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 1.11 by root, Sat Mar 24 19:42:14 2007 UTC vs.
Revision 1.47 by root, Sun Jul 1 14:08:03 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
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 27#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL // NYI
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35
36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 38
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 40#define F_DEFAULT (9UL << S_MAXDEPTH)
20 41
21#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level
44
45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 46
23#define SB do { 47#define SB do {
24#define SE } while (0) 48#define SE } while (0)
25 49
26static HV *json_stash; 50#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value))
52# define inline inline
53#else
54# define expect(expr,value) (expr)
55# define inline static
56#endif
57
58#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1)
60
61static HV *json_stash, *json_boolean_stash; // JSON::XS::
62static SV *json_true, *json_false;
63
64typedef struct json {
65 U32 flags;
66} JSON__XS;
67
68/////////////////////////////////////////////////////////////////////////////
69// utility functions
70
71static UV *
72SvJSON (SV *sv)
73{
74 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
75 croak ("object is not of type JSON::XS");
76
77 return &SvUVX (SvRV (sv));
78}
79
80static void
81shrink (SV *sv)
82{
83 sv_utf8_downgrade (sv, 1);
84 if (SvLEN (sv) > SvCUR (sv) + 1)
85 {
86#ifdef SvPV_shrink_to_cur
87 SvPV_shrink_to_cur (sv);
88#elif defined (SvPV_renew)
89 SvPV_renew (sv, SvCUR (sv) + 1);
90#endif
91 }
92}
93
94// decode an utf-8 character and return it, or (UV)-1 in
95// case of an error.
96// we special-case "safe" characters from U+80 .. U+7FF,
97// but use the very good perl function to parse anything else.
98// note that we never call this function for a ascii codepoints
99inline UV
100decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
101{
102 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
103 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
104 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
105 {
106 *clen = 2;
107 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
108 }
109 else
110 {
111 *clen = (STRLEN)-1;
112 return (UV)-1;
113 }
114}
115
116/////////////////////////////////////////////////////////////////////////////
117// encoder
27 118
28// structure used for encoding JSON 119// structure used for encoding JSON
29typedef struct 120typedef struct
30{ 121{
31 char *cur; 122 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 123 char *end; // SvEND (sv)
34 SV *sv; 124 SV *sv; // result scalar
35 UV flags; 125 struct json json;
36 int max_recurse; 126 U32 indent; // indentation level
37 int indent; 127 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 128} enc_t;
39 129
40// structure used for decoding JSON 130inline void
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58static void
59shrink (SV *sv)
60{
61 sv_utf8_downgrade (sv, 1);
62#ifdef SvPV_shrink_to_cur
63 SvPV_shrink_to_cur (sv);
64#endif
65}
66
67/////////////////////////////////////////////////////////////////////////////
68
69static void
70need (enc_t *enc, STRLEN len) 131need (enc_t *enc, STRLEN len)
71{ 132{
72 if (enc->cur + len >= enc->end) 133 if (expect_false (enc->cur + len >= enc->end))
73 { 134 {
74 STRLEN cur = enc->cur - SvPVX (enc->sv); 135 STRLEN cur = enc->cur - SvPVX (enc->sv);
75 SvGROW (enc->sv, cur + len + 1); 136 SvGROW (enc->sv, cur + len + 1);
76 enc->cur = SvPVX (enc->sv) + cur; 137 enc->cur = SvPVX (enc->sv) + cur;
77 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 138 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
78 } 139 }
79} 140}
80 141
81static void 142inline void
82encode_ch (enc_t *enc, char ch) 143encode_ch (enc_t *enc, char ch)
83{ 144{
84 need (enc, 1); 145 need (enc, 1);
85 *enc->cur++ = ch; 146 *enc->cur++ = ch;
86} 147}
… …
94 155
95 while (str < end) 156 while (str < end)
96 { 157 {
97 unsigned char ch = *(unsigned char *)str; 158 unsigned char ch = *(unsigned char *)str;
98 159
99 if (ch >= 0x20 && ch < 0x80) // most common case 160 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
100 { 161 {
101 if (ch == '"') // but with slow exceptions 162 if (expect_false (ch == '"')) // but with slow exceptions
102 { 163 {
103 need (enc, len += 1); 164 need (enc, len += 1);
104 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
105 *enc->cur++ = '"'; 166 *enc->cur++ = '"';
106 } 167 }
107 else if (ch == '\\') 168 else if (expect_false (ch == '\\'))
108 { 169 {
109 need (enc, len += 1); 170 need (enc, len += 1);
110 *enc->cur++ = '\\'; 171 *enc->cur++ = '\\';
111 *enc->cur++ = '\\'; 172 *enc->cur++ = '\\';
112 } 173 }
… …
130 STRLEN clen; 191 STRLEN clen;
131 UV uch; 192 UV uch;
132 193
133 if (is_utf8) 194 if (is_utf8)
134 { 195 {
135 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 196 uch = decode_utf8 (str, end - str, &clen);
136 if (clen == (STRLEN)-1) 197 if (clen == (STRLEN)-1)
137 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 198 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
138 } 199 }
139 else 200 else
140 { 201 {
… …
143 } 204 }
144 205
145 if (uch > 0x10FFFFUL) 206 if (uch > 0x10FFFFUL)
146 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 207 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
147 208
148 if (uch < 0x80 || enc->flags & F_ASCII) 209 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
149 { 210 {
150 if (uch > 0xFFFFUL) 211 if (uch > 0xFFFFUL)
151 { 212 {
152 need (enc, len += 11); 213 need (enc, len += 11);
153 sprintf (enc->cur, "\\u%04x\\u%04x", 214 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
167 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 228 *enc->cur++ = hexdigit [(uch >> 0) & 15];
168 } 229 }
169 230
170 str += clen; 231 str += clen;
171 } 232 }
233 else if (enc->json.flags & F_LATIN1)
234 {
235 *enc->cur++ = uch;
236 str += clen;
237 }
172 else if (is_utf8) 238 else if (is_utf8)
173 { 239 {
174 need (enc, len += clen); 240 need (enc, len += clen);
175 do 241 do
176 { 242 {
… …
178 } 244 }
179 while (--clen); 245 while (--clen);
180 } 246 }
181 else 247 else
182 { 248 {
183 need (enc, len += 10); // never more than 11 bytes needed 249 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
184 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 250 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
185 ++str; 251 ++str;
186 } 252 }
187 } 253 }
188 } 254 }
… …
190 256
191 --len; 257 --len;
192 } 258 }
193} 259}
194 260
195#define INDENT SB \ 261inline void
262encode_indent (enc_t *enc)
263{
196 if (enc->flags & F_INDENT) \ 264 if (enc->json.flags & F_INDENT)
197 { \ 265 {
198 int i_; \ 266 int spaces = enc->indent * INDENT_STEP;
199 need (enc, enc->indent); \ 267
200 for (i_ = enc->indent * 3; i_--; )\ 268 need (enc, spaces);
269 memset (enc->cur, ' ', spaces);
270 enc->cur += spaces;
271 }
272}
273
274inline void
275encode_space (enc_t *enc)
276{
277 need (enc, 1);
278 encode_ch (enc, ' ');
279}
280
281inline void
282encode_nl (enc_t *enc)
283{
284 if (enc->json.flags & F_INDENT)
285 {
286 need (enc, 1);
201 encode_ch (enc, ' '); \ 287 encode_ch (enc, '\n');
202 } \ 288 }
203 SE 289}
204 290
205#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 291inline void
206#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 292encode_comma (enc_t *enc)
207#define COMMA SB \ 293{
208 encode_ch (enc, ','); \ 294 encode_ch (enc, ',');
295
209 if (enc->flags & F_INDENT) \ 296 if (enc->json.flags & F_INDENT)
210 NL; \ 297 encode_nl (enc);
211 else if (enc->flags & F_SPACE_AFTER) \ 298 else if (enc->json.flags & F_SPACE_AFTER)
212 SPACE; \ 299 encode_space (enc);
213 SE 300}
214 301
215static void encode_sv (enc_t *enc, SV *sv); 302static void encode_sv (enc_t *enc, SV *sv);
216 303
217static void 304static void
218encode_av (enc_t *enc, AV *av) 305encode_av (enc_t *enc, AV *av)
219{ 306{
220 int i, len = av_len (av); 307 int i, len = av_len (av);
221 308
222 encode_ch (enc, '['); NL; 309 if (enc->indent >= enc->maxdepth)
310 croak ("data structure too deep (hit recursion limit)");
311
312 encode_ch (enc, '['); encode_nl (enc);
223 ++enc->indent; 313 ++enc->indent;
224 314
225 for (i = 0; i <= len; ++i) 315 for (i = 0; i <= len; ++i)
226 { 316 {
227 INDENT; 317 encode_indent (enc);
228 encode_sv (enc, *av_fetch (av, i, 0)); 318 encode_sv (enc, *av_fetch (av, i, 0));
229 319
230 if (i < len) 320 if (i < len)
231 COMMA; 321 encode_comma (enc);
232 } 322 }
233 323
234 NL; 324 encode_nl (enc);
235 325
236 --enc->indent; 326 --enc->indent;
237 INDENT; encode_ch (enc, ']'); 327 encode_indent (enc); encode_ch (enc, ']');
238} 328}
239 329
240static void 330static void
241encode_he (enc_t *enc, HE *he) 331encode_he (enc_t *enc, HE *he)
242{ 332{
… …
256 else 346 else
257 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 347 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
258 348
259 encode_ch (enc, '"'); 349 encode_ch (enc, '"');
260 350
261 if (enc->flags & F_SPACE_BEFORE) SPACE; 351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
262 encode_ch (enc, ':'); 352 encode_ch (enc, ':');
263 if (enc->flags & F_SPACE_AFTER ) SPACE; 353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
264 encode_sv (enc, HeVAL (he)); 354 encode_sv (enc, HeVAL (he));
265} 355}
266 356
267// compare hash entries, used when all keys are bytestrings 357// compare hash entries, used when all keys are bytestrings
268static int 358static int
… …
292static void 382static void
293encode_hv (enc_t *enc, HV *hv) 383encode_hv (enc_t *enc, HV *hv)
294{ 384{
295 int count, i; 385 int count, i;
296 386
387 if (enc->indent >= enc->maxdepth)
388 croak ("data structure too deep (hit recursion limit)");
389
297 encode_ch (enc, '{'); NL; ++enc->indent; 390 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
298 391
299 if ((count = hv_iterinit (hv))) 392 if ((count = hv_iterinit (hv)))
300 { 393 {
301 // for canonical output we have to sort by keys first 394 // for canonical output we have to sort by keys first
302 // actually, this is mostly due to the stupid so-called 395 // actually, this is mostly due to the stupid so-called
303 // security workaround added somewhere in 5.8.x. 396 // security workaround added somewhere in 5.8.x.
304 // that randomises hash orderings 397 // that randomises hash orderings
305 if (enc->flags & F_CANONICAL) 398 if (enc->json.flags & F_CANONICAL)
306 { 399 {
307 HE *he, *hes [count];
308 int fast = 1; 400 int fast = 1;
401 HE *he;
402#if defined(__BORLANDC__) || defined(_MSC_VER)
403 HE **hes = _alloca (count * sizeof (HE));
404#else
405 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
406#endif
309 407
310 i = 0; 408 i = 0;
311 while ((he = hv_iternext (hv))) 409 while ((he = hv_iternext (hv)))
312 { 410 {
313 hes [i++] = he; 411 hes [i++] = he;
… …
337 LEAVE; 435 LEAVE;
338 } 436 }
339 437
340 for (i = 0; i < count; ++i) 438 for (i = 0; i < count; ++i)
341 { 439 {
342 INDENT; 440 encode_indent (enc);
343 encode_he (enc, hes [i]); 441 encode_he (enc, hes [i]);
344 442
345 if (i < count - 1) 443 if (i < count - 1)
346 COMMA; 444 encode_comma (enc);
347 }
348
349 NL; 445 }
446
447 encode_nl (enc);
350 } 448 }
351 else 449 else
352 { 450 {
353 SV *sv;
354 HE *he = hv_iternext (hv); 451 HE *he = hv_iternext (hv);
355 452
356 for (;;) 453 for (;;)
357 { 454 {
358 INDENT; 455 encode_indent (enc);
359 encode_he (enc, he); 456 encode_he (enc, he);
360 457
361 if (!(he = hv_iternext (hv))) 458 if (!(he = hv_iternext (hv)))
362 break; 459 break;
363 460
364 COMMA; 461 encode_comma (enc);
365 }
366
367 NL; 462 }
463
464 encode_nl (enc);
368 } 465 }
369 } 466 }
370 467
371 --enc->indent; INDENT; encode_ch (enc, '}'); 468 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
469}
470
471// encode objects, arrays and special \0=false and \1=true values.
472static void
473encode_rv (enc_t *enc, SV *sv)
474{
475 svtype svt;
476
477 SvGETMAGIC (sv);
478 svt = SvTYPE (sv);
479
480 if (expect_false (SvOBJECT (sv)))
481 {
482 if (SvSTASH (sv) == json_boolean_stash)
483 {
484 if (SvIV (sv) == 0)
485 encode_str (enc, "false", 5, 0);
486 else
487 encode_str (enc, "true", 4, 0);
488 }
489 else
490 {
491#if 0
492 if (0 && sv_derived_from (rv, "JSON::Literal"))
493 {
494 // not yet
495 }
496#endif
497 if (enc->json.flags & F_CONV_BLESSED)
498 {
499 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
501
502 if (to_json)
503 {
504 dSP;
505 ENTER;
506 SAVETMPS;
507 PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509
510 // calling with G_SCALAR ensures that we always get a 1 reutrn value
511 // check anyways.
512 PUTBACK;
513 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
514 SPAGAIN;
515
516 encode_sv (enc, POPs);
517
518 FREETMPS;
519 LEAVE;
520 }
521 else if (enc->json.flags & F_ALLOW_BLESSED)
522 encode_str (enc, "null", 4, 0);
523 else
524 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
525 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
526 }
527 else if (enc->json.flags & F_ALLOW_BLESSED)
528 encode_str (enc, "null", 4, 0);
529 else
530 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
531 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
532 }
533 }
534 else if (svt == SVt_PVHV)
535 encode_hv (enc, (HV *)sv);
536 else if (svt == SVt_PVAV)
537 encode_av (enc, (AV *)sv);
538 else if (svt < SVt_PVAV)
539 {
540 if (SvNIOK (sv) && SvIV (sv) == 0)
541 encode_str (enc, "false", 5, 0);
542 else if (SvNIOK (sv) && SvIV (sv) == 1)
543 encode_str (enc, "true", 4, 0);
544 else
545 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547 }
548 else
549 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
550 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
372} 551}
373 552
374static void 553static void
375encode_sv (enc_t *enc, SV *sv) 554encode_sv (enc_t *enc, SV *sv)
376{ 555{
… …
384 encode_str (enc, str, len, SvUTF8 (sv)); 563 encode_str (enc, str, len, SvUTF8 (sv));
385 encode_ch (enc, '"'); 564 encode_ch (enc, '"');
386 } 565 }
387 else if (SvNOKp (sv)) 566 else if (SvNOKp (sv))
388 { 567 {
568 // trust that perl will do the right thing w.r.t. JSON syntax.
389 need (enc, NV_DIG + 32); 569 need (enc, NV_DIG + 32);
390 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 570 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
391 enc->cur += strlen (enc->cur); 571 enc->cur += strlen (enc->cur);
392 } 572 }
393 else if (SvIOKp (sv)) 573 else if (SvIOKp (sv))
394 { 574 {
395 need (enc, 64); 575 // we assume we can always read an IV as a UV
576 if (SvUV (sv) & ~(UV)0x7fff)
577 {
578 // large integer, use the (rather slow) snprintf way.
579 need (enc, sizeof (UV) * 3);
396 enc->cur += 580 enc->cur +=
397 SvIsUV(sv) 581 SvIsUV(sv)
398 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 582 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
399 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 583 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
584 }
585 else
586 {
587 // optimise the "small number case"
588 // code will likely be branchless and use only a single multiplication
589 I32 i = SvIV (sv);
590 U32 u;
591 char digit, nz = 0;
592
593 need (enc, 6);
594
595 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
596 u = i < 0 ? -i : i;
597
598 // convert to 4.28 fixed-point representation
599 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
600
601 // now output digit by digit, each time masking out the integer part
602 // and multiplying by 5 while moving the decimal point one to the right,
603 // resulting in a net multiplication by 10.
604 // we always write the digit to memory but conditionally increment
605 // the pointer, to ease the usage of conditional move instructions.
606 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
607 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
608 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
609 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
610 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
611 }
400 } 612 }
401 else if (SvROK (sv)) 613 else if (SvROK (sv))
402 { 614 encode_rv (enc, SvRV (sv));
403 SV *rv = SvRV (sv);
404
405 if (!--enc->max_recurse)
406 croak ("data structure too deep (hit recursion limit)");
407
408 switch (SvTYPE (rv))
409 {
410 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
411 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
412
413 default:
414 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
415 SvPV_nolen (sv));
416 }
417 }
418 else if (!SvOK (sv)) 615 else if (!SvOK (sv))
419 encode_str (enc, "null", 4, 0); 616 encode_str (enc, "null", 4, 0);
420 else 617 else
421 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 618 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
422 SvPV_nolen (sv), SvFLAGS (sv)); 619 SvPV_nolen (sv), SvFLAGS (sv));
423} 620}
424 621
425static SV * 622static SV *
426encode_json (SV *scalar, UV flags) 623encode_json (SV *scalar, struct json *json)
427{ 624{
625 enc_t enc;
626
428 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 627 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
429 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 628 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
430 629
431 enc_t enc; 630 enc.json = *json;
432 enc.flags = flags;
433 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
434 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
435 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
436 enc.max_recurse = 0;
437 enc.indent = 0; 634 enc.indent = 0;
635 enc.maxdepth = DEC_DEPTH (enc.json.flags);
438 636
439 SvPOK_only (enc.sv); 637 SvPOK_only (enc.sv);
440 encode_sv (&enc, scalar); 638 encode_sv (&enc, scalar);
441 639
640 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
641 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
642
442 if (!(flags & (F_ASCII | F_UTF8))) 643 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
443 SvUTF8_on (enc.sv); 644 SvUTF8_on (enc.sv);
444 645
445 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
446
447 if (enc.flags & F_SHRINK) 646 if (enc.json.flags & F_SHRINK)
448 shrink (enc.sv); 647 shrink (enc.sv);
449 648
450 return enc.sv; 649 return enc.sv;
451} 650}
452 651
453///////////////////////////////////////////////////////////////////////////// 652/////////////////////////////////////////////////////////////////////////////
653// decoder
454 654
455#define WS \ 655// structure used for decoding JSON
656typedef struct
657{
658 char *cur; // current parser pointer
659 char *end; // end of input string
660 const char *err; // parse error, if != 0
661 struct json json;
662 U32 depth; // recursion depth
663 U32 maxdepth; // recursion depth limit
664} dec_t;
665
666inline void
667decode_ws (dec_t *dec)
668{
456 for (;;) \ 669 for (;;)
457 { \ 670 {
458 char ch = *dec->cur; \ 671 char ch = *dec->cur;
672
459 if (ch > 0x20 \ 673 if (ch > 0x20
460 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 674 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
461 break; \ 675 break;
676
462 ++dec->cur; \ 677 ++dec->cur;
463 } 678 }
679}
464 680
465#define ERR(reason) SB dec->err = reason; goto fail; SE 681#define ERR(reason) SB dec->err = reason; goto fail; SE
682
466#define EXPECT_CH(ch) SB \ 683#define EXPECT_CH(ch) SB \
467 if (*dec->cur != ch) \ 684 if (*dec->cur != ch) \
468 ERR (# ch " expected"); \ 685 ERR (# ch " expected"); \
469 ++dec->cur; \ 686 ++dec->cur; \
470 SE 687 SE
471 688
689#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
690#define DEC_DEC_DEPTH --dec->depth
691
472static SV *decode_sv (dec_t *dec); 692static SV *decode_sv (dec_t *dec);
473 693
474static signed char decode_hexdigit[256]; 694static signed char decode_hexdigit[256];
475 695
476static UV 696static UV
477decode_4hex (dec_t *dec) 697decode_4hex (dec_t *dec)
478{ 698{
479 signed char d1, d2, d3, d4; 699 signed char d1, d2, d3, d4;
700 unsigned char *cur = (unsigned char *)dec->cur;
480 701
481 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 702 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
482 if (d1 < 0) ERR ("four hexadecimal digits expected"); 703 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
483 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 704 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
484 if (d2 < 0) ERR ("four hexadecimal digits expected"); 705 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
485 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
486 if (d3 < 0) ERR ("four hexadecimal digits expected");
487 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
488 if (d4 < 0) ERR ("four hexadecimal digits expected");
489 706
490 dec->cur += 4; 707 dec->cur += 4;
491 708
492 return ((UV)d1) << 12 709 return ((UV)d1) << 12
493 | ((UV)d2) << 8 710 | ((UV)d2) << 8
… …
496 713
497fail: 714fail:
498 return (UV)-1; 715 return (UV)-1;
499} 716}
500 717
501#define APPEND_GROW(n) SB \
502 if (cur + (n) >= end) \
503 { \
504 STRLEN ofs = cur - SvPVX (sv); \
505 SvGROW (sv, ofs + (n) + 1); \
506 cur = SvPVX (sv) + ofs; \
507 end = SvEND (sv); \
508 } \
509 SE
510
511#define APPEND_CH(ch) SB \
512 APPEND_GROW (1); \
513 *cur++ = (ch); \
514 SE
515
516static SV * 718static SV *
517decode_str (dec_t *dec) 719decode_str (dec_t *dec)
518{ 720{
519 SV *sv = NEWSV (0,2); 721 SV *sv = 0;
520 int utf8 = 0; 722 int utf8 = 0;
521 char *cur = SvPVX (sv); 723 char *dec_cur = dec->cur;
522 char *end = SvEND (sv);
523 724
524 for (;;) 725 do
525 { 726 {
526 unsigned char ch = *(unsigned char *)dec->cur; 727 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
728 char *cur = buf;
527 729
528 if (ch == '"') 730 do
529 break;
530 else if (ch == '\\')
531 { 731 {
532 switch (*++dec->cur) 732 unsigned char ch = *(unsigned char *)dec_cur++;
733
734 if (expect_false (ch == '"'))
735 {
736 --dec_cur;
737 break;
533 { 738 }
534 case '\\': 739 else if (expect_false (ch == '\\'))
535 case '/': 740 {
536 case '"': APPEND_CH (*dec->cur++); break; 741 switch (*dec_cur)
537
538 case 'b': APPEND_CH ('\010'); ++dec->cur; break;
539 case 't': APPEND_CH ('\011'); ++dec->cur; break;
540 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
541 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
542 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
543
544 case 'u':
545 { 742 {
546 UV lo, hi; 743 case '\\':
547 ++dec->cur; 744 case '/':
745 case '"': *cur++ = *dec_cur++; break;
548 746
549 hi = decode_4hex (dec); 747 case 'b': ++dec_cur; *cur++ = '\010'; break;
550 if (hi == (UV)-1) 748 case 't': ++dec_cur; *cur++ = '\011'; break;
551 goto fail; 749 case 'n': ++dec_cur; *cur++ = '\012'; break;
750 case 'f': ++dec_cur; *cur++ = '\014'; break;
751 case 'r': ++dec_cur; *cur++ = '\015'; break;
552 752
553 // possibly a surrogate pair 753 case 'u':
554 if (hi >= 0xd800 && hi < 0xdc00)
555 { 754 {
556 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 755 UV lo, hi;
557 ERR ("missing low surrogate character in surrogate pair"); 756 ++dec_cur;
558 757
559 dec->cur += 2; 758 dec->cur = dec_cur;
560
561 lo = decode_4hex (dec); 759 hi = decode_4hex (dec);
760 dec_cur = dec->cur;
562 if (lo == (UV)-1) 761 if (hi == (UV)-1)
563 goto fail; 762 goto fail;
564 763
764 // possibly a surrogate pair
765 if (hi >= 0xd800)
766 if (hi < 0xdc00)
767 {
768 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
769 ERR ("missing low surrogate character in surrogate pair");
770
771 dec_cur += 2;
772
773 dec->cur = dec_cur;
774 lo = decode_4hex (dec);
775 dec_cur = dec->cur;
776 if (lo == (UV)-1)
777 goto fail;
778
565 if (lo < 0xdc00 || lo >= 0xe000) 779 if (lo < 0xdc00 || lo >= 0xe000)
566 ERR ("surrogate pair expected"); 780 ERR ("surrogate pair expected");
567 781
568 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 782 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
783 }
784 else if (hi < 0xe000)
785 ERR ("missing high surrogate character in surrogate pair");
786
787 if (hi >= 0x80)
788 {
789 utf8 = 1;
790
791 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
792 }
793 else
794 *cur++ = hi;
569 } 795 }
570 else if (hi >= 0xdc00 && hi < 0xe000)
571 ERR ("missing high surrogate character in surrogate pair");
572
573 if (hi >= 0x80)
574 { 796 break;
575 utf8 = 1;
576 797
577 APPEND_GROW (4); // at most 4 bytes for 21 bits
578 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
579 }
580 else 798 default:
581 APPEND_CH (hi); 799 --dec_cur;
800 ERR ("illegal backslash escape sequence in string");
582 } 801 }
583 break; 802 }
803 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
804 *cur++ = ch;
805 else if (ch >= 0x80)
806 {
807 STRLEN clen;
808 UV uch;
584 809
585 default:
586 --dec->cur; 810 --dec_cur;
587 ERR ("illegal backslash escape sequence in string"); 811
812 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
813 if (clen == (STRLEN)-1)
814 ERR ("malformed UTF-8 character in JSON string");
815
816 do
817 *cur++ = *dec_cur++;
818 while (--clen);
819
820 utf8 = 1;
821 }
822 else
823 {
824 --dec_cur;
825
826 if (!ch)
827 ERR ("unexpected end of string while parsing JSON string");
828 else
829 ERR ("invalid character encountered while parsing JSON string");
588 } 830 }
589 } 831 }
590 else if (ch >= 0x20 && ch <= 0x7f) 832 while (cur < buf + SHORT_STRING_LEN);
591 APPEND_CH (*dec->cur++); 833
592 else if (ch >= 0x80)
593 { 834 {
594 STRLEN clen; 835 STRLEN len = cur - buf;
595 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
596 if (clen == (STRLEN)-1)
597 ERR ("malformed UTF-8 character in JSON string");
598 836
599 APPEND_GROW (clen); 837 if (sv)
600 do
601 { 838 {
602 *cur++ = *dec->cur++; 839 SvGROW (sv, SvCUR (sv) + len + 1);
840 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
841 SvCUR_set (sv, SvCUR (sv) + len);
603 } 842 }
604 while (--clen);
605
606 utf8 = 1;
607 }
608 else if (dec->cur == dec->end)
609 ERR ("unexpected end of string while parsing json string");
610 else 843 else
611 ERR ("invalid character encountered"); 844 sv = newSVpvn (buf, len);
612 } 845 }
846 }
847 while (*dec_cur != '"');
613 848
614 ++dec->cur; 849 ++dec_cur;
615 850
616 SvCUR_set (sv, cur - SvPVX (sv)); 851 if (sv)
617 852 {
618 SvPOK_only (sv); 853 SvPOK_only (sv);
619 *SvEND (sv) = 0; 854 *SvEND (sv) = 0;
620 855
621 if (utf8) 856 if (utf8)
622 SvUTF8_on (sv); 857 SvUTF8_on (sv);
858 }
859 else
860 sv = newSVpvn ("", 0);
623 861
624 if (dec->flags & F_SHRINK) 862 dec->cur = dec_cur;
625 shrink (sv);
626
627 return sv; 863 return sv;
628 864
629fail: 865fail:
630 SvREFCNT_dec (sv); 866 dec->cur = dec_cur;
631 return 0; 867 return 0;
632} 868}
633 869
634static SV * 870static SV *
635decode_num (dec_t *dec) 871decode_num (dec_t *dec)
… …
693 is_nv = 1; 929 is_nv = 1;
694 } 930 }
695 931
696 if (!is_nv) 932 if (!is_nv)
697 { 933 {
698 UV uv; 934 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
699 int numtype = grok_number (start, dec->cur - start, &uv); 935 if (*start == '-')
700 if (numtype & IS_NUMBER_IN_UV) 936 switch (dec->cur - start)
701 if (numtype & IS_NUMBER_NEG)
702 { 937 {
703 if (uv < (UV)IV_MIN) 938 case 2: return newSViv (-( start [1] - '0' * 1));
704 return newSViv (-(IV)uv); 939 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
940 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
941 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
705 } 942 }
943 else
944 switch (dec->cur - start)
945 {
946 case 1: return newSViv ( start [0] - '0' * 1);
947 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
948 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
949 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
950 }
951
952 {
953 UV uv;
954 int numtype = grok_number (start, dec->cur - start, &uv);
955 if (numtype & IS_NUMBER_IN_UV)
956 if (numtype & IS_NUMBER_NEG)
957 {
958 if (uv < (UV)IV_MIN)
959 return newSViv (-(IV)uv);
960 }
706 else 961 else
707 return newSVuv (uv); 962 return newSVuv (uv);
963
964 // here would likely be the place for bigint support
708 } 965 }
966 }
709 967
968 // if we ever support bigint or bigfloat, this is the place for bigfloat
710 return newSVnv (Atof (start)); 969 return newSVnv (Atof (start));
711 970
712fail: 971fail:
713 return 0; 972 return 0;
714} 973}
… …
716static SV * 975static SV *
717decode_av (dec_t *dec) 976decode_av (dec_t *dec)
718{ 977{
719 AV *av = newAV (); 978 AV *av = newAV ();
720 979
721 WS; 980 DEC_INC_DEPTH;
981 decode_ws (dec);
982
722 if (*dec->cur == ']') 983 if (*dec->cur == ']')
723 ++dec->cur; 984 ++dec->cur;
724 else 985 else
725 for (;;) 986 for (;;)
726 { 987 {
… …
730 if (!value) 991 if (!value)
731 goto fail; 992 goto fail;
732 993
733 av_push (av, value); 994 av_push (av, value);
734 995
735 WS; 996 decode_ws (dec);
736 997
737 if (*dec->cur == ']') 998 if (*dec->cur == ']')
738 { 999 {
739 ++dec->cur; 1000 ++dec->cur;
740 break; 1001 break;
… …
744 ERR (", or ] expected while parsing array"); 1005 ERR (", or ] expected while parsing array");
745 1006
746 ++dec->cur; 1007 ++dec->cur;
747 } 1008 }
748 1009
1010 DEC_DEC_DEPTH;
749 return newRV_noinc ((SV *)av); 1011 return newRV_noinc ((SV *)av);
750 1012
751fail: 1013fail:
752 SvREFCNT_dec (av); 1014 SvREFCNT_dec (av);
1015 DEC_DEC_DEPTH;
753 return 0; 1016 return 0;
754} 1017}
755 1018
756static SV * 1019static SV *
757decode_hv (dec_t *dec) 1020decode_hv (dec_t *dec)
758{ 1021{
759 HV *hv = newHV (); 1022 HV *hv = newHV ();
760 1023
761 WS; 1024 DEC_INC_DEPTH;
1025 decode_ws (dec);
1026
762 if (*dec->cur == '}') 1027 if (*dec->cur == '}')
763 ++dec->cur; 1028 ++dec->cur;
764 else 1029 else
765 for (;;) 1030 for (;;)
766 { 1031 {
1032 decode_ws (dec); EXPECT_CH ('"');
1033
1034 // heuristic: assume that
1035 // a) decode_str + hv_store_ent are abysmally slow.
1036 // b) most hash keys are short, simple ascii text.
1037 // => try to "fast-match" such strings to avoid
1038 // the overhead of decode_str + hv_store_ent.
1039 {
767 SV *key, *value; 1040 SV *value;
1041 char *p = dec->cur;
1042 char *e = p + 24; // only try up to 24 bytes
768 1043
769 WS; EXPECT_CH ('"'); 1044 for (;;)
770
771 key = decode_str (dec);
772 if (!key)
773 goto fail;
774
775 WS; EXPECT_CH (':');
776
777 value = decode_sv (dec);
778 if (!value)
779 { 1045 {
1046 // the >= 0x80 is true on most architectures
1047 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1048 {
1049 // slow path, back up and use decode_str
1050 SV *key = decode_str (dec);
1051 if (!key)
1052 goto fail;
1053
1054 decode_ws (dec); EXPECT_CH (':');
1055
1056 value = decode_sv (dec);
1057 if (!value)
1058 {
1059 SvREFCNT_dec (key);
1060 goto fail;
1061 }
1062
1063 hv_store_ent (hv, key, value, 0);
780 SvREFCNT_dec (key); 1064 SvREFCNT_dec (key);
1065
1066 break;
1067 }
1068 else if (*p == '"')
1069 {
1070 // fast path, got a simple key
1071 char *key = dec->cur;
1072 int len = p - key;
1073 dec->cur = p + 1;
1074
1075 decode_ws (dec); EXPECT_CH (':');
1076
1077 value = decode_sv (dec);
1078 if (!value)
781 goto fail; 1079 goto fail;
1080
1081 hv_store (hv, key, len, value, 0);
1082
1083 break;
1084 }
1085
1086 ++p;
782 } 1087 }
783
784 //TODO: optimise
785 hv_store_ent (hv, key, value, 0);
786
787 WS; 1088 }
1089
1090 decode_ws (dec);
788 1091
789 if (*dec->cur == '}') 1092 if (*dec->cur == '}')
790 { 1093 {
791 ++dec->cur; 1094 ++dec->cur;
792 break; 1095 break;
… …
796 ERR (", or } expected while parsing object/hash"); 1099 ERR (", or } expected while parsing object/hash");
797 1100
798 ++dec->cur; 1101 ++dec->cur;
799 } 1102 }
800 1103
1104 DEC_DEC_DEPTH;
801 return newRV_noinc ((SV *)hv); 1105 return newRV_noinc ((SV *)hv);
802 1106
803fail: 1107fail:
804 SvREFCNT_dec (hv); 1108 SvREFCNT_dec (hv);
1109 DEC_DEC_DEPTH;
805 return 0; 1110 return 0;
806} 1111}
807 1112
808static SV * 1113static SV *
809decode_sv (dec_t *dec) 1114decode_sv (dec_t *dec)
810{ 1115{
811 WS; 1116 decode_ws (dec);
1117
1118 // the beauty of JSON: you need exactly one character lookahead
1119 // to parse anything.
812 switch (*dec->cur) 1120 switch (*dec->cur)
813 { 1121 {
814 case '"': ++dec->cur; return decode_str (dec); 1122 case '"': ++dec->cur; return decode_str (dec);
815 case '[': ++dec->cur; return decode_av (dec); 1123 case '[': ++dec->cur; return decode_av (dec);
816 case '{': ++dec->cur; return decode_hv (dec); 1124 case '{': ++dec->cur; return decode_hv (dec);
… …
822 1130
823 case 't': 1131 case 't':
824 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1132 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
825 { 1133 {
826 dec->cur += 4; 1134 dec->cur += 4;
827 return newSViv (1); 1135 return SvREFCNT_inc (json_true);
828 } 1136 }
829 else 1137 else
830 ERR ("'true' expected"); 1138 ERR ("'true' expected");
831 1139
832 break; 1140 break;
833 1141
834 case 'f': 1142 case 'f':
835 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1143 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
836 { 1144 {
837 dec->cur += 5; 1145 dec->cur += 5;
838 return newSViv (0); 1146 return SvREFCNT_inc (json_false);
839 } 1147 }
840 else 1148 else
841 ERR ("'false' expected"); 1149 ERR ("'false' expected");
842 1150
843 break; 1151 break;
… …
852 ERR ("'null' expected"); 1160 ERR ("'null' expected");
853 1161
854 break; 1162 break;
855 1163
856 default: 1164 default:
857 ERR ("malformed json string, neither array, object, number, string or atom"); 1165 ERR ("malformed JSON string, neither array, object, number, string or atom");
858 break; 1166 break;
859 } 1167 }
860 1168
861fail: 1169fail:
862 return 0; 1170 return 0;
863} 1171}
864 1172
865static SV * 1173static SV *
866decode_json (SV *string, UV flags) 1174decode_json (SV *string, struct json *json, UV *offset_return)
867{ 1175{
1176 dec_t dec;
1177 UV offset;
868 SV *sv; 1178 SV *sv;
869 1179
1180 SvGETMAGIC (string);
1181 SvUPGRADE (string, SVt_PV);
1182
1183 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1184 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1185 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1186
870 if (flags & F_UTF8) 1187 if (json->flags & F_UTF8)
871 sv_utf8_downgrade (string, 0); 1188 sv_utf8_downgrade (string, 0);
872 else 1189 else
873 sv_utf8_upgrade (string); 1190 sv_utf8_upgrade (string);
874 1191
875 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1192 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
876 1193
877 dec_t dec; 1194 dec.json = *json;
878 dec.flags = flags;
879 dec.cur = SvPVX (string); 1195 dec.cur = SvPVX (string);
880 dec.end = SvEND (string); 1196 dec.end = SvEND (string);
881 dec.err = 0; 1197 dec.err = 0;
1198 dec.depth = 0;
1199 dec.maxdepth = DEC_DEPTH (dec.json.flags);
882 1200
1201 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
883 sv = decode_sv (&dec); 1202 sv = decode_sv (&dec);
884 1203
1204 if (!(offset_return || !sv))
1205 {
1206 // check for trailing garbage
1207 decode_ws (&dec);
1208
1209 if (*dec.cur)
1210 {
1211 dec.err = "garbage after JSON object";
1212 SvREFCNT_dec (sv);
1213 sv = 0;
1214 }
1215 }
1216
1217 if (offset_return || !sv)
1218 {
1219 offset = dec.json.flags & F_UTF8
1220 ? dec.cur - SvPVX (string)
1221 : utf8_distance (dec.cur, SvPVX (string));
1222
1223 if (offset_return)
1224 *offset_return = offset;
1225 }
1226
885 if (!sv) 1227 if (!sv)
886 { 1228 {
887 IV offset = dec.flags & F_UTF8
888 ? dec.cur - SvPVX (string)
889 : utf8_distance (dec.cur, SvPVX (string));
890 SV *uni = sv_newmortal (); 1229 SV *uni = sv_newmortal ();
891 1230
892 // horrible hack to silence warning inside pv_uni_display 1231 // horrible hack to silence warning inside pv_uni_display
893 COP cop = *PL_curcop; 1232 COP cop = *PL_curcop;
894 cop.cop_warnings = pWARN_NONE; 1233 cop.cop_warnings = pWARN_NONE;
… …
896 SAVEVPTR (PL_curcop); 1235 SAVEVPTR (PL_curcop);
897 PL_curcop = &cop; 1236 PL_curcop = &cop;
898 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1237 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
899 LEAVE; 1238 LEAVE;
900 1239
901 croak ("%s, at character offset %d (%s)", 1240 croak ("%s, at character offset %d [\"%s\"]",
902 dec.err, 1241 dec.err,
903 (int)offset, 1242 (int)offset,
904 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1243 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
905 } 1244 }
906 1245
907 sv = sv_2mortal (sv); 1246 sv = sv_2mortal (sv);
908 1247
909 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1248 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
910 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1249 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
911 1250
912 return sv; 1251 return sv;
913} 1252}
914 1253
1254/////////////////////////////////////////////////////////////////////////////
1255// XS interface functions
1256
915MODULE = JSON::XS PACKAGE = JSON::XS 1257MODULE = JSON::XS PACKAGE = JSON::XS
916 1258
917BOOT: 1259BOOT:
918{ 1260{
919 int i; 1261 int i;
920 1262
921 memset (decode_hexdigit, 0xff, 256);
922 for (i = 10; i--; ) 1263 for (i = 0; i < 256; ++i)
923 decode_hexdigit ['0' + i] = i; 1264 decode_hexdigit [i] =
1265 i >= '0' && i <= '9' ? i - '0'
1266 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1267 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1268 : -1;
924 1269
925 for (i = 7; i--; )
926 {
927 decode_hexdigit ['a' + i] = 10 + i;
928 decode_hexdigit ['A' + i] = 10 + i;
929 }
930
931 json_stash = gv_stashpv ("JSON::XS", 1); 1270 json_stash = gv_stashpv ("JSON::XS" , 1);
1271 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1272
1273 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1274 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
932} 1275}
933 1276
934PROTOTYPES: DISABLE 1277PROTOTYPES: DISABLE
935 1278
936SV *new (char *dummy) 1279SV *new (char *dummy)
… …
939 OUTPUT: 1282 OUTPUT:
940 RETVAL 1283 RETVAL
941 1284
942SV *ascii (SV *self, int enable = 1) 1285SV *ascii (SV *self, int enable = 1)
943 ALIAS: 1286 ALIAS:
944 ascii = F_ASCII 1287 ascii = F_ASCII
1288 latin1 = F_LATIN1
945 utf8 = F_UTF8 1289 utf8 = F_UTF8
946 indent = F_INDENT 1290 indent = F_INDENT
947 canonical = F_CANONICAL 1291 canonical = F_CANONICAL
948 space_before = F_SPACE_BEFORE 1292 space_before = F_SPACE_BEFORE
949 space_after = F_SPACE_AFTER 1293 space_after = F_SPACE_AFTER
950 pretty = F_PRETTY 1294 pretty = F_PRETTY
951 allow_nonref = F_ALLOW_NONREF 1295 allow_nonref = F_ALLOW_NONREF
952 shrink = F_SHRINK 1296 shrink = F_SHRINK
1297 allow_blessed = F_ALLOW_BLESSED
1298 convert_blessed = F_CONV_BLESSED
953 CODE: 1299 CODE:
954{ 1300{
955 UV *uv = SvJSON (self); 1301 UV *uv = SvJSON (self);
956 if (enable) 1302 if (enable)
957 *uv |= ix; 1303 *uv |= ix;
… …
961 RETVAL = newSVsv (self); 1307 RETVAL = newSVsv (self);
962} 1308}
963 OUTPUT: 1309 OUTPUT:
964 RETVAL 1310 RETVAL
965 1311
1312SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1313 CODE:
1314{
1315 UV *uv = SvJSON (self);
1316 UV log2 = 0;
1317
1318 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1319
1320 while ((1UL << log2) < max_depth)
1321 ++log2;
1322
1323 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1324
1325 RETVAL = newSVsv (self);
1326}
1327 OUTPUT:
1328 RETVAL
1329
1330SV *max_size (SV *self, UV max_size = 0)
1331 CODE:
1332{
1333 UV *uv = SvJSON (self);
1334 UV log2 = 0;
1335
1336 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1337 if (max_size == 1) max_size = 2;
1338
1339 while ((1UL << log2) < max_size)
1340 ++log2;
1341
1342 *uv = *uv & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1343
1344 RETVAL = newSVsv (self);
1345}
1346 OUTPUT:
1347 RETVAL
1348
966void encode (SV *self, SV *scalar) 1349void encode (SV *self, SV *scalar)
967 PPCODE: 1350 PPCODE:
1351{
1352 struct json json = { *SvJSON (self) };
968 XPUSHs (encode_json (scalar, *SvJSON (self))); 1353 XPUSHs (encode_json (scalar, &json));
1354}
969 1355
970void decode (SV *self, SV *jsonstr) 1356void decode (SV *self, SV *jsonstr)
971 PPCODE: 1357 PPCODE:
1358{
1359 struct json json = { *SvJSON (self) };
972 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1360 XPUSHs (decode_json (jsonstr, &json, 0));
1361}
1362
1363void decode_prefix (SV *self, SV *jsonstr)
1364 PPCODE:
1365{
1366 UV offset;
1367 struct json json = { *SvJSON (self) };
1368 EXTEND (SP, 2);
1369 PUSHs (decode_json (jsonstr, &json, &offset));
1370 PUSHs (sv_2mortal (newSVuv (offset)));
1371}
973 1372
974PROTOTYPES: ENABLE 1373PROTOTYPES: ENABLE
975 1374
976void to_json (SV *scalar) 1375void to_json (SV *scalar)
977 PPCODE: 1376 PPCODE:
1377{
1378 struct json json = { F_DEFAULT | F_UTF8 };
978 XPUSHs (encode_json (scalar, F_UTF8)); 1379 XPUSHs (encode_json (scalar, &json));
1380}
979 1381
980void from_json (SV *jsonstr) 1382void from_json (SV *jsonstr)
981 PPCODE: 1383 PPCODE:
1384{
1385 struct json json = { F_DEFAULT | F_UTF8 };
982 XPUSHs (decode_json (jsonstr, F_UTF8)); 1386 XPUSHs (decode_json (jsonstr, &json, 0));
1387}
983 1388

Diff Legend

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