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 1.10 by root, Sat Mar 24 01:15:22 2007 UTC vs.
Revision 1.29 by root, Wed May 9 15:34:04 2007 UTC

3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
9#define F_ASCII 0x00000001 14#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 15#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 16#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 17#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 18#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 19#define F_SPACE_AFTER 0x00000020UL
15#define F_ALLOW_NONREF 0x00000080 20#define F_ALLOW_NONREF 0x00000080UL
16#define F_SHRINK 0x00000100 21#define F_SHRINK 0x00000100UL
22#define F_MAXDEPTH 0xf8000000UL
23#define S_MAXDEPTH 27
24
25#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
26
27// F_SELFCONVERT? <=> to_json/toJson
28// F_BLESSED? <=> { $__class__$ => }
17 29
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 30#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 31#define F_DEFAULT (9UL << S_MAXDEPTH)
20 32
21#define INIT_SIZE 32 // initial scalar size to be allocated 33#define INIT_SIZE 32 // initial scalar size to be allocated
34#define INDENT_STEP 3 // spaces per indentation level
35
36#define SHORT_STRING_LEN 512 // special-case strings of up to this size
22 37
23#define SB do { 38#define SB do {
24#define SE } while (0) 39#define SE } while (0)
25 40
26static HV *json_stash; 41static HV *json_stash; // JSON::XS::
42
43/////////////////////////////////////////////////////////////////////////////
44// utility functions
45
46static UV *
47SvJSON (SV *sv)
48{
49 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
50 croak ("object is not of type JSON::XS");
51
52 return &SvUVX (SvRV (sv));
53}
54
55static void
56shrink (SV *sv)
57{
58 sv_utf8_downgrade (sv, 1);
59 if (SvLEN (sv) > SvCUR (sv) + 1)
60 {
61#ifdef SvPV_shrink_to_cur
62 SvPV_shrink_to_cur (sv);
63#elif defined (SvPV_renew)
64 SvPV_renew (sv, SvCUR (sv) + 1);
65#endif
66 }
67}
68
69// decode an utf-8 character and return it, or (UV)-1 in
70// case of an error.
71// we special-case "safe" characters from U+80 .. U+7FF,
72// but use the very good perl function to parse anything else.
73// note that we never call this function for a ascii codepoints
74static UV
75decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
76{
77 if (s[0] > 0xdf || s[0] < 0xc2)
78 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
79 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
80 {
81 *clen = 2;
82 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
83 }
84 else
85 {
86 *clen = (STRLEN)-1;
87 return (UV)-1;
88 }
89}
90
91/////////////////////////////////////////////////////////////////////////////
92// encoder
27 93
28// structure used for encoding JSON 94// structure used for encoding JSON
29typedef struct 95typedef struct
30{ 96{
31 char *cur; 97 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 98 char *end; // SvEND (sv)
34 SV *sv; 99 SV *sv; // result scalar
35 UV flags; 100 U32 flags; // F_*
36 int max_recurse; 101 U32 indent; // indentation level
37 int indent; 102 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 103} enc_t;
39
40// structure used for decoding JSON
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58static void
59shrink (SV *sv)
60{
61 sv_utf8_downgrade (sv, 1);
62#ifdef SvPV_shrink_to_cur
63 SvPV_shrink_to_cur (sv);
64#endif
65}
66
67/////////////////////////////////////////////////////////////////////////////
68 104
69static void 105static void
70need (enc_t *enc, STRLEN len) 106need (enc_t *enc, STRLEN len)
