… | |
… | |
27 | #define F_SPACE_AFTER 0x00000040UL |
27 | #define F_SPACE_AFTER 0x00000040UL |
28 | #define F_ALLOW_NONREF 0x00000100UL |
28 | #define F_ALLOW_NONREF 0x00000100UL |
29 | #define F_SHRINK 0x00000200UL |
29 | #define F_SHRINK 0x00000200UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
|
|
32 | #define F_RELAXED 0x00001000UL |
|
|
33 | |
32 | #define F_MAXDEPTH 0xf8000000UL |
34 | #define F_MAXDEPTH 0xf8000000UL |
33 | #define S_MAXDEPTH 27 |
35 | #define S_MAXDEPTH 27 |
34 | #define F_MAXSIZE 0x01f00000UL |
36 | #define F_MAXSIZE 0x01f00000UL |
35 | #define S_MAXSIZE 20 |
37 | #define S_MAXSIZE 20 |
36 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
38 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
… | |
… | |
310 | int i, len = av_len (av); |
312 | int i, len = av_len (av); |
311 | |
313 | |
312 | if (enc->indent >= enc->maxdepth) |
314 | if (enc->indent >= enc->maxdepth) |
313 | croak ("data structure too deep (hit recursion limit)"); |
315 | croak ("data structure too deep (hit recursion limit)"); |
314 | |
316 | |
315 | encode_ch (enc, '['); encode_nl (enc); |
317 | encode_ch (enc, '['); |
316 | ++enc->indent; |
318 | |
|
|
319 | if (len >= 0) |
|
|
320 | { |
|
|
321 | encode_nl (enc); ++enc->indent; |
317 | |
322 | |
318 | for (i = 0; i <= len; ++i) |
323 | for (i = 0; i <= len; ++i) |
319 | { |
324 | { |
320 | SV **svp = av_fetch (av, i, 0); |
325 | SV **svp = av_fetch (av, i, 0); |
321 | |
326 | |
322 | encode_indent (enc); |
327 | encode_indent (enc); |
323 | |
328 | |
324 | if (svp) |
329 | if (svp) |
325 | encode_sv (enc, *svp); |
330 | encode_sv (enc, *svp); |
326 | else |
331 | else |
327 | encode_str (enc, "null", 4, 0); |
332 | encode_str (enc, "null", 4, 0); |
328 | |
333 | |
329 | if (i < len) |
334 | if (i < len) |
330 | encode_comma (enc); |
335 | encode_comma (enc); |
331 | } |
336 | } |
332 | |
337 | |
|
|
338 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
339 | } |
|
|
340 | |
333 | encode_nl (enc); |
341 | encode_ch (enc, ']'); |
334 | |
|
|
335 | --enc->indent; |
|
|
336 | encode_indent (enc); encode_ch (enc, ']'); |
|
|
337 | } |
342 | } |
338 | |
343 | |
339 | static void |
344 | static void |
340 | encode_hk (enc_t *enc, HE *he) |
345 | encode_hk (enc_t *enc, HE *he) |
341 | { |
346 | { |
… | |
… | |
394 | int count; |
399 | int count; |
395 | |
400 | |
396 | if (enc->indent >= enc->maxdepth) |
401 | if (enc->indent >= enc->maxdepth) |
397 | croak ("data structure too deep (hit recursion limit)"); |
402 | croak ("data structure too deep (hit recursion limit)"); |
398 | |
403 | |
399 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
404 | encode_ch (enc, '{'); |
400 | |
405 | |
401 | // for canonical output we have to sort by keys first |
406 | // for canonical output we have to sort by keys first |
402 | // actually, this is mostly due to the stupid so-called |
407 | // actually, this is mostly due to the stupid so-called |
403 | // security workaround added somewhere in 5.8.x. |
408 | // security workaround added somewhere in 5.8.x. |
404 | // that randomises hash orderings |
409 | // that randomises hash orderings |
… | |
… | |
457 | |
462 | |
458 | FREETMPS; |
463 | FREETMPS; |
459 | LEAVE; |
464 | LEAVE; |
460 | } |
465 | } |
461 | |
466 | |
|
|
467 | encode_nl (enc); ++enc->indent; |
|
|
468 | |
462 | while (count--) |
469 | while (count--) |
463 | { |
470 | { |
464 | encode_indent (enc); |
471 | encode_indent (enc); |
465 | he = hes [count]; |
