plan).toBeUndefined();
+  });
+
+  it('uses the release title and author for manual books', () => {
+    const manual: Book = { ...book, provider: 'manual', title: 'ignored' };
+    const withAuthor = { ...release, extra: { author: 'Release Author' } };
+    const payload = buildReleaseDownloadPayload(manual, withAuthor, 'audiobook');
+    expect(payload.title).toBe(release.title);
+    expect(payload.author).toBe('Release Author');
+  });
+
+  it('flags a manual multi-book pack', () => {
+    const payload = buildReleaseDownloadPayload(book, release, 'audiobook', { multiBook: true });
+    expect(payload.multi_book).toBe(true);
+    expect(payload.book_plan).toBeUndefined();
+  });
+
+  it('attaches the approved book plan', () => {
+    const plan = [{ title: 'Leviathan Wakes', series_position: 1, year: 2011, files: ['a.m4b'] }];
+    const payload = buildReleaseDownloadPayload(book, release, 'audiobook', {
+      multiBook: true,
+      bookPlan: plan,
+    });
+    expect(payload.multi_book).toBe(true);
+    expect(payload.book_plan).toEqual(plan);
+  });
+});
@@ -0,0 +1,77 @@
+import { describe, it, expect, vi, afterEach } from 'vitest';
+
+import { searchBooks } from '../services/api';
+import {
+  describeSearchFailure,
+  CLIENT_TIMEOUT_MESSAGE,
+  UNREACHABLE_SOURCE_MESSAGE,
+} from '../utils/searchFailureMessage';
+
+/**
+ * What a failed direct-mode search tells the user.
+ *
+ * Every non-auth failure used to be relabelled "Unable to reach download source. Network
+ * may be restricted or mirrors blocked.", which threw away the server's explanation and
+ * blamed the user's network for a protection challenge. See issue #1285.
+ */
+
+const jsonResponse = (body: unknown, status: number): Response =>
+  new Response(JSON.stringify(body), {
+    status,
+    statusText: 'SERVICE UNAVAILABLE',
+    headers: { 'Content-Type': 'application/json' },
+  });
+
+/** Drive a real searchBooks() failure so the error is the one the hook actually sees. */
+const failedSearch = async (respond: () => Promise<Response>): Promise<unknown> => {
+  vi.stubGlobal('fetch', vi.fn(respond));
+  return searchBooks('q=dune').catch((error: unknown) => error);
+};
+
+describe('describeSearchFailure', () => {
+  afterEach(() => {
+    vi.unstubAllGlobals();
+  });
+
+  it('shows the sentence the server sent', async () => {
+    const sentence = 'The release search ran out of time (300s).';
+    const error = await failedSearch(() => Promise.resolve(jsonResponse({ error: sentence }, 503)));
+
+    expect(describeSearchFailure(error)).toBe(sentence);
+  });
+
+  it('names the wait when the client gave up first', async () => {
+    // The client's abort is the backstop for a server that never answered. It tells us
+    // nothing about mirrors or the network, and the old chain reported it as if it did.
+    const abort = Object.assign(new Error('The operation was aborted.'), { name: 'AbortError' });
+    const error = await failedSearch(() => Promise.reject(abort));
+
+    expect(describeSearchFailure(error)).toBe(CLIENT_TIMEOUT_MESSAGE);
+    expect(describeSearchFailure(error)).not.toBe(UNREACHABLE_SOURCE_MESSAGE);
+    expect(describeSearchFailure(error)).not.toContain('mirrors');
+  });
+
+  it('falls back to the mirrors line only when nothing explained itself', async () => {
+    const error = await failedSearch(() => Promise.resolve(jsonResponse({}, 503)));
+
+    expect(describeSearchFailure(error)).toBe(UNREACHABLE_SOURCE_MESSAGE);
+  });
+
+  it('keeps a reachability message that already says the right thing', () => {
+    const error = new Error('Unable to reach download source. Every mirror was quarantined.');
+
+    expect(describeSearchFailure(error)).toBe(error.message);
+  });
+
+  it('never produces an empty sentence from a blank server message', async () => {
+    // `{"message": ""}` is not the server explaining itself. Treating it as one used to
+    // reach showToast('') and render an empty error toast.
+    const error = await failedSearch(() => Promise.resolve(jsonResponse({ message: '' }, 503)));
+
+    expect(describeSearchFailure(error)).toBe(UNREACHABLE_SOURCE_MESSAGE);
+  });
+
+  it('handles a non-Error rejection without inventing detail', () => {
+    expect(describeSearchFailure('something odd')).toBe(UNREACHABLE_SOURCE_MESSAGE);
+  });
+});
@@ -0,0 +1,197 @@
+import { describe, it, expect, vi, afterEach, beforeEach } from 'vitest';
+
+import {
+  getConfig,
+  getSearchTimeoutMs,
+  setSearchTimeoutFromConfig,
+  searchBooks,
+  isApiResponseError,
+} from '../services/api';
+
+/**
+ * The client's abort must fire *after* the server's own search deadline.
+ *
+ * `/api/releases` bounds itself with RELEASE_SEARCH_TIMEOUT and answers a spent budget
+ * with a message naming the real cause. The client aborted at a fixed 180s against a
+ * 300s default, so it always won the race and replaced that message with "Request timed
+ * out. Check your network connection or proxy configuration." Raising the setting had no
+ * visible effect either, the 180s being baked into the hashed bundle. See issue #1285.
+ */
+
+// Must stay ahead of what the server still has to do after its budget trips: the deadline
+// is cooperative, and internal_bypasser._CDP_UNWIND_GRACE_SECONDS alone allows 15s for a
+// cancelled solve to close its browser before the response is even built.
+const MARGIN_MS = 45_000;
+const SERVER_UNWIND_GRACE_MS = 15_000;
+
+const jsonResponse = (body: unknown, status = 200): Response =>
+  new Response(JSON.stringify(body), {
+    status,
+    statusText: status === 200 ? 'OK' : 'SERVICE UNAVAILABLE',
+    headers: { 'Content-Type': 'application/json' },
+  });
+
+const configBody = (releaseSearchTimeout: number): Record<string, unknown> => ({
+  release_search_timeout: releaseSearchTimeout,
+});
+
+describe('release search timeout', () => {
+  beforeEach(() => {
+    setSearchTimeoutFromConfig(300);
+  });
+
+  afterEach(() => {
+    vi.unstubAllGlobals();
+    setSearchTimeoutFromConfig(300);
+  });
+
+  it('defaults behind the server default rather than ahead of it', () => {
+    expect(getSearchTimeoutMs()).toBe(300 * 1000 + MARGIN_MS);
+  });
+
+  it('follows the budget the server reports', async () => {
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(() => Promise.resolve(jsonResponse(configBody(900)))),
+    );
+
+    await getConfig();
+
+    expect(getSearchTimeoutMs()).toBe(900 * 1000 + MARGIN_MS);
+  });
+
+  it('still outlasts the server when the budget is lowered', async () => {
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(() => Promise.resolve(jsonResponse(configBody(30)))),
+    );
+
+    await getConfig();
+
+    // Exact, not a lower bound: `> 30_000` is also satisfied by the 345_000 left over
+    // from the previous budget, so a setter that silently stopped applying the config
+    // would pass it.
+    expect(getSearchTimeoutMs()).toBe(30 * 1000 + MARGIN_MS);
+  });
+
+  it('leaves the server room to unwind a cancelled solve and answer', async () => {
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(() => Promise.resolve(jsonResponse(configBody(300)))),
+    );
+
+    await getConfig();
+
+    // The failure this exists for is a budget spent mid-solve. The server then has to
+    // close a browser before it can serialize anything, so a margin merely equal to that
+    // unwind is entirely spent by it and the client aborts first all over again.
