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 1.7 by root, Fri Mar 23 15:57:18 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_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 20#define F_ALLOW_NONREF 0x00000080UL
17#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? <=> { $__class__$ => }
18 29
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 30#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 31#define F_DEFAULT (9UL << S_MAXDEPTH)
21 32
22#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
23 37
24#define SB do { 38#define SB do {
25#define SE } while (0) 39#define SE } while (0)
26 40
27static 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
28 93
29// structure used for encoding JSON 94// structure used for encoding JSON
30typedef struct 95typedef struct
31{ 96{
32 char *cur; 97 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 98 char *end; // SvEND (sv)
35 SV *sv; 99 SV *sv; // result scalar
36 UV flags; 100 U32 flags; // F_*
37 int max_recurse; 101 U32 indent; // indentation level
38 int indent; 102 U32 maxdepth; // max. indentation/recursion level
39} enc_t; 103} enc_t;
40
41// structure used for decoding JSON
42typedef struct
43{
44 char *cur;
45 char *end;
46 const char *err;
47 UV flags;
48} dec_t;
49
50static UV *
51SvJSON (SV *sv)
52{
53 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
54 croak ("object is not of type JSON::XS");
55
56 return &SvUVX (SvRV (sv));
57}
58
59static void
60shrink (SV *sv)
61{
62 sv_utf8_downgrade (sv, 1);
63#ifdef SvPV_shrink_to_cur
64 SvPV_shrink_to_cur (sv);
65#endif
66}
67
68/////////////////////////////////////////////////////////////////////////////
69 104
70static void 105static void
71need (enc_t *enc, STRLEN len) 106need (enc_t *enc, STRLEN len)
72{ 107{
73 if (enc->cur + len >= enc->end) 108 if (enc->cur + len >= enc->end)
74 { 109 {
75 STRLEN cur = enc->cur - SvPVX (enc->sv); 110 STRLEN cur = enc->cur - SvPVX (enc->sv);
76 SvGROW (enc->sv, cur + len + 1); 111 SvGROW (enc->sv, cur + len + 1);
77 enc->cur = SvPVX (enc->sv) + cur; 112 enc->cur = SvPVX (enc->sv) + cur;
78 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 113 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
79 } 114 }
80} 115}
81 116
82static void 117static void
83encode_ch (enc_t *enc, char ch) 118encode_ch (enc_t *enc, char ch)
… …
131 STRLEN clen; 166 STRLEN clen;
132 UV uch; 167 UV uch;
133 168
134 if (is_utf8) 169 if (is_utf8)
135 { 170 {
136 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);
137 if (clen == (STRLEN)-1) 173 if (clen == (STRLEN)-1)
138 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 174 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
139 } 175 }
140 else 176 else
141 { 177 {
142 uch = ch; 178 uch = ch;
143 clen = 1; 179 clen = 1;
144 } 180 }
145 181
182 if (uch > 0x10FFFFUL)
183 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
184
146 if (uch < 0x80 || enc->flags & F_ASCII) 185 if (uch < 0x80 || enc->flags & F_ASCII)
147 { 186 {
148 if (uch > 0xFFFFUL) 187 if (uch > 0xFFFFUL)
149 { 188 {
150 need (enc, len += 11); 189 need (enc, len += 11);
151 sprintf (enc->cur, "\\u%04x\\u%04x", 190 sprintf (enc->cur, "\\u%04x\\u%04x",
152 (uch - 0x10000) / 0x400 + 0xD800, 191 (int)((uch - 0x10000) / 0x400 + 0xD800),
153 (uch - 0x10000) % 0x400 + 0xDC00); 192 (int)((uch - 0x10000) % 0x400 + 0xDC00));
154 enc->cur += 12; 193 enc->cur += 12;
155 } 194 }
156 else 195 else
157 { 196 {
158 static char hexdigit [16] = "0123456789abcdef"; 197 static char hexdigit [16] = "0123456789abcdef";
… …
176 } 215 }
177 while (--clen); 216 while (--clen);
178 } 217 }
179 else 218 else
180 { 219 {
181 need (enc, len += 10); // never more than 11 bytes needed 220 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
182 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 221 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
183 ++str; 222 ++str;
184 } 223 }
185 } 224 }
186 } 225 }
… …
188 227
189 --len; 228 --len;
