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 1.92 by root, Mon Sep 22 07:29:29 2008 UTC vs.
Revision 1.117 by root, Fri Oct 25 20:19:57 2013 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
… …
41#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
42#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
43 44
44#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
45 46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
48
46#define SB do { 49#define SB do {
47#define SE } while (0) 50#define SE } while (0)
48 51
49#if __GNUC__ >= 3 52#if __GNUC__ >= 3
50# define expect(expr,value) __builtin_expect ((expr), (value)) 53# define expect(expr,value) __builtin_expect ((expr), (value))
… …
69#else 72#else
70# define JSON_SLOW 0 73# define JSON_SLOW 0
71# define JSON_STASH json_stash 74# define JSON_STASH json_stash
72#endif 75#endif
73 76
77// the amount of HEs to allocate on the stack, when sorting keys
78#define STACK_HES 64
79
74static HV *json_stash, *json_boolean_stash; // JSON::XS:: 80static HV *json_stash, *json_boolean_stash; // JSON::XS::
75static SV *json_true, *json_false; 81static SV *json_true, *json_false;
76 82
77enum { 83enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0 84 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string 85 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash 86 INCR_M_BS, // inside backslash
87 INCR_M_C0, // inside comment in initial whitespace sequence
88 INCR_M_C1, // inside comment in other places
81 INCR_M_JSON // outside anything, count nesting 89 INCR_M_JSON // outside anything, count nesting
82}; 90};
83 91
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) 92#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85 93
… …
175 *s++ = 0x80 | ( ch & 0x3f); 183 *s++ = 0x80 | ( ch & 0x3f);
176 184
177 return s; 185 return s;
178} 186}
179 187
188// convert offset pointer to character index, sv must be string
189static STRLEN
190ptr_to_index (SV *sv, char *offset)
191{
192 return SvUTF8 (sv)
193 ? utf8_distance (offset, SvPVX (sv))
194 : offset - SvPVX (sv);
195}
196
197/////////////////////////////////////////////////////////////////////////////
198// fp hell
199
200// scan a group of digits, and a trailing exponent
201static void
202json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
203{
204 UV uaccum = 0;
205 int eaccum = 0;
206
207 // if we recurse too deep, skip all remaining digits
208 // to avoid a stack overflow attack
209 if (expect_false (--maxdepth <= 0))
210 while (((U8)*s - '0') < 10)
211 ++s;
212
213 for (;;)
214 {
215 U8 dig = (U8)*s - '0';
216
217 if (expect_false (dig >= 10))
218 {
219 if (dig == (U8)((U8)'.' - (U8)'0'))
220 {
221 ++s;
222 json_atof_scan1 (s, accum, expo, 1, maxdepth);
223 }
224 else if ((dig | ' ') == 'e' - '0')
225 {
226 int exp2 = 0;
227 int neg = 0;
228
229 ++s;
230
231 if (*s == '-')
232 {
233 ++s;
234 neg = 1;
235 }
236 else if (*s == '+')
237 ++s;
238
239 while ((dig = (U8)*s - '0') < 10)
240 exp2 = exp2 * 10 + *s++ - '0';
241
242 *expo += neg ? -exp2 : exp2;
243 }
244
245 break;
246 }
247
248 ++s;
249
250 uaccum = uaccum * 10 + dig;
251 ++eaccum;
252
253 // if we have too many digits, then recurse for more
254 // we actually do this for rather few digits
255 if (uaccum >= (UV_MAX - 9) / 10)
256 {
257 if (postdp) *expo -= eaccum;
258 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
259 if (postdp) *expo += eaccum;
260
261 break;
262 }
263 }
264
265 // this relies greatly on the quality of the pow ()
266 // implementation of the platform, but a good
267 // implementation is hard to beat.
268 // (IEEE 754 conformant ones are required to be exact)
269 if (postdp) *expo -= eaccum;
270 *accum += uaccum * Perl_pow (10., *expo);
271 *expo += eaccum;
272}
273
274static NV
275json_atof (const char *s)
276{
277 NV accum = 0.;
278 int expo = 0;
279 int neg = 0;
280
281 if (*s == '-')
282 {
283 ++s;
284 neg = 1;
285 }
286
287 // a recursion depth of ten gives us >>500 bits
288 json_atof_scan1 (s, &accum, &expo, 0, 10);
289
290 return neg ? -accum : accum;
291}
180///////////////////////////////////////////////////////////////////////////// 292/////////////////////////////////////////////////////////////////////////////
181// encoder 293// encoder
182 294
183// structure used for encoding JSON 295// structure used for encoding JSON
184typedef struct 296typedef struct
… …
194INLINE void 306INLINE void
195need (enc_t *enc, STRLEN len) 307need (enc_t *enc, STRLEN len)
196{ 308{
197 if (expect_false (enc->cur + len >= enc->end)) 309 if (expect_false (enc->cur + len >= enc->end))
198 { 310 {
199 STRLEN cur = enc->cur - SvPVX (enc->sv); 311 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
200 SvGROW (enc->sv, cur + len + 1); 312 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
201 enc->cur = SvPVX (enc->sv) + cur; 313 enc->cur = SvPVX (enc->sv) + cur;
202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 314 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
203 } 315 }
204} 316}
205 317
… …
371 483
372 if (enc->indent >= enc->json.max_depth) 484 if (enc->indent >= enc->json.max_depth)
373 croak (ERR_NESTING_EXCEEDED); 485 croak (ERR_NESTING_EXCEEDED);
374 486
375 encode_ch (enc, '['); 487 encode_ch (enc, '[');
376 488
377 if (len >= 0) 489 if (len >= 0)
378 { 490 {
379 encode_nl (enc); ++enc->indent; 491 encode_nl (enc); ++enc->indent;
380 492
381 for (i = 0; i <= len; ++i) 493 for (i = 0; i <= len; ++i)
… …
393 encode_comma (enc); 505 encode_comma (enc);
394 } 506 }
395 507
396 encode_nl (enc); --enc->indent; encode_indent (enc); 508 encode_nl (enc); --enc->indent; encode_indent (enc);
397 } 509 }
398 510
399 encode_ch (enc, ']'); 511 encode_ch (enc, ']');
400} 512}
401 513
402static void 514static void
403encode_hk (enc_t *enc, HE *he) 515encode_hk (enc_t *enc, HE *he)
… …
407 if (HeKLEN (he) == HEf_SVKEY) 519 if (HeKLEN (he) == HEf_SVKEY)
408 { 520 {
409 SV *sv = HeSVKEY (he); 521 SV *sv = HeSVKEY (he);
410 STRLEN len; 522 STRLEN len;
411 char *str; 523 char *str;
412 524
413 SvGETMAGIC (sv); 525 SvGETMAGIC (sv);
414 str = SvPV (sv, len); 526 str = SvPV (sv, len);
415 527
416 encode_str (enc, str, len, SvUTF8 (sv)); 528 encode_str (enc, str, len, SvUTF8 (sv));
417 } 529 }
… …
462 574
463 // for canonical output we have to sort by keys first 575 // for canonical output we have to sort by keys first
464 // actually, this is mostly due to the stupid so-called 576 // actually, this is mostly due to the stupid so-called
465 // security workaround added somewhere in 5.8.x 577 // security workaround added somewhere in 5.8.x
466 // that randomises hash orderings 578 // that randomises hash orderings
467 if (enc->json.flags & F_CANONICAL) 579 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
468 { 580 {
469 int count = hv_iterinit (hv); 581 int count = hv_iterinit (hv);
470 582
471 if (SvMAGICAL (hv)) 583 if (SvMAGICAL (hv))
472 { 584 {
… …
483 } 595 }
484 596
485 if (count) 597 if (count)
486 { 598 {
487 int i, fast = 1; 599 int i, fast = 1;
488#if defined(__BORLANDC__) || defined(_MSC_VER) 600 HE *hes_stack [STACK_HES];
489 HE **hes = _alloca (count * sizeof (HE)); 601 HE **hes = hes_stack;
490#else 602
491 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 603 // allocate larger arrays on the heap
492#endif 604 if (count > STACK_HES)
605 {
606 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
607 hes = (HE **)SvPVX (sv);
608 }
493 609
494 i = 0; 610 i = 0;
495 while ((he = hv_iternext (hv))) 611 while ((he = hv_iternext (hv)))
496 { 612 {
497 hes [i++] = he; 613 hes [i++] = he;
… …
716 } 832 }
717 else 833 else
718 { 834 {
719 // large integer, use the (rather slow) snprintf way. 835 // large integer, use the (rather slow) snprintf way.