+    expect(getSearchTimeoutMs() - 300 * 1000).toBeGreaterThan(SERVER_UNWIND_GRACE_MS);
+  });
+
+  it('ignores a missing or nonsensical budget instead of disabling the backstop', () => {
+    const before = getSearchTimeoutMs();
+
+    setSearchTimeoutFromConfig(undefined);
+    setSearchTimeoutFromConfig(0);
+    setSearchTimeoutFromConfig(-1);
+    setSearchTimeoutFromConfig('600');
+    setSearchTimeoutFromConfig(Number.NaN);
+
+    expect(getSearchTimeoutMs()).toBe(before);
+  });
+
+  it('applies the derived timeout to the direct_download search', async () => {
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(() => Promise.resolve(jsonResponse(configBody(600)))),
+    );
+    await getConfig();
+
+    const seen: Array<AbortSignal | undefined> = [];
+    vi.stubGlobal(
+      'fetch',
+      vi.fn((_url: string, init: RequestInit) => {
+        seen.push(init.signal ?? undefined);
+        return Promise.resolve(jsonResponse({ releases: [] }));
+      }),
+    );
+
+    await searchBooks('q=dune');
+
+    // The request carries an abort signal, and it is not yet aborted: the point is that
+    // the clock it runs on is the server's, not a constant.
+    expect(seen).toHaveLength(1);
+    expect(seen[0]?.aborted).toBe(false);
+    expect(getSearchTimeoutMs()).toBe(600 * 1000 + MARGIN_MS);
+  });
+});
+
+describe('server-provided failure messages', () => {
+  afterEach(() => {
+    vi.unstubAllGlobals();
+  });
+
+  it('carries the server sentence through instead of a status placeholder', async () => {
+    const sentence =
+      'The release search ran out of time (300s). Anna’s Archive is behind a ' +
+      'protection challenge the bypasser could not solve in that window.';
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(() => Promise.resolve(jsonResponse({ error: sentence }, 503))),
+    );
+
+    const error = await searchBooks('q=dune').catch((e: unknown) => e);
+
+    expect(isApiResponseError(error)).toBe(true);
+    if (isApiResponseError(error)) {
+      expect(error.serverMessage).toBe(sentence);
+    }
+  });
+
+  it('leaves serverMessage unset when the server explained nothing', async () => {
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(() => Promise.resolve(jsonResponse({}, 503))),
+    );
+
+    const error = await searchBooks('q=dune').catch((e: unknown) => e);
+
+    expect(isApiResponseError(error)).toBe(true);
+    if (isApiResponseError(error)) {
+      expect(error.serverMessage).toBeUndefined();
+      // Without this the UI would show a bare "503 SERVICE UNAVAILABLE".
+      expect(error.message).toContain('Server unavailable');
+    }
+  });
+
+  it('treats a blank message as no explanation at all', async () => {
+    // An empty string is not the server explaining itself. Taking it as one suppresses
+    // the placeholder below *and* survives a `??` fallback, leaving an empty toast.
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(() => Promise.resolve(jsonResponse({ message: '   ' }, 503))),
+    );
+
+    const error = await searchBooks('q=dune').catch((e: unknown) => e);
+
+    expect(isApiResponseError(error)).toBe(true);
+    if (isApiResponseError(error)) {
+      expect(error.serverMessage).toBeUndefined();
+      expect(error.message).toContain('Server unavailable');
+    }
+  });
+
+  it('falls back to `error` when `message` is blank', async () => {
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(() => Promise.resolve(jsonResponse({ message: '', error: 'the real reason' }, 503))),
+    );
+
+    const error = await searchBooks('q=dune').catch((e: unknown) => e);
+
+    expect(isApiResponseError(error)).toBe(true);
+    if (isApiResponseError(error)) {
+      expect(error.serverMessage).toBe('the real reason');
+    }
+  });
+});
@@ -285,6 +285,7 @@ export interface AppConfig {
  auto_open_downloads_sidebar: boolean; // Auto-open sidebar when download is queued
  hardcover_auto_remove_on_download: boolean; // Auto-remove from active Hardcover list on download
  download_to_browser_content_types: string[]; // Auto-download completed files to browser for selected content types
+  release_search_timeout: number; // Server-side budget for one release search, in seconds
  settings_enabled: boolean; // Whether config directory is mounted and writable
  onboarding_complete: boolean; // Whether the user has completed initial setup
  default_sort: string; // Default sort for direct mode
@@ -460,6 +461,27 @@ export interface SourceSearchInfo {
 }

 // Response from /api/releases endpoint
+/** One book split out of a multi-book pack release, files as release-relative paths. */
+export interface PackBook {
+  title: string;
+  series_position: number | null;
+  year: number | null;
+  files: string[];
+}
+
+export interface PackPlan {
+  is_pack: boolean;
+  books: PackBook[];
+  ignored: string[];
+}
+
+export interface InspectReleaseResponse {
+  inspected: boolean;
+  reason: string | null;
+  files: { path: string; size: number | null }[];
+  plan: PackPlan | null;
+}
+
 export interface ReleasesResponse {
  releases: Release[];
  book: {
@@ -146,7 +146,13 @@ export interface TableFieldColumnOption {
  childOf?: string;
 }

-export type TableFieldColumnType = 'text' | 'select' | 'multiselect' | 'checkbox' | 'path';
+export type TableFieldColumnType =
+  | 'text'
+  | 'password'
+  | 'select'
+  | 'multiselect'
+  | 'checkbox'
+  | 'path';

 export interface TableFieldColumn {
  key: string;
@@ -1,4 +1,4 @@
-type BookTargetChangeEvent = {
+export type BookTargetChangeEvent = {
  provider: string;
  bookId: string;
  target: string;
@@ -27,6 +27,24 @@ export const normalizeLanguageSelection = (selected: string[]): string[] => {
  return unique.length ? unique : [LANGUAGE_OPTION_DEFAULT];
 };

+/**
+ * Resolve the language codes the "Default" filter option stands for.
+ *
+ * An explicitly empty list is a deliberate "no default filter" and is returned as-is;
+ * only a missing value falls back to the first supported language. Substituting a
+ * language for the empty list would make the UI filter by a language the backend
+ * does not apply.
+ */
+export const resolveDefaultLanguageCodes = (
+  configuredDefault: string[] | null | undefined,
+  supportedLanguages: Language[],
+): string[] => {
+  if (Array.isArray(configuredDefault)) {
+    return configuredDefault;
+  }
+  return [supportedLanguages[0]?.code || 'en'];
+};
+
 export const getLanguageFilterValues = (
  selection: string[],
  supportedLanguages: Language[],
@@ -0,0 +1,39 @@
+import type { PackBook } from '../types';
+
+/** Return a copy of `books` with one entry patched; the input is not mutated. */
+export function updateReviewBook(
+  books: PackBook[],
+  index: number,
+  patch: Partial<PackBook>,
+): PackBook[] {
+  return books.map((book, i) => (i === index ? { ...book, ...patch } : book));
+}
+
+/** Parse a series-position text field: "3" → 3, "2.5" → 2.5, blank/junk → null. */
+export function parseSeriesPositionInput(value: string): number | null {
+  const trimmed = value.trim();
+  if (!trimmed) return null;
+  const parsed = Number(trimmed);
+  return Number.isFinite(parsed) ? parsed : null;
+}
+
+/** The plan sent with the download: trimmed titles, untitled books dropped. */
+export function toBookPlanPayload(books: PackBook[]): PackBook[] {
+  return books
+    .map((book) => ({ ...book, title: book.title.trim() }))
+    .filter((book) => book.title.length > 0 && book.files.length > 0);
+}
+
+function plural(count: number, noun: string): string {
+  return `${count} ${noun}${count === 1 ? '' : 's'}`;
+}
+
+/** "2 books · 3 files · 2 files ignored" */
+export function describePackPlan(books: PackBook[], ignored: string[]): string {
+  const fileCount = books.reduce((sum, book) => sum + book.files.length, 0);
+  const parts = [plural(books.length, 'book'), plural(fileCount, 'file')];
+  if (ignored.length > 0) {
+    parts.push(`${plural(ignored.length, 'file')} ignored`);
+  }
+  return parts.join(' · ');
+}
@@ -33,3 +33,27 @@ export function getReleaseFormats(release: Release): string[] {

  return formats;
 }
+
+/**
+ * Format tokens the indexer declared but the backend could not map to a known
+ * book/audiobook format (e.g. MyAnonamouse "[ENG / AVI]"). Such a release will
+ * download but fail post-processing, so the UI warns instead of showing a bare
+ * content-type icon.
