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 1.9 by root, Fri Mar 23 17:40:29 2007 UTC vs.
Revision 1.32 by root, Wed May 9 16:41:12 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
15#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 16#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 17#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 18#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 19#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 20#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 21#define F_ALLOW_NONREF 0x00000100UL
17#define F_SHRINK 0x00000100 22#define F_SHRINK 0x00000200UL
23#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27
25
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
18 30
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 32#define F_DEFAULT (9UL << S_MAXDEPTH)
21 33
22#define INIT_SIZE 32 // initial scalar size to be allocated 34#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level
36
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size
23 38
24#define SB do { 39#define SB do {
25#define SE } while (0) 40#define SE } while (0)
26 41
27static HV *json_stash; 42static HV *json_stash; // JSON::XS::
43
44/////////////////////////////////////////////////////////////////////////////
45// utility functions
46
47static UV *
48SvJSON (SV *sv)
49{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
51 croak ("object is not of type JSON::XS");
52
53 return &SvUVX (SvRV (sv));
54}
55
56static void
57shrink (SV *sv)
58{
59 sv_utf8_downgrade (sv, 1);
60 if (SvLEN (sv) > SvCUR (sv) + 1)
61 {
62#ifdef SvPV_shrink_to_cur
63 SvPV_shrink_to_cur (sv);
64#elif defined (SvPV_renew)
65 SvPV_renew (sv, SvCUR (sv) + 1);
66#endif
67 }
68}
69
70// decode an utf-8 character and return it, or (UV)-1 in
71// case of an error.
72// we special-case "safe" characters from U+80 .. U+7FF,
73// but use the very good perl function to parse anything else.
74// note that we never call this function for a ascii codepoints
75static UV
76decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
77{
78 if (s[0] > 0xdf || s[0] < 0xc2)
79 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
80 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
81 {
82 *clen = 2;
83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
84 }
85 else
86 {
87 *clen = (STRLEN)-1;
88 return (UV)-1;
89 }
90}
91
92/////////////////////////////////////////////////////////////////////////////
93// encoder
28 94
29// structure used for encoding JSON 95// structure used for encoding JSON
30typedef struct 96typedef struct
31{ 97{
32 char *cur; 98 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 99 char *end; // SvEND (sv)
35 SV *sv; 100 SV *sv; // result scalar
36 UV flags; 101 U32 flags; // F_*
37 int max_recurse; 102 U32 indent; // indentation level
38 int indent; 103 U32 maxdepth; // max. indentation/recursion level
39} enc_t; 104} 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/////////////////////////////////////////////////////////////////////////////
69 105
70static void 106static void
71need (enc_t *enc, STRLEN len) 107need (enc_t *enc, STRLEN len)
72{ 108{
73 if (enc->cur + len >= enc->end) 109 if (enc->cur + len >= enc->end)
74 { 110 {
75 STRLEN cur = enc->cur - SvPVX (enc->sv); 111 STRLEN cur = enc->cur - SvPVX (enc->sv);
76 SvGROW (enc->sv, cur + len + 1); 112 SvGROW (enc->sv, cur + len + 1);
77 enc->cur = SvPVX (enc->sv) + cur; 113 enc->cur = SvPVX (enc->sv) + cur;
78 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
79 } 115 }
80} 116}
81 117
82static void 118static void
83encode_ch (enc_t *enc, char ch) 119encode_ch (enc_t *enc, char ch)
… …
131 STRLEN clen; 167 STRLEN clen;
132 UV uch; 168 UV uch;
133 169
134 if (is_utf8) 170 if (is_utf8)
135 { 171 {
136 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 172 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173 uch = decode_utf8 (str, end - str, &clen);
137 if (clen == (STRLEN)-1) 174 if (clen == (STRLEN)-1)
138 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 175 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
