ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.84 by root, Thu Mar 27 06:37:35 2008 UTC vs.
Revision 1.93 by root, Mon Sep 29 03:09:27 2008 UTC

… …
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
… …
86 INCR_M_STR, // inside string 79 INCR_M_STR, // inside string
87 INCR_M_BS, // inside backslash 80 INCR_M_BS, // inside backslash
88 INCR_M_JSON // outside anything, count nesting 81 INCR_M_JSON // outside anything, count nesting
89}; 82};
90 83
91#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)
92 85
93typedef struct { 86typedef 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 int incr_nest; // {[]}-nesting level
102 int incr_mode; 98 unsigned char incr_mode;
103} JSON; 99} JSON;
100
101INLINE void
102json_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
108INLINE SV * 111INLINE SV *
… …
118 121
119INLINE void 122INLINE void
120shrink (SV *sv) 123shrink (SV *sv)
121{ 124{
122 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
123 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
124 { 128 {
125#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
126 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
127#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
… …
182 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
183 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
184 SV *sv; // result scalar 188 SV *sv; // result scalar
185 JSON json; 189 JSON json;
186 U32 indent; // indentation level 190 U32 indent; // indentation level
187 U32 maxdepth; // max. indentation/recursion level
188 UV limit; // escape character values >= this value when encoding 191 UV limit; // escape character values >= this value when encoding
189} enc_t; 192} enc_t;
190 193
191INLINE void 194INLINE void
192need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
193{ 196{
194 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
195 { 198 {
196 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
197 SvGROW (enc->sv, cur + len + 1); 200 SvGROW (enc->sv, cur + len + 1);
198 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
199 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
200 } 203 }
201} 204}
… …
277 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
278 enc->cur += 12; 281 enc->cur += 12;
279 } 282 }
280 else 283 else
281 { 284 {
282 static char hexdigit [16] = "0123456789abcdef";
283 need (enc, len += 5); 285 need (enc, len += 5);
284 *enc->cur++ = '\\'; 286 *enc->cur++ = '\\';
285 *enc->cur++ = 'u'; 287 *enc->cur++ = 'u';
286 *enc->cur++ = hexdigit [ uch >> 12 ]; 288 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
287 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 289 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
288 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 290 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
289 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 291 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
290 } 292 }
291 293
292 str += clen; 294 str += clen;
293 } 295 }
294 else if (enc->json.flags & F_LATIN1) 296 else if (enc->json.flags & F_LATIN1)
… …
365static void 367static void
366encode_av (enc_t *enc, AV *av) 368encode_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 {
… …
451static void 453static void
452encode_hv (enc_t *enc, HV *hv) 454encode_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
463 // security workaround added somewhere in 5.8.x. 465 // security workaround added somewhere in 5.8.x
464 // that randomises hash orderings 466 // that randomises hash orderings
465 if (enc->json.flags & F_CANONICAL) 467 if (enc->json.flags & F_CANONICAL)
466 { 468 {
467 int count = hv_iterinit (hv); 469 int count = hv_iterinit (hv);
468 470
… …
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
827static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
828 829
829static signed char decode_hexdigit[256]; 830static 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 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1419 * assertion with -DDEBUGGING, although SvCUR is documented to
1420 * return the xpv_cur field which certainly exists after upgrading.
1421 * according to nicholas clark, calling SvPOK fixes this.
1422 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1423 * and hope for the best.
1424 * Damnit, SvPV_nolen still trips over yet another assertion. This
1425 * assertion business is seriously broken, try yet another workaround
1426 * for the broken -DDEBUGGING.
1427 */
1428#ifdef DEBUGGING
1429 offset = SvOK (string) ? sv_len (string) : 0;
1430#else
1431 offset = SvCUR (string);
1432#endif
1433
1434 if (offset > json->max_size && json->max_size)
1418 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1435 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)); 1436 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1420 1437
1421 if (json->flags & F_UTF8) 1438 if (json->flags & F_UTF8)
1422 sv_utf8_downgrade (string, 0); 1439 sv_utf8_downgrade (string, 0);
1423 else 1440 else
1424 sv_utf8_upgrade (string); 1441 sv_utf8_upgrade (string);
1425 1442
1426 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1443 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1427 1444
1428 dec.json = *json; 1445 dec.json = *json;
1429 dec.cur = SvPVX (string); 1446 dec.cur = SvPVX (string);
1430 dec.end = SvEND (string); 1447 dec.end = SvEND (string);
1431 dec.err = 0; 1448 dec.err = 0;
1432 dec.depth = 0; 1449 dec.depth = 0;
1433 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1434 1450
1435 if (dec.json.cb_object || dec.json.cb_sk_object) 1451 if (dec.json.cb_object || dec.json.cb_sk_object)
1436 dec.json.flags |= F_HOOK; 1452 dec.json.flags |= F_HOOK;
1437 1453
1438 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1454 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
… …
1584 self->incr_mode = INCR_M_STR; 1600 self->incr_mode = INCR_M_STR;
1585 goto incr_m_str; 1601 goto incr_m_str;
1586 1602
1587 case '[': 1603 case '[':
1588 case '{': 1604 case '{':
1589 ++self->incr_nest; 1605 if (++self->incr_nest > self->max_depth)
1606 croak (ERR_NESTING_EXCEEDED);
1590 break; 1607 break;
1591 1608
1592 case ']': 1609 case ']':
1593 case '}': 1610 case '}':
1594 if (!--self->incr_nest) 1611 if (--self->incr_nest <= 0)
1595 goto interrupt; 1612 goto interrupt;
1596 } 1613 }
1597 } 1614 }
1598 } 1615 }
1599 1616
… …
1639void new (char *klass) 1656void new (char *klass)
1640 PPCODE: 1657 PPCODE:
1641{ 1658{
1642 SV *pv = NEWSV (0, sizeof (JSON)); 1659 SV *pv = NEWSV (0, sizeof (JSON));
1643 SvPOK_only (pv); 1660 SvPOK_only (pv);
1644 Zero (SvPVX (pv), 1, JSON); 1661 json_init ((JSON *)SvPVX (pv));
1645 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1646 XPUSHs (sv_2mortal (sv_bless ( 1662 XPUSHs (sv_2mortal (sv_bless (
1647 newRV_noinc (pv), 1663 newRV_noinc (pv),
1648 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1664 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1649 ))); 1665 )));
1650} 1666}
… …
1691 get_relaxed = F_RELAXED 1707 get_relaxed = F_RELAXED
1692 get_allow_unknown = F_ALLOW_UNKNOWN 1708 get_allow_unknown = F_ALLOW_UNKNOWN
1693 PPCODE: 1709 PPCODE:
1694 XPUSHs (boolSV (self->flags & ix)); 1710 XPUSHs (boolSV (self->flags & ix));
1695 1711
1696void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1712void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1697 PPCODE: 1713 PPCODE:
1698{ 1714 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)); 1715 XPUSHs (ST (0));
1709}
1710 1716
1711U32 get_max_depth (JSON *self) 1717U32 get_max_depth (JSON *self)
1712 CODE: 1718 CODE:
1713 RETVAL = DEC_DEPTH (self->flags); 1719 RETVAL = self->max_depth;
1714 OUTPUT: 1720 OUTPUT:
1715 RETVAL 1721 RETVAL
1716 1722
1717void max_size (JSON *self, UV max_size = 0) 1723void max_size (JSON *self, U32 max_size = 0)
1718 PPCODE: 1724 PPCODE:
1719{ 1725 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)); 1726 XPUSHs (ST (0));
1731}
1732 1727
1733int get_max_size (JSON *self) 1728int get_max_size (JSON *self)
1734 CODE: 1729 CODE:
1735 RETVAL = DEC_SIZE (self->flags); 1730 RETVAL = self->max_size;
1736 OUTPUT: 1731 OUTPUT:
1737 RETVAL 1732 RETVAL
1738 1733
1739void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1734void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1740 PPCODE: 1735 PPCODE:
… …
1819 STRLEN offset; 1814 STRLEN offset;
1820 1815
1821 if (!INCR_DONE (self)) 1816 if (!INCR_DONE (self))
1822 { 1817 {
1823 incr_parse (self); 1818 incr_parse (self);
1819
1820 if (self->incr_pos > self->max_size && self->max_size)
1821 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1822 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1823
1824 if (!INCR_DONE (self)) 1824 if (!INCR_DONE (self))
1825 break; 1825 break;
1826 } 1826 }
1827 1827
1828 XPUSHs (decode_json (self->incr_text, self, &offset)); 1828 XPUSHs (decode_json (self->incr_text, self, &offset));
… …
1857 self->incr_nest = 0; 1857 self->incr_nest = 0;
1858 self->incr_mode = 0; 1858 self->incr_mode = 0;
1859 } 1859 }
1860} 1860}
1861 1861
1862void incr_reset (JSON *self)
1863 CODE:
1864{
1865 SvREFCNT_dec (self->incr_text);
1866 self->incr_text = 0;
1867 self->incr_pos = 0;
1868 self->incr_nest = 0;
1869 self->incr_mode = 0;
1870}
1871
1862void DESTROY (JSON *self) 1872void DESTROY (JSON *self)
1863 CODE: 1873 CODE:
1864 SvREFCNT_dec (self->cb_sk_object); 1874 SvREFCNT_dec (self->cb_sk_object);
1865 SvREFCNT_dec (self->cb_object); 1875 SvREFCNT_dec (self->cb_object);
1866 SvREFCNT_dec (self->incr_text); 1876 SvREFCNT_dec (self->incr_text);
… …
1871 ALIAS: 1881 ALIAS:
1872 to_json_ = 0 1882 to_json_ = 0
1873 encode_json = F_UTF8 1883 encode_json = F_UTF8
1874 PPCODE: 1884 PPCODE:
1875{ 1885{
1876 JSON json = { F_DEFAULT | ix }; 1886 JSON json;
1887 json_init (&json);
1888 json.flags |= ix;
1877 XPUSHs (encode_json (scalar, &json)); 1889 XPUSHs (encode_json (scalar, &json));
1878} 1890}
1879 1891
1880void decode_json (SV *jsonstr) 1892void decode_json (SV *jsonstr)
1881 ALIAS: 1893 ALIAS:
1882 from_json_ = 0 1894 from_json_ = 0
1883 decode_json = F_UTF8 1895 decode_json = F_UTF8
1884 PPCODE: 1896 PPCODE:
1885{ 1897{
1886 JSON json = { F_DEFAULT | ix }; 1898 JSON json;
1899 json_init (&json);
1900 json.flags |= ix;
1887 XPUSHs (decode_json (jsonstr, &json, 0)); 1901 XPUSHs (decode_json (jsonstr, &json, 0));
1888} 1902}
1889 1903
1890 1904

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines