… | |
… | |
173 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
173 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
174 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
174 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
175 | *s++ = 0x80 | ( ch & 0x3f); |
175 | *s++ = 0x80 | ( ch & 0x3f); |
176 | |
176 | |
177 | return s; |
177 | return s; |
|
|
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); |
178 | } |
187 | } |
179 | |
188 | |
180 | ///////////////////////////////////////////////////////////////////////////// |
189 | ///////////////////////////////////////////////////////////////////////////// |
181 | // encoder |
190 | // encoder |
182 | |
191 | |
… | |
… | |
1408 | fail: |
1417 | fail: |
1409 | return 0; |
1418 | return 0; |
1410 | } |
1419 | } |
1411 | |
1420 | |
1412 | static SV * |
1421 | static SV * |
1413 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1422 | decode_json (SV *string, JSON *json, char **offset_return) |
1414 | { |
1423 | { |
1415 | dec_t dec; |
1424 | dec_t dec; |
1416 | STRLEN offset; |
|
|
1417 | SV *sv; |
1425 | SV *sv; |
1418 | |
1426 | |
1419 | /* work around bugs in 5.10 where manipulating magic values |
1427 | /* work around bugs in 5.10 where manipulating magic values |
1420 | * will perl ignore the magic in subsequent accesses |
1428 | * will perl ignore the magic in subsequent accesses |
1421 | */ |
1429 | */ |
… | |
… | |
1433 | * and hope for the best. |
1441 | * and hope for the best. |
1434 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
1442 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
1435 | * assertion business is seriously broken, try yet another workaround |
1443 | * assertion business is seriously broken, try yet another workaround |
1436 | * for the broken -DDEBUGGING. |
1444 | * for the broken -DDEBUGGING. |
1437 | */ |
1445 | */ |
|
|
1446 | { |
1438 | #ifdef DEBUGGING |
1447 | #ifdef DEBUGGING |
1439 | offset = SvOK (string) ? sv_len (string) : 0; |
1448 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
1440 | #else |
1449 | #else |
1441 | offset = SvCUR (string); |
1450 | STRLEN offset = SvCUR (string); |
1442 | #endif |
1451 | #endif |
1443 | |
1452 | |
1444 | if (offset > json->max_size && json->max_size) |
1453 | if (offset > json->max_size && json->max_size) |
1445 | 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", |
1446 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1455 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1456 | } |
1447 | |
1457 | |
1448 | if (json->flags & F_UTF8) |
1458 | if (json->flags & F_UTF8) |
1449 | sv_utf8_downgrade (string, 0); |
1459 | sv_utf8_downgrade (string, 0); |
1450 | else |
1460 | else |
1451 | sv_utf8_upgrade (string); |
1461 | sv_utf8_upgrade (string); |
… | |
… | |
1463 | |
1473 | |
1464 | *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 |
1465 | |
1475 | |
1466 | decode_ws (&dec); |
1476 | decode_ws (&dec); |
1467 | sv = decode_sv (&dec); |
1477 | sv = decode_sv (&dec); |
|
|
1478 | |
|
|
1479 | if (offset_return) |
|
|
1480 | *offset_return = dec.cur; |
1468 | |
1481 | |
1469 | if (!(offset_return || !sv)) |
1482 | if (!(offset_return || !sv)) |
1470 | { |
1483 | { |
1471 | // check for trailing garbage |
1484 | // check for trailing garbage |
1472 | decode_ws (&dec); |
1485 | decode_ws (&dec); |
… | |
… | |
1475 | { |
1488 | { |
1476 | dec.err = "garbage after JSON object"; |
1489 | dec.err = "garbage after JSON object"; |
1477 | SvREFCNT_dec (sv); |
1490 | SvREFCNT_dec (sv); |
1478 | sv = 0; |
1491 | sv = 0; |
1479 | } |
1492 | } |
1480 | } |
|
|
1481 | |
|
|
1482 | if (offset_return || !sv) |
|
|
1483 | { |
|
|
1484 | offset = dec.json.flags & F_UTF8 |
|
|
1485 | ? dec.cur - SvPVX (string) |
|
|
1486 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1487 | |
|
|
1488 | if (offset_return) |
|
|
1489 | *offset_return = offset; |
|
|
1490 | } |
1493 | } |
1491 | |
1494 | |
1492 | if (!sv) |
1495 | if (!sv) |
1493 | { |
1496 | { |
1494 | SV *uni = sv_newmortal (); |
1497 | SV *uni = sv_newmortal (); |
… | |
… | |
1500 | SAVEVPTR (PL_curcop); |
1503 | SAVEVPTR (PL_curcop); |
1501 | PL_curcop = &cop; |
1504 | PL_curcop = &cop; |
1502 | 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); |
1503 | LEAVE; |
1506 | LEAVE; |
1504 | |
1507 | |
1505 | croak ("%s, at character offset %d [\"%s\"]", |
1508 | croak ("%s, at character offset %d (before \"%s\")", |
1506 | dec.err, |
1509 | dec.err, |
1507 | (int)offset, |
1510 | ptr_to_index (string, dec.cur), |
1508 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1511 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1509 | } |
1512 | } |
1510 | |
1513 | |
1511 | sv = sv_2mortal (sv); |
1514 | sv = sv_2mortal (sv); |
1512 | |
1515 | |
… | |
… | |
1781 | XPUSHs (decode_json (jsonstr, self, 0)); |
1784 | XPUSHs (decode_json (jsonstr, self, 0)); |
1782 | |
1785 | |
1783 | void decode_prefix (JSON *self, SV *jsonstr) |
1786 | void decode_prefix (JSON *self, SV *jsonstr) |
1784 | PPCODE: |
1787 | PPCODE: |
1785 | { |
1788 | { |
1786 | STRLEN offset; |
1789 | char *offset; |
1787 | EXTEND (SP, 2); |
1790 | EXTEND (SP, 2); |
1788 | PUSHs (decode_json (jsonstr, self, &offset)); |
1791 | PUSHs (decode_json (jsonstr, self, &offset)); |
1789 | PUSHs (sv_2mortal (newSVuv (offset))); |
1792 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
1790 | } |
1793 | } |
1791 | |
1794 | |
1792 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1795 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1793 | PPCODE: |
1796 | PPCODE: |
1794 | { |
1797 | { |
… | |
… | |
1796 | self->incr_text = newSVpvn ("", 0); |
1799 | self->incr_text = newSVpvn ("", 0); |
1797 | |
1800 | |
1798 | // append data, if any |
1801 | // append data, if any |
1799 | if (jsonstr) |
1802 | if (jsonstr) |
1800 | { |
1803 | { |
1801 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
1804 | if (SvUTF8 (jsonstr)) |
1802 | { |
1805 | { |
|
|
1806 | if (!SvUTF8 (self->incr_text)) |
|
|
1807 | { |
1803 | /* utf-8-ness differs, need to upgrade */ |
1808 | /* utf-8-ness differs, need to upgrade */ |
1804 | sv_utf8_upgrade (self->incr_text); |
1809 | sv_utf8_upgrade (self->incr_text); |
1805 | |
1810 | |
1806 | if (self->incr_pos) |
1811 | if (self->incr_pos) |
1807 | 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) |
1808 | - (U8 *)SvPVX (self->incr_text); |
1813 | - (U8 *)SvPVX (self->incr_text); |
|
|
1814 | } |
1809 | } |
1815 | } |
|
|
1816 | else if (SvUTF8 (self->incr_text)) |
|
|
1817 | sv_utf8_upgrade (jsonstr); |
1810 | |
1818 | |
1811 | { |
1819 | { |
1812 | STRLEN len; |
1820 | STRLEN len; |
1813 | const char *str = SvPV (jsonstr, len); |
1821 | const char *str = SvPV (jsonstr, len); |
1814 | STRLEN cur = SvCUR (self->incr_text); |
1822 | STRLEN cur = SvCUR (self->incr_text); |
… | |
… | |
1823 | } |
1831 | } |
1824 | |
1832 | |
1825 | if (GIMME_V != G_VOID) |
1833 | if (GIMME_V != G_VOID) |
1826 | do |
1834 | do |
1827 | { |
1835 | { |
1828 | STRLEN offset; |
1836 | char *offset; |
1829 | |
1837 | |
1830 | if (!INCR_DONE (self)) |
1838 | if (!INCR_DONE (self)) |
1831 | { |
1839 | { |
1832 | incr_parse (self); |
1840 | incr_parse (self); |
1833 | |
1841 | |
… | |
… | |
1839 | break; |
1847 | break; |
1840 | } |
1848 | } |
1841 | |
1849 | |
1842 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1850 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1843 | |
1851 | |
1844 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1845 | self->incr_pos -= offset; |
1852 | self->incr_pos -= offset - SvPVX (self->incr_text); |
1846 | self->incr_nest = 0; |
1853 | self->incr_nest = 0; |
1847 | self->incr_mode = 0; |
1854 | self->incr_mode = 0; |
|
|
1855 | |
|
|
1856 | sv_chop (self->incr_text, offset); |
1848 | } |
1857 | } |
1849 | while (GIMME_V == G_ARRAY); |
1858 | while (GIMME_V == G_ARRAY); |
1850 | } |
1859 | } |
1851 | |
1860 | |
1852 | SV *incr_text (JSON *self) |
1861 | SV *incr_text (JSON *self) |