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 1.21 by root, Thu Mar 29 02:45:49 2007 UTC

… …
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 8
9#define F_ASCII 0x00000001 9#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 15#define F_ALLOW_NONREF 0x00000080UL
17#define F_SHRINK 0x00000100 16#define F_SHRINK 0x00000100UL
17#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27
19
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
18 24
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 26#define F_DEFAULT (12UL << S_MAXDEPTH)
21 27
22#define INIT_SIZE 32 // initial scalar size to be allocated 28#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level
30
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size
23 33
24#define SB do { 34#define SB do {
25#define SE } while (0) 35#define SE } while (0)
26 36
27static HV *json_stash; 37static HV *json_stash; // JSON::XS::
38
39/////////////////////////////////////////////////////////////////////////////
40// utility functions
41
42static UV *
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv)
53{
54 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1)
56 {
57#ifdef SvPV_shrink_to_cur
58 SvPV_shrink_to_cur (sv);
59#elif defined (SvPV_renew)
60 SvPV_renew (sv, SvCUR (sv) + 1);
61#endif
62 }
63}
64
65// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints
70static UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{
73 if (s[0] > 0xdf || s[0] < 0xc2)
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 {
77 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 }
80 else
81 return (UV)-1;
82}
83
84/////////////////////////////////////////////////////////////////////////////
85// encoder
28 86
29// structure used for encoding JSON 87// structure used for encoding JSON
30typedef struct 88typedef struct
31{ 89{
32 char *cur; 90 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 91 char *end; // SvEND (sv)
35 SV *sv; 92 SV *sv; // result scalar
36 UV flags; 93 U32 flags; // F_*
37 int max_recurse; 94 U32 indent; // indentation level
38 int indent; 95 U32 maxdepth; // max. indentation/recursion level
39} enc_t; 96} 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 97
70static void 98static void
71need (enc_t *enc, STRLEN len) 99need (enc_t *enc, STRLEN len)
72{ 100{
73 if (enc->cur + len >= enc->end) 101 if (enc->cur + len >= enc->end)
… …
131 STRLEN clen; 159 STRLEN clen;
132 UV uch; 160 UV uch;
133 161
134 if (is_utf8) 162 if (is_utf8)
135 { 163 {
136 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen);
137 if (clen == (STRLEN)-1) 166 if (clen == (STRLEN)-1)
138 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
139 } 168 }
140 else 169 else
141 { 170 {
142 uch = ch; 171 uch = ch;
143 clen = 1; 172 clen = 1;
144 } 173 }
145 174
175 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177
146 if (uch < 0x80 || enc->flags & F_ASCII) 178 if (uch < 0x80 || enc->flags & F_ASCII)
147 { 179 {
148 if (uch > 0xFFFFUL) 180 if (uch > 0xFFFFUL)
149 { 181 {
150 need (enc, len += 11); 182 need (enc, len += 11);
151 sprintf (enc->cur, "\\u%04x\\u%04x", 183 sprintf (enc->cur, "\\u%04x\\u%04x",
152 (uch - 0x10000) / 0x400 + 0xD800, 184 (int)((uch - 0x10000) / 0x400 + 0xD800),
153 (uch - 0x10000) % 0x400 + 0xDC00); 185 (int)((uch - 0x10000) % 0x400 + 0xDC00));
154 enc->cur += 12; 186 enc->cur += 12;
155 } 187 }
156 else 188 else
157 { 189 {
158 static char hexdigit [16] = "0123456789abcdef"; 190 static char hexdigit [16] = "0123456789abcdef";
… …
176 } 208 }
177 while (--clen); 209 while (--clen);
178 } 210 }
179 else 211 else
180 { 212 {
181 need (enc, len += 10); // never more than 11 bytes needed 213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
182 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
183 ++str; 215 ++str;
184 } 216 }
185 } 217 }
186 } 218 }
… …
188 220
189 --len; 221 --len;
190 } 222 }
191} 223}
192 224
193#define INDENT SB \ 225static void
226encode_indent (enc_t *enc)
227{
194 if (enc->flags & F_INDENT) \ 228 if (enc->flags & F_INDENT)
195 { \ 229 {
