Leahnaya/pokeplatinum
1
0
Fork 0
mirror of https://github.com/pret/pokeplatinum.git synced 2026-03-21 17:55:13 -05:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
41c0c2f864
pokeplatinum/subprojects/packagefiles/yyjson_patch/feat-value-spans.patch

1030 lines
39 KiB
Diff
Raw Blame History

diff --git a/src/yyjson.c b/src/yyjson.c
index c16d925..e67cbaf 100644
--- a/src/yyjson.c
+++ b/src/yyjson.c
@@ -18,6 +18,9 @@
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
+
+ Additional changes to add per-value byte-spans made by <lhearachel@proton.me>
+ Copyright (c) 2025
*============================================================================*/
#include "yyjson.h"
@@ -3101,10 +3104,12 @@ static_inline bool has_rflag(yyjson_read_flag flg, yyjson_read_flag chk,
*============================================================================*/
/** Read `true` literal, `*ptr[0]` should be `t`. */
-static_inline bool read_true(u8 **ptr, yyjson_val *val) {
+static_inline bool read_true(u8 **ptr, yyjson_val *val, u8 *beg) {
u8 *cur = *ptr;
if (likely(byte_match_4(cur, "true"))) {
val->tag = YYJSON_TYPE_BOOL | YYJSON_SUBTYPE_TRUE;
+ val->beg = cur - beg;
+ val->end = val->beg + 3;
*ptr = cur + 4;
return true;
}
@@ -3112,10 +3117,12 @@ static_inline bool read_true(u8 **ptr, yyjson_val *val) {
}
/** Read `false` literal, `*ptr[0]` should be `f`. */
-static_inline bool read_false(u8 **ptr, yyjson_val *val) {
+static_inline bool read_false(u8 **ptr, yyjson_val *val, u8 *beg) {
u8 *cur = *ptr;
if (likely(byte_match_4(cur + 1, "alse"))) {
val->tag = YYJSON_TYPE_BOOL | YYJSON_SUBTYPE_FALSE;
+ val->beg = cur - beg;
+ val->end = val->beg + 4;
*ptr = cur + 5;
return true;
}
@@ -3123,10 +3130,12 @@ static_inline bool read_false(u8 **ptr, yyjson_val *val) {
}
/** Read `null` literal, `*ptr[0]` should be `n`. */
-static_inline bool read_null(u8 **ptr, yyjson_val *val) {
+static_inline bool read_null(u8 **ptr, yyjson_val *val, u8 *beg) {
u8 *cur = *ptr;
if (likely(byte_match_4(cur, "null"))) {
val->tag = YYJSON_TYPE_NULL;
+ val->beg = cur - beg;
+ val->end = val->beg + 4;
*ptr = cur + 4;
return true;
}
@@ -3134,7 +3143,7 @@ static_inline bool read_null(u8 **ptr, yyjson_val *val) {
}
/** Read `Inf` or `Infinity` literal (ignoring case). */
-static_inline bool read_inf(u8 **ptr, u8 **pre,
+static_inline bool read_inf(u8 **ptr, u8 **pre, u8 *beg,
yyjson_read_flag flg, yyjson_val *val) {
u8 *hdr = *ptr;
u8 *cur = *ptr;
@@ -3163,9 +3172,13 @@ static_inline bool read_inf(u8 **ptr, u8 **pre,
*pre = cur; /* save end position for current raw string */
val->tag = ((u64)(cur - hdr) << YYJSON_TAG_BIT) | YYJSON_TYPE_RAW;
val->uni.str = (const char *)hdr;
+ val->beg = hdr - beg;
+ val->end = cur - beg - 1;
} else {
val->tag = YYJSON_TYPE_NUM | YYJSON_SUBTYPE_REAL;
val->uni.u64 = f64_bits_inf(sign);
+ val->beg = hdr - beg;
+ val->end = cur - beg - 1;
}
return true;
}
@@ -3173,7 +3186,7 @@ static_inline bool read_inf(u8 **ptr, u8 **pre,
}
/** Read `NaN` literal (ignoring case). */
-static_inline bool read_nan(u8 **ptr, u8 **pre,
+static_inline bool read_nan(u8 **ptr, u8 **pre, u8 *beg,
yyjson_read_flag flg, yyjson_val *val) {
u8 *hdr = *ptr;
u8 *cur = *ptr;
@@ -3191,9 +3204,13 @@ static_inline bool read_nan(u8 **ptr, u8 **pre,
*pre = cur; /* save end position for current raw string */
val->tag = ((u64)(cur - hdr) << YYJSON_TAG_BIT) | YYJSON_TYPE_RAW;
val->uni.str = (const char *)hdr;
+ val->beg = hdr - beg;
+ val->end = cur - beg - 1;
} else {
val->tag = YYJSON_TYPE_NUM | YYJSON_SUBTYPE_REAL;
val->uni.u64 = f64_bits_nan(sign);
+ val->beg = hdr - beg;
+ val->end = cur - beg - 1;
}
return true;
}
@@ -3201,15 +3218,15 @@ static_inline bool read_nan(u8 **ptr, u8 **pre,
}
/** Read `Inf`, `Infinity` or `NaN` literal (ignoring case). */
-static_inline bool read_inf_or_nan(u8 **ptr, u8 **pre,
+static_inline bool read_inf_or_nan(u8 **ptr, u8 **pre, u8 *beg,
yyjson_read_flag flg, yyjson_val *val) {
- if (read_inf(ptr, pre, flg, val)) return true;
- if (read_nan(ptr, pre, flg, val)) return true;
+ if (read_inf(ptr, pre, beg, flg, val)) return true;
+ if (read_nan(ptr, pre, beg, flg, val)) return true;
return false;
}
/** Read a JSON number as raw string. */
-static_noinline bool read_num_raw(u8 **ptr, u8 **pre, yyjson_read_flag flg,
+static_noinline bool read_num_raw(u8 **ptr, u8 **pre, u8 *beg, yyjson_read_flag flg,
yyjson_val *val, const char **msg) {