720 need (enc, IVUV_MAXCHARS); 836 need (enc, IVUV_MAXCHARS);
721 enc->cur += 837 enc->cur +=
722 SvIsUV(sv) 838 SvIsUV(sv)
723 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 839 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
724 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 840 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
725 } 841 }
726 } 842 }
727 else if (SvROK (sv)) 843 else if (SvROK (sv))
728 encode_rv (enc, SvRV (sv)); 844 encode_rv (enc, SvRV (sv));
729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 845 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
730 encode_str (enc, "null", 4, 0); 846 encode_str (enc, "null", 4, 0);
731 else 847 else
732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 848 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
733 SvPV_nolen (sv), SvFLAGS (sv)); 849 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
734} 850}
735 851
736static SV * 852static SV *
737encode_json (SV *scalar, JSON *json) 853encode_json (SV *scalar, JSON *json)
738{ 854{
… …
750 : enc.json.flags & F_LATIN1 ? 0x000100UL 866 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL; 867 : 0x110000UL;
752 868
753 SvPOK_only (enc.sv); 869 SvPOK_only (enc.sv);
754 encode_sv (&enc, scalar); 870 encode_sv (&enc, scalar);
871 encode_nl (&enc);
755 872
756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 873 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 874 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
758 875
759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 876 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
939 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1056 else if (expect_true (ch >= 0x20 && ch < 0x80))
940 *cur++ = ch; 1057 *cur++ = ch;
941 else if (ch >= 0x80) 1058 else if (ch >= 0x80)
942 { 1059 {
943 STRLEN clen; 1060 STRLEN clen;
944 UV uch;
945 1061
946 --dec_cur; 1062 --dec_cur;
947 1063
948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1064 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949 if (clen == (STRLEN)-1) 1065 if (clen == (STRLEN)-1)
950 ERR ("malformed UTF-8 character in JSON string"); 1066 ERR ("malformed UTF-8 character in JSON string");
951 1067
952 do 1068 do
953 *cur++ = *dec_cur++; 1069 *cur++ = *dec_cur++;
… …
970 { 1086 {
971 STRLEN len = cur - buf; 1087 STRLEN len = cur - buf;
972 1088
973 if (sv) 1089 if (sv)
974 { 1090 {
975 SvGROW (sv, SvCUR (sv) + len + 1); 1091 STRLEN cur = SvCUR (sv);
1092
1093 if (SvLEN (sv) <= cur + len)
1094 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1095
976 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1096 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
977 SvCUR_set (sv, SvCUR (sv) + len); 1097 SvCUR_set (sv, SvCUR (sv) + len);
978 } 1098 }
979 else 1099 else
980 sv = newSVpvn (buf, len); 1100 sv = newSVpvn (buf, len);
… …
1071 1191
1072 // special case the rather common 1..5-digit-int case 1192 // special case the rather common 1..5-digit-int case
1073 if (*start == '-') 1193 if (*start == '-')
1074 switch (len) 1194 switch (len)
1075 { 1195 {
1076 case 2: return newSViv (-( start [1] - '0' * 1)); 1196 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1197 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)); 1198 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)); 1199 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)); 1200 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1081 } 1201 }
1082 else 1202 else
1083 switch (len) 1203 switch (len)
1084 { 1204 {
1085 case 1: return newSViv ( start [0] - '0' * 1); 1205 case 1: return newSViv ( start [0] - '0' * 1);
1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1206 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); 1207 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); 1208 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); 1209 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1090 } 1210 }
1091 1211
1092 { 1212 {
1093 UV uv; 1213 UV uv;
1094 int numtype = grok_number (start, len, &uv); 1214 int numtype = grok_number (start, len, &uv);
… …
1103 } 1223 }
1104 1224
1105 len -= *start == '-' ? 1 : 0; 1225 len -= *start == '-' ? 1 : 0;
1106 1226
1107 // does not fit into IV or UV, try NV 1227 // does not fit into IV or UV, try NV
1108 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1228 if (len <= NV_DIG)
1109 #if defined (LDBL_DIG)
1110 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1111 #endif
1112 )
1113 // fits into NV without loss of precision 1229 // fits into NV without loss of precision
1114 return newSVnv (Atof (start)); 1230 return newSVnv (json_atof (start));
1115 1231
1116 // everything else fails, convert it to a string 1232 // everything else fails, convert it to a string
1117 return newSVpvn (start, dec->cur - start); 1233 return newSVpvn (start, dec->cur - start);
1118 } 1234 }
1119 1235
1120 // loss of precision here 1236 // loss of precision here
1121 return newSVnv (Atof (start)); 1237 return newSVnv (json_atof (start));
1122 1238
1123fail: 1239fail:
1124 return 0; 1240 return 0;
1125} 1241}
1126 1242
… …
1150 if (*dec->cur == ']') 1266 if (*dec->cur == ']')
1151 { 1267 {
1152 ++dec->cur; 1268 ++dec->cur;
1153 break; 1269 break;
1154 } 1270 }
1155 1271
1156 if (*dec->cur != ',') 1272 if (*dec->cur != ',')
1157 ERR (", or ] expected while parsing array"); 1273 ERR (", or ] expected while parsing array");
1158 1274
1159 ++dec->cur; 1275 ++dec->cur;
1160 1276
… …
1295 dSP; 1411 dSP;
1296 int count; 1412 int count;
1297 1413
1298 ENTER; SAVETMPS; PUSHMARK (SP); 1414 ENTER; SAVETMPS; PUSHMARK (SP);
1299 XPUSHs (HeVAL (he)); 1415 XPUSHs (HeVAL (he));
1416 sv_2mortal (sv);
1300 1417
1301 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1418 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1302 1419
1303 if (count == 1) 1420 if (count == 1)
1304 { 1421 {
1305 sv = newSVsv (POPs); 1422 sv = newSVsv (POPs);
1306 FREETMPS; LEAVE; 1423 FREETMPS; LEAVE;
1307 return sv; 1424 return sv;
1308 } 1425 }
1309 1426
1427 SvREFCNT_inc (sv);
1310 FREETMPS; LEAVE; 1428 FREETMPS; LEAVE;
1311 } 1429 }
1312 } 1430 }
1313 1431
1314 if (dec->json.cb_object) 1432 if (dec->json.cb_object)
… …
1346{ 1464{
1347 // the beauty of JSON: you need exactly one character lookahead 1465 // the beauty of JSON: you need exactly one character lookahead
1348 // to parse everything. 1466 // to parse everything.
1349 switch (*dec->cur) 1467 switch (*dec->cur)
1350 { 1468 {
1351 case '"': ++dec->cur; return decode_str (dec); 1469 case '"': ++dec->cur; return decode_str (dec);
1352 case '[': ++dec->cur; return decode_av (dec); 1470 case '[': ++dec->cur; return decode_av (dec);
1353 case '{': ++dec->cur; return decode_hv (dec); 1471 case '{': ++dec->cur; return decode_hv (dec);
1354 1472
1355 case '-': 1473 case '-':
1356 case '0': case '1': case '2': case '3': case '4': 1474 case '0': case '1': case '2': case '3': case '4':
1357 case '5': case '6': case '7': case '8': case '9': 1475 case '5': case '6': case '7': case '8': case '9':
… …
1404fail: 1522fail:
1405 return 0; 1523 return 0;
1406} 1524}
1407 1525
1408static SV * 1526static SV *
1409decode_json (SV *string, JSON *json, STRLEN *offset_return) 1527decode_json (SV *string, JSON *json, char **offset_return)
1410{ 1528{
1411 dec_t dec; 1529 dec_t dec;
1412 STRLEN offset;
1413 SV *sv; 1530 SV *sv;
1414 1531
1532 /* work around bugs in 5.10 where manipulating magic values
1533 * makes perl ignore the magic in subsequent accesses.
1534 * also make a copy of non-PV values, to get them into a clean
1535 * state (SvPV should do that, but it's buggy, see below).
1536 */
1415 SvGETMAGIC (string); 1537 /*SvGETMAGIC (string);*/
1538 if (SvMAGICAL (string) || !SvPOK (string))
1539 string = sv_2mortal (newSVsv (string));
1540
1416 SvUPGRADE (string, SVt_PV); 1541 SvUPGRADE (string, SVt_PV);
1417 1542
1418 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1543 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1419 * assertion with -DDEBUGGING, although SvCUR is documented to 1544 * assertion with -DDEBUGGING, although SvCUR is documented to
1420 * return the xpv_cur field which certainly exists after upgrading. 1545 * return the xpv_cur field which certainly exists after upgrading.
… …
1423 * and hope for the best. 1548 * and hope for the best.
1424 * Damnit, SvPV_nolen still trips over yet another assertion. This 1549 * Damnit, SvPV_nolen still trips over yet another assertion. This
1425 * assertion business is seriously broken, try yet another workaround 1550 * assertion business is seriously broken, try yet another workaround
1426 * for the broken -DDEBUGGING. 1551 * for the broken -DDEBUGGING.
1427 */ 1552 */
1553 {
1428#ifdef DEBUGGING 1554#ifdef DEBUGGING
1429 offset = SvOK (string) ? sv_len (string) : 0; 1555 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1430#else 1556#else
1431 offset = SvCUR (string); 1557 STRLEN offset = SvCUR (string);
1432#endif 1558#endif
1433 1559
1434 if (offset > json->max_size && json->max_size) 1560 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", 1561 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); 1562 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1563 }
1437 1564
1438 if (json->flags & F_UTF8) 1565 if (DECODE_WANTS_OCTETS (json))
1439 sv_utf8_downgrade (string, 0); 1566 sv_utf8_downgrade (string, 0);
1440 else 1567 else
1441 sv_utf8_upgrade (string); 1568 sv_utf8_upgrade (string);
1442 1569
1443 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1570 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
… …
1453 1580
1454 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1581 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1455 1582
1456 decode_ws (&dec); 1583 decode_ws (&dec);
1457 sv = decode_sv (&dec); 1584 sv = decode_sv (&dec);
1585
1586 if (offset_return)
1587 *offset_return = dec.cur;
1458 1588
1459 if (!(offset_return || !sv)) 1589 if (!(offset_return || !sv))
1460 { 1590 {
1461 // check for trailing garbage 1591 // check for trailing garbage
1462 decode_ws (&dec); 1592 decode_ws (&dec);
… …
1465 { 1595 {
1466 dec.err = "garbage after JSON object"; 1596 dec.err = "garbage after JSON object";
1467 SvREFCNT_dec (sv); 1597 SvREFCNT_dec (sv);
1468 sv = 0; 1598 sv = 0;
1469 } 1599 }
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 } 1600 }
1481 1601
1482 if (!sv) 1602 if (!sv)
1483 { 1603 {
1484 SV *uni = sv_newmortal (); 1604 SV *uni = sv_newmortal ();
… …
1490 SAVEVPTR (PL_curcop); 1610 SAVEVPTR (PL_curcop);
1491 PL_curcop = &cop; 1611 PL_curcop = &cop;
1492 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1612 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1493 LEAVE; 1613 LEAVE;
1494 1614
1495 croak ("%s, at character offset %d [\"%s\"]", 1615 croak ("%s, at character offset %d (before \"%s\")",
1496 dec.err, 1616 dec.err,
1497 (int)offset, 1617 (int)ptr_to_index (string, dec.cur),
1498 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1618 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1499 } 1619 }
1500 1620
1501 sv = sv_2mortal (sv); 1621 sv = sv_2mortal (sv);
1502 1622
… …
1511 1631
1512static void 1632static void
1513incr_parse (JSON *self) 1633incr_parse (JSON *self)
1514{ 1634{
1515 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1635 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1636
1637 // the state machine here is a bit convoluted and could be simplified a lot
1638 // but this would make it slower, so...
1516 1639
1517 for (;;) 1640 for (;;)
1518 { 1641 {
1519 //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 1642 //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
1520 switch (self->incr_mode) 1643 switch (self->incr_mode)
1521 { 1644 {
1522 // only used for intiial whitespace skipping 1645 // only used for initial whitespace skipping
1523 case INCR_M_WS: 1646 case INCR_M_WS:
1524 for (;;) 1647 for (;;)
1525 { 1648 {
1526 if (*p > 0x20) 1649 if (*p > 0x20)
1527 { 1650 {
1651 if (*p == '#')
1652 {
1653 self->incr_mode = INCR_M_C0;
1654 goto incr_m_c;
1655 }
1656 else
1657 {
1528 self->incr_mode = INCR_M_JSON; 1658 self->incr_mode = INCR_M_JSON;
1529 goto incr_m_json; 1659 goto incr_m_json;
1660 }
1530 } 1661 }
1531 else if (!*p) 1662 else if (!*p)
1532 goto interrupt; 1663 goto interrupt;
1533 1664
1534 ++p; 1665 ++p;
… …
1540 goto interrupt; 1671 goto interrupt;
1541 1672
1542 ++p; 1673 ++p;
1543 self->incr_mode = INCR_M_STR; 1674 self->incr_mode = INCR_M_STR;
1544 goto incr_m_str; 1675 goto incr_m_str;
1676
1677 // inside #-style comments
1678 case INCR_M_C0:
1679 case INCR_M_C1:
1680 incr_m_c:
1681 for (;;)
1682 {
1683 if (*p == '\n')
1684 {
1685 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1686 break;
1687 }
1688 else if (!*p)
1689 goto interrupt;
1690
1691 ++p;
1692 }
1693
1694 break;
1545 1695
1546 // inside a string 1696 // inside a string
1547 case INCR_M_STR: 1697 case INCR_M_STR:
1548 incr_m_str: 1698 incr_m_str:
1549 for (;;) 1699 for (;;)
… …
1608 1758
1609 case ']': 1759 case ']':
1610 case '}': 1760 case '}':
1611 if (--self->incr_nest <= 0) 1761 if (--self->incr_nest <= 0)
1612 goto interrupt; 1762 goto interrupt;
1763 break;
1764
1765 case '#':
1766 self->incr_mode = INCR_M_C1;
1767 goto incr_m_c;
1613 } 1768 }
1614 } 1769 }
1615 } 1770 }
1616 1771
1617 modechange: 1772 modechange:
1618 ; 1773 ;
1619 } 1774 }
1620 1775
1621interrupt: 1776interrupt:
1622 self->incr_pos = p - SvPVX (self->incr_text); 1777 self->incr_pos = p - SvPVX (self->incr_text);
1778 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1623 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1779 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1624} 1780}
1625 1781
1626///////////////////////////////////////////////////////////////////////////// 1782/////////////////////////////////////////////////////////////////////////////
1627// XS interface functions 1783// XS interface functions
… …
1642 json_stash = gv_stashpv ("JSON::XS" , 1); 1798 json_stash = gv_stashpv ("JSON::XS" , 1);
1643 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1799 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1644 1800
1645 json_true = get_bool ("JSON::XS::true"); 1801 json_true = get_bool ("JSON::XS::true");
1646 json_false = get_bool ("JSON::XS::false"); 1802 json_false = get_bool ("JSON::XS::false");
1803
1804 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1647} 1805}
1648 1806
1649PROTOTYPES: DISABLE 1807PROTOTYPES: DISABLE
1650 1808
1651void CLONE (...) 1809void CLONE (...)