+ */
+export function getUnrecognizedReleaseFormats(release: Release): string[] {
+  const raw = release.extra?.unrecognized_formats;
+  const values = Array.isArray(raw) ? raw : [raw];
+  const formats: string[] = [];
+  const seen = new Set<string>();
+
+  values.forEach((value) => {
+    const normalized = normalizeFormatValue(value);
+    if (!normalized || seen.has(normalized)) {
+      return;
+    }
+    seen.add(normalized);
+    formats.push(normalized);
+  });
+
+  return formats;
+}
@@ -0,0 +1,29 @@
+export interface EnterAnimationState {
+  key: string | null;
+  animate: boolean;
+}
+
+export const INITIAL_ENTER_ANIMATION: EnterAnimationState = { key: null, animate: true };
+
+/**
+ * Decide whether a release modal session should play its enter animation.
+ *
+ * The decision is made once, when a session key first appears, and then held for
+ * that session's lifetime — re-rendering mid-session must not restart the
+ * animation. Swapping between sessions in combined mode is a step transition
+ * rather than an entrance, so it does not animate.
+ */
+export const nextEnterAnimation = (
+  current: EnterAnimationState,
+  sessionKey: string | null,
+  isCombinedMode: boolean,
+): EnterAnimationState => {
+  if (current.key === sessionKey) {
+    return current;
+  }
+
+  return {
+    key: sessionKey,
+    animate: !isCombinedMode || current.key === null,
+  };
+};
@@ -0,0 +1,55 @@
+import type { DownloadReleasePayload } from '../services/api';
+import type { Book, ContentType, PackBook, Release } from '../types';
+
+export interface ReleaseDownloadOptions {
+  /** Ask post-processing to split the release into one book per subfolder/file. */
+  multiBook?: boolean;
+  /** The split the user approved in the pack review panel. */
+  bookPlan?: PackBook[];
+}
+
+/** Build the body for /api/releases/download (and /api/releases/inspect). */
+export function buildReleaseDownloadPayload(
+  book: Book,
+  release: Release,
+  releaseContentType: ContentType,
+  options: ReleaseDownloadOptions = {},
+): DownloadReleasePayload {
+  const isManual = book.provider === 'manual';
+  const releasePreview =
+    typeof release.extra?.preview === 'string' ? release.extra.preview : undefined;
+  const releaseAuthor =
+    typeof release.extra?.author === 'string' ? release.extra.author : undefined;
+
+  const payload: DownloadReleasePayload = {
+    source: release.source,
+    source_id: release.source_id,
+    title: isManual ? release.title : book.title,
+    author: isManual ? releaseAuthor || '' : book.author,
+    year: book.year,
+    format: release.format,
+    size: release.size,
+    size_bytes: release.size_bytes,
+    download_url: release.download_url,
+    protocol: release.protocol,
+    indexer: release.indexer,
+    seeders: release.seeders,
+    extra: release.extra,
+    preview: isManual ? releasePreview || undefined : book.preview,
+    content_type: releaseContentType,
+    series_name: book.series_name,
+    series_position: book.series_position,
+    subtitle: book.subtitle,
+    // From the release, never the book: book.language is the provider's
+    // canonical edition, which would mislabel a translated release.
+    language: release.language ?? undefined,
+  };
+
+  if (options.multiBook || options.bookPlan) {
+    payload.multi_book = true;
+  }
+  if (options.bookPlan) {
+    payload.book_plan = options.bookPlan;
+  }
+  return payload;
+}
@@ -0,0 +1,39 @@
+import { isApiResponseError, isTimeoutError } from '../services/api';
+
+// Shown when we genuinely have nothing better: the request failed without the server
+// saying why, which really can mean blocked mirrors or a restricted network.
+export const UNREACHABLE_SOURCE_MESSAGE =
+  'Unable to reach download source. Network may be restricted or mirrors blocked.';
+
+// Shown when the client's own abort fired. It is the backstop for a server that never
+// answered at all, and it says nothing about mirrors - the budget the client waited out
+// is the server's own, which the user can raise. See issue #1285.
+export const CLIENT_TIMEOUT_MESSAGE =
+  'The search took longer than the server said it would. Raise the release search ' +
+  'timeout if your setup is simply slow.';
+
+/**
+ * The sentence to show for a failed direct-mode search.
+ *
+ * The server knows why a search failed - a spent search budget, an unsolved protection
+ * challenge - and blanket-replacing that with the mirrors line told users their network
+ * was broken when it was not. So: the server's own words when it explained itself, the
+ * timeout line when we gave up before it answered, and the mirrors line only when
+ * neither applies. See issue #1285.
+ */
+export const describeSearchFailure = (error: unknown): string => {
+  if (isTimeoutError(error)) {
+    return CLIENT_TIMEOUT_MESSAGE;
+  }
+
+  if (isApiResponseError(error) && error.serverMessage) {
+    return error.serverMessage;
+  }
+
+  const message = error instanceof Error ? error.message : '';
+  if (message.includes('Network restricted') || message.includes('Unable to reach')) {
+    return message;
+  }
+
+  return UNREACHABLE_SOURCE_MESSAGE;
+};
@@ -0,0 +1,13 @@
+"""AudiobookBay test fixtures."""
+
+import pytest
+
+from shelfmark.release_sources.audiobookbay import scraper
+
+
+@pytest.fixture(autouse=True)
+def _clear_detail_page_cache():
+    """Detail pages are cached briefly in production; tests must not share them."""
+    scraper.clear_detail_page_cache()
+    yield
+    scraper.clear_detail_page_cache()
@@ -0,0 +1,102 @@
+"""Tests for reading the torrent file list off an AudiobookBay detail page."""
+
+from shelfmark.download.postprocess.packs import PackFile
+from shelfmark.release_sources.audiobookbay import scraper
+
+# Trimmed from a real detail page (2026-08): the file rows sit between the
+# "Multifile Torrent" marker and the "Combined File Size" row.
+MULTIFILE_DETAIL_HTML = """
+<table>
+<tr><td>Tracker:</td><td>udp://tracker.torrent.eu.org:451/announce</td></tr>
+<tr><td>Creation Date:</td><td>Sun, 29 Mar 2026 21:09:39 +0200</td></tr>
+<tr><td colspan='2'>This is a Multifile Torrent</td></tr>
+<tr><td colspan='2'>The Expanse 9.0 - Leviathan Falls (2021).m4b 1.05 GBs</td></tr>
+<tr><td colspan='2'>The Expanse 0.1 - An Expanse Novella - Drive (2012).txt 340 Bytes</td></tr>
+<tr><td colspan='2'>The Expanse 0.2 - An Expanse Novella - The Churn (2014).m4b 125.72 MBs</td></tr>
+<tr><td colspan='2'>The Expanse 2.0 - Caliban’s War (2012).m4b 578.97 MBs</td></tr>
+<tr><td>Combined File Size:</td><td><span style='color:#00f;'>7.87</span> GBs</td></tr>
+<tr><td>Info Hash:</td><td>e4a5538e26987ee58a43aa629ec2c4f2b2d46526</td></tr>
+</table>
+"""
+
+SINGLE_FILE_DETAIL_HTML = """
+<table>
+<tr><td>Creation Date:</td><td>Sun, 29 Mar 2026 21:09:39 +0200</td></tr>
+<tr><td colspan='2'>Drive.m4b 41.55 MBs</td></tr>
+<tr><td>Combined File Size:</td><td><span style='color:#00f;'>41.55</span> MBs</td></tr>
+<tr><td>Info Hash:</td><td>e4a5538e26987ee58a43aa629ec2c4f2b2d46526</td></tr>
+</table>
+"""
+
+
+def test_extracts_multifile_rows_with_byte_sizes():
+    files = scraper.extract_file_list(MULTIFILE_DETAIL_HTML)
+    assert files == [
+        PackFile("The Expanse 9.0 - Leviathan Falls (2021).m4b", int(1.05 * 1024**3)),
+        PackFile("The Expanse 0.1 - An Expanse Novella - Drive (2012).txt", 340),
+        PackFile(
+            "The Expanse 0.2 - An Expanse Novella - The Churn (2014).m4b", int(125.72 * 1024**2)
+        ),
+        PackFile("The Expanse 2.0 - Caliban’s War (2012).m4b", int(578.97 * 1024**2)),
+    ]
+
+
+def test_single_file_torrent_lists_the_row_before_combined_size():
+    assert scraper.extract_file_list(SINGLE_FILE_DETAIL_HTML) == [
+        PackFile("Drive.m4b", int(41.55 * 1024**2))
+    ]
+
+
+def test_page_without_file_table_returns_none():
+    assert scraper.extract_file_list("<html><body><p>nothing here</p></body></html>") is None
+
+
+class TestHandlerListFiles:
+    def test_lists_files_from_detail_page(self):
+        from unittest.mock import patch
+
+        from shelfmark.release_sources.audiobookbay.handler import AudiobookBayHandler
+
+        with patch(
+            "shelfmark.release_sources.audiobookbay.handler.scraper.fetch_detail_html",
+            return_value=SINGLE_FILE_DETAIL_HTML,
+        ) as fetch:
+            files = AudiobookBayHandler().list_files(
+                {"source_id": "abc", "download_url": "https://audiobookbay.lu/abss/drive/"}
+            )
+        assert files == [PackFile("Drive.m4b", int(41.55 * 1024**2))]
+        fetch.assert_called_once_with("https://audiobookbay.lu/abss/drive/", "audiobookbay.lu")
+
+    def test_rejects_detail_url_on_other_host(self):
+        from unittest.mock import patch
+
+        from shelfmark.release_sources.audiobookbay.handler import AudiobookBayHandler
+
+        with patch(
+            "shelfmark.release_sources.audiobookbay.handler.scraper.fetch_detail_html"
+        ) as fetch:
+            files = AudiobookBayHandler().list_files(
+                {"source_id": "abc", "download_url": "https://evil.example/abss/drive/"}
+            )
+        assert files is None
+        fetch.assert_not_called()
+
+
+def test_extract_magnet_link_and_file_list_share_one_page_fetch():
+    from unittest.mock import patch
+
+    page = MULTIFILE_DETAIL_HTML
+    with (
+        patch(
+            "shelfmark.release_sources.audiobookbay.scraper.downloader.html_get_page",
+            return_value=page,
+        ) as get_page,
+        patch("shelfmark.release_sources.audiobookbay.scraper._bootstrap_abb_session"),
+    ):
+        scraper.clear_detail_page_cache()
+        url = "https://audiobookbay.lu/abss/expanse/"
+        assert scraper.fetch_detail_html(url, "audiobookbay.lu") == page
+        magnet = scraper.extract_magnet_link(url, "audiobookbay.lu")
+    assert magnet is not None
+    assert "e4a5538e26987ee58a43aa629ec2c4f2b2d46526".upper() in magnet
+    assert get_page.call_count == 1
@@ -0,0 +1,315 @@
+"""How long the internal bypasser is allowed to spend, and on what.
