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 1.103 by root, Wed Jan 6 08:01:39 2010 UTC

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
32#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL 36#define F_RELAXED 0x00001000UL
36#define F_ALLOW_UNKNOWN 0x00002000UL 37#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 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
43 39
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 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
48#define F_DEFAULT (9UL << S_MAXDEPTH)
49 41
50#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
51#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
52 44
53#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
68 60
69#define IN_RANGE_INC(type,val,beg,end) \ 61#define IN_RANGE_INC(type,val,beg,end) \
70 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
71 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
72 64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
73#ifdef USE_ITHREADS 67#ifdef USE_ITHREADS
74# define JSON_SLOW 1 68# define JSON_SLOW 1
75# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
76#else 70#else
77# define JSON_SLOW 0 71# define JSON_SLOW 0
83 77
84enum { 78enum {
85 INCR_M_WS = 0, // initial whitespace skipping, must be 0 79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
86 INCR_M_STR, // inside string 80 INCR_M_STR, // inside string
87 INCR_M_BS, // inside backslash 81 INCR_M_BS, // inside backslash
82 INCR_M_C0, // inside comment in initial whitespace sequence
83 INCR_M_C1, // inside comment in other places
88 INCR_M_JSON // outside anything, count nesting 84 INCR_M_JSON // outside anything, count nesting
89}; 85};
90 86
91#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) 87#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
92 88
93typedef struct { 89typedef struct {
94 U32 flags; 90 U32 flags;
91 U32 max_depth;
92 STRLEN max_size;
93
95 SV *cb_object; 94 SV *cb_object;
96 HV *cb_sk_object; 95 HV *cb_sk_object;
97 96
98 // for the incremental parser 97 // for the incremental parser
99 SV *incr_text; // the source text so far 98 SV *incr_text; // the source text so far
100 STRLEN incr_pos; // the current offset into the text 99 STRLEN incr_pos; // the current offset into the text
101 int incr_nest; // {[]}-nesting level 100 int incr_nest; // {[]}-nesting level
102 int incr_mode; 101 unsigned char incr_mode;
103} JSON; 102} JSON;
103
104INLINE void
105json_init (JSON *json)
106{
107 Zero (json, 1, JSON);
108 json->max_depth = 512;
109}
104 110
105///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
106// utility functions 112// utility functions
107 113
108INLINE SV * 114INLINE SV *
118 124
119INLINE void 125INLINE void
120shrink (SV *sv) 126shrink (SV *sv)
121{ 127{
122 sv_utf8_downgrade (sv, 1); 128 sv_utf8_downgrade (sv, 1);
129
123 if (SvLEN (sv) > SvCUR (sv) + 1) 130 if (SvLEN (sv) > SvCUR (sv) + 1)
124 { 131 {
125#ifdef SvPV_shrink_to_cur 132#ifdef SvPV_shrink_to_cur
126 SvPV_shrink_to_cur (sv); 133 SvPV_shrink_to_cur (sv);
127#elif defined (SvPV_renew) 134#elif defined (SvPV_renew)
171 *s++ = 0x80 | ( ch & 0x3f); 178 *s++ = 0x80 | ( ch & 0x3f);
172 179
173 return s; 180 return s;
174} 181}
175 182
183// convert offset pointer to character index, sv must be string
184static STRLEN
185ptr_to_index (SV *sv, char *offset)
186{
187 return SvUTF8 (sv)
188 ? utf8_distance (offset, SvPVX (sv))
189 : offset - SvPVX (sv);
190}
191
176///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
177// encoder 193// encoder
178 194
179// structure used for encoding JSON 195// structure used for encoding JSON
180typedef struct 196typedef struct
182 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
183 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
184 SV *sv; // result scalar 200 SV *sv; // result scalar
185 JSON json; 201 JSON json;
186 U32 indent; // indentation level 202 U32 indent; // indentation level
187 U32 maxdepth; // max. indentation/recursion level
188 UV limit; // escape character values >= this value when encoding 203 UV limit; // escape character values >= this value when encoding
189} enc_t; 204} enc_t;
190 205
191INLINE void 206INLINE void
192need (enc_t *enc, STRLEN len) 207need (enc_t *enc, STRLEN len)