472 | he = hes [count]; |
466 | encode_hk (enc, he); |
473 | encode_hk (enc, he); |
467 | encode_sv (enc, SvMAGICAL (hv) ? hv_iterval (hv, he) : HeVAL (he)); |
474 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
468 | |
475 | |
469 | if (count) |
476 | if (count) |
470 | encode_comma (enc); |
477 | encode_comma (enc); |
471 | } |
478 | } |
|
|
479 | |
|
|
480 | encode_nl (enc); --enc->indent; encode_indent (enc); |
472 | } |
481 | } |
473 | } |
482 | } |
474 | else |
483 | else |
475 | { |
484 | { |
476 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
485 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
477 | if ((he = hv_iternext (hv))) |
486 | if ((he = hv_iternext (hv))) |
|
|
487 | { |
|
|
488 | encode_nl (enc); ++enc->indent; |
|
|
489 | |
478 | for (;;) |
490 | for (;;) |
479 | { |
491 | { |
480 | encode_indent (enc); |
492 | encode_indent (enc); |
481 | encode_hk (enc, he); |
493 | encode_hk (enc, he); |
482 | encode_sv (enc, SvMAGICAL (hv) ? hv_iterval (hv, he) : HeVAL (he)); |
494 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
483 | |
495 | |
484 | if (!(he = hv_iternext (hv))) |
496 | if (!(he = hv_iternext (hv))) |
485 | break; |
497 | break; |
486 | |
498 | |
487 | encode_comma (enc); |
499 | encode_comma (enc); |
488 | } |
500 | } |
489 | } |
|
|
490 | |
501 | |
|
|
502 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
503 | } |
|
|
504 | } |
|
|
505 | |
491 | encode_nl (enc); |
506 | encode_ch (enc, '}'); |
492 | |
|
|
493 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
|
|
494 | } |
507 | } |
495 | |
508 | |
496 | // encode objects, arrays and special \0=false and \1=true values. |
509 | // encode objects, arrays and special \0=false and \1=true values. |
497 | static void |
510 | static void |
498 | encode_rv (enc_t *enc, SV *sv) |
511 | encode_rv (enc_t *enc, SV *sv) |
… | |
… | |
698 | U32 depth; // recursion depth |
711 | U32 depth; // recursion depth |
699 | U32 maxdepth; // recursion depth limit |
712 | U32 maxdepth; // recursion depth limit |
700 | } dec_t; |
713 | } dec_t; |
701 | |
714 | |
702 | inline void |
715 | inline void |
|
|
716 | decode_comment (dec_t *dec) |
|
|
717 | { |
|
|
718 | // only '#'-style comments allowed a.t.m. |
|
|
719 | |
|
|
720 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
|
|
721 | ++dec->cur; |
|
|
722 | } |
|
|
723 | |
|
|
724 | inline void |
703 | decode_ws (dec_t *dec) |
725 | decode_ws (dec_t *dec) |
704 | { |
726 | { |
705 | for (;;) |
727 | for (;;) |
706 | { |
728 | { |
707 | char ch = *dec->cur; |
729 | char ch = *dec->cur; |
708 | |
730 | |
709 | if (ch > 0x20 |
731 | if (ch > 0x20) |
|
|
732 | { |
|
|
733 | if (expect_false (ch == '#')) |
|
|
734 | { |
|
|
735 | if (dec->json.flags & F_RELAXED) |
|
|
736 | decode_comment (dec); |
|
|
737 | else |
|
|
738 | break; |
|
|
739 | } |
|
|
740 | else |
|
|
741 | break; |
|
|
742 | } |
710 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
743 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
711 | break; |
744 | break; // parse error, but let higher level handle it, gives better error messages |
712 | |
745 | |
713 | ++dec->cur; |
746 | ++dec->cur; |
714 | } |
747 | } |
715 | } |
748 | } |
716 | |
749 | |
… | |
… | |
1053 | |
1086 | |
1054 | if (*dec->cur != ',') |
1087 | if (*dec->cur != ',') |
1055 | ERR (", or ] expected while parsing array"); |
1088 | ERR (", or ] expected while parsing array"); |
1056 | |
1089 | |
1057 | ++dec->cur; |
1090 | ++dec->cur; |
|
|