71{ 107{
72 if (enc->cur + len >= enc->end) 108 if (enc->cur + len >= enc->end)
73 { 109 {
74 STRLEN cur = enc->cur - SvPVX (enc->sv); 110 STRLEN cur = enc->cur - SvPVX (enc->sv);
75 SvGROW (enc->sv, cur + len + 1); 111 SvGROW (enc->sv, cur + len + 1);
76 enc->cur = SvPVX (enc->sv) + cur; 112 enc->cur = SvPVX (enc->sv) + cur;
77 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 113 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
78 } 114 }
79} 115}
80 116
81static void 117static void
82encode_ch (enc_t *enc, char ch) 118encode_ch (enc_t *enc, char ch)
130 STRLEN clen; 166 STRLEN clen;
131 UV uch; 167 UV uch;
132 168
133 if (is_utf8) 169 if (is_utf8)
134 { 170 {
135 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 171 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
172 uch = decode_utf8 (str, end - str, &clen);
136 if (clen == (STRLEN)-1) 173 if (clen == (STRLEN)-1)
137 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 174 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
138 } 175 }
139 else 176 else
140 { 177 {
178 } 215 }
179 while (--clen); 216 while (--clen);
180 } 217 }
181 else 218 else
182 { 219 {
183 need (enc, len += 10); // never more than 11 bytes needed 220 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
184 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 221 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
185 ++str; 222 ++str;
186 } 223 }
187 } 224 }
188 } 225 }
190 227
191 --len; 228 --len;
192 } 229 }
193} 230}
194 231
195#define INDENT SB \ 232static void
233encode_indent (enc_t *enc)
234{
196 if (enc->flags & F_INDENT) \ 235 if (enc->flags & F_INDENT)
197 { \ 236 {
198 int i_; \ 237 int spaces = enc->indent * INDENT_STEP;
199 need (enc, enc->indent); \
200 for (i_ = enc->indent * 3; i_--; )\
201 encode_ch (enc, ' '); \
202 } \
203 SE
204 238
205#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 239 need (enc, spaces);
206#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 240 memset (enc->cur, ' ', spaces);
207#define COMMA SB \ 241 enc->cur += spaces;
242 }
243}
244
245static void
246encode_space (enc_t *enc)
247{
248 need (enc, 1);
208 encode_ch (enc, ','); \ 249 encode_ch (enc, ' ');
250}
251
252static void
253encode_nl (enc_t *enc)
254{
209 if (enc->flags & F_INDENT) \ 255 if (enc->flags & F_INDENT)
210 NL; \ 256 {
257 need (enc, 1);
258 encode_ch (enc, '\n');
259 }
260}
261
262static void
263encode_comma (enc_t *enc)
264{
265 encode_ch (enc, ',');
266
267 if (enc->flags & F_INDENT)
268 encode_nl (enc);
211 else if (enc->flags & F_SPACE_AFTER) \ 269 else if (enc->flags & F_SPACE_AFTER)
212 SPACE; \ 270 encode_space (enc);
213 SE 271}
214 272
215static void encode_sv (enc_t *enc, SV *sv); 273static void encode_sv (enc_t *enc, SV *sv);
216 274
217static void 275static void
218encode_av (enc_t *enc, AV *av) 276encode_av (enc_t *enc, AV *av)
219{ 277{
220 int i, len = av_len (av); 278 int i, len = av_len (av);
221 279
222 encode_ch (enc, '['); NL; 280 if (enc->indent >= enc->maxdepth)
281 croak ("data structure too deep (hit recursion limit)");
282
283 encode_ch (enc, '['); encode_nl (enc);
223 ++enc->indent; 284 ++enc->indent;
224 285
225 for (i = 0; i <= len; ++i) 286 for (i = 0; i <= len; ++i)
226 { 287 {
227 INDENT; 288 encode_indent (enc);
228 encode_sv (enc, *av_fetch (av, i, 0)); 289 encode_sv (enc, *av_fetch (av, i, 0));
229 290
230 if (i < len) 291 if (i < len)
231 COMMA; 292 encode_comma (enc);
232 } 293 }
233 294
234 NL; 295 encode_nl (enc);
235 296
236 --enc->indent; 297 --enc->indent;
237 INDENT; encode_ch (enc, ']'); 298 encode_indent (enc); encode_ch (enc, ']');
238} 299}
239 300
240static void 301static void
241encode_he (enc_t *enc, HE *he) 302encode_he (enc_t *enc, HE *he)
242{ 303{
256 else 317 else
257 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 318 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
258 319
259 encode_ch (enc, '"'); 320 encode_ch (enc, '"');
260 321
261 if (enc->flags & F_SPACE_BEFORE) SPACE; 322 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
262 encode_ch (enc, ':'); 323 encode_ch (enc, ':');
263 if (enc->flags & F_SPACE_AFTER ) SPACE; 324 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
264 encode_sv (enc, HeVAL (he)); 325 encode_sv (enc, HeVAL (he));
265} 326}
266 327
267// compare hash entries, used when all keys are bytestrings 328// compare hash entries, used when all keys are bytestrings
268static int 329static int
274 HE *b = *(HE **)b_; 335 HE *b = *(HE **)b_;
275 336
276 STRLEN la = HeKLEN (a); 337 STRLEN la = HeKLEN (a);
277 STRLEN lb = HeKLEN (b); 338 STRLEN lb = HeKLEN (b);
278 339
279 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 340 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
280 cmp = la < lb ? -1 : la == lb ? 0 : 1; 341 cmp = la - lb;
281 342
282 return cmp; 343 return cmp;
283} 344}
284 345
285// compare hash entries, used when some keys are sv's or utf-x 346// compare hash entries, used when some keys are sv's or utf-x
292static void 353static void
293encode_hv (enc_t *enc, HV *hv) 354encode_hv (enc_t *enc, HV *hv)
294{ 355{
295 int count, i; 356 int count, i;
296 357
358 if (enc->indent >= enc->maxdepth)
359 croak ("data structure too deep (hit recursion limit)");
360
297 encode_ch (enc, '{'); NL; ++enc->indent; 361 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
298 362
299 if ((count = hv_iterinit (hv))) 363 if ((count = hv_iterinit (hv)))
300 { 364 {
301 // for canonical output we have to sort by keys first 365 // for canonical output we have to sort by keys first
302 // actually, this is mostly due to the stupid so-called 366 // actually, this is mostly due to the stupid so-called
303 // security workaround added somewhere in 5.8.x. 367 // security workaround added somewhere in 5.8.x.
304 // that randomises hash orderings 368 // that randomises hash orderings
305 if (enc->flags & F_CANONICAL) 369 if (enc->flags & F_CANONICAL)
306 { 370 {
307 HE *he, *hes [count]; 371 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
308 int fast = 1; 372 int fast = 1;
309 373
310 i = 0; 374 i = 0;
311 while ((he = hv_iternext (hv))) 375 while ((he = hv_iternext (hv)))
312 { 376 {
337 LEAVE; 401 LEAVE;
338 } 402 }
339 403
340 for (i = 0; i < count; ++i) 404 for (i = 0; i < count; ++i)
341 { 405 {
342 INDENT; 406 encode_indent (enc);
343 encode_he (enc, hes [i]); 407 encode_he (enc, hes [i]);
344 408
345 if (i < count - 1) 409 if (i < count - 1)
346 COMMA; 410 encode_comma (enc);
347 } 411 }
348 412
349 NL; 413 encode_nl (enc);
350 } 414 }
351 else 415 else
352 { 416 {
353 SV *sv;
354 HE *he = hv_iternext (hv); 417 HE *he = hv_iternext (hv);
355 418
356 for (;;) 419 for (;;)
357 { 420 {
358 INDENT; 421 encode_indent (enc);
359 encode_he (enc, he); 422 encode_he (enc, he);
360 423
361 if (!(he = hv_iternext (hv))) 424 if (!(he = hv_iternext (hv)))
362 break; 425 break;
363 426
364 COMMA; 427 encode_comma (enc);
365 } 428 }
366 429
367 NL; 430 encode_nl (enc);
368 } 431 }
369 } 432 }
370 433
371 --enc->indent; INDENT; encode_ch (enc, '}'); 434 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
435}
436
437// encode objects, arrays and special \0=false and \1=true values.
438static void
439encode_rv (enc_t *enc, SV *sv)
440{
441 svtype svt;
442
443 SvGETMAGIC (sv);
444 svt = SvTYPE (sv);
445
446 if (svt == SVt_PVHV)
447 encode_hv (enc, (HV *)sv);
448 else if (svt == SVt_PVAV)
449 encode_av (enc, (AV *)sv);
450 else if (svt < SVt_PVAV)
451 {
452 if (SvNIOK (sv) && SvIV (sv) == 0)
453 encode_str (enc, "false", 5, 0);
454 else if (SvNIOK (sv) && SvIV (sv) == 1)
455 encode_str (enc, "true", 4, 0);
456 else
457 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
458 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
459 }
460 else
461 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
462 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
372} 463}
373 464
374static void 465static void
375encode_sv (enc_t *enc, SV *sv) 466encode_sv (enc_t *enc, SV *sv)
376{ 467{
397 SvIsUV(sv) 488 SvIsUV(sv)
398 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 489 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
399 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 490 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
400 } 491 }
401 else if (SvROK (sv)) 492 else if (SvROK (sv))
402 { 493 encode_rv (enc, SvRV (sv));
403 SV *rv = SvRV (sv);
404
405 if (!--enc->max_recurse)
406 croak ("data structure too deep (hit recursion limit)");
407
408 switch (SvTYPE (rv))
409 {
410 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
411 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
412
413 default:
414 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
415 SvPV_nolen (sv));
416 }
417 }
418 else if (!SvOK (sv)) 494 else if (!SvOK (sv))
419 encode_str (enc, "null", 4, 0); 495 encode_str (enc, "null", 4, 0);
420 else 496 else
421 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 497 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
422 SvPV_nolen (sv), SvFLAGS (sv)); 498 SvPV_nolen (sv), SvFLAGS (sv));
423} 499}
424 500
425static SV * 501static SV *
426encode_json (SV *scalar, UV flags) 502encode_json (SV *scalar, U32 flags)
427{ 503{
504 enc_t enc;
505
428 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 506 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
429 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 507 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
430 508
431 enc_t enc;
432 enc.flags = flags; 509 enc.flags = flags;
433 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 510 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
434 enc.cur = SvPVX (enc.sv); 511 enc.cur = SvPVX (enc.sv);
435 enc.end = SvEND (enc.sv); 512 enc.end = SvEND (enc.sv);
436 enc.max_recurse = 0;
437 enc.indent = 0; 513 enc.indent = 0;
514 enc.maxdepth = DEC_DEPTH (flags);
438 515
439 SvPOK_only (enc.sv); 516 SvPOK_only (enc.sv);
440 encode_sv (&enc, scalar); 517 encode_sv (&enc, scalar);
441 518
442 if (!(flags & (F_ASCII | F_UTF8))) 519 if (!(flags & (F_ASCII | F_UTF8)))
443 SvUTF8_on (enc.sv); 520 SvUTF8_on (enc.sv);
444 521
445 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 522 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
523 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
446 524
447 if (enc.flags & F_SHRINK) 525 if (enc.flags & F_SHRINK)
448 shrink (enc.sv); 526 shrink (enc.sv);
449 527
450 return enc.sv; 528 return enc.sv;
451} 529}
452 530
453///////////////////////////////////////////////////////////////////////////// 531/////////////////////////////////////////////////////////////////////////////
532// decoder
454 533
455#define WS \ 534// structure used for decoding JSON
535typedef struct
536{
537 char *cur; // current parser pointer
538 char *end; // end of input string
539 const char *err; // parse error, if != 0
540 U32 flags; // F_*
541 U32 depth; // recursion depth
542 U32 maxdepth; // recursion depth limit
543} dec_t;
544
545static void
546decode_ws (dec_t *dec)
