… | |
… | |
32 | #define F_SHRINK 0x00000200UL |
32 | #define F_SHRINK 0x00000200UL |
33 | #define F_ALLOW_BLESSED 0x00000400UL |
33 | #define F_ALLOW_BLESSED 0x00000400UL |
34 | #define F_CONV_BLESSED 0x00000800UL |
34 | #define F_CONV_BLESSED 0x00000800UL |
35 | #define F_RELAXED 0x00001000UL |
35 | #define F_RELAXED 0x00001000UL |
36 | #define F_ALLOW_UNKNOWN 0x00002000UL |
36 | #define F_ALLOW_UNKNOWN 0x00002000UL |
37 | |
|
|
38 | #define F_MAXDEPTH 0xf8000000UL |
|
|
39 | #define S_MAXDEPTH 27 |
|
|
40 | #define F_MAXSIZE 0x01f00000UL |
|
|
41 | #define S_MAXSIZE 20 |
|
|
42 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
37 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
43 | |
38 | |
44 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
45 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
|
|
46 | |
|
|
47 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
39 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
48 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
|
|
49 | |
40 | |
50 | #define INIT_SIZE 32 // initial scalar size to be allocated |
41 | #define INIT_SIZE 32 // initial scalar size to be allocated |
51 | #define INDENT_STEP 3 // spaces per indentation level |
42 | #define INDENT_STEP 3 // spaces per indentation level |
52 | |
43 | |
53 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
44 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
… | |
… | |
68 | |
59 | |
69 | #define IN_RANGE_INC(type,val,beg,end) \ |
60 | #define IN_RANGE_INC(type,val,beg,end) \ |
70 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
61 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
71 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
62 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
72 | |
63 | |
|
|
64 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
|
|
65 | |
73 | #ifdef USE_ITHREADS |
66 | #ifdef USE_ITHREADS |
74 | # define JSON_SLOW 1 |
67 | # define JSON_SLOW 1 |
75 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
68 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
76 | #else |
69 | #else |
77 | # define JSON_SLOW 0 |
70 | # define JSON_SLOW 0 |
… | |
… | |
90 | |
83 | |
91 | #define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) |
84 | #define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) |
92 | |
85 | |
93 | typedef struct { |
86 | typedef struct { |
94 | U32 flags; |
87 | U32 flags; |
|
|
88 | U32 max_depth; |
|
|
89 | STRLEN max_size; |
|
|
90 | |
95 | SV *cb_object; |
91 | SV *cb_object; |
96 | HV *cb_sk_object; |
92 | HV *cb_sk_object; |
97 | |
93 | |
98 | // for the incremental parser |
94 | // for the incremental parser |
99 | SV *incr_text; // the source text so far |
95 | SV *incr_text; // the source text so far |
100 | STRLEN incr_pos; // the current offset into the text |
96 | STRLEN incr_pos; // the current offset into the text |
101 | int incr_nest; // {[]}-nesting level |
97 | unsigned char incr_nest; // {[]}-nesting level |
102 | int incr_mode; |
98 | unsigned char incr_mode; |
103 | } JSON; |
99 | } JSON; |
|
|
100 | |
|
|
101 | INLINE void |
|
|
102 | json_init (JSON *json) |
|
|
103 | { |
|
|
104 | Zero (json, 1, JSON); |
|
|
105 | json->max_depth = 512; |
|
|
106 | } |
104 | |
107 | |
105 | ///////////////////////////////////////////////////////////////////////////// |
108 | ///////////////////////////////////////////////////////////////////////////// |
106 | // utility functions |
109 | // utility functions |
107 | |
110 | |
108 | INLINE SV * |
111 | INLINE SV * |
… | |
… | |
182 | char *cur; // SvPVX (sv) + current output position |
185 | char *cur; // SvPVX (sv) + current output position |
183 | char *end; // SvEND (sv) |
186 | char *end; // SvEND (sv) |
184 | SV *sv; // result scalar |
187 | SV *sv; // result scalar |
185 | JSON json; |
188 | JSON json; |
186 | U32 indent; // indentation level |
189 | U32 indent; // indentation level |
187 | U32 maxdepth; // max. indentation/recursion level |
|
|
188 | UV limit; // escape character values >= this value when encoding |
190 | UV limit; // escape character values >= this value when encoding |
189 | } enc_t; |
191 | } enc_t; |
190 | |
192 | |
191 | INLINE void |
193 | INLINE void |
192 | need (enc_t *enc, STRLEN len) |
194 | need (enc_t *enc, STRLEN len) |
… | |
… | |
365 | static void |
367 | static void |
366 | encode_av (enc_t *enc, AV *av) |
368 | encode_av (enc_t *enc, AV *av) |
367 | { |
369 | { |
368 | int i, len = av_len (av); |
370 | int i, len = av_len (av); |
369 | |
371 | |
370 | if (enc->indent >= enc->maxdepth) |
372 | if (enc->indent >= enc->json.max_depth) |
371 | croak ("data structure too deep (hit recursion limit)"); |
373 | croak (ERR_NESTING_EXCEEDED); |
372 | |
374 | |
373 | encode_ch (enc, '['); |
375 | encode_ch (enc, '['); |
374 | |
376 | |
375 | if (len >= 0) |
377 | if (len >= 0) |
376 | { |
378 | { |
… | |
… | |
451 | static void |
453 | static void |
452 | encode_hv (enc_t *enc, HV *hv) |
454 | encode_hv (enc_t *enc, HV *hv) |
453 | { |
455 | { |
454 | HE *he; |
456 | HE *he; |
455 | |
457 | |
456 | if (enc->indent >= enc->maxdepth) |
458 | if (enc->indent >= enc->json.max_depth) |
457 | croak ("data structure too deep (hit recursion limit)"); |
459 | croak (ERR_NESTING_EXCEEDED); |
458 | |
460 | |
459 | encode_ch (enc, '{'); |
461 | encode_ch (enc, '{'); |
460 | |
462 | |
461 | // for canonical output we have to sort by keys first |
463 | // for canonical output we have to sort by keys first |
462 | // actually, this is mostly due to the stupid so-called |
464 | // actually, this is mostly due to the stupid so-called |
… | |
… | |
742 | enc.json = *json; |
744 | enc.json = *json; |
743 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
745 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
744 | enc.cur = SvPVX (enc.sv); |
746 | enc.cur = SvPVX (enc.sv); |
745 | enc.end = SvEND (enc.sv); |
747 | enc.end = SvEND (enc.sv); |
746 | enc.indent = 0; |
748 | enc.indent = 0; |
747 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
|
|
748 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
749 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
749 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
750 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
750 | : 0x110000UL; |
751 | : 0x110000UL; |
751 | |
752 | |
752 | SvPOK_only (enc.sv); |
753 | SvPOK_only (enc.sv); |
… | |
… | |
819 | if (*dec->cur != ch) \ |
820 | if (*dec->cur != ch) \ |
820 | ERR (# ch " expected"); \ |
821 | ERR (# ch " expected"); \ |
821 | ++dec->cur; \ |
822 | ++dec->cur; \ |
822 | SE |
823 | SE |
823 | |
824 | |
824 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
825 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
825 | #define DEC_DEC_DEPTH --dec->depth |
826 | #define DEC_DEC_DEPTH --dec->depth |
826 | |
827 | |
827 | static SV *decode_sv (dec_t *dec); |
828 | static SV *decode_sv (dec_t *dec); |
828 | |
829 | |
829 | static signed char decode_hexdigit[256]; |
830 | static signed char decode_hexdigit[256]; |
… | |
… | |
1412 | SV *sv; |
1413 | SV *sv; |
1413 | |
1414 | |
1414 | SvGETMAGIC (string); |
1415 | SvGETMAGIC (string); |
1415 | SvUPGRADE (string, SVt_PV); |
1416 | SvUPGRADE (string, SVt_PV); |
1416 | |
1417 | |
1417 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
1418 | if (SvCUR (string) > json->max_size && json->max_size) |
1418 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1419 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1419 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
1420 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1420 | |
1421 | |
1421 | if (json->flags & F_UTF8) |
1422 | if (json->flags & F_UTF8) |
1422 | sv_utf8_downgrade (string, 0); |
1423 | sv_utf8_downgrade (string, 0); |
1423 | else |
1424 | else |
1424 | sv_utf8_upgrade (string); |
1425 | sv_utf8_upgrade (string); |
1425 | |
1426 | |
1426 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1427 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1427 | |
1428 | |
1428 | dec.json = *json; |
1429 | dec.json = *json; |
1429 | dec.cur = SvPVX (string); |
1430 | dec.cur = SvPVX (string); |
1430 | dec.end = SvEND (string); |
1431 | dec.end = SvEND (string); |
1431 | dec.err = 0; |
1432 | dec.err = 0; |
1432 | dec.depth = 0; |
1433 | dec.depth = 0; |
1433 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1434 | |
1434 | |
1435 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1435 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1436 | dec.json.flags |= F_HOOK; |
1436 | dec.json.flags |= F_HOOK; |
1437 | |
1437 | |
1438 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1438 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
… | |
… | |
1584 | self->incr_mode = INCR_M_STR; |
1584 | self->incr_mode = INCR_M_STR; |
1585 | goto incr_m_str; |
1585 | goto incr_m_str; |
1586 | |
1586 | |
1587 | case '[': |
1587 | case '[': |
1588 | case '{': |
1588 | case '{': |
1589 | ++self->incr_nest; |
1589 | if (++self->incr_nest > self->max_depth) |
|
|
1590 | croak (ERR_NESTING_EXCEEDED); |
1590 | break; |
1591 | break; |
1591 | |
1592 | |
1592 | case ']': |
1593 | case ']': |
1593 | case '}': |
1594 | case '}': |
1594 | if (!--self->incr_nest) |
1595 | if (!--self->incr_nest) |
… | |
… | |
1639 | void new (char *klass) |
1640 | void new (char *klass) |
1640 | PPCODE: |
1641 | PPCODE: |
1641 | { |
1642 | { |
1642 | SV *pv = NEWSV (0, sizeof (JSON)); |
1643 | SV *pv = NEWSV (0, sizeof (JSON)); |
1643 | SvPOK_only (pv); |
1644 | SvPOK_only (pv); |
1644 | Zero (SvPVX (pv), 1, JSON); |
1645 | json_init ((JSON *)SvPVX (pv)); |
1645 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1646 | XPUSHs (sv_2mortal (sv_bless ( |
1646 | XPUSHs (sv_2mortal (sv_bless ( |
1647 | newRV_noinc (pv), |
1647 | newRV_noinc (pv), |
1648 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1648 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1649 | ))); |
1649 | ))); |
1650 | } |
1650 | } |
… | |
… | |
1691 | get_relaxed = F_RELAXED |
1691 | get_relaxed = F_RELAXED |
1692 | get_allow_unknown = F_ALLOW_UNKNOWN |
1692 | get_allow_unknown = F_ALLOW_UNKNOWN |
1693 | PPCODE: |
1693 | PPCODE: |
1694 | XPUSHs (boolSV (self->flags & ix)); |
1694 | XPUSHs (boolSV (self->flags & ix)); |
1695 | |
1695 | |
1696 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1696 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1697 | PPCODE: |