190 } 229 }
191} 230}
192 231
193#define INDENT SB \ 232static void
233encode_indent (enc_t *enc)
234{
194 if (enc->flags & F_INDENT) \ 235 if (enc->flags & F_INDENT)
195 { \ 236 {
196 int i_; \ 237 int spaces = enc->indent * INDENT_STEP;
197 need (enc, enc->indent); \
198 for (i_ = enc->indent * 3; i_--; )\
199 encode_ch (enc, ' '); \
200 } \
201 SE
202 238
203#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 239 need (enc, spaces);
204#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 240 memset (enc->cur, ' ', spaces);
205#define COMMA SB \ 241 enc->cur += spaces;
242 }
243}
244
245static void
246encode_space (enc_t *enc)
247{
248 need (enc, 1);
206 encode_ch (enc, ','); \ 249 encode_ch (enc, ' ');
250}
251
252static void
253encode_nl (enc_t *enc)
254{
207 if (enc->flags & F_INDENT) \ 255 if (enc->flags & F_INDENT)
208 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);
209 else if (enc->flags & F_SPACE_AFTER) \ 269 else if (enc->flags & F_SPACE_AFTER)
210 SPACE; \ 270 encode_space (enc);
211 SE 271}
212 272
213static void encode_sv (enc_t *enc, SV *sv); 273static void encode_sv (enc_t *enc, SV *sv);
214 274
215static void 275static void
216encode_av (enc_t *enc, AV *av) 276encode_av (enc_t *enc, AV *av)
217{ 277{
218 int i, len = av_len (av); 278 int i, len = av_len (av);
219 279
220 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);
221 ++enc->indent; 284 ++enc->indent;
222 285
223 for (i = 0; i <= len; ++i) 286 for (i = 0; i <= len; ++i)
224 { 287 {
225 INDENT; 288 encode_indent (enc);
226 encode_sv (enc, *av_fetch (av, i, 0)); 289 encode_sv (enc, *av_fetch (av, i, 0));
227 290
228 if (i < len) 291 if (i < len)
229 COMMA; 292 encode_comma (enc);
230 } 293 }
231 294
232 NL; 295 encode_nl (enc);
233 296
234 --enc->indent; 297 --enc->indent;
235 INDENT; encode_ch (enc, ']'); 298 encode_indent (enc); encode_ch (enc, ']');
236} 299}
237 300
238static void 301static void
239encode_he (enc_t *enc, HE *he) 302encode_he (enc_t *enc, HE *he)
240{ 303{
… …
254 else 317 else
255 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 318 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
256 319
257 encode_ch (enc, '"'); 320 encode_ch (enc, '"');
258 321
259 if (enc->flags & F_SPACE_BEFORE) SPACE; 322 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
260 encode_ch (enc, ':'); 323 encode_ch (enc, ':');
261 if (enc->flags & F_SPACE_AFTER ) SPACE; 324 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
262 encode_sv (enc, HeVAL (he)); 325 encode_sv (enc, HeVAL (he));
263} 326}
264 327
265// compare hash entries, used when all keys are bytestrings 328// compare hash entries, used when all keys are bytestrings
266static int 329static int
… …
272 HE *b = *(HE **)b_; 335 HE *b = *(HE **)b_;
273 336
274 STRLEN la = HeKLEN (a); 337 STRLEN la = HeKLEN (a);
275 STRLEN lb = HeKLEN (b); 338 STRLEN lb = HeKLEN (b);
276 339
277 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 340 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
278 cmp = la < lb ? -1 : la == lb ? 0 : 1; 341 cmp = la - lb;
279 342
280 return cmp; 343 return cmp;
281} 344}
282 345
283// 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
… …
290static void 353static void
291encode_hv (enc_t *enc, HV *hv) 354encode_hv (enc_t *enc, HV *hv)
292{ 355{
293 int count, i; 356 int count, i;
294 357
358 if (enc->indent >= enc->maxdepth)
359 croak ("data structure too deep (hit recursion limit)");
360
295 encode_ch (enc, '{'); NL; ++enc->indent; 361 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
296 362
297 if ((count = hv_iterinit (hv))) 363 if ((count = hv_iterinit (hv)))
298 { 364 {
299 // for canonical output we have to sort by keys first 365 // for canonical output we have to sort by keys first
300 // actually, this is mostly due to the stupid so-called 366 // actually, this is mostly due to the stupid so-called
301 // security workaround added somewhere in 5.8.x. 367 // security workaround added somewhere in 5.8.x.
