… | |
… | |
175 | *s++ = 0x80 | ( ch & 0x3f); |
175 | *s++ = 0x80 | ( ch & 0x3f); |
176 | |
176 | |
177 | return s; |
177 | return s; |
178 | } |
178 | } |
179 | |
179 | |
|
|
180 | // convert offset pointer to character index, sv must be string |
|
|
181 | static STRLEN |
|
|
182 | ptr_to_index (SV *sv, char *offset) |
|
|
183 | { |
|
|
184 | return SvUTF8 (sv) |
|
|
185 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
186 | : offset - SvPVX (sv); |
|
|
187 | } |
|
|
188 | |
180 | ///////////////////////////////////////////////////////////////////////////// |
189 | ///////////////////////////////////////////////////////////////////////////// |
181 | // encoder |
190 | // encoder |
182 | |
191 | |
183 | // structure used for encoding JSON |
192 | // structure used for encoding JSON |
184 | typedef struct |
193 | typedef struct |
… | |
… | |
195 | need (enc_t *enc, STRLEN len) |
204 | need (enc_t *enc, STRLEN len) |
196 | { |
205 | { |
197 | if (expect_false (enc->cur + len >= enc->end)) |
206 | if (expect_false (enc->cur + len >= enc->end)) |
198 | { |
207 | { |
199 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
208 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
200 | SvGROW (enc->sv, cur + len + 1); |
209 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
201 | enc->cur = SvPVX (enc->sv) + cur; |
210 | enc->cur = SvPVX (enc->sv) + cur; |
202 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
211 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
203 | } |
212 | } |
204 | } |
213 | } |
205 | |
214 | |
… | |
… | |
462 | |
471 | |
463 | // for canonical output we have to sort by keys first |
472 | // for canonical output we have to sort by keys first |
464 | // actually, this is mostly due to the stupid so-called |
473 | // actually, this is mostly due to the stupid so-called |
465 | // security workaround added somewhere in 5.8.x |
474 | // security workaround added somewhere in 5.8.x |
466 | // that randomises hash orderings |
475 | // that randomises hash orderings |
467 | if (enc->json.flags & F_CANONICAL) |
476 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
468 | { |
477 | { |
469 | int count = hv_iterinit (hv); |
478 | int count = hv_iterinit (hv); |
470 | |
479 | |
471 | if (SvMAGICAL (hv)) |
480 | if (SvMAGICAL (hv)) |
472 | { |
481 | { |
… | |
… | |
970 | { |
979 | { |
971 | STRLEN len = cur - buf; |
980 | STRLEN len = cur - buf; |
972 | |
981 | |
973 | if (sv) |
982 | if (sv) |
974 | { |
983 | { |
975 | SvGROW (sv, SvCUR (sv) + len + 1); |
984 | STRLEN cur = SvCUR (sv); |
|
|
985 | |
|
|
986 | if (SvLEN (sv) <= cur + len) |
|
|
987 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
988 | |
976 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
989 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
977 | SvCUR_set (sv, SvCUR (sv) + len); |
990 | SvCUR_set (sv, SvCUR (sv) + len); |
978 | } |
991 | } |
979 | else |
992 | else |
980 | sv = newSVpvn (buf, len); |
993 | sv = newSVpvn (buf, len); |
… | |
… | |
1071 | |
1084 | |
1072 | // special case the rather common 1..5-digit-int case |
1085 | // special case the rather common 1..5-digit-int case |
1073 | if (*start == '-') |
1086 | if (*start == '-') |
1074 | switch (len) |
1087 | switch (len) |
1075 | { |
1088 | { |
1076 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1089 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
1077 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1090 | case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11)); |
1078 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1091 | case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1079 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1092 | case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1080 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1093 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1081 | } |
1094 | } |
1082 | else |
1095 | else |
1083 | switch (len) |
1096 | switch (len) |
1084 | { |
1097 | { |
1085 | case 1: return newSViv ( start [0] - '0' * 1); |
1098 | case 1: return newSViv ( start [0] - '0' * 1); |
