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 1.8 by root, Fri Mar 23 16:13:59 2007 UTC vs.
Revision 1.27 by root, Mon Apr 9 05:09:57 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 UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
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 UTF-8 character in string, cannot convert to JSON"); 175 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
139 } 176 }
140 else 177 else
141 { 178 {
142 uch = ch; 179 uch = ch;
143 clen = 1; 180 clen = 1;
144 } 181 }
145 182
183 if (uch > 0x10FFFFUL)
184 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
185
146 if (uch < 0x80 || enc->flags & F_ASCII) 186 if (uch < 0x80 || enc->flags & F_ASCII)
147 { 187 {
148 if (uch > 0xFFFFUL) 188 if (uch > 0xFFFFUL)
149 { 189 {
150 need (enc, len += 11); 190 need (enc, len += 11);
151 sprintf (enc->cur, "\\u%04x\\u%04x", 191 sprintf (enc->cur, "\\u%04x\\u%04x",
152 (uch - 0x10000) / 0x400 + 0xD800, 192 (int)((uch - 0x10000) / 0x400 + 0xD800),
153 (uch - 0x10000) % 0x400 + 0xDC00); 193 (int)((uch - 0x10000) % 0x400 + 0xDC00));
154 enc->cur += 12; 194 enc->cur += 12;
155 } 195 }
156 else 196 else
157 { 197 {
158 static char hexdigit [16] = "0123456789abcdef"; 198 static char hexdigit [16] = "0123456789abcdef";
… …
176 } 216 }
177 while (--clen); 217 while (--clen);
178 } 218 }
179 else 219 else
180 { 220 {
181 need (enc, len += 10); // never more than 11 bytes needed 221 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
182 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 222 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
183 ++str; 223 ++str;
184 } 224 }
185 } 225 }
186 } 226 }
… …
188 228
189 --len; 229 --len;
190 } 230 }
191} 231}
192 232
193#define INDENT SB \ 233static void
234encode_indent (enc_t *enc)
235{
194 if (enc->flags & F_INDENT) \ 236 if (enc->flags & F_INDENT)
195 { \ 237 {
196 int i_; \ 238 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
202 239
203#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 240 need (enc, spaces);
204#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 241 memset (enc->cur, ' ', spaces);
205#define COMMA SB \ 242 enc->cur += spaces;
243 }
244}
245
246static void
247encode_space (enc_t *enc)
248{
249 need (enc, 1);
206 encode_ch (enc, ','); \ 250 encode_ch (enc, ' ');
251}
252
253static void
254encode_nl (enc_t *enc)
255{
207 if (enc->flags & F_INDENT) \ 256 if (enc->flags & F_INDENT)
208 NL; \ 257 {
258 need (enc, 1);
259 encode_ch (enc, '\n');
260 }
261}
262
263static void
264encode_comma (enc_t *enc)
265{
266 encode_ch (enc, ',');
267
268 if (enc->flags & F_INDENT)
269 encode_nl (enc);
209 else if (enc->flags & F_SPACE_AFTER) \ 270 else if (enc->flags & F_SPACE_AFTER)
210 SPACE; \ 271 encode_space (enc);
211 SE 272}
212 273
213static void encode_sv (enc_t *enc, SV *sv); 274static void encode_sv (enc_t *enc, SV *sv);
214 275
215static void 276static void
216encode_av (enc_t *enc, AV *av) 277encode_av (enc_t *enc, AV *av)
217{ 278{
218 int i, len = av_len (av); 279 int i, len = av_len (av);
219 280
220 encode_ch (enc, '['); NL; 281 if (enc->indent >= enc->maxdepth)
282 croak ("data structure too deep (hit recursion limit)");
283
284 encode_ch (enc, '['); encode_nl (enc);
221 ++enc->indent; 285 ++enc->indent;
222 286
223 for (i = 0; i <= len; ++i) 287 for (i = 0; i <= len; ++i)
224 { 288 {
225 INDENT; 289 encode_indent (enc);
226 encode_sv (enc, *av_fetch (av, i, 0)); 290 encode_sv (enc, *av_fetch (av, i, 0));
227 291
228 if (i < len) 292 if (i < len)
229 COMMA; 293 encode_comma (enc);
230 } 294 }
231 295
232 NL; 296 encode_nl (enc);
233 297
234 --enc->indent; 298 --enc->indent;