1697 | PPCODE: |
1698 | { |
1698 | self->max_depth = max_depth; |
1699 | UV log2 = 0; |
|
|
1700 | |
|
|
1701 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1702 | |
|
|
1703 | while ((1UL << log2) < max_depth) |
|
|
1704 | ++log2; |
|
|
1705 | |
|
|
1706 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1707 | |
|
|
1708 | XPUSHs (ST (0)); |
1699 | XPUSHs (ST (0)); |
1709 | } |
|
|
1710 | |
1700 | |
1711 | U32 get_max_depth (JSON *self) |
1701 | U32 get_max_depth (JSON *self) |
1712 | CODE: |
1702 | CODE: |
1713 | RETVAL = DEC_DEPTH (self->flags); |
1703 | RETVAL = self->max_depth; |
1714 | OUTPUT: |
1704 | OUTPUT: |
1715 | RETVAL |
1705 | RETVAL |
1716 | |
1706 | |
1717 | void max_size (JSON *self, UV max_size = 0) |
1707 | void max_size (JSON *self, U32 max_size = 0) |
1718 | PPCODE: |
1708 | PPCODE: |
1719 | { |
1709 | self->max_size = max_size; |
1720 | UV log2 = 0; |
|
|
1721 | |
|
|
1722 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1723 | if (max_size == 1) max_size = 2; |
|
|
1724 | |
|
|
1725 | while ((1UL << log2) < max_size) |
|
|
1726 | ++log2; |
|
|
1727 | |
|
|
1728 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1729 | |
|
|
1730 | XPUSHs (ST (0)); |
1710 | XPUSHs (ST (0)); |
1731 | } |
|
|
1732 | |
1711 | |
1733 | int get_max_size (JSON *self) |
1712 | int get_max_size (JSON *self) |
1734 | CODE: |
1713 | CODE: |
1735 | RETVAL = DEC_SIZE (self->flags); |
1714 | RETVAL = self->max_size; |
1736 | OUTPUT: |
1715 | OUTPUT: |
1737 | RETVAL |
1716 | RETVAL |
1738 | |
1717 | |
1739 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1718 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1740 | PPCODE: |
1719 | PPCODE: |
… | |
… | |
1819 | STRLEN offset; |
1798 | STRLEN offset; |
1820 | |
1799 | |
1821 | if (!INCR_DONE (self)) |
1800 | if (!INCR_DONE (self)) |
1822 | { |
1801 | { |
1823 | incr_parse (self); |
1802 | incr_parse (self); |
|
|
1803 | |
|
|
1804 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
1805 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1806 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
1807 | |
1824 | if (!INCR_DONE (self)) |
1808 | if (!INCR_DONE (self)) |
1825 | break; |
1809 | break; |
1826 | } |
1810 | } |
1827 | |
1811 | |
1828 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1812 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
… | |
… | |
1871 | ALIAS: |
1855 | ALIAS: |
1872 | to_json_ = 0 |
1856 | to_json_ = 0 |
1873 | encode_json = F_UTF8 |
1857 | encode_json = F_UTF8 |
1874 | PPCODE: |
1858 | PPCODE: |
1875 | { |
1859 | { |
1876 | JSON json = { F_DEFAULT | ix }; |
1860 | JSON json; |
|
|
1861 | json_init (&json); |
|
|
1862 | json.flags |= ix; |
1877 | XPUSHs (encode_json (scalar, &json)); |
1863 | XPUSHs (encode_json (scalar, &json)); |
1878 | } |
1864 | } |
1879 | |
1865 | |
1880 | void decode_json (SV *jsonstr) |
1866 | void decode_json (SV *jsonstr) |
1881 | ALIAS: |
1867 | ALIAS: |
1882 | from_json_ = 0 |
1868 | from_json_ = 0 |
1883 | decode_json = F_UTF8 |
1869 | decode_json = F_UTF8 |
1884 | PPCODE: |
1870 | PPCODE: |
1885 | { |
1871 | { |
1886 | JSON json = { F_DEFAULT | ix }; |
1872 | JSON json; |
|
|
1873 | json_init (&json); |
|
|
1874 | json.flags |= ix; |
1887 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1875 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1888 | } |
1876 | } |
1889 | |
1877 | |
1890 | |
1878 | |