… | |
… | |
79 | INCR_M_STR, // inside string |
79 | INCR_M_STR, // inside string |
80 | INCR_M_BS, // inside backslash |
80 | INCR_M_BS, // inside backslash |
81 | INCR_M_JSON // outside anything, count nesting |
81 | INCR_M_JSON // outside anything, count nesting |
82 | }; |
82 | }; |
83 | |
83 | |
84 | #define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) |
84 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
85 | |
85 | |
86 | typedef struct { |
86 | typedef struct { |
87 | U32 flags; |
87 | U32 flags; |
88 | U32 max_depth; |
88 | U32 max_depth; |
89 | STRLEN max_size; |
89 | STRLEN max_size; |
… | |
… | |
92 | HV *cb_sk_object; |
92 | HV *cb_sk_object; |
93 | |
93 | |
94 | // for the incremental parser |
94 | // for the incremental parser |
95 | SV *incr_text; // the source text so far |
95 | SV *incr_text; // the source text so far |
96 | STRLEN incr_pos; // the current offset into the text |
96 | STRLEN incr_pos; // the current offset into the text |
97 | unsigned char incr_nest; // {[]}-nesting level |
97 | int incr_nest; // {[]}-nesting level |
98 | unsigned char incr_mode; |
98 | unsigned char incr_mode; |
99 | } JSON; |
99 | } JSON; |
100 | |
100 | |
101 | INLINE void |
101 | INLINE void |
102 | json_init (JSON *json) |
102 | json_init (JSON *json) |
… | |
… | |
121 | |
121 | |
122 | INLINE void |
122 | INLINE void |
123 | shrink (SV *sv) |
123 | shrink (SV *sv) |
124 | { |
124 | { |
125 | sv_utf8_downgrade (sv, 1); |
125 | sv_utf8_downgrade (sv, 1); |
|
|
126 | |
126 | if (SvLEN (sv) > SvCUR (sv) + 1) |
127 | if (SvLEN (sv) > SvCUR (sv) + 1) |
127 | { |
128 | { |
128 | #ifdef SvPV_shrink_to_cur |
129 | #ifdef SvPV_shrink_to_cur |
129 | SvPV_shrink_to_cur (sv); |
130 | SvPV_shrink_to_cur (sv); |
130 | #elif defined (SvPV_renew) |
131 | #elif defined (SvPV_renew) |
… | |
… | |
174 | *s++ = 0x80 | ( ch & 0x3f); |
175 | *s++ = 0x80 | ( ch & 0x3f); |
175 | |
176 | |
176 | return s; |
177 | return s; |
177 | } |
178 | } |
178 | |
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 | |
179 | ///////////////////////////////////////////////////////////////////////////// |
189 | ///////////////////////////////////////////////////////////////////////////// |
180 | // encoder |
190 | // encoder |
181 | |
191 | |
182 | // structure used for encoding JSON |
192 | // structure used for encoding JSON |
183 | typedef struct |
193 | typedef struct |
… | |
… | |
193 | INLINE void |
203 | INLINE void |
194 | need (enc_t *enc, STRLEN len) |
204 | need (enc_t *enc, STRLEN len) |
195 | { |
205 | { |
196 | if (expect_false (enc->cur + len >= enc->end)) |
206 | if (expect_false (enc->cur + len >= enc->end)) |
197 | { |
207 | { |
198 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
208 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
199 | SvGROW (enc->sv, cur + len + 1); |
209 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
200 | enc->cur = SvPVX (enc->sv) + cur; |
210 | enc->cur = SvPVX (enc->sv) + cur; |
201 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
211 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
202 | } |
212 | } |
203 | } |
213 | } |
204 | |
214 | |
… | |
… | |
279 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
289 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
280 | enc->cur += 12; |
290 | enc->cur += 12; |
281 | } |
291 | } |
282 | else |
292 | else |
283 | { |
293 | { |
284 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
285 | need (enc, len += 5); |
294 | need (enc, len += 5); |
286 | *enc->cur++ = '\\'; |
295 | *enc->cur++ = '\\'; |
287 | *enc->cur++ = 'u'; |
296 | *enc->cur++ = 'u'; |
288 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
297 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
289 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
298 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
290 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
299 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
291 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
300 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
292 | } |
301 | } |
293 | |
302 | |
294 | str += clen; |
303 | str += clen; |
295 | } |
304 | } |
296 | else if (enc->json.flags & F_LATIN1) |
305 | else if (enc->json.flags & F_LATIN1) |
… | |
… | |
460 | |
469 | |
461 | encode_ch (enc, '{'); |
470 | encode_ch (enc, '{'); |