235 INDENT; encode_ch (enc, ']'); 299 encode_indent (enc); encode_ch (enc, ']');
236} 300}
237 301
238static void 302static void
239encode_he (enc_t *enc, HE *he) 303encode_he (enc_t *enc, HE *he)
240{ 304{
… …
254 else 318 else
255 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 319 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
256 320
257 encode_ch (enc, '"'); 321 encode_ch (enc, '"');
258 322
259 if (enc->flags & F_SPACE_BEFORE) SPACE; 323 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
260 encode_ch (enc, ':'); 324 encode_ch (enc, ':');
261 if (enc->flags & F_SPACE_AFTER ) SPACE; 325 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
262 encode_sv (enc, HeVAL (he)); 326 encode_sv (enc, HeVAL (he));
263} 327}
264 328
265// compare hash entries, used when all keys are bytestrings 329// compare hash entries, used when all keys are bytestrings
266static int 330static int
… …
272 HE *b = *(HE **)b_; 336 HE *b = *(HE **)b_;
273 337
274 STRLEN la = HeKLEN (a); 338 STRLEN la = HeKLEN (a);
275 STRLEN lb = HeKLEN (b); 339 STRLEN lb = HeKLEN (b);
276 340
277 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 341 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
278 cmp = la < lb ? -1 : la == lb ? 0 : 1; 342 cmp = la - lb;
279 343
280 return cmp; 344 return cmp;
281} 345}
282 346
283// compare hash entries, used when some keys are sv's or utf-x 347// compare hash entries, used when some keys are sv's or utf-x
… …
290static void 354static void
291encode_hv (enc_t *enc, HV *hv) 355encode_hv (enc_t *enc, HV *hv)
292{ 356{
293 int count, i; 357 int count, i;
294 358
359 if (enc->indent >= enc->maxdepth)
360 croak ("data structure too deep (hit recursion limit)");
361
295 encode_ch (enc, '{'); NL; ++enc->indent; 362 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
296 363
297 if ((count = hv_iterinit (hv))) 364 if ((count = hv_iterinit (hv)))
298 { 365 {
299 // for canonical output we have to sort by keys first 366 // for canonical output we have to sort by keys first
300 // actually, this is mostly due to the stupid so-called 367 // actually, this is mostly due to the stupid so-called
301 // security workaround added somewhere in 5.8.x. 368 // security workaround added somewhere in 5.8.x.
302 // that randomises hash orderings 369 // that randomises hash orderings
303 if (enc->flags & F_CANONICAL) 370 if (enc->flags & F_CANONICAL)
304 { 371 {
305 HE *he, *hes [count]; 372 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
306 int fast = 1; 373 int fast = 1;
307 374
308 i = 0; 375 i = 0;
309 while ((he = hv_iternext (hv))) 376 while ((he = hv_iternext (hv)))
310 { 377 {
… …
335 LEAVE; 402 LEAVE;
336 } 403 }
337 404
338 for (i = 0; i < count; ++i) 405 for (i = 0; i < count; ++i)
339 { 406 {
340 INDENT; 407 encode_indent (enc);
341 encode_he (enc, hes [i]); 408 encode_he (enc, hes [i]);
342 409
343 if (i < count - 1) 410 if (i < count - 1)
344 COMMA; 411 encode_comma (enc);
345 } 412 }
346 413
347 NL; 414 encode_nl (enc);
348 } 415 }
349 else 416 else
350 { 417 {
351 SV *sv;
352 HE *he = hv_iternext (hv); 418 HE *he = hv_iternext (hv);
353 419
354 for (;;) 420 for (;;)
355 { 421 {
356 INDENT; 422 encode_indent (enc);
357 encode_he (enc, he); 423 encode_he (enc, he);
358 424
359 if (!(he = hv_iternext (hv))) 425 if (!(he = hv_iternext (hv)))
360 break; 426 break;
361 427
362 COMMA; 428 encode_comma (enc);
363 } 429 }
364 430
365 NL; 431 encode_nl (enc);
366 } 432 }
367 } 433 }
368 434
369 --enc->indent; INDENT; encode_ch (enc, '}'); 435 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
436}
437
438// encode objects, arrays and special \0=false and \1=true values.
