Files
json/docs/mkdocs/docs/features/binary_values.md
T
Niels Lohmann d5da84249f Skip CBOR tags in front of map keys, validate indefinite text chunks, fix CBOR docs (#5927)
* Skip CBOR tags in front of map keys and validate indefinite text chunks

A tag in front of a text string map key was rejected with parse_error.113
even with cbor_tag_handler_t::ignore or store, although tags in front of
values are handled. The tags are now skipped iteratively before the key is
read; the handler error still rejects them.

The chunks of an indefinite-length text string were concatenated before
the single UTF-8 check, so a code point split over two chunks, or an
invalid chunk, was accepted. RFC 8949, Section 3.2.3 requires every chunk
to be valid on its own. With error_handler_t::strict, replace and ignore
each chunk is now checked; keep is unchanged. The chunks are appended with
append(ptr, len), so StringType needs nothing new.

Signed-off-by: Niels Lohmann <mail@nlohmann.me>

* docs: correct and extend the CBOR documentation

Fix the negative-integer rows (they used MessagePack boundaries), the
number_integer range of the 0x1B row, and the half-precision entry in the
list of unused types; correct the NaN history (0xF9 since 3.8.0) and the
tagged-item wording (only the tag directly in front of a byte string is
stored, and only in the 0xD8..0xDB form). Document that to_cbor is not
RFC 8949 deterministic, that duplicate keys are accepted (last wins),
tags in front of keys, per-chunk UTF-8 validation, the exceptions
parse_error.113, out_of_range.406/408 of from_cbor, the pointer and
C array input traps, the partially written output on exceptions, and the
store/ignore results in the binary values example. Drop the claim that
the CBOR reader needs StringType::append(const StringType&) and
BinaryType::insert(pos, first, last).

Signed-off-by: Niels Lohmann <mail@nlohmann.me>

* Regenerate single_include and apply astyle

Result of make amalgamate (including make pretty) for the preceding changes.

Signed-off-by: Niels Lohmann <mail@nlohmann.me>

* Address review of the CBOR changes

Cite RFC 8949 Section 5.3.1 for duplicate keys, describe the out_of_range.408 condition precisely, rewrap a long line, and update the stale get_string comment (per-chunk UTF-8 check). single_include regenerated.

Signed-off-by: Niels Lohmann <mail@nlohmann.me>

* docs: drop discarded-value remarks

Remove the mention that a discarded value cannot be serialized to CBOR
(type_error.321) from the exception-safety and mapping sections. The
integer table and all added prose already follow the table-format and
Oxford-comma rules.

Signed-off-by: Niels Lohmann <mail@nlohmann.me>

---------

Signed-off-by: Niels Lohmann <mail@nlohmann.me>
2026-10-26 18:25:44 +01:00

13 KiB

Binary Values

The library implements several binary formats that encode JSON in an efficient way. Most of these formats support binary values; that is, values that have semantics defined outside the library and only define a sequence of bytes to be stored.

JSON itself does not have a binary value. As such, binary values are an extension that this library implements to store values received by a binary format. Binary values are never created by the JSON parser and are only part of a serialized JSON text if they have been created manually or via a binary format.

API for binary values

classDiagram

class binary_t ["json::binary_t"] {
    +void set_subtype(std::uint64_t subtype)
    +void clear_subtype()
    +std::uint64_t subtype() const
    +bool has_subtype() const
}

class vector ["std::vector<uint8_t>"]

vector <|-- binary_t

By default, binary values are stored as std::vector<std::uint8_t>. This type can be changed by providing a template parameter to the basic_json type. To store binary subtypes, the storage type is extended and exposed as json::binary_t:

auto binary = json::binary_t({0xCA, 0xFE, 0xBA, 0xBE});
auto binary_with_subtype = json::binary_t({0xCA, 0xFE, 0xBA, 0xBE}, 42);

There are several convenience functions to check and set the subtype:

binary.has_subtype();                   // returns false
binary_with_subtype.has_subtype();      // returns true

binary_with_subtype.clear_subtype();
binary_with_subtype.has_subtype();      // returns false

binary_with_subtype.set_subtype(42);
binary.set_subtype(23);

binary.subtype();                       // returns 23

As json::binary_t is subclassing std::vector<std::uint8_t>, all member functions are available:

binary.size();  // returns 4
binary[1];      // returns 0xFE

JSON values can be constructed from json::binary_t:

json j = binary;

Binary values are primitive values just like numbers or strings, as reflected by is_binary() and is_primitive():

j.is_binary();    // returns true
j.is_primitive(); // returns true

Given a binary JSON value, the binary_t can be accessed by reference via get_binary():

j.get_binary().has_subtype();  // returns true
j.get_binary().size();         // returns 4

For convenience, binary JSON values can be constructed via json::binary:

auto j2 = json::binary({0xCA, 0xFE, 0xBA, 0xBE}, 23);
auto j3 = json::binary({0xCA, 0xFE, 0xBA, 0xBE});

j2 == j;                        // returns true
j3.get_binary().has_subtype();  // returns false
j3.get_binary().subtype();      // returns std::uint64_t(-1) as j3 has no subtype

Serialization

Binary values are serialized differently according to the formats.