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) |
468 | { |
477 | { |
469 | int count = hv_iterinit (hv); |
478 | int count = hv_iterinit (hv); |
470 | |
479 | |
… | |
… | |
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); |
… | |
… | |
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 | |
|
|
1427 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1428 | * will perl ignore the magic in subsequent accesses |
|
|
1429 | */ |
1415 | SvGETMAGIC (string); |
1430 | /*SvGETMAGIC (string);*/ |
|
|
1431 | if (SvMAGICAL (string)) |
|
|
1432 | string = sv_2mortal (newSVsv (string)); |
|
|
1433 | |
1416 | SvUPGRADE (string, SVt_PV); |
1434 | SvUPGRADE (string, SVt_PV); |
1417 | |
1435 | |
|
|
1436 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1437 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1438 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1439 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1440 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1441 | * and hope for the best. |
|
|
1442 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1443 | * assertion business is seriously broken, try yet another workaround |
|
|
1444 | * for the broken -DDEBUGGING. |
|
|
1445 | */ |
|
|
1446 | { |
|
|
1447 | #ifdef DEBUGGING |
|
|
1448 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1449 | #else |
|
|
1450 | STRLEN offset = SvCUR (string); |
|
|
1451 | #endif |
|
|
1452 | |
1418 | if (SvCUR (string) > json->max_size && json->max_size) |
1453 | if (offset > json->max_size && json->max_size) |
1419 | 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", |
1420 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1455 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1456 | } |
1421 | |
1457 | |
1422 | if (json->flags & F_UTF8) |
1458 | if (json->flags & F_UTF8) |
1423 | sv_utf8_downgrade (string, 0); |
1459 | sv_utf8_downgrade (string, 0); |
1424 | else |
1460 | else |
1425 | sv_utf8_upgrade (string); |
1461 | sv_utf8_upgrade (string); |
… | |
… | |
1437 | |
1473 | |
1438 | *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 |
1439 | |
1475 | |
1440 | decode_ws (&dec); |
1476 | decode_ws (&dec); |
1441 | sv = decode_sv (&dec); |
1477 | sv = decode_sv (&dec); |
|
|
1478 | |
|
|
1479 | if (offset_return) |
|
|
1480 | *offset_return = dec.cur; |
1442 | |
1481 | |
1443 | if (!(offset_return || !sv)) |
1482 | if (!(offset_return || !sv)) |
1444 | { |
1483 | { |
1445 | // check for trailing garbage |
1484 | // check for trailing garbage |
1446 | decode_ws (&dec); |
1485 | decode_ws (&dec); |
… | |
… | |
1449 | { |
1488 | { |
1450 | dec.err = "garbage after JSON object"; |
1489 | dec.err = "garbage after JSON object"; |
1451 | SvREFCNT_dec (sv); |
1490 | SvREFCNT_dec (sv); |
1452 | sv = 0; |
1491 | sv = 0; |
1453 | } |
1492 | } |
1454 | } |
|
|
1455 | |
|
|
1456 | if (offset_return || !sv) |
|
|
1457 | { |
|
|
1458 | offset = dec.json.flags & F_UTF8 |
|
|
1459 | ? dec.cur - SvPVX (string) |
|
|
1460 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1461 | |
|
|
1462 | if (offset_return) |
|
|
1463 | *offset_return = offset; |
|
|
1464 | } |
1493 | } |
1465 | |
1494 | |
1466 | if (!sv) |
1495 | if (!sv) |
1467 | { |
1496 | { |
1468 | SV *uni = sv_newmortal (); |
1497 | SV *uni = sv_newmortal (); |
… | |
… | |
1474 | SAVEVPTR (PL_curcop); |
1503 | SAVEVPTR (PL_curcop); |
1475 | PL_curcop = &cop; |
1504 | PL_curcop = &cop; |
1476 | 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); |
1477 | LEAVE; |
1506 | LEAVE; |
1478 | |
1507 | |
1479 | croak ("%s, at character offset %d [\"%s\"]", |
1508 | croak ("%s, at character offset %d (before \"%s\")", |
1480 | dec.err, |
1509 | dec.err, |
1481 | (int)offset, |
1510 | ptr_to_index (string, dec.cur), |
1482 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1511 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1483 | } |
1512 | } |
1484 | |
1513 | |
1485 | sv = sv_2mortal (sv); |
1514 | sv = sv_2mortal (sv); |
1486 | |
1515 | |
… | |
… | |
1590 | croak (ERR_NESTING_EXCEEDED); |
1619 | croak (ERR_NESTING_EXCEEDED); |
1591 | break; |
1620 | break; |
1592 | |
1621 | |
1593 | case ']': |
1622 | case ']': |
1594 | case '}': |
1623 | case '}': |
1595 | if (!--self->incr_nest) |
1624 | if (--self->incr_nest <= 0) |
1596 | goto interrupt; |
1625 | goto interrupt; |
1597 | } |
1626 | } |
1598 | } |