439static void
440encode_rv (enc_t *enc, SV *sv)
441{
442 svtype svt;
443
444 SvGETMAGIC (sv);
445 svt = SvTYPE (sv);
446
447 if (svt == SVt_PVHV)
448 encode_hv (enc, (HV *)sv);
449 else if (svt == SVt_PVAV)
450 encode_av (enc, (AV *)sv);
451 else if (svt < SVt_PVAV)
452 {
453 if (SvNIOK (sv) && SvIV (sv) == 0)
454 encode_str (enc, "false", 5, 0);
455 else if (SvNIOK (sv) && SvIV (sv) == 1)
456 encode_str (enc, "true", 4, 0);
457 else
458 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
459 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
460 }
461 else
462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
370} 464}
371 465
372static void 466static void
373encode_sv (enc_t *enc, SV *sv) 467encode_sv (enc_t *enc, SV *sv)
374{ 468{
… …
395 SvIsUV(sv) 489 SvIsUV(sv)
396 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 490 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
397 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 491 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
398 } 492 }
399 else if (SvROK (sv)) 493 else if (SvROK (sv))
400 { 494 encode_rv (enc, SvRV (sv));
401 if (!--enc->max_recurse)
402 croak ("data structure too deep (hit recursion limit)");
403
404 sv = SvRV (sv);
405
406 switch (SvTYPE (sv))
407 {
408 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
409 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
410
411 default:
412 croak ("JSON can only represent references to arrays or hashes");
413 }
414 }
415 else if (!SvOK (sv)) 495 else if (!SvOK (sv))
416 encode_str (enc, "null", 4, 0); 496 encode_str (enc, "null", 4, 0);
417 else 497 else
418 croak ("encountered perl type that JSON cannot handle"); 498 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
499 SvPV_nolen (sv), SvFLAGS (sv));
419} 500}
420 501
421static SV * 502static SV *
422encode_json (SV *scalar, UV flags) 503encode_json (SV *scalar, U32 flags)
423{ 504{
505 enc_t enc;
506
424 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 507 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
425 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 508 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
426 509
427 enc_t enc;
428 enc.flags = flags; 510 enc.flags = flags;
429 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 511 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
430 enc.cur = SvPVX (enc.sv); 512 enc.cur = SvPVX (enc.sv);
431 enc.end = SvEND (enc.sv); 513 enc.end = SvEND (enc.sv);
432 enc.max_recurse = 0;
433 enc.indent = 0; 514 enc.indent = 0;
515 enc.maxdepth = DEC_DEPTH (flags);
434 516
435 SvPOK_only (enc.sv); 517 SvPOK_only (enc.sv);
436 encode_sv (&enc, scalar); 518 encode_sv (&enc, scalar);
437 519
438 if (!(flags & (F_ASCII | F_UTF8))) 520 if (!(flags & (F_ASCII | F_UTF8)))
439 SvUTF8_on (enc.sv); 521 SvUTF8_on (enc.sv);
440 522
441 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 523 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
524 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
442 525
443 if (enc.flags & F_SHRINK) 526 if (enc.flags & F_SHRINK)
444 shrink (enc.sv); 527 shrink (enc.sv);
445 528
446 return enc.sv; 529 return enc.sv;
447} 530}
448 531
449///////////////////////////////////////////////////////////////////////////// 532/////////////////////////////////////////////////////////////////////////////
533// decoder
450 534
451#define WS \ 535// structure used for decoding JSON
536typedef struct
537{
538 char *cur; // current parser pointer
539 char *end; // end of input string
540 const char *err; // parse error, if != 0
541 U32 flags; // F_*
542 U32 depth; // recursion depth
543 U32 maxdepth; // recursion depth limit
544} dec_t;
545
546static void
547decode_ws (dec_t *dec)
548{
452 for (;;) \ 549 for (;;)
453 { \ 550 {
454 char ch = *dec->cur; \ 551 char ch = *dec->cur;
552
455 if (ch > 0x20 \ 553 if (ch > 0x20
456 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 554 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
457 break; \ 555 break;
556
458 ++dec->cur; \ 557 ++dec->cur;
459 } 558 }
559}
460 560
461#define ERR(reason) SB dec->err = reason; goto fail; SE 561#define ERR(reason) SB dec->err = reason; goto fail; SE
562
462#define EXPECT_CH(ch) SB \ 563#define EXPECT_CH(ch) SB \
463 if (*dec->cur != ch) \ 564 if (*dec->cur != ch) \
464 ERR (# ch " expected"); \ 565 ERR (# ch " expected"); \
465 ++dec->cur; \ 566 ++dec->cur; \
466 SE 567 SE
467 568
569#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
570#define DEC_DEC_DEPTH --dec->depth
571
468static SV *decode_sv (dec_t *dec); 572static SV *decode_sv (dec_t *dec);
469 573
470static signed char decode_hexdigit[256]; 574static signed char decode_hexdigit[256];
471 575
472static UV 576static UV
473decode_4hex (dec_t *dec) 577decode_4hex (dec_t *dec)
474{ 578{
475 signed char d1, d2, d3, d4; 579 signed char d1, d2, d3, d4;
580 unsigned char *cur = (unsigned char *)dec->cur;
476 581
477 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
478 if (d1 < 0) ERR ("four hexadecimal digits expected"); 582 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
479 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
480 if (d2 < 0) ERR ("four hexadecimal digits expected"); 583 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
481 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
482 if (d3 < 0) ERR ("four hexadecimal digits expected"); 584 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
483 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
484 if (d4 < 0) ERR ("four hexadecimal digits expected"); 585 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
485 586
486 dec->cur += 4; 587 dec->cur += 4;
487 588
488 return ((UV)d1) << 12 589 return ((UV)d1) << 12
489 | ((UV)d2) << 8 590 | ((UV)d2) << 8
… …
492 593
493fail: 594fail:
494 return (UV)-1; 595 return (UV)-1;
495} 596}
496 597
497#define APPEND_GROW(n) SB \
498 if (cur + (n) >= end) \
499 { \
500 STRLEN ofs = cur - SvPVX (sv); \
501 SvGROW (sv, ofs + (n) + 1); \
502 cur = SvPVX (sv) + ofs; \
503 end = SvEND (sv); \
504 } \
505 SE
506
507#define APPEND_CH(ch) SB \
508 APPEND_GROW (1); \
509 *cur++ = (ch); \
510 SE
511
512static SV * 598static SV *
513decode_str (dec_t *dec) 599decode_str (dec_t *dec)
514{ 600{
515 SV *sv = NEWSV (0,2); 601 SV *sv = 0;
516 int utf8 = 0; 602 int utf8 = 0;
517 char *cur = SvPVX (sv);
518 char *end = SvEND (sv);
519 603
520 for (;;) 604 do
521 { 605 {
522 unsigned char ch = *(unsigned char *)dec->cur; 606 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
607 char *cur = buf;
523 608
524 if (ch == '"') 609 do
525 break;
526 else if (ch == '\\')
527 { 610 {
528 switch (*++dec->cur) 611 unsigned char ch = *(unsigned char *)dec->cur++;
612
613 if (ch == '"')
529 { 614 {
530 case '\\': 615 --dec->cur;
531 case '/': 616 break;
532 case '"': APPEND_CH (*dec->cur++); break; 617 }
533 618 else if (ch == '\\')
534 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 619 {
535 case 't': APPEND_CH ('\011'); ++dec->cur; break; 620 switch (*dec->cur)
536 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
537 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
538 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
539
540 case 'u':
541 { 621 {
542 UV lo, hi; 622 case '\\':
543 ++dec->cur; 623 case '/':
624 case '"': *cur++ = *dec->cur++; break;
544 625
545 hi = decode_4hex (dec); 626 case 'b': ++dec->cur; *cur++ = '\010'; break;
546 if (hi == (UV)-1) 627 case 't': ++dec->cur; *cur++ = '\011'; break;
547 goto fail; 628 case 'n': ++dec->cur; *cur++ = '\012'; break;
629 case 'f': ++dec->cur; *cur++ = '\014'; break;
630 case 'r': ++dec->cur; *cur++ = '\015'; break;
548 631
549 // possibly a surrogate pair 632 case 'u':
550 if (hi >= 0xd800 && hi < 0xdc00)
551 { 633 {
552 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 634 UV lo, hi;
553 ERR ("missing low surrogate character in surrogate pair");
554
555 dec->cur += 2; 635 ++dec->cur;
556 636
557 lo = decode_4hex (dec); 637 hi = decode_4hex (dec);
558 if (lo == (UV)-1) 638 if (hi == (UV)-1)
559 goto fail; 639 goto fail;
560 640
641 // possibly a surrogate pair
642 if (hi >= 0xd800)
643 if (hi < 0xdc00)
644 {
645 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
646 ERR ("missing low surrogate character in surrogate pair");
647
648 dec->cur += 2;
649
650 lo = decode_4hex (dec);
651 if (lo == (UV)-1)
652 goto fail;
653
561 if (lo < 0xdc00 || lo >= 0xe000) 654 if (lo < 0xdc00 || lo >= 0xe000)
562 ERR ("surrogate pair expected"); 655 ERR ("surrogate pair expected");
563 656
564 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 657 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
658 }
659 else if (hi < 0xe000)
660 ERR ("missing high surrogate character in surrogate pair");
661
662 if (hi >= 0x80)
663 {
664 utf8 = 1;
665
666 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
667 }
668 else
669 *cur++ = hi;
565 } 670 }
566 else if (hi >= 0xdc00 && hi < 0xe000)
567 ERR ("missing high surrogate character in surrogate pair");
568
569 if (hi >= 0x80)
570 { 671 break;
571 utf8 = 1;
572 672
573 APPEND_GROW (4); // at most 4 bytes for 21 bits
574 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
575 }
576 else 673 default:
577 APPEND_CH (hi); 674 --dec->cur;
675 ERR ("illegal backslash escape sequence in string");
578 } 676 }
579 break; 677 }
678 else if (ch >= 0x20 && ch <= 0x7f)
679 *cur++ = ch;
680 else if (ch >= 0x80)
681 {
682 STRLEN clen;
683 UV uch;
580 684
581 default:
582 --dec->cur; 685 --dec->cur;
583 ERR ("illegal backslash escape sequence in string"); 686
687 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
688 if (clen == (STRLEN)-1)
689 ERR ("malformed UTF-8 character in JSON string");
690
691 do
692 *cur++ = *dec->cur++;
693 while (--clen);
694
695 utf8 = 1;
696 }
697 else
698 {
699 --dec->cur;
700
701 if (!ch)
702 ERR ("unexpected end of string while parsing JSON string");
703 else
704 ERR ("invalid character encountered while parsing JSON string");
584 } 705 }
585 } 706 }
586 else if (ch >= 0x20 && ch <= 0x7f) 707 while (cur < buf + SHORT_STRING_LEN);
587 APPEND_CH (*dec->cur++); 708
588 else if (ch >= 0x80)
589 { 709 {
590 STRLEN clen; 710 STRLEN len = cur - buf;
591 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
592 if (clen == (STRLEN)-1)
593 ERR ("malformed UTF-8 character in JSON string");
594 711
595 APPEND_GROW (clen); 712 if (sv)
596 do
597 { 713 {
598 *cur++ = *dec->cur++; 714 SvGROW (sv, SvCUR (sv) + len + 1);
715 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
716 SvCUR_set (sv, SvCUR (sv) + len);
599 } 717 }
600 while (--clen);
601
602 utf8 = 1;
603 }
604 else if (dec->cur == dec->end)
605 ERR ("unexpected end of string while parsing json string");
606 else 718 else
607 ERR ("invalid character encountered"); 719 sv = newSVpvn (buf, len);
608 } 720 }
721 }
722 while (*dec->cur != '"');
609 723
610 ++dec->cur; 724 ++dec->cur;
611 725
612 SvCUR_set (sv, cur - SvPVX (sv)); 726 if (sv)
613 727 {
614 SvPOK_only (sv); 728 SvPOK_only (sv);
615 *SvEND (sv) = 0; 729 *SvEND (sv) = 0;
616 730
617 if (utf8) 731 if (utf8)
618 SvUTF8_on (sv); 732 SvUTF8_on (sv);
619 733 }
620 if (dec->flags & F_SHRINK) 734 else
621 shrink (sv); 735 sv = newSVpvn ("", 0);
622 736
623 return sv; 737 return sv;
624 738
625fail: 739fail:
626 SvREFCNT_dec (sv);
627 return 0; 740 return 0;
628} 741}
629 742
630static SV * 743static SV *
631decode_num (dec_t *dec) 744decode_num (dec_t *dec)
… …
712static SV * 825static SV *
713decode_av (dec_t *dec) 826decode_av (dec_t *dec)
714{ 827{
715 AV *av = newAV (); 828 AV *av = newAV ();
716 829
717 WS; 830 DEC_INC_DEPTH;
831 decode_ws (dec);
832
718 if (*dec->cur == ']') 833 if (*dec->cur == ']')
719 ++dec->cur; 834 ++dec->cur;
720 else 835 else
721 for (;;) 836 for (;;)
722 { 837 {
… …
726 if (!value) 841 if (!value)
727 goto fail; 842 goto fail;
728 843
729 av_push (av, value); 844 av_push (av, value);
730 845
731 WS; 846 decode_ws (dec);
732 847
733 if (*dec->cur == ']') 848 if (*dec->cur == ']')
734 { 849 {
735 ++dec->cur; 850 ++dec->cur;
736 break; 851 break;
… …
740 ERR (", or ] expected while parsing array"); 855 ERR (", or ] expected while parsing array");
741 856
742 ++dec->cur; 857 ++dec->cur;
743 } 858 }
744 859
860 DEC_DEC_DEPTH;
745 return newRV_noinc ((SV *)av); 861 return newRV_noinc ((SV *)av);
746 862
747fail: 863fail:
748 SvREFCNT_dec (av); 864 SvREFCNT_dec (av);
865 DEC_DEC_DEPTH;
749 return 0; 866 return 0;
750} 867}
751 868
752static SV * 869static SV *
753decode_hv (dec_t *dec) 870decode_hv (dec_t *dec)
754{ 871{
755 HV *hv = newHV (); 872 HV *hv = newHV ();
756 873
757 WS; 874 DEC_INC_DEPTH;
875 decode_ws (dec);
876
758 if (*dec->cur == '}') 877 if (*dec->cur == '}')
759 ++dec->cur; 878 ++dec->cur;
760 else 879 else
761 for (;;) 880 for (;;)
762 { 881 {
763 SV *key, *value; 882 SV *key, *value;
764 883
765 WS; EXPECT_CH ('"'); 884 decode_ws (dec); EXPECT_CH ('"');
766 885
767 key = decode_str (dec); 886 key = decode_str (dec);
768 if (!key) 887 if (!key)
769 goto fail; 888 goto fail;
770 889
771 WS; EXPECT_CH (':'); 890 decode_ws (dec); EXPECT_CH (':');
772 891
773 value = decode_sv (dec); 892 value = decode_sv (dec);
774 if (!value) 893 if (!value)
775 { 894 {
776 SvREFCNT_dec (key); 895 SvREFCNT_dec (key);
777 goto fail; 896 goto fail;
778 } 897 }
779 898
780 //TODO: optimise
781 hv_store_ent (hv, key, value, 0); 899 hv_store_ent (hv, key, value, 0);
900 SvREFCNT_dec (key);
782 901
783 WS; 902 decode_ws (dec);
784 903
785 if (*dec->cur == '}') 904 if (*dec->cur == '}')
786 { 905 {
787 ++dec->cur; 906 ++dec->cur;
788 break; 907 break;
… …
792 ERR (", or } expected while parsing object/hash"); 911 ERR (", or } expected while parsing object/hash");
793 912
794 ++dec->cur; 913 ++dec->cur;
795 } 914 }
796 915
916 DEC_DEC_DEPTH;
797 return newRV_noinc ((SV *)hv); 917 return newRV_noinc ((SV *)hv);
798 918
799fail: 919fail:
800 SvREFCNT_dec (hv); 920 SvREFCNT_dec (hv);
921 DEC_DEC_DEPTH;
801 return 0; 922 return 0;
802} 923}
803 924
804static SV * 925static SV *
805decode_sv (dec_t *dec) 926decode_sv (dec_t *dec)
806{ 927{
807 WS; 928 decode_ws (dec);
808 switch (*dec->cur) 929 switch (*dec->cur)
809 { 930 {
810 case '"': ++dec->cur; return decode_str (dec); 931 case '"': ++dec->cur; return decode_str (dec);
811 case '[': ++dec->cur; return decode_av (dec); 932 case '[': ++dec->cur; return decode_av (dec);
812 case '{': ++dec->cur; return decode_hv (dec); 933 case '{': ++dec->cur; return decode_hv (dec);
… …
848 ERR ("'null' expected"); 969 ERR ("'null' expected");
849 970
850 break; 971 break;
851 972
852 default: 973 default:
853 ERR ("malformed json string, neither array, object, number, string or atom"); 974 ERR ("malformed JSON string, neither array, object, number, string or atom");
854 break; 975 break;
855 } 976 }
856 977
857fail: 978fail:
858 return 0; 979 return 0;
859} 980}
860 981
861static SV * 982static SV *
862decode_json (SV *string, UV flags) 983decode_json (SV *string, U32 flags)
863{ 984{
985 dec_t dec;
864 SV *sv; 986 SV *sv;
987
988 SvUPGRADE (string, SVt_PV);
865 989
866 if (flags & F_UTF8) 990 if (flags & F_UTF8)
867 sv_utf8_downgrade (string, 0); 991 sv_utf8_downgrade (string, 0);
868 else 992 else
869 sv_utf8_upgrade (string); 993 sv_utf8_upgrade (string);
870 994
871 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 995 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
872 996
873 dec_t dec;
874 dec.flags = flags; 997 dec.flags = flags;
875 dec.cur = SvPVX (string); 998 dec.cur = SvPVX (string);
876 dec.end = SvEND (string); 999 dec.end = SvEND (string);
877 dec.err = 0; 1000 dec.err = 0;
1001 dec.depth = 0;
1002 dec.maxdepth = DEC_DEPTH (dec.flags);
878 1003
1004 *dec.end = 0; // this should basically be a nop, too, but make sure its there
879 sv = decode_sv (&dec); 1005 sv = decode_sv (&dec);
880 1006
881 if (!sv) 1007 if (!sv)
882 { 1008 {
883 IV offset = dec.flags & F_UTF8 1009 IV offset = dec.flags & F_UTF8
… …
892 SAVEVPTR (PL_curcop); 1018 SAVEVPTR (PL_curcop);
893 PL_curcop = &cop; 1019 PL_curcop = &cop;
894 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);