196 int i_; \ 230 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
202 231
203#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 232 need (enc, spaces);
204#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 233 memset (enc->cur, ' ', spaces);
205#define COMMA SB \ 234 enc->cur += spaces;
235 }
236}
237
238static void
239encode_space (enc_t *enc)
240{
241 need (enc, 1);
206 encode_ch (enc, ','); \ 242 encode_ch (enc, ' ');
243}
244
245static void
246encode_nl (enc_t *enc)
247{
207 if (enc->flags & F_INDENT) \ 248 if (enc->flags & F_INDENT)
208 NL; \ 249 {
250 need (enc, 1);
251 encode_ch (enc, '\n');
252 }
253}
254
255static void
256encode_comma (enc_t *enc)
257{
258 encode_ch (enc, ',');
259
260 if (enc->flags & F_INDENT)
261 encode_nl (enc);
209 else if (enc->flags & F_SPACE_AFTER) \ 262 else if (enc->flags & F_SPACE_AFTER)
210 SPACE; \ 263 encode_space (enc);
211 SE 264}
212 265
213static void encode_sv (enc_t *enc, SV *sv); 266static void encode_sv (enc_t *enc, SV *sv);
214 267
215static void 268static void
216encode_av (enc_t *enc, AV *av) 269encode_av (enc_t *enc, AV *av)
217{ 270{
218 int i, len = av_len (av); 271 int i, len = av_len (av);
219 272
220 encode_ch (enc, '['); NL; 273 if (enc->indent >= enc->maxdepth)
274 croak ("data structure too deep (hit recursion limit)");
275
276 encode_ch (enc, '['); encode_nl (enc);
221 ++enc->indent; 277 ++enc->indent;
222 278
223 for (i = 0; i <= len; ++i) 279 for (i = 0; i <= len; ++i)
224 { 280 {
225 INDENT; 281 encode_indent (enc);
226 encode_sv (enc, *av_fetch (av, i, 0)); 282 encode_sv (enc, *av_fetch (av, i, 0));
227 283
228 if (i < len) 284 if (i < len)
229 COMMA; 285 encode_comma (enc);
230 } 286 }
231 287
232 NL; 288 encode_nl (enc);
233 289
234 --enc->indent; 290 --enc->indent;
235 INDENT; encode_ch (enc, ']'); 291 encode_indent (enc); encode_ch (enc, ']');
236} 292}
237 293
238static void 294static void
239encode_he (enc_t *enc, HE *he) 295encode_he (enc_t *enc, HE *he)
240{ 296{
… …
254 else 310 else
255 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 311 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
256 312
257 encode_ch (enc, '"'); 313 encode_ch (enc, '"');
258 314
259 if (enc->flags & F_SPACE_BEFORE) SPACE; 315 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
260 encode_ch (enc, ':'); 316 encode_ch (enc, ':');
261 if (enc->flags & F_SPACE_AFTER ) SPACE; 317 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
262 encode_sv (enc, HeVAL (he)); 318 encode_sv (enc, HeVAL (he));
263} 319}
264 320
265// compare hash entries, used when all keys are bytestrings 321// compare hash entries, used when all keys are bytestrings
266static int 322static int
… …
272 HE *b = *(HE **)b_; 328 HE *b = *(HE **)b_;
273 329
274 STRLEN la = HeKLEN (a); 330 STRLEN la = HeKLEN (a);
275 STRLEN lb = HeKLEN (b); 331 STRLEN lb = HeKLEN (b);
276 332
277 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 333 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
278 cmp = la < lb ? -1 : la == lb ? 0 : 1; 334 cmp = la - lb;
279 335
280 return cmp; 336 return cmp;
281} 337}
282 338
283// compare hash entries, used when some keys are sv's or utf-x 339// compare hash entries, used when some keys are sv's or utf-x
… …
290static void 346static void
291encode_hv (enc_t *enc, HV *hv) 347encode_hv (enc_t *enc, HV *hv)
292{ 348{
293 int count, i; 349 int count, i;
294 350
351 if (enc->indent >= enc->maxdepth)
352 croak ("data structure too deep (hit recursion limit)");
353
295 encode_ch (enc, '{'); NL; ++enc->indent; 354 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
296 355
297 if ((count = hv_iterinit (hv))) 356 if ((count = hv_iterinit (hv)))
298 { 357 {
299 // for canonical output we have to sort by keys first 358 // for canonical output we have to sort by keys first
300 // actually, this is mostly due to the stupid so-called 359 // actually, this is mostly due to the stupid so-called
301 // security workaround added somewhere in 5.8.x. 360 // security workaround added somewhere in 5.8.x.