547{
456 for (;;) \ 548 for (;;)
457 { \ 549 {
458 char ch = *dec->cur; \ 550 char ch = *dec->cur;
551
459 if (ch > 0x20 \ 552 if (ch > 0x20
460 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 553 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
461 break; \ 554 break;
555
462 ++dec->cur; \ 556 ++dec->cur;
463 } 557 }
558}
464 559
465#define ERR(reason) SB dec->err = reason; goto fail; SE 560#define ERR(reason) SB dec->err = reason; goto fail; SE
561
466#define EXPECT_CH(ch) SB \ 562#define EXPECT_CH(ch) SB \
467 if (*dec->cur != ch) \ 563 if (*dec->cur != ch) \
468 ERR (# ch " expected"); \ 564 ERR (# ch " expected"); \
469 ++dec->cur; \ 565 ++dec->cur; \
470 SE 566 SE
471 567
568#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
569#define DEC_DEC_DEPTH --dec->depth
570
472static SV *decode_sv (dec_t *dec); 571static SV *decode_sv (dec_t *dec);
473 572
474static signed char decode_hexdigit[256]; 573static signed char decode_hexdigit[256];
475 574
476static UV 575static UV
477decode_4hex (dec_t *dec) 576decode_4hex (dec_t *dec)
478{ 577{
479 signed char d1, d2, d3, d4; 578 signed char d1, d2, d3, d4;
579 unsigned char *cur = (unsigned char *)dec->cur;
480 580
481 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
482 if (d1 < 0) ERR ("four hexadecimal digits expected"); 581 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
483 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
484 if (d2 < 0) ERR ("four hexadecimal digits expected"); 582 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
485 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
486 if (d3 < 0) ERR ("four hexadecimal digits expected"); 583 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
487 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
488 if (d4 < 0) ERR ("four hexadecimal digits expected"); 584 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
489 585
490 dec->cur += 4; 586 dec->cur += 4;
491 587
492 return ((UV)d1) << 12 588 return ((UV)d1) << 12
493 | ((UV)d2) << 8 589 | ((UV)d2) << 8
496 592
497fail: 593fail:
498 return (UV)-1; 594 return (UV)-1;
499} 595}
500 596
501#define APPEND_GROW(n) SB \
502 if (cur + (n) >= end) \
503 { \
504 STRLEN ofs = cur - SvPVX (sv); \
505 SvGROW (sv, ofs + (n) + 1); \
506 cur = SvPVX (sv) + ofs; \
507 end = SvEND (sv); \
508 } \
509 SE
510
511#define APPEND_CH(ch) SB \
512 APPEND_GROW (1); \
513 *cur++ = (ch); \
514 SE
515
516static SV * 597static SV *
517decode_str (dec_t *dec) 598decode_str (dec_t *dec)
518{ 599{
519 SV *sv = NEWSV (0,2); 600 SV *sv = 0;
520 int utf8 = 0; 601 int utf8 = 0;
521 char *cur = SvPVX (sv);
522 char *end = SvEND (sv);
523 602
524 for (;;) 603 do
525 { 604 {
526 unsigned char ch = *(unsigned char *)dec->cur; 605 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
606 char *cur = buf;
527 607
528 if (ch == '"') 608 do
529 break;
530 else if (ch == '\\')
531 { 609 {
532 switch (*++dec->cur) 610 unsigned char ch = *(unsigned char *)dec->cur++;
611
612 if (ch == '"')
533 { 613 {
534 case '\\': 614 --dec->cur;
535 case '/': 615 break;
536 case '"': APPEND_CH (*dec->cur++); break; 616 }
537 617 else if (ch == '\\')
538 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 618 {
539 case 't': APPEND_CH ('\011'); ++dec->cur; break; 619 switch (*dec->cur)
540 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
541 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
542 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
543
544 case 'u':
545 { 620 {
546 UV lo, hi; 621 case '\\':
547 ++dec->cur; 622 case '/':
623 case '"': *cur++ = *dec->cur++; break;
548 624
549 hi = decode_4hex (dec); 625 case 'b': ++dec->cur; *cur++ = '\010'; break;
550 if (hi == (UV)-1) 626 case 't': ++dec->cur; *cur++ = '\011'; break;
551 goto fail; 627 case 'n': ++dec->cur; *cur++ = '\012'; break;