302 // that randomises hash orderings 368 // that randomises hash orderings
303 if (enc->flags & F_CANONICAL) 369 if (enc->flags & F_CANONICAL)
304 { 370 {
305 HE *he, *hes [count]; 371 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
306 int fast = 1; 372 int fast = 1;
307 373
308 i = 0; 374 i = 0;
309 while ((he = hv_iternext (hv))) 375 while ((he = hv_iternext (hv)))
310 { 376 {
… …
317 383
318 if (fast) 384 if (fast)
319 qsort (hes, count, sizeof (HE *), he_cmp_fast); 385 qsort (hes, count, sizeof (HE *), he_cmp_fast);
320 else 386 else
321 { 387 {
322 // hack to disable "use bytes" 388 // hack to forcefully disable "use bytes"
323 COP *oldcop = PL_curcop, cop; 389 COP cop = *PL_curcop;
324 cop.op_private = 0; 390 cop.op_private = 0;
391
392 ENTER;
393 SAVETMPS;
394
395 SAVEVPTR (PL_curcop);
325 PL_curcop = &cop; 396 PL_curcop = &cop;
326 397
327 SAVETMPS;
328 qsort (hes, count, sizeof (HE *), he_cmp_slow); 398 qsort (hes, count, sizeof (HE *), he_cmp_slow);
399
329 FREETMPS; 400 FREETMPS;
330 401 LEAVE;
331 PL_curcop = oldcop;
332 } 402 }
333 403
334 for (i = 0; i < count; ++i) 404 for (i = 0; i < count; ++i)
335 { 405 {
336 INDENT; 406 encode_indent (enc);
337 encode_he (enc, hes [i]); 407 encode_he (enc, hes [i]);
338 408
339 if (i < count - 1) 409 if (i < count - 1)
340 COMMA; 410 encode_comma (enc);
341 } 411 }
342 412
343 NL; 413 encode_nl (enc);
344 } 414 }
345 else 415 else
346 { 416 {
347 SV *sv;
348 HE *he = hv_iternext (hv); 417 HE *he = hv_iternext (hv);
349 418
350 for (;;) 419 for (;;)
351 { 420 {
352 INDENT; 421 encode_indent (enc);
353 encode_he (enc, he); 422 encode_he (enc, he);
354 423
355 if (!(he = hv_iternext (hv))) 424 if (!(he = hv_iternext (hv)))
356 break; 425 break;
357 426
358 COMMA; 427 encode_comma (enc);
359 } 428 }
360 429
361 NL; 430 encode_nl (enc);
362 } 431 }
363 } 432 }
364 433
365 --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))));
366} 463}
367 464
368static void 465static void
369encode_sv (enc_t *enc, SV *sv) 466encode_sv (enc_t *enc, SV *sv)
370{ 467{
… …
391 SvIsUV(sv) 488 SvIsUV(sv)
392 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 489 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
393 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 490 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
394 } 491 }
395 else if (SvROK (sv)) 492 else if (SvROK (sv))
396 { 493 encode_rv (enc, SvRV (sv));
397 if (!--enc->max_recurse)
398 croak ("data structure too deep (hit recursion limit)");
399
400 sv = SvRV (sv);
401
402 switch (SvTYPE (sv))
403 {
404 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
405 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
406
407 default:
408 croak ("JSON can only represent references to arrays or hashes");
409 }
410 }
411 else if (!SvOK (sv)) 494 else if (!SvOK (sv))
412 encode_str (enc, "null", 4, 0); 495 encode_str (enc, "null", 4, 0);
413 else 496 else
414 croak ("encountered perl type that JSON cannot handle"); 497 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
498 SvPV_nolen (sv), SvFLAGS (sv));
415} 499}
416 500
417static SV * 501static SV *
418encode_json (SV *scalar, UV flags) 502encode_json (SV *scalar, U32 flags)
419{ 503{
504 enc_t enc;
505
420 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 506 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
421 croak ("hash- or arraref required (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)");
422 508
423 enc_t enc;
424 enc.flags = flags; 509 enc.flags = flags;
425 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 510 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
426 enc.cur = SvPVX (enc.sv); 511 enc.cur = SvPVX (enc.sv);
427 enc.end = SvEND (enc.sv); 512 enc.end = SvEND (enc.sv);
428 enc.max_recurse = 0;
429 enc.indent = 0; 513 enc.indent = 0;
514 enc.maxdepth = DEC_DEPTH (flags);
430 515
431 SvPOK_only (enc.sv); 516 SvPOK_only (enc.sv);
432 encode_sv (&enc, scalar); 517 encode_sv (&enc, scalar);
433 518
434 if (!(flags & (F_ASCII | F_UTF8))) 519 if (!(flags & (F_ASCII | F_UTF8)))
435 SvUTF8_on (enc.sv); 520 SvUTF8_on (enc.sv);
436 521
437 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
438 524
439 if (enc.flags & F_SHRINK) 525 if (enc.flags & F_SHRINK)
440 shrink (enc.sv); 526 shrink (enc.sv);
441 527
442 return enc.sv; 528 return enc.sv;
443} 529}
444 530
445///////////////////////////////////////////////////////////////////////////// 531/////////////////////////////////////////////////////////////////////////////
532// decoder
446 533
447#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{
448 for (;;) \ 548 for (;;)
449 { \ 549 {
450 char ch = *dec->cur; \ 550 char ch = *dec->cur;
551
451 if (ch > 0x20 \ 552 if (ch > 0x20
452 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 553 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
453 break; \ 554 break;
555
454 ++dec->cur; \ 556 ++dec->cur;
455 } 557 }
558}
456 559
457#define ERR(reason) SB dec->err = reason; goto fail; SE 560#define ERR(reason) SB dec->err = reason; goto fail; SE
561
458#define EXPECT_CH(ch) SB \ 562#define EXPECT_CH(ch) SB \
459 if (*dec->cur != ch) \ 563 if (*dec->cur != ch) \
460 ERR (# ch " expected"); \ 564 ERR (# ch " expected"); \
461 ++dec->cur; \ 565 ++dec->cur; \
462 SE 566 SE
463 567
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
464static SV *decode_sv (dec_t *dec); 571static SV *decode_sv (dec_t *dec);
465 572
466static signed char decode_hexdigit[256]; 573static signed char decode_hexdigit[256];
467 574
468static UV 575static UV
469decode_4hex (dec_t *dec) 576decode_4hex (dec_t *dec)
470{ 577{
471 signed char d1, d2, d3, d4; 578 signed char d1, d2, d3, d4;
579 unsigned char *cur = (unsigned char *)dec->cur;
472 580
473 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
474 if (d1 < 0) ERR ("four hexadecimal digits expected"); 581 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
475 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
476 if (d2 < 0) ERR ("four hexadecimal digits expected"); 582 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
477 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
478 if (d3 < 0) ERR ("four hexadecimal digits expected"); 583 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
479 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
480 if (d4 < 0) ERR ("four hexadecimal digits expected"); 584 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
481 585
482 dec->cur += 4; 586 dec->cur += 4;
483 587
484 return ((UV)d1) << 12 588 return ((UV)d1) << 12
485 | ((UV)d2) << 8 589 | ((UV)d2) << 8
… …
488 592
489fail: 593fail:
490 return (UV)-1; 594 return (UV)-1;
491} 595}
492 596
493#define APPEND_GROW(n) SB \
494 if (cur + (n) >= end) \
495 { \
496 STRLEN ofs = cur - SvPVX (sv); \
497 SvGROW (sv, ofs + (n) + 1); \
498 cur = SvPVX (sv) + ofs; \
499 end = SvEND (sv); \
500 } \
501 SE
502
503#define APPEND_CH(ch) SB \
504 APPEND_GROW (1); \
505 *cur++ = (ch); \
506 SE
507
508static SV * 597static SV *
509decode_str (dec_t *dec) 598decode_str (dec_t *dec)
510{ 599{
511 SV *sv = NEWSV (0,2); 600 SV *sv = 0;
512 int utf8 = 0; 601 int utf8 = 0;
513 char *cur = SvPVX (sv);
514 char *end = SvEND (sv);
515 602
516 for (;;) 603 do
517 { 604 {
518 unsigned char ch = *(unsigned char *)dec->cur; 605 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
606 char *cur = buf;
519 607
520 if (ch == '"') 608 do
521 break;
522 else if (ch == '\\')
523 { 609 {
524 switch (*++dec->cur) 610 unsigned char ch = *(unsigned char *)dec->cur++;
611
612 if (ch == '"')
525 { 613 {
526 case '\\': 614 --dec->cur;
527 case '/': 615 break;
528 case '"': APPEND_CH (*dec->cur++); break; 616 }
529 617 else if (ch == '\\')
530 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 618 {
531 case 't': APPEND_CH ('\011'); ++dec->cur; break; 619 switch (*dec->cur)
532 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
533 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
534 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
535
536 case 'u':
537 { 620 {
538 UV lo, hi; 621 case '\\':
539 ++dec->cur; 622 case '/':
623 case '"': *cur++ = *dec->cur++; break;
540 624
541 hi = decode_4hex (dec); 625 case 'b': ++dec->cur; *cur++ = '\010'; break;
542 if (hi == (UV)-1) 626 case 't': ++dec->cur; *cur++ = '\011'; break;
543 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;
544 630
545 // possibly a surrogate pair 631 case 'u':
546 if (hi >= 0xd800 && hi < 0xdc00)
547 { 632 {
548 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 633 UV lo, hi;
549 ERR ("missing low surrogate character in surrogate pair");
550
551 dec->cur += 2; 634 ++dec->cur;
552 635
553 lo = decode_4hex (dec); 636 hi = decode_4hex (dec);
554 if (lo == (UV)-1) 637 if (hi == (UV)-1)
555 goto fail; 638 goto fail;
556 639
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
557 if (lo < 0xdc00 || lo >= 0xe000) 653 if (lo < 0xdc00 || lo >= 0xe000)
558 ERR ("surrogate pair expected"); 654 ERR ("surrogate pair expected");
559 655
560 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;
561 } 669 }
562 else if (hi >= 0xdc00 && hi < 0xe000)
563 ERR ("missing high surrogate character in surrogate pair");
564
565 if (hi >= 0x80)
566 { 670 break;
567 utf8 = 1;
568 671
569 APPEND_GROW (4); // at most 4 bytes for 21 bits
570 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
571 }
572 else 672 default:
573 APPEND_CH (hi); 673 --dec->cur;
674 ERR ("illegal backslash escape sequence in string");
574 } 675 }
575 break; 676 }
677 else if (ch >= 0x20 && ch <= 0x7f)
678 *cur++ = ch;
679 else if (ch >= 0x80)
680 {
681 STRLEN clen;
682 UV uch;
576 683
577 default:
578 --dec->cur; 684 --dec->cur;
579 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");
580 } 704 }
581 } 705 }
582 else if (ch >= 0x20 && ch <= 0x7f) 706 while (cur < buf + SHORT_STRING_LEN);
583 APPEND_CH (*dec->cur++); 707
584 else if (ch >= 0x80)
585 { 708 {
586 STRLEN clen; 709 STRLEN len = cur - buf;
587 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
588 if (clen == (STRLEN)-1)
589 ERR ("malformed UTF-8 character in JSON string");
590 710
591 APPEND_GROW (clen); 711 if (sv)
592 do
593 { 712 {
594 *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);
595 } 716 }
596 while (--clen);
597
598 utf8 = 1;
599 }
600 else if (dec->cur == dec->end)
601 ERR ("unexpected end of string while parsing json string");
602 else 717 else
603 ERR ("invalid character encountered"); 718 sv = newSVpvn (buf, len);
604 } 719 }
720 }
721 while (*dec->cur != '"');
605 722
606 ++dec->cur; 723 ++dec->cur;
607 724
608 SvCUR_set (sv, cur - SvPVX (sv)); 725 if (sv)
609 726 {
610 SvPOK_only (sv); 727 SvPOK_only (sv);
611 *SvEND (sv) = 0; 728 *SvEND (sv) = 0;
612 729
613 if (utf8) 730 if (utf8)
614 SvUTF8_on (sv); 731 SvUTF8_on (sv);
615 732 }
616 if (dec->flags & F_SHRINK) 733 else
617 shrink (sv); 734 sv = newSVpvn ("", 0);
618 735
619 return sv; 736 return sv;
620 737
621fail: 738fail:
622 SvREFCNT_dec (sv);
623 return 0; 739 return 0;
624} 740}
625 741
626static SV * 742static SV *
627decode_num (dec_t *dec) 743decode_num (dec_t *dec)
… …
708static SV * 824static SV *
709decode_av (dec_t *dec) 825decode_av (dec_t *dec)
710{ 826{
711 AV *av = newAV (); 827 AV *av = newAV ();
712 828
713 WS; 829 DEC_INC_DEPTH;
830 decode_ws (dec);
831
714 if (*dec->cur == ']') 832 if (*dec->cur == ']')
715 ++dec->cur; 833 ++dec->cur;
716 else 834 else
717 for (;;) 835 for (;;)
718 { 836 {
… …
722 if (!value) 840 if (!value)
723 goto fail; 841 goto fail;
724 842
725 av_push (av, value); 843 av_push (av, value);
726 844
727 WS; 845 decode_ws (dec);
728 846
729 if (*dec->cur == ']') 847 if (*dec->cur == ']')
730 { 848 {
731 ++dec->cur; 849 ++dec->cur;
732 break; 850 break;
… …
736 ERR (", or ] expected while parsing array"); 854 ERR (", or ] expected while parsing array");
737 855
738 ++dec->cur; 856 ++dec->cur;
739 } 857 }
740 858
859 DEC_DEC_DEPTH;
741 return newRV_noinc ((SV *)av); 860 return newRV_noinc ((SV *)av);
742 861
743fail: 862fail:
744 SvREFCNT_dec (av); 863 SvREFCNT_dec (av);
864 DEC_DEC_DEPTH;
745 return 0; 865 return 0;
746} 866}
747 867
748static SV * 868static SV *
749decode_hv (dec_t *dec) 869decode_hv (dec_t *dec)
750{ 870{
751 HV *hv = newHV (); 871 HV *hv = newHV ();
752 872
753 WS; 873 DEC_INC_DEPTH;
874 decode_ws (dec);
875
754 if (*dec->cur == '}') 876 if (*dec->cur == '}')
755 ++dec->cur; 877 ++dec->cur;
756 else 878 else
757 for (;;) 879 for (;;)
758 { 880 {
759 SV *key, *value; 881 SV *key, *value;
760 882
761 WS; EXPECT_CH ('"'); 883 decode_ws (dec); EXPECT_CH ('"');
762 884
763 key = decode_str (dec); 885 key = decode_str (dec);
764 if (!key) 886 if (!key)
765 goto fail; 887 goto fail;
766 888
767 WS; EXPECT_CH (':'); 889 decode_ws (dec); EXPECT_CH (':');
768 890
769 value = decode_sv (dec); 891 value = decode_sv (dec);
770 if (!value) 892 if (!value)
771 { 893 {
772 SvREFCNT_dec (key); 894 SvREFCNT_dec (key);
773 goto fail; 895 goto fail;
774 } 896 }
775 897
776 //TODO: optimise
777 hv_store_ent (hv, key, value, 0); 898 hv_store_ent (hv, key, value, 0);
899 SvREFCNT_dec (key);
778 900
779 WS; 901 decode_ws (dec);
780 902
781 if (*dec->cur == '}') 903 if (*dec->cur == '}')
782 { 904 {
783 ++dec->cur; 905 ++dec->cur;
784 break; 906 break;
… …
788 ERR (", or } expected while parsing object/hash"); 910 ERR (", or } expected while parsing object/hash");
789 911
790 ++dec->cur; 912 ++dec->cur;
791 } 913 }
792 914
915 DEC_DEC_DEPTH;
793 return newRV_noinc ((SV *)hv); 916 return newRV_noinc ((SV *)hv);
794 917
795fail: 918fail:
796 SvREFCNT_dec (hv); 919 SvREFCNT_dec (hv);
920 DEC_DEC_DEPTH;
797 return 0; 921 return 0;
798} 922}
799 923
800static SV * 924static SV *
801decode_sv (dec_t *dec) 925decode_sv (dec_t *dec)
802{ 926{
803 WS; 927 decode_ws (dec);
804 switch (*dec->cur) 928 switch (*dec->cur)
805 { 929 {
806 case '"': ++dec->cur; return decode_str (dec); 930 case '"': ++dec->cur; return decode_str (dec);
807 case '[': ++dec->cur; return decode_av (dec); 931 case '[': ++dec->cur; return decode_av (dec);
808 case '{': ++dec->cur; return decode_hv (dec); 932 case '{': ++dec->cur; return decode_hv (dec);
… …
844 ERR ("'null' expected"); 968 ERR ("'null' expected");
845 969
846 break; 970 break;
847 971
848 default: 972 default:
849 ERR ("malformed json string, neither array, object, number, string or atom"); 973 ERR ("malformed JSON string, neither array, object, number, string or atom");
850 break; 974 break;
851 } 975 }
852 976
853fail: 977fail:
854 return 0; 978 return 0;
855} 979}
856 980
857static SV * 981static SV *
858decode_json (SV *string, UV flags) 982decode_json (SV *string, U32 flags)
859{ 983{
984 dec_t dec;
860 SV *sv; 985 SV *sv;
986
987 SvGETMAGIC (string);
988 SvUPGRADE (string, SVt_PV);
861 989
862 if (flags & F_UTF8) 990 if (flags & F_UTF8)
863 sv_utf8_downgrade (string, 0); 991 sv_utf8_downgrade (string, 0);
864 else 992 else
865 sv_utf8_upgrade (string); 993 sv_utf8_upgrade (string);
866 994
867 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 995 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
868 996
869 dec_t dec;
870 dec.flags = flags; 997 dec.flags = flags;
871 dec.cur = SvPVX (string); 998 dec.cur = SvPVX (string);
872 dec.end = SvEND (string); 999 dec.end = SvEND (string);
873 dec.err = 0; 1000 dec.err = 0;
1001 dec.depth = 0;
1002 dec.maxdepth = DEC_DEPTH (dec.flags);
874 1003
1004 *dec.end = 0; // this should basically be a nop, too, but make sure its there
875 sv = decode_sv (&dec); 1005 sv = decode_sv (&dec);
876 1006
877 if (!sv) 1007 if (!sv)
878 { 1008 {
879 IV offset = dec.flags & F_UTF8 1009 IV offset = dec.flags & F_UTF8
880 ? dec.cur - SvPVX (string) 1010 ? dec.cur - SvPVX (string)
881 : utf8_distance (dec.cur, SvPVX (string)); 1011 : utf8_distance (dec.cur, SvPVX (string));
882 SV *uni = sv_newmortal (); 1012 SV *uni = sv_newmortal ();
1013
883 // horrible hack to silence warning inside pv_uni_display 1014 // horrible hack to silence warning inside pv_uni_display
884 COP cop; 1015 COP cop = *PL_curcop;
885 memset (&cop, 0, sizeof (cop));
886 cop.cop_warnings = pWARN_NONE; 1016 cop.cop_warnings = pWARN_NONE;
1017 ENTER;
887 SAVEVPTR (PL_curcop); 1018 SAVEVPTR (PL_curcop);
888 PL_curcop = &cop; 1019 PL_curcop = &cop;
889
890 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);
1021 LEAVE;
1022
891 croak ("%s, at character offset %d (%s)", 1023 croak ("%s, at character offset %d [\"%s\"]",
892 dec.err, 1024 dec.err,
893 (int)offset, 1025 (int)offset,
894 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1026 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
895 } 1027 }
896 1028
897 sv = sv_2mortal (sv); 1029 sv = sv_2mortal (sv);
898 1030
899 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1031 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
900 croak ("JSON object or array expected (but number, string, true, false or null found, 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)");
901 1033
902 return sv; 1034 return sv;
903} 1035}
904 1036
1037/////////////////////////////////////////////////////////////////////////////
1038// XS interface functions
1039
905MODULE = JSON::XS PACKAGE = JSON::XS 1040MODULE = JSON::XS PACKAGE = JSON::XS
906 1041
907BOOT: 1042BOOT:
908{ 1043{
909 int i; 1044 int i;
910 1045
911 memset (decode_hexdigit, 0xff, 256); 1046 memset (decode_hexdigit, 0xff, 256);
1047
912 for (i = 10; i--; ) 1048 for (i = 0; i < 256; ++i)
913 decode_hexdigit ['0' + i] = i; 1049 decode_hexdigit [i] =
914 1050 i >= '0' && i <= '9' ? i - '0'
915 for (i = 7; i--; ) 1051 : i >= 'a' && i <= 'f' ? i - 'a' + 10
916 { 1052 : i >= 'A' && i <= 'F' ? i - 'A' + 10
917 decode_hexdigit ['a' + i] = 10 + i; 1053 : -1;
918 decode_hexdigit ['A' + i] = 10 + i;
919 }
920 1054
921 json_stash = gv_stashpv ("JSON::XS", 1); 1055 json_stash = gv_stashpv ("JSON::XS", 1);
922} 1056}
923 1057
924PROTOTYPES: DISABLE 1058PROTOTYPES: DISABLE
… …
935 utf8 = F_UTF8 1069 utf8 = F_UTF8
936 indent = F_INDENT 1070 indent = F_INDENT
937 canonical = F_CANONICAL 1071 canonical = F_CANONICAL
938 space_before = F_SPACE_BEFORE 1072 space_before = F_SPACE_BEFORE
939 space_after = F_SPACE_AFTER 1073 space_after = F_SPACE_AFTER
940 json_rpc = F_JSON_RPC
941 pretty = F_PRETTY 1074 pretty = F_PRETTY
942 allow_nonref = F_ALLOW_NONREF 1075 allow_nonref = F_ALLOW_NONREF
943 shrink = F_SHRINK 1076 shrink = F_SHRINK
944 CODE: 1077 CODE:
945{ 1078{
… …
952 RETVAL = newSVsv (self); 1085 RETVAL = newSVsv (self);
953} 1086}
954 OUTPUT: 1087 OUTPUT:
955 RETVAL 1088 RETVAL
956 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
957void encode (SV *self, SV *scalar) 1108void encode (SV *self, SV *scalar)
958 PPCODE: 1109 PPCODE:
959 XPUSHs (encode_json (scalar, *SvJSON (self))); 1110 XPUSHs (encode_json (scalar, *SvJSON (self)));
960 1111
961void decode (SV *self, SV *jsonstr) 1112void decode (SV *self, SV *jsonstr)
… …
963 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1114 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
964 1115
965PROTOTYPES: ENABLE 1116PROTOTYPES: ENABLE
966 1117
967void to_json (SV *scalar) 1118void to_json (SV *scalar)
1119 ALIAS:
1120 objToJson = 0
968 PPCODE: 1121 PPCODE:
969 XPUSHs (encode_json (scalar, F_UTF8)); 1122 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
970 1123
971void from_json (SV *jsonstr) 1124void from_json (SV *jsonstr)
1125 ALIAS:
1126 jsonToObj = 0
972 PPCODE: 1127 PPCODE:
973 XPUSHs (decode_json (jsonstr, F_UTF8)); 1128 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
974 1129

Diff Legend

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