1627 | } |
1599 | } |
1628 | } |
1600 | |
1629 | |
… | |
… | |
1755 | XPUSHs (decode_json (jsonstr, self, 0)); |
1784 | XPUSHs (decode_json (jsonstr, self, 0)); |
1756 | |
1785 | |
1757 | void decode_prefix (JSON *self, SV *jsonstr) |
1786 | void decode_prefix (JSON *self, SV *jsonstr) |
1758 | PPCODE: |
1787 | PPCODE: |
1759 | { |
1788 | { |
1760 | STRLEN offset; |
1789 | char *offset; |
1761 | EXTEND (SP, 2); |
1790 | EXTEND (SP, 2); |
1762 | PUSHs (decode_json (jsonstr, self, &offset)); |
1791 | PUSHs (decode_json (jsonstr, self, &offset)); |
1763 | PUSHs (sv_2mortal (newSVuv (offset))); |
1792 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
1764 | } |
1793 | } |
1765 | |
1794 | |
1766 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1795 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1767 | PPCODE: |
1796 | PPCODE: |
1768 | { |
1797 | { |
… | |
… | |
1770 | self->incr_text = newSVpvn ("", 0); |
1799 | self->incr_text = newSVpvn ("", 0); |
1771 | |
1800 | |
1772 | // append data, if any |
1801 | // append data, if any |
1773 | if (jsonstr) |
1802 | if (jsonstr) |
1774 | { |
1803 | { |
1775 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
1804 | if (SvUTF8 (jsonstr)) |
1776 | { |
1805 | { |
|
|
1806 | if (!SvUTF8 (self->incr_text)) |
|
|
1807 | { |
1777 | /* utf-8-ness differs, need to upgrade */ |
1808 | /* utf-8-ness differs, need to upgrade */ |
1778 | sv_utf8_upgrade (self->incr_text); |
1809 | sv_utf8_upgrade (self->incr_text); |
1779 | |
1810 | |
1780 | if (self->incr_pos) |
1811 | if (self->incr_pos) |
1781 | 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) |
1782 | - (U8 *)SvPVX (self->incr_text); |
1813 | - (U8 *)SvPVX (self->incr_text); |
|
|
1814 | } |
1783 | } |
1815 | } |
|
|
1816 | else if (SvUTF8 (self->incr_text)) |
|
|
1817 | sv_utf8_upgrade (jsonstr); |
1784 | |
1818 | |
1785 | { |
1819 | { |
1786 | STRLEN len; |
1820 | STRLEN len; |
1787 | const char *str = SvPV (jsonstr, len); |
1821 | const char *str = SvPV (jsonstr, len); |
1788 | 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 | |
1789 | Move (str, SvEND (self->incr_text), len, char); |
1827 | Move (str, SvEND (self->incr_text), len, char); |
1790 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1828 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1791 | *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 |
1792 | } |
1830 | } |
1793 | } |
1831 | } |
1794 | |
1832 | |
1795 | if (GIMME_V != G_VOID) |
1833 | if (GIMME_V != G_VOID) |
1796 | do |
1834 | do |
1797 | { |
1835 | { |
1798 | STRLEN offset; |
1836 | char *offset; |
1799 | |
1837 | |
1800 | if (!INCR_DONE (self)) |
1838 | if (!INCR_DONE (self)) |
1801 | { |
1839 | { |
1802 | incr_parse (self); |
1840 | incr_parse (self); |
1803 | |
1841 | |
… | |
… | |
1809 | break; |
1847 | break; |
1810 | } |
1848 | } |
1811 | |
1849 | |
1812 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1850 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1813 | |
1851 | |
1814 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1815 | self->incr_pos -= offset; |
1852 | self->incr_pos -= offset - SvPVX (self->incr_text); |
1816 | self->incr_nest = 0; |
1853 | self->incr_nest = 0; |
1817 | self->incr_mode = 0; |
1854 | self->incr_mode = 0; |
|
|
1855 | |
|
|
1856 | sv_chop (self->incr_text, offset); |
1818 | } |
1857 | } |
1819 | while (GIMME_V == G_ARRAY); |
1858 | while (GIMME_V == G_ARRAY); |
1820 | } |
1859 | } |
1821 | |
1860 | |
1822 | SV *incr_text (JSON *self) |
1861 | SV *incr_text (JSON *self) |
… | |
… | |
1841 | self->incr_nest = 0; |
1880 | self->incr_nest = 0; |
1842 | self->incr_mode = 0; |
1881 | self->incr_mode = 0; |
1843 | } |
1882 | } |
1844 | } |
1883 | } |
1845 | |
1884 | |
|
|
1885 | void incr_reset (JSON *self) |
|
|
1886 | CODE: |
|
|
1887 | { |
|
|
1888 | SvREFCNT_dec (self->incr_text); |
|
|
1889 | self->incr_text = 0; |
|
|
1890 | self->incr_pos = 0; |
|
|
1891 | self->incr_nest = 0; |
|
|
1892 | self->incr_mode = 0; |
|
|
1893 | } |
|
|
1894 | |
1846 | void DESTROY (JSON *self) |
1895 | void DESTROY (JSON *self) |
1847 | CODE: |
1896 | CODE: |
1848 | SvREFCNT_dec (self->cb_sk_object); |
1897 | SvREFCNT_dec (self->cb_sk_object); |
1849 | SvREFCNT_dec (self->cb_object); |
1898 | SvREFCNT_dec (self->cb_object); |
1850 | SvREFCNT_dec (self->incr_text); |
1899 | SvREFCNT_dec (self->incr_text); |