628 case 'f': ++dec->cur; *cur++ = '\014'; break;
629 case 'r': ++dec->cur; *cur++ = '\015'; break;
552 630
553 // possibly a surrogate pair 631 case 'u':
554 if (hi >= 0xd800 && hi < 0xdc00)
555 { 632 {
556 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 633 UV lo, hi;
557 ERR ("missing low surrogate character in surrogate pair");
558
559 dec->cur += 2; 634 ++dec->cur;
560 635
561 lo = decode_4hex (dec); 636 hi = decode_4hex (dec);
562 if (lo == (UV)-1) 637 if (hi == (UV)-1)
563 goto fail; 638 goto fail;
564 639
640 // possibly a surrogate pair
641 if (hi >= 0xd800)
642 if (hi < 0xdc00)
643 {
644 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
645 ERR ("missing low surrogate character in surrogate pair");
646
647 dec->cur += 2;
648
649 lo = decode_4hex (dec);
650 if (lo == (UV)-1)
651 goto fail;
652
565 if (lo < 0xdc00 || lo >= 0xe000) 653 if (lo < 0xdc00 || lo >= 0xe000)
566 ERR ("surrogate pair expected"); 654 ERR ("surrogate pair expected");
567 655
568 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 656 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
657 }
658 else if (hi < 0xe000)
659 ERR ("missing high surrogate character in surrogate pair");
660
661 if (hi >= 0x80)
662 {
663 utf8 = 1;
664
665 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
666 }
667 else
668 *cur++ = hi;
569 } 669 }
570 else if (hi >= 0xdc00 && hi < 0xe000)
571 ERR ("missing high surrogate character in surrogate pair");
572
573 if (hi >= 0x80)
574 { 670 break;
575 utf8 = 1;
576 671
577 APPEND_GROW (4); // at most 4 bytes for 21 bits
578 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
579 }
580 else 672 default:
581 APPEND_CH (hi); 673 --dec->cur;
674 ERR ("illegal backslash escape sequence in string");
582 } 675 }
583 break; 676 }
677 else if (ch >= 0x20 && ch <= 0x7f)
678 *cur++ = ch;
679 else if (ch >= 0x80)
680 {
681 STRLEN clen;
682 UV uch;
584 683
585 default:
586 --dec->cur; 684 --dec->cur;
587 ERR ("illegal backslash escape sequence in string"); 685
686 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
687 if (clen == (STRLEN)-1)
688 ERR ("malformed UTF-8 character in JSON string");
689
690 do
691 *cur++ = *dec->cur++;
692 while (--clen);
693
694 utf8 = 1;
695 }
696 else
697 {
698 --dec->cur;
699
700 if (!ch)
701 ERR ("unexpected end of string while parsing JSON string");
702 else
703 ERR ("invalid character encountered while parsing JSON string");
588 } 704 }
589 } 705 }
590 else if (ch >= 0x20 && ch <= 0x7f) 706 while (cur < buf + SHORT_STRING_LEN);
591 APPEND_CH (*dec->cur++); 707
592 else if (ch >= 0x80)
593 { 708 {
594 STRLEN clen; 709 STRLEN len = cur - buf;
595 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
596 if (clen == (STRLEN)-1)
597 ERR ("malformed UTF-8 character in JSON string");
598 710
599 APPEND_GROW (clen); 711 if (sv)
600 do
601 { 712 {
602 *cur++ = *dec->cur++; 713 SvGROW (sv, SvCUR (sv) + len + 1);
714 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
715 SvCUR_set (sv, SvCUR (sv) + len);
603 } 716 }
604 while (--clen);
605
606 utf8 = 1;
607 }
608 else if (dec->cur == dec->end)
609 ERR ("unexpected end of string while parsing json string");
610 else 717 else
611 ERR ("invalid character encountered"); 718 sv = newSVpvn (buf, len);
612 } 719 }
720 }
721 while (*dec->cur != '"');
613 722
614 ++dec->cur; 723 ++dec->cur;
615 724
616 SvCUR_set (sv, cur - SvPVX (sv)); 725 if (sv)
617 726 {
618 SvPOK_only (sv); 727 SvPOK_only (sv);
619 *SvEND (sv) = 0; 728 *SvEND (sv) = 0;
620 729
621 if (utf8) 730 if (utf8)
622 SvUTF8_on (sv); 731 SvUTF8_on (sv);
623 732 }
624 if (dec->flags & F_SHRINK) 733 else
625 shrink (sv); 734 sv = newSVpvn ("", 0);
626 735
627 return sv; 736 return sv;
628 737
629fail: 738fail:
630 SvREFCNT_dec (sv);
631 return 0; 739 return 0;
632} 740}
633 741
634static SV * 742static SV *