895 LEAVE; 1021 LEAVE;
896 1022
897 croak ("%s, at character offset %d (%s)", 1023 croak ("%s, at character offset %d [\"%s\"]",
898 dec.err, 1024 dec.err,
899 (int)offset, 1025 (int)offset,
900 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1026 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
901 } 1027 }
902 1028
903 sv = sv_2mortal (sv); 1029 sv = sv_2mortal (sv);
904 1030
905 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1031 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
906 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)");
907 1033
908 return sv; 1034 return sv;
909} 1035}
910 1036
1037/////////////////////////////////////////////////////////////////////////////
1038// XS interface functions
1039
911MODULE = JSON::XS PACKAGE = JSON::XS 1040MODULE = JSON::XS PACKAGE = JSON::XS
912 1041
913BOOT: 1042BOOT:
914{ 1043{
915 int i; 1044 int i;
916 1045
917 memset (decode_hexdigit, 0xff, 256); 1046 memset (decode_hexdigit, 0xff, 256);
1047
918 for (i = 10; i--; ) 1048 for (i = 0; i < 256; ++i)
919 decode_hexdigit ['0' + i] = i; 1049 decode_hexdigit [i] =
920 1050 i >= '0' && i <= '9' ? i - '0'
921 for (i = 7; i--; ) 1051 : i >= 'a' && i <= 'f' ? i - 'a' + 10
922 { 1052 : i >= 'A' && i <= 'F' ? i - 'A' + 10
923 decode_hexdigit ['a' + i] = 10 + i; 1053 : -1;
924 decode_hexdigit ['A' + i] = 10 + i;
925 }
926 1054
927 json_stash = gv_stashpv ("JSON::XS", 1); 1055 json_stash = gv_stashpv ("JSON::XS", 1);
928} 1056}
929 1057
930PROTOTYPES: DISABLE 1058PROTOTYPES: DISABLE
… …
941 utf8 = F_UTF8 1069 utf8 = F_UTF8
942 indent = F_INDENT 1070 indent = F_INDENT
943 canonical = F_CANONICAL 1071 canonical = F_CANONICAL
944 space_before = F_SPACE_BEFORE 1072 space_before = F_SPACE_BEFORE
945 space_after = F_SPACE_AFTER 1073 space_after = F_SPACE_AFTER
946 json_rpc = F_JSON_RPC
947 pretty = F_PRETTY 1074 pretty = F_PRETTY
948 allow_nonref = F_ALLOW_NONREF 1075 allow_nonref = F_ALLOW_NONREF
949 shrink = F_SHRINK 1076 shrink = F_SHRINK
950 CODE: 1077 CODE:
951{ 1078{
… …
958 RETVAL = newSVsv (self); 1085 RETVAL = newSVsv (self);
959} 1086}
960 OUTPUT: 1087 OUTPUT:
961 RETVAL 1088 RETVAL
962 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
963void encode (SV *self, SV *scalar) 1108void encode (SV *self, SV *scalar)
964 PPCODE: 1109 PPCODE:
965 XPUSHs (encode_json (scalar, *SvJSON (self))); 1110 XPUSHs (encode_json (scalar, *SvJSON (self)));
966 1111
967void decode (SV *self, SV *jsonstr) 1112void decode (SV *self, SV *jsonstr)
… …
969 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1114 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
970 1115
971PROTOTYPES: ENABLE 1116PROTOTYPES: ENABLE
972 1117
973void to_json (SV *scalar) 1118void to_json (SV *scalar)
1119 ALIAS:
1120 objToJson = 0
974 PPCODE: 1121 PPCODE:
975 XPUSHs (encode_json (scalar, F_UTF8)); 1122 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
976 1123
977void from_json (SV *jsonstr) 1124void from_json (SV *jsonstr)
1125 ALIAS:
1126 jsonToObj = 0
978 PPCODE: 1127 PPCODE:
979 XPUSHs (decode_json (jsonstr, F_UTF8)); 1128 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
980 1129

Diff Legend

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