… | |
… | |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
14 | #endif |
14 | #endif |
15 | |
15 | |
16 | // some old perls do not have this, try to make it work, no |
16 | // some old perls do not have this, try to make it work, no |
17 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | // guarantees, though. if it breaks, you get to keep the pieces. |
18 | #ifndef UTF8_MAXBYTES |
18 | #ifndef UTF8_MAXBYTES |
19 | # define UTF8_MAXBYTES 13 |
19 | # define UTF8_MAXBYTES 13 |
20 | #endif |
20 | #endif |
21 | |
21 | |
|
|
22 | // three extra for rounding, sign, and end of string |
22 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
23 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
23 | |
24 | |
24 | #define F_ASCII 0x00000001UL |
25 | #define F_ASCII 0x00000001UL |
25 | #define F_LATIN1 0x00000002UL |
26 | #define F_LATIN1 0x00000002UL |
26 | #define F_UTF8 0x00000004UL |
27 | #define F_UTF8 0x00000004UL |
27 | #define F_INDENT 0x00000008UL |
28 | #define F_INDENT 0x00000008UL |
… | |
… | |
175 | *s++ = 0x80 | ( ch & 0x3f); |
176 | *s++ = 0x80 | ( ch & 0x3f); |
176 | |
177 | |
177 | return s; |
178 | return s; |
178 | } |
179 | } |
179 | |
180 | |
|
|
181 | // convert offset pointer to character index, sv must be string |
|
|
182 | static STRLEN |
|
|
183 | ptr_to_index (SV *sv, char *offset) |
|
|
184 | { |
|
|
185 | return SvUTF8 (sv) |
|
|
186 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
187 | : offset - SvPVX (sv); |
|
|
188 | } |
|
|
189 | |
180 | ///////////////////////////////////////////////////////////////////////////// |
190 | ///////////////////////////////////////////////////////////////////////////// |
181 | // encoder |
191 | // encoder |
182 | |
192 | |
183 | // structure used for encoding JSON |
193 | // structure used for encoding JSON |
184 | typedef struct |
194 | typedef struct |
… | |
… | |
195 | need (enc_t *enc, STRLEN len) |
205 | need (enc_t *enc, STRLEN len) |
196 | { |
206 | { |
197 | if (expect_false (enc->cur + len >= enc->end)) |
207 | if (expect_false (enc->cur + len >= enc->end)) |
198 | { |
208 | { |
199 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
209 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
200 | SvGROW (enc->sv, cur + len + 1); |
210 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
201 | enc->cur = SvPVX (enc->sv) + cur; |
211 | enc->cur = SvPVX (enc->sv) + cur; |
202 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
212 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
203 | } |
213 | } |
204 | } |
214 | } |
205 | |
215 | |
… | |
… | |
462 | |
472 | |
463 | // for canonical output we have to sort by keys first |
473 | // for canonical output we have to sort by keys first |
464 | // actually, this is mostly due to the stupid so-called |
474 | // actually, this is mostly due to the stupid so-called |
465 | // security workaround added somewhere in 5.8.x |
475 | // security workaround added somewhere in 5.8.x |
466 | // that randomises hash orderings |
476 | // that randomises hash orderings |
467 | if (enc->json.flags & F_CANONICAL) |
477 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
468 | { |
478 | { |
469 | int count = hv_iterinit (hv); |
479 | int count = hv_iterinit (hv); |
470 | |
480 | |
471 | if (SvMAGICAL (hv)) |
481 | if (SvMAGICAL (hv)) |
472 | { |
482 | { |
… | |
… | |
750 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
760 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
751 | : 0x110000UL; |
761 | : 0x110000UL; |
752 | |
762 | |
753 | SvPOK_only (enc.sv); |
763 | SvPOK_only (enc.sv); |
754 | encode_sv (&enc, scalar); |
764 | encode_sv (&enc, scalar); |
|
|
765 | encode_nl (&enc); |
755 | |
766 | |
756 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
767 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
757 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
768 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
758 | |
769 | |
759 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
770 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
… | |
… | |
939 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
950 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
940 | *cur++ = ch; |
951 | *cur++ = ch; |
941 | else if (ch >= 0x80) |
952 | else if (ch >= 0x80) |
942 | { |
953 | { |
943 | STRLEN clen; |
954 | STRLEN clen; |
944 | UV uch; |
|
|
945 | |
955 | |
946 | --dec_cur; |
956 | --dec_cur; |
947 | |
957 | |
948 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
958 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
949 | if (clen == (STRLEN)-1) |
959 | if (clen == (STRLEN)-1) |
950 | ERR ("malformed UTF-8 character in JSON string"); |
960 | ERR ("malformed UTF-8 character in JSON string"); |
951 | |
961 | |
952 | do |
962 | do |
953 | *cur++ = *dec_cur++; |
963 | *cur++ = *dec_cur++; |
… | |
… | |
970 | { |
980 | { |
971 | STRLEN len = cur - buf; |
981 | STRLEN len = cur - buf; |
972 | |
982 | |
973 | if (sv) |
983 | if (sv) |
974 | { |
984 | { |
975 | SvGROW (sv, SvCUR (sv) + len + 1); |
985 | STRLEN cur = SvCUR (sv); |
|
|
986 | |
|
|
987 | if (SvLEN (sv) <= cur + len) |
|
|
988 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
989 | |
976 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
990 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
977 | SvCUR_set (sv, SvCUR (sv) + len); |
991 | SvCUR_set (sv, SvCUR (sv) + len); |
978 | } |
992 | } |
979 | else |
993 | else |
980 | sv = newSVpvn (buf, len); |
994 | sv = newSVpvn (buf, len); |
… | |
… | |
1071 | |
1085 | |
1072 | // special case the rather common 1..5-digit-int case |
1086 | // special case the rather common 1..5-digit-int case |
1073 | if (*start == '-') |
1087 | if (*start == '-') |
1074 | switch (len) |
1088 | switch (len) |
1075 | { |
1089 | { |
1076 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1090 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
1077 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1091 | 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)); |
1092 | 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)); |
1093 | 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)); |
1094 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1081 | } |
1095 | } |
1082 | else |
1096 | else |
1083 | switch (len) |
1097 | switch (len) |
1084 | { |
1098 | { |
1085 | case 1: return newSViv ( start [0] - '0' * 1); |
1099 | case 1: return newSViv ( start [0] - '0' * 1); |
1086 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1100 | 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); |
1101 | 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); |
1102 | 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); |
1103 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1090 | } |
1104 | } |
1091 | |
1105 | |
1092 | { |
1106 | { |
1093 | UV uv; |
1107 | UV uv; |
1094 | int numtype = grok_number (start, len, &uv); |
1108 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1404 | fail: |
1418 | fail: |
1405 | return 0; |
1419 | return 0; |
1406 | } |
1420 | } |
1407 | |
1421 | |
1408 | static SV * |
1422 | static SV * |
1409 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1423 | decode_json (SV *string, JSON *json, char **offset_return) |
1410 | { |
1424 | { |
1411 | dec_t dec; |
1425 | dec_t dec; |
1412 | STRLEN offset; |
|
|
1413 | SV *sv; |
1426 | SV *sv; |
1414 | |
1427 | |
|
|
1428 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1429 | * will perl ignore the magic in subsequent accesses |
|
|
1430 | */ |
1415 | SvGETMAGIC (string); |
1431 | /*SvGETMAGIC (string);*/ |
|
|
1432 | if (SvMAGICAL (string)) |
|
|
1433 | string = sv_2mortal (newSVsv (string)); |
|
|
1434 | |
1416 | SvUPGRADE (string, SVt_PV); |
1435 | SvUPGRADE (string, SVt_PV); |
1417 | |
1436 | |
1418 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
1437 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
1419 | * assertion with -DDEBUGGING, although SvCUR is documented to |
1438 | * assertion with -DDEBUGGING, although SvCUR is documented to |
1420 | * return the xpv_cur field which certainly exists after upgrading. |
1439 | * return the xpv_cur field which certainly exists after upgrading. |
… | |
… | |
1423 | * and hope for the best. |
1442 | * and hope for the best. |
1424 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
1443 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
1425 | * assertion business is seriously broken, try yet another workaround |
1444 | * assertion business is seriously broken, try yet another workaround |
1426 | * for the broken -DDEBUGGING. |
1445 | * for the broken -DDEBUGGING. |
1427 | */ |
1446 | */ |
|
|
1447 | { |
1428 | #ifdef DEBUGGING |
1448 | #ifdef DEBUGGING |
1429 | offset = SvOK (string) ? sv_len (string) : 0; |
1449 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
1430 | #else |
1450 | #else |
1431 | offset = SvCUR (string); |
1451 | STRLEN offset = SvCUR (string); |
1432 | #endif |
1452 | #endif |
1433 | |
1453 | |
1434 | if (offset > json->max_size && json->max_size) |
1454 | if (offset > json->max_size && json->max_size) |
1435 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1455 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1436 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1456 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1457 | } |
1437 | |
1458 | |
1438 | if (json->flags & F_UTF8) |
1459 | if (json->flags & F_UTF8) |
1439 | sv_utf8_downgrade (string, 0); |
1460 | sv_utf8_downgrade (string, 0); |
1440 | else |
1461 | else |
1441 | sv_utf8_upgrade (string); |
1462 | sv_utf8_upgrade (string); |
… | |
… | |
1453 | |
1474 | |
1454 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1475 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1455 | |
1476 | |
1456 | decode_ws (&dec); |
1477 | decode_ws (&dec); |
1457 | sv = decode_sv (&dec); |
1478 | sv = decode_sv (&dec); |
|
|
1479 | |
|
|
1480 | if (offset_return) |
|
|
1481 | *offset_return = dec.cur; |
1458 | |
1482 | |
1459 | if (!(offset_return || !sv)) |
1483 | if (!(offset_return || !sv)) |
1460 | { |
1484 | { |
1461 | // check for trailing garbage |
1485 | // check for trailing garbage |
1462 | decode_ws (&dec); |
1486 | decode_ws (&dec); |
… | |
… | |
1465 | { |
1489 | { |
1466 | dec.err = "garbage after JSON object"; |
1490 | dec.err = "garbage after JSON object"; |
1467 | SvREFCNT_dec (sv); |
1491 | SvREFCNT_dec (sv); |
1468 | sv = 0; |
1492 | sv = 0; |
1469 | } |
1493 | } |
1470 | } |
|
|
1471 | |
|
|
1472 | if (offset_return || !sv) |
|
|
1473 | { |
|
|
1474 | offset = dec.json.flags & F_UTF8 |
|
|
1475 | ? dec.cur - SvPVX (string) |
|
|
1476 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1477 | |
|
|
1478 | if (offset_return) |
|
|
1479 | *offset_return = offset; |
|
|
1480 | } |
1494 | } |
1481 | |
1495 | |
1482 | if (!sv) |
1496 | if (!sv) |
1483 | { |
1497 | { |
1484 | SV *uni = sv_newmortal (); |
1498 | SV *uni = sv_newmortal (); |
… | |
… | |
1490 | SAVEVPTR (PL_curcop); |
1504 | SAVEVPTR (PL_curcop); |
1491 | PL_curcop = &cop; |
1505 | PL_curcop = &cop; |
1492 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1506 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1493 | LEAVE; |
1507 | LEAVE; |
1494 | |
1508 | |
1495 | croak ("%s, at character offset %d [\"%s\"]", |
1509 | croak ("%s, at character offset %d (before \"%s\")", |
1496 | dec.err, |
1510 | dec.err, |
1497 | (int)offset, |
1511 | ptr_to_index (string, dec.cur), |
1498 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1512 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1499 | } |
1513 | } |
1500 | |
1514 | |
1501 | sv = sv_2mortal (sv); |
1515 | sv = sv_2mortal (sv); |
1502 | |
1516 | |
… | |
… | |
1642 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1656 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1643 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1657 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1644 | |
1658 | |
1645 | json_true = get_bool ("JSON::XS::true"); |
1659 | json_true = get_bool ("JSON::XS::true"); |
1646 | json_false = get_bool ("JSON::XS::false"); |
1660 | json_false = get_bool ("JSON::XS::false"); |
|
|
1661 | |
|
|
1662 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1647 | } |
1663 | } |
1648 | |
1664 | |
1649 | PROTOTYPES: DISABLE |
1665 | PROTOTYPES: DISABLE |
1650 | |
1666 | |
1651 | void CLONE (...) |
1667 | void CLONE (...) |
… | |
… | |
1771 | XPUSHs (decode_json (jsonstr, self, 0)); |
1787 | XPUSHs (decode_json (jsonstr, self, 0)); |
1772 | |
1788 | |
1773 | void decode_prefix (JSON *self, SV *jsonstr) |
1789 | void decode_prefix (JSON *self, SV *jsonstr) |
1774 | PPCODE: |
1790 | PPCODE: |
1775 | { |
1791 | { |
1776 | STRLEN offset; |
1792 | char *offset; |
1777 | EXTEND (SP, 2); |
1793 | EXTEND (SP, 2); |
1778 | PUSHs (decode_json (jsonstr, self, &offset)); |
1794 | PUSHs (decode_json (jsonstr, self, &offset)); |
1779 | PUSHs (sv_2mortal (newSVuv (offset))); |
1795 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
1780 | } |
1796 | } |
1781 | |
1797 | |
1782 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1798 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1783 | PPCODE: |
1799 | PPCODE: |
1784 | { |
1800 | { |
… | |
… | |
1786 | self->incr_text = newSVpvn ("", 0); |
1802 | self->incr_text = newSVpvn ("", 0); |
1787 | |
1803 | |
1788 | // append data, if any |
1804 | // append data, if any |
1789 | if (jsonstr) |
1805 | if (jsonstr) |
1790 | { |
1806 | { |
1791 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
1807 | if (SvUTF8 (jsonstr)) |
1792 | { |
1808 | { |
|
|
1809 | if (!SvUTF8 (self->incr_text)) |
|
|
1810 | { |
1793 | /* utf-8-ness differs, need to upgrade */ |
1811 | /* utf-8-ness differs, need to upgrade */ |
1794 | sv_utf8_upgrade (self->incr_text); |
1812 | sv_utf8_upgrade (self->incr_text); |
1795 | |
1813 | |
1796 | if (self->incr_pos) |
1814 | if (self->incr_pos) |
1797 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
1815 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
1798 | - (U8 *)SvPVX (self->incr_text); |
1816 | - (U8 *)SvPVX (self->incr_text); |
|
|
1817 | } |
1799 | } |
1818 | } |
|
|
1819 | else if (SvUTF8 (self->incr_text)) |
|
|
1820 | sv_utf8_upgrade (jsonstr); |
1800 | |
1821 | |
1801 | { |
1822 | { |
1802 | STRLEN len; |
1823 | STRLEN len; |
1803 | const char *str = SvPV (jsonstr, len); |
1824 | const char *str = SvPV (jsonstr, len); |
1804 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
1825 | STRLEN cur = SvCUR (self->incr_text); |
|
|
1826 | |
|
|
1827 | if (SvLEN (self->incr_text) <= cur + len) |
|
|
1828 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
1829 | |
1805 | Move (str, SvEND (self->incr_text), len, char); |
1830 | Move (str, SvEND (self->incr_text), len, char); |
1806 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1831 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1807 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
1832 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
1808 | } |
1833 | } |
1809 | } |
1834 | } |
1810 | |
1835 | |
1811 | if (GIMME_V != G_VOID) |
1836 | if (GIMME_V != G_VOID) |
1812 | do |
1837 | do |
1813 | { |
1838 | { |
1814 | STRLEN offset; |
1839 | char *offset; |
1815 | |
1840 | |
1816 | if (!INCR_DONE (self)) |
1841 | if (!INCR_DONE (self)) |
1817 | { |
1842 | { |
1818 | incr_parse (self); |
1843 | incr_parse (self); |
1819 | |
1844 | |
… | |
… | |
1825 | break; |
1850 | break; |
1826 | } |
1851 | } |
1827 | |
1852 | |
1828 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1853 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1829 | |
1854 | |
1830 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1831 | self->incr_pos -= offset; |
1855 | self->incr_pos -= offset - SvPVX (self->incr_text); |
1832 | self->incr_nest = 0; |
1856 | self->incr_nest = 0; |
1833 | self->incr_mode = 0; |
1857 | self->incr_mode = 0; |
|
|
1858 | |
|
|
1859 | sv_chop (self->incr_text, offset); |
1834 | } |
1860 | } |
1835 | while (GIMME_V == G_ARRAY); |
1861 | while (GIMME_V == G_ARRAY); |
1836 | } |
1862 | } |
1837 | |
1863 | |
1838 | SV *incr_text (JSON *self) |
1864 | SV *incr_text (JSON *self) |