139 } 176 }
140 else 177 else
141 { 178 {
… …
144 } 181 }
145 182
146 if (uch > 0x10FFFFUL) 183 if (uch > 0x10FFFFUL)
147 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 184 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
148 185
149 if (uch < 0x80 || enc->flags & F_ASCII) 186 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
150 { 187 {
151 if (uch > 0xFFFFUL) 188 if (uch > 0xFFFFUL)
152 { 189 {
153 need (enc, len += 11); 190 need (enc, len += 11);
154 sprintf (enc->cur, "\\u%04x\\u%04x", 191 sprintf (enc->cur, "\\u%04x\\u%04x",
155 (uch - 0x10000) / 0x400 + 0xD800, 192 (int)((uch - 0x10000) / 0x400 + 0xD800),
156 (uch - 0x10000) % 0x400 + 0xDC00); 193 (int)((uch - 0x10000) % 0x400 + 0xDC00));
157 enc->cur += 12; 194 enc->cur += 12;
158 } 195 }
159 else 196 else
160 { 197 {
161 static char hexdigit [16] = "0123456789abcdef"; 198 static char hexdigit [16] = "0123456789abcdef";
… …
168 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 205 *enc->cur++ = hexdigit [(uch >> 0) & 15];
169 } 206 }
170 207
171 str += clen; 208 str += clen;
172 } 209 }
210 else if (enc->flags & F_LATIN1)
211 {
212 *enc->cur++ = uch;
213 str += clen;
214 }
173 else if (is_utf8) 215 else if (is_utf8)
174 { 216 {
175 need (enc, len += clen); 217 need (enc, len += clen);
176 do 218 do
177 { 219 {
… …
179 } 221 }
180 while (--clen); 222 while (--clen);
181 } 223 }
182 else 224 else
183 { 225 {
184 need (enc, len += 10); // never more than 11 bytes needed 226 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
185 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 227 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
186 ++str; 228 ++str;
187 } 229 }
188 } 230 }
189 } 231 }
… …
191 233
192 --len; 234 --len;
193 } 235 }
194} 236}
195 237
196#define INDENT SB \ 238static void
239encode_indent (enc_t *enc)
240{
197 if (enc->flags & F_INDENT) \ 241 if (enc->flags & F_INDENT)
198 { \ 242 {
199 int i_; \ 243 int spaces = enc->indent * INDENT_STEP;
200 need (enc, enc->indent); \
201 for (i_ = enc->indent * 3; i_--; )\
202 encode_ch (enc, ' '); \
203 } \
204 SE
205 244
206#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 245 need (enc, spaces);
207#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 246 memset (enc->cur, ' ', spaces);
208#define COMMA SB \ 247 enc->cur += spaces;
248 }
249}
250
251static void
252encode_space (enc_t *enc)
253{
254 need (enc, 1);
209 encode_ch (enc, ','); \ 255 encode_ch (enc, ' ');
256}
257
258static void
259encode_nl (enc_t *enc)
260{
210 if (enc->flags & F_INDENT) \ 261 if (enc->flags & F_INDENT)
211 NL; \ 262 {
263 need (enc, 1);
264 encode_ch (enc, '\n');
265 }
266}
267
268static void
269encode_comma (enc_t *enc)
270{
271 encode_ch (enc, ',');
272
273 if (enc->flags & F_INDENT)
274 encode_nl (enc);
212 else if (enc->flags & F_SPACE_AFTER) \ 275 else if (enc->flags & F_SPACE_AFTER)
213 SPACE; \ 276 encode_space (enc);
214 SE 277}
215 278
216static void encode_sv (enc_t *enc, SV *sv); 279static void encode_sv (enc_t *enc, SV *sv);
217 280
218static void 281static void
219encode_av (enc_t *enc, AV *av) 282encode_av (enc_t *enc, AV *av)
220{ 283{
221 int i, len = av_len (av); 284 int i, len = av_len (av);
222 285
223 encode_ch (enc, '['); NL; 286 if (enc->indent >= enc->maxdepth)
287 croak ("data structure too deep (hit recursion limit)");
288
289 encode_ch (enc, '['); encode_nl (enc);
224 ++enc->indent; 290 ++enc->indent;
225 291
226 for (i = 0; i <= len; ++i) 292 for (i = 0; i <= len; ++i)
227 { 293 {
228 INDENT; 294 encode_indent (enc);
229 encode_sv (enc, *av_fetch (av, i, 0)); 295 encode_sv (enc, *av_fetch (av, i, 0));
230 296
231 if (i < len) 297 if (i < len)
232 COMMA; 298 encode_comma (enc);
233 } 299 }
234 300
235 NL; 301 encode_nl (enc);
236 302
237 --enc->indent; 303 --enc->indent;
238 INDENT; encode_ch (enc, ']'); 304 encode_indent (enc); encode_ch (enc, ']');
239} 305}
240 306
241static void 307static void
242encode_he (enc_t *enc, HE *he) 308encode_he (enc_t *enc, HE *he)
243{ 309{
… …
257 else 323 else
258 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 324 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
259 325
260 encode_ch (enc, '"'); 326 encode_ch (enc, '"');
261 327
262 if (enc->flags & F_SPACE_BEFORE) SPACE; 328 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
263 encode_ch (enc, ':'); 329 encode_ch (enc, ':');
264 if (enc->flags & F_SPACE_AFTER ) SPACE; 330 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
265 encode_sv (enc, HeVAL (he)); 331 encode_sv (enc, HeVAL (he));
266} 332}
267 333
268// compare hash entries, used when all keys are bytestrings 334// compare hash entries, used when all keys are bytestrings
269static int 335static int
… …
275 HE *b = *(HE **)b_; 341 HE *b = *(HE **)b_;
276 342
277 STRLEN la = HeKLEN (a); 343 STRLEN la = HeKLEN (a);
278 STRLEN lb = HeKLEN (b); 344 STRLEN lb = HeKLEN (b);
279 345
280 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 346 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
281 cmp = la < lb ? -1 : la == lb ? 0 : 1; 347 cmp = la - lb;
282 348
283 return cmp; 349 return cmp;
284} 350}
285 351
286// compare hash entries, used when some keys are sv's or utf-x 352// compare hash entries, used when some keys are sv's or utf-x
… …
293static void 359static void
294encode_hv (enc_t *enc, HV *hv) 360encode_hv (enc_t *enc, HV *hv)
295{ 361{
296 int count, i; 362 int count, i;
297 363
364 if (enc->indent >= enc->maxdepth)
365 croak ("data structure too deep (hit recursion limit)");
366
298 encode_ch (enc, '{'); NL; ++enc->indent; 367 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
299 368
300 if ((count = hv_iterinit (hv))) 369 if ((count = hv_iterinit (hv)))
301 { 370 {
302 // for canonical output we have to sort by keys first 371 // for canonical output we have to sort by keys first
303 // actually, this is mostly due to the stupid so-called 372 // actually, this is mostly due to the stupid so-called
304 // security workaround added somewhere in 5.8.x. 373 // security workaround added somewhere in 5.8.x.
305 // that randomises hash orderings 374 // that randomises hash orderings
306 if (enc->flags & F_CANONICAL) 375 if (enc->flags & F_CANONICAL)
307 { 376 {
308 HE *he, *hes [count]; 377 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
309 int fast = 1; 378 int fast = 1;
310 379
311 i = 0; 380 i = 0;
312 while ((he = hv_iternext (hv))) 381 while ((he = hv_iternext (hv)))
313 { 382 {
… …
338 LEAVE; 407 LEAVE;
339 } 408 }
340 409
341 for (i = 0; i < count; ++i) 410 for (i = 0; i < count; ++i)
342 { 411 {
343 INDENT; 412 encode_indent (enc);
344 encode_he (enc, hes [i]); 413 encode_he (enc, hes [i]);
345 414
346 if (i < count - 1) 415 if (i < count - 1)
347 COMMA; 416 encode_comma (enc);
348 } 417 }
349 418
350 NL; 419 encode_nl (enc);
351 } 420 }
352 else 421 else
353 { 422 {
354 SV *sv;
355 HE *he = hv_iternext (hv); 423 HE *he = hv_iternext (hv);
356 424
357 for (;;) 425 for (;;)
358 { 426 {
359 INDENT; 427 encode_indent (enc);
360 encode_he (enc, he); 428 encode_he (enc, he);
361 429
362 if (!(he = hv_iternext (hv))) 430 if (!(he = hv_iternext (hv)))
363 break; 431 break;
364 432
365 COMMA; 433 encode_comma (enc);
366 } 434 }
367 435
368 NL; 436 encode_nl (enc);
369 } 437 }
370 } 438 }
371 439
372 --enc->indent; INDENT; encode_ch (enc, '}'); 440 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
441}
442
443// encode objects, arrays and special \0=false and \1=true values.
444static void
445encode_rv (enc_t *enc, SV *sv)
446{
447 svtype svt;
448
449 SvGETMAGIC (sv);
450 svt = SvTYPE (sv);
451
452 if (svt == SVt_PVHV)
453 encode_hv (enc, (HV *)sv);
454 else if (svt == SVt_PVAV)
455 encode_av (enc, (AV *)sv);
456 else if (svt < SVt_PVAV)
457 {
458 if (SvNIOK (sv) && SvIV (sv) == 0)
459 encode_str (enc, "false", 5, 0);
460 else if (SvNIOK (sv) && SvIV (sv) == 1)
461 encode_str (enc, "true", 4, 0);
462 else
463 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
464 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
465 }
466 else
467 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