302 // that randomises hash orderings 361 // that randomises hash orderings
303 if (enc->flags & F_CANONICAL) 362 if (enc->flags & F_CANONICAL)
304 { 363 {
305 HE *he, *hes [count]; 364 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
306 int fast = 1; 365 int fast = 1;
307 366
308 i = 0; 367 i = 0;
309 while ((he = hv_iternext (hv))) 368 while ((he = hv_iternext (hv)))
310 { 369 {
… …
317 376
318 if (fast) 377 if (fast)
319 qsort (hes, count, sizeof (HE *), he_cmp_fast); 378 qsort (hes, count, sizeof (HE *), he_cmp_fast);
320 else 379 else
321 { 380 {
322 // hack to disable "use bytes" 381 // hack to forcefully disable "use bytes"
323 COP *oldcop = PL_curcop, cop; 382 COP cop = *PL_curcop;
324 cop.op_private = 0; 383 cop.op_private = 0;
384
385 ENTER;
386 SAVETMPS;
387
388 SAVEVPTR (PL_curcop);
325 PL_curcop = &cop; 389 PL_curcop = &cop;
326 390
327 SAVETMPS;
328 qsort (hes, count, sizeof (HE *), he_cmp_slow); 391 qsort (hes, count, sizeof (HE *), he_cmp_slow);
392
329 FREETMPS; 393 FREETMPS;
330 394 LEAVE;
331 PL_curcop = oldcop;
332 } 395 }
333 396
334 for (i = 0; i < count; ++i) 397 for (i = 0; i < count; ++i)
335 { 398 {
336 INDENT; 399 encode_indent (enc);
337 encode_he (enc, hes [i]); 400 encode_he (enc, hes [i]);
338 401
339 if (i < count - 1) 402 if (i < count - 1)
340 COMMA; 403 encode_comma (enc);
341 } 404 }
342 405
343 NL; 406 encode_nl (enc);
344 } 407 }
345 else 408 else
346 { 409 {
347 SV *sv; 410 SV *sv;
348 HE *he = hv_iternext (hv); 411 HE *he = hv_iternext (hv);
349 412
350 for (;;) 413 for (;;)
351 { 414 {
352 INDENT; 415 encode_indent (enc);
353 encode_he (enc, he); 416 encode_he (enc, he);
354 417
355 if (!(he = hv_iternext (hv))) 418 if (!(he = hv_iternext (hv)))
356 break; 419 break;
357 420
358 COMMA; 421 encode_comma (enc);
359 } 422 }
360 423
361 NL; 424 encode_nl (enc);
362 } 425 }
363 } 426 }
364 427
365 --enc->indent; INDENT; encode_ch (enc, '}'); 428 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
429}
430
431// encode objects, arrays and special \0=false and \1=true values.