#define return_err(_pos, _msg) do { \
*msg = _msg; *end = _pos; return false; \
@@ -3218,6 +3235,8 @@ static_noinline bool read_num_raw(u8 **ptr, u8 **pre, yyjson_read_flag flg,
#define return_raw() do { \
val->tag = ((u64)(cur - hdr) << YYJSON_TAG_BIT) | YYJSON_TYPE_RAW; \
val->uni.str = (const char *)hdr; \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
**pre = '\0'; *pre = cur; *end = cur; return true; \
} while (false)
@@ -3240,7 +3259,7 @@ static_noinline bool read_num_raw(u8 **ptr, u8 **pre, yyjson_read_flag flg,
}
}
if (has_allow(INF_AND_NAN)) {
- if (read_inf_or_nan(ptr, pre, flg, val)) return true;
+ if (read_inf_or_nan(ptr, pre, beg, flg, val)) return true;
}
return_err(cur, "no digit after sign");
}
@@ -3293,7 +3312,7 @@ read_double:
}
/** Read a hex number. */
-static_noinline bool read_num_hex(u8 **ptr, u8 **pre, yyjson_read_flag flg,
+static_noinline bool read_num_hex(u8 **ptr, u8 **pre, u8 *beg, yyjson_read_flag flg,
yyjson_val *val, const char **msg) {
u8 *hdr = *ptr;
u8 *cur = *ptr;
@@ -3331,6 +3350,8 @@ static_noinline bool read_num_hex(u8 **ptr, u8 **pre, yyjson_read_flag flg,
**pre = '\0';
val->tag = ((u64)(cur - hdr) << YYJSON_TAG_BIT) | YYJSON_TYPE_RAW;
val->uni.str = (const char *)hdr;
+ val->beg = hdr - beg;
+ val->end = cur - beg - 1;
*pre = cur; *end = cur;
return true;
}
@@ -3338,6 +3359,8 @@ static_noinline bool read_num_hex(u8 **ptr, u8 **pre, yyjson_read_flag flg,
val->tag = YYJSON_TYPE_NUM | (u64)((u8)sign << 3);
val->uni.u64 = (u64)(sign ? (u64)(~(sig) + 1) : (u64)(sig));
+ val->beg = hdr - beg;
+ val->end = cur + i - beg - 1;
*end = cur + i;
return true;
}
@@ -3813,7 +3836,7 @@ static_inline u64 diy_fp_to_ieee_raw(diy_fp fp) {
number is infinite, the return value is based on flag.
3. This function (with inline attribute) may generate a lot of instructions.
*/
-static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
+static_inline bool read_num(u8 **ptr, u8 **pre, u8 *beg, yyjson_read_flag flg,
yyjson_val *val, const char **msg) {
#define return_err(_pos, _msg) do { \
*msg = _msg; \
@@ -3824,24 +3847,32 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
#define return_0() do { \
val->tag = YYJSON_TYPE_NUM | (u8)((u8)sign << 3); \
val->uni.u64 = 0; \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*end = cur; return true; \
} while (false)
#define return_i64(_v) do { \
val->tag = YYJSON_TYPE_NUM | (u8)((u8)sign << 3); \
val->uni.u64 = (u64)(sign ? (u64)(~(_v) + 1) : (u64)(_v)); \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*end = cur; return true; \
} while (false)
#define return_f64(_v) do { \
val->tag = YYJSON_TYPE_NUM | YYJSON_SUBTYPE_REAL; \
val->uni.f64 = sign ? -(f64)(_v) : (f64)(_v); \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*end = cur; return true; \
} while (false)
#define return_f64_bin(_v) do { \
val->tag = YYJSON_TYPE_NUM | YYJSON_SUBTYPE_REAL; \
val->uni.u64 = ((u64)sign << 63) | (u64)(_v); \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*end = cur; return true; \
} while (false)
@@ -3855,6 +3886,8 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
**pre = '\0'; /* add null-terminator for previous raw string */ \
val->tag = ((u64)(cur - hdr) << YYJSON_TAG_BIT) | YYJSON_TYPE_RAW; \
val->uni.str = (const char *)hdr; \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*pre = cur; *end = cur; return true; \
} while (false)
@@ -3878,7 +3911,7 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
/* read number as raw string if has `YYJSON_READ_NUMBER_AS_RAW` flag */
if (has_flg(NUMBER_AS_RAW)) {
- return read_num_raw(ptr, pre, flg, val, msg);
+ return read_num_raw(ptr, pre, beg, flg, val, msg);
}
sign = (*hdr == '-');
@@ -3897,14 +3930,14 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
}
}
if (has_allow(INF_AND_NAN)) {
- if (read_inf_or_nan(ptr, pre, flg, val)) return true;
+ if (read_inf_or_nan(ptr, pre, beg, flg, val)) return true;
}
return_err(cur, "no digit after sign");
}
/* begin with 0 */
if (likely(!char_is_digit_or_fp(*++cur))) {
if (has_allow(EXT_NUMBER) && char_to_lower(*cur) == 'x') { /* hex */
- return read_num_hex(ptr, pre, flg, val, msg);
+ return read_num_hex(ptr, pre, beg, flg, val, msg);
}
return_0();
}
@@ -4447,7 +4480,7 @@ digi_finish:
This is a fallback function if the custom number reader is disabled.
This function use libc's strtod() to read floating-point number.
*/
-static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
+static_inline bool read_num(u8 **ptr, u8 **pre, u8 *beg, yyjson_read_flag flg,
yyjson_val *val, const char **msg) {
#define return_err(_pos, _msg) do { \
*msg = _msg; \
@@ -4458,24 +4491,32 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
#define return_0() do { \
val->tag = YYJSON_TYPE_NUM | (u64)((u8)sign << 3); \
val->uni.u64 = 0; \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*end = cur; return true; \
} while (false)
#define return_i64(_v) do { \
val->tag = YYJSON_TYPE_NUM | (u64)((u8)sign << 3); \
val->uni.u64 = (u64)(sign ? (u64)(~(_v) + 1) : (u64)(_v)); \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*end = cur; return true; \
} while (false)
#define return_f64(_v) do { \
val->tag = YYJSON_TYPE_NUM | YYJSON_SUBTYPE_REAL; \
val->uni.f64 = sign ? -(f64)(_v) : (f64)(_v); \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*end = cur; return true; \
} while (false)
#define return_f64_bin(_v) do { \
val->tag = YYJSON_TYPE_NUM | YYJSON_SUBTYPE_REAL; \
val->uni.u64 = ((u64)sign << 63) | (u64)(_v); \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
*end = cur; return true; \
} while (false)
@@ -4488,6 +4529,8 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
#define return_raw() do { \
val->tag = ((u64)(cur - hdr) << YYJSON_TAG_BIT) | YYJSON_TYPE_RAW; \
val->uni.str = (const char *)hdr; \
+ val->beg = hdr - beg; \
+ val->end = cur - beg - 1; \
**pre = '\0'; *pre = cur; *end = cur; return true; \
} while (false)
@@ -4501,7 +4544,7 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
/* read number as raw string if has `YYJSON_READ_NUMBER_AS_RAW` flag */
if (has_flg(NUMBER_AS_RAW)) {
- return read_num_raw(ptr, pre, flg, val, msg);
+ return read_num_raw(ptr, pre, beg, flg, val, msg);
}
sign = (*hdr == '-');
@@ -4521,7 +4564,7 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
}
}
if (has_allow(INF_AND_NAN)) {
- if (read_inf_or_nan(ptr, pre, flg, val)) return true;
+ if (read_inf_or_nan(ptr, pre, beg, flg, val)) return true;
}
return_err(cur, "no digit after sign");
}
@@ -4533,7 +4576,7 @@ static_inline bool read_num(u8 **ptr, u8 **pre, yyjson_read_flag flg,
if (!char_is_fp(*cur)) {
if (has_allow(EXT_NUMBER) &&
(*cur == 'x' || *cur == 'X')) { /* hex integer */
- return read_num_hex(ptr, pre, flg, val, msg);
+ return read_num_hex(ptr, pre, beg, flg, val, msg);
}
return_0();
}
@@ -4721,8 +4764,9 @@ static_inline bool read_uni_esc(u8 **src_ptr, u8 **dst_ptr, const char **msg) {
@param con Continuation for incremental parsing.
@return Whether success.
*/
-static_inline bool read_str_opt(u8 quo, u8 **ptr, u8 *eof, yyjson_read_flag flg,
- yyjson_val *val, const char **msg, u8 *con[2]) {
+static_inline bool read_str_opt(u8 quo, u8 **ptr, u8 *beg, u8 *eof,
+ yyjson_read_flag flg, yyjson_val *val,
+ const char **msg, u8 *con[2]) {
/*
GCC may sometimes load variables into registers too early, causing
unnecessary instructions and performance degradation. This inline assembly
@@ -4808,6 +4852,8 @@ skip_ascii_end:
val->tag = ((u64)(src - hdr) << YYJSON_TAG_BIT) | YYJSON_TYPE_STR |
(quo == '"' ? YYJSON_SUBTYPE_NOESC : 0);
val->uni.str = (const char *)hdr;
+ val->beg = hdr - beg - 1;
+ val->end = src - beg + 1;
*src = '\0';
*end = src + 1;
if (con) con[0] = con[1] = NULL;
@@ -4930,6 +4976,8 @@ copy_escape:
} else if (likely(*src == quo)) {
val->tag = ((u64)(dst - hdr) << YYJSON_TAG_BIT) | YYJSON_TYPE_STR;
val->uni.str = (const char *)hdr;
+ val->beg = hdr - beg - 1;
+ val->end = src - beg + 1;
*dst = '\0';
*end = src + 1;
if (con) con[0] = con[1] = NULL;
@@ -5039,23 +5087,25 @@ copy_utf8:
#undef return_err
}
-static_inline bool read_str(u8 **ptr, u8 *eof, yyjson_read_flag flg,
+static_inline bool read_str(u8 **ptr, u8 *beg, u8 *eof, yyjson_read_flag flg,
yyjson_val *val, const char **msg) {
- return read_str_opt('\"', ptr, eof, flg, val, msg, NULL);
+ return read_str_opt('\"', ptr, beg, eof, flg, val, msg, NULL);
}
-static_inline bool read_str_con(u8 **ptr, u8 *eof, yyjson_read_flag flg,
- yyjson_val *val, const char **msg, u8 **con) {
- return read_str_opt('\"', ptr, eof, flg, val, msg, con);
+static_inline bool read_str_con(u8 **ptr, u8 *beg, u8 *eof,
+ yyjson_read_flag flg, yyjson_val *val,
+ const char **msg, u8 **con) {
+ return read_str_opt('\"', ptr, beg, eof, flg, val, msg, con);
}
-static_noinline bool read_str_sq(u8 **ptr, u8 *eof, yyjson_read_flag flg,
- yyjson_val *val, const char **msg) {
- return read_str_opt('\'', ptr, eof, flg, val, msg, NULL);
+static_noinline bool read_str_sq(u8 **ptr, u8 *beg, u8 *eof,
+ yyjson_read_flag flg, yyjson_val *val,
+ const char **msg) {
+ return read_str_opt('\'', ptr, beg, eof, flg, val, msg, NULL);
}
/** Read unquoted key (identifier name). */
-static_noinline bool read_str_id(u8 **ptr, u8 *eof, yyjson_read_flag flg,
+static_noinline bool read_str_id(u8 **ptr, u8 *beg, u8 *eof, yyjson_read_flag flg,
u8 **pre, yyjson_val *val, const char **msg) {
#define return_err(_end, _msg) do { \
*msg = _msg; \
@@ -5067,6 +5117,8 @@ static_noinline bool read_str_id(u8 **ptr, u8 *eof, yyjson_read_flag flg,
val->tag = ((u64)(_str_end - hdr) << YYJSON_TAG_BIT) | \
(u64)(YYJSON_TYPE_STR); \
val->uni.str = (const char *)hdr; \
+ val->beg = hdr - beg; \
+ val->end = _str_end - beg - 1; \
*pre = _str_end; *end = _cur_end; \
return true; \
} while (false)
@@ -5245,33 +5297,33 @@ static_noinline yyjson_doc *read_root_single(u8 *hdr, u8 *cur, u8 *eof,
val = val_hdr + hdr_len;
if (char_is_num(*cur)) {
- if (likely(read_num(&cur, pre, flg, val, &msg))) goto doc_end;
+ if (likely(read_num(&cur, pre, hdr, flg, val, &msg))) goto doc_end;
goto fail_number;
}
if (*cur == '"') {
- if (likely(read_str(&cur, eof, flg, val, &msg))) goto doc_end;
+ if (likely(read_str(&cur, hdr, eof, flg, val, &msg))) goto doc_end;
goto fail_string;
}
if (*cur == 't') {
- if (likely(read_true(&cur, val))) goto doc_end;
+ if (likely(read_true(&cur, val, hdr))) goto doc_end;
goto fail_literal_true;
}
if (*cur == 'f') {
- if (likely(read_false(&cur, val))) goto doc_end;
+ if (likely(read_false(&cur, val, hdr))) goto doc_end;
goto fail_literal_false;
}
if (*cur == 'n') {
- if (likely(read_null(&cur, val))) goto doc_end;
+ if (likely(read_null(&cur, val, hdr))) goto doc_end;
if (has_allow(INF_AND_NAN)) {
- if (read_nan(&cur, pre, flg, val)) goto doc_end;
+ if (read_nan(&cur, pre, hdr, flg, val)) goto doc_end;
}
goto fail_literal_null;
}
if (has_allow(INF_AND_NAN)) {
- if (read_inf_or_nan(&cur, pre, flg, val)) goto doc_end;
+ if (read_inf_or_nan(&cur, pre, hdr, flg, val)) goto doc_end;
}
if (has_allow(SINGLE_QUOTED_STR) && *cur == '\'') {
- if (likely(read_str_sq(&cur, eof, flg, val, &msg))) goto doc_end;
+ if (likely(read_str_sq(&cur, hdr, eof, flg, val, &msg))) goto doc_end;
goto fail_string;
}
goto fail_character;
@@ -5383,10 +5435,12 @@ static_inline yyjson_doc *read_root_minify(u8 *hdr, u8 *cur, u8 *eof,
if (*cur++ == '{') {
ctn->tag = YYJSON_TYPE_OBJ;
ctn->uni.ofs = 0;
+ ctn->beg = cur - hdr;
goto obj_key_begin;
} else {
ctn->tag = YYJSON_TYPE_ARR;
ctn->uni.ofs = 0;
+ ctn->beg = cur - hdr;
goto arr_val_begin;
}
@@ -5399,6 +5453,7 @@ arr_begin:
val_incr();
val->tag = YYJSON_TYPE_ARR;
val->uni.ofs = (usize)((u8 *)val - (u8 *)ctn);
+ val->beg = cur - hdr - 1;
/* push the new array value as current container */
ctn = val;
@@ -5416,33 +5471,33 @@ arr_val_begin:
if (char_is_num(*cur)) {
val_incr();
ctn_len++;
- if (likely(read_num(&cur, pre, flg, val, &msg))) goto arr_val_end;
+ if (likely(read_num(&cur, pre, hdr, flg, val, &msg))) goto arr_val_end;
goto fail_number;
}
if (*cur == '"') {
val_incr();
ctn_len++;
- if (likely(read_str(&cur, eof, flg, val, &msg))) goto arr_val_end;
+ if (likely(read_str(&cur, hdr, eof, flg, val, &msg))) goto arr_val_end;
goto fail_string;
}
if (*cur == 't') {
val_incr();
ctn_len++;
- if (likely(read_true(&cur, val))) goto arr_val_end;
+ if (likely(read_true(&cur, val, hdr))) goto arr_val_end;
goto fail_literal_true;
}
if (*cur == 'f') {
val_incr();
ctn_len++;
- if (likely(read_false(&cur, val))) goto arr_val_end;
+ if (likely(read_false(&cur, val, hdr))) goto arr_val_end;
goto fail_literal_false;
}
if (*cur == 'n') {
val_incr();
ctn_len++;
- if (likely(read_null(&cur, val))) goto arr_val_end;
+ if (likely(read_null(&cur, val, hdr))) goto arr_val_end;
if (has_allow(INF_AND_NAN)) {
- if (read_nan(&cur, pre, flg, val)) goto arr_val_end;
+ if (read_nan(&cur, pre, hdr, flg, val)) goto arr_val_end;
}
goto fail_literal_null;
}
@@ -5461,13 +5516,13 @@ arr_val_begin:
(*cur == 'i' || *cur == 'I' || *cur == 'N')) {
val_incr();
ctn_len++;
- if (read_inf_or_nan(&cur, pre, flg, val)) goto arr_val_end;
+ if (read_inf_or_nan(&cur, pre, hdr, flg, val)) goto arr_val_end;
goto fail_character_val;
}
if (has_allow(SINGLE_QUOTED_STR) && *cur == '\'') {
val_incr();
ctn_len++;
- if (likely(read_str_sq(&cur, eof, flg, val, &msg))) goto arr_val_end;
+ if (likely(read_str_sq(&cur, hdr, eof, flg, val, &msg))) goto arr_val_end;
goto fail_string;
}
if (has_allow(TRIVIA) && char_is_trivia(*cur)) {
@@ -5502,6 +5557,7 @@ arr_end:
/* save the next sibling value offset */
ctn->uni.ofs = (usize)((u8 *)val - (u8 *)ctn) + sizeof(yyjson_val);
ctn->tag = ((ctn_len) << YYJSON_TAG_BIT) | YYJSON_TYPE_ARR;
+ ctn->end = cur - hdr - 1;
if (unlikely(ctn == ctn_parent)) goto doc_end;
/* pop parent as current container */
@@ -5521,6 +5577,7 @@ obj_begin:
val->tag = YYJSON_TYPE_OBJ;
/* offset to the parent */
val->uni.ofs = (usize)((u8 *)val - (u8 *)ctn);
+ val->beg = cur - hdr - 1;
ctn = val;
ctn_len = 0;
@@ -5528,7 +5585,7 @@ obj_key_begin:
if (likely(*cur == '"')) {
val_incr();
ctn_len++;
- if (likely(read_str(&cur, eof, flg, val, &msg))) goto obj_key_end;
+ if (likely(read_str(&cur, hdr, eof, flg, val, &msg))) goto obj_key_end;
goto fail_string;
}
if (likely(*cur == '}')) {
@@ -5545,13 +5602,13 @@ obj_key_begin:
if (has_allow(SINGLE_QUOTED_STR) && *cur == '\'') {
val_incr();
ctn_len++;
- if (likely(read_str_sq(&cur, eof, flg, val, &msg))) goto obj_key_end;
+ if (likely(read_str_sq(&cur, hdr, eof, flg, val, &msg))) goto obj_key_end;
goto fail_string;
}
if (has_allow(UNQUOTED_KEY) && char_is_id_start(*cur)) {
val_incr();
ctn_len++;
- if (read_str_id(&cur, eof, flg, pre, val, &msg)) goto obj_key_end;
+ if (read_str_id(&cur, hdr, eof, flg, pre, val, &msg)) goto obj_key_end;
goto fail_string;
}
if (has_allow(TRIVIA) && char_is_trivia(*cur)) {
@@ -5579,13 +5636,13 @@ obj_val_begin:
if (*cur == '"') {
val++;
ctn_len++;
- if (likely(read_str(&cur, eof, flg, val, &msg))) goto obj_val_end;
+ if (likely(read_str(&cur, hdr, eof, flg, val, &msg))) goto obj_val_end;
goto fail_string;
}
if (char_is_num(*cur)) {
val++;
ctn_len++;
- if (likely(read_num(&cur, pre, flg, val, &msg))) goto obj_val_end;
+ if (likely(read_num(&cur, pre, hdr, flg, val, &msg))) goto obj_val_end;
goto fail_number;
}
if (*cur == '{') {
@@ -5599,21 +5656,21 @@ obj_val_begin:
if (*cur == 't') {
val++;
ctn_len++;
- if (likely(read_true(&cur, val))) goto obj_val_end;
+ if (likely(read_true(&cur, val, hdr))) goto obj_val_end;
goto fail_literal_true;
}
if (*cur == 'f') {
val++;
ctn_len++;
- if (likely(read_false(&cur, val))) goto obj_val_end;
+ if (likely(read_false(&cur, val, hdr))) goto obj_val_end;
goto fail_literal_false;
}
if (*cur == 'n') {
val++;
ctn_len++;
- if (likely(read_null(&cur, val))) goto obj_val_end;
+ if (likely(read_null(&cur, val, hdr))) goto obj_val_end;
if (has_allow(INF_AND_NAN)) {
- if (read_nan(&cur, pre, flg, val)) goto obj_val_end;
+ if (read_nan(&cur, pre, hdr, flg, val)) goto obj_val_end;
}
goto fail_literal_null;
}
@@ -5625,13 +5682,13 @@ obj_val_begin:
(*cur == 'i' || *cur == 'I' || *cur == 'N')) {
val++;
ctn_len++;
- if (read_inf_or_nan(&cur, pre, flg, val)) goto obj_val_end;
+ if (read_inf_or_nan(&cur, pre, hdr, flg, val)) goto obj_val_end;
goto fail_character_val;
}
if (has_allow(SINGLE_QUOTED_STR) && *cur == '\'') {
val++;
ctn_len++;
- if (likely(read_str_sq(&cur, eof, flg, val, &msg))) goto obj_val_end;
+ if (likely(read_str_sq(&cur, hdr, eof, flg, val, &msg))) goto obj_val_end;
goto fail_string;
}
if (has_allow(TRIVIA) && char_is_trivia(*cur)) {
@@ -5665,6 +5722,7 @@ obj_end:
/* point to the next value */
ctn->uni.ofs = (usize)((u8 *)val - (u8 *)ctn) + sizeof(yyjson_val);
ctn->tag = (ctn_len << (YYJSON_TAG_BIT - 1)) | YYJSON_TYPE_OBJ;
+ ctn->end = cur - hdr - 1;
if (unlikely(ctn == ctn_parent)) goto doc_end;
ctn = ctn_parent;
ctn_len = (usize)(ctn->tag >> YYJSON_TAG_BIT);
@@ -5787,11 +5845,13 @@ static_inline yyjson_doc *read_root_pretty(u8 *hdr, u8 *cur, u8 *eof,
if (*cur++ == '{') {
ctn->tag = YYJSON_TYPE_OBJ;
ctn->uni.ofs = 0;
+ ctn->beg = cur - hdr - 1;
if (*cur == '\n') cur++;
goto obj_key_begin;
} else {
ctn->tag = YYJSON_TYPE_ARR;
ctn->uni.ofs = 0;
+ ctn->beg = cur - hdr - 1;
if (*cur == '\n') cur++;
goto arr_val_begin;
}
@@ -5805,6 +5865,7 @@ arr_begin:
val_incr();
val->tag = YYJSON_TYPE_ARR;
val->uni.ofs = (usize)((u8 *)val - (u8 *)ctn);
+ val->beg = cur - hdr - 1;
/* push the new array value as current container */
ctn = val;
@@ -5835,33 +5896,33 @@ arr_val_begin:
if (char_is_num(*cur)) {
val_incr();
ctn_len++;
- if (likely(read_num(&cur, pre, flg, val, &msg))) goto arr_val_end;
+ if (likely(read_num(&cur, pre, hdr, flg, val, &msg))) goto arr_val_end;
goto fail_number;
}
if (*cur == '"') {
val_incr();
ctn_len++;
- if (likely(read_str(&cur, eof, flg, val, &msg))) goto arr_val_end;
+ if (likely(read_str(&cur, hdr, eof, flg, val, &msg))) goto arr_val_end;
goto fail_string;
}
if (*cur == 't') {
val_incr();
ctn_len++;
- if (likely(read_true(&cur, val))) goto arr_val_end;
+ if (likely(read_true(&cur, val, hdr))) goto arr_val_end;
goto fail_literal_true;
}
if (*cur == 'f') {
val_incr();
ctn_len++;
- if (likely(read_false(&cur, val))) goto arr_val_end;
+ if (likely(read_false(&cur, val, hdr))) goto arr_val_end;
goto fail_literal_false;
}
if (*cur == 'n') {
val_incr();
ctn_len++;
- if (likely(read_null(&cur, val))) goto arr_val_end;
+ if (likely(read_null(&cur, val, hdr))) goto arr_val_end;
if (has_allow(INF_AND_NAN)) {
- if (read_nan(&cur, pre, flg, val)) goto arr_val_end;
+ if (read_nan(&cur, pre, hdr, flg, val)) goto arr_val_end;
}
goto fail_literal_null;
}
@@ -5880,13 +5941,13 @@ arr_val_begin:
(*cur == 'i' || *cur == 'I' || *cur == 'N')) {
val_incr();
ctn_len++;
- if (read_inf_or_nan(&cur, pre, flg, val)) goto arr_val_end;
+ if (read_inf_or_nan(&cur, pre, hdr, flg, val)) goto arr_val_end;
goto fail_character_val;
}
if (has_allow(SINGLE_QUOTED_STR) && *cur == '\'') {
val_incr();
ctn_len++;
- if (likely(read_str_sq(&cur, eof, flg, val, &msg))) goto arr_val_end;
+ if (likely(read_str_sq(&cur, hdr, eof, flg, val, &msg))) goto arr_val_end;
goto fail_string;
}
if (has_allow(TRIVIA) && char_is_trivia(*cur)) {
@@ -5925,6 +5986,7 @@ arr_end:
/* save the next sibling value offset */
ctn->uni.ofs = (usize)((u8 *)val - (u8 *)ctn) + sizeof(yyjson_val);
ctn->tag = ((ctn_len) << YYJSON_TAG_BIT) | YYJSON_TYPE_ARR;
+ ctn->end = cur - hdr - 1;
if (unlikely(ctn == ctn_parent)) goto doc_end;
/* pop parent as current container */
@@ -5945,6 +6007,7 @@ obj_begin:
val->tag = YYJSON_TYPE_OBJ;
/* offset to the parent */
val->uni.ofs = (usize)((u8 *)val - (u8 *)ctn);
+ val->beg = cur - hdr - 1;
ctn = val;
ctn_len = 0;
if (*cur == '\n') cur++;
@@ -5964,7 +6027,7 @@ obj_key_begin:
if (likely(*cur == '"')) {
val_incr();
ctn_len++;
- if (likely(read_str(&cur, eof, flg, val, &msg))) goto obj_key_end;
+ if (likely(read_str(&cur, hdr, eof, flg, val, &msg))) goto obj_key_end;
goto fail_string;
}
if (likely(*cur == '}')) {
@@ -5981,13 +6044,13 @@ obj_key_begin:
if (has_allow(SINGLE_QUOTED_STR) && *cur == '\'') {
val_incr();
ctn_len++;
- if (likely(read_str_sq(&cur, eof, flg, val, &msg))) goto obj_key_end;
+ if (likely(read_str_sq(&cur, hdr, eof, flg, val, &msg))) goto obj_key_end;
goto fail_string;
}
if (has_allow(UNQUOTED_KEY) && char_is_id_start(*cur)) {
val_incr();
ctn_len++;
- if (read_str_id(&cur, eof, flg, pre, val, &msg)) goto obj_key_end;
+ if (read_str_id(&cur, hdr, eof, flg, pre, val, &msg)) goto obj_key_end;
goto fail_string;
}
if (has_allow(TRIVIA) && char_is_trivia(*cur)) {
@@ -6019,13 +6082,13 @@ obj_val_begin:
if (*cur == '"') {
val++;
ctn_len++;
- if (likely(read_str(&cur, eof, flg, val, &msg))) goto obj_val_end;
+ if (likely(read_str(&cur, hdr, eof, flg, val, &msg))) goto obj_val_end;
goto fail_string;
}
if (char_is_num(*cur)) {
val++;
ctn_len++;
- if (likely(read_num(&cur, pre, flg, val, &msg))) goto obj_val_end;
+ if (likely(read_num(&cur, pre, hdr, flg, val, &msg))) goto obj_val_end;
goto fail_number;
}
if (*cur == '{') {
@@ -6039,21 +6102,21 @@ obj_val_begin:
if (*cur == 't') {
val++;
ctn_len++;
- if (likely(read_true(&cur, val))) goto obj_val_end;
+ if (likely(read_true(&cur, val, hdr))) goto obj_val_end;
goto fail_literal_true;
}
if (*cur == 'f') {
val++;
ctn_len++;
- if (likely(read_false(&cur, val))) goto obj_val_end;
+ if (likely(read_false(&cur, val, hdr))) goto obj_val_end;
goto fail_literal_false;
}
if (*cur == 'n') {
val++;
ctn_len++;
- if (likely(read_null(&cur, val))) goto obj_val_end;
+ if (likely(read_null(&cur, val, hdr))) goto obj_val_end;
if (has_allow(INF_AND_NAN)) {
- if (read_nan(&cur, pre, flg, val)) goto obj_val_end;
+ if (read_nan(&cur, pre, hdr, flg, val)) goto obj_val_end;
}
goto fail_literal_null;
}
@@ -6065,13 +6128,13 @@ obj_val_begin:
(*cur == 'i' || *cur == 'I' || *cur == 'N')) {
val++;
ctn_len++;
- if (read_inf_or_nan(&cur, pre, flg, val)) goto obj_val_end;
+ if (read_inf_or_nan(&cur, pre, hdr, flg, val)) goto obj_val_end;
goto fail_character_val;
}
if (has_allow(SINGLE_QUOTED_STR) && *cur == '\'') {
val++;
ctn_len++;
- if (likely(read_str_sq(&cur, eof, flg, val, &msg))) goto obj_val_end;
+ if (likely(read_str_sq(&cur, hdr, eof, flg, val, &msg))) goto obj_val_end;
goto fail_string;
}
if (has_allow(TRIVIA) && char_is_trivia(*cur)) {
@@ -6109,6 +6172,7 @@ obj_end:
/* point to the next value */
ctn->uni.ofs = (usize)((u8 *)val - (u8 *)ctn) + sizeof(yyjson_val);
ctn->tag = (ctn_len << (YYJSON_TAG_BIT - 1)) | YYJSON_TYPE_OBJ;
+ ctn->end = cur - hdr - 1;
if (unlikely(ctn == ctn_parent)) goto doc_end;
ctn = ctn_parent;
ctn_len = (usize)(ctn->tag >> YYJSON_TAG_BIT);
@@ -6228,9 +6292,9 @@ yyjson_doc *yyjson_read_opts(char *dat, usize len,
/* read json document */
if (likely(char_is_ctn(*cur))) {
if (char_is_space(cur[1]) && char_is_space(cur[2])) {
- doc = read_root_pretty(hdr, cur, eof, alc, flg, err);
+ doc = read_root_pretty(hdr, cur, eof, alc, flg, err); // read function
} else {
- doc = read_root_minify(hdr, cur, eof, alc, flg, err);
+ doc = read_root_minify(hdr, cur, eof, alc, flg, err); // read function
}
} else {
doc = read_root_single(hdr, cur, eof, alc, flg, err);
@@ -6425,7 +6489,7 @@ const char *yyjson_read_number(const char *dat,
#endif
#if YYJSON_DISABLE_FAST_FP_CONV
- if (!read_num(&cur, pre, flg, val, &msg)) {
+ if (!read_num(&cur, pre, hdr, flg, val, &msg)) {
if (dat_len >= sizeof(buf)) alc->free(alc->ctx, hdr);
return_err(cur, INVALID_NUMBER, msg);
}
@@ -6433,7 +6497,7 @@ const char *yyjson_read_number(const char *dat,
if (yyjson_is_raw(val)) val->uni.str = dat;
return dat + (cur - hdr);
#else
- if (!read_num(&cur, pre, flg, val, &msg)) {
+ if (!read_num(&cur, pre, hdr, flg, val, &msg)) {
return_err(cur, INVALID_NUMBER, msg);
}
return (const char *)cur;
@@ -6703,23 +6767,23 @@ doc_begin:
goto arr_val_begin;
}
if (char_is_num(*cur)) {
- if (likely(read_num(&cur, pre, flg, val, &msg))) goto doc_end;
+ if (likely(read_num(&cur, pre, hdr, flg, val, &msg))) goto doc_end;
goto fail_number;
}
if (*cur == '"') {
- if (likely(read_str_con(&cur, end, flg, val, &msg, con))) goto doc_end;
+ if (likely(read_str_con(&cur, hdr, end, flg, val, &msg, con))) goto doc_end;
goto fail_string;
}
if (*cur == 't') {
- if (likely(read_true(&cur, val))) goto doc_end;
+ if (likely(read_true(&cur, val, hdr))) goto doc_end;
goto fail_literal_true;
}
if (*cur == 'f') {
- if (likely(read_false(&cur, val))) goto doc_end;
+ if (likely(read_false(&cur, val, hdr))) goto doc_end;
goto fail_literal_false;
}
if (*cur == 'n') {
- if (likely(read_null(&cur, val))) goto doc_end;
+ if (likely(read_null(&cur, val, hdr))) goto doc_end;
goto fail_literal_null;
}
@@ -6760,32 +6824,32 @@ arr_val_continue:
if (char_is_num(*cur)) {
val_incr();
ctn_len++;
- if (likely(read_num(&cur, pre, flg, val, &msg))) goto arr_val_maybe_end;
+ if (likely(read_num(&cur, pre, hdr, flg, val, &msg))) goto arr_val_maybe_end;
goto fail_number;
}
if (*cur == '"') {
val_incr();
ctn_len++;
- if (likely(read_str_con(&cur, end, flg, val, &msg, con)))
+ if (likely(read_str_con(&cur, hdr, end, flg, val, &msg, con)))
goto arr_val_end;
goto fail_string;
}
if (*cur == 't') {
val_incr();
ctn_len++;
- if (likely(read_true(&cur, val))) goto arr_val_end;
+ if (likely(read_true(&cur, val, hdr))) goto arr_val_end;
goto fail_literal_true;
}
if (*cur == 'f') {
val_incr();
ctn_len++;
- if (likely(read_false(&cur, val))) goto arr_val_end;
+ if (likely(read_false(&cur, val, hdr))) goto arr_val_end;
goto fail_literal_false;
}
if (*cur == 'n') {
val_incr();
ctn_len++;
- if (likely(read_null(&cur, val))) goto arr_val_end;
+ if (likely(read_null(&cur, val, hdr))) goto arr_val_end;
goto fail_literal_null;
}
if (*cur == ']') {
@@ -6856,7 +6920,7 @@ obj_key_continue:
if (likely(*cur == '"')) {
val_incr();
ctn_len++;
- if (likely(read_str_con(&cur, end, flg, val, &msg, con)))
+ if (likely(read_str_con(&cur, hdr, end, flg, val, &msg, con)))
goto obj_key_end;
goto fail_string;
}
@@ -6890,14 +6954,14 @@ obj_val_continue:
if (*cur == '"') {
val++;
ctn_len++;
- if (likely(read_str_con(&cur, end, flg, val, &msg, con)))
+ if (likely(read_str_con(&cur, hdr, end, flg, val, &msg, con)))
goto obj_val_end;
goto fail_string;
}
if (char_is_num(*cur)) {
val++;
ctn_len++;
- if (likely(read_num(&cur, pre, flg, val, &msg))) goto obj_val_maybe_end;
+ if (likely(read_num(&cur, pre, hdr, flg, val, &msg))) goto obj_val_maybe_end;
goto fail_number;
}
if (*cur == '{') {
@@ -6911,19 +6975,19 @@ obj_val_continue:
if (*cur == 't') {
val++;
ctn_len++;
- if (likely(read_true(&cur, val))) goto obj_val_end;
+ if (likely(read_true(&cur, val, hdr))) goto obj_val_end;
goto fail_literal_true;
}
if (*cur == 'f') {
val++;
ctn_len++;
- if (likely(read_false(&cur, val))) goto obj_val_end;
+ if (likely(read_false(&cur, val, hdr))) goto obj_val_end;
goto fail_literal_false;
}
if (*cur == 'n') {
val++;
ctn_len++;
- if (likely(read_null(&cur, val))) goto obj_val_end;
+ if (likely(read_null(&cur, val, hdr))) goto obj_val_end;
goto fail_literal_null;
}
if (char_is_space(*cur)) {
diff --git a/src/yyjson.h b/src/yyjson.h
index 5eb6d46..e172efc 100644
--- a/src/yyjson.h
+++ b/src/yyjson.h
@@ -18,12 +18,18 @@
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
+
+ Additional changes to add per-value byte-spans made by <lhearachel@proton.me>
+ Copyright (c) 2025
*============================================================================*/
/**
@file yyjson.h
@date 2019-03-09
@author YaoYuan
+
+ Additional modifications made by github.com/lhearachel to support the location
+ of value-spans.
*/
#ifndef YYJSON_H
@@ -1094,17 +1100,19 @@ yyjson_api_inline size_t yyjson_read_max_memory_usage(size_t len,
for example: "[1,2,3,4]" size is 9, value count is 5.
2. Some broken JSON may cost more memory during reading, but fail at end,
for example: "[[[[[[[[".
- 3. yyjson use 16 bytes per value, see struct yyjson_val.
+ 3. yyjson use 32 bytes per value, see struct yyjson_val.
4. yyjson use dynamic memory with a growth factor of 1.5.
- The max memory size is (json_size / 2 * 16 * 1.5 + padding).
+ The max memory size is (json_size / 2 * 32 * 1.5 + padding).
*/
- size_t mul = (size_t)12 + !(flg & YYJSON_READ_INSITU);
+#define memsize (size_t)(32 * 3 / 4)
+ size_t mul = memsize + !(flg & YYJSON_READ_INSITU);
size_t pad = 256;
size_t max = (size_t)(~(size_t)0);
if (flg & YYJSON_READ_STOP_WHEN_DONE) len = len < 256 ? 256 : len;
if (len >= (max - pad - mul) / mul) return 0;
return len * mul + pad;
+#undef memsize
}
/**
@@ -4760,11 +4768,13 @@ typedef union yyjson_val_uni {
} yyjson_val_uni;
/**
- Immutable JSON value, 16 bytes.
+ Immutable JSON value, 32 bytes.
*/
struct yyjson_val {
uint64_t tag; /**< type, subtype and length */
yyjson_val_uni uni; /**< payload */
+ size_t beg; /**< beginning of value-span */
+ size_t end; /**< ending of value-span */
};
struct yyjson_doc {
@@ -5233,6 +5243,14 @@ yyjson_api_inline const char *yyjson_get_type_desc(yyjson_val *val) {
}
}
+yyjson_api_inline size_t yyjson_dist_beg(yyjson_val *val) {
+ return val == NULL ? 0 : val->beg;
+}
+
+yyjson_api_inline size_t yyjson_dist_end(yyjson_val *val) {
+ return val == NULL ? 0 : val->end;
+}
+
yyjson_api_inline const char *yyjson_get_raw(yyjson_val *val) {
return yyjson_is_raw(val) ? unsafe_yyjson_get_raw(val) : NULL;
}
Reference in New Issue View Git Blame Copy Permalink