468 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
373} 469}
374 470
375static void 471static void
376encode_sv (enc_t *enc, SV *sv) 472encode_sv (enc_t *enc, SV *sv)
377{ 473{
… …
398 SvIsUV(sv) 494 SvIsUV(sv)
399 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 495 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
400 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 496 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
401 } 497 }
402 else if (SvROK (sv)) 498 else if (SvROK (sv))
403 { 499 encode_rv (enc, SvRV (sv));
404 SV *rv = SvRV (sv);
405
406 if (!--enc->max_recurse)
407 croak ("data structure too deep (hit recursion limit)");
408
409 switch (SvTYPE (rv))
410 {
411 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
412 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
413
414 default:
415 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
416 SvPV_nolen (sv));
417 }
418 }
419 else if (!SvOK (sv)) 500 else if (!SvOK (sv))
420 encode_str (enc, "null", 4, 0); 501 encode_str (enc, "null", 4, 0);
421 else 502 else
422 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 503 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
423 SvPV_nolen (sv), SvFLAGS (sv)); 504 SvPV_nolen (sv), SvFLAGS (sv));
424} 505}
425 506
426static SV * 507static SV *
427encode_json (SV *scalar, UV flags) 508encode_json (SV *scalar, U32 flags)
428{ 509{
510 enc_t enc;
511
429 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 512 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
430 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 513 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
431 514
432 enc_t enc;
433 enc.flags = flags; 515 enc.flags = flags;
434 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 516 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
435 enc.cur = SvPVX (enc.sv); 517 enc.cur = SvPVX (enc.sv);
436 enc.end = SvEND (enc.sv); 518 enc.end = SvEND (enc.sv);
437 enc.max_recurse = 0;
438 enc.indent = 0; 519 enc.indent = 0;
520 enc.maxdepth = DEC_DEPTH (flags);
439 521
440 SvPOK_only (enc.sv); 522 SvPOK_only (enc.sv);
441 encode_sv (&enc, scalar); 523 encode_sv (&enc, scalar);
442 524
525 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
526 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
527
443 if (!(flags & (F_ASCII | F_UTF8))) 528 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
444 SvUTF8_on (enc.sv); 529 SvUTF8_on (enc.sv);
445
446 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
447 530
448 if (enc.flags & F_SHRINK) 531 if (enc.flags & F_SHRINK)
449 shrink (enc.sv); 532 shrink (enc.sv);
450 533
451 return enc.sv; 534 return enc.sv;
452} 535}
453 536
454///////////////////////////////////////////////////////////////////////////// 537/////////////////////////////////////////////////////////////////////////////
538// decoder
455 539
456#define WS \ 540// structure used for decoding JSON
541typedef struct
542{
543 char *cur; // current parser pointer
544 char *end; // end of input string
545 const char *err; // parse error, if != 0
546 U32 flags; // F_*
547 U32 depth; // recursion depth
548 U32 maxdepth; // recursion depth limit
549} dec_t;
550
551static void
552decode_ws (dec_t *dec)
553{
457 for (;;) \ 554 for (;;)
458 { \ 555 {
459 char ch = *dec->cur; \ 556 char ch = *dec->cur;
557
460 if (ch > 0x20 \ 558 if (ch > 0x20
461 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 559 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
462 break; \ 560 break;
561
463 ++dec->cur; \ 562 ++dec->cur;
464 } 563 }
564}
465 565
466#define ERR(reason) SB dec->err = reason; goto fail; SE 566#define ERR(reason) SB dec->err = reason; goto fail; SE
567
467#define EXPECT_CH(ch) SB \ 568#define EXPECT_CH(ch) SB \
468 if (*dec->cur != ch) \ 569 if (*dec->cur != ch) \
469 ERR (# ch " expected"); \ 570 ERR (# ch " expected"); \
470 ++dec->cur; \ 571 ++dec->cur; \
471 SE 572 SE
472 573
574#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
575#define DEC_DEC_DEPTH --dec->depth
576
473static SV *decode_sv (dec_t *dec); 577static SV *decode_sv (dec_t *dec);
474 578
475static signed char decode_hexdigit[256]; 579static signed char decode_hexdigit[256];
476 580
477static UV 581static UV
478decode_4hex (dec_t *dec) 582decode_4hex (dec_t *dec)
479{ 583{
480 signed char d1, d2, d3, d4; 584 signed char d1, d2, d3, d4;
585 unsigned char *cur = (unsigned char *)dec->cur;
481 586
482 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
483 if (d1 < 0) ERR ("four hexadecimal digits expected"); 587 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
484 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
485 if (d2 < 0) ERR ("four hexadecimal digits expected"); 588 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
486 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
487 if (d3 < 0) ERR ("four hexadecimal digits expected"); 589 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
488 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
489 if (d4 < 0) ERR ("four hexadecimal digits expected"); 590 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
490 591
491 dec->cur += 4; 592 dec->cur += 4;
492 593
493 return ((UV)d1) << 12 594 return ((UV)d1) << 12
494 | ((UV)d2) << 8 595 | ((UV)d2) << 8
… …
497 598
498fail: 599fail:
499 return (UV)-1; 600 return (UV)-1;
500} 601}
501 602
502#define APPEND_GROW(n) SB \
503 if (cur + (n) >= end) \
504 { \
505 STRLEN ofs = cur - SvPVX (sv); \
506 SvGROW (sv, ofs + (n) + 1); \
507 cur = SvPVX (sv) + ofs; \
508 end = SvEND (sv); \
509 } \
510 SE
511
512#define APPEND_CH(ch) SB \
513 APPEND_GROW (1); \
514 *cur++ = (ch); \
515 SE
516
517static SV * 603static SV *
518decode_str (dec_t *dec) 604decode_str (dec_t *dec)
519{ 605{
520 SV *sv = NEWSV (0,2); 606 SV *sv = 0;
521 int utf8 = 0; 607 int utf8 = 0;
522 char *cur = SvPVX (sv);
523 char *end = SvEND (sv);
524 608
525 for (;;) 609 do
526 { 610 {
527 unsigned char ch = *(unsigned char *)dec->cur; 611 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
612 char *cur = buf;
528 613
529 if (ch == '"') 614 do
530 break;
531 else if (ch == '\\')
532 { 615 {
533 switch (*++dec->cur) 616 unsigned char ch = *(unsigned char *)dec->cur++;
617
618 if (ch == '"')
534 { 619 {
535 case '\\': 620 --dec->cur;
536 case '/': 621 break;
537 case '"': APPEND_CH (*dec->cur++); break; 622 }
538 623 else if (ch == '\\')
539 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 624 {
540 case 't': APPEND_CH ('\011'); ++dec->cur; break; 625 switch (*dec->cur)
541 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
542 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
543 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
544
545 case 'u':
546 { 626 {
547 UV lo, hi; 627 case '\\':
548 ++dec->cur; 628 case '/':
629 case '"': *cur++ = *dec->cur++; break;
549 630
550 hi = decode_4hex (dec); 631 case 'b': ++dec->cur; *cur++ = '\010'; break;
551 if (hi == (UV)-1) 632 case 't': ++dec->cur; *cur++ = '\011'; break;
552 goto fail; 633 case 'n': ++dec->cur; *cur++ = '\012'; break;
634 case 'f': ++dec->cur; *cur++ = '\014'; break;
635 case 'r': ++dec->cur; *cur++ = '\015'; break;
553 636
554 // possibly a surrogate pair 637 case 'u':
555 if (hi >= 0xd800 && hi < 0xdc00)
556 { 638 {
557 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 639 UV lo, hi;
558 ERR ("missing low surrogate character in surrogate pair");
559
560 dec->cur += 2; 640 ++dec->cur;
561 641
562 lo = decode_4hex (dec); 642 hi = decode_4hex (dec);
563 if (lo == (UV)-1) 643 if (hi == (UV)-1)
564 goto fail; 644 goto fail;
565 645
646 // possibly a surrogate pair
647 if (hi >= 0xd800)
648 if (hi < 0xdc00)
649 {
650 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
651 ERR ("missing low surrogate character in surrogate pair");
652
653 dec->cur += 2;
654
655 lo = decode_4hex (dec);
656 if (lo == (UV)-1)
657 goto fail;
658
566 if (lo < 0xdc00 || lo >= 0xe000) 659 if (lo < 0xdc00 || lo >= 0xe000)
567 ERR ("surrogate pair expected"); 660 ERR ("surrogate pair expected");
568 661
569 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 662 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
663 }
664 else if (hi < 0xe000)
665 ERR ("missing high surrogate character in surrogate pair");
666
667 if (hi >= 0x80)
668 {
669 utf8 = 1;
670
671 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
672 }
673 else
674 *cur++ = hi;
570 } 675 }
571 else if (hi >= 0xdc00 && hi < 0xe000)
572 ERR ("missing high surrogate character in surrogate pair");
573
574 if (hi >= 0x80)
575 { 676 break;
576 utf8 = 1;
577 677
578 APPEND_GROW (4); // at most 4 bytes for 21 bits
579 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
580 }
581 else 678 default:
582 APPEND_CH (hi); 679 --dec->cur;
680 ERR ("illegal backslash escape sequence in string");
583 } 681 }
584 break; 682 }
683 else if (ch >= 0x20 && ch <= 0x7f)
684 *cur++ = ch;
685 else if (ch >= 0x80)
686 {
687 STRLEN clen;
688 UV uch;
585 689
586 default:
587 --dec->cur; 690 --dec->cur;
588 ERR ("illegal backslash escape sequence in string"); 691
692 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
693 if (clen == (STRLEN)-1)
694 ERR ("malformed UTF-8 character in JSON string");
695
696 do
697 *cur++ = *dec->cur++;
698 while (--clen);
699
700 utf8 = 1;
701 }
702 else
703 {
704 --dec->cur;
705
706 if (!ch)
707 ERR ("unexpected end of string while parsing JSON string");
708 else
709 ERR ("invalid character encountered while parsing JSON string");
589 } 710 }
590 } 711 }
591 else if (ch >= 0x20 && ch <= 0x7f) 712 while (cur < buf + SHORT_STRING_LEN);
592 APPEND_CH (*dec->cur++); 713
593 else if (ch >= 0x80)
594 { 714 {
595 STRLEN clen; 715 STRLEN len = cur - buf;
596 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
597 if (clen == (STRLEN)-1)
598 ERR ("malformed UTF-8 character in JSON string");
599 716
600 APPEND_GROW (clen); 717 if (sv)
601 do
602 { 718 {
603 *cur++ = *dec->cur++; 719 SvGROW (sv, SvCUR (sv) + len + 1);
720 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
721 SvCUR_set (sv, SvCUR (sv) + len);
604 } 722 }
605 while (--clen);
606
607 utf8 = 1;
608 }
609 else if (dec->cur == dec->end)
610 ERR ("unexpected end of string while parsing json string");
611 else 723 else
612 ERR ("invalid character encountered"); 724 sv = newSVpvn (buf, len);
613 } 725 }
726 }
727 while (*dec->cur != '"');
614 728
615 ++dec->cur; 729 ++dec->cur;
616 730
617 SvCUR_set (sv, cur - SvPVX (sv)); 731 if (sv)
618 732 {
619 SvPOK_only (sv); 733 SvPOK_only (sv);
620 *SvEND (sv) = 0; 734 *SvEND (sv) = 0;
621 735
622 if (utf8) 736 if (utf8)
623 SvUTF8_on (sv); 737 SvUTF8_on (sv);
624 738 }
625 if (dec->flags & F_SHRINK) 739 else
626 shrink (sv); 740 sv = newSVpvn ("", 0);
627 741
628 return sv; 742 return sv;
629 743
630fail: 744fail:
631 SvREFCNT_dec (sv);
632 return 0; 745 return 0;
633} 746}
634 747
635static SV * 748static SV *
636decode_num (dec_t *dec) 749decode_num (dec_t *dec)
… …
717static SV * 830static SV *
718decode_av (dec_t *dec) 831decode_av (dec_t *dec)
719{ 832{
720 AV *av = newAV (); 833 AV *av = newAV ();
721 834
722 WS; 835 DEC_INC_DEPTH;
836 decode_ws (dec);
837
723 if (*dec->cur == ']') 838 if (*dec->cur == ']')
724 ++dec->cur; 839 ++dec->cur;
725 else 840 else
726 for (;;) 841 for (;;)
727 { 842 {
… …
731 if (!value) 846 if (!value)
732 goto fail; 847 goto fail;
733 848
734 av_push (av, value); 849 av_push (av, value);
735 850
736 WS; 851 decode_ws (dec);
737 852
738 if (*dec->cur == ']') 853 if (*dec->cur == ']')
739 { 854 {
740 ++dec->cur; 855 ++dec->cur;
741 break; 856 break;
… …
745 ERR (", or ] expected while parsing array"); 860 ERR (", or ] expected while parsing array");
746 861
747 ++dec->cur; 862 ++dec->cur;
748 } 863 }
749 864
865 DEC_DEC_DEPTH;
750 return newRV_noinc ((SV *)av); 866 return newRV_noinc ((SV *)av);
751 867
752fail: 868fail:
753 SvREFCNT_dec (av); 869 SvREFCNT_dec (av);
870 DEC_DEC_DEPTH;
754 return 0; 871 return 0;
755} 872}
756 873
757static SV * 874static SV *
758decode_hv (dec_t *dec) 875decode_hv (dec_t *dec)
759{ 876{
760 HV *hv = newHV (); 877 HV *hv = newHV ();
761 878
762 WS; 879 DEC_INC_DEPTH;
880 decode_ws (dec);
881
763 if (*dec->cur == '}') 882 if (*dec->cur == '}')
764 ++dec->cur; 883 ++dec->cur;
765 else 884 else
766 for (;;) 885 for (;;)
767 { 886 {
768 SV *key, *value; 887 SV *key, *value;
769 888
770 WS; EXPECT_CH ('"'); 889 decode_ws (dec); EXPECT_CH ('"');
771 890
772 key = decode_str (dec); 891 key = decode_str (dec);
773 if (!key) 892 if (!key)
774 goto fail; 893 goto fail;
775 894
776 WS; EXPECT_CH (':'); 895 decode_ws (dec); EXPECT_CH (':');
777 896
778 value = decode_sv (dec); 897 value = decode_sv (dec);
779 if (!value) 898 if (!value)
780 { 899 {
781 SvREFCNT_dec (key); 900 SvREFCNT_dec (key);
782 goto fail; 901 goto fail;
783 } 902 }
784 903
785 //TODO: optimise
786 hv_store_ent (hv, key, value, 0); 904 hv_store_ent (hv, key, value, 0);
905 SvREFCNT_dec (key);
787 906
788 WS; 907 decode_ws (dec);
789 908
790 if (*dec->cur == '}') 909 if (*dec->cur == '}')
791 { 910 {
792 ++dec->cur; 911 ++dec->cur;
793 break; 912 break;
… …
797 ERR (", or } expected while parsing object/hash"); 916 ERR (", or } expected while parsing object/hash");
798 917
799 ++dec->cur; 918 ++dec->cur;
800 } 919 }
801 920
921 DEC_DEC_DEPTH;
802 return newRV_noinc ((SV *)hv); 922 return newRV_noinc ((SV *)hv);
803 923
804fail: 924fail:
805 SvREFCNT_dec (hv); 925 SvREFCNT_dec (hv);
926 DEC_DEC_DEPTH;
806 return 0; 927 return 0;
807} 928}
808 929
809static SV * 930static SV *
810decode_sv (dec_t *dec) 931decode_sv (dec_t *dec)
811{ 932{
812 WS; 933 decode_ws (dec);
813 switch (*dec->cur) 934 switch (*dec->cur)
814 { 935 {
815 case '"': ++dec->cur; return decode_str (dec); 936 case '"': ++dec->cur; return decode_str (dec);
816 case '[': ++dec->cur; return decode_av (dec); 937 case '[': ++dec->cur; return decode_av (dec);
817 case '{': ++dec->cur; return decode_hv (dec); 938 case '{': ++dec->cur; return decode_hv (dec);
… …
853 ERR ("'null' expected"); 974 ERR ("'null' expected");
854 975
855 break; 976 break;
856 977
857 default: 978 default:
858 ERR ("malformed json string, neither array, object, number, string or atom"); 979 ERR ("malformed JSON string, neither array, object, number, string or atom");
859 break; 980 break;
860 } 981 }
861 982
862fail: 983fail:
863 return 0; 984 return 0;
864} 985}
865 986
866static SV * 987static SV *
867decode_json (SV *string, UV flags) 988decode_json (SV *string, U32 flags, UV *offset_return)
868{ 989{
990 dec_t dec;
991 UV offset;
869 SV *sv; 992 SV *sv;
993
994 SvGETMAGIC (string);
995 SvUPGRADE (string, SVt_PV);
870 996
871 if (flags & F_UTF8) 997 if (flags & F_UTF8)
872 sv_utf8_downgrade (string, 0); 998 sv_utf8_downgrade (string, 0);
873 else 999 else
874 sv_utf8_upgrade (string); 1000 sv_utf8_upgrade (string);
875 1001
876 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1002 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
877 1003
878 dec_t dec;
879 dec.flags = flags; 1004 dec.flags = flags;
880 dec.cur = SvPVX (string); 1005 dec.cur = SvPVX (string);
881 dec.end = SvEND (string); 1006 dec.end = SvEND (string);
882 dec.err = 0; 1007 dec.err = 0;
1008 dec.depth = 0;
1009 dec.maxdepth = DEC_DEPTH (dec.flags);
883 1010
1011 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
884 sv = decode_sv (&dec); 1012 sv = decode_sv (&dec);
885 1013
1014 if (!(offset_return || !sv))
1015 {
1016 // check for trailing garbage
1017 decode_ws (&dec);
1018
1019 if (*dec.cur)
1020 {
1021 dec.err = "garbage after JSON object";
1022 SvREFCNT_dec (sv);
1023 sv = 0;
1024 }
1025 }
1026
1027 if (offset_return || !sv)
1028 {
1029 offset = dec.flags & F_UTF8
1030 ? dec.cur - SvPVX (string)
1031 : utf8_distance (dec.cur, SvPVX (string));
1032
1033 if (offset_return)
1034 *offset_return = offset;
1035 }
1036
886 if (!sv) 1037 if (!sv)
887 { 1038 {
888 IV offset = dec.flags & F_UTF8
889 ? dec.cur - SvPVX (string)
890 : utf8_distance (dec.cur, SvPVX (string));
891 SV *uni = sv_newmortal (); 1039 SV *uni = sv_newmortal ();
892 1040
893 // horrible hack to silence warning inside pv_uni_display 1041 // horrible hack to silence warning inside pv_uni_display
894 COP cop = *PL_curcop; 1042 COP cop = *PL_curcop;
895 cop.cop_warnings = pWARN_NONE; 1043 cop.cop_warnings = pWARN_NONE;