432static void
433encode_rv (enc_t *enc, SV *sv)
434{
435 SvGETMAGIC (sv);
436
437 svtype svt = SvTYPE (sv);
438
439 if (svt == SVt_PVHV)
440 encode_hv (enc, (HV *)sv);
441 else if (svt == SVt_PVAV)
442 encode_av (enc, (AV *)sv);
443 else if (svt < SVt_PVAV)
444 {
445 if (SvNIOK (sv) && SvIV (sv) == 0)
446 encode_str (enc, "false", 5, 0);
447 else if (SvNIOK (sv) && SvIV (sv) == 1)
448 encode_str (enc, "true", 4, 0);
449 else
450 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
451 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
452 }
453 else
454 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
455 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
366} 456}
367 457
368static void 458static void
369encode_sv (enc_t *enc, SV *sv) 459encode_sv (enc_t *enc, SV *sv)
370{ 460{
… …
391 SvIsUV(sv) 481 SvIsUV(sv)
392 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 482 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
393 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 483 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
394 } 484 }
395 else if (SvROK (sv)) 485 else if (SvROK (sv))
396 { 486 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)) 487 else if (!SvOK (sv))
412 encode_str (enc, "null", 4, 0); 488 encode_str (enc, "null", 4, 0);
413 else 489 else
414 croak ("encountered perl type that JSON cannot handle"); 490 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
491 SvPV_nolen (sv), SvFLAGS (sv));
415} 492}
416 493
417static SV * 494static SV *
418encode_json (SV *scalar, UV flags) 495encode_json (SV *scalar, U32 flags)
419{ 496{
420 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 497 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
421 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 498 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
422 499
423 enc_t enc; 500 enc_t enc;
424 enc.flags = flags; 501 enc.flags = flags;
425 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 502 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
426 enc.cur = SvPVX (enc.sv); 503 enc.cur = SvPVX (enc.sv);
427 enc.end = SvEND (enc.sv); 504 enc.end = SvEND (enc.sv);
428 enc.max_recurse = 0;
429 enc.indent = 0; 505 enc.indent = 0;
506 enc.maxdepth = DEC_DEPTH (flags);
430 507
431 SvPOK_only (enc.sv); 508 SvPOK_only (enc.sv);
432 encode_sv (&enc, scalar); 509 encode_sv (&enc, scalar);
433 510
434 if (!(flags & (F_ASCII | F_UTF8))) 511 if (!(flags & (F_ASCII | F_UTF8)))
… …
441 518
442 return enc.sv; 519 return enc.sv;
443} 520}
444 521
445///////////////////////////////////////////////////////////////////////////// 522/////////////////////////////////////////////////////////////////////////////
523// decoder
446 524
447#define WS \ 525// structure used for decoding JSON
526typedef struct
527{
528 char *cur; // current parser pointer
529 char *end; // end of input string
530 const char *err; // parse error, if != 0
531 U32 flags; // F_*
532 U32 depth; // recursion depth
533 U32 maxdepth; // recursion depth limit
534} dec_t;
535
536static void
537decode_ws (dec_t *dec)
538{
448 for (;;) \ 539 for (;;)
449 { \ 540 {
450 char ch = *dec->cur; \ 541 char ch = *dec->cur;
542
451 if (ch > 0x20 \ 543 if (ch > 0x20
452 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 544 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
453 break; \ 545 break;
546
454 ++dec->cur; \ 547 ++dec->cur;
455 } 548 }
549}
456 550
457#define ERR(reason) SB dec->err = reason; goto fail; SE 551#define ERR(reason) SB dec->err = reason; goto fail; SE
552
458#define EXPECT_CH(ch) SB \ 553#define EXPECT_CH(ch) SB \
459 if (*dec->cur != ch) \ 554 if (*dec->cur != ch) \