193{ 208{
194 if (expect_false (enc->cur + len >= enc->end)) 209 if (expect_false (enc->cur + len >= enc->end))
195 { 210 {
196 STRLEN cur = enc->cur - SvPVX (enc->sv); 211 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
197 SvGROW (enc->sv, cur + len + 1); 212 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
198 enc->cur = SvPVX (enc->sv) + cur; 213 enc->cur = SvPVX (enc->sv) + cur;
199 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 214 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
200 } 215 }
201} 216}
202 217
277 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 292 (int)((uch - 0x10000) % 0x400 + 0xDC00));
278 enc->cur += 12; 293 enc->cur += 12;
279 } 294 }
280 else 295 else
281 { 296 {
282 static char hexdigit [16] = "0123456789abcdef";
283 need (enc, len += 5); 297 need (enc, len += 5);
284 *enc->cur++ = '\\'; 298 *enc->cur++ = '\\';
285 *enc->cur++ = 'u'; 299 *enc->cur++ = 'u';
286 *enc->cur++ = hexdigit [ uch >> 12 ]; 300 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
287 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 301 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
288 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 302 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
289 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 303 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
290 } 304 }
291 305
292 str += clen; 306 str += clen;
293 } 307 }
294 else if (enc->json.flags & F_LATIN1) 308 else if (enc->json.flags & F_LATIN1)
365static void 379static void
366encode_av (enc_t *enc, AV *av) 380encode_av (enc_t *enc, AV *av)
367{ 381{
368 int i, len = av_len (av); 382 int i, len = av_len (av);
369 383
370 if (enc->indent >= enc->maxdepth) 384 if (enc->indent >= enc->json.max_depth)
371 croak ("data structure too deep (hit recursion limit)"); 385 croak (ERR_NESTING_EXCEEDED);
372 386
373 encode_ch (enc, '['); 387 encode_ch (enc, '[');
374 388
375 if (len >= 0) 389 if (len >= 0)
376 { 390 {
451static void 465static void
452encode_hv (enc_t *enc, HV *hv) 466encode_hv (enc_t *enc, HV *hv)
453{ 467{
454 HE *he; 468 HE *he;
455 469
456 if (enc->indent >= enc->maxdepth) 470 if (enc->indent >= enc->json.max_depth)
457 croak ("data structure too deep (hit recursion limit)"); 471 croak (ERR_NESTING_EXCEEDED);
458 472
459 encode_ch (enc, '{'); 473 encode_ch (enc, '{');
460 474
461 // for canonical output we have to sort by keys first 475 // for canonical output we have to sort by keys first
462 // actually, this is mostly due to the stupid so-called 476 // actually, this is mostly due to the stupid so-called
463 // security workaround added somewhere in 5.8.x. 477 // security workaround added somewhere in 5.8.x
464 // that randomises hash orderings 478 // that randomises hash orderings
465 if (enc->json.flags & F_CANONICAL) 479 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
466 { 480 {
467 int count = hv_iterinit (hv); 481 int count = hv_iterinit (hv);
468 482
469 if (SvMAGICAL (hv)) 483 if (SvMAGICAL (hv))
470 { 484 {
742 enc.json = *json; 756 enc.json = *json;
743 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 757 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
744 enc.cur = SvPVX (enc.sv); 758 enc.cur = SvPVX (enc.sv);
745 enc.end = SvEND (enc.sv); 759 enc.end = SvEND (enc.sv);
746 enc.indent = 0; 760 enc.indent = 0;
747 enc.maxdepth = DEC_DEPTH (enc.json.flags);
748 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL 761 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
749 : enc.json.flags & F_LATIN1 ? 0x000100UL 762 : enc.json.flags & F_LATIN1 ? 0x000100UL
750 : 0x110000UL; 763 : 0x110000UL;
751 764
752 SvPOK_only (enc.sv); 765 SvPOK_only (enc.sv);
753 encode_sv (&enc, scalar); 766 encode_sv (&enc, scalar);
767 encode_nl (&enc);
754 768
755 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 769 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
756 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 770 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
757 771
758 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 772 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
819 if (*dec->cur != ch) \ 833 if (*dec->cur != ch) \