1086 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1099 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1087 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1100 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1088 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1101 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1089 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1102 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1090 | } |
1103 | } |
1091 | |
1104 | |
1092 | { |
1105 | { |
1093 | UV uv; |
1106 | UV uv; |
1094 | int numtype = grok_number (start, len, &uv); |
1107 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1404 | fail: |
1417 | fail: |
1405 | return 0; |
1418 | return 0; |
1406 | } |
1419 | } |
1407 | |
1420 | |
1408 | static SV * |
1421 | static SV * |
1409 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1422 | decode_json (SV *string, JSON *json, char **offset_return) |
1410 | { |
1423 | { |
1411 | dec_t dec; |
1424 | dec_t dec; |
1412 | STRLEN offset; |
|
|
1413 | SV *sv; |
1425 | SV *sv; |
1414 | |
1426 | |
1415 | /* work around bugs in 5.10 where manipulating magic values |
1427 | /* work around bugs in 5.10 where manipulating magic values |
1416 | * will perl ignore the magic in subsequent accesses |
1428 | * will perl ignore the magic in subsequent accesses |
1417 | */ |
1429 | */ |
… | |
… | |
1429 | * and hope for the best. |
1441 | * and hope for the best. |
1430 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
1442 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
1431 | * assertion business is seriously broken, try yet another workaround |
1443 | * assertion business is seriously broken, try yet another workaround |
1432 | * for the broken -DDEBUGGING. |
1444 | * for the broken -DDEBUGGING. |
1433 | */ |
1445 | */ |
|
|
1446 | { |
1434 | #ifdef DEBUGGING |
1447 | #ifdef DEBUGGING |
1435 | offset = SvOK (string) ? sv_len (string) : 0; |
1448 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
1436 | #else |
1449 | #else |
1437 | offset = SvCUR (string); |
1450 | STRLEN offset = SvCUR (string); |
1438 | #endif |
1451 | #endif |
1439 | |
1452 | |
1440 | if (offset > json->max_size && json->max_size) |
1453 | if (offset > json->max_size && json->max_size) |
1441 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1454 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1442 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1455 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1456 | } |
1443 | |
1457 | |
1444 | if (json->flags & F_UTF8) |
1458 | if (json->flags & F_UTF8) |
1445 | sv_utf8_downgrade (string, 0); |
1459 | sv_utf8_downgrade (string, 0); |
1446 | else |
1460 | else |
1447 | sv_utf8_upgrade (string); |
1461 | sv_utf8_upgrade (string); |
… | |
… | |
1459 | |
1473 | |
1460 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1474 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1461 | |
1475 | |
1462 | decode_ws (&dec); |
1476 | decode_ws (&dec); |
1463 | sv = decode_sv (&dec); |
1477 | sv = decode_sv (&dec); |
|
|
1478 | |
|
|
1479 | if (offset_return) |
|
|
1480 | *offset_return = dec.cur; |
1464 | |
1481 | |
1465 | if (!(offset_return || !sv)) |
1482 | if (!(offset_return || !sv)) |
1466 | { |
1483 | { |
1467 | // check for trailing garbage |
1484 | // check for trailing garbage |
1468 | decode_ws (&dec); |
1485 | decode_ws (&dec); |
… | |
… | |
1471 | { |
1488 | { |
1472 | dec.err = "garbage after JSON object"; |
1489 | dec.err = "garbage after JSON object"; |
1473 | SvREFCNT_dec (sv); |
1490 | SvREFCNT_dec (sv); |
1474 | sv = 0; |
1491 | sv = 0; |
1475 | } |
1492 | } |
1476 | } |
|
|
1477 | |
|
|
1478 | if (offset_return || !sv) |
|
|
1479 | { |
|
|
1480 | offset = dec.json.flags & F_UTF8 |
|
|
1481 | ? dec.cur - SvPVX (string) |
|
|
1482 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1483 | |
|
|
1484 | if (offset_return) |
|
|
1485 | *offset_return = offset; |
|
|
1486 | } |
1493 | } |
1487 | |
1494 | |
1488 | if (!sv) |
1495 | if (!sv) |
1489 | { |
1496 | { |
1490 | SV *uni = sv_newmortal (); |
1497 | SV *uni = sv_newmortal (); |
… | |
… | |
1496 | SAVEVPTR (PL_curcop); |
1503 | SAVEVPTR (PL_curcop); |
1497 | PL_curcop = &cop; |
1504 | PL_curcop = &cop; |
1498 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1505 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1499 | LEAVE; |
1506 | LEAVE; |
1500 | |
1507 | |
1501 | croak ("%s, at character offset %d [\"%s\"]", |
1508 | croak ("%s, at character offset %d (before \"%s\")", |
1502 | dec.err, |
1509 | dec.err, |
1503 | (int)offset, |
1510 | ptr_to_index (string, dec.cur), |
1504 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1511 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1505 | } |
1512 | } |
1506 | |
1513 | |
1507 | sv = sv_2mortal (sv); |
1514 | sv = sv_2mortal (sv); |
1508 | |
1515 | |
… | |
… | |
1777 | XPUSHs (decode_json (jsonstr, self, 0)); |
1784 | XPUSHs (decode_json (jsonstr, self, 0)); |
1778 | |
1785 | |
1779 | void decode_prefix (JSON *self, SV *jsonstr) |
1786 | void decode_prefix (JSON *self, SV *jsonstr) |
1780 | PPCODE: |
1787 | PPCODE: |
1781 | { |
1788 | { |
1782 | STRLEN offset; |
1789 | char *offset; |
1783 | EXTEND (SP, 2); |
1790 | EXTEND (SP, 2); |
1784 | PUSHs (decode_json (jsonstr, self, &offset)); |
1791 | PUSHs (decode_json (jsonstr, self, &offset)); |
1785 | PUSHs (sv_2mortal (newSVuv (offset))); |
1792 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
1786 | } |
1793 | } |
1787 | |
1794 | |
1788 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1795 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1789 | PPCODE: |
1796 | PPCODE: |
1790 | { |
1797 | { |
… | |
… | |
1792 | self->incr_text = newSVpvn ("", 0); |
1799 | self->incr_text = newSVpvn ("", 0); |
1793 | |
1800 | |
1794 | // append data, if any |
1801 | // append data, if any |
1795 | if (jsonstr) |
1802 | if (jsonstr) |
1796 | { |
1803 | { |
1797 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
1804 | if (SvUTF8 (jsonstr)) |
1798 | { |
1805 | { |
|
|
1806 | if (!SvUTF8 (self->incr_text)) |
|
|
1807 | { |
1799 | /* utf-8-ness differs, need to upgrade */ |
1808 | /* utf-8-ness differs, need to upgrade */ |
1800 | sv_utf8_upgrade (self->incr_text); |
1809 | sv_utf8_upgrade (self->incr_text); |
1801 | |
1810 | |
1802 | if (self->incr_pos) |
1811 | if (self->incr_pos) |
1803 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
1812 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
1804 | - (U8 *)SvPVX (self->incr_text); |
1813 | - (U8 *)SvPVX (self->incr_text); |
|
|
1814 | } |
1805 | } |
1815 | } |
|
|
1816 | else if (SvUTF8 (self->incr_text)) |
|
|
1817 | sv_utf8_upgrade (jsonstr); |
1806 | |
1818 | |
1807 | { |
1819 | { |
1808 | STRLEN len; |
1820 | STRLEN len; |
1809 | const char *str = SvPV (jsonstr, len); |
1821 | const char *str = SvPV (jsonstr, len); |
1810 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
1822 | STRLEN cur = SvCUR (self->incr_text); |
|
|
1823 | |
|
|
1824 | if (SvLEN (self->incr_text) <= cur + len) |
|
|
1825 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
1826 | |
1811 | Move (str, SvEND (self->incr_text), len, char); |
1827 | Move (str, SvEND (self->incr_text), len, char); |
1812 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1828 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1813 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
1829 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
1814 | } |
1830 | } |
1815 | } |
1831 | } |
1816 | |
1832 | |
1817 | if (GIMME_V != G_VOID) |
1833 | if (GIMME_V != G_VOID) |
1818 | do |
1834 | do |
1819 | { |
1835 | { |
1820 | STRLEN offset; |
1836 | char *offset; |
1821 | |
1837 | |
1822 | if (!INCR_DONE (self)) |
1838 | if (!INCR_DONE (self)) |
1823 | { |
1839 | { |
1824 | incr_parse (self); |
1840 | incr_parse (self); |
1825 | |
1841 | |
… | |
… | |
1831 | break; |
1847 | break; |
1832 | } |
1848 | } |
1833 | |
1849 | |
1834 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1850 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1835 | |
1851 | |
1836 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1837 | self->incr_pos -= offset; |
1852 | self->incr_pos -= offset - SvPVX (self->incr_text); |
1838 | self->incr_nest = 0; |
1853 | self->incr_nest = 0; |
1839 | self->incr_mode = 0; |
1854 | self->incr_mode = 0; |
|
|
1855 | |
|
|
1856 | sv_chop (self->incr_text, offset); |
1840 | } |
1857 | } |
1841 | while (GIMME_V == G_ARRAY); |
1858 | while (GIMME_V == G_ARRAY); |
1842 | } |
1859 | } |
1843 | |
1860 | |
1844 | SV *incr_text (JSON *self) |
1861 | SV *incr_text (JSON *self) |