… …
1654 json_boolean_stash = 0; 1812 json_boolean_stash = 0;
1655 1813
1656void new (char *klass) 1814void new (char *klass)
1657 PPCODE: 1815 PPCODE:
1658{ 1816{
1659 SV *pv = NEWSV (0, sizeof (JSON)); 1817 SV *pv = NEWSV (0, sizeof (JSON));
1660 SvPOK_only (pv); 1818 SvPOK_only (pv);
1661 json_init ((JSON *)SvPVX (pv)); 1819 json_init ((JSON *)SvPVX (pv));
1662 XPUSHs (sv_2mortal (sv_bless ( 1820 XPUSHs (sv_2mortal (sv_bless (
1663 newRV_noinc (pv), 1821 newRV_noinc (pv),
1664 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1822 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
… …
1741} 1899}
1742 1900
1743void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 1901void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1744 PPCODE: 1902 PPCODE:
1745{ 1903{
1746 if (!self->cb_sk_object) 1904 if (!self->cb_sk_object)
1747 self->cb_sk_object = newHV (); 1905 self->cb_sk_object = newHV ();
1748 1906
1749 if (SvOK (cb)) 1907 if (SvOK (cb))
1750 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 1908 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1751 else 1909 else
… …
1762 XPUSHs (ST (0)); 1920 XPUSHs (ST (0));
1763} 1921}
1764 1922
1765void encode (JSON *self, SV *scalar) 1923void encode (JSON *self, SV *scalar)
1766 PPCODE: 1924 PPCODE:
1767 XPUSHs (encode_json (scalar, self)); 1925 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
1926 XPUSHs (scalar);
1768 1927
1769void decode (JSON *self, SV *jsonstr) 1928void decode (JSON *self, SV *jsonstr)
1770 PPCODE: 1929 PPCODE:
1771 XPUSHs (decode_json (jsonstr, self, 0)); 1930 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
1931 XPUSHs (jsonstr);
1772 1932
1773void decode_prefix (JSON *self, SV *jsonstr) 1933void decode_prefix (JSON *self, SV *jsonstr)
1774 PPCODE: 1934 PPCODE:
1775{ 1935{
1776 STRLEN offset; 1936 SV *sv;
1937 char *offset;
1938 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1777 EXTEND (SP, 2); 1939 EXTEND (SP, 2);
1778 PUSHs (decode_json (jsonstr, self, &offset)); 1940 PUSHs (sv);
1779 PUSHs (sv_2mortal (newSVuv (offset))); 1941 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1780} 1942}
1781 1943
1782void incr_parse (JSON *self, SV *jsonstr = 0) 1944void incr_parse (JSON *self, SV *jsonstr = 0)
1783 PPCODE: 1945 PPCODE:
1784{ 1946{
1785 if (!self->incr_text) 1947 if (!self->incr_text)
1786 self->incr_text = newSVpvn ("", 0); 1948 self->incr_text = newSVpvn ("", 0);
1949
1950 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1951 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1952 if (DECODE_WANTS_OCTETS (self))
1953 {
1954 if (self->incr_pos)
1955 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1956 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1957
1958 sv_utf8_downgrade (self->incr_text, 0);
1959 }
1960 else
1961 {
1962 sv_utf8_upgrade (self->incr_text);
1963
1964 if (self->incr_pos)
1965 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1966 - (U8 *)SvPVX (self->incr_text);
1967 }
1787 1968
1788 // append data, if any 1969 // append data, if any
1789 if (jsonstr) 1970 if (jsonstr)
1790 { 1971 {
1972 /* make sure both strings have same encoding */
1791 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1973 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1974 if (SvUTF8 (jsonstr))
1975 sv_utf8_downgrade (jsonstr, 0);
1792 { 1976 else
1793 /* utf-8-ness differs, need to upgrade */
1794 sv_utf8_upgrade (self->incr_text); 1977 sv_utf8_upgrade (jsonstr);
1795 1978
1796 if (self->incr_pos) 1979 /* and then just blindly append */
1797 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1798 - (U8 *)SvPVX (self->incr_text);
1799 }
1800
1801 { 1980 {
1802 STRLEN len; 1981 STRLEN len;
1803 const char *str = SvPV (jsonstr, len); 1982 const char *str = SvPV (jsonstr, len);
1804 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1983 STRLEN cur = SvCUR (self->incr_text);
1984
1985 if (SvLEN (self->incr_text) <= cur + len)
1986 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1987
1805 Move (str, SvEND (self->incr_text), len, char); 1988 Move (str, SvEND (self->incr_text), len, char);
1806 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1989 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 1990 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1808 } 1991 }
1809 } 1992 }
1810 1993
1811 if (GIMME_V != G_VOID) 1994 if (GIMME_V != G_VOID)
1812 do 1995 do
1813 { 1996 {
1997 SV *sv;
1814 STRLEN offset; 1998 char *offset;
1815 1999
1816 if (!INCR_DONE (self)) 2000 if (!INCR_DONE (self))
1817 { 2001 {
1818 incr_parse (self); 2002 incr_parse (self);
1819 2003
1820 if (self->incr_pos > self->max_size && self->max_size) 2004 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", 2005 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); 2006 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1823 2007
1824 if (!INCR_DONE (self)) 2008 if (!INCR_DONE (self))
2009 {
2010 // as an optimisation, do not accumulate white space in the incr buffer
2011 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2012 {
2013 self->incr_pos = 0;
2014 SvCUR_set (self->incr_text, 0);
2015 }
2016
1825 break; 2017 break;
2018 }
1826 } 2019 }
1827 2020
1828 XPUSHs (decode_json (self->incr_text, self, &offset)); 2021 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2022 XPUSHs (sv);
1829 2023
1830 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1831 self->incr_pos -= offset; 2024 self->incr_pos -= offset - SvPVX (self->incr_text);
1832 self->incr_nest = 0; 2025 self->incr_nest = 0;
1833 self->incr_mode = 0; 2026 self->incr_mode = 0;
2027
2028 sv_chop (self->incr_text, offset);
1834 } 2029 }
1835 while (GIMME_V == G_ARRAY); 2030 while (GIMME_V == G_ARRAY);
1836} 2031}
1837 2032
1838SV *incr_text (JSON *self) 2033SV *incr_text (JSON *self)
… …
1884 PPCODE: 2079 PPCODE:
1885{ 2080{
1886 JSON json; 2081 JSON json;
1887 json_init (&json); 2082 json_init (&json);
1888 json.flags |= ix; 2083 json.flags |= ix;
1889 XPUSHs (encode_json (scalar, &json)); 2084 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2085 XPUSHs (scalar);
1890} 2086}
1891 2087
1892void decode_json (SV *jsonstr) 2088void decode_json (SV *jsonstr)
1893 ALIAS: 2089 ALIAS:
1894 from_json_ = 0 2090 from_json_ = 0
… …
1896 PPCODE: 2092 PPCODE:
1897{ 2093{
1898 JSON json; 2094 JSON json;
1899 json_init (&json); 2095 json_init (&json);
1900 json.flags |= ix; 2096 json.flags |= ix;
1901 XPUSHs (decode_json (jsonstr, &json, 0)); 2097 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2098 XPUSHs (jsonstr);
1902} 2099}
1903 2100
1904

Diff Legend

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