JSON

JSON does not have a binary type, and this library does not introduce a new type as this would break conformance. Instead, binary values are serialized as an object with two keys: bytes holds an array of integers, and subtype is an integer or null.

??? example "Example: serialize a binary value to JSON"

Code:

```cpp
// create a binary value of subtype 42
json j;
j["binary"] = json::binary({0xCA, 0xFE, 0xBA, 0xBE}, 42);

// serialize to standard output
std::cout << j.dump(2) << std::endl;
```

Output:

```json
{
  "binary": {
    "bytes": [202, 254, 186, 190],
    "subtype": 42
  }
}
```

!!! warning "No roundtrip for binary values"

The JSON parser will not parse the objects generated by binary values back to binary values. This is by design to
remain standards compliant. Serializing binary values to JSON is only implemented for debugging purposes.

BJData

BJData neither supports binary values nor subtypes and proposes to serialize binary values as an array of uint8 values. The library implements this translation.

??? example "Example: serialize a binary value to BJData"

Code:

```cpp
// create a binary value of subtype 42 (will be ignored in BJData)
json j;
j["binary"] = json::binary({0xCA, 0xFE, 0xBA, 0xBE}, 42);

// convert to BJData
auto v = json::to_bjdata(j);      
```
        
`v` is a `std::vector<std::uint8_t>` with the following 20 elements:

```c
0x7B                                             // '{'
    0x69 0x06                                    // i 6 (length of the key)
    0x62 0x69 0x6E 0x61 0x72 0x79                // "binary"
    0x5B                                         // '['
        0x55 0xCA 0x55 0xFE 0x55 0xBA 0x55 0xBE  // content (each byte prefixed with 'U')
    0x5D                                         // ']'
0x7D                                             // '}'
```

The following code uses the type and size optimization for BJData:

```cpp
// convert to BJData using the size and type optimization
auto v = json::to_bjdata(j, true, true);
```

The resulting vector has 22 elements; the optimization is not effective for examples with few values:

```c
0x7B                                // '{'
    0x23 0x69 0x01                  // '#' 'i' type of the array elements: unsigned integers
    0x69 0x06                       // i 6 (length of the key)
    0x62 0x69 0x6E 0x61 0x72 0x79   // "binary"
    0x5B                            // '[' array
        0x24 0x55                   // '$' 'U' type of the array elements: unsigned integers
        0x23 0x69 0x04              // '#' i 4 number of array elements
        0xCA 0xFE 0xBA 0xBE         // content
```

Note that subtype (42) is **not** serialized and that BJData has **no binary type**, and deserializing `v` would
yield the following value:

```json
{
  "binary": [202, 254, 186, 190]
}
```

BON8

BON8 neither supports binary values nor subtypes. The library serializes binary values as an array of integers.

??? example "Example: serialize a binary value to BON8"

Code:

```cpp
// create a binary value of subtype 42 (will be ignored in BON8)
json j;
j["binary"] = json::binary({0xCA, 0xFE, 0xBA, 0xBE}, 42);

// convert to BON8
auto v = json::to_bon8(j);
```

`v` is a `std::vector<std::uint8_t>` with the following 16 elements:

```c
0x87                                     // object with 1 member
    0x62 0x69 0x6E 0x61 0x72 0x79        // "binary"
    0x84                                 // array with 4 elements
        0xC3 0x22 0xC3 0x56 0xC3 0x12 0xC3 0x16  // content (each byte as a 2-byte integer)
```

Note that the subtype is lost, and deserializing `v` would yield the following value:

```json
{
  "binary": [202, 254, 186, 190]
}
```

BSON

BSON supports binary values and subtypes. If a subtype is given, it is used and added as an unsigned 8-bit integer. If no subtype is given, the generic binary subtype 0x00 is used.

??? example "Example: serialize a binary value to BSON"

Code:

```cpp
// create a binary value of subtype 42
json j;
j["binary"] = json::binary({0xCA, 0xFE, 0xBA, 0xBE}, 42);

// convert to BSON
auto v = json::to_bson(j);      
```
        
`v` is a `std::vector<std::uint8_t>` with the following 22 elements:

```c
0x16 0x00 0x00 0x00                         // number of bytes in the document
    0x05                                    // binary value
        0x62 0x69 0x6E 0x61 0x72 0x79 0x00  // key "binary" + null byte
        0x04 0x00 0x00 0x00                 // number of bytes
        0x2a                                // subtype
        0xCA 0xFE 0xBA 0xBE                 // content
0x00                                        // end of the document
```

Note that the serialization preserves the subtype, and deserializing `v` would yield the following value:

```json
{
  "binary": {
    "bytes": [202, 254, 186, 190],
    "subtype": 42
  }
}
```

CBOR

CBOR supports binary values and subtypes. Any binary value will be serialized as a byte string; a subtype is serialized as a tag preceding it. When parsing, tags are only read back as subtypes with cbor_tag_handler_t::store (see from_cbor). The library will choose the smallest representation using the length of the byte array.