820 ERR (# ch " expected"); \ 834 ERR (# ch " expected"); \
821 ++dec->cur; \ 835 ++dec->cur; \
822 SE 836 SE
823 837
824#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 838#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
825#define DEC_DEC_DEPTH --dec->depth 839#define DEC_DEC_DEPTH --dec->depth
826 840
827static SV *decode_sv (dec_t *dec); 841static SV *decode_sv (dec_t *dec);
828 842
829static signed char decode_hexdigit[256]; 843static signed char decode_hexdigit[256];
938 else if (expect_true (ch >= 0x20 && ch < 0x80)) 952 else if (expect_true (ch >= 0x20 && ch < 0x80))
939 *cur++ = ch; 953 *cur++ = ch;
940 else if (ch >= 0x80) 954 else if (ch >= 0x80)
941 { 955 {
942 STRLEN clen; 956 STRLEN clen;
943 UV uch;
944 957
945 --dec_cur; 958 --dec_cur;
946 959
947 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 960 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
948 if (clen == (STRLEN)-1) 961 if (clen == (STRLEN)-1)
949 ERR ("malformed UTF-8 character in JSON string"); 962 ERR ("malformed UTF-8 character in JSON string");
950 963
951 do 964 do
952 *cur++ = *dec_cur++; 965 *cur++ = *dec_cur++;
969 { 982 {
970 STRLEN len = cur - buf; 983 STRLEN len = cur - buf;
971 984
972 if (sv) 985 if (sv)
973 { 986 {
974 SvGROW (sv, SvCUR (sv) + len + 1); 987 STRLEN cur = SvCUR (sv);
988
989 if (SvLEN (sv) <= cur + len)
990 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
991
975 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 992 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
976 SvCUR_set (sv, SvCUR (sv) + len); 993 SvCUR_set (sv, SvCUR (sv) + len);
977 } 994 }
978 else 995 else
979 sv = newSVpvn (buf, len); 996 sv = newSVpvn (buf, len);
1070 1087
1071 // special case the rather common 1..5-digit-int case 1088 // special case the rather common 1..5-digit-int case
1072 if (*start == '-') 1089 if (*start == '-')
1073 switch (len) 1090 switch (len)
1074 { 1091 {
1075 case 2: return newSViv (-( start [1] - '0' * 1)); 1092 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1076 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1093 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1077 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1094 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1078 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1095 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1079 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1096 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1080 } 1097 }
1081 else 1098 else
1082 switch (len) 1099 switch (len)
1083 { 1100 {
1084 case 1: return newSViv ( start [0] - '0' * 1); 1101 case 1: return newSViv ( start [0] - '0' * 1);
1085 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1102 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1086 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1103 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1087 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1104 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1088 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1105 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1089 } 1106 }
1090 1107
1091 { 1108 {
1092 UV uv; 1109 UV uv;
1093 int numtype = grok_number (start, len, &uv); 1110 int numtype = grok_number (start, len, &uv);
1403fail: 1420fail:
1404 return 0; 1421 return 0;
1405} 1422}
1406 1423
1407static SV * 1424static SV *
1408decode_json (SV *string, JSON *json, STRLEN *offset_return) 1425decode_json (SV *string, JSON *json, char **offset_return)
1409{ 1426{
1410 dec_t dec; 1427 dec_t dec;
1411 STRLEN offset;
1412 SV *sv; 1428 SV *sv;
1413 1429
1430 /* work around bugs in 5.10 where manipulating magic values
1431 * will perl ignore the magic in subsequent accesses
1432 */
1414 SvGETMAGIC (string); 1433 /*SvGETMAGIC (string);*/
1434 if (SvMAGICAL (string))
1435 string = sv_2mortal (newSVsv (string));
1436
1415 SvUPGRADE (string, SVt_PV); 1437 SvUPGRADE (string, SVt_PV);
1416 1438
1417 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1439 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1440 * assertion with -DDEBUGGING, although SvCUR is documented to
1441 * return the xpv_cur field which certainly exists after upgrading.
1442 * according to nicholas clark, calling SvPOK fixes this.
1443 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1444 * and hope for the best.
1445 * Damnit, SvPV_nolen still trips over yet another assertion. This
1446 * assertion business is seriously broken, try yet another workaround
1447 * for the broken -DDEBUGGING.
1448 */
1449 {
1450#ifdef DEBUGGING
1451 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1452#else
1453 STRLEN offset = SvCUR (string);
1454#endif
1455
1456 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", 1457 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)); 1458 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1459 }
1420 1460
1421 if (json->flags & F_UTF8) 1461 if (json->flags & F_UTF8)
1422 sv_utf8_downgrade (string, 0); 1462 sv_utf8_downgrade (string, 0);
1423 else 1463 else
1424 sv_utf8_upgrade (string); 1464 sv_utf8_upgrade (string);
1425 1465
1426 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1466 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1427 1467
1428 dec.json = *json; 1468 dec.json = *json;
1429 dec.cur = SvPVX (string); 1469 dec.cur = SvPVX (string);
1430 dec.end = SvEND (string); 1470 dec.end = SvEND (string);
1431 dec.err = 0; 1471 dec.err = 0;
1432 dec.depth = 0; 1472 dec.depth = 0;
1433 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1434 1473
1435 if (dec.json.cb_object || dec.json.cb_sk_object) 1474 if (dec.json.cb_object || dec.json.cb_sk_object)
1436 dec.json.flags |= F_HOOK; 1475 dec.json.flags |= F_HOOK;
1437 1476
1438 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1477 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1439 1478
1440 decode_ws (&dec); 1479 decode_ws (&dec);
1441 sv = decode_sv (&dec); 1480 sv = decode_sv (&dec);
1481
1482 if (offset_return)
1483 *offset_return = dec.cur;
1442 1484
1443 if (!(offset_return || !sv)) 1485 if (!(offset_return || !sv))
1444 { 1486 {
1445 // check for trailing garbage 1487 // check for trailing garbage
1446 decode_ws (&dec); 1488 decode_ws (&dec);
1449 { 1491 {
1450 dec.err = "garbage after JSON object"; 1492 dec.err = "garbage after JSON object";
1451 SvREFCNT_dec (sv); 1493 SvREFCNT_dec (sv);
1452 sv = 0; 1494 sv = 0;
1453 } 1495 }
1454 }
1455
1456 if (offset_return || !sv)
1457 {
1458 offset = dec.json.flags & F_UTF8
1459 ? dec.cur - SvPVX (string)
1460 : utf8_distance (dec.cur, SvPVX (string));
1461
1462 if (offset_return)
1463 *offset_return = offset;
1464 } 1496 }
1465 1497
1466 if (!sv) 1498 if (!sv)
1467 { 1499 {
1468 SV *uni = sv_newmortal (); 1500 SV *uni = sv_newmortal ();
1474 SAVEVPTR (PL_curcop); 1506 SAVEVPTR (PL_curcop);
1475 PL_curcop = &cop; 1507 PL_curcop = &cop;
1476 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1508 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1477 LEAVE; 1509 LEAVE;
1478 1510
1479 croak ("%s, at character offset %d [\"%s\"]", 1511 croak ("%s, at character offset %d (before \"%s\")",
1480 dec.err, 1512 dec.err,
1481 (int)offset, 1513 ptr_to_index (string, dec.cur),
1482 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1514 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1483 } 1515 }
1484 1516
1485 sv = sv_2mortal (sv); 1517 sv = sv_2mortal (sv);
1486 1518
1495 1527
1496static void 1528static void
1497incr_parse (JSON *self) 1529incr_parse (JSON *self)
1498{ 1530{
1499 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1531 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1532
1533 // the state machine here is a bit convoluted and could be simplified a lot
1534 // but this would make it slower, so...
1500 1535
1501 for (;;) 1536 for (;;)
1502 { 1537 {
1503 //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 1538 //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
1504 switch (self->incr_mode) 1539 switch (self->incr_mode)
1505 { 1540 {
1506 // only used for intiial whitespace skipping 1541 // only used for initial whitespace skipping
1507 case INCR_M_WS: 1542 case INCR_M_WS:
1508 for (;;) 1543 for (;;)
1509 { 1544 {
1510 if (*p > 0x20) 1545 if (*p > 0x20)
1511 { 1546 {
1547 if (*p == '#')
1548 {
1549 self->incr_mode = INCR_M_C0;
1550 goto incr_m_c;
1551 }
1552 else
1553 {
1512 self->incr_mode = INCR_M_JSON; 1554 self->incr_mode = INCR_M_JSON;
1513 goto incr_m_json; 1555 goto incr_m_json;
1556 }
1514 } 1557 }
1515 else if (!*p) 1558 else if (!*p)
1516 goto interrupt; 1559 goto interrupt;
1517 1560
1518 ++p; 1561 ++p;
1524 goto interrupt; 1567 goto interrupt;
1525 1568
1526 ++p; 1569 ++p;
1527 self->incr_mode = INCR_M_STR; 1570 self->incr_mode = INCR_M_STR;
1528 goto incr_m_str; 1571 goto incr_m_str;
1572
1573 // inside #-style comments
1574 case INCR_M_C0:
1575 case INCR_M_C1:
1576 incr_m_c:
1577 for (;;)
1578 {
1579 if (*p == '\n')
1580 {
1581 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1582 break;
1583 }
1584 else if (!*p)
1585 goto interrupt;
1586
1587 ++p;
1588 }
1589
1590 break;
1529 1591
1530 // inside a string 1592 // inside a string
1531 case INCR_M_STR: 1593 case INCR_M_STR:
1532 incr_m_str: 1594 incr_m_str:
1533 for (;;) 1595 for (;;)
1584 self->incr_mode = INCR_M_STR; 1646 self->incr_mode = INCR_M_STR;
1585 goto incr_m_str; 1647 goto incr_m_str;
1586 1648
1587 case '[': 1649 case '[':
1588 case '{': 1650 case '{':
1589 ++self->incr_nest; 1651 if (++self->incr_nest > self->max_depth)
1652 croak (ERR_NESTING_EXCEEDED);
1590 break; 1653 break;
1591 1654
1592 case ']': 1655 case ']':
1593 case '}': 1656 case '}':
1594 if (!--self->incr_nest) 1657 if (--self->incr_nest <= 0)
1595 goto interrupt; 1658 goto interrupt;
1659 break;
1660
1661 case '#':
1662 self->incr_mode = INCR_M_C1;
1663 goto incr_m_c;
1596 } 1664 }
1597 } 1665 }
1598 } 1666 }
1599 1667
1600 modechange: 1668 modechange:
1601 ; 1669 ;
1602 } 1670 }
1603 1671
1604interrupt: 1672interrupt:
1605 self->incr_pos = p - SvPVX (self->incr_text); 1673 self->incr_pos = p - SvPVX (self->incr_text);
1674 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1606 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1675 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1607} 1676}
1608 1677
1609///////////////////////////////////////////////////////////////////////////// 1678/////////////////////////////////////////////////////////////////////////////
1610// XS interface functions 1679// XS interface functions
1625 json_stash = gv_stashpv ("JSON::XS" , 1); 1694 json_stash = gv_stashpv ("JSON::XS" , 1);
1626 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1695 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1627 1696
1628 json_true = get_bool ("JSON::XS::true"); 1697 json_true = get_bool ("JSON::XS::true");
1629 json_false = get_bool ("JSON::XS::false"); 1698 json_false = get_bool ("JSON::XS::false");
1699
1700 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1630} 1701}
1631 1702
1632PROTOTYPES: DISABLE 1703PROTOTYPES: DISABLE
1633 1704
1634void CLONE (...) 1705void CLONE (...)
1639void new (char *klass) 1710void new (char *klass)
1640 PPCODE: 1711 PPCODE:
1641{ 1712{
1642 SV *pv = NEWSV (0, sizeof (JSON)); 1713 SV *pv = NEWSV (0, sizeof (JSON));
1643 SvPOK_only (pv); 1714 SvPOK_only (pv);
1644 Zero (SvPVX (pv), 1, JSON); 1715 json_init ((JSON *)SvPVX (pv));
1645 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1646 XPUSHs (sv_2mortal (sv_bless ( 1716 XPUSHs (sv_2mortal (sv_bless (
1647 newRV_noinc (pv), 1717 newRV_noinc (pv),
1648 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1718 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1649 ))); 1719 )));
1650} 1720}
1691 get_relaxed = F_RELAXED 1761 get_relaxed = F_RELAXED
1692 get_allow_unknown = F_ALLOW_UNKNOWN 1762 get_allow_unknown = F_ALLOW_UNKNOWN
1693 PPCODE: 1763 PPCODE:
1694 XPUSHs (boolSV (self->flags & ix)); 1764 XPUSHs (boolSV (self->flags & ix));
1695 1765
1696void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1766void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1697 PPCODE: 1767 PPCODE:
1698{ 1768 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)); 1769 XPUSHs (ST (0));
1709}
1710 1770
1711U32 get_max_depth (JSON *self) 1771U32 get_max_depth (JSON *self)
1712 CODE: 1772 CODE:
1713 RETVAL = DEC_DEPTH (self->flags); 1773 RETVAL = self->max_depth;
1714 OUTPUT: 1774 OUTPUT:
1715 RETVAL 1775 RETVAL
1716 1776
1717void max_size (JSON *self, UV max_size = 0) 1777void max_size (JSON *self, U32 max_size = 0)
1718 PPCODE: 1778 PPCODE:
1719{ 1779 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)); 1780 XPUSHs (ST (0));
1731}
1732 1781
1733int get_max_size (JSON *self) 1782int get_max_size (JSON *self)
1734 CODE: 1783 CODE:
1735 RETVAL = DEC_SIZE (self->flags); 1784 RETVAL = self->max_size;
1736 OUTPUT: 1785 OUTPUT:
1737 RETVAL 1786 RETVAL
1738 1787
1739void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1788void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1740 PPCODE: 1789 PPCODE:
1776 XPUSHs (decode_json (jsonstr, self, 0)); 1825 XPUSHs (decode_json (jsonstr, self, 0));
1777 1826
1778void decode_prefix (JSON *self, SV *jsonstr) 1827void decode_prefix (JSON *self, SV *jsonstr)
1779 PPCODE: 1828 PPCODE:
1780{ 1829{
1781 STRLEN offset; 1830 char *offset;
1782 EXTEND (SP, 2); 1831 EXTEND (SP, 2);
1783 PUSHs (decode_json (jsonstr, self, &offset)); 1832 PUSHs (decode_json (jsonstr, self, &offset));
1784 PUSHs (sv_2mortal (newSVuv (offset))); 1833 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1785} 1834}
1786 1835
1787void incr_parse (JSON *self, SV *jsonstr = 0) 1836void incr_parse (JSON *self, SV *jsonstr = 0)
1788 PPCODE: 1837 PPCODE:
1789{ 1838{
1791 self->incr_text = newSVpvn ("", 0); 1840 self->incr_text = newSVpvn ("", 0);
1792 1841
1793 // append data, if any 1842 // append data, if any
1794 if (jsonstr) 1843 if (jsonstr)
1795 { 1844 {
1796 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1845 if (SvUTF8 (jsonstr))
1797 { 1846 {
1847 if (!SvUTF8 (self->incr_text))
1848 {
1798 /* utf-8-ness differs, need to upgrade */ 1849 /* utf-8-ness differs, need to upgrade */
1799 sv_utf8_upgrade (self->incr_text); 1850 sv_utf8_upgrade (self->incr_text);
1800 1851
1801 if (self->incr_pos) 1852 if (self->incr_pos)
1802 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) 1853 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1803 - (U8 *)SvPVX (self->incr_text); 1854 - (U8 *)SvPVX (self->incr_text);
1855 }
1804 } 1856 }
1857 else if (SvUTF8 (self->incr_text))
1858 sv_utf8_upgrade (jsonstr);
1805 1859
1806 { 1860 {
1807 STRLEN len; 1861 STRLEN len;
1808 const char *str = SvPV (jsonstr, len); 1862 const char *str = SvPV (jsonstr, len);
1809 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1863 STRLEN cur = SvCUR (self->incr_text);
1864
1865 if (SvLEN (self->incr_text) <= cur + len)
1866 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1867
1810 Move (str, SvEND (self->incr_text), len, char); 1868 Move (str, SvEND (self->incr_text), len, char);
1811 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1869 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1812 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 1870 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1813 } 1871 }
1814 } 1872 }
1815 1873
1816 if (GIMME_V != G_VOID) 1874 if (GIMME_V != G_VOID)
1817 do 1875 do
1818 { 1876 {
1819 STRLEN offset; 1877 char *offset;
1820 1878
1821 if (!INCR_DONE (self)) 1879 if (!INCR_DONE (self))
1822 { 1880 {
1823 incr_parse (self); 1881 incr_parse (self);
1882
1883 if (self->incr_pos > self->max_size && self->max_size)
1884 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1885 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1886
1824 if (!INCR_DONE (self)) 1887 if (!INCR_DONE (self))
1825 break; 1888 break;
1826 } 1889 }
1827 1890
1828 XPUSHs (decode_json (self->incr_text, self, &offset)); 1891 XPUSHs (decode_json (self->incr_text, self, &offset));
1829 1892
1830 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1831 self->incr_pos -= offset; 1893 self->incr_pos -= offset - SvPVX (self->incr_text);
1832 self->incr_nest = 0; 1894 self->incr_nest = 0;
1833 self->incr_mode = 0; 1895 self->incr_mode = 0;
1896
1897 sv_chop (self->incr_text, offset);
1834 } 1898 }
1835 while (GIMME_V == G_ARRAY); 1899 while (GIMME_V == G_ARRAY);
1836} 1900}
1837 1901
1838SV *incr_text (JSON *self) 1902SV *incr_text (JSON *self)
1857 self->incr_nest = 0; 1921 self->incr_nest = 0;
1858 self->incr_mode = 0; 1922 self->incr_mode = 0;
1859 } 1923 }
1860} 1924}
1861 1925
1926void incr_reset (JSON *self)
1927 CODE:
1928{
1929 SvREFCNT_dec (self->incr_text);
1930 self->incr_text = 0;
1931 self->incr_pos = 0;
1932 self->incr_nest = 0;
1933 self->incr_mode = 0;
1934}
1935
1862void DESTROY (JSON *self) 1936void DESTROY (JSON *self)
1863 CODE: 1937 CODE:
1864 SvREFCNT_dec (self->cb_sk_object); 1938 SvREFCNT_dec (self->cb_sk_object);
1865 SvREFCNT_dec (self->cb_object); 1939 SvREFCNT_dec (self->cb_object);
1866 SvREFCNT_dec (self->incr_text); 1940 SvREFCNT_dec (self->incr_text);
1871 ALIAS: 1945 ALIAS:
1872 to_json_ = 0 1946 to_json_ = 0
1873 encode_json = F_UTF8 1947 encode_json = F_UTF8
1874 PPCODE: 1948 PPCODE:
1875{ 1949{
1876 JSON json = { F_DEFAULT | ix }; 1950 JSON json;
1951 json_init (&json);
1952 json.flags |= ix;
1877 XPUSHs (encode_json (scalar, &json)); 1953 XPUSHs (encode_json (scalar, &json));
1878} 1954}
1879 1955
1880void decode_json (SV *jsonstr) 1956void decode_json (SV *jsonstr)
1881 ALIAS: 1957 ALIAS:
1882 from_json_ = 0 1958 from_json_ = 0
1883 decode_json = F_UTF8 1959 decode_json = F_UTF8
1884 PPCODE: 1960 PPCODE:
1885{ 1961{
1886 JSON json = { F_DEFAULT | ix }; 1962 JSON json;
1963 json_init (&json);
1964 json.flags |= ix;
1887 XPUSHs (decode_json (jsonstr, &json, 0)); 1965 XPUSHs (decode_json (jsonstr, &json, 0));
1888} 1966}
1889 1967
1890 1968

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines