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 1.12 by root, Sat Mar 24 22:10:08 2007 UTC vs.
Revision 1.50 by root, Mon Jul 2 00:29:38 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
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 39
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 41#define F_DEFAULT (9UL << S_MAXDEPTH)
20 42
21#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
23 45
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#define SHORT_STRING_LEN 256 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
26 47
27#define SB do { 48#define SB do {
28#define SE } while (0) 49#define SE } while (0)
29 50
51#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value))
53# define inline inline
54#else
55# define expect(expr,value) (expr)
56# define inline static
57#endif
58
59#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1)
61
30static HV *json_stash; // JSON::XS:: 62static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false;
64
65typedef struct {
66 U32 flags;
67 SV *cb_object, *cb_sk_object;
68} JSON;
31 69
32///////////////////////////////////////////////////////////////////////////// 70/////////////////////////////////////////////////////////////////////////////
33// utility functions 71// utility functions
34 72
35static UV * 73inline void
36SvJSON (SV *sv)
37{
38 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
39 croak ("object is not of type JSON::XS");
40
41 return &SvUVX (SvRV (sv));
42}
43
44static void
45shrink (SV *sv) 74shrink (SV *sv)
46{ 75{
47 sv_utf8_downgrade (sv, 1); 76 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 77 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 78 {
… …
53 SvPV_renew (sv, SvCUR (sv) + 1); 82 SvPV_renew (sv, SvCUR (sv) + 1);
54#endif 83#endif
55 } 84 }
56} 85}
57 86
87// decode an utf-8 character and return it, or (UV)-1 in
88// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints
92inline UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
98 {
99 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 }
102 else
103 {
104 *clen = (STRLEN)-1;
105 return (UV)-1;
106 }
107}
108
58///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
59// encoder 110// encoder
60 111
61// structure used for encoding JSON 112// structure used for encoding JSON
62typedef struct 113typedef struct
63{ 114{
64 char *cur; // SvPVX (sv) + current output position 115 char *cur; // SvPVX (sv) + current output position
65 char *end; // SvEND (sv) 116 char *end; // SvEND (sv)
66 SV *sv; // result scalar 117 SV *sv; // result scalar
67 UV flags; // F_* 118 JSON json;
68 int indent; // indentation level 119 U32 indent; // indentation level
69 int max_depth; // max. recursion level 120 U32 maxdepth; // max. indentation/recursion level
70} enc_t; 121} enc_t;
71 122
72static void 123inline void
73need (enc_t *enc, STRLEN len) 124need (enc_t *enc, STRLEN len)
74{ 125{
75 if (enc->cur + len >= enc->end) 126 if (expect_false (enc->cur + len >= enc->end))
76 { 127 {
77 STRLEN cur = enc->cur - SvPVX (enc->sv); 128 STRLEN cur = enc->cur - SvPVX (enc->sv);
78 SvGROW (enc->sv, cur + len + 1); 129 SvGROW (enc->sv, cur + len + 1);
79 enc->cur = SvPVX (enc->sv) + cur; 130 enc->cur = SvPVX (enc->sv) + cur;
80 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
81 } 132 }
82} 133}
83 134
84static void 135inline void
85encode_ch (enc_t *enc, char ch) 136encode_ch (enc_t *enc, char ch)
86{ 137{
87 need (enc, 1); 138 need (enc, 1);
88 *enc->cur++ = ch; 139 *enc->cur++ = ch;
89} 140}
… …
97 148
98 while (str < end) 149 while (str < end)
99 { 150 {
100 unsigned char ch = *(unsigned char *)str; 151 unsigned char ch = *(unsigned char *)str;
101 152
102 if (ch >= 0x20 && ch < 0x80) // most common case 153 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
103 { 154 {
104 if (ch == '"') // but with slow exceptions 155 if (expect_false (ch == '"')) // but with slow exceptions
105 { 156 {
106 need (enc, len += 1); 157 need (enc, len += 1);
107 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
108 *enc->cur++ = '"'; 159 *enc->cur++ = '"';
109 } 160 }
110 else if (ch == '\\') 161 else if (expect_false (ch == '\\'))
111 { 162 {
112 need (enc, len += 1); 163 need (enc, len += 1);
113 *enc->cur++ = '\\'; 164 *enc->cur++ = '\\';
114 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
115 } 166 }
… …
133 STRLEN clen; 184 STRLEN clen;
134 UV uch; 185 UV uch;
135 186
136 if (is_utf8) 187 if (is_utf8)
137 { 188 {
138 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 189 uch = decode_utf8 (str, end - str, &clen);
139 if (clen == (STRLEN)-1) 190 if (clen == (STRLEN)-1)
140 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 191 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
141 } 192 }
142 else 193 else
143 { 194 {
… …
146 } 197 }
147 198
148 if (uch > 0x10FFFFUL) 199 if (uch > 0x10FFFFUL)
149 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
150 201
151 if (uch < 0x80 || enc->flags & F_ASCII) 202 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
152 { 203 {
153 if (uch > 0xFFFFUL) 204 if (uch > 0xFFFFUL)
154 { 205 {
155 need (enc, len += 11); 206 need (enc, len += 11);
156 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
170 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
171 } 222 }
172 223
173 str += clen; 224 str += clen;
174 } 225 }
226 else if (enc->json.flags & F_LATIN1)
227 {
228 *enc->cur++ = uch;
229 str += clen;
230 }
175 else if (is_utf8) 231 else if (is_utf8)
176 { 232 {
177 need (enc, len += clen); 233 need (enc, len += clen);
178 do 234 do
179 { 235 {
… …
181 } 237 }
182 while (--clen); 238 while (--clen);
183 } 239 }
184 else 240 else
185 { 241 {
186 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
187 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
188 ++str; 244 ++str;
189 } 245 }
190 } 246 }
191 } 247 }
… …
193 249
194 --len; 250 --len;
195 } 251 }
196} 252}
197 253
198static void 254inline void
199encode_indent (enc_t *enc) 255encode_indent (enc_t *enc)
200{ 256{
201 if (enc->flags & F_INDENT) 257 if (enc->json.flags & F_INDENT)
202 { 258 {
203 int spaces = enc->indent * INDENT_STEP; 259 int spaces = enc->indent * INDENT_STEP;
204 260
205 need (enc, spaces); 261 need (enc, spaces);
206 memset (enc->cur, ' ', spaces); 262 memset (enc->cur, ' ', spaces);
207 enc->cur += spaces; 263 enc->cur += spaces;
208 } 264 }
209} 265}
210 266
211static void 267inline void
212encode_space (enc_t *enc) 268encode_space (enc_t *enc)
213{ 269{
214 need (enc, 1); 270 need (enc, 1);
215 encode_ch (enc, ' '); 271 encode_ch (enc, ' ');
216} 272}
217 273
218static void 274inline void
219encode_nl (enc_t *enc) 275encode_nl (enc_t *enc)
220{ 276{
221 if (enc->flags & F_INDENT) 277 if (enc->json.flags & F_INDENT)
222 { 278 {
223 need (enc, 1); 279 need (enc, 1);
224 encode_ch (enc, '\n'); 280 encode_ch (enc, '\n');
225 } 281 }
226} 282}
227 283
228static void 284inline void
229encode_comma (enc_t *enc) 285encode_comma (enc_t *enc)
230{ 286{
231 encode_ch (enc, ','); 287 encode_ch (enc, ',');
232 288
233 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
234 encode_nl (enc); 290 encode_nl (enc);
235 else if (enc->flags & F_SPACE_AFTER) 291 else if (enc->json.flags & F_SPACE_AFTER)
236 encode_space (enc); 292 encode_space (enc);
237} 293}
238 294
239static void encode_sv (enc_t *enc, SV *sv); 295static void encode_sv (enc_t *enc, SV *sv);
240 296
241static void 297static void
242encode_av (enc_t *enc, AV *av) 298encode_av (enc_t *enc, AV *av)
243{ 299{
244 int i, len = av_len (av); 300 int i, len = av_len (av);
301
302 if (enc->indent >= enc->maxdepth)
303 croak ("data structure too deep (hit recursion limit)");
245 304
246 encode_ch (enc, '['); encode_nl (enc); 305 encode_ch (enc, '['); encode_nl (enc);
247 ++enc->indent; 306 ++enc->indent;
248 307
249 for (i = 0; i <= len; ++i) 308 for (i = 0; i <= len; ++i)
… …
280 else 339 else
281 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
282 341
283 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
284 343
285 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 344 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
286 encode_ch (enc, ':'); 345 encode_ch (enc, ':');
287 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 346 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
288 encode_sv (enc, HeVAL (he)); 347 encode_sv (enc, HeVAL (he));
289} 348}
290 349
291// compare hash entries, used when all keys are bytestrings 350// compare hash entries, used when all keys are bytestrings
292static int 351static int
… …
315 374
316static void 375static void
317encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
318{ 377{
319 int count, i; 378 int count, i;
379
380 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)");
320 382
321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
322 384
323 if ((count = hv_iterinit (hv))) 385 if ((count = hv_iterinit (hv)))
324 { 386 {
325 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
326 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
327 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
328 // that randomises hash orderings 390 // that randomises hash orderings
329 if (enc->flags & F_CANONICAL) 391 if (enc->json.flags & F_CANONICAL)
330 { 392 {
331 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
332 int fast = 1; 393 int fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE));
397#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif
333 400
334 i = 0; 401 i = 0;
335 while ((he = hv_iternext (hv))) 402 while ((he = hv_iternext (hv)))
336 { 403 {
337 hes [i++] = he; 404 hes [i++] = he;
… …
372 439
373 encode_nl (enc); 440 encode_nl (enc);
374 } 441 }
375 else 442 else
376 { 443 {
377 SV *sv;
378 HE *he = hv_iternext (hv); 444 HE *he = hv_iternext (hv);
379 445
380 for (;;) 446 for (;;)
381 { 447 {
382 encode_indent (enc); 448 encode_indent (enc);
… …
391 encode_nl (enc); 457 encode_nl (enc);
392 } 458 }
393 } 459 }
394 460
395 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 461 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462}
463
464// encode objects, arrays and special \0=false and \1=true values.
465static void
466encode_rv (enc_t *enc, SV *sv)
467{
468 svtype svt;
469
470 SvGETMAGIC (sv);
471 svt = SvTYPE (sv);
472
473 if (expect_false (SvOBJECT (sv)))
474 {
475 if (SvSTASH (sv) == json_boolean_stash)
476 {
477 if (SvIV (sv) == 0)
478 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 }
482 else
483 {
484#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 {
487 // not yet
488 }
489#endif
490 if (enc->json.flags & F_CONV_BLESSED)
491 {
492 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
494
495 if (to_json)
496 {
497 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
498 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
499
500 // calling with G_SCALAR ensures that we always get a 1 reutrn value
501 // check anyways.
502 PUTBACK;
503 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
504 SPAGAIN;
505
506 encode_sv (enc, POPs);
507
508 FREETMPS; LEAVE;
509 }
510 else if (enc->json.flags & F_ALLOW_BLESSED)
511 encode_str (enc, "null", 4, 0);
512 else
513 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
514 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
515 }
516 else if (enc->json.flags & F_ALLOW_BLESSED)
517 encode_str (enc, "null", 4, 0);
518 else
519 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
520 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
521 }
522 }
523 else if (svt == SVt_PVHV)
524 encode_hv (enc, (HV *)sv);
525 else if (svt == SVt_PVAV)
526 encode_av (enc, (AV *)sv);
527 else if (svt < SVt_PVAV)
528 {
529 STRLEN len = 0;
530 char *pv = svt ? SvPV (sv, len) : 0;
531
532 if (len == 1 && *pv == '0')
533 encode_str (enc, "false", 5, 0);
534 else if (len == 1 && *pv == '1')
535 encode_str (enc, "true", 4, 0);
536 else
537 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
538 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
539 }
540 else
541 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
396} 543}
397 544
398static void 545static void
399encode_sv (enc_t *enc, SV *sv) 546encode_sv (enc_t *enc, SV *sv)
400{ 547{
… …
408 encode_str (enc, str, len, SvUTF8 (sv)); 555 encode_str (enc, str, len, SvUTF8 (sv));
409 encode_ch (enc, '"'); 556 encode_ch (enc, '"');
410 } 557 }
411 else if (SvNOKp (sv)) 558 else if (SvNOKp (sv))
412 { 559 {
560 // trust that perl will do the right thing w.r.t. JSON syntax.
413 need (enc, NV_DIG + 32); 561 need (enc, NV_DIG + 32);
414 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 562 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
415 enc->cur += strlen (enc->cur); 563 enc->cur += strlen (enc->cur);
416 } 564 }
417 else if (SvIOKp (sv)) 565 else if (SvIOKp (sv))
418 { 566 {
419 need (enc, 64); 567 // we assume we can always read an IV as a UV
568 if (SvUV (sv) & ~(UV)0x7fff)
569 {
570 // large integer, use the (rather slow) snprintf way.
571 need (enc, sizeof (UV) * 3);
420 enc->cur += 572 enc->cur +=
421 SvIsUV(sv) 573 SvIsUV(sv)
422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 574 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 575 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
576 }
577 else
578 {
579 // optimise the "small number case"
580 // code will likely be branchless and use only a single multiplication
581 I32 i = SvIV (sv);
582 U32 u;
583 char digit, nz = 0;
584
585 need (enc, 6);
586
587 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
588 u = i < 0 ? -i : i;
589
590 // convert to 4.28 fixed-point representation
591 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
592
593 // now output digit by digit, each time masking out the integer part
594 // and multiplying by 5 while moving the decimal point one to the right,
595 // resulting in a net multiplication by 10.
596 // we always write the digit to memory but conditionally increment
597 // the pointer, to ease the usage of conditional move instructions.
598 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
599 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
600 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
601 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
602 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
603 }
424 } 604 }
425 else if (SvROK (sv)) 605 else if (SvROK (sv))
426 { 606 encode_rv (enc, SvRV (sv));
427 SV *rv = SvRV (sv);
428
429 if (enc->indent >= enc->max_depth)
430 croak ("data structure too deep (hit recursion limit)");
431
432 switch (SvTYPE (rv))
433 {
434 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
435 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
436
437 default:
438 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
439 SvPV_nolen (sv));
440 }
441 }
442 else if (!SvOK (sv)) 607 else if (!SvOK (sv))
443 encode_str (enc, "null", 4, 0); 608 encode_str (enc, "null", 4, 0);
444 else 609 else
445 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 610 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
446 SvPV_nolen (sv), SvFLAGS (sv)); 611 SvPV_nolen (sv), SvFLAGS (sv));
447} 612}
448 613
449static SV * 614static SV *
450encode_json (SV *scalar, UV flags) 615encode_json (SV *scalar, JSON *json)
451{ 616{
617 enc_t enc;
618
452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 619 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
453 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 620 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
454 621
455 enc_t enc; 622 enc.json = *json;
456 enc.flags = flags;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 623 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 624 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 625 enc.end = SvEND (enc.sv);
460 enc.indent = 0; 626 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL; 627 enc.maxdepth = DEC_DEPTH (enc.json.flags);
462 628
463 SvPOK_only (enc.sv); 629 SvPOK_only (enc.sv);
464 encode_sv (&enc, scalar); 630 encode_sv (&enc, scalar);
465 631
632 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
633 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
634
466 if (!(flags & (F_ASCII | F_UTF8))) 635 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
467 SvUTF8_on (enc.sv); 636 SvUTF8_on (enc.sv);
468 637
469 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
470
471 if (enc.flags & F_SHRINK) 638 if (enc.json.flags & F_SHRINK)
472 shrink (enc.sv); 639 shrink (enc.sv);
473 640
474 return enc.sv; 641 return enc.sv;
475} 642}
476 643
… …
481typedef struct 648typedef struct
482{ 649{
483 char *cur; // current parser pointer 650 char *cur; // current parser pointer
484 char *end; // end of input string 651 char *end; // end of input string
485 const char *err; // parse error, if != 0 652 const char *err; // parse error, if != 0
486 UV flags; // F_* 653 JSON json;
654 U32 depth; // recursion depth
655 U32 maxdepth; // recursion depth limit
487} dec_t; 656} dec_t;
488 657
489static void 658inline void
490decode_ws (dec_t *dec) 659decode_ws (dec_t *dec)
491{ 660{
492 for (;;) 661 for (;;)
493 { 662 {
494 char ch = *dec->cur; 663 char ch = *dec->cur;
… …
500 ++dec->cur; 669 ++dec->cur;
501 } 670 }
502} 671}
503 672
504#define ERR(reason) SB dec->err = reason; goto fail; SE 673#define ERR(reason) SB dec->err = reason; goto fail; SE
674
505#define EXPECT_CH(ch) SB \ 675#define EXPECT_CH(ch) SB \
506 if (*dec->cur != ch) \ 676 if (*dec->cur != ch) \
507 ERR (# ch " expected"); \ 677 ERR (# ch " expected"); \
508 ++dec->cur; \ 678 ++dec->cur; \
509 SE 679 SE
510 680
681#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
682#define DEC_DEC_DEPTH --dec->depth
683
511static SV *decode_sv (dec_t *dec); 684static SV *decode_sv (dec_t *dec);
512 685
513static signed char decode_hexdigit[256]; 686static signed char decode_hexdigit[256];
514 687
515static UV 688static UV
516decode_4hex (dec_t *dec) 689decode_4hex (dec_t *dec)
517{ 690{
518 signed char d1, d2, d3, d4; 691 signed char d1, d2, d3, d4;
519 unsigned char *cur = (unsigned char *)dec->cur; 692 unsigned char *cur = (unsigned char *)dec->cur;
520 693
521 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 694 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
522 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 695 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
523 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 696 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
524 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 697 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
525 698
526 dec->cur += 4; 699 dec->cur += 4;
527 700
528 return ((UV)d1) << 12 701 return ((UV)d1) << 12
529 | ((UV)d2) << 8 702 | ((UV)d2) << 8
… …
537static SV * 710static SV *
538decode_str (dec_t *dec) 711decode_str (dec_t *dec)
539{ 712{
540 SV *sv = 0; 713 SV *sv = 0;
541 int utf8 = 0; 714 int utf8 = 0;
715 char *dec_cur = dec->cur;
542 716
543 do 717 do
544 { 718 {
545 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 719 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
546 char *cur = buf; 720 char *cur = buf;
547 721
548 do 722 do
549 { 723 {
550 unsigned char ch = *(unsigned char *)dec->cur++; 724 unsigned char ch = *(unsigned char *)dec_cur++;
551 725
552 if (ch == '"') 726 if (expect_false (ch == '"'))
553 { 727 {
554 --dec->cur; 728 --dec_cur;
555 break; 729 break;
556 } 730 }
557 else if (ch == '\\') 731 else if (expect_false (ch == '\\'))
558 { 732 {
559 switch (*dec->cur) 733 switch (*dec_cur)
560 { 734 {
561 case '\\': 735 case '\\':
562 case '/': 736 case '/':
563 case '"': *cur++ = *dec->cur++; break; 737 case '"': *cur++ = *dec_cur++; break;
564 738
565 case 'b': ++dec->cur; *cur++ = '\010'; break; 739 case 'b': ++dec_cur; *cur++ = '\010'; break;
566 case 't': ++dec->cur; *cur++ = '\011'; break; 740 case 't': ++dec_cur; *cur++ = '\011'; break;
567 case 'n': ++dec->cur; *cur++ = '\012'; break; 741 case 'n': ++dec_cur; *cur++ = '\012'; break;
568 case 'f': ++dec->cur; *cur++ = '\014'; break; 742 case 'f': ++dec_cur; *cur++ = '\014'; break;
569 case 'r': ++dec->cur; *cur++ = '\015'; break; 743 case 'r': ++dec_cur; *cur++ = '\015'; break;
570 744
571 case 'u': 745 case 'u':
572 { 746 {
573 UV lo, hi; 747 UV lo, hi;
574 ++dec->cur; 748 ++dec_cur;
575 749
750 dec->cur = dec_cur;
576 hi = decode_4hex (dec); 751 hi = decode_4hex (dec);
752 dec_cur = dec->cur;
577 if (hi == (UV)-1) 753 if (hi == (UV)-1)
578 goto fail; 754 goto fail;
579 755
580 // possibly a surrogate pair 756 // possibly a surrogate pair
581 if (hi >= 0xd800) 757 if (hi >= 0xd800)
582 if (hi < 0xdc00) 758 if (hi < 0xdc00)
583 { 759 {
584 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 760 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
585 ERR ("missing low surrogate character in surrogate pair"); 761 ERR ("missing low surrogate character in surrogate pair");
586 762
587 dec->cur += 2; 763 dec_cur += 2;
588 764
765 dec->cur = dec_cur;
589 lo = decode_4hex (dec); 766 lo = decode_4hex (dec);
767 dec_cur = dec->cur;
590 if (lo == (UV)-1) 768 if (lo == (UV)-1)
591 goto fail; 769 goto fail;
592 770
593 if (lo < 0xdc00 || lo >= 0xe000) 771 if (lo < 0xdc00 || lo >= 0xe000)
594 ERR ("surrogate pair expected"); 772 ERR ("surrogate pair expected");
… …
608 *cur++ = hi; 786 *cur++ = hi;
609 } 787 }
610 break; 788 break;
611 789
612 default: 790 default:
613 --dec->cur; 791 --dec_cur;
614 ERR ("illegal backslash escape sequence in string"); 792 ERR ("illegal backslash escape sequence in string");
615 } 793 }
616 } 794 }
617 else if (ch >= 0x20 && ch <= 0x7f) 795 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
618 *cur++ = ch; 796 *cur++ = ch;
619 else if (ch >= 0x80) 797 else if (ch >= 0x80)
620 { 798 {
621 --dec->cur;
622
623 STRLEN clen; 799 STRLEN clen;
624 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 800 UV uch;
801
802 --dec_cur;
803
804 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
625 if (clen == (STRLEN)-1) 805 if (clen == (STRLEN)-1)
626 ERR ("malformed UTF-8 character in JSON string"); 806 ERR ("malformed UTF-8 character in JSON string");
627 807
628 do 808 do
629 {
630 *cur++ = *dec->cur++; 809 *cur++ = *dec_cur++;
631 }
632 while (--clen); 810 while (--clen);
633 811
634 utf8 = 1; 812 utf8 = 1;
635 } 813 }
636 else if (!ch)
637 ERR ("unexpected end of string while parsing json string");
638 else 814 else
815 {
816 --dec_cur;
817
818 if (!ch)
819 ERR ("unexpected end of string while parsing JSON string");
820 else
639 ERR ("invalid character encountered"); 821 ERR ("invalid character encountered while parsing JSON string");
640 822 }
641 } 823 }
642 while (cur < buf + SHORT_STRING_LEN); 824 while (cur < buf + SHORT_STRING_LEN);
643 825
826 {
644 STRLEN len = cur - buf; 827 STRLEN len = cur - buf;
645 828
646 if (sv) 829 if (sv)
647 { 830 {
648 SvGROW (sv, SvCUR (sv) + len + 1); 831 SvGROW (sv, SvCUR (sv) + len + 1);
649 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 832 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
650 SvCUR_set (sv, SvCUR (sv) + len); 833 SvCUR_set (sv, SvCUR (sv) + len);
651 } 834 }
652 else 835 else
653 sv = newSVpvn (buf, len); 836 sv = newSVpvn (buf, len);
654 } 837 }
838 }
655 while (*dec->cur != '"'); 839 while (*dec_cur != '"');
656 840
657 ++dec->cur; 841 ++dec_cur;
658 842
659 if (sv) 843 if (sv)
660 { 844 {
661 SvPOK_only (sv); 845 SvPOK_only (sv);
662 *SvEND (sv) = 0; 846 *SvEND (sv) = 0;
… …
665 SvUTF8_on (sv); 849 SvUTF8_on (sv);
666 } 850 }
667 else 851 else
668 sv = newSVpvn ("", 0); 852 sv = newSVpvn ("", 0);
669 853
854 dec->cur = dec_cur;
670 return sv; 855 return sv;
671 856
672fail: 857fail:
858 dec->cur = dec_cur;
673 return 0; 859 return 0;
674} 860}
675 861
676static SV * 862static SV *
677decode_num (dec_t *dec) 863decode_num (dec_t *dec)
… …
735 is_nv = 1; 921 is_nv = 1;
736 } 922 }
737 923
738 if (!is_nv) 924 if (!is_nv)
739 { 925 {
740 UV uv; 926 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
741 int numtype = grok_number (start, dec->cur - start, &uv); 927 if (*start == '-')
742 if (numtype & IS_NUMBER_IN_UV) 928 switch (dec->cur - start)
743 if (numtype & IS_NUMBER_NEG)
744 { 929 {
745 if (uv < (UV)IV_MIN) 930 case 2: return newSViv (-( start [1] - '0' * 1));
746 return newSViv (-(IV)uv); 931 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
932 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
933 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
747 } 934 }
935 else
936 switch (dec->cur - start)
937 {
938 case 1: return newSViv ( start [0] - '0' * 1);
939 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
940 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
941 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
942 }
943
944 {
945 UV uv;
946 int numtype = grok_number (start, dec->cur - start, &uv);
947 if (numtype & IS_NUMBER_IN_UV)
948 if (numtype & IS_NUMBER_NEG)
949 {
950 if (uv < (UV)IV_MIN)
951 return newSViv (-(IV)uv);
952 }
748 else 953 else
749 return newSVuv (uv); 954 return newSVuv (uv);
955
956 // here would likely be the place for bigint support
750 } 957 }
958 }
751 959
960 // if we ever support bigint or bigfloat, this is the place for bigfloat
752 return newSVnv (Atof (start)); 961 return newSVnv (Atof (start));
753 962
754fail: 963fail:
755 return 0; 964 return 0;
756} 965}
… …
758static SV * 967static SV *
759decode_av (dec_t *dec) 968decode_av (dec_t *dec)
760{ 969{
761 AV *av = newAV (); 970 AV *av = newAV ();
762 971
972 DEC_INC_DEPTH;
763 decode_ws (dec); 973 decode_ws (dec);
974
764 if (*dec->cur == ']') 975 if (*dec->cur == ']')
765 ++dec->cur; 976 ++dec->cur;
766 else 977 else
767 for (;;) 978 for (;;)
768 { 979 {
… …
786 ERR (", or ] expected while parsing array"); 997 ERR (", or ] expected while parsing array");
787 998
788 ++dec->cur; 999 ++dec->cur;
789 } 1000 }
790 1001
1002 DEC_DEC_DEPTH;
791 return newRV_noinc ((SV *)av); 1003 return newRV_noinc ((SV *)av);
792 1004
793fail: 1005fail:
794 SvREFCNT_dec (av); 1006 SvREFCNT_dec (av);
1007 DEC_DEC_DEPTH;
795 return 0; 1008 return 0;
796} 1009}
797 1010
798static SV * 1011static SV *
799decode_hv (dec_t *dec) 1012decode_hv (dec_t *dec)
800{ 1013{
1014 SV *sv;
801 HV *hv = newHV (); 1015 HV *hv = newHV ();
802 1016
1017 DEC_INC_DEPTH;
803 decode_ws (dec); 1018 decode_ws (dec);
1019
804 if (*dec->cur == '}') 1020 if (*dec->cur == '}')
805 ++dec->cur; 1021 ++dec->cur;
806 else 1022 else
807 for (;;) 1023 for (;;)
808 { 1024 {
809 SV *key, *value;
810
811 decode_ws (dec); EXPECT_CH ('"'); 1025 decode_ws (dec); EXPECT_CH ('"');
812 1026
813 key = decode_str (dec); 1027 // heuristic: assume that
814 if (!key) 1028 // a) decode_str + hv_store_ent are abysmally slow.
815 goto fail; 1029 // b) most hash keys are short, simple ascii text.
1030 // => try to "fast-match" such strings to avoid
1031 // the overhead of decode_str + hv_store_ent.
1032 {
1033 SV *value;
1034 char *p = dec->cur;
1035 char *e = p + 24; // only try up to 24 bytes
816 1036
817 decode_ws (dec); EXPECT_CH (':'); 1037 for (;;)
818
819 value = decode_sv (dec);
820 if (!value)
821 { 1038 {
1039 // the >= 0x80 is true on most architectures
1040 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1041 {
1042 // slow path, back up and use decode_str
1043 SV *key = decode_str (dec);
1044 if (!key)
1045 goto fail;
1046
1047 decode_ws (dec); EXPECT_CH (':');
1048
1049 value = decode_sv (dec);
1050 if (!value)
1051 {
1052 SvREFCNT_dec (key);
1053 goto fail;
1054 }
1055
1056 hv_store_ent (hv, key, value, 0);
822 SvREFCNT_dec (key); 1057 SvREFCNT_dec (key);
1058
1059 break;
1060 }
1061 else if (*p == '"')
1062 {
1063 // fast path, got a simple key
1064 char *key = dec->cur;
1065 int len = p - key;
1066 dec->cur = p + 1;
1067
1068 decode_ws (dec); EXPECT_CH (':');
1069
1070 value = decode_sv (dec);
1071 if (!value)
823 goto fail; 1072 goto fail;
1073
1074 hv_store (hv, key, len, value, 0);
1075
1076 break;
1077 }
1078
1079 ++p;
824 } 1080 }
825 1081 }
826 //TODO: optimise
827 hv_store_ent (hv, key, value, 0);
828 1082
829 decode_ws (dec); 1083 decode_ws (dec);
830 1084
831 if (*dec->cur == '}') 1085 if (*dec->cur == '}')
832 { 1086 {
… …
838 ERR (", or } expected while parsing object/hash"); 1092 ERR (", or } expected while parsing object/hash");
839 1093
840 ++dec->cur; 1094 ++dec->cur;
841 } 1095 }
842 1096
1097 DEC_DEC_DEPTH;
1098 sv = newRV_noinc ((SV *)hv);
1099
1100 // check filter callbacks
1101 if (dec->json.flags & F_HOOK)
1102 {
1103 ENTER; SAVETMPS;
1104
1105 if (HvKEYS (hv) == 1 && dec->json.cb_sk_object)
1106 {
1107 int count;
1108
1109 dSP; PUSHMARK (SP);
1110 XPUSHs (sv_2mortal (sv));
1111
1112 PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN;
1113
1114 if (count == 1)
1115 sv = newSVsv (POPs);
1116 else
1117 SvREFCNT_inc (sv);
1118 }
1119
1120 if (dec->json.cb_object)
1121 {
1122 int count;
1123
1124 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
1125 XPUSHs (sv_2mortal (sv));
1126
1127 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1128
1129 if (count == 1)
1130 sv = newSVsv (POPs);
1131 else
1132 SvREFCNT_inc (sv);
1133 }
1134
1135 FREETMPS; LEAVE;
1136 }
1137
843 return newRV_noinc ((SV *)hv); 1138 return newRV_noinc ((SV *)hv);
844 1139
845fail: 1140fail:
846 SvREFCNT_dec (hv); 1141 SvREFCNT_dec (hv);
1142 DEC_DEC_DEPTH;
847 return 0; 1143 return 0;
848} 1144}
849 1145
850static SV * 1146static SV *
851decode_sv (dec_t *dec) 1147decode_sv (dec_t *dec)
852{ 1148{
853 decode_ws (dec); 1149 decode_ws (dec);
1150
1151 // the beauty of JSON: you need exactly one character lookahead
1152 // to parse anything.
854 switch (*dec->cur) 1153 switch (*dec->cur)
855 { 1154 {
856 case '"': ++dec->cur; return decode_str (dec); 1155 case '"': ++dec->cur; return decode_str (dec);
857 case '[': ++dec->cur; return decode_av (dec); 1156 case '[': ++dec->cur; return decode_av (dec);
858 case '{': ++dec->cur; return decode_hv (dec); 1157 case '{': ++dec->cur; return decode_hv (dec);
… …
864 1163
865 case 't': 1164 case 't':
866 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1165 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
867 { 1166 {
868 dec->cur += 4; 1167 dec->cur += 4;
869 return newSViv (1); 1168 return SvREFCNT_inc (json_true);
870 } 1169 }
871 else 1170 else
872 ERR ("'true' expected"); 1171 ERR ("'true' expected");
873 1172
874 break; 1173 break;
875 1174
876 case 'f': 1175 case 'f':
877 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1176 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
878 { 1177 {
879 dec->cur += 5; 1178 dec->cur += 5;
880 return newSViv (0); 1179 return SvREFCNT_inc (json_false);
881 } 1180 }
882 else 1181 else
883 ERR ("'false' expected"); 1182 ERR ("'false' expected");
884 1183
885 break; 1184 break;
… …
894 ERR ("'null' expected"); 1193 ERR ("'null' expected");
895 1194
896 break; 1195 break;
897 1196
898 default: 1197 default:
899 ERR ("malformed json string, neither array, object, number, string or atom"); 1198 ERR ("malformed JSON string, neither array, object, number, string or atom");
900 break; 1199 break;
901 } 1200 }
902 1201
903fail: 1202fail:
904 return 0; 1203 return 0;
905} 1204}
906 1205
907static SV * 1206static SV *
908decode_json (SV *string, UV flags) 1207decode_json (SV *string, JSON *json, UV *offset_return)
909{ 1208{
1209 dec_t dec;
1210 UV offset;
910 SV *sv; 1211 SV *sv;
911 1212
1213 SvGETMAGIC (string);
1214 SvUPGRADE (string, SVt_PV);
1215
1216 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1217 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1218 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1219
912 if (flags & F_UTF8) 1220 if (json->flags & F_UTF8)
913 sv_utf8_downgrade (string, 0); 1221 sv_utf8_downgrade (string, 0);
914 else 1222 else
915 sv_utf8_upgrade (string); 1223 sv_utf8_upgrade (string);
916 1224
917 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1225 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
918 1226
919 dec_t dec; 1227 dec.json = *json;
920 dec.flags = flags;
921 dec.cur = SvPVX (string); 1228 dec.cur = SvPVX (string);
922 dec.end = SvEND (string); 1229 dec.end = SvEND (string);
923 dec.err = 0; 1230 dec.err = 0;
1231 dec.depth = 0;
1232 dec.maxdepth = DEC_DEPTH (dec.json.flags);
924 1233
1234 if (dec.json.cb_object || dec.json.cb_sk_object)
1235 dec.json.flags |= F_HOOK;
1236
1237 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
925 sv = decode_sv (&dec); 1238 sv = decode_sv (&dec);
926 1239
1240 if (!(offset_return || !sv))
1241 {
1242 // check for trailing garbage
1243 decode_ws (&dec);
1244
1245 if (*dec.cur)
1246 {
1247 dec.err = "garbage after JSON object";
1248 SvREFCNT_dec (sv);
1249 sv = 0;
1250 }
1251 }
1252
1253 if (offset_return || !sv)
1254 {
1255 offset = dec.json.flags & F_UTF8
1256 ? dec.cur - SvPVX (string)
1257 : utf8_distance (dec.cur, SvPVX (string));
1258
1259 if (offset_return)
1260 *offset_return = offset;
1261 }
1262
927 if (!sv) 1263 if (!sv)
928 { 1264 {
929 IV offset = dec.flags & F_UTF8
930 ? dec.cur - SvPVX (string)
931 : utf8_distance (dec.cur, SvPVX (string));
932 SV *uni = sv_newmortal (); 1265 SV *uni = sv_newmortal ();
933 1266
934 // horrible hack to silence warning inside pv_uni_display 1267 // horrible hack to silence warning inside pv_uni_display
935 COP cop = *PL_curcop; 1268 COP cop = *PL_curcop;
936 cop.cop_warnings = pWARN_NONE; 1269 cop.cop_warnings = pWARN_NONE;
… …
938 SAVEVPTR (PL_curcop); 1271 SAVEVPTR (PL_curcop);
939 PL_curcop = &cop; 1272 PL_curcop = &cop;
940 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1273 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
941 LEAVE; 1274 LEAVE;
942 1275
943 croak ("%s, at character offset %d (%s)", 1276 croak ("%s, at character offset %d [\"%s\"]",
944 dec.err, 1277 dec.err,
945 (int)offset, 1278 (int)offset,
946 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1279 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
947 } 1280 }
948 1281
949 sv = sv_2mortal (sv); 1282 sv = sv_2mortal (sv);
950 1283
951 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1284 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
952 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1285 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
953 1286
954 return sv; 1287 return sv;
955} 1288}
956 1289
… …
961 1294
962BOOT: 1295BOOT:
963{ 1296{
964 int i; 1297 int i;
965 1298
966 memset (decode_hexdigit, 0xff, 256);
967 for (i = 10; i--; ) 1299 for (i = 0; i < 256; ++i)
968 decode_hexdigit ['0' + i] = i; 1300 decode_hexdigit [i] =
1301 i >= '0' && i <= '9' ? i - '0'
1302 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1303 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1304 : -1;
969 1305
970 for (i = 7; i--; ) 1306 json_stash = gv_stashpv ("JSON::XS" , 1);
1307 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1308
1309 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1310 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1311}
1312
1313PROTOTYPES: DISABLE
1314
1315void new (char *klass)
1316 PPCODE:
1317{
1318 SV *pv = NEWSV (0, sizeof (JSON));
1319 SvPOK_only (pv);
1320 Zero (SvPVX (pv), 1, JSON);
1321 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1322 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1323}
1324
1325void ascii (JSON *self, int enable = 1)
1326 ALIAS:
1327 ascii = F_ASCII
1328 latin1 = F_LATIN1
1329 utf8 = F_UTF8
1330 indent = F_INDENT
1331 canonical = F_CANONICAL
1332 space_before = F_SPACE_BEFORE
1333 space_after = F_SPACE_AFTER
1334 pretty = F_PRETTY
1335 allow_nonref = F_ALLOW_NONREF
1336 shrink = F_SHRINK
1337 allow_blessed = F_ALLOW_BLESSED
1338 convert_blessed = F_CONV_BLESSED
1339 PPCODE:
1340{
1341 if (enable)
1342 self->flags |= ix;
1343 else
1344 self->flags &= ~ix;
1345
1346 XPUSHs (ST (0));
1347}
1348
1349void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1350 PPCODE:
1351{
1352 UV log2 = 0;
1353
1354 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1355
1356 while ((1UL << log2) < max_depth)
1357 ++log2;
1358
1359 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1360
1361 XPUSHs (ST (0));
1362}
1363
1364void max_size (JSON *self, UV max_size = 0)
1365 PPCODE:
1366{
1367 UV log2 = 0;
1368
1369 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1370 if (max_size == 1) max_size = 2;
1371
1372 while ((1UL << log2) < max_size)
1373 ++log2;
1374
1375 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1376
1377 XPUSHs (ST (0));
1378}
1379
1380void filter_json_objects (JSON *self, SV *cb = &PL_sv_undef)
1381 ALIAS:
1382 filter_sk_json_objects = 1
1383 PPCODE:
1384{
1385 if (!SvOK (cb))
1386 cb = 0;
1387
1388 switch (ix)
971 { 1389 {
972 decode_hexdigit ['a' + i] = 10 + i; 1390 case 0: self->cb_object = cb; break;
973 decode_hexdigit ['A' + i] = 10 + i; 1391 case 1: self->cb_sk_object = cb; break;
974 } 1392 }
975 1393
976 json_stash = gv_stashpv ("JSON::XS", 1); 1394 XPUSHs (ST (0));
977} 1395}
978 1396
979PROTOTYPES: DISABLE
980
981SV *new (char *dummy)
982 CODE:
983 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
984 OUTPUT:
985 RETVAL
986
987SV *ascii (SV *self, int enable = 1)
988 ALIAS:
989 ascii = F_ASCII
990 utf8 = F_UTF8
991 indent = F_INDENT
992 canonical = F_CANONICAL
993 space_before = F_SPACE_BEFORE
994 space_after = F_SPACE_AFTER
995 pretty = F_PRETTY
996 allow_nonref = F_ALLOW_NONREF
997 shrink = F_SHRINK
998 CODE:
999{
1000 UV *uv = SvJSON (self);
1001 if (enable)
1002 *uv |= ix;
1003 else
1004 *uv &= ~ix;
1005
1006 RETVAL = newSVsv (self);
1007}
1008 OUTPUT:
1009 RETVAL
1010
1011void encode (SV *self, SV *scalar) 1397void encode (JSON *self, SV *scalar)
1012 PPCODE: 1398 PPCODE:
1013 XPUSHs (encode_json (scalar, *SvJSON (self))); 1399 XPUSHs (encode_json (scalar, self));
1014 1400
1015void decode (SV *self, SV *jsonstr) 1401void decode (JSON *self, SV *jsonstr)
1016 PPCODE: 1402 PPCODE:
1017 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1403 XPUSHs (decode_json (jsonstr, self, 0));
1404
1405void decode_prefix (JSON *self, SV *jsonstr)
1406 PPCODE:
1407{
1408 UV offset;
1409 EXTEND (SP, 2);
1410 PUSHs (decode_json (jsonstr, self, &offset));
1411 PUSHs (sv_2mortal (newSVuv (offset)));
1412}
1018 1413
1019PROTOTYPES: ENABLE 1414PROTOTYPES: ENABLE
1020 1415
1021void to_json (SV *scalar) 1416void to_json (SV *scalar)
1022 PPCODE: 1417 PPCODE:
1418{
1419 JSON json = { F_DEFAULT | F_UTF8 };
1023 XPUSHs (encode_json (scalar, F_UTF8)); 1420 XPUSHs (encode_json (scalar, &json));
1421}
1024 1422
1025void from_json (SV *jsonstr) 1423void from_json (SV *jsonstr)
1026 PPCODE: 1424 PPCODE:
1425{
1426 JSON json = { F_DEFAULT | F_UTF8 };
1027 XPUSHs (decode_json (jsonstr, F_UTF8)); 1427 XPUSHs (decode_json (jsonstr, &json, 0));
1428}
1028 1429

Diff Legend

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