+
+Issue #1276: MAX_RETRY drove *both* the outer page-load loop and the per-page method
+loop, so the default of 10 meant ~40 solve attempts on one browser. That overran the
+worker deadline, and the failure reached the user as `RuntimeError: TimeoutError` - a
+message that says nothing about a protection challenge and sent people looking at their
+reverse proxy instead.
+
+Also covered: the undisturbed window a passive challenge gets before anything touches the
+page. Anna's Archive's DDoS-Guard check has no click target and clears itself; going
+straight to the click/reload methods meant the one thing that solves it was never tried.
+"""
+
+import asyncio
+
+import pytest
+
+
+@pytest.fixture
+def bypass(monkeypatch):
+    """internal_bypasser with sleeps and jitter removed."""
+    import shelfmark.bypass.internal_bypasser as internal_bypasser
+
+    async def _no_sleep(_seconds) -> None:
+        return None
+
+    monkeypatch.setattr(internal_bypasser.asyncio, "sleep", _no_sleep)
+    monkeypatch.setattr(internal_bypasser._RNG, "uniform", lambda _a, _b: 0)
+    return internal_bypasser
+
+
+def _recording_methods(calls: list[str], count: int = 4):
+    def _make(name: str):
+        async def _method(_page) -> bool:
+            calls.append(name)
+            return False
+
+        _method.__name__ = name
+        return _method
+
+    return [_make(f"m{i}") for i in range(count)]
+
+
+def _stub_page_state(monkeypatch, bypass, *, bypassed=False, challenge="ddos_guard"):
+    async def _is_bypassed(*_args, **_kwargs) -> bool:
+        return bypassed
+
+    async def _detect(*_args, **_kwargs) -> str:
+        return challenge
+
+    monkeypatch.setattr(bypass, "_is_bypassed", _is_bypassed)
+    monkeypatch.setattr(bypass, "_detect_challenge_type", _detect)
+
+
+# --------------------------------------------------------------------------- #
+# The method loop must not read MAX_RETRY
+# --------------------------------------------------------------------------- #
+def test_method_loop_budget_is_independent_of_max_retry(monkeypatch, bypass):
+    """MAX_RETRY is the outer page-load retry; reading it here squared the budget.
+
+    Exercised against a challenge whose *type* keeps changing, because that is the case
+    where max_retries is what bounds the loop: the stuck-challenge guard only fires on a
+    run of the same type, so with a stable challenge it hid the real budget entirely.
+    """
+    monkeypatch.setattr(type(bypass.app_config), "MAX_RETRY", 50, raising=False)
+
+    types = iter(["ddos_guard", "cloudflare"] * 100)
+
+    async def _alternating(*_args, **_kwargs) -> str:
+        return next(types)
+
+    calls: list[str] = []
+    monkeypatch.setattr(bypass, "BYPASS_METHODS", _recording_methods(calls))
+    monkeypatch.setattr(bypass, "_wait_for_passive_solve", _never_passes)
+    _stub_page_state(monkeypatch, bypass)
+    monkeypatch.setattr(bypass, "_detect_challenge_type", _alternating)
+
+    assert asyncio.run(bypass._bypass(object())) is False
+    assert len(calls) == bypass._BYPASS_METHOD_ATTEMPTS
+    assert len(calls) < 50, "MAX_RETRY must not reach the method loop"
+
+
+def test_method_attempt_budget_is_reachable(bypass):
+    """The number reported as `attempt N/X` must be a number the loop can reach.
+
+    It used to be MAX_RETRY (10) while the stuck-challenge guard capped the loop at 5,
+    so logs showed `4/10` and stopped, which reads like six lost attempts.
+    """
+    assert bypass._BYPASS_METHOD_ATTEMPTS == len(bypass.BYPASS_METHODS) + 1
+    assert bypass._BYPASS_METHOD_ATTEMPTS >= (
+        max(bypass.MAX_CONSECUTIVE_SAME_CHALLENGE, len(bypass.BYPASS_METHODS) + 1)
+    )
+
+
+# --------------------------------------------------------------------------- #
+# A passive challenge gets an undisturbed window first
+# --------------------------------------------------------------------------- #
+async def _never_passes(*_args, **_kwargs) -> bool:
+    return False
+
+
+def test_passive_challenge_is_given_time_before_any_method_runs(monkeypatch, bypass):
+    """DDoS-Guard's JS check clears itself; nothing should click or reload first."""
+    calls: list[str] = []
+    monkeypatch.setattr(bypass, "BYPASS_METHODS", _recording_methods(calls))
+    _stub_page_state(monkeypatch, bypass)
+
+    async def _passes(*_args, **_kwargs) -> bool:
+        return True
+
+    monkeypatch.setattr(bypass, "_wait_for_passive_solve", _passes)
+
+    assert asyncio.run(bypass._bypass(object())) is True
+    assert calls == [], "the page must not be touched while the check can still pass"
+
+
+def test_passive_wait_happens_once_not_before_every_method(monkeypatch, bypass):
+    """It is a settling window, not a delay bolted onto each attempt."""
+    waits: list[int] = []
+    calls: list[str] = []
+
+    async def _count_wait(*_args, **_kwargs) -> bool:
+        waits.append(1)
+        return False
+
+    monkeypatch.setattr(bypass, "BYPASS_METHODS", _recording_methods(calls))
+    monkeypatch.setattr(bypass, "_wait_for_passive_solve", _count_wait)
+    _stub_page_state(monkeypatch, bypass)
+
+    asyncio.run(bypass._bypass(object()))
+
+    assert len(waits) == 1
+    assert calls == ["m0", "m1", "m2", "m3"]
+
+
+def test_no_passive_wait_when_no_challenge_is_detected(monkeypatch, bypass):
+    """The 'none' branch has its own settle-and-refresh handling."""
+    waits: list[int] = []
+
+    async def _count_wait(*_args, **_kwargs) -> bool:
+        waits.append(1)
+        return False
+
+    monkeypatch.setattr(bypass, "_wait_for_passive_solve", _count_wait)
+    _stub_page_state(monkeypatch, bypass, challenge="none")
+
+    class _Page:
+        async def reload(self, **_kwargs) -> None:
+            return None
+
+    asyncio.run(bypass._bypass(_Page(), max_retries=1))
+
+    assert waits == []
+
+
+def test_wait_for_passive_solve_returns_as_soon_as_the_page_clears(monkeypatch, bypass):
+    polls = {"n": 0}
+
+    async def _is_bypassed(*_args, **_kwargs) -> bool:
+        polls["n"] += 1
+        return polls["n"] >= 3
+
+    monkeypatch.setattr(bypass, "_is_bypassed", _is_bypassed)
+
+    assert asyncio.run(bypass._wait_for_passive_solve(object())) is True
+    assert polls["n"] == 3
+
+
+def test_wait_for_passive_solve_gives_up_at_the_window(monkeypatch, bypass):
+    """It must not poll forever - the methods still need their share of the budget."""
+    clock = {"now": 0.0}
+    monkeypatch.setattr(bypass.time, "monotonic", lambda: clock["now"])
+
+    async def _tick(*_args, **_kwargs) -> bool:
+        clock["now"] += 1.0
+        return False
+
+    monkeypatch.setattr(bypass, "_is_bypassed", _tick)
+
+    assert asyncio.run(bypass._wait_for_passive_solve(object())) is False
+    assert clock["now"] >= bypass._PASSIVE_SOLVE_SECONDS
+
+
+def test_wait_for_passive_solve_honours_cancellation(monkeypatch, bypass):
+    import threading
+
+    from shelfmark.bypass import BypassCancelledError
+
+    cancel = threading.Event()
+    cancel.set()
+    monkeypatch.setattr(bypass, "_is_bypassed", _never_passes)
+
+    with pytest.raises(BypassCancelledError):
+        asyncio.run(bypass._wait_for_passive_solve(object(), cancel))
+
+
+# --------------------------------------------------------------------------- #
+# The page-load loop stops while there is still time to report a real failure
+# --------------------------------------------------------------------------- #
+def test_page_load_loop_stops_before_the_worker_deadline(monkeypatch, bypass):
+    """A stubborn challenge must produce "bypass failed", not a cancelled coroutine."""
+    clock = {"now": 0.0}
+    monkeypatch.setattr(bypass.time, "monotonic", lambda: clock["now"])
+    monkeypatch.delenv(bypass._BYPASS_CHILD_ENV, raising=False)
+
+    attempts = {"n": 0}
+
+    async def _create(_url):
+        return object()
+
+    async def _get(_url, _driver, _cancel=None) -> str:
+        attempts["n"] += 1
+        # Each pass eats a realistic slice of the budget.
+        clock["now"] += 120.0
+        return ""
+
+    async def _close(_driver) -> None:
+        return None
+
+    monkeypatch.setattr(bypass, "_create_cdp_browser", _create)
+    monkeypatch.setattr(bypass, "_get", _get)
+    monkeypatch.setattr(bypass, "_close_cdp_driver", _close)
+
+    class _RealWorker:
+        def run(self, coro, timeout=None):
+            return asyncio.run(coro)
+
+    monkeypatch.setattr(bypass, "_CDP_WORKER", _RealWorker())
+
+    result = bypass._run_bypass_in_current_process("https://example.com", 10)
+
+    assert result == ""
+    # Well short of the 10 it was asked for, and short of the deadline it had.
+    assert attempts["n"] < 10
+    budget = bypass._IN_PROCESS_BYPASS_TIMEOUT_SECONDS
+    assert clock["now"] < budget, "the loop must leave room to report the failure"
+
+
+def test_page_load_loop_still_makes_one_attempt_on_a_spent_budget(monkeypatch, bypass):
+    """The deadline check must never skip the request entirely."""
+    clock = {"now": 10_000.0}
+    monkeypatch.setattr(bypass.time, "monotonic", lambda: clock["now"])
+    monkeypatch.delenv(bypass._BYPASS_CHILD_ENV, raising=False)
+
+    attempts = {"n": 0}
+
+    async def _create(_url):
+        return object()
+
+    async def _get(_url, _driver, _cancel=None) -> str:
+        attempts["n"] += 1
+        return "<html>solved</html>"
+
+    async def _close(_driver) -> None:
+        return None
+
+    monkeypatch.setattr(bypass, "_create_cdp_browser", _create)
+    monkeypatch.setattr(bypass, "_get", _get)
+    monkeypatch.setattr(bypass, "_close_cdp_driver", _close)
+
+    class _RealWorker:
+        def run(self, coro, timeout=None):
+            return asyncio.run(coro)
+
+    monkeypatch.setattr(bypass, "_CDP_WORKER", _RealWorker())
+
+    assert bypass._run_bypass_in_current_process("https://example.com", 10) == "<html>solved</html>"
+    assert attempts["n"] == 1
+
+
+class _FakeElement:
+    async def get_html_async(self) -> str:
+        return "<html>solved</html>"
+
+
+class _FakePage:
+    """A page that only produces its document after `ready_after` seconds of waiting."""
+
+    def __init__(self, ready_after: float = 0.0) -> None:
+        self.ready_after = ready_after
+        self.waited_with: list[float] = []
+
+    async def find(self, selector: str, timeout: float = 1):
+        self.waited_with.append(timeout)
+        if timeout < self.ready_after:
+            msg = f"Time ran out while waiting for: {{{selector}}}"
+            raise TimeoutError(msg)
+        return _FakeElement()
+
+
+def test_page_source_waits_longer_than_seleniumbases_one_second(bypass):
+    """A page still navigating after a solve must not lose the solve.
+
+    SeleniumBase's get_page_source() allows one second for the document. Anna's Archive
+    hands back a redirect to the real content instead, so the read raised TimeoutError
+    while the challenge had in fact been cleared.
+    """
+    page = _FakePage(ready_after=5.0)
+
+    assert asyncio.run(bypass._read_page_source(page)) == "<html>solved</html>"
+    assert page.waited_with == [bypass._PAGE_SOURCE_TIMEOUT_DEFAULT]
+
+
+def test_page_source_timeout_is_configurable(bypass, monkeypatch):
+    """BYPASS_PAGE_SOURCE_TIMEOUT overrides the default for slow or fast setups."""
+    monkeypatch.setattr(
+        bypass.app_config,
+        "get",
+        lambda key, default=None: 45 if key == "BYPASS_PAGE_SOURCE_TIMEOUT" else default,
+    )
+    page = _FakePage()
+
+    asyncio.run(bypass._read_page_source(page))
+
+    assert page.waited_with == [45.0]
@@ -0,0 +1,88 @@
+"""No method in the list may be a step another method already takes first.
+
+`BYPASS_METHODS` used to open with a solve-only entry that called `page.solve_captcha()`
+and checked the result. `_bypass_method_cdp_gui_click`, the entry behind it, opens by
+doing exactly that and returns the moment it works - so against a challenge that
+`solve_captcha()` cannot clear, the first method could only repeat the half that had
+already failed, then charge the loop's backoff before the method that does work started.
+Measured on Anna's Archive at 0/19 successes and ~5.5s of the ~26s each solve cost
+(issue #1285).
+
+The passive-solve window added in v1.3.13 keeps most solves away from this loop entirely,
+so this is about what the loop costs when it does run.
+"""
+
+import asyncio
+
+import pytest
+
+import shelfmark.bypass.internal_bypasser as ib
+
+
+@pytest.fixture
+def no_sleep(monkeypatch):
+    async def _no_sleep(_seconds) -> None:
+        return None
+
+    monkeypatch.setattr(ib.asyncio, "sleep", _no_sleep)
+    monkeypatch.setattr(ib._RNG, "uniform", lambda _a, _b: 0)
+
+
+class _Page:
+    """Records what a method asked the page to do."""
+
+    def __init__(self, *, solve_clears: bool) -> None:
+        self.solve_clears = solve_clears
+        self.calls: list[str] = []
+
+    async def solve_captcha(self) -> None:
+        self.calls.append("solve_captcha")
+
+    async def is_element_visible(self, selector: str) -> bool:
+        self.calls.append(f"visible:{selector}")
+        return False
+
+    async def click_with_offset(self, selector: str, _x, _y, center=True) -> None:
+        self.calls.append(f"click:{selector}")
+
+
+def _stub_is_bypassed(monkeypatch, page: _Page) -> None:
+    async def _is_bypassed(*_a, **_kw) -> bool:
+        return page.solve_clears and "solve_captcha" in page.calls
+
+    monkeypatch.setattr(ib, "_is_bypassed", _is_bypassed)
+
+
+def test_no_solve_only_method_remains_in_the_list():
+    names = [method.__name__ for method in ib.BYPASS_METHODS]
+    assert "_bypass_method_cdp_solve" not in names
+    assert names[0] == "_bypass_method_cdp_gui_click"
+
+
+def test_the_first_method_still_tries_solve_captcha_first(monkeypatch, no_sleep):
+    """Coverage is only preserved because gui_click opens with the same call."""
+    page = _Page(solve_clears=True)
+    _stub_is_bypassed(monkeypatch, page)
+
+    assert asyncio.run(ib.BYPASS_METHODS[0](page)) is True
+    assert page.calls == ["solve_captcha"], "it must return before touching any selector"
+
+
+def test_it_falls_through_to_clicking_when_solve_does_not_clear(monkeypatch, no_sleep):
+    """The half that actually works on DDoS-Guard still runs in the same attempt."""
+    page = _Page(solve_clears=False)
+    _stub_is_bypassed(monkeypatch, page)
+
+    assert asyncio.run(ib.BYPASS_METHODS[0](page)) is False
+    assert page.calls[0] == "solve_captcha"
+    assert any(call.startswith("visible:") for call in page.calls), (
+        "the selector pass should have been reached in the same attempt"
+    )
+
+
+def test_the_derived_budgets_follow_the_shortened_list():
+    """Both budgets are computed from the list, so removing an entry must not strand them."""
+    assert ib._BYPASS_METHOD_ATTEMPTS == len(ib.BYPASS_METHODS) + 1
+    assert ib._BYPASS_METHOD_ATTEMPTS >= len(ib.BYPASS_METHODS), (
+        "every method must still get a turn"
+    )
@@ -43,6 +43,37 @@ def _store(cookies, url="https://annas-archive.gl/search"):
    cs.store_extracted_cookies(url=url, cookies=cookies, user_agent="UA/1.0")


+@pytest.fixture
+def cookie_store_logs():
+    """Collect cookie-store log messages.
+
+    The store's logger is built outside the standard hierarchy, so its records never
+    reach the root handler caplog installs.
+    """
+    import logging
+
+    messages: list[str] = []
+
+    class _Capture(logging.Handler):
+        def emit(self, record: logging.LogRecord) -> None:
+            messages.append(record.getMessage())
+
+    handler = _Capture()
+    cs.logger.addHandler(handler)
+    previous = cs.logger.level
+    cs.logger.setLevel(logging.DEBUG)
+    # setup_logger builds its loggers with CustomLogger(name) rather than getLogger, so
+    # they are not in the manager's hierarchy - and Logger.setLevel only invalidates the
+    # is-enabled cache *through* the manager. Without this the logger keeps answering
+    # "DEBUG is off" from a cache entry made while it was at INFO.
+    cs.logger._cache.clear()
+    try:
+        yield messages
+    finally:
+        cs.logger.removeHandler(handler)
+        cs.logger.setLevel(previous)
+
+
 # --------------------------------------------------------------------------- #
 # Per-check cookies must not be persisted for replay
 # --------------------------------------------------------------------------- #
@@ -220,3 +251,56 @@ def test_failure_only_clears_the_failing_host(monkeypatch):

    assert ib.get_cf_cookies_for_domain("annas-archive.gl") == {}
    assert ib.get_cf_cookies_for_domain("other-site.test") == {"__ddg1_": "other"}
+
+
+# --------------------------------------------------------------------------- #
+# Settling what DDoS-Guard actually treats as clearance (issue #1276)
+# --------------------------------------------------------------------------- #
+def test_per_check_cookies_can_be_kept_for_a_field_test(monkeypatch):
+    """Which __ddg* cookies are clearance is not settled, so it has to be testable.
+
+    The store's premise - that __ddg8_/__ddg9_/__ddg10_ describe one check and must not
+    be replayed - is contradicted by the field reports on #1276, where every request
+    after a successful solve was challenged again. This env-only switch is how that gets
+    answered against a live host without building a branch.
+    """
+    from shelfmark.config import env
+
+    monkeypatch.setattr(env, "DDG_REPLAY_PER_CHECK_COOKIES", True)
+    _store(
+        [
+            _Cookie("__ddg1_", "clearance"),
+            _Cookie("__ddg8_", "opaque"),
+            _Cookie("__ddg9_", "203.0.113.7"),
+            _Cookie("__ddg10_", "1786826304"),
+        ]
+    )
+
+    stored = ib.get_cf_cookies_for_domain("annas-archive.gl")
+
+    assert set(stored) == {"__ddg1_", "__ddg8_", "__ddg9_", "__ddg10_"}
+
+
+def test_dropping_per_check_cookies_is_the_default(monkeypatch):
+    """The switch is for reproducing the question, not a behaviour change."""
+    from shelfmark.config import env
+
+    assert env.DDG_REPLAY_PER_CHECK_COOKIES is False
+    _store([_Cookie("__ddg1_", "clearance"), _Cookie("__ddg9_", "203.0.113.7")])
+
+    assert set(ib.get_cf_cookies_for_domain("annas-archive.gl")) == {"__ddg1_"}
+
+
+def test_a_solve_logs_which_cookies_it_won_and_which_were_held_back(cookie_store_logs):
+    """Without this, a debug log shows a solve succeed and the next request challenged,
+    with nothing in between to explain why."""
+    _store([_Cookie("__ddg1_", "clearance"), _Cookie("__ddg9_", "203.0.113.7")])
+
+    messages = cookie_store_logs
+    line = next((m for m in messages if "won" in m and "dropping" in m), None)
+    assert line is not None, messages
+    assert "__ddg1_" in line
+    assert "__ddg9_" in line
+    # Names only - a clearance cookie's value is a credential.
+    assert "clearance" not in line
+    assert "203.0.113.7" not in line
@@ -0,0 +1,110 @@
+"""A recording that never happened must say why.
+
+Issue #1276: the debug bundle's recording/ directory was empty, and the only trace was
+three "FFmpeg already stopped" debug lines - one per bypass, each logged 20-56s after
+the recorder was started, meaning ffmpeg had exited almost immediately every time. It ran
+with `-loglevel 0` and no stderr capture, so nothing anywhere recorded the reason. The
+screen recording is the single most useful artifact for diagnosing a bypass failure.
+"""
+
+import subprocess
+
+import pytest
+
+import shelfmark.bypass.internal_bypasser as ib
+
+
+@pytest.fixture(autouse=True)
+def _clean_display():
+    before = dict(ib.DISPLAY)
+    ib.DISPLAY["ffmpeg"] = None
+    ib.DISPLAY["ffmpeg_output"] = None
+    ib.DISPLAY["ffmpeg_error_log"] = None
+    yield
+    ib.DISPLAY.update(before)
+
+
+class _Proc:
+    def __init__(self, returncode):
+        self.returncode = returncode
+
+    def poll(self):
+        return self.returncode
+
+
+def test_ffmpeg_errors_are_captured_to_a_file_beside_the_recording(monkeypatch, tmp_path):
+    monkeypatch.setattr(ib, "RECORDING_DIR", tmp_path)
+    captured: dict[str, object] = {}
+
+    def fake_popen(cmd, **kwargs):
+        captured["cmd"] = cmd
+        captured["stderr"] = kwargs.get("stderr")
+        return _Proc(None)
+
+    monkeypatch.setattr(ib.subprocess, "Popen", fake_popen)
+
+    ib._start_ffmpeg_recording(display=":99")
+
+    cmd = captured["cmd"]
+    # Errors must not be thrown away any more.
+    assert "-loglevel" in cmd
+    assert cmd[cmd.index("-loglevel") + 1] == "error"
+    # stderr goes to a real file, not a pipe nothing would drain.
+    assert captured["stderr"] is not None
+    assert captured["stderr"] is not subprocess.PIPE
+
+    error_log = ib.DISPLAY["ffmpeg_error_log"]
+    assert error_log is not None
+    assert error_log.parent == tmp_path
+    # It sits beside the mp4, so it travels in the debug bundle.
+    assert error_log.name.startswith("screen_recording_")
+
+
+def test_an_early_exit_is_reported_with_ffmpegs_own_reason(monkeypatch, tmp_path, caplog):
+    reason = "[x11grab @ 0x1] Cannot open display :99, error 1."
+    error_log = tmp_path / "screen_recording_x.ffmpeg.log"
+    error_log.write_text(reason, encoding="utf-8")
+
+    ib.DISPLAY["ffmpeg"] = _Proc(1)
+    ib.DISPLAY["ffmpeg_output"] = tmp_path / "screen_recording_x.mp4"
+    ib.DISPLAY["ffmpeg_error_log"] = error_log
+
+    messages: list[str] = []
+
+    class _Capture:
+        def emit(self, record):
+            messages.append(record.getMessage())
+
+    import logging
+
+    handler = logging.Handler()
+    handler.emit = _Capture().emit  # type: ignore[method-assign]
+    ib.logger.addHandler(handler)
+    previous = ib.logger.level
+    ib.logger.setLevel(logging.DEBUG)
+    ib.logger._cache.clear()
+    try:
+        ib._stop_ffmpeg_recording()
+    finally:
+        ib.logger.removeHandler(handler)
+        ib.logger.setLevel(previous)
+
+    line = next((m for m in messages if "exited early" in m), None)
+    assert line is not None, messages
+    assert "code 1" in line
+    assert "Cannot open display" in line
+    assert ib.DISPLAY["ffmpeg"] is None
+
+
+def test_summary_is_explicit_when_ffmpeg_logged_nothing(tmp_path):
+    empty = tmp_path / "screen_recording_y.ffmpeg.log"
+    empty.write_text("", encoding="utf-8")
+    ib.DISPLAY["ffmpeg_error_log"] = empty
+
+    assert "logged nothing" in ib._ffmpeg_error_summary()
+
+
+def test_summary_survives_a_missing_log():
+    ib.DISPLAY["ffmpeg_error_log"] = None
+
+    assert "No FFmpeg error log" in ib._ffmpeg_error_summary()
@@ -0,0 +1,113 @@
+"""A 429 is throttling, not a dead clearance cookie.
+
+Issue #1276: every rejection of the cached cookies took the same exit, which cleared the
+host's clearance. That is right for a 403 and for the ?check=1 redirect loop - being
+challenged while presenting a cookie proves the cookie is dead - and wrong for a 429,
+where the origin is rate-limiting the IP and would answer a real browser holding the very
+same cookies identically.
+
+The cost in the reported bundle: a solve completed at 13:41:23 and stored five cookies;
+six seconds later a 429 threw them away, and the next query bought its own 56-second
+browser solve. Reuse rate across the whole log was 0 of 2.
+"""
+
+import pytest
+
+import shelfmark.bypass.cookie_store as cs
+import shelfmark.bypass.internal_bypasser as ib
+
+URL = "https://annas-archive.gl/search?q=dune"
+HOST = "annas-archive.gl"
+
+
+class _Resp:
+    def __init__(self, status_code, text="page"):
+        self.status_code = status_code
+        self.text = text
+
+
+@pytest.fixture(autouse=True)
+def _clean(monkeypatch):
+    monkeypatch.setattr(cs, "_cf_cookies", {})
+    monkeypatch.setattr(cs, "_cf_user_agents", {})
+    monkeypatch.setattr(cs, "_get_full_cookie_domains", set)
+    monkeypatch.setattr(ib, "get_proxies", lambda _url: None)
+    monkeypatch.setattr(ib, "get_ssl_verify", lambda _url: True)
+    cs._cf_cookies[HOST] = {
+        "__ddg1_": {"value": "clearance", "expiry": None},
+        "__ddg2_": {"value": "c2", "expiry": None},
+    }
+
+
+def _cooldowns(monkeypatch):
+    """Record note_rate_limited calls without arming the real per-host ladder."""
+    armed: list[str] = []
+    monkeypatch.setattr(ib.network, "note_rate_limited", lambda url: armed.append(url) or 120.0)
+    return armed
+
+
+def test_429_keeps_the_clearance(monkeypatch):
+    armed = _cooldowns(monkeypatch)
+    monkeypatch.setattr(ib.requests, "get", lambda *a, **k: _Resp(429))
+
+    assert ib._try_with_cached_cookies(URL, HOST) is None
+    assert ib.get_cf_cookies_for_domain(HOST) == {"__ddg1_": "clearance", "__ddg2_": "c2"}
+    assert armed == [URL], "the backoff must still be armed"
+
+
+def test_403_still_discards_the_clearance(monkeypatch):
+    """The pre-existing behaviour for a genuine rejection must not regress."""
+    _cooldowns(monkeypatch)
+    monkeypatch.setattr(ib.requests, "get", lambda *a, **k: _Resp(403))
+
+    assert ib._try_with_cached_cookies(URL, HOST) is None
+    assert ib.get_cf_cookies_for_domain(HOST) == {}
+
+
+def test_redirect_loop_still_discards_the_clearance(monkeypatch):
+    _cooldowns(monkeypatch)
+
+    def boom(*_a, **_k):
+        raise ib.requests.exceptions.TooManyRedirects("Exceeded 30 redirects")
+
+    monkeypatch.setattr(ib.requests, "get", boom)
+
+    assert ib._try_with_cached_cookies(URL, HOST) is None
+    assert ib.get_cf_cookies_for_domain(HOST) == {}
+
+
+def test_a_throttled_host_is_not_handed_a_browser_solve(monkeypatch):
+    """A solve cannot clear a throttle, and is itself more traffic at a host asking for
+    less. get_bypassed_page checks the cooldown before the queue; get() has to re-check
+    after it, because a request can hold for LOCKED while another collects the 429."""
+    monkeypatch.setattr(ib.requests, "get", lambda *a, **k: _Resp(429))
+    monkeypatch.setattr(ib.network, "note_rate_limited", lambda _url: 120.0)
+    monkeypatch.setattr(ib.network, "host_cooldown_remaining", lambda _url: 118.0)
+
+    solved: list[str] = []
+    monkeypatch.setattr(
+        ib, "_run_bypass_in_current_process", lambda url, *a, **k: solved.append(url) or "html"
+    )
+    monkeypatch.setattr(ib.env, "DOCKERMODE", False)
+
+    with pytest.raises(ib.network.RateLimitedError) as excinfo:
+        ib.get(URL, retry=1)
+
+    assert solved == [], "no browser should have been started"
+    assert "rate-limited" in str(excinfo.value)
+    # And the clearance survives, ready for when the cooldown clears.
+    assert ib.get_cf_cookies_for_domain(HOST) == {"__ddg1_": "clearance", "__ddg2_": "c2"}
+
+
+def test_a_host_that_is_not_throttled_still_solves(monkeypatch):
+    monkeypatch.setattr(ib.requests, "get", lambda *a, **k: _Resp(403))
+    monkeypatch.setattr(ib.network, "host_cooldown_remaining", lambda _url: 0.0)
+
+    solved: list[str] = []
+    monkeypatch.setattr(
+        ib, "_run_bypass_in_current_process", lambda url, *a, **k: solved.append(url) or "html"
+    )
+    monkeypatch.setattr(ib.env, "DOCKERMODE", False)
+
+    assert ib.get(URL, retry=1) == "html"
+    assert solved == [URL]
@@ -0,0 +1,214 @@
+"""A challenge page is a failed solve, whatever verdict the solver reports on itself.
+
+Regression for #1292. FlareSolverr answers "Challenge solved!" for anything it does not
+recognise as a Cloudflare challenge, and DDoS-Guard's manual CAPTCHA page is one such
+thing. The external bypasser logged a warning that the solve had not cleared the
+protection and then returned the page as a success anyway, which had three consequences:
+the retry-and-rotate loop that could still have reached a working mirror was never
+entered, the CAPTCHA page's own __ddg cookies were filed as that host's clearance, and
+the user was told to go and check a bypasser that was working perfectly.
+"""
+
+import pytest
+
+from shelfmark.bypass import ChallengeNotSolvedError
+
+# Verbatim from the annas-archive.pk page in #1292, trimmed to the markers. This is the
+# *manual* CAPTCHA - "could not verify your browser automatically" - not the ~900 byte
+# JS interstitial that a browser clears on its own.
+DDOS_GUARD_CAPTCHA = (
+    '<html><head><title>DDOS-GUARD</title><meta charset="utf-8">'
+    '<link rel="stylesheet" href="/.well-known/ddos-guard/ddg-captcha-page/index.css">'
+    '<script defer="defer" src="/.well-known/ddos-guard/ddg-captcha-page/index.js"></script>'
+    '</head><body><div class="container"><h1 id="title">Checking your browser before '
+    'accessing annas-archive.pk</h1><p id="description">Sorry, we could not verify your '
+    "browser automatically. Complete the manual check to continue</p>"
+    '<div id="ddg-captcha"></div></div></body></html>'
+)
+
+
+class _FakeResponse:
+    def __init__(self, payload: dict) -> None:
+        self._payload = payload
+
+    def raise_for_status(self) -> None:
+        return None
+
+    def json(self) -> dict:
+        return self._payload
+
+
+def _stub_solution(monkeypatch, external_bypasser, solution: dict) -> None:
+    """Answer every bypass with `solution`, with config and SSL stubbed out."""
+
+    def fake_get(key, default=""):
+        values = {
+            "EXT_BYPASSER_URL": "https://bypass.example",
+            "EXT_BYPASSER_PATH": "/v1",
+            "EXT_BYPASSER_TIMEOUT": 60000,
+        }
+        return values.get(key, default)
+
+    monkeypatch.setattr(external_bypasser.config, "get", fake_get)
+    monkeypatch.setattr(
+        external_bypasser.requests,
+        "post",
+        # "Challenge solved!" is the solver's verdict; the page is the evidence.
+        lambda *_a, **_k: _FakeResponse(
+            {"status": "ok", "message": "Challenge solved!", "solution": solution}
+        ),
+    )
+    monkeypatch.setattr(external_bypasser, "get_ssl_verify", lambda _url: False)
+
+
+def test_a_captcha_page_is_reported_as_unsolved_not_returned(monkeypatch):
+    import shelfmark.bypass.external_bypasser as external_bypasser
+
+    _stub_solution(monkeypatch, external_bypasser, {"response": DDOS_GUARD_CAPTCHA})
+
+    with pytest.raises(ChallengeNotSolvedError) as excinfo:
+        external_bypasser._fetch_via_bypasser("https://annas-archive.pk/search?q=dune")
+
+    # The marker travels with the failure so the user-facing message can name it.
+    assert str(excinfo.value) == "/.well-known/ddos-guard/"
+
+
+def test_cookies_from_a_captcha_page_are_never_filed_as_clearance(monkeypatch):
+    """They belong to an unsolved check, so replaying them only re-arms the gate."""
+    import shelfmark.bypass.cookie_store as cookie_store
+    import shelfmark.bypass.external_bypasser as external_bypasser
+
+    monkeypatch.setattr(cookie_store, "_cf_cookies", {})
+    monkeypatch.setattr(cookie_store, "_cf_user_agents", {})
+    _stub_solution(
+        monkeypatch,
+        external_bypasser,
+        {
+            "response": DDOS_GUARD_CAPTCHA,
+            "userAgent": "Mozilla/5.0 (solver)",
+            "cookies": [{"name": "__ddg1_", "value": "from-a-captcha"}],
+        },
+    )
+
+    with pytest.raises(ChallengeNotSolvedError):
+        external_bypasser._fetch_via_bypasser("https://annas-archive.pk/search?q=dune")
+
+    assert cookie_store.get_cf_cookies_for_domain("annas-archive.pk") == {}
+    assert cookie_store.get_cf_user_agent_for_domain("annas-archive.pk") is None
+
+
+class _FakeSelector:
+    """Two mirrors, rotated on demand - each is its own DDoS-Guard host."""
+
+    def __init__(self) -> None:
+        self.current_base = "https://mirror-one.example"
+        self.rotate_calls = 0
+
+    def rewrite(self, url: str) -> str:
+        return url.replace("https://orig.example", self.current_base, 1)
+
+    def next_mirror_or_rotate_dns(self) -> tuple[str | None, str]:
+        self.rotate_calls += 1
+        self.current_base = "https://mirror-two.example"
+        return self.current_base, "mirror"
+
+
+def _no_sleeping(monkeypatch, external_bypasser) -> None:
+    monkeypatch.setattr(external_bypasser, "_sleep_with_cancellation", lambda _seconds, _flag: None)
+
+
+def test_an_unsolved_challenge_rotates_to_the_next_mirror(monkeypatch):
+    """The recovery the old code skipped by calling the CAPTCHA page a success."""
+    import shelfmark.bypass.external_bypasser as external_bypasser
+
+    _no_sleeping(monkeypatch, external_bypasser)
+    fetched: list[str] = []
+
+    def fake_fetch(url: str) -> str | None:
+        fetched.append(url)
+        if "mirror-one" in url:
+            raise ChallengeNotSolvedError("/.well-known/ddos-guard/")
+        return "<html>real page</html>"
+
+    monkeypatch.setattr(external_bypasser, "_fetch_via_bypasser", fake_fetch)
+
+    selector = _FakeSelector()
+    result = external_bypasser.get_bypassed_page("https://orig.example/search", selector=selector)
+
+    assert result == "<html>real page</html>"
+    assert fetched == [
+        "https://mirror-one.example/search",
+        "https://mirror-two.example/search",
+    ]
+    assert selector.rotate_calls == 1
+
+
+def test_every_attempt_challenged_blames_the_host_not_the_bypasser(monkeypatch):
+    import shelfmark.bypass.external_bypasser as external_bypasser
+
+    _no_sleeping(monkeypatch, external_bypasser)
+
+    def always_challenged(_url: str) -> str | None:
+        raise ChallengeNotSolvedError("/.well-known/ddos-guard/")
+
+    monkeypatch.setattr(external_bypasser, "_fetch_via_bypasser", always_challenged)
+
+    with pytest.raises(ChallengeNotSolvedError) as excinfo:
+        external_bypasser.get_bypassed_page("https://orig.example/search", selector=_FakeSelector())
+
+    message = str(excinfo.value)
+    assert "manual CAPTCHA" in message
+    assert "the bypasser itself is working" in message
+
+
+def test_an_unreachable_bypasser_still_reports_as_such(monkeypatch):
+    """The other cause must stay distinguishable: None, not an unsolved challenge."""
+    import shelfmark.bypass.external_bypasser as external_bypasser
+
+    _no_sleeping(monkeypatch, external_bypasser)
+    monkeypatch.setattr(external_bypasser, "_fetch_via_bypasser", lambda _url: None)
+
+    assert (
+        external_bypasser.get_bypassed_page("https://orig.example/search", selector=_FakeSelector())
+        is None
+    )
+
+
+def test_html_get_page_surfaces_the_host_as_the_cause(monkeypatch):
+    """The message the user actually reads must not send them to fix FlareSolverr.
+
+    `_run_bypasser`'s generic handler says "the protection bypasser failed", and the
+    search layer's give-up used to add "check that the bypasser is reachable and
+    working" - which is what #1292 spent its investigation doing.
+    """
+    import shelfmark.download.http as http
+    import shelfmark.download.network as network
+
+    monkeypatch.setattr(http, "_is_cf_bypass_enabled", lambda: True)
+    monkeypatch.setattr(http.network, "should_rotate_dns_for_url", lambda _url: True)
+
+    def challenged(*_args, **_kwargs):
+        msg = "the site kept answering with a protection challenge - manual CAPTCHA"
+        raise ChallengeNotSolvedError(msg)
+
+    monkeypatch.setattr(http, "get_bypassed_page", challenged)
+
+    statuses: list[tuple[str, str | None]] = []
+    selector = network.AAMirrorSelector()
+
+    html = http.html_get_page(
+        "https://annas-archive.pk/search?q=dune",
+        retry=1,
+        selector=selector,
+        status_callback=lambda stage, detail: statuses.append((stage, detail)),
+        use_bypasser=True,
+        success_delay=0,
+    )
+
+    assert html == ""
+    assert selector.last_failure is not None
+    assert "manual CAPTCHA" in selector.last_failure
+    assert "reachable" not in selector.last_failure
+    assert ("error", "the site kept answering with a protection challenge - manual CAPTCHA") in (
+        statuses
+    )
--- a/Show More
+++ b/Show More