460 ERR (# ch " expected"); \ 555 ERR (# ch " expected"); \
461 ++dec->cur; \ 556 ++dec->cur; \
462 SE 557 SE
463 558
559#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
560#define DEC_DEC_DEPTH --dec->depth
561
464static SV *decode_sv (dec_t *dec); 562static SV *decode_sv (dec_t *dec);
465 563
466static signed char decode_hexdigit[256]; 564static signed char decode_hexdigit[256];
467 565
468static UV 566static UV
469decode_4hex (dec_t *dec) 567decode_4hex (dec_t *dec)
470{ 568{
471 signed char d1, d2, d3, d4; 569 signed char d1, d2, d3, d4;
570 unsigned char *cur = (unsigned char *)dec->cur;
472 571
473 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
474 if (d1 < 0) ERR ("four hexadecimal digits expected"); 572 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"); 573 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"); 574 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"); 575 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
481 576
482 dec->cur += 4; 577 dec->cur += 4;
483 578
484 return ((UV)d1) << 12 579 return ((UV)d1) << 12
485 | ((UV)d2) << 8 580 | ((UV)d2) << 8
… …
488 583
489fail: 584fail:
490 return (UV)-1; 585 return (UV)-1;
491} 586}
492 587
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 * 588static SV *
509decode_str (dec_t *dec) 589decode_str (dec_t *dec)
510{ 590{
511 SV *sv = NEWSV (0,2); 591 SV *sv = 0;
512 int utf8 = 0; 592 int utf8 = 0;
513 char *cur = SvPVX (sv);
514 char *end = SvEND (sv);
515 593
516 for (;;) 594 do
517 { 595 {
518 unsigned char ch = *(unsigned char *)dec->cur; 596 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
597 char *cur = buf;
519 598
520 if (ch == '"') 599 do
521 break;
522 else if (ch == '\\')
523 { 600 {
524 switch (*++dec->cur) 601 unsigned char ch = *(unsigned char *)dec->cur++;
602
603 if (ch == '"')
525 { 604 {
526 case '\\': 605 --dec->cur;
527 case '/': 606 break;
528 case '"': APPEND_CH (*dec->cur++); break; 607 }
529 608 else if (ch == '\\')
530 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 609 {
531 case 't': APPEND_CH ('\011'); ++dec->cur; break; 610 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 { 611 {
538 UV lo, hi; 612 case '\\':
539 ++dec->cur; 613 case '/':
614 case '"': *cur++ = *dec->cur++; break;
540 615
541 hi = decode_4hex (dec); 616 case 'b': ++dec->cur; *cur++ = '\010'; break;
542 if (hi == (UV)-1) 617 case 't': ++dec->cur; *cur++ = '\011'; break;
543 goto fail; 618 case 'n': ++dec->cur; *cur++ = '\012'; break;
619 case 'f': ++dec->cur; *cur++ = '\014'; break;
620 case 'r': ++dec->cur; *cur++ = '\015'; break;
544 621
545 // possibly a surrogate pair 622 case 'u':
546 if (hi >= 0xd800 && hi < 0xdc00)
547 { 623 {
548 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 624 UV lo, hi;
549 ERR ("missing low surrogate character in surrogate pair");
550
551 dec->cur += 2; 625 ++dec->cur;
552 626
553 lo = decode_4hex (dec); 627 hi = decode_4hex (dec);
554 if (lo == (UV)-1) 628 if (hi == (UV)-1)
555 goto fail; 629 goto fail;
556 630
631 // possibly a surrogate pair
632 if (hi >= 0xd800)
633 if (hi < 0xdc00)
634 {
635 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
636 ERR ("missing low surrogate character in surrogate pair");
637
638 dec->cur += 2;
639
640 lo = decode_4hex (dec);
641 if (lo == (UV)-1)
642 goto fail;
643
557 if (lo < 0xdc00 || lo >= 0xe000) 644 if (lo < 0xdc00 || lo >= 0xe000)
558 ERR ("surrogate pair expected"); 645 ERR ("surrogate pair expected");
559 646
560 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 647 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
648 }
649 else if (hi < 0xe000)
650 ERR ("missing high surrogate character in surrogate pair");
651
652 if (hi >= 0x80)
653 {
654 utf8 = 1;
655
656 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
657 }
658 else
659 *cur++ = hi;
561 } 660 }
562 else if (hi >= 0xdc00 && hi < 0xe000)
563 ERR ("missing high surrogate character in surrogate pair");
564
565 if (hi >= 0x80)
566 { 661 break;
567 utf8 = 1;
568 662
569 APPEND_GROW (4); // at most 4 bytes for 21 bits
570 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
571 }
572 else 663 default:
573 APPEND_CH (hi); 664 --dec->cur;
665 ERR ("illegal backslash escape sequence in string");
574 } 666 }
575 break;
576
577 default:
578 --dec->cur;
579 ERR ("illegal backslash escape sequence in string");
580 } 667 }
668 else if (ch >= 0x20 && ch <= 0x7f)
669 *cur++ = ch;
670 else if (ch >= 0x80)
671 {
672 --dec->cur;
673
674 STRLEN clen;
675 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
676 if (clen == (STRLEN)-1)
677 ERR ("malformed UTF-8 character in JSON string");
678
679 do
680 {
681 *cur++ = *dec->cur++;
682 }
683 while (--clen);
684
685 utf8 = 1;
686 }
687 else if (!ch)
688 ERR ("unexpected end of string while parsing json string");
689 else
690 ERR ("invalid character encountered");
691
581 } 692 }
582 else if (ch >= 0x20 && ch <= 0x7f) 693 while (cur < buf + SHORT_STRING_LEN);
583 APPEND_CH (*dec->cur++); 694
584 else if (ch >= 0x80) 695 STRLEN len = cur - buf;
696
697 if (sv)
585 { 698 {
586 STRLEN clen; 699 SvGROW (sv, SvCUR (sv) + len + 1);
587 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 700 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
588 if (clen == (STRLEN)-1) 701 SvCUR_set (sv, SvCUR (sv) + len);
589 ERR ("malformed UTF-8 character in JSON string");
590
591 APPEND_GROW (clen);
592 do
593 {
594 *cur++ = *dec->cur++;
595 }
596 while (--clen);
597
598 utf8 = 1;
599 } 702 }
600 else if (dec->cur == dec->end)
601 ERR ("unexpected end of string while parsing json string");
602 else 703 else
603 ERR ("invalid character encountered"); 704 sv = newSVpvn (buf, len);
604 } 705 }
706 while (*dec->cur != '"');
605 707
606 ++dec->cur; 708 ++dec->cur;
607 709
608 SvCUR_set (sv, cur - SvPVX (sv)); 710 if (sv)
609 711 {
610 SvPOK_only (sv); 712 SvPOK_only (sv);
611 *SvEND (sv) = 0; 713 *SvEND (sv) = 0;
612 714
613 if (utf8) 715 if (utf8)
614 SvUTF8_on (sv); 716 SvUTF8_on (sv);
615 717 }
616 if (dec->flags & F_SHRINK) 718 else
617 shrink (sv); 719 sv = newSVpvn ("", 0);
618 720
619 return sv; 721 return sv;
620 722
621fail: 723fail:
622 SvREFCNT_dec (sv);
623 return 0; 724 return 0;
624} 725}
625 726
626static SV * 727static SV *
627decode_num (dec_t *dec) 728decode_num (dec_t *dec)
… …
708static SV * 809static SV *
709decode_av (dec_t *dec) 810decode_av (dec_t *dec)
710{ 811{
711 AV *av = newAV (); 812 AV *av = newAV ();
712 813
713 WS; 814 DEC_INC_DEPTH;
815 decode_ws (dec);
816
714 if (*dec->cur == ']') 817 if (*dec->cur == ']')
715 ++dec->cur; 818 ++dec->cur;
716 else 819 else
717 for (;;) 820 for (;;)
718 { 821 {
… …
722 if (!value) 825 if (!value)
723 goto fail; 826 goto fail;
724 827
725 av_push (av, value); 828 av_push (av, value);
726 829
727 WS; 830 decode_ws (dec);
728 831
729 if (*dec->cur == ']') 832 if (*dec->cur == ']')
730 { 833 {
731 ++dec->cur; 834 ++dec->cur;
732 break; 835 break;
… …
736 ERR (", or ] expected while parsing array"); 839 ERR (", or ] expected while parsing array");
737 840
738 ++dec->cur; 841 ++dec->cur;
739 } 842 }
740 843
844 DEC_DEC_DEPTH;
741 return newRV_noinc ((SV *)av); 845 return newRV_noinc ((SV *)av);
742 846
743fail: 847fail:
744 SvREFCNT_dec (av); 848 SvREFCNT_dec (av);
849 DEC_DEC_DEPTH;
745 return 0; 850 return 0;
746} 851}
747 852
748static SV * 853static SV *