… …
897 SAVEVPTR (PL_curcop); 1045 SAVEVPTR (PL_curcop);
898 PL_curcop = &cop; 1046 PL_curcop = &cop;
899 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1047 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
900 LEAVE; 1048 LEAVE;
901 1049
902 croak ("%s, at character offset %d (%s)", 1050 croak ("%s, at character offset %d [\"%s\"]",
903 dec.err, 1051 dec.err,
904 (int)offset, 1052 (int)offset,
905 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1053 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
906 } 1054 }
907 1055
… …
911 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1059 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
912 1060
913 return sv; 1061 return sv;
914} 1062}
915 1063
1064/////////////////////////////////////////////////////////////////////////////
1065// XS interface functions
1066
916MODULE = JSON::XS PACKAGE = JSON::XS 1067MODULE = JSON::XS PACKAGE = JSON::XS
917 1068
918BOOT: 1069BOOT:
919{ 1070{
920 int i; 1071 int i;
921 1072
922 memset (decode_hexdigit, 0xff, 256); 1073 memset (decode_hexdigit, 0xff, 256);
1074
923 for (i = 10; i--; ) 1075 for (i = 0; i < 256; ++i)
924 decode_hexdigit ['0' + i] = i; 1076 decode_hexdigit [i] =
925 1077 i >= '0' && i <= '9' ? i - '0'
926 for (i = 7; i--; ) 1078 : i >= 'a' && i <= 'f' ? i - 'a' + 10
927 { 1079 : i >= 'A' && i <= 'F' ? i - 'A' + 10
928 decode_hexdigit ['a' + i] = 10 + i; 1080 : -1;
929 decode_hexdigit ['A' + i] = 10 + i;
930 }
931 1081
932 json_stash = gv_stashpv ("JSON::XS", 1); 1082 json_stash = gv_stashpv ("JSON::XS", 1);
933} 1083}
934 1084
935PROTOTYPES: DISABLE 1085PROTOTYPES: DISABLE
… …
941 RETVAL 1091 RETVAL
942 1092
943SV *ascii (SV *self, int enable = 1) 1093SV *ascii (SV *self, int enable = 1)
944 ALIAS: 1094 ALIAS:
945 ascii = F_ASCII 1095 ascii = F_ASCII
1096 latin1 = F_LATIN1
946 utf8 = F_UTF8 1097 utf8 = F_UTF8
947 indent = F_INDENT 1098 indent = F_INDENT
948 canonical = F_CANONICAL 1099 canonical = F_CANONICAL
949 space_before = F_SPACE_BEFORE 1100 space_before = F_SPACE_BEFORE
950 space_after = F_SPACE_AFTER 1101 space_after = F_SPACE_AFTER
951 json_rpc = F_JSON_RPC
952 pretty = F_PRETTY 1102 pretty = F_PRETTY
953 allow_nonref = F_ALLOW_NONREF 1103 allow_nonref = F_ALLOW_NONREF
954 shrink = F_SHRINK 1104 shrink = F_SHRINK
955 CODE: 1105 CODE:
956{ 1106{
… …
963 RETVAL = newSVsv (self); 1113 RETVAL = newSVsv (self);
964} 1114}
965 OUTPUT: 1115 OUTPUT:
966 RETVAL 1116 RETVAL
967 1117
1118SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1119 CODE:
1120{
1121 UV *uv = SvJSON (self);
1122 UV log2 = 0;
1123
1124 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1125
1126 while ((1UL << log2) < max_depth)
1127 ++log2;
1128
1129 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1130
1131 RETVAL = newSVsv (self);
1132}
1133 OUTPUT:
1134 RETVAL
1135
968void encode (SV *self, SV *scalar) 1136void encode (SV *self, SV *scalar)
969 PPCODE: 1137 PPCODE:
970 XPUSHs (encode_json (scalar, *SvJSON (self))); 1138 XPUSHs (encode_json (scalar, *SvJSON (self)));
971 1139
972void decode (SV *self, SV *jsonstr) 1140void decode (SV *self, SV *jsonstr)
973 PPCODE: 1141 PPCODE:
974 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1142 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1143
1144void decode_prefix (SV *self, SV *jsonstr)
1145 PPCODE:
1146{
1147 UV offset;
1148 EXTEND (SP, 2);
1149 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1150 PUSHs (sv_2mortal (newSVuv (offset)));
1151}
975 1152
976PROTOTYPES: ENABLE 1153PROTOTYPES: ENABLE
977 1154
978void to_json (SV *scalar) 1155void to_json (SV *scalar)
1156 ALIAS:
1157 objToJson = 0
979 PPCODE: 1158 PPCODE:
980 XPUSHs (encode_json (scalar, F_UTF8)); 1159 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
981 1160
982void from_json (SV *jsonstr) 1161void from_json (SV *jsonstr)
1162 ALIAS:
1163 jsonToObj = 0
983 PPCODE: 1164 PPCODE:
984 XPUSHs (decode_json (jsonstr, F_UTF8)); 1165 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
985 1166

Diff Legend

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