??? example "Example: serialize a binary value to CBOR"

Code:

```cpp
// create a binary value of subtype 42
json j;
j["binary"] = json::binary({0xCA, 0xFE, 0xBA, 0xBE}, 42);

// convert to CBOR
auto v = json::to_cbor(j);      
```
        
`v` is a `std::vector<std::uint8_t>` with the following 15 elements:

```c
0xA1                                   // map(1)
    0x66                               // text(6)
        0x62 0x69 0x6E 0x61 0x72 0x79  // "binary"
    0xD8 0x2A                          // tag(42)
    0x44                               // bytes(4)
        0xCA 0xFE 0xBA 0xBE            // content
```

Note that the subtype is serialized as tag. However, parsing tagged values yield a parse error unless
`json::cbor_tag_handler_t::ignore` or `json::cbor_tag_handler_t::store` is passed to
[`json::from_cbor`](../api/basic_json/from_cbor.md) (see
[`cbor_tag_handler_t`](../api/basic_json/cbor_tag_handler_t.md)).

With `ignore`, the subtype is dropped (`"subtype": null`); with `store`, the tag is kept (`"subtype": 42`):

```json
{
  "binary": {
    "bytes": [202, 254, 186, 190],
    "subtype": null
  }
}
```

```json
{
  "binary": {
    "bytes": [202, 254, 186, 190],
    "subtype": 42
  }
}
```

MessagePack

MessagePack supports binary values and subtypes. If a subtype is given, the ext family is used. The library will choose the smallest representation among fixext1, fixext2, fixext4, fixext8, ext8, ext16, and ext32. The subtype is then added as a signed 8-bit integer.

If no subtype is given, the bin family (bin8, bin16, bin32) is used.

??? example "Example: serialize a binary value to MessagePack"

Code:

```cpp
// create a binary value of subtype 42
json j;
j["binary"] = json::binary({0xCA, 0xFE, 0xBA, 0xBE}, 42);

// convert to MessagePack
auto v = json::to_msgpack(j);      
```
        
`v` is a `std::vector<std::uint8_t>` with the following 14 elements:

```c
0x81                                   // fixmap1
    0xA6                               // fixstr6
        0x62 0x69 0x6E 0x61 0x72 0x79  // "binary"
    0xD6                               // fixext4
        0x2A                           // subtype
        0xCA 0xFE 0xBA 0xBE            // content
```

Note that the serialization preserves the subtype, and deserializing `v` would yield the following value:

```json
{
  "binary": {
    "bytes": [202, 254, 186, 190],
    "subtype": 42
  }
}
```

UBJSON

UBJSON neither supports binary values nor subtypes and proposes to serialize binary values as an array of uint8 values. The library implements this translation.

??? example "Example: serialize a binary value to UBJSON"

Code:

```cpp
// create a binary value of subtype 42 (will be ignored in UBJSON)
json j;
j["binary"] = json::binary({0xCA, 0xFE, 0xBA, 0xBE}, 42);

// convert to UBJSON
auto v = json::to_ubjson(j);      
```
        
`v` is a `std::vector<std::uint8_t>` with the following 20 elements:

```c
0x7B                                             // '{'
    0x69 0x06                                    // i 6 (length of the key)
    0x62 0x69 0x6E 0x61 0x72 0x79                // "binary"
    0x5B                                         // '['
        0x55 0xCA 0x55 0xFE 0x55 0xBA 0x55 0xBE  // content (each byte prefixed with 'U')
    0x5D                                         // ']'
0x7D                                             // '}'
```

The following code uses the type and size optimization for UBJSON:

```cpp
// convert to UBJSON using the size and type optimization
auto v = json::to_ubjson(j, true, true);
```

The resulting vector has 23 elements; the optimization is not effective for examples with few values:

```c
0x7B                                // '{'
    0x24                            // '$' type of the object elements
    0x5B                            // '[' array
    0x23 0x69 0x01                  // '#' i 1 number of object elements
    0x69 0x06                       // i 6 (length of the key)
    0x62 0x69 0x6E 0x61 0x72 0x79   // "binary"
        0x24 0x55                   // '$' 'U' type of the array elements: unsigned integers
        0x23 0x69 0x04              // '#' i 4 number of array elements
        0xCA 0xFE 0xBA 0xBE         // content
```

Note that subtype (42) is **not** serialized and that UBJSON has **no binary type**, and deserializing `v` would
yield the following value:

```json
{
  "binary": [202, 254, 186, 190]
}
```