635decode_num (dec_t *dec) 743decode_num (dec_t *dec)
716static SV * 824static SV *
717decode_av (dec_t *dec) 825decode_av (dec_t *dec)
718{ 826{
719 AV *av = newAV (); 827 AV *av = newAV ();
720 828
721 WS; 829 DEC_INC_DEPTH;
830 decode_ws (dec);
831
722 if (*dec->cur == ']') 832 if (*dec->cur == ']')
723 ++dec->cur; 833 ++dec->cur;
724 else 834 else
725 for (;;) 835 for (;;)
726 { 836 {
730 if (!value) 840 if (!value)
731 goto fail; 841 goto fail;
732 842
733 av_push (av, value); 843 av_push (av, value);
734 844
735 WS; 845 decode_ws (dec);
736 846
737 if (*dec->cur == ']') 847 if (*dec->cur == ']')
738 { 848 {
739 ++dec->cur; 849 ++dec->cur;
740 break; 850 break;
744 ERR (", or ] expected while parsing array"); 854 ERR (", or ] expected while parsing array");
745 855
746 ++dec->cur; 856 ++dec->cur;
747 } 857 }
748 858
859 DEC_DEC_DEPTH;
749 return newRV_noinc ((SV *)av); 860 return newRV_noinc ((SV *)av);
750 861
751fail: 862fail:
752 SvREFCNT_dec (av); 863 SvREFCNT_dec (av);
864 DEC_DEC_DEPTH;
753 return 0; 865 return 0;
754} 866}
755 867
756static SV * 868static SV *
757decode_hv (dec_t *dec) 869decode_hv (dec_t *dec)
758{ 870{
759 HV *hv = newHV (); 871 HV *hv = newHV ();
760 872
761 WS; 873 DEC_INC_DEPTH;
874 decode_ws (dec);
875
762 if (*dec->cur == '}') 876 if (*dec->cur == '}')
763 ++dec->cur; 877 ++dec->cur;
764 else 878 else
765 for (;;) 879 for (;;)
766 { 880 {
767 SV *key, *value; 881 SV *key, *value;
768 882
769 WS; EXPECT_CH ('"'); 883 decode_ws (dec); EXPECT_CH ('"');
770 884
771 key = decode_str (dec); 885 key = decode_str (dec);
772 if (!key) 886 if (!key)
773 goto fail; 887 goto fail;
774 888
775 WS; EXPECT_CH (':'); 889 decode_ws (dec); EXPECT_CH (':');
776 890
777 value = decode_sv (dec); 891 value = decode_sv (dec);
778 if (!value) 892 if (!value)
779 { 893 {
780 SvREFCNT_dec (key); 894 SvREFCNT_dec (key);
781 goto fail; 895 goto fail;
782 } 896 }
783 897
784 //TODO: optimise
785 hv_store_ent (hv, key, value, 0); 898 hv_store_ent (hv, key, value, 0);
899 SvREFCNT_dec (key);
786 900
787 WS; 901 decode_ws (dec);
788 902
789 if (*dec->cur == '}') 903 if (*dec->cur == '}')
790 { 904 {
791 ++dec->cur; 905 ++dec->cur;
792 break; 906 break;
796 ERR (", or } expected while parsing object/hash"); 910 ERR (", or } expected while parsing object/hash");
797 911
798 ++dec->cur; 912 ++dec->cur;
799 } 913 }
800 914
915 DEC_DEC_DEPTH;
801 return newRV_noinc ((SV *)hv); 916 return newRV_noinc ((SV *)hv);
802 917
803fail: 918fail:
804 SvREFCNT_dec (hv); 919 SvREFCNT_dec (hv);
920 DEC_DEC_DEPTH;
805 return 0; 921 return 0;
806} 922}
807 923
808static SV * 924static SV *
809decode_sv (dec_t *dec) 925decode_sv (dec_t *dec)
810{ 926{
811 WS; 927 decode_ws (dec);
812 switch (*dec->cur) 928 switch (*dec->cur)
813 { 929 {
814 case '"': ++dec->cur; return decode_str (dec); 930 case '"': ++dec->cur; return decode_str (dec);
815 case '[': ++dec->cur; return decode_av (dec); 931 case '[': ++dec->cur; return decode_av (dec);
816 case '{': ++dec->cur; return decode_hv (dec); 932 case '{': ++dec->cur; return decode_hv (dec);
852 ERR ("'null' expected"); 968 ERR ("'null' expected");
853 969
854 break; 970 break;
855 971
856 default: 972 default:
857 ERR ("malformed json string, neither array, object, number, string or atom"); 973 ERR ("malformed JSON string, neither array, object, number, string or atom");
858 break; 974 break;
859 } 975 }
860 976
861fail: 977fail:
862 return 0; 978 return 0;
863} 979}
864 980
865static SV * 981static SV *
866decode_json (SV *string, UV flags) 982decode_json (SV *string, U32 flags)
867{ 983{
984 dec_t dec;
868 SV *sv; 985 SV *sv;
986
987 SvGETMAGIC (string);
988 SvUPGRADE (string, SVt_PV);
869 989
870 if (flags & F_UTF8) 990 if (flags & F_UTF8)
871 sv_utf8_downgrade (string, 0); 991 sv_utf8_downgrade (string, 0);
872 else 992 else
873 sv_utf8_upgrade (string); 993 sv_utf8_upgrade (string);
874 994
875 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 995 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
876 996
877 dec_t dec;
878 dec.flags = flags; 997 dec.flags = flags;
879 dec.cur = SvPVX (string); 998 dec.cur = SvPVX (string);
880 dec.end = SvEND (string); 999 dec.end = SvEND (string);
881 dec.err = 0; 1000 dec.err = 0;
1001 dec.depth = 0;
1002 dec.maxdepth = DEC_DEPTH (dec.flags);
882 1003
1004 *dec.end = 0; // this should basically be a nop, too, but make sure its there
883 sv = decode_sv (&dec); 1005 sv = decode_sv (&dec);
884 1006
885 if (!sv) 1007 if (!sv)
886 { 1008 {
887 IV offset = dec.flags & F_UTF8 1009 IV offset = dec.flags & F_UTF8
896 SAVEVPTR (PL_curcop); 1018 SAVEVPTR (PL_curcop);
897 PL_curcop = &cop; 1019 PL_curcop = &cop;
898 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1020 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
899 LEAVE; 1021 LEAVE;
900 1022
901 croak ("%s, at character offset %d (%s)", 1023 croak ("%s, at character offset %d [\"%s\"]",
902 dec.err, 1024 dec.err,
903 (int)offset, 1025 (int)offset,
904 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1026 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
905 } 1027 }
906 1028
910 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1032 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
911 1033
912 return sv; 1034 return sv;
913} 1035}
914 1036
1037/////////////////////////////////////////////////////////////////////////////
1038// XS interface functions
1039
915MODULE = JSON::XS PACKAGE = JSON::XS 1040MODULE = JSON::XS PACKAGE = JSON::XS
916 1041
917BOOT: 1042BOOT:
918{ 1043{
919 int i; 1044 int i;
920 1045
921 memset (decode_hexdigit, 0xff, 256); 1046 memset (decode_hexdigit, 0xff, 256);
1047
922 for (i = 10; i--; ) 1048 for (i = 0; i < 256; ++i)
923 decode_hexdigit ['0' + i] = i; 1049 decode_hexdigit [i] =
924 1050 i >= '0' && i <= '9' ? i - '0'
925 for (i = 7; i--; ) 1051 : i >= 'a' && i <= 'f' ? i - 'a' + 10
926 { 1052 : i >= 'A' && i <= 'F' ? i - 'A' + 10
927 decode_hexdigit ['a' + i] = 10 + i; 1053 : -1;
928 decode_hexdigit ['A' + i] = 10 + i;
929 }
930 1054
931 json_stash = gv_stashpv ("JSON::XS", 1); 1055 json_stash = gv_stashpv ("JSON::XS", 1);
932} 1056}
933 1057
934PROTOTYPES: DISABLE 1058PROTOTYPES: DISABLE
961 RETVAL = newSVsv (self); 1085 RETVAL = newSVsv (self);
962} 1086}
963 OUTPUT: 1087 OUTPUT:
964 RETVAL 1088 RETVAL
965 1089
1090SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1091 CODE:
1092{
1093 UV *uv = SvJSON (self);
1094 UV log2 = 0;
1095
1096 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1097
1098 while ((1UL << log2) < max_depth)
1099 ++log2;
1100
1101 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1102
1103 RETVAL = newSVsv (self);
1104}
1105 OUTPUT:
1106 RETVAL
1107
966void encode (SV *self, SV *scalar) 1108void encode (SV *self, SV *scalar)
967 PPCODE: 1109 PPCODE:
968 XPUSHs (encode_json (scalar, *SvJSON (self))); 1110 XPUSHs (encode_json (scalar, *SvJSON (self)));
969 1111
970void decode (SV *self, SV *jsonstr) 1112void decode (SV *self, SV *jsonstr)
972 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1114 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
973 1115
974PROTOTYPES: ENABLE 1116PROTOTYPES: ENABLE
975 1117
976void to_json (SV *scalar) 1118void to_json (SV *scalar)
1119 ALIAS:
1120 objToJson = 0
977 PPCODE: 1121 PPCODE:
978 XPUSHs (encode_json (scalar, F_UTF8)); 1122 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
979 1123
980void from_json (SV *jsonstr) 1124void from_json (SV *jsonstr)
1125 ALIAS:
1126 jsonToObj = 0
981 PPCODE: 1127 PPCODE:
982 XPUSHs (decode_json (jsonstr, F_UTF8)); 1128 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
983 1129

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines