This document contains all the API usage and examples for the yyjson library.

API Design

API prefix

All public functions and structs are prefixed with yyjson_, and all constants are prefixed with YYJSON_.

API for immutable/mutable data

The library has 2 types of data structures: immutable and mutable:

	Immutable	Mutable
Document	yyjson_doc	yyjson_mut_doc
Value	yyjson_val	yyjson_mut_val

When reading a JSON, yyjson returns immutable documents and values.
When building a JSON, yyjson creates mutable documents and values.
The document holds the memory for all its JSON values and strings.

For most immutable APIs, you can just add a mut after yyjson_ to get the mutable version, for example:

char *yyjson_write(yyjson_doc *doc, ...);
char *yyjson_mut_write(yyjson_mut_doc *doc, ...);
 
bool yyjson_is_str(const yyjson_val *val);
bool yyjson_mut_is_str(const yyjson_mut_val *val);

The library also provides some functions to convert values between immutable and mutable:

// doc -> mut_doc
yyjson_mut_doc *yyjson_doc_mut_copy(const yyjson_doc *doc, ...);
// val -> mut_val
yyjson_mut_val *yyjson_val_mut_copy(const yyjson_val *val, ...);
 
// mut_doc -> doc
yyjson_doc *yyjson_mut_doc_imut_copy(const yyjson_mut_doc *doc, ...);
// mut_val -> val
yyjson_doc *yyjson_mut_val_imut_copy(const yyjson_mut_val *val, ...);

API for string

The library supports strings with or without null-terminator (\0).
When you need to use a string without a null-terminator or when you explicitly know the length of the string, you can use the function that ends with n, for example:

// null-terminator is required
bool yyjson_equals_str(const yyjson_val *val, const char *str);
// null-terminator is optional
bool yyjson_equals_strn(const yyjson_val *val, const char *str, size_t len);

When creating JSON, yyjson treats strings as constants for better performance. However, if your string will be modified, you should use a function with a cpy to copy the string to the document, for example:

// reference only, null-terminated is required
yyjson_mut_val *yyjson_mut_str(yyjson_mut_doc *doc, const char *str);
// reference only, null-terminator is optional
yyjson_mut_val *yyjson_mut_strn(yyjson_mut_doc *doc, const char *str, size_t len);
 
// copied, null-terminated is required
yyjson_mut_val *yyjson_mut_strcpy(yyjson_mut_doc *doc, const char *str);
// copied, null-terminator is optional
yyjson_mut_val *yyjson_mut_strncpy(yyjson_mut_doc *doc, const char *str, size_t len);

Reading JSON

The library provides 5 functions for reading JSON.
Each function accepts an input of UTF-8 data or a file,
returns a document if successful or NULL if it fails.

Read JSON from string

The dat should be a UTF-8 string, null-terminator is not required.
The len is the byte length of dat.
The flg is reader flag, pass 0 if you don't need it, see reader flag for details.
Returns NULL if dat is NULL or len is 0.

yyjson_doc *yyjson_read(const char *dat, 
                        size_t len, 
                        yyjson_read_flag flg);

Sample code:

const char *str = "[1,2,3,4]";
yyjson_doc *doc = yyjson_read(str, strlen(str), 0);
if (doc) {...}
yyjson_doc_free(doc);

Read JSON from file

The path is the JSON file path. This should be a null-terminated string using the system's native encoding.
The flg is reader flag, pass 0 if you don't need it, see reader flag for details.
The alc is memory allocator, pass NULL if you don't need it, see memory allocator for details.
The err is a pointer to receive error message, pass NULL if you don't need it.
Returns NULL if path is NULL or invalid.

yyjson_doc *yyjson_read_file(const char *path,
                             yyjson_read_flag flg,
                             const yyjson_alc *alc,
                             yyjson_read_err *err);

Sample code:

yyjson_doc *doc = yyjson_read_file("/tmp/test.json", 0, NULL, NULL);
if (doc) {...}
yyjson_doc_free(doc);

Read JSON from file pointer

The fp is file pointer. The data will be read from the current position of the FILE to the end.
The flg is reader flag, pass 0 if you don't need it, see reader flag for details.
The alc is memory allocator, pass NULL if you don't need it, see memory allocator for details.
The err is a pointer to receive error message, pass NULL if you don't need it.
Returns NULL if fp is NULL or invalid.

yyjson_doc *yyjson_read_fp(FILE *fp,
                           yyjson_read_flag flg,
                           const yyjson_alc *alc,
                           yyjson_read_err *err);

Sample code:

FILE *fp = fdopen(fd, "rb"); // POSIX file descriptor (fd)
yyjson_doc *doc = yyjson_read_fp(fp, 0, NULL, NULL);
if (fp) fclose(fp);
if (doc) {...}
yyjson_doc_free(doc);

Read JSON with options

The dat should be a UTF-8 string, you can pass a const string if you don't use YYJSON_READ_INSITU flag.
The len is the dat's length in bytes.
The flg is reader flag, pass 0 if you don't need it, see reader flag for details.
The alc is memory allocator, pass NULL if you don't need it, see memory allocator for details.
The err is a pointer to receive error message, pass NULL if you don't need it.

yyjson_doc *yyjson_read_opts(char *dat, 
                             size_t len, 
                             yyjson_read_flag flg,
                             const yyjson_alc *alc, 
                             yyjson_read_err *err);

Sample code:

const char *dat = your_file.bytes;
size_t len = your_file.size;
 
yyjson_read_flag flg = YYJSON_READ_ALLOW_COMMENTS | YYJSON_READ_ALLOW_INF_AND_NAN;
yyjson_doc *doc = yyjson_read_opts((char *)dat, len, flg, NULL, NULL);
 
if (doc) {...}
 
yyjson_doc_free(doc);

Read JSON incrementally

Reading a very large JSON document can freeze the program for a short while. If this is not acceptable, incremental reading can be used.

Incremental reading is recommended only for large documents and only when the program needs to be responsive. Incremental reading is slightly slower than yyjson_read() and yyjson_read_opts().

Note: The incremental JSON reader only supports standard JSON. Flags for non-standard features (e.g. comments, trailing commas) are ignored.

To read a large JSON document incrementally:

Call yyjson_incr_new() to create the state for incremental reading.
Call yyjson_incr_read() repeatedly.
Call yyjson_incr_free() to free the state.

Create the state for incremental reading

The buf should be a UTF-8 string, null-terminator is not required. You can pass a const string if you don't use the YYJSON_READ_INSITU flag.
The buf_len is the length of buf in bytes. The flg is reader flag. Pass 0 if you don't need it. See reader flag for details. The alc is memory allocator, pass NULL if you don't need it. See memory allocator for details.

The function returns a new state, or NULL if memory allocation fails.

yyjson_incr_state *yyjson_incr_new(char *buf, size_t buf_len, yyjson_read_flag flg, const yyjson_alc *alc);

yyjson_incr_state

struct yyjson_incr_state yyjson_incr_state

Definition yyjson.h:1086

yyjson_incr_new

yyjson_api yyjson_incr_state * yyjson_incr_new(char *buf, size_t buf_len, yyjson_read_flag flg, const yyjson_alc *alc)

Perform incremental read

Performs incremental read of up to len bytes.

The state for incremental reading is created using yyjson_incr_new().
The len is the maximum number of bytes to read, counting from the start of the JSON data.
The err is a pointer to receive the error information. Required.

The function returns a document object when the reading is complete and NULL otherwise. If err->code is set to YYJSON_READ_ERROR_MORE, it indicates that parsing is not yet complete. Then, increase len by some kilobytes and call this function again. Continue increasing len until len == buf_len (the total length of the input buffer) or until an error other than YYJSON_READ_ERROR_MORE is returned.

Note: Parsing in very small increments is not efficient. An increment of several kilobytes or megabytes is recommended.

yyjson_doc *yyjson_incr_read(yyjson_incr_state *state, size_t len, yyjson_read_err *err);

yyjson_incr_read

yyjson_api yyjson_doc * yyjson_incr_read(yyjson_incr_state *state, size_t len, yyjson_read_err *err)

Free the state used for incremental reading

Free the state created by yyjson_incr_new().

void yyjson_incr_free(yyjson_incr_state *state);

yyjson_incr_free

yyjson_api void yyjson_incr_free(yyjson_incr_state *state)

Sample code

const char *dat = your_file.bytes;
size_t len = your_file.size;
 
yyjson_read_flag flg = YYJSON_READ_NOFLAG;
yyjson_incr_state *state = yyjson_incr_new(dat, len, flg, NULL);
yyjson_doc *doc;
yyjson_read_err err;
size_t read_so_far = 0;
do {
    read_so_far += 100000;
    if (read_so_far > len)
        read_so_far = len;
    doc = yyjson_incr_read(state, read_so_far, &err);
    if (err.code != YYJSON_READ_ERROR_MORE)
        break;
} while (read_so_far < len);
yyjson_incr_free(state);
 
if (doc != NULL) { ... }
 
yyjson_doc_free(doc);

Reader error handling

When reading JSON fails, and you need error information, you can pass a yyjson_read_err pointer to the yyjson_read_xxx() functions to receive the error details.

Sample code:

char *dat = ...;
size_t dat_len = ...;
yyjson_read_err err;
yyjson_doc *doc = yyjson_read_opts(dat, dat_len, 0, NULL, &err);
 
if (!doc) {
    printf("read error: %s, code: %u at byte position: %lu\n", 
            err.msg, err.code, err.pos);
    // printed:
    // read error: trailing comma is not allowed, code: 7, at byte position: 40
}
 
yyjson_doc_free(doc);

The pos in the error information indicates the byte position where the error occurred. If you need the line and column number of the error, you can use the yyjson_locate_pos() function. Note that line and column start from 1, while character starts from 0. All values are calculated based on Unicode characters to ensure compatibility with various text editors.