1091 | |
|
|
1092 | decode_ws (dec); |
|
|
1093 | |
|
|
1094 | if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
|
|
1095 | { |
|
|
1096 | ++dec->cur; |
|
|
1097 | break; |
|
|
1098 | } |
1058 | } |
1099 | } |
1059 | |
1100 | |
1060 | DEC_DEC_DEPTH; |
1101 | DEC_DEC_DEPTH; |
1061 | return newRV_noinc ((SV *)av); |
1102 | return newRV_noinc ((SV *)av); |
1062 | |
1103 | |
… | |
… | |
1078 | if (*dec->cur == '}') |
1119 | if (*dec->cur == '}') |
1079 | ++dec->cur; |
1120 | ++dec->cur; |
1080 | else |
1121 | else |
1081 | for (;;) |
1122 | for (;;) |
1082 | { |
1123 | { |
1083 | decode_ws (dec); EXPECT_CH ('"'); |
1124 | EXPECT_CH ('"'); |
1084 | |
1125 | |
1085 | // heuristic: assume that |
1126 | // heuristic: assume that |
1086 | // a) decode_str + hv_store_ent are abysmally slow. |
1127 | // a) decode_str + hv_store_ent are abysmally slow. |
1087 | // b) most hash keys are short, simple ascii text. |
1128 | // b) most hash keys are short, simple ascii text. |
1088 | // => try to "fast-match" such strings to avoid |
1129 | // => try to "fast-match" such strings to avoid |
… | |
… | |
1102 | if (!key) |
1143 | if (!key) |
1103 | goto fail; |
1144 | goto fail; |
1104 | |
1145 | |
1105 | decode_ws (dec); EXPECT_CH (':'); |
1146 | decode_ws (dec); EXPECT_CH (':'); |
1106 | |
1147 | |
|
|
1148 | decode_ws (dec); |
1107 | value = decode_sv (dec); |
1149 | value = decode_sv (dec); |
1108 | if (!value) |
1150 | if (!value) |
1109 | { |
1151 | { |
1110 | SvREFCNT_dec (key); |
1152 | SvREFCNT_dec (key); |
1111 | goto fail; |
1153 | goto fail; |
… | |
… | |
1123 | int len = p - key; |
1165 | int len = p - key; |
1124 | dec->cur = p + 1; |
1166 | dec->cur = p + 1; |
1125 | |
1167 | |
1126 | decode_ws (dec); EXPECT_CH (':'); |
1168 | decode_ws (dec); EXPECT_CH (':'); |
1127 | |
1169 | |
|
|
1170 | decode_ws (dec); |
1128 | value = decode_sv (dec); |
1171 | value = decode_sv (dec); |
1129 | if (!value) |
1172 | if (!value) |
1130 | goto fail; |
1173 | goto fail; |
1131 | |
1174 | |
1132 | hv_store (hv, key, len, value, 0); |
1175 | hv_store (hv, key, len, value, 0); |
… | |
… | |
1148 | |
1191 | |
1149 | if (*dec->cur != ',') |
1192 | if (*dec->cur != ',') |
1150 | ERR (", or } expected while parsing object/hash"); |
1193 | ERR (", or } expected while parsing object/hash"); |
1151 | |
1194 | |
1152 | ++dec->cur; |
1195 | ++dec->cur; |
|
|
1196 | |
|
|
1197 | decode_ws (dec); |
|
|
1198 | |
|
|
1199 | if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
|
|
1200 | { |
|
|
1201 | ++dec->cur; |
|
|
1202 | break; |
|
|
1203 | } |
1153 | } |
1204 | } |
1154 | |
1205 | |
1155 | DEC_DEC_DEPTH; |
1206 | DEC_DEC_DEPTH; |
1156 | sv = newRV_noinc ((SV *)hv); |
1207 | sv = newRV_noinc ((SV *)hv); |
1157 | |
1208 | |
… | |
… | |
1222 | } |
1273 | } |
1223 | |
1274 | |
1224 | static SV * |
1275 | static SV * |
1225 | decode_sv (dec_t *dec) |
1276 | decode_sv (dec_t *dec) |
1226 | { |
1277 | { |
1227 | decode_ws (dec); |
|
|
1228 | |
|
|
1229 | // the beauty of JSON: you need exactly one character lookahead |
1278 | // the beauty of JSON: you need exactly one character lookahead |
1230 | // to parse anything. |
1279 | // to parse anything. |
1231 | switch (*dec->cur) |
1280 | switch (*dec->cur) |
1232 | { |
1281 | { |
1233 | case '"': ++dec->cur; return decode_str (dec); |
1282 | case '"': ++dec->cur; return decode_str (dec); |
… | |
… | |
1317 | |
1366 | |
1318 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1367 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1319 | dec.json.flags |= F_HOOK; |
1368 | dec.json.flags |= F_HOOK; |
1320 | |
1369 | |
1321 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1370 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1371 | |
|
|
1372 | decode_ws (&dec); |
1322 | sv = decode_sv (&dec); |
1373 | sv = decode_sv (&dec); |
1323 | |
1374 | |
1324 | if (!(offset_return || !sv)) |
1375 | if (!(offset_return || !sv)) |
1325 | { |
1376 | { |
1326 | // check for trailing garbage |
1377 | // check for trailing garbage |
… | |
… | |
1423 | pretty = F_PRETTY |
1474 | pretty = F_PRETTY |
1424 | allow_nonref = F_ALLOW_NONREF |
1475 | allow_nonref = F_ALLOW_NONREF |
1425 | shrink = F_SHRINK |
1476 | shrink = F_SHRINK |
1426 | allow_blessed = F_ALLOW_BLESSED |
1477 | allow_blessed = F_ALLOW_BLESSED |
1427 | convert_blessed = F_CONV_BLESSED |
1478 | convert_blessed = F_CONV_BLESSED |
|
|
1479 | relaxed = F_RELAXED |
1428 | PPCODE: |
1480 | PPCODE: |
1429 | { |
1481 | { |
1430 | if (enable) |
1482 | if (enable) |
1431 | self->flags |= ix; |
1483 | self->flags |= ix; |
1432 | else |
1484 | else |
1433 | self->flags &= ~ix; |
1485 | self->flags &= ~ix; |
1434 | |
1486 | |
1435 | XPUSHs (ST (0)); |
1487 | XPUSHs (ST (0)); |
1436 | } |
1488 | } |
1437 | |
1489 | |
|
|
1490 | void get_ascii (JSON *self) |
|
|
1491 | ALIAS: |
|
|
1492 | get_ascii = F_ASCII |
|
|
1493 | get_latin1 = F_LATIN1 |
|
|
1494 | get_utf8 = F_UTF8 |
|
|
1495 | get_indent = F_INDENT |
|
|
1496 | get_canonical = F_CANONICAL |
|
|
1497 | get_space_before = F_SPACE_BEFORE |
|
|
1498 | get_space_after = F_SPACE_AFTER |
|
|
1499 | get_pretty = F_PRETTY |
|
|
1500 | get_allow_nonref = F_ALLOW_NONREF |
|
|
1501 | get_shrink = F_SHRINK |
|
|
1502 | get_allow_blessed = F_ALLOW_BLESSED |
|
|
1503 | get_convert_blessed = F_CONV_BLESSED |
|
|
1504 | get_relaxed = F_RELAXED |
|
|
1505 | PPCODE: |
|
|
1506 | XPUSHs (boolSV (self->flags & ix)); |
|
|
1507 | |
1438 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1508 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1439 | PPCODE: |
1509 | PPCODE: |
1440 | { |
1510 | { |
1441 | UV log2 = 0; |
1511 | UV log2 = 0; |
1442 | |
1512 | |
… | |
… | |
1448 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1518 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1449 | |
1519 | |
1450 | XPUSHs (ST (0)); |
1520 | XPUSHs (ST (0)); |
1451 | } |
1521 | } |
1452 | |
1522 | |
|
|
1523 | U32 get_max_depth (JSON *self) |
|
|
1524 | CODE: |
|
|
1525 | RETVAL = DEC_DEPTH (self->flags); |
|
|
1526 | OUTPUT: |
|
|
1527 | RETVAL |
|
|
1528 | |
1453 | void max_size (JSON *self, UV max_size = 0) |
1529 | void max_size (JSON *self, UV max_size = 0) |
1454 | PPCODE: |
1530 | PPCODE: |
1455 | { |
1531 | { |
1456 | UV log2 = 0; |
1532 | UV log2 = 0; |
1457 | |
1533 | |
… | |
… | |
1463 | |
1539 | |
1464 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
1540 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
1465 | |
1541 | |
1466 | XPUSHs (ST (0)); |
1542 | XPUSHs (ST (0)); |
1467 | } |
1543 | } |
|
|
1544 | |
|
|
1545 | int get_max_size (JSON *self) |
|
|
1546 | CODE: |
|
|
1547 | RETVAL = DEC_SIZE (self->flags); |
|
|
1548 | OUTPUT: |
|
|
1549 | RETVAL |
1468 | |
1550 | |
1469 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1551 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1470 | PPCODE: |
1552 | PPCODE: |
1471 | { |
1553 | { |
1472 | SvREFCNT_dec (self->cb_object); |
1554 | SvREFCNT_dec (self->cb_object); |