… | |
… | |
78 | #endif |
78 | #endif |
79 | |
79 | |
80 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
80 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
81 | static SV *json_true, *json_false; |
81 | static SV *json_true, *json_false; |
82 | |
82 | |
|
|
83 | enum { |
|
|
84 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
85 | INCR_M_STR, // inside string |
|
|
86 | INCR_M_BS, // inside backslash |
|
|
87 | INCR_M_JSON // outside anything, count nesting |
|
|
88 | }; |
|
|
89 | |
|
|
90 | #define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) |
|
|
91 | |
83 | typedef struct { |
92 | typedef struct { |
84 | U32 flags; |
93 | U32 flags; |
85 | SV *cb_object; |
94 | SV *cb_object; |
86 | HV *cb_sk_object; |
95 | HV *cb_sk_object; |
|
|
96 | |
|
|
97 | // for the incremental parser |
|
|
98 | SV *incr_text; // the source text so far |
|
|
99 | STRLEN incr_pos; // the current offset into the text |
|
|
100 | int incr_nest; // {[]}-nesting level |
|
|
101 | int incr_mode; |
87 | } JSON; |
102 | } JSON; |
88 | |
103 | |
89 | ///////////////////////////////////////////////////////////////////////////// |
104 | ///////////////////////////////////////////////////////////////////////////// |
90 | // utility functions |
105 | // utility functions |
|
|
106 | |
|
|
107 | INLINE SV * |
|
|
108 | get_bool (const char *name) |
|
|
109 | { |
|
|
110 | SV *sv = get_sv (name, 1); |
|
|
111 | |
|
|
112 | SvREADONLY_on (sv); |
|
|
113 | SvREADONLY_on (SvRV (sv)); |
|
|
114 | |
|
|
115 | return sv; |
|
|
116 | } |
91 | |
117 | |
92 | INLINE void |
118 | INLINE void |
93 | shrink (SV *sv) |
119 | shrink (SV *sv) |
94 | { |
120 | { |
95 | sv_utf8_downgrade (sv, 1); |
121 | sv_utf8_downgrade (sv, 1); |
… | |
… | |
423 | |
449 | |
424 | static void |
450 | static void |
425 | encode_hv (enc_t *enc, HV *hv) |
451 | encode_hv (enc_t *enc, HV *hv) |
426 | { |
452 | { |
427 | HE *he; |
453 | HE *he; |
428 | int count; |
|
|
429 | |
454 | |
430 | if (enc->indent >= enc->maxdepth) |
455 | if (enc->indent >= enc->maxdepth) |
431 | croak ("data structure too deep (hit recursion limit)"); |
456 | croak ("data structure too deep (hit recursion limit)"); |
432 | |
457 | |
433 | encode_ch (enc, '{'); |
458 | encode_ch (enc, '{'); |
… | |
… | |
1171 | char *p = dec->cur; |
1196 | char *p = dec->cur; |
1172 | char *e = p + 24; // only try up to 24 bytes |
1197 | char *e = p + 24; // only try up to 24 bytes |
1173 | |
1198 | |
1174 | for (;;) |
1199 | for (;;) |
1175 | { |
1200 | { |
1176 | // the >= 0x80 is true on most architectures |
1201 | // the >= 0x80 is false on most architectures |
1177 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1202 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1178 | { |
1203 | { |
1179 | // slow path, back up and use decode_str |
1204 | // slow path, back up and use decode_str |
1180 | SV *key = decode_str (dec); |
1205 | SV *key = decode_str (dec); |
1181 | if (!key) |
1206 | if (!key) |
… | |
… | |
1329 | case 't': |
1354 | case 't': |
1330 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1355 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1331 | { |
1356 | { |
1332 | dec->cur += 4; |
1357 | dec->cur += 4; |
1333 | #if JSON_SLOW |
1358 | #if JSON_SLOW |
1334 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1359 | json_true = get_bool ("JSON::XS::true"); |
1335 | #endif |
1360 | #endif |
1336 | return SvREFCNT_inc (json_true); |
1361 | return newSVsv (json_true); |
1337 | } |
1362 | } |
1338 | else |
1363 | else |
1339 | ERR ("'true' expected"); |
1364 | ERR ("'true' expected"); |
1340 | |
1365 | |
1341 | break; |
1366 | break; |
… | |
… | |
1343 | case 'f': |
1368 | case 'f': |
1344 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1369 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1345 | { |
1370 | { |
1346 | dec->cur += 5; |
1371 | dec->cur += 5; |
1347 | #if JSON_SLOW |
1372 | #if JSON_SLOW |
1348 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1373 | json_false = get_bool ("JSON::XS::false"); |
1349 | #endif |
1374 | #endif |
1350 | return SvREFCNT_inc (json_false); |
1375 | return newSVsv (json_false); |
1351 | } |
1376 | } |
1352 | else |
1377 | else |
1353 | ERR ("'false' expected"); |
1378 | ERR ("'false' expected"); |
1354 | |
1379 | |
1355 | break; |
1380 | break; |
… | |
… | |
1373 | fail: |
1398 | fail: |
1374 | return 0; |
1399 | return 0; |
1375 | } |
1400 | } |
1376 | |
1401 | |
1377 | static SV * |
1402 | static SV * |
1378 | decode_json (SV *string, JSON *json, UV *offset_return) |
1403 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1379 | { |
1404 | { |
1380 | dec_t dec; |
1405 | dec_t dec; |
1381 | UV offset; |
1406 | STRLEN offset; |
1382 | SV *sv; |
1407 | SV *sv; |
1383 | |
1408 | |
1384 | SvGETMAGIC (string); |
1409 | SvGETMAGIC (string); |
1385 | SvUPGRADE (string, SVt_PV); |
1410 | SvUPGRADE (string, SVt_PV); |
1386 | |
1411 | |
… | |
… | |
1459 | |
1484 | |
1460 | return sv; |
1485 | return sv; |
1461 | } |
1486 | } |
1462 | |
1487 | |
1463 | ///////////////////////////////////////////////////////////////////////////// |
1488 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1489 | // incremental parser |
|
|
1490 | |
|
|
1491 | static void |
|
|
1492 | incr_parse (JSON *self) |
|
|
1493 | { |
|
|
1494 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1495 | |
|
|
1496 | for (;;) |
|
|
1497 | { |
|
|
1498 | //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D |
|
|
1499 | switch (self->incr_mode) |
|
|
1500 | { |
|
|
1501 | // only used for intiial whitespace skipping |
|
|
1502 | case INCR_M_WS: |
|
|
1503 | for (;;) |
|
|
1504 | { |
|
|
1505 | if (*p > 0x20) |
|
|
1506 | { |
|
|
1507 | self->incr_mode = INCR_M_JSON; |
|
|
1508 | goto incr_m_json; |
|
|
1509 | } |
|
|
1510 | else if (!*p) |
|
|
1511 | goto interrupt; |
|
|
1512 | |
|
|
1513 | ++p; |
|
|
1514 | } |
|
|
1515 | |
|
|
1516 | // skip a single char inside a string (for \\-processing) |
|
|
1517 | case INCR_M_BS: |
|
|
1518 | if (!*p) |
|
|
1519 | goto interrupt; |
|
|
1520 | |
|
|
1521 | ++p; |
|
|
1522 | self->incr_mode = INCR_M_STR; |
|
|
1523 | goto incr_m_str; |
|
|
1524 | |
|
|
1525 | // inside a string |
|
|
1526 | case INCR_M_STR: |
|
|
1527 | incr_m_str: |
|
|
1528 | for (;;) |
|
|
1529 | { |
|
|
1530 | if (*p == '"') |
|
|
1531 | { |
|
|
1532 | ++p; |
|
|
1533 | self->incr_mode = INCR_M_JSON; |
|
|
1534 | |
|
|
1535 | if (!self->incr_nest) |
|
|
1536 | goto interrupt; |
|
|
1537 | |
|
|
1538 | goto incr_m_json; |
|
|
1539 | } |
|
|
1540 | else if (*p == '\\') |
|
|
1541 | { |
|
|
1542 | ++p; // "virtually" consumes character after \ |
|
|
1543 | |
|
|
1544 | if (!*p) // if at end of string we have to switch modes |
|
|
1545 | { |
|
|
1546 | self->incr_mode = INCR_M_BS; |
|
|
1547 | goto interrupt; |
|
|
1548 | } |
|
|
1549 | } |
|
|
1550 | else if (!*p) |
|
|
1551 | goto interrupt; |
|
|
1552 | |
|
|
1553 | ++p; |
|
|
1554 | } |
|
|
1555 | |
|
|
1556 | // after initial ws, outside string |
|
|
1557 | case INCR_M_JSON: |
|
|
1558 | incr_m_json: |
|
|
1559 | for (;;) |
|
|
1560 | { |
|
|
1561 | switch (*p++) |
|
|
1562 | { |
|
|
1563 | case 0: |
|
|
1564 | --p; |
|
|
1565 | goto interrupt; |
|
|
1566 | |
|
|
1567 | case 0x09: |
|
|
1568 | case 0x0a: |
|
|
1569 | case 0x0d: |
|
|
1570 | case 0x20: |
|
|
1571 | if (!self->incr_nest) |
|
|
1572 | { |
|
|
1573 | --p; // do not eat the whitespace, let the next round do it |
|
|
1574 | goto interrupt; |
|
|
1575 | } |
|
|
1576 | break; |
|
|
1577 | |
|
|
1578 | case '"': |
|
|
1579 | self->incr_mode = INCR_M_STR; |
|
|
1580 | goto incr_m_str; |
|
|
1581 | |
|
|
1582 | case '[': |
|
|
1583 | case '{': |
|
|
1584 | ++self->incr_nest; |
|
|
1585 | break; |
|
|
1586 | |
|
|
1587 | case ']': |
|
|
1588 | case '}': |
|
|
1589 | if (!--self->incr_nest) |
|
|
1590 | goto interrupt; |
|
|
1591 | } |
|
|
1592 | } |
|
|
1593 | } |
|
|
1594 | |
|
|
1595 | modechange: |
|
|
1596 | ; |
|
|
1597 | } |
|
|
1598 | |
|
|
1599 | interrupt: |
|
|
1600 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1601 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
|
|
1602 | } |
|
|
1603 | |
|
|
1604 | ///////////////////////////////////////////////////////////////////////////// |
1464 | // XS interface functions |
1605 | // XS interface functions |
1465 | |
1606 | |
1466 | MODULE = JSON::XS PACKAGE = JSON::XS |
1607 | MODULE = JSON::XS PACKAGE = JSON::XS |
1467 | |
1608 | |
1468 | BOOT: |
1609 | BOOT: |
… | |
… | |
1477 | : -1; |
1618 | : -1; |
1478 | |
1619 | |
1479 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1620 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1480 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1621 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1481 | |
1622 | |
1482 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
1623 | json_true = get_bool ("JSON::XS::true"); |
1483 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1624 | json_false = get_bool ("JSON::XS::false"); |
1484 | } |
1625 | } |
1485 | |
1626 | |
1486 | PROTOTYPES: DISABLE |
1627 | PROTOTYPES: DISABLE |
1487 | |
1628 | |
1488 | void CLONE (...) |
1629 | void CLONE (...) |
… | |
… | |
1628 | XPUSHs (decode_json (jsonstr, self, 0)); |
1769 | XPUSHs (decode_json (jsonstr, self, 0)); |
1629 | |
1770 | |
1630 | void decode_prefix (JSON *self, SV *jsonstr) |
1771 | void decode_prefix (JSON *self, SV *jsonstr) |
1631 | PPCODE: |
1772 | PPCODE: |
1632 | { |
1773 | { |
1633 | UV offset; |
1774 | STRLEN offset; |
1634 | EXTEND (SP, 2); |
1775 | EXTEND (SP, 2); |
1635 | PUSHs (decode_json (jsonstr, self, &offset)); |
1776 | PUSHs (decode_json (jsonstr, self, &offset)); |
1636 | PUSHs (sv_2mortal (newSVuv (offset))); |
1777 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1778 | } |
|
|
1779 | |
|
|
1780 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
1781 | PPCODE: |
|
|
1782 | { |
|
|
1783 | if (!self->incr_text) |
|
|
1784 | self->incr_text = newSVpvn ("", 0); |
|
|
1785 | |
|
|
1786 | // append data, if any |
|
|
1787 | if (jsonstr) |
|
|
1788 | { |
|
|
1789 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
|
|
1790 | { |
|
|
1791 | /* utf-8-ness differs, need to upgrade */ |
|
|
1792 | sv_utf8_upgrade (self->incr_text); |
|
|
1793 | |
|
|
1794 | if (self->incr_pos) |
|
|
1795 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1796 | - (U8 *)SvPVX (self->incr_text); |
|
|
1797 | } |
|
|
1798 | |
|
|
1799 | { |
|
|
1800 | STRLEN len; |
|
|
1801 | const char *str = SvPV (jsonstr, len); |
|
|
1802 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
|
|
1803 | Move (str, SvEND (self->incr_text), len, char); |
|
|
1804 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
1805 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1806 | } |
|
|
1807 | } |
|
|
1808 | |
|
|
1809 | if (GIMME_V != G_VOID) |
|
|
1810 | do |
|
|
1811 | { |
|
|
1812 | STRLEN offset; |
|
|
1813 | |
|
|
1814 | if (!INCR_DONE (self)) |
|
|
1815 | { |
|
|
1816 | incr_parse (self); |
|
|
1817 | if (!INCR_DONE (self)) |
|
|
1818 | break; |
|
|
1819 | } |
|
|
1820 | |
|
|
1821 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
|
|
1822 | |
|
|
1823 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1824 | self->incr_pos -= offset; |
|
|
1825 | self->incr_nest = 0; |
|
|
1826 | self->incr_mode = 0; |
|
|
1827 | } |
|
|
1828 | while (GIMME_V == G_ARRAY); |
|
|
1829 | } |
|
|
1830 | |
|
|
1831 | SV *incr_text (JSON *self) |
|
|
1832 | ATTRS: lvalue |
|
|
1833 | CODE: |
|
|
1834 | { |
|
|
1835 | if (self->incr_pos) |
|
|
1836 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
1837 | |
|
|
1838 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
1839 | } |
|
|
1840 | OUTPUT: |
|
|
1841 | RETVAL |
|
|
1842 | |
|
|
1843 | void incr_skip (JSON *self) |
|
|
1844 | CODE: |
|
|
1845 | { |
|
|
1846 | if (self->incr_pos) |
|
|
1847 | { |
|
|
1848 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
1849 | self->incr_pos = 0; |
|
|
1850 | self->incr_nest = 0; |
|
|
1851 | self->incr_mode = 0; |
|
|
1852 | } |
1637 | } |
1853 | } |
1638 | |
1854 | |
1639 | void DESTROY (JSON *self) |
1855 | void DESTROY (JSON *self) |
1640 | CODE: |
1856 | CODE: |
1641 | SvREFCNT_dec (self->cb_sk_object); |
1857 | SvREFCNT_dec (self->cb_sk_object); |
1642 | SvREFCNT_dec (self->cb_object); |
1858 | SvREFCNT_dec (self->cb_object); |
|
|
1859 | SvREFCNT_dec (self->incr_text); |
1643 | |
1860 | |
1644 | PROTOTYPES: ENABLE |
1861 | PROTOTYPES: ENABLE |
1645 | |
1862 | |
1646 | void encode_json (SV *scalar) |
1863 | void encode_json (SV *scalar) |
1647 | ALIAS: |
1864 | ALIAS: |
… | |
… | |
1661 | { |
1878 | { |
1662 | JSON json = { F_DEFAULT | ix }; |
1879 | JSON json = { F_DEFAULT | ix }; |
1663 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1880 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1664 | } |
1881 | } |
1665 | |
1882 | |
|
|
1883 | |