Sample code:

char *dat = ...;
size_t dat_len = ...;
yyjson_read_err err = ...;
 
size_t line, col, chr;
if (yyjson_locate_pos(dat, dat_len, err.pos, &line, &col, &chr)) {
    printf("error at line: %lu, column: %lu, character index: %lu\n",
           line, col, chr);
    // printed:
    // error at line: 3, column: 5, character index: 32
}

The complete list of error codes (yyjson_read_code):

Code	Name	Description
0	YYJSON_READ_SUCCESS	Success, no error.
1	YYJSON_READ_ERROR_INVALID_PARAMETER	Invalid parameter, such as NULL input string or 0 input length.
2	YYJSON_READ_ERROR_MEMORY_ALLOCATION	Memory allocation failure.
3	YYJSON_READ_ERROR_EMPTY_CONTENT	Input JSON string is empty.
4	YYJSON_READ_ERROR_UNEXPECTED_CONTENT	Unexpected content after document end, such as [123]abc.
5	YYJSON_READ_ERROR_UNEXPECTED_END	Unexpected end of input; the parsed part is valid, such as [123.
6	YYJSON_READ_ERROR_UNEXPECTED_CHARACTER	Unexpected character inside the document, such as [abc].
7	YYJSON_READ_ERROR_JSON_STRUCTURE	Invalid JSON structure, such as [1,].
8	YYJSON_READ_ERROR_INVALID_COMMENT	Invalid comment (deprecated, mapped to UNEXPECTED_END).
9	YYJSON_READ_ERROR_INVALID_NUMBER	Invalid number, such as 123.e12 or 000.
10	YYJSON_READ_ERROR_INVALID_STRING	Invalid string, such as an invalid escape sequence.
11	YYJSON_READ_ERROR_LITERAL	Invalid JSON literal, such as truu.
12	YYJSON_READ_ERROR_FILE_OPEN	Failed to open a file.
13	YYJSON_READ_ERROR_FILE_READ	Failed to read a file.
14	YYJSON_READ_ERROR_MORE	Incomplete input during incremental parsing; state is preserved for continuation.
15	YYJSON_READ_ERROR_DEPTH	Nesting depth exceeded YYJSON_READER_DEPTH_LIMIT.

Reader flag

The library provides a set of flags for JSON reader.

You can use a single flag, or combine multiple flags with bitwise | operator.

Non-standard flags (such as YYJSON_READ_JSON5) have no performance impact when reading standard JSON input.

YYJSON_READ_NOFLAG = 0

This is the default flag for JSON reader (RFC-8259 or ECMA-404 compliant):

Read positive integer as uint64_t.
Read negative integer as int64_t.
Read floating-point number as double with correct rounding.
Read integer which cannot fit in uint64_t or int64_t as double.
Report error if double number is infinity.
Report error if string contains invalid UTF-8 character or BOM.
Report error on trailing commas, comments, Inf and NaN literals.

YYJSON_READ_INSITU

Read the input data in-situ.

This option allows the reader to modify and use the input data to store string values, which can slightly improve reading speed. However, the caller must ensure that the input data is held until the document is freed. The input data must be padded with at least YYJSON_PADDING_SIZE bytes. For example: [1,2] should be [1,2]\0\0\0\0, input length should be 5.

Sample code:

size_t dat_len = ...;
char *buf = malloc(dat_len + YYJSON_PADDING_SIZE); // create a buffer larger than (len + 4)
read_from_socket(buf, ...);
memset(buf + dat_len, 0, YYJSON_PADDING_SIZE); // set 4-byte padding after data
 
yyjson_doc *doc = yyjson_read_opts(buf, dat_len, YYJSON_READ_INSITU, NULL, NULL);
if (doc) {...}
yyjson_doc_free(doc);
free(buf); // the input data should be freed after the document.

YYJSON_READ_STOP_WHEN_DONE

Stop parsing when reaching the end of a JSON document instead of issuing an error if there's additional content after it.

This option is useful for parsing small pieces of JSON within larger data, such as NDJSON.

Sample code:

// Single file with multiple JSON, such as:
// [1,2,3] [4,5,6] {"a":"b"}
 
size_t file_size = ...;
char *dat = malloc(file_size + YYJSON_PADDING_SIZE);
your_read_file(dat, file);
memset(dat + file_size, 0, YYJSON_PADDING_SIZE); // add padding
    
char *hdr = dat;
char *end = dat + file_size;
yyjson_read_flag flg = YYJSON_READ_INSITU | YYJSON_READ_STOP_WHEN_DONE;
 
while (true) {
    yyjson_doc *doc = yyjson_read_opts(hdr, end - hdr, flg, NULL, NULL);
    if (!doc) break;
    your_doc_process(doc);
    hdr += yyjson_doc_get_read_size(doc); // move to next position
    yyjson_doc_free(doc);
}
free(dat);

YYJSON_READ_ALLOW_TRAILING_COMMAS

Allow a single trailing comma at the end of an object or array (non-standard), for example:

{
    "a": 1,
    "b": 2,
}
 
[
    "a",
    "b",
]

YYJSON_READ_ALLOW_COMMENTS

Allow C-style single-line and multi-line comments (non-standard), for example:

{
    "name": "Harry", // single-line comment
    "id": /* multi-line comment */ 123
}

YYJSON_READ_ALLOW_INF_AND_NAN

Allow nan/inf number or case-insensitive literal (non-standard), for example:

{
    "large": 123e999,
    "nan1": NaN,
    "nan2": nan,
    "inf1": Inf,
    "inf2": -Infinity
}

YYJSON_READ_NUMBER_AS_RAW

Read all numbers as raw strings without parsing.

This flag is useful if you want to handle number parsing yourself. You can use the following functions to extract raw strings:

bool yyjson_is_raw(const yyjson_val *val);
const char *yyjson_get_raw(const yyjson_val *val);
size_t yyjson_get_len(const yyjson_val *val);

YYJSON_READ_BIGNUM_AS_RAW

Read big numbers as raw strings.

This flag is useful if you want to parse these big numbers yourself. These big numbers include integers that cannot be represented by int64_t and uint64_t, and floating-point numbers that cannot be represented by finite double.

Note that this flag will be overridden by YYJSON_READ_NUMBER_AS_RAW flag.

YYJSON_READ_ALLOW_INVALID_UNICODE

Allow reading invalid unicode when parsing string values (non-standard), for example:

"\x80xyz"

"\xF0\x81\x81\x81"

This flag permits invalid characters to appear in the string values, but it still reports errors for invalid escape sequences. It does not impact the performance of correctly encoded strings.

Warning: when using this option, be aware that strings within JSON values may contain incorrect encoding, so you need to handle these strings carefully to avoid security risks.

YYJSON_READ_ALLOW_BOM

Allow UTF-8 BOM and skip it before parsing if any (non-standard).

YYJSON_READ_ALLOW_EXT_NUMBER

Allow extended number formats (non-standard):

Hexadecimal numbers, such as 0x7B.
Numbers with leading or trailing decimal point, such as .123, 123..
Numbers with a leading plus sign, such as +123.

YYJSON_READ_ALLOW_EXT_ESCAPE

Allow extended escape sequences in strings (non-standard):

Additional escapes: \a, \e, \v, \', \?, \0.
Hex escapes: \xNN, such as \x7B.
Line continuation: backslash followed by line terminator sequences.
Unknown escape: if backslash is followed by an unsupported character, the backslash will be removed and the character will be kept as-is. However, \1-\9 will still trigger an error.

YYJSON_READ_ALLOW_EXT_WHITESPACE

Allow extended whitespace characters (non-standard):

Vertical tab \v and form feed \f.
Line separator \u2028 and paragraph separator \u2029.
Non-breaking space \xA0.
Byte order mark: \uFEFF.
Other Unicode characters in the Zs (Separator, space) category.

YYJSON_READ_ALLOW_SINGLE_QUOTED_STR

Allow strings enclosed in single quotes (non-standard), such as 'ab'.

YYJSON_READ_ALLOW_UNQUOTED_KEY

Allow object keys without quotes (non-standard), such as {a:1,b:2}. This extends the ECMAScript IdentifierName rule by allowing any non-whitespace character with code point above U+007F.

YYJSON_READ_JSON5

Allow JSON5 format, see: https://json5.org.

This flag supports all JSON5 features with some additional extensions:

Accepts more escape sequences than JSON5 (e.g. \a, \e).
Unquoted keys are not limited to ECMAScript IdentifierName.
Allow case-insensitive NaN, Inf and Infinity literals.

For example:

{
    /* JSON5 example */
    id: 123,
    name: 'Harry',
    color: 0x66CCFF,
    min: .001,
    max: Inf,
    data: '\x00\xAA\xFF',
}

Writing JSON

The library provides 5 sets of functions for writing JSON.
Each function accepts an input of JSON document or root value, and returns a UTF-8 string or file.

Write JSON to string

The doc/val is the JSON document or root value. If it is NULL, returns NULL.
The flg is writer flag, pass 0 if you don't need it, see writer flag for details.
The len is a pointer to receive output length (not including the null-terminator), pass NULL if you don't need it.
This function returns a new JSON string, or NULL if an error occurs.
The string is encoded as UTF-8 with a null-terminator.
You should use free() or alc->free() to release it when it's no longer needed.

// doc -> str
char *yyjson_write(const yyjson_doc *doc, yyjson_write_flag flg, size_t *len);
// mut_doc -> str
char *yyjson_mut_write(const yyjson_mut_doc *doc, yyjson_write_flag flg, size_t *len);
// val -> str
char *yyjson_val_write(const yyjson_val *val, yyjson_write_flag flg, size_t *len);
// mut_val -> str
char *yyjson_mut_val_write(const yyjson_mut_val *val, yyjson_write_flag flg, size_t *len);

Sample code 1:

yyjson_doc *doc = yyjson_read("[1,2,3]", 7, 0);
char *json = yyjson_write(doc, YYJSON_WRITE_PRETTY, NULL);
printf("%s\n", json);
free(json);

Sample code 2:

yyjson_mut_doc *doc = yyjson_mut_doc_new(NULL);
yyjson_mut_val *arr = yyjson_mut_arr(doc);
yyjson_mut_doc_set_root(doc, arr);
yyjson_mut_arr_add_int(doc, arr, 1);
yyjson_mut_arr_add_int(doc, arr, 2);
yyjson_mut_arr_add_int(doc, arr, 3);
    
char *json = yyjson_mut_write(doc, YYJSON_WRITE_PRETTY, NULL);
printf("%s\n", json);
free(json);

Write JSON to file

The path is the output JSON file path. This should be a null-terminated string using the system's native encoding. If path is NULL or invalid, returns false. If the file is not empty, its content is discarded.
The doc/val is the JSON document or root value. If it is NULL, returns false.
The flg is writer flag, pass 0 if you don't need it, see writer flag for details.
The alc is memory allocator, pass NULL if you don't need it, see memory allocator for details.
The err is a pointer to receive error message, pass NULL if you don't need it.
This function returns true on success, or false if an error occurs.

// doc -> file
bool yyjson_write_file(const char *path, const yyjson_doc *doc, yyjson_write_flag flg, const yyjson_alc *alc, yyjson_write_err *err);
// mut_doc -> file
bool yyjson_mut_write_file(const char *path, const yyjson_mut_doc *doc, yyjson_write_flag flg, const yyjson_alc *alc, yyjson_write_err *err);
// val -> file
bool yyjson_val_write_file(const char *path, const yyjson_val *val, yyjson_write_flag flg, const yyjson_alc *alc, yyjson_write_err *err);
// mut_val -> file
bool yyjson_mut_val_write_file(const char *path, const yyjson_mut_val *val, yyjson_write_flag flg, const yyjson_alc *alc, yyjson_write_err *err);

Sample code:

yyjson_doc *doc = yyjson_read_file("/tmp/test.json", 0, NULL, NULL);
bool suc = yyjson_write_file("tmp/test.json", doc, YYJSON_WRITE_PRETTY, NULL, NULL);
if (suc) printf("OK");

Write JSON to file pointer

The fp is the output file pointer. The data will be written to the current position of the file.
If fp is NULL or invalid, returns false.
The doc/val is the JSON document or root value. If it is NULL, returns false.
The flg is writer flag, pass 0 if you don't need it, see writer flag for details.
The alc is memory allocator, pass NULL if you don't need it, see memory allocator for details.
The err is a pointer to receive error message, pass NULL if you don't need it.
This function returns true on success, or false if an error occurs.

// doc -> file
bool yyjson_write_fp(FILE *fp, const yyjson_doc *doc, yyjson_write_flag flg, const yyjson_alc *alc, yyjson_write_err *err);
// mut_doc -> file
bool yyjson_mut_write_fp(FILE *fp, const yyjson_mut_doc *doc, yyjson_write_flag flg, const yyjson_alc *alc, yyjson_write_err *err);
// val -> file
bool yyjson_val_write_fp(FILE *fp, const yyjson_val *val, yyjson_write_flag flg, const yyjson_alc *alc, yyjson_write_err *err);
// mut_val -> file
bool yyjson_mut_val_write_fp(FILE *fp, const yyjson_mut_val *val, yyjson_write_flag flg, const yyjson_alc *alc, yyjson_write_err *err);

Sample code:

FILE *fp = fdopen(fd, "wb"); // POSIX file descriptor (fd)
bool suc = yyjson_write_fp(fp, doc, YYJSON_WRITE_PRETTY, NULL, NULL);
if (fp) fclose(fp);
if (suc) printf("OK");

Write JSON to buffer

The buf is the output buffer. If buf is NULL, returns 0.
The buf_len is the buffer length. If buf_len is too small, returns 0.
The doc/val is the JSON document or root value. If it is NULL, returns 0.
The flg is writer flag, pass 0 if you don't need it, see writer flag for details.
The err is a pointer to receive error message, pass NULL if you don't need it.
This function returns the number of bytes written (excluding the null terminator), or 0 on failure.

This function does not allocate memory, but the buffer must be larger than the final JSON size to allow temporary space.

The extra space is needed temporarily for each value while it is written, and is reused for later values:

Number: 40
String: 16 + (str_len * 6)
Other values: 16
Nesting depth: 16 * max_json_depth

// doc -> buffer
size_t yyjson_write_buf(char *buf, size_t buf_len, const yyjson_doc *doc, yyjson_write_flag flg, yyjson_write_err *err);
// mut_doc -> buffer
size_t yyjson_mut_write_buf(char *buf, size_t buf_len, const yyjson_mut_doc *doc, yyjson_write_flag flg, yyjson_write_err *err);
// val -> buffer
size_t yyjson_val_write_buf(char *buf, size_t buf_len, const yyjson_val *val, yyjson_write_flag flg, yyjson_write_err *err);
// mut_val -> buffer
size_t yyjson_mut_val_write_buf(char *buf, size_t buf_len, const yyjson_mut_val *val, yyjson_write_flag flg, yyjson_write_err *err);

Sample code:

char buf[512];
size_t len = yyjson_write_buf(buf, sizeof(buf), doc, YYJSON_WRITE_PRETTY, NULL);
if (len > 0) printf("OK, output:\n%s\n", buf);

Write JSON with options

The doc/val is the JSON document or root value. If it is NULL, returns NULL.
The flg is writer flag, pass 0 if you don't need it, see writer flag for details.
The alc is memory allocator, pass NULL if you don't need it, see memory allocator for details.
The len is a pointer to receive output length (not including the null-terminator), pass NULL if you don't need it.
The err is a pointer to receive error message, pass NULL if you don't need it.

This function returns a new JSON string, or NULL if an error occurs.
The string is encoded as UTF-8 with a null-terminator.
You should use free() or alc->free() to release it when it's no longer needed.

char *yyjson_write_opts(const yyjson_doc *doc, yyjson_write_flag flg, const yyjson_alc *alc, size_t *len, yyjson_write_err *err);
 
char *yyjson_mut_write_opts(const yyjson_mut_doc *doc, yyjson_write_flag flg, const yyjson_alc *alc, size_t *len, yyjson_write_err *err);
 
char *yyjson_val_write_opts(const yyjson_val *val, yyjson_write_flag flg, const yyjson_alc *alc, size_t *len, yyjson_write_err *err);
 
char *yyjson_mut_val_write_opts(const yyjson_mut_val *val, yyjson_write_flag flg, const yyjson_alc *alc, size_t *len, yyjson_write_err *err);

Sample code:

yyjson_doc *doc = ...;
 
// init an allocator with stack memory
char buf[64 * 1024];
yyjson_alc alc;
yyjson_alc_pool_init(&alc, buf, sizeof(buf));
 
// write
size_t len;
yyjson_write_err err;
char *json = yyjson_write_opts(doc, YYJSON_WRITE_PRETTY | YYJSON_WRITE_ESCAPE_UNICODE, &alc, &len, &err);
 
// get result
if (json) {
    printf("suc: %lu\n%s\n", len, json);
} else {
    printf("err: %u msg:%s\n", err.code, err.msg);
}
alc.free(alc.ctx, json);

The complete list of error codes (yyjson_write_code):

Code	Name	Description
0	YYJSON_WRITE_SUCCESS	Success, no error.
1	YYJSON_WRITE_ERROR_INVALID_PARAMETER	Invalid parameter, such as NULL document.
2	YYJSON_WRITE_ERROR_MEMORY_ALLOCATION	Memory allocation failure.
3	YYJSON_WRITE_ERROR_INVALID_VALUE_TYPE	Invalid value type in JSON document.
4	YYJSON_WRITE_ERROR_NAN_OR_INF	NaN or Infinity number occurs.
5	YYJSON_WRITE_ERROR_FILE_OPEN	Failed to open a file.
6	YYJSON_WRITE_ERROR_FILE_WRITE	Failed to write a file.
7	YYJSON_WRITE_ERROR_INVALID_STRING	Invalid unicode in string.
8	YYJSON_WRITE_ERROR_DEPTH	Nesting depth exceeded YYJSON_WRITER_DEPTH_LIMIT.

Writer flag

The library provides a set of flags for JSON writer.
You can use a single flag, or combine multiple flags with bitwise | operator.

YYJSON_WRITE_NOFLAG = 0

This is the default flag for JSON writer:

Writes JSON in minified format.
Reports an error on encountering inf or nan number.
Reports an error on encountering invalid UTF-8 strings.
Does not escape unicode or slashes.

YYJSON_WRITE_PRETTY

Writes JSON with a pretty format using a 4-space indent.

YYJSON_WRITE_PRETTY_TWO_SPACES

Writes JSON with a pretty format using a 2-space indent. This flag will override YYJSON_WRITE_PRETTY flag.

YYJSON_WRITE_ESCAPE_UNICODE

Escape unicode as \uXXXX, making the output ASCII-only, for example:

["Alizée, 😊"]

["Aliz\\u00E9e, \\uD83D\\uDE0A"]

YYJSON_WRITE_LOWERCASE_HEX

Use lowercase hex digits in \uXXXX escape sequences instead of the default uppercase. Only effective when YYJSON_WRITE_ESCAPE_UNICODE is also set.

YYJSON_WRITE_ESCAPE_SLASHES

Escapes the forward slash character / as \/, for example:

["https://github.com"]

["https:\/\/github.com"]

YYJSON_WRITE_ALLOW_INF_AND_NAN

Writes inf/nan numbers as Infinity and NaN literals instead of reporting errors.

Note that this output is NOT standard JSON and may be rejected by other JSON libraries, for example:

{"not_a_number":NaN,"large_number":Infinity}

YYJSON_WRITE_INF_AND_NAN_AS_NULL

Writes inf/nan numbers as null literals instead of reporting errors.
This flag will override YYJSON_WRITE_ALLOW_INF_AND_NAN flag, for example:

{"not_a_number":null,"large_number":null}

YYJSON_WRITE_ALLOW_INVALID_UNICODE

Allows invalid unicode when encoding string values.

Invalid characters within string values will be copied byte by byte. If YYJSON_WRITE_ESCAPE_UNICODE flag is also set, invalid characters will be escaped as \uFFFD (replacement character).

This flag does not affect the performance of correctly encoded strings.

YYJSON_WRITE_NEWLINE_AT_END

Adds a newline character \n at the end of the JSON. This can be helpful for text editors or NDJSON.

YYJSON_WRITE_FP_TO_FLOAT

Write floating-point numbers using single-precision (float). This casts double to float before serialization. This will produce shorter output, but may lose some precision. This flag is ignored if YYJSON_WRITE_FP_TO_FIXED(prec) is also used.

YYJSON_WRITE_FP_TO_FIXED(prec)

Write floating-point number using fixed-point notation. This is similar to ECMAScript Number.prototype.toFixed(prec), but with trailing zeros removed. The prec ranges from 1 to 15. This will produce shorter output but may lose some precision.

Accessing JSON Document

JSON Document

You can access the content of a document with the following functions:

// Get the root value of this JSON document.
yyjson_val *yyjson_doc_get_root(const yyjson_doc *doc);
 
// Get how many bytes are read when parsing JSON.
// e.g. "[1,2,3]" returns 7.
size_t yyjson_doc_get_read_size(const yyjson_doc *doc);
 
// Get total value count in this JSON document.
// e.g. "[1,2,3]" returns 4 (1 array and 3 numbers).
size_t yyjson_doc_get_val_count(const yyjson_doc *doc);

A document holds all the memory for its internal values and strings. When you no longer need it, you should release the document and free up all the memory:

// Free the document; if NULL is passed in, do nothing.

void yyjson_doc_free(yyjson_doc *doc);

JSON Value

Each JSON Value has a type and subtype, as specified in the table:

Type	Subtype
YYJSON_TYPE_NONE		Invalid value
YYJSON_TYPE_RAW		Raw string
YYJSON_TYPE_NULL		null literal
YYJSON_TYPE_BOOL	YYJSON_SUBTYPE_FALSE	false literal
YYJSON_TYPE_BOOL	YYJSON_SUBTYPE_TRUE	true literal
YYJSON_TYPE_NUM	YYJSON_SUBTYPE_UINT	uint64_t number
YYJSON_TYPE_NUM	YYJSON_SUBTYPE_SINT	int64_t number
YYJSON_TYPE_NUM	YYJSON_SUBTYPE_REAL	double number
YYJSON_TYPE_STR		String value
YYJSON_TYPE_STR	YYJSON_SUBTYPE_NOESC	String value, no-escape
YYJSON_TYPE_ARR		Array value
YYJSON_TYPE_OBJ		Object value

YYJSON_TYPE_NONE means invalid value, it does not appear when the JSON is successfully parsed.
YYJSON_TYPE_RAW only appears when the corresponding flag YYJSON_READ_XXX_AS_RAW is used.
YYJSON_SUBTYPE_NOESC is used to optimize the writing speed of strings that do not need to be escaped. This subtype is used internally, and the user does not need to handle it.

The following functions can be used to determine the type of JSON value.

// Returns the type and subtype of a JSON value.
// Returns 0 if `val` is NULL.
yyjson_type yyjson_get_type(const yyjson_val *val);
yyjson_subtype yyjson_get_subtype(const yyjson_val *val);
 
// Returns value's tag, see `Data Structures` doc for details.
uint8_t yyjson_get_tag(const yyjson_val *val);
 
// returns type description, such as:  
// "null", "string", "array", "object", "true", "false",
// "uint", "sint", "real", "unknown"
const char *yyjson_get_type_desc(const yyjson_val *val);
 
// Returns true if the JSON value is specified type.
// Returns false if `val` is NULL or type does not match.
bool yyjson_is_null(const yyjson_val *val);  // null
bool yyjson_is_true(const yyjson_val *val);  // true
bool yyjson_is_false(const yyjson_val *val); // false
bool yyjson_is_bool(const yyjson_val *val);  // true/false
bool yyjson_is_uint(const yyjson_val *val);  // uint64_t
bool yyjson_is_sint(const yyjson_val *val);  // int64_t
bool yyjson_is_int(const yyjson_val *val);   // uint64_t/int64_t
bool yyjson_is_real(const yyjson_val *val);  // double
bool yyjson_is_num(const yyjson_val *val);   // uint64_t/int64_t/double
bool yyjson_is_str(const yyjson_val *val);   // string
bool yyjson_is_arr(const yyjson_val *val);   // array
bool yyjson_is_obj(const yyjson_val *val);   // object
bool yyjson_is_ctn(const yyjson_val *val);   // array/object
bool yyjson_is_raw(const yyjson_val *val);   // raw string

The following functions can be used to get the contents of the JSON value.

// Returns the raw string, or NULL if `val` is not raw type.
const char *yyjson_get_raw(const yyjson_val *val);
 
// Returns bool value, or false if `val` is not bool type.
bool yyjson_get_bool(const yyjson_val *val);
 
// Returns uint64_t value (cast), or 0 if `val` is not uint/sint type.
uint64_t yyjson_get_uint(const yyjson_val *val);
 
// Returns int64_t value (cast), or 0 if `val` is not uint/sint type.
int64_t yyjson_get_sint(const yyjson_val *val);
 
// Returns int value (cast, may overflow), or 0 if `val` is not uint/sint type.
int yyjson_get_int(const yyjson_val *val);
 
// Returns double value, or 0 if `val` is not real type.
double yyjson_get_real(const yyjson_val *val);
 
// Returns double value (cast), or 0 if `val` is not uint/sint/real type.
double yyjson_get_num(const yyjson_val *val);
 
// Returns the string value, or NULL if `val` is not string type.
const char *yyjson_get_str(const yyjson_val *val);
 
// Returns the content length for raw/string/array/object values.
// Returns 0 if `val` is NULL. The return value is unspecified for other types.
size_t yyjson_get_len(const yyjson_val *val);
 
// Returns whether the value is equal to a string.
// Returns false if `val` is NULL or is not string.
bool yyjson_equals_str(const yyjson_val *val, const char *str);
bool yyjson_equals_strn(const yyjson_val *val, const char *str, size_t len);
 
// Returns whether two JSON values are equal (deep compare).
// Returns false if `lhs` or `rhs` is NULL.
// Note: result may be inaccurate if an object has duplicate keys.
// Warning: this function is recursive and may cause a stack overflow
//          if the object/array nesting level is too deep.
bool yyjson_equals(const yyjson_val *lhs, const yyjson_val *rhs);
bool yyjson_mut_equals(const yyjson_mut_val *lhs, const yyjson_mut_val *rhs);

The following functions can be used to modify the content of a JSON value.

Warning: For immutable documents, these functions will break the immutable convention, you should use this set of APIs with caution (e.g. make sure the document is only accessed in a single thread).

// Set the value to new type and content.
// Returns false if `val` is NULL or is object or array.
bool yyjson_set_raw(yyjson_val *val, const char *raw, size_t len);
bool yyjson_set_null(yyjson_val *val);
bool yyjson_set_bool(yyjson_val *val, bool num);
bool yyjson_set_uint(yyjson_val *val, uint64_t num);
bool yyjson_set_sint(yyjson_val *val, int64_t num);
bool yyjson_set_int(yyjson_val *val, int64_t num);
bool yyjson_set_float(yyjson_val *val, float num);
bool yyjson_set_double(yyjson_val *val, double num);
bool yyjson_set_real(yyjson_val *val, double num);
 
// The string is not copied, should be held by caller.
bool yyjson_set_str(yyjson_val *val, const char *str);
bool yyjson_set_strn(yyjson_val *val, const char *str, size_t len);

JSON Array

The following functions can be used to access a JSON array.

Note that accessing elements by index may take a linear search time. Therefore, if you need to iterate through an array, it is recommended to use the iterator API.

// Returns the number of elements in this array.
// Returns 0 if `arr` is NULL or is not an array.
size_t yyjson_arr_size(const yyjson_val *arr);
 
// Returns the element at the specified position (linear search time).
// Returns NULL if `arr` is NULL, is not an array, or `idx` is out of bounds.
yyjson_val *yyjson_arr_get(const yyjson_val *arr, size_t idx);
 
// Returns the first element of this array (constant time).
// Returns NULL if `arr` is NULL, is not an array, or is empty.
yyjson_val *yyjson_arr_get_first(const yyjson_val *arr);
 
// Returns the last element of this array (linear search time).
// Returns NULL if `arr` is NULL, is not an array, or is empty.
yyjson_val *yyjson_arr_get_last(const yyjson_val *arr);

JSON Array Iterator

There are two ways to traverse an array:

Sample code 1 (iterator API):

yyjson_val *arr; // the array to be traversed
 
yyjson_val *val;
yyjson_arr_iter iter = yyjson_arr_iter_with(arr);
while ((val = yyjson_arr_iter_next(&iter))) {
    your_func(val);
}

Sample code 2 (foreach macro):

yyjson_val *arr; // the array to be traversed
 
size_t idx, max;
yyjson_val *val;
yyjson_arr_foreach(arr, idx, max, val) {
    your_func(idx, val);
}

There's also a mutable version of the API to traverse a mutable array:

Sample code 1 (mutable iterator API):

yyjson_mut_val *arr; // the array to be traversed
 
yyjson_mut_val *val;
yyjson_mut_arr_iter iter = yyjson_mut_arr_iter_with(arr);
while ((val = yyjson_mut_arr_iter_next(&iter))) {
    if (your_val_is_unused(val)) {
        // you can remove current value inside iteration
        yyjson_mut_arr_iter_remove(&iter); 
    }
}

Sample code 2 (mutable foreach macro):

yyjson_mut_val *arr; // the array to be traversed
 
size_t idx, max;
yyjson_mut_val *val;
yyjson_mut_arr_foreach(arr, idx, max, val) {
    your_func(idx, val);
}

JSON Object

The following functions can be used to access a JSON object.

Note that accessing elements by key may take a linear search time. Therefore, if you need to iterate through an object, it is recommended to use the iterator API.

// Returns the number of key-value pairs in this object.
// Returns 0 if `obj` is NULL or is not an object.
size_t yyjson_obj_size(const yyjson_val *obj);
 
// Returns the value to which the specified key is mapped.
// Returns NULL if this object contains no mapping for the key.
yyjson_val *yyjson_obj_get(const yyjson_val *obj, const char *key);
yyjson_val *yyjson_obj_getn(const yyjson_val *obj, const char *key, size_t key_len);
 
// If the order of the object's keys is known at compile-time,
// you can use this method to avoid searching the entire object.
// e.g. { "x":1, "y":2, "z":3 }
yyjson_val *obj = ...;
yyjson_obj_iter iter = yyjson_obj_iter_with(obj);
 
yyjson_val *x = yyjson_obj_iter_get(&iter, "x");
yyjson_val *z = yyjson_obj_iter_get(&iter, "z");

JSON Object Iterator

There are two ways to traverse an object:

Sample code 1 (iterator API):

yyjson_val *obj; // the object to be traversed
 
yyjson_val *key, *val;
yyjson_obj_iter iter = yyjson_obj_iter_with(obj);
while ((key = yyjson_obj_iter_next(&iter))) {
    val = yyjson_obj_iter_get_val(key);
    your_func(key, val);
}

Sample code 2 (foreach macro):

yyjson_val *obj; // this is your object
 
size_t idx, max;
yyjson_val *key, *val;
yyjson_obj_foreach(obj, idx, max, key, val) {
    your_func(key, val);
}

There's also a mutable version of the API to traverse a mutable object:

Sample code 1 (mutable iterator API):

yyjson_mut_val *obj; // the object to be traversed
 
yyjson_mut_val *key, *val;
yyjson_mut_obj_iter iter = yyjson_mut_obj_iter_with(obj);
while ((key = yyjson_mut_obj_iter_next(&iter))) {
    val = yyjson_mut_obj_iter_get_val(key);
    if (your_key_is_unused(key)) {
        // you can remove current kv pair inside iteration
        yyjson_mut_obj_iter_remove(&iter);
    }
}

Sample code 2 (mutable foreach macro):

yyjson_mut_val *obj; // the object to be traversed
 
size_t idx, max;
yyjson_mut_val *key, *val;
yyjson_mut_obj_foreach(obj, idx, max, key, val) {
    your_func(key, val);
}

Creating JSON Document

The yyjson_mut_doc and related APIs are used to build JSON documents.

Please note that yyjson_mut_doc uses a memory pool to hold all strings and values. The pool can only be created, grown, or freed in its entirety. Therefore, yyjson_mut_doc is more suitable for write-once than mutation of an existing document.

JSON objects and arrays are composed of linked lists, so each yyjson_mut_val can only be added to one object or array.

Sample code:

// Build this JSON:
//     {
//        "page": 123,
//        "names": [ "Harry", "Ron", "Hermione" ]
//     }
 
// Create a mutable document.
yyjson_mut_doc *doc = yyjson_mut_doc_new(NULL);
 
// Create an object, the value's memory is held by doc.
yyjson_mut_val *root = yyjson_mut_obj(doc);
 
// Create key and value, add to the root object.
yyjson_mut_val *key = yyjson_mut_str(doc, "page");
yyjson_mut_val *num = yyjson_mut_int(doc, 123);
yyjson_mut_obj_add(root, key, num);
 
// Create 3 string values, add to the array object.
yyjson_mut_val *names = yyjson_mut_arr(doc);
yyjson_mut_val *name1 = yyjson_mut_str(doc, "Harry");
yyjson_mut_val *name2 = yyjson_mut_str(doc, "Ron");
yyjson_mut_val *name3 = yyjson_mut_str(doc, "Hermione");
yyjson_mut_arr_append(names, name1);
yyjson_mut_arr_append(names, name2);
yyjson_mut_arr_append(names, name3);
yyjson_mut_obj_add(root, yyjson_mut_str(doc, "names"), names);
 
// ❌ Wrong! the value is already added to another container.
yyjson_mut_obj_add(root, key, name1);
 
// Set the document's root value.
yyjson_mut_doc_set_root(doc, root);
 
// Write to JSON string
const char *json = yyjson_mut_write(doc, 0, NULL);
 
// Free the memory of doc and all values created from this doc.
yyjson_mut_doc_free(doc);

Mutable Document

The following functions are used to create, modify, copy, and destroy a JSON document.

// Creates and returns a new mutable JSON document.
// Returns NULL on error (e.g. memory allocation failure).
// If `alc` is NULL, the default allocator will be used.
yyjson_mut_doc *yyjson_mut_doc_new(const yyjson_alc *alc);
 
// Delete the JSON document, free the memory of this doc
// and all values created from this doc
void yyjson_mut_doc_free(yyjson_mut_doc *doc);
 
// Set the internal memory pool size (string length and value count).
// It can be used to reserve memory for the next string and value creation.
bool yyjson_mut_doc_set_str_pool_size(yyjson_mut_doc *doc, size_t len);
bool yyjson_mut_doc_set_val_pool_size(yyjson_mut_doc *doc, size_t count);
 
// Get or set the root value of this JSON document.
yyjson_mut_val *yyjson_mut_doc_get_root(yyjson_mut_doc *doc);
void yyjson_mut_doc_set_root(yyjson_mut_doc *doc, yyjson_mut_val *root);
 
// Copies and returns a new mutable document/value from input.
// Returns NULL on error (e.g. memory allocation failure).
 
// doc -> mut_doc
yyjson_mut_doc *yyjson_doc_mut_copy(const yyjson_doc *doc, const yyjson_alc *alc);
// val -> mut_val
yyjson_mut_val *yyjson_val_mut_copy(yyjson_mut_doc *doc, const yyjson_val *val);
// mut_doc -> mut_doc
yyjson_mut_doc *yyjson_mut_doc_mut_copy(const yyjson_mut_doc *doc, const yyjson_alc *alc);
// mut_val -> mut_val
yyjson_mut_val *yyjson_mut_val_mut_copy(yyjson_mut_doc *doc, const yyjson_mut_val *val);
// mut_doc -> doc
yyjson_doc *yyjson_mut_doc_imut_copy(const yyjson_mut_doc *doc, const yyjson_alc *alc);
// mut_val -> doc
yyjson_doc *yyjson_mut_val_imut_copy(const yyjson_mut_val *val, const yyjson_alc *alc);

JSON Value Creation

The following functions are used to create mutable JSON value, the value's memory is held by the document.

// Creates and returns a new value, returns NULL on error.
yyjson_mut_val *yyjson_mut_null(yyjson_mut_doc *doc);
yyjson_mut_val *yyjson_mut_true(yyjson_mut_doc *doc);
yyjson_mut_val *yyjson_mut_false(yyjson_mut_doc *doc);
yyjson_mut_val *yyjson_mut_bool(yyjson_mut_doc *doc, bool val);
yyjson_mut_val *yyjson_mut_uint(yyjson_mut_doc *doc, uint64_t num);
yyjson_mut_val *yyjson_mut_sint(yyjson_mut_doc *doc, int64_t num);
yyjson_mut_val *yyjson_mut_int(yyjson_mut_doc *doc, int64_t num);
yyjson_mut_val *yyjson_mut_float(yyjson_mut_doc *doc, float num);
yyjson_mut_val *yyjson_mut_double(yyjson_mut_doc *doc, double num);
yyjson_mut_val *yyjson_mut_real(yyjson_mut_doc *doc, double num);
 
// Creates a string value, the input string is NOT copied.
yyjson_mut_val *yyjson_mut_str(yyjson_mut_doc *doc, const char *str);
yyjson_mut_val *yyjson_mut_strn(yyjson_mut_doc *doc, const char *str, size_t len);
 
// Creates a string value, the input string is copied and held by the document.
yyjson_mut_val *yyjson_mut_strcpy(yyjson_mut_doc *doc, const char *str);
yyjson_mut_val *yyjson_mut_strncpy(yyjson_mut_doc *doc, const char *str, size_t len);

JSON Array Creation

The following functions are used to create mutable JSON array.

// Creates and returns an empty mutable array, returns NULL on error.
yyjson_mut_val *yyjson_mut_arr(yyjson_mut_doc *doc);
 
// Creates and returns a mutable array with a C array.
yyjson_mut_val *yyjson_mut_arr_with_bool(yyjson_mut_doc *doc, bool *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_sint(yyjson_mut_doc *doc, int64_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_uint(yyjson_mut_doc *doc, uint64_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_real(yyjson_mut_doc *doc, double *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_sint8(yyjson_mut_doc *doc, int8_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_sint16(yyjson_mut_doc *doc, int16_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_sint32(yyjson_mut_doc *doc, int32_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_sint64(yyjson_mut_doc *doc, int64_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_uint8(yyjson_mut_doc *doc, uint8_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_uint16(yyjson_mut_doc *doc, uint16_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_uint32(yyjson_mut_doc *doc, uint32_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_uint64(yyjson_mut_doc *doc, uint64_t *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_float(yyjson_mut_doc *doc, float *vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_double(yyjson_mut_doc *doc, double *vals, size_t count);
// sample code:
int32_t vals[3] = {-1, 0, 1};
yyjson_mut_val *arr = yyjson_mut_arr_with_sint32(doc, vals, 3);
 
// Creates and returns a mutable array with strings,
// the strings should be encoded as UTF-8.
yyjson_mut_val *yyjson_mut_arr_with_str(yyjson_mut_doc *doc, const char **vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_strn(yyjson_mut_doc *doc, const char **vals, const size_t *lens, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_strcpy(yyjson_mut_doc *doc, const char **vals, size_t count);
yyjson_mut_val *yyjson_mut_arr_with_strncpy(yyjson_mut_doc *doc, const char **vals, const size_t *lens, size_t count);
// sample code:
const char *strs[3] = {"Jan", "Feb", "Mar"};
yyjson_mut_val *arr = yyjson_mut_arr_with_str(doc, strs, 3);

JSON Array Modification

The following functions are used to modify the contents of a JSON array.

// Inserts a value into an array at a given index.
// Returns false on error (e.g. out of bounds).
// Note that this function takes a linear search time.
bool yyjson_mut_arr_insert(yyjson_mut_val *arr, yyjson_mut_val *val, size_t idx);
 
// Inserts a val at the end of the array, returns false on error.
bool yyjson_mut_arr_append(yyjson_mut_val *arr, yyjson_mut_val *val);
 
// Inserts a val at the head of the array, returns false on error.
bool yyjson_mut_arr_prepend(yyjson_mut_val *arr, yyjson_mut_val *val);
 
// Replaces a value at index and returns old value, returns NULL on error.
// Note that this function takes a linear search time.
yyjson_mut_val *yyjson_mut_arr_replace(yyjson_mut_val *arr, size_t idx, yyjson_mut_val *val);
 
// Removes and returns a value at index, returns NULL on error.
// Note that this function takes a linear search time.
yyjson_mut_val *yyjson_mut_arr_remove(yyjson_mut_val *arr, size_t idx);
 
// Removes and returns the first value in this array, returns NULL on error.
yyjson_mut_val *yyjson_mut_arr_remove_first(yyjson_mut_val *arr);
 
// Removes and returns the last value in this array, returns NULL on error.
yyjson_mut_val *yyjson_mut_arr_remove_last(yyjson_mut_val *arr);
 
// Removes all values within a specified range in the array.
// Note that this function takes a linear search time.
bool yyjson_mut_arr_remove_range(yyjson_mut_val *arr, size_t idx, size_t len);
 
// Removes all values in this array.
bool yyjson_mut_arr_clear(yyjson_mut_val *arr);
 
// Convenience API:
// Adds a value at the end of this array, returns false on error.
bool yyjson_mut_arr_add_val(yyjson_mut_val *arr, yyjson_mut_val *val);
bool yyjson_mut_arr_add_null(yyjson_mut_doc *doc, yyjson_mut_val *arr);
bool yyjson_mut_arr_add_true(yyjson_mut_doc *doc, yyjson_mut_val *arr);
bool yyjson_mut_arr_add_false(yyjson_mut_doc *doc, yyjson_mut_val *arr);
bool yyjson_mut_arr_add_bool(yyjson_mut_doc *doc, yyjson_mut_val *arr, bool val);
bool yyjson_mut_arr_add_uint(yyjson_mut_doc *doc, yyjson_mut_val *arr, uint64_t num);
bool yyjson_mut_arr_add_sint(yyjson_mut_doc *doc, yyjson_mut_val *arr, int64_t num);
bool yyjson_mut_arr_add_int(yyjson_mut_doc *doc, yyjson_mut_val *arr, int64_t num);
bool yyjson_mut_arr_add_float(yyjson_mut_doc *doc, yyjson_mut_val *arr, float num);
bool yyjson_mut_arr_add_double(yyjson_mut_doc *doc, yyjson_mut_val *arr, double num);
bool yyjson_mut_arr_add_real(yyjson_mut_doc *doc, yyjson_mut_val *arr, double num);
bool yyjson_mut_arr_add_str(yyjson_mut_doc *doc, yyjson_mut_val *arr, const char *str);
bool yyjson_mut_arr_add_strn(yyjson_mut_doc *doc, yyjson_mut_val *arr, const char *str, size_t len);
bool yyjson_mut_arr_add_strcpy(yyjson_mut_doc *doc, yyjson_mut_val *arr, const char *str);
bool yyjson_mut_arr_add_strncpy(yyjson_mut_doc *doc, yyjson_mut_val *arr, const char *str, size_t len);
 
// Convenience API:
// Creates and adds a new array at the end of the array.
// Returns the new array, or NULL on error.
yyjson_mut_val *yyjson_mut_arr_add_arr(yyjson_mut_doc *doc, yyjson_mut_val *arr);
 
// Convenience API:
// Creates and adds a new object at the end of the array.
// Returns the new object, or NULL on error.
yyjson_mut_val *yyjson_mut_arr_add_obj(yyjson_mut_doc *doc, yyjson_mut_val *arr);

JSON Object Creation

The following functions are used to create mutable JSON object.

// Creates and returns a mutable object, returns NULL on error.
yyjson_mut_val *yyjson_mut_obj(yyjson_mut_doc *doc);
 
// Creates and returns a mutable object with keys and values,
// returns NULL on error. The keys and values are NOT copied.
// The strings should be encoded as UTF-8 with null-terminator.
yyjson_mut_val *yyjson_mut_obj_with_str(yyjson_mut_doc *doc,
                                        const char **keys,
                                        const char **vals,
                                        size_t count);
// sample code:
const char *keys[] = {"name", "type", "id"};
const char *vals[] = {"Harry", "student", "123456"};
yyjson_mut_obj_with_str(doc, keys, vals, 3);
 
// Creates and returns a mutable object with key-value pairs,
// returns NULL on error. The keys and values are NOT copied.
// The strings should be encoded as UTF-8 with null-terminator.
yyjson_mut_val *yyjson_mut_obj_with_kv(yyjson_mut_doc *doc,
                                       const char **kv_pairs,
                                       size_t pair_count);
// sample code:
const char *pairs[] = {"name", "Harry", "type", "student", "id", "123456"};
yyjson_mut_obj_with_kv(doc, pairs, 3);

JSON Object Modification

The following functions are used to modify the contents of a JSON object.

// Adds a key-value pair at the end of the object. 
// The key must be a string value.
// This function allows duplicate keys in one object.
bool yyjson_mut_obj_add(yyjson_mut_val *obj, yyjson_mut_val *key, yyjson_mut_val *val);
 
// Adds a key-value pair to the object.
// The key must be a string value.
// This function may remove all key-value pairs for the given key before adding.
// Note that this function takes a linear search time.
bool yyjson_mut_obj_put(yyjson_mut_val *obj, yyjson_mut_val *key, yyjson_mut_val *val);
 
// Removes a key-value pair from the object with a given key.
// Note that this function takes a linear search time.
bool yyjson_mut_obj_remove(yyjson_mut_val *obj, yyjson_mut_val *key);
 
// Removes all key-value pairs in this object.
bool yyjson_mut_obj_clear(yyjson_mut_val *obj);
 
// Convenience API:
// Adds a key-value pair at the end of the object. The key is not copied.
// Note that these functions allow duplicate keys in one object.
bool yyjson_mut_obj_add_null(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key);
bool yyjson_mut_obj_add_true(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key);
bool yyjson_mut_obj_add_false(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key);
bool yyjson_mut_obj_add_bool(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, bool val);
bool yyjson_mut_obj_add_uint(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, uint64_t val);
bool yyjson_mut_obj_add_sint(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, int64_t val);
bool yyjson_mut_obj_add_int(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, int64_t val);
bool yyjson_mut_obj_add_float(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, float val);
bool yyjson_mut_obj_add_double(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, double val);
bool yyjson_mut_obj_add_real(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, double val);
bool yyjson_mut_obj_add_str(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, const char *val);
bool yyjson_mut_obj_add_strn(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, const char *val, size_t len);
bool yyjson_mut_obj_add_strcpy(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, const char *val);
bool yyjson_mut_obj_add_strncpy(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, const char *val, size_t len);
yyjson_mut_val *yyjson_mut_obj_add_arr(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *_key);
yyjson_mut_val *yyjson_mut_obj_add_obj(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *_key);
                              
// Convenience API:
// Removes all key-value pairs for the given key.
// Note that this function takes a linear search time.
bool yyjson_mut_obj_remove_str(yyjson_mut_val *obj, const char *key);
bool yyjson_mut_obj_remove_strn(yyjson_mut_val *obj, const char *key, size_t len);
 
// Convenience API:
// Replaces all matching keys with the new key.
// Returns true if at least one key was renamed.
// This function takes a linear search time.
yyjson_api_inline bool yyjson_mut_obj_rename_key(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, const char *new_key);
yyjson_api_inline bool yyjson_mut_obj_rename_keyn(yyjson_mut_doc *doc, yyjson_mut_val *obj, const char *key, size_t len, const char *new_key, size_t new_len);

JSON Pointer and Patch

JSON Pointer

The library supports querying JSON values using JSON Pointer (RFC 6901).

// `JSON pointer` is a null-terminated string.
yyjson_val *yyjson_ptr_get(const yyjson_val *val, const char *ptr);
yyjson_val *yyjson_doc_ptr_get(const yyjson_doc *doc, const char *ptr);
yyjson_mut_val *yyjson_mut_ptr_get(const yyjson_mut_val *val, const char *ptr);
yyjson_mut_val *yyjson_mut_doc_ptr_get(const yyjson_mut_doc *doc, const char *ptr);
 
// `JSON pointer` with string length, allow NUL (Unicode U+0000) characters inside.
yyjson_val *yyjson_ptr_getn(const yyjson_val *val, const char *ptr, size_t len);
yyjson_val *yyjson_doc_ptr_getn(const yyjson_doc *doc, const char *ptr, size_t len);
yyjson_mut_val *yyjson_mut_ptr_getn(const yyjson_mut_val *val, const char *ptr, size_t len);
yyjson_mut_val *yyjson_mut_doc_ptr_getn(const yyjson_mut_doc *doc, const char *ptr, size_t len);
 
// `JSON pointer` with string length, context and error information.
yyjson_val *yyjson_ptr_getx(const yyjson_val *val, const char *ptr, size_t len, yyjson_ptr_err *err);
yyjson_val *yyjson_doc_ptr_getx(const yyjson_doc *doc, const char *ptr, size_t len, yyjson_ptr_err *err);
yyjson_mut_val *yyjson_mut_ptr_getx(const yyjson_mut_val *val, const char *ptr, size_t len, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);
yyjson_mut_val *yyjson_mut_doc_ptr_getx(const yyjson_mut_doc *doc, const char *ptr, size_t len, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);

For example, given the JSON document:

{
    "size" : 3,
    "users" : [
        {"id": 1, "name": "Harry"},
        {"id": 2, "name": "Ron"},
        {"id": 3, "name": "Hermione"}
    ]
}

The following JSON strings evaluate to the accompanying values:

Pointer	Matched Value
""	the whole document
"/size"	3
"/users/0"	{"id": 1, "name": "Harry"}
"/users/1/name"	"Ron"
"/no_match"	NULL
"no_slash"	NULL
"/"	NULL (match to empty key: root[""])

yyjson_doc *doc = ...;
yyjson_val *val = yyjson_doc_ptr_get(doc, "/users/1/name");
printf("%s\n", yyjson_get_str(val)); // Ron
 
yyjson_ptr_err err;
yyjson_val *val2 = yyjson_doc_ptr_getx(doc, "/", 1, &err);
if (!val2) printf("err %d: %s\n", err.code, err.msg); // err 3: cannot be resolved

The library also supports modifying JSON values using JSON Pointer.

// Add or insert a new value.
bool yyjson_mut_ptr_add(yyjson_mut_val *val, const char *ptr, yyjson_mut_val *new_val, yyjson_mut_doc *doc);
bool yyjson_mut_ptr_addn(yyjson_mut_val *val, const char *ptr, size_t len, yyjson_mut_val *new_val, yyjson_mut_doc *doc);
bool yyjson_mut_ptr_addx(yyjson_mut_val *val, const char *ptr, size_t len, yyjson_mut_val *new_val, yyjson_mut_doc *doc, bool create_parent, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);
                                           
bool yyjson_mut_doc_ptr_add(yyjson_mut_doc *doc, const char *ptr, yyjson_mut_val *new_val);
bool yyjson_mut_doc_ptr_addn(yyjson_mut_doc *doc, const char *ptr, size_t len, yyjson_mut_val *new_val);
bool yyjson_mut_doc_ptr_addx(yyjson_mut_doc *doc, const char *ptr, size_t len, yyjson_mut_val *new_val, bool create_parent, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);
 
// Set a new value (add if it doesn't exist, replace if it does).
bool yyjson_mut_ptr_set(yyjson_mut_val *val, const char *ptr, yyjson_mut_val *new_val, yyjson_mut_doc *doc);
bool yyjson_mut_ptr_setn(yyjson_mut_val *val, const char *ptr, size_t len, yyjson_mut_val *new_val, yyjson_mut_doc *doc);
bool yyjson_mut_ptr_setx(yyjson_mut_val *val, const char *ptr, size_t len, yyjson_mut_val *new_val, yyjson_mut_doc *doc, bool create_parent, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);
                                             
bool yyjson_mut_doc_ptr_set(yyjson_mut_doc *doc, const char *ptr, yyjson_mut_val *new_val);
bool yyjson_mut_doc_ptr_setn(yyjson_mut_doc *doc, const char *ptr, size_t len, yyjson_mut_val *new_val);
bool yyjson_mut_doc_ptr_setx(yyjson_mut_doc *doc, const char *ptr, size_t len, yyjson_mut_val *new_val, bool create_parent, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);
 
// Replace an existing value.
yyjson_mut_val *yyjson_mut_ptr_replace(yyjson_mut_val *val, const char *ptr, yyjson_mut_val *new_val);
yyjson_mut_val *yyjson_mut_ptr_replacen(yyjson_mut_val *val, const char *ptr, size_t len, yyjson_mut_val *new_val);
yyjson_mut_val *yyjson_mut_ptr_replacex(yyjson_mut_val *val, const char *ptr, size_t len, yyjson_mut_val *new_val, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);
    
yyjson_mut_val *yyjson_mut_doc_ptr_replace(yyjson_mut_doc *doc, const char *ptr, yyjson_mut_val *new_val);
yyjson_mut_val *yyjson_mut_doc_ptr_replacen(yyjson_mut_doc *doc, const char *ptr, size_t len, yyjson_mut_val *new_val);
yyjson_mut_val *yyjson_mut_doc_ptr_replacex(yyjson_mut_doc *doc, const char *ptr, size_t len, yyjson_mut_val *new_val, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);
 
// Remove an existing value.
yyjson_mut_val *yyjson_mut_ptr_remove(yyjson_mut_val *val, const char *ptr);
yyjson_mut_val *yyjson_mut_ptr_removen(yyjson_mut_val *val, const char *ptr, size_t len);
yyjson_mut_val *yyjson_mut_ptr_removex(yyjson_mut_val *val, const char *ptr, size_t len, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);
 
yyjson_mut_val *yyjson_mut_doc_ptr_remove(yyjson_mut_doc *doc, const char *ptr);
yyjson_mut_val *yyjson_mut_doc_ptr_removen(yyjson_mut_doc *doc, const char *ptr, size_t len);
yyjson_mut_val *yyjson_mut_doc_ptr_removex(yyjson_mut_doc *doc, const char *ptr, size_t len, yyjson_ptr_ctx *ctx, yyjson_ptr_err *err);

For example:

yyjson_mut_doc *doc = ...;
// doc: {"a":0,"b":[1,2,3]}
 
yyjson_mut_doc_ptr_set(doc, "/a", yyjson_mut_int(doc, 9));
// now: {"a":9,"b":[1,2,3]}
 
yyjson_mut_doc_ptr_add(doc, "/b/-", yyjson_mut_int(doc, 4));
// now: {"a":9,"b":[1,2,3,4]}
 
yyjson_mut_doc_ptr_remove(doc, "/b");
// now: {"a":9}

All the above functions ending with x can be used to get the result context ctx, and the error message err. For example:

// doc: {"a":0,"b":[null,2,3]}
yyjson_mut_doc *doc = ...;
 
// get error code and message
yyjson_ptr_err err;
yyjson_mut_doc_ptr_setx(doc, "/b/99", 4, yyjson_mut_int(doc, 99), true, NULL, &err);
if (err.code) printf("err: %s\n", err.msg); // err: cannot resolve
 
// get target value's context
// perform some operations without reparsing the JSON Pointer
yyjson_mut_val *val = yyjson_mut_doc_ptr_getx(doc, "/b/0", 4, &ctx, &err);
if (yyjson_mut_is_null(val)) yyjson_ptr_ctx_remove(&ctx);
// now: {"a":0,"b":[2,3]}

JSON Patch

The library supports JSON Patch (RFC 6902). Specification and example: https://tools.ietf.org/html/rfc6902

// Creates and returns a patched JSON value.
// Returns NULL if the patch could not be applied.
yyjson_mut_val *yyjson_patch(yyjson_mut_doc *doc,
                             const yyjson_val *orig,
                             const yyjson_val *patch,
                             yyjson_patch_err *err);
 
yyjson_mut_val *yyjson_mut_patch(yyjson_mut_doc *doc,
                                 const yyjson_mut_val *orig,
                                 const yyjson_mut_val *patch,
                                 yyjson_patch_err *err);

JSON Merge Patch

The library supports JSON Merge Patch (RFC 7386). Specification and example: https://tools.ietf.org/html/rfc7386

// Creates and returns a merge-patched JSON value.
// Returns NULL if the patch could not be applied.
yyjson_mut_val *yyjson_merge_patch(yyjson_mut_doc *doc,
                                   const yyjson_val *orig,
                                   const yyjson_val *patch);
 
yyjson_mut_val *yyjson_mut_merge_patch(yyjson_mut_doc *doc,
                                       const yyjson_mut_val *orig,
                                       const yyjson_mut_val *patch);

Number Processing

Number reader

The library has a built-in high-performance number reader,
it will read numbers according to these rules by default:

Positive integers are read as uint64_t. If an overflow occurs, it is converted to double.
Negative integers are read as int64_t. If an overflow occurs, it is converted to double.
Floating-point numbers are read as double with correct rounding.
If a double number overflows (reaches infinity), an error is reported.
If a number does not conform to the JSON standard, an error is reported.

There are 3 flags that can be used to adjust the number parsing strategy:

YYJSON_READ_ALLOW_INF_AND_NAN: read nan/inf number or literal as double (non-standard).
YYJSON_READ_NUMBER_AS_RAW: read all numbers as raw strings without parsing.
YYJSON_READ_BIGNUM_AS_RAW: read big numbers (overflow or infinity) as raw strings without parsing.

See the Reader flag section for more details.

Number writer

The library has a built-in high-performance number writer,
it will write numbers according to these rules by default:

Positive integers are written without a sign.
Negative integers are written with a negative sign.
Floating-point numbers are written using the ECMAScript format, with the following modifications:
- If the number is Infinity or NaN, an error is reported.
- The negative sign of -0.0 is preserved to maintain input information.
- The positive sign in the exponent part is removed.
The floating-point number writer will generate the shortest correctly rounded decimal representation.

There are several flags that can be used to adjust the number writing strategy:

YYJSON_WRITE_ALLOW_INF_AND_NAN writes inf/nan numbers as Infinity and NaN literals without error (non-standard).
YYJSON_WRITE_INF_AND_NAN_AS_NULL writes inf/nan numbers as null literal.
YYJSON_WRITE_FP_TO_FLOAT writes real numbers as float instead of double.
YYJSON_WRITE_FP_TO_FIXED(prec) writes real numbers using fixed-point notation.

See the Writer flag section for more details.

There are also some helper functions to control the output format of individual values:

yyjson_set_fp_to_float(yyjson_val *val, bool flt) and yyjson_mut_set_fp_to_float(yyjson_mut_val *val, bool flt) write this real number with float or double precision.
yyjson_set_fp_to_fixed(yyjson_val *val, int prec) and yyjson_mut_set_fp_to_fixed(yyjson_mut_val *val, int prec) write this real number using fixed-point notation, the prec should be in the range of 1 to 15.

Number conversion function

There are also two utility functions that provide direct access to the library's internal number conversion logic.
They are intended for standalone use and typically do not allocate memory.

// parse a number from string
const char *yyjson_read_number(const char *dat,
                               yyjson_val *val,
                               yyjson_read_flag flg,
                               const yyjson_alc *alc,
                               yyjson_read_err *err);
// write a number to string
char *yyjson_write_number(const yyjson_val *val, char *buf);

Text Processing

Character Encoding

By default, this library supports UTF-8 encoding without a BOM, as specified in RFC 8259:

JSON text exchanged between systems that are not part of a closed ecosystem MUST be encoded using UTF-8. Implementations MUST NOT add a byte order mark (U+FEFF) to the beginning of a networked-transmitted JSON text.

This library performs strict UTF-8 encoding validation on input strings by default. If an invalid character is encountered, an error will be reported.

To allow a BOM, use the YYJSON_READ_ALLOW_BOM or YYJSON_READ_ALLOW_EXT_WHITESPACE flags.

To allow invalid Unicode encoding, use the YYJSON_READ_ALLOW_INVALID_UNICODE and YYJSON_WRITE_ALLOW_INVALID_UNICODE flags. Note: Enabling these flags may result in yyjson producing values that contain invalid characters, which could be processed by other code and potentially introduce security risks.

To mark a string as not requiring escaping during JSON writing, use yyjson_set_str_noesc(yyjson_val *val, bool noesc) or yyjson_mut_set_str_noesc(yyjson_mut_val *val, bool noesc). This can improve string-writing performance and preserve the original string bytes.

NUL Character

This library supports the NUL character (also known as the null terminator, or Unicode U+0000, ASCII \0) inside strings.

When reading JSON, \u0000 will be unescaped to NUL character. If a string contains the NUL character, the length obtained with strlen() will be inaccurate, and you should use yyjson_get_len() to get the actual length.

When building JSON, the input string is treated as null-terminated by default. If you need to pass in a string that contains the NUL character, you should use the API with the n suffix and provide the actual length of the string.

For example:

// null-terminated string
yyjson_mut_str(doc, str);
yyjson_obj_get(obj, str);
 
// any string, with or without null terminator
yyjson_mut_strn(doc, str, len);
yyjson_obj_getn(obj, str, len);
 
// C++ string
std::string sstr = ...;
yyjson_obj_getn(obj, sstr.data(), sstr.length());

Memory Allocator

The library does not directly call libc's memory allocation functions (malloc/realloc/free). Instead, when memory allocation is required, yyjson's API takes a parameter named alc that allows the caller to pass in an allocator. If the alc is NULL, yyjson will use the default memory allocator, which is a simple wrapper of libc's functions.

Using a custom memory allocator allows you to have more control over memory allocation, here are a few examples:

Single allocator for multiple JSON

If you need to parse multiple small JSON one by one, you can use a single allocator to avoid multiple memory allocations.

Sample code:

// max data size for single JSON
size_t max_json_size = 64 * 1024;
// calculate the max memory usage for a single JSON
size_t buf_size = yyjson_read_max_memory_usage(max_json_size, 0);
// create a buffer for allocator
void *buf = malloc(buf_size);
// set up the allocator with buffer
yyjson_alc alc;
yyjson_alc_pool_init(&alc, buf, buf_size);
 
// read multiple JSON using one allocator
for (int i = 0; i < your_json_file_count; i++) {
    const char *your_json_file_path = ...;
    yyjson_doc *doc = yyjson_read_file(your_json_file_path, 0, &alc, NULL);
    ...
    yyjson_doc_free(doc);
}
 
// free the buffer
free(buf);

If you are not sure about the amount of memory required to process JSON, you can use the dynamic allocator.

// create a dynamic allocator
yyjson_alc *alc = yyjson_alc_dyn_new();
 
// read multiple JSON using one allocator
for (int i = 0; i < your_json_file_count; i++) {
    const char *your_json_file_path = ...;
    yyjson_doc *doc = yyjson_read_file(your_json_file_path, 0, alc, NULL);
    ...
    yyjson_doc_free(doc);
}
 
// free the allocator
yyjson_alc_dyn_free(alc);

Stack memory allocator

If the JSON is small enough, you can use stack memory to read or write it.

Sample code:

char buf[128 * 1024]; // stack buffer
yyjson_alc alc;
yyjson_alc_pool_init(&alc, buf, sizeof(buf));
 
yyjson_doc *doc = yyjson_read_opts(dat, len, 0, &alc, NULL);
...
yyjson_doc_free(doc); // this is optional, as the memory is on stack

Use a third-party allocator library

You can use a third-party high-performance memory allocator for yyjson, such as jemalloc, tcmalloc, mimalloc. You can also refer to the following code to implement your own allocator.

Sample code:

// Use https://github.com/microsoft/mimalloc
 
#include <mimalloc.h>
 
// same as malloc(size)
static void *priv_malloc(void *ctx, size_t size) {
    return mi_malloc(size);
}
 
// same as realloc(ptr, size)
// `old_size` is the size of the originally allocated memory
static void *priv_realloc(void *ctx, void *ptr, size_t old_size, size_t size) {
    return mi_realloc(ptr, size);
}
 
// same as free(ptr)
static void priv_free(void *ctx, void *ptr) {
    mi_free(ptr);
}
 
// the allocator object
static const yyjson_alc PRIV_ALC = {
    priv_malloc,
    priv_realloc,
    priv_free,
    NULL // `ctx` which will be passed into the functions above
};
 
// Read with custom allocator
yyjson_doc *doc = yyjson_read_opts(dat, len, 0, &PRIV_ALC, NULL);
...
yyjson_doc_free(doc);
 
// Write with custom allocator
yyjson_alc *alc = &PRIV_ALC;
char *json = yyjson_write_opts(doc, 0, alc, NULL, NULL);
...
alc->free(alc->ctx, json);

Stack Memory Usage

Most functions in the library use fixed-size stack memory. This includes functions for JSON reading and writing, as well as JSON Pointer handling.

However, a few functions use recursion and may cause a stack overflow if the nesting level is too deep. These functions are marked with the following warning in the header file:

Warning
This function is recursive and may cause a stack overflow if the object level is too deep.

Null Check

The library's public APIs perform a null check for every input parameter to prevent crashes.

For example, when reading a JSON, you don't need to perform null checks or type checks on each value:

yyjson_doc *doc = yyjson_read(NULL, 0, 0); // doc is NULL
yyjson_val *val = yyjson_doc_get_root(doc); // val is NULL
const char *str = yyjson_get_str(val); // str is NULL
if (!str) printf("err!");
yyjson_doc_free(doc); // do nothing

However, if you are certain that a value is non-null and matches the expected type, you can use the unsafe prefix API to avoid the null check.

For example, when iterating over an array or object, the value and key must be non-null:

size_t idx, max;
yyjson_val *key, *val;
yyjson_obj_foreach(obj, idx, max, key, val) {
    // this is a valid JSON, so the key must be a valid string
    if (unsafe_yyjson_equals_str(key, "id") &&
        unsafe_yyjson_is_uint(val) &&
        unsafe_yyjson_get_uint(val) == 1234) {
        ...
    }
}

Thread Safety

The library does not use global variables. Therefore, if you can ensure that the input parameters of a function are thread-safe, then the function calls are also thread-safe.

In general, yyjson_doc and yyjson_val are immutable and thread-safe, while yyjson_mut_doc and yyjson_mut_val are mutable and not thread-safe.

Locale Independence

The library is designed to be locale-independent.

However, there are certain conditions that you should be aware of:

You use libc's setlocale() function to change the locale.
Your environment does not adhere to the IEEE 754 floating-point standard (e.g. some IBM mainframes), or you explicitly set YYJSON_DISABLE_FAST_FP_CONV during build, in which case yyjson will use strtod() to parse floating-point numbers.

If both of these conditions are met, it is recommended to avoid calling setlocale() while another thread is parsing JSON. Otherwise, an error may be returned during JSON floating-point number parsing.