749decode_hv (dec_t *dec) 854decode_hv (dec_t *dec)
750{ 855{
751 HV *hv = newHV (); 856 HV *hv = newHV ();
752 857
753 WS; 858 DEC_INC_DEPTH;
859 decode_ws (dec);
860
754 if (*dec->cur == '}') 861 if (*dec->cur == '}')
755 ++dec->cur; 862 ++dec->cur;
756 else 863 else
757 for (;;) 864 for (;;)
758 { 865 {
759 SV *key, *value; 866 SV *key, *value;
760 867
761 WS; EXPECT_CH ('"'); 868 decode_ws (dec); EXPECT_CH ('"');
762 869
763 key = decode_str (dec); 870 key = decode_str (dec);
764 if (!key) 871 if (!key)
765 goto fail; 872 goto fail;
766 873
767 WS; EXPECT_CH (':'); 874 decode_ws (dec); EXPECT_CH (':');
768 875
769 value = decode_sv (dec); 876 value = decode_sv (dec);
770 if (!value) 877 if (!value)
771 { 878 {
772 SvREFCNT_dec (key); 879 SvREFCNT_dec (key);
773 goto fail; 880 goto fail;
774 } 881 }
775 882
776 //TODO: optimise
777 hv_store_ent (hv, key, value, 0); 883 hv_store_ent (hv, key, value, 0);
884 SvREFCNT_dec (key);
778 885
779 WS; 886 decode_ws (dec);
780 887
781 if (*dec->cur == '}') 888 if (*dec->cur == '}')
782 { 889 {
783 ++dec->cur; 890 ++dec->cur;
784 break; 891 break;
… …
788 ERR (", or } expected while parsing object/hash"); 895 ERR (", or } expected while parsing object/hash");
789 896
790 ++dec->cur; 897 ++dec->cur;
791 } 898 }
792 899
900 DEC_DEC_DEPTH;
793 return newRV_noinc ((SV *)hv); 901 return newRV_noinc ((SV *)hv);
794 902
795fail: 903fail:
796 SvREFCNT_dec (hv); 904 SvREFCNT_dec (hv);
905 DEC_DEC_DEPTH;
797 return 0; 906 return 0;
798} 907}
799 908
800static SV * 909static SV *
801decode_sv (dec_t *dec) 910decode_sv (dec_t *dec)
802{ 911{
803 WS; 912 decode_ws (dec);
804 switch (*dec->cur) 913 switch (*dec->cur)
805 { 914 {
806 case '"': ++dec->cur; return decode_str (dec); 915 case '"': ++dec->cur; return decode_str (dec);
807 case '[': ++dec->cur; return decode_av (dec); 916 case '[': ++dec->cur; return decode_av (dec);
808 case '{': ++dec->cur; return decode_hv (dec); 917 case '{': ++dec->cur; return decode_hv (dec);
… …
853fail: 962fail:
854 return 0; 963 return 0;
855} 964}
856 965
857static SV * 966static SV *
858decode_json (SV *string, UV flags) 967decode_json (SV *string, U32 flags)
859{ 968{
860 SV *sv; 969 SV *sv;
861 970
862 if (flags & F_UTF8) 971 if (flags & F_UTF8)
863 sv_utf8_downgrade (string, 0); 972 sv_utf8_downgrade (string, 0);
… …
865 sv_utf8_upgrade (string); 974 sv_utf8_upgrade (string);
866 975
867 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 976 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
868 977
869 dec_t dec; 978 dec_t dec;
870 dec.flags = flags; 979 dec.flags = flags;
871 dec.cur = SvPVX (string); 980 dec.cur = SvPVX (string);
872 dec.end = SvEND (string); 981 dec.end = SvEND (string);
873 dec.err = 0; 982 dec.err = 0;
983 dec.depth = 0;
984 dec.maxdepth = DEC_DEPTH (dec.flags);
874 985
986 *dec.end = 0; // this should basically be a nop, too, but make sure its there
875 sv = decode_sv (&dec); 987 sv = decode_sv (&dec);
876 988
877 if (!sv) 989 if (!sv)
878 { 990 {
879 IV offset = dec.flags & F_UTF8 991 IV offset = dec.flags & F_UTF8
880 ? dec.cur - SvPVX (string) 992 ? dec.cur - SvPVX (string)
881 : utf8_distance (dec.cur, SvPVX (string)); 993 : utf8_distance (dec.cur, SvPVX (string));
882 SV *uni = sv_newmortal (); 994 SV *uni = sv_newmortal ();
995
883 // horrible hack to silence warning inside pv_uni_display 996 // horrible hack to silence warning inside pv_uni_display
884 COP cop; 997 COP cop = *PL_curcop;
885 memset (&cop, 0, sizeof (cop));
886 cop.cop_warnings = pWARN_NONE; 998 cop.cop_warnings = pWARN_NONE;
999 ENTER;
887 SAVEVPTR (PL_curcop); 1000 SAVEVPTR (PL_curcop);
888 PL_curcop = &cop; 1001 PL_curcop = &cop;
889
890 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1002 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1003 LEAVE;
1004
891 croak ("%s, at character offset %d (%s)", 1005 croak ("%s, at character offset %d (%s)",
892 dec.err, 1006 dec.err,
893 (int)offset, 1007 (int)offset,
894 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1008 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
895 } 1009 }
896 1010
897 sv = sv_2mortal (sv); 1011 sv = sv_2mortal (sv);
898 1012
899 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1013 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)"); 1014 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
901 1015
902 return sv; 1016 return sv;
903} 1017}
904 1018
1019/////////////////////////////////////////////////////////////////////////////
1020// XS interface functions
1021
905MODULE = JSON::XS PACKAGE = JSON::XS 1022MODULE = JSON::XS PACKAGE = JSON::XS
906 1023
907BOOT: 1024BOOT:
908{ 1025{
909 int i; 1026 int i;
910 1027
911 memset (decode_hexdigit, 0xff, 256); 1028 memset (decode_hexdigit, 0xff, 256);
1029
912 for (i = 10; i--; ) 1030 for (i = 0; i < 256; ++i)
913 decode_hexdigit ['0' + i] = i; 1031 decode_hexdigit [i] =
914 1032 i >= '0' && i <= '9' ? i - '0'
915 for (i = 7; i--; ) 1033 : i >= 'a' && i <= 'f' ? i - 'a' + 10
916 { 1034 : i >= 'A' && i <= 'F' ? i - 'A' + 10
917 decode_hexdigit ['a' + i] = 10 + i; 1035 : -1;
918 decode_hexdigit ['A' + i] = 10 + i;
919 }
920 1036
921 json_stash = gv_stashpv ("JSON::XS", 1); 1037 json_stash = gv_stashpv ("JSON::XS", 1);
922} 1038}
923 1039
924PROTOTYPES: DISABLE 1040PROTOTYPES: DISABLE
… …
935 utf8 = F_UTF8 1051 utf8 = F_UTF8
936 indent = F_INDENT 1052 indent = F_INDENT
937 canonical = F_CANONICAL 1053 canonical = F_CANONICAL
938 space_before = F_SPACE_BEFORE 1054 space_before = F_SPACE_BEFORE
939 space_after = F_SPACE_AFTER 1055 space_after = F_SPACE_AFTER
940 json_rpc = F_JSON_RPC
941 pretty = F_PRETTY 1056 pretty = F_PRETTY
942 allow_nonref = F_ALLOW_NONREF 1057 allow_nonref = F_ALLOW_NONREF
943 shrink = F_SHRINK 1058 shrink = F_SHRINK
944 CODE: 1059 CODE:
945{ 1060{
… …
952 RETVAL = newSVsv (self); 1067 RETVAL = newSVsv (self);
953} 1068}
954 OUTPUT: 1069 OUTPUT:
955 RETVAL 1070 RETVAL
956 1071
1072SV *max_depth (SV *self, int max_depth = 0x80000000UL)
1073 CODE:
1074{
1075 UV *uv = SvJSON (self);
1076 UV log2 = 0;
1077
1078 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1079
1080 while ((1UL << log2) < max_depth)
1081 ++log2;
1082
1083 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1084
1085 RETVAL = newSVsv (self);
1086}
1087 OUTPUT:
1088 RETVAL
1089
957void encode (SV *self, SV *scalar) 1090void encode (SV *self, SV *scalar)
958 PPCODE: 1091 PPCODE:
959 XPUSHs (encode_json (scalar, *SvJSON (self))); 1092 XPUSHs (encode_json (scalar, *SvJSON (self)));
960 1093
961void decode (SV *self, SV *jsonstr) 1094void decode (SV *self, SV *jsonstr)
… …
963 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1096 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
964 1097
965PROTOTYPES: ENABLE 1098PROTOTYPES: ENABLE
966 1099
967void to_json (SV *scalar) 1100void to_json (SV *scalar)
1101 ALIAS:
1102 objToJson = 0
968 PPCODE: 1103 PPCODE:
969 XPUSHs (encode_json (scalar, F_UTF8)); 1104 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
970 1105
971void from_json (SV *jsonstr) 1106void from_json (SV *jsonstr)
1107 ALIAS:
1108 jsonToObj = 0
972 PPCODE: 1109 PPCODE:
973 XPUSHs (decode_json (jsonstr, F_UTF8)); 1110 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
974 1111

Diff Legend

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