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 1.33 by root, Wed May 23 20:26:40 2007 UTC vs.
Revision 1.62 by root, Sun Aug 26 22:27:32 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
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#include <stdio.h>
9#include <float.h>
9 10
10#if defined(__BORLANDC__) || defined(_MSC_VER) 11#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 12# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 13#endif
13 14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
14#define F_ASCII 0x00000001UL 21#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 22#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 23#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 24#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 25#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 26#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 27#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 28#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 32#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27 33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25 37
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27 39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 40
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH) 42#define F_DEFAULT (9UL << S_MAXDEPTH)
33 43
34#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 45#define INDENT_STEP 3 // spaces per indentation level
36 46
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size 47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38 48
39#define SB do { 49#define SB do {
40#define SE } while (0) 50#define SE } while (0)
41 51
52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value))
54# define inline inline
55#else
56# define expect(expr,value) (expr)
57# define inline static
58#endif
59
60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#ifdef USE_ITHREADS
64# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else
67# define JSON_SLOW 0
68# define JSON_STASH json_stash
69#endif
70
42static HV *json_stash; // JSON::XS:: 71static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false;
73
74typedef struct {
75 U32 flags;
76 SV *cb_object;
77 HV *cb_sk_object;
78} JSON;
43 79
44///////////////////////////////////////////////////////////////////////////// 80/////////////////////////////////////////////////////////////////////////////
45// utility functions 81// utility functions
46 82
47static UV * 83inline void
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) 84shrink (SV *sv)
58{ 85{
59 sv_utf8_downgrade (sv, 1); 86 sv_utf8_downgrade (sv, 1);
60 if (SvLEN (sv) > SvCUR (sv) + 1) 87 if (SvLEN (sv) > SvCUR (sv) + 1)
61 { 88 {
… …
70// decode an utf-8 character and return it, or (UV)-1 in 97// decode an utf-8 character and return it, or (UV)-1 in
71// case of an error. 98// case of an error.
72// we special-case "safe" characters from U+80 .. U+7FF, 99// we special-case "safe" characters from U+80 .. U+7FF,
73// but use the very good perl function to parse anything else. 100// but use the very good perl function to parse anything else.
74// note that we never call this function for a ascii codepoints 101// note that we never call this function for a ascii codepoints
75static UV 102inline UV
76decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
77{ 104{
78 if (s[0] > 0xdf || s[0] < 0xc2) 105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
79 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
80 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
81 { 108 {
82 *clen = 2; 109 *clen = 2;
83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
… …
96typedef struct 123typedef struct
97{ 124{
98 char *cur; // SvPVX (sv) + current output position 125 char *cur; // SvPVX (sv) + current output position
99 char *end; // SvEND (sv) 126 char *end; // SvEND (sv)
100 SV *sv; // result scalar 127 SV *sv; // result scalar
101 U32 flags; // F_* 128 JSON json;
102 U32 indent; // indentation level 129 U32 indent; // indentation level
103 U32 maxdepth; // max. indentation/recursion level 130 U32 maxdepth; // max. indentation/recursion level
104} enc_t; 131} enc_t;
105 132
106static void 133inline void
107need (enc_t *enc, STRLEN len) 134need (enc_t *enc, STRLEN len)
108{ 135{
109 if (enc->cur + len >= enc->end) 136 if (expect_false (enc->cur + len >= enc->end))
110 { 137 {
111 STRLEN cur = enc->cur - SvPVX (enc->sv); 138 STRLEN cur = enc->cur - SvPVX (enc->sv);
112 SvGROW (enc->sv, cur + len + 1); 139 SvGROW (enc->sv, cur + len + 1);
113 enc->cur = SvPVX (enc->sv) + cur; 140 enc->cur = SvPVX (enc->sv) + cur;
114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
115 } 142 }
116} 143}
117 144
118static void 145inline void
119encode_ch (enc_t *enc, char ch) 146encode_ch (enc_t *enc, char ch)
120{ 147{
121 need (enc, 1); 148 need (enc, 1);
122 *enc->cur++ = ch; 149 *enc->cur++ = ch;
123} 150}
… …
131 158
132 while (str < end) 159 while (str < end)
133 { 160 {
134 unsigned char ch = *(unsigned char *)str; 161 unsigned char ch = *(unsigned char *)str;
135 162
136 if (ch >= 0x20 && ch < 0x80) // most common case 163 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
137 { 164 {
138 if (ch == '"') // but with slow exceptions 165 if (expect_false (ch == '"')) // but with slow exceptions
139 { 166 {
140 need (enc, len += 1); 167 need (enc, len += 1);
141 *enc->cur++ = '\\'; 168 *enc->cur++ = '\\';
142 *enc->cur++ = '"'; 169 *enc->cur++ = '"';
143 } 170 }
144 else if (ch == '\\') 171 else if (expect_false (ch == '\\'))
145 { 172 {
146 need (enc, len += 1); 173 need (enc, len += 1);
147 *enc->cur++ = '\\'; 174 *enc->cur++ = '\\';
148 *enc->cur++ = '\\'; 175 *enc->cur++ = '\\';
149 } 176 }
… …
167 STRLEN clen; 194 STRLEN clen;
168 UV uch; 195 UV uch;
169 196
170 if (is_utf8) 197 if (is_utf8)
171 { 198 {
172 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173 uch = decode_utf8 (str, end - str, &clen); 199 uch = decode_utf8 (str, end - str, &clen);
174 if (clen == (STRLEN)-1) 200 if (clen == (STRLEN)-1)
175 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 201 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
176 } 202 }
177 else 203 else
… …
181 } 207 }
182 208
183 if (uch > 0x10FFFFUL) 209 if (uch > 0x10FFFFUL)
184 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
185 211
186 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) 212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
187 { 213 {
188 if (uch > 0xFFFFUL) 214 if (uch > 0xFFFFUL)
189 { 215 {
190 need (enc, len += 11); 216 need (enc, len += 11);
191 sprintf (enc->cur, "\\u%04x\\u%04x", 217 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
205 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 231 *enc->cur++ = hexdigit [(uch >> 0) & 15];
206 } 232 }
207 233
208 str += clen; 234 str += clen;
209 } 235 }
210 else if (enc->flags & F_LATIN1) 236 else if (enc->json.flags & F_LATIN1)
211 { 237 {
212 *enc->cur++ = uch; 238 *enc->cur++ = uch;
213 str += clen; 239 str += clen;
214 } 240 }
215 else if (is_utf8) 241 else if (is_utf8)
… …
233 259
234 --len; 260 --len;
235 } 261 }
236} 262}
237 263
238static void 264inline void
239encode_indent (enc_t *enc) 265encode_indent (enc_t *enc)
240{ 266{
241 if (enc->flags & F_INDENT) 267 if (enc->json.flags & F_INDENT)
242 { 268 {
243 int spaces = enc->indent * INDENT_STEP; 269 int spaces = enc->indent * INDENT_STEP;
244 270
245 need (enc, spaces); 271 need (enc, spaces);
246 memset (enc->cur, ' ', spaces); 272 memset (enc->cur, ' ', spaces);
247 enc->cur += spaces; 273 enc->cur += spaces;
248 } 274 }
249} 275}
250 276
251static void 277inline void
252encode_space (enc_t *enc) 278encode_space (enc_t *enc)
253{ 279{
254 need (enc, 1); 280 need (enc, 1);
255 encode_ch (enc, ' '); 281 encode_ch (enc, ' ');
256} 282}
257 283
258static void 284inline void
259encode_nl (enc_t *enc) 285encode_nl (enc_t *enc)
260{ 286{
261 if (enc->flags & F_INDENT) 287 if (enc->json.flags & F_INDENT)
262 { 288 {
263 need (enc, 1); 289 need (enc, 1);
264 encode_ch (enc, '\n'); 290 encode_ch (enc, '\n');
265 } 291 }
266} 292}
267 293
268static void 294inline void
269encode_comma (enc_t *enc) 295encode_comma (enc_t *enc)
270{ 296{
271 encode_ch (enc, ','); 297 encode_ch (enc, ',');
272 298
273 if (enc->flags & F_INDENT) 299 if (enc->json.flags & F_INDENT)
274 encode_nl (enc); 300 encode_nl (enc);
275 else if (enc->flags & F_SPACE_AFTER) 301 else if (enc->json.flags & F_SPACE_AFTER)
276 encode_space (enc); 302 encode_space (enc);
277} 303}
278 304
279static void encode_sv (enc_t *enc, SV *sv); 305static void encode_sv (enc_t *enc, SV *sv);
280 306
… …
289 encode_ch (enc, '['); encode_nl (enc); 315 encode_ch (enc, '['); encode_nl (enc);
290 ++enc->indent; 316 ++enc->indent;
291 317
292 for (i = 0; i <= len; ++i) 318 for (i = 0; i <= len; ++i)
293 { 319 {
320 SV **svp = av_fetch (av, i, 0);
321
294 encode_indent (enc); 322 encode_indent (enc);
295 encode_sv (enc, *av_fetch (av, i, 0)); 323
324 if (svp)
325 encode_sv (enc, *svp);
326 else
327 encode_str (enc, "null", 4, 0);
296 328
297 if (i < len) 329 if (i < len)
298 encode_comma (enc); 330 encode_comma (enc);
299 } 331 }
300 332
… …
303 --enc->indent; 335 --enc->indent;
304 encode_indent (enc); encode_ch (enc, ']'); 336 encode_indent (enc); encode_ch (enc, ']');
305} 337}
306 338
307static void 339static void
308encode_he (enc_t *enc, HE *he) 340encode_hk (enc_t *enc, HE *he)
309{ 341{
310 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
311 343
312 if (HeKLEN (he) == HEf_SVKEY) 344 if (HeKLEN (he) == HEf_SVKEY)
313 { 345 {
… …
323 else 355 else
324 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 356 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
325 357
326 encode_ch (enc, '"'); 358 encode_ch (enc, '"');
327 359
328 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 360 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
329 encode_ch (enc, ':'); 361 encode_ch (enc, ':');
330 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 362 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
331 encode_sv (enc, HeVAL (he));
332} 363}
333 364
334// compare hash entries, used when all keys are bytestrings 365// compare hash entries, used when all keys are bytestrings
335static int 366static int
336he_cmp_fast (const void *a_, const void *b_) 367he_cmp_fast (const void *a_, const void *b_)
… …
341 HE *b = *(HE **)b_; 372 HE *b = *(HE **)b_;
342 373
343 STRLEN la = HeKLEN (a); 374 STRLEN la = HeKLEN (a);
344 STRLEN lb = HeKLEN (b); 375 STRLEN lb = HeKLEN (b);
345 376
346 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 377 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
347 cmp = la - lb; 378 cmp = lb - la;
348 379
349 return cmp; 380 return cmp;
350} 381}
351 382
352// compare hash entries, used when some keys are sv's or utf-x 383// compare hash entries, used when some keys are sv's or utf-x
353static int 384static int
354he_cmp_slow (const void *a, const void *b) 385he_cmp_slow (const void *a, const void *b)
355{ 386{
356 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 387 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
357} 388}
358 389
359static void 390static void
360encode_hv (enc_t *enc, HV *hv) 391encode_hv (enc_t *enc, HV *hv)
361{ 392{
393 HE *he;
362 int count, i; 394 int count;
363 395
364 if (enc->indent >= enc->maxdepth) 396 if (enc->indent >= enc->maxdepth)
365 croak ("data structure too deep (hit recursion limit)"); 397 croak ("data structure too deep (hit recursion limit)");
366 398
367 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
368 400
369 if ((count = hv_iterinit (hv)))
370 {
371 // for canonical output we have to sort by keys first 401 // for canonical output we have to sort by keys first
372 // actually, this is mostly due to the stupid so-called 402 // actually, this is mostly due to the stupid so-called
373 // security workaround added somewhere in 5.8.x. 403 // security workaround added somewhere in 5.8.x.
374 // that randomises hash orderings 404 // that randomises hash orderings
375 if (enc->flags & F_CANONICAL) 405 if (enc->json.flags & F_CANONICAL)
406 {
407 int count = hv_iterinit (hv);
408
409 if (SvMAGICAL (hv))
376 { 410 {
411 // need to count by iterating. could improve by dynamically building the vector below
412 // but I don't care for the speed of this special case.
413 // note also that we will run into undefined behaviour when the two iterations
414 // do not result in the same count, something I might care for in some later release.
415
416 count = 0;
417 while (hv_iternext (hv))
418 ++count;
419
420 hv_iterinit (hv);
421 }
422
423 if (count)
424 {
377 int fast = 1; 425 int i, fast = 1;
378 HE *he; 426#if defined(__BORLANDC__) || defined(_MSC_VER)
379#if WIN32
380 HE **hes = _alloca (count * sizeof (HE)); 427 HE **hes = _alloca (count * sizeof (HE));
381#else 428#else
382 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 429 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
383#endif 430#endif
384 431
… …
410 457
411 FREETMPS; 458 FREETMPS;
412 LEAVE; 459 LEAVE;
413 } 460 }
414 461
415 for (i = 0; i < count; ++i) 462 while (count--)
416 { 463 {
417 encode_indent (enc); 464 encode_indent (enc);
465 he = hes [count];
418 encode_he (enc, hes [i]); 466 encode_hk (enc, he);
467 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
419 468
420 if (i < count - 1) 469 if (count)
421 encode_comma (enc); 470 encode_comma (enc);
422 } 471 }
423 472 }
473 }
474 else
475 {
476 if (hv_iterinit (hv) || SvMAGICAL (hv))
477 if ((he = hv_iternext (hv)))
478 for (;;)
479 {
424 encode_nl (enc); 480 encode_indent (enc);
481 encode_hk (enc, he);
482 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
483
484 if (!(he = hv_iternext (hv)))
485 break;
486
487 encode_comma (enc);
488 }
489 }
490
491 encode_nl (enc);
492
493 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
494}
495
496// encode objects, arrays and special \0=false and \1=true values.
497static void
498encode_rv (enc_t *enc, SV *sv)
499{
500 svtype svt;
501
502 SvGETMAGIC (sv);
503 svt = SvTYPE (sv);
504
505 if (expect_false (SvOBJECT (sv)))
506 {
507 HV *stash = !JSON_SLOW || json_boolean_stash
508 ? json_boolean_stash
509 : gv_stashpv ("JSON::XS::Boolean", 1);
510
511 if (SvSTASH (sv) == stash)
512 {
513 if (SvIV (sv))
514 encode_str (enc, "true", 4, 0);
515 else
516 encode_str (enc, "false", 5, 0);
425 } 517 }
426 else 518 else
427 { 519 {
428 HE *he = hv_iternext (hv); 520#if 0
429 521 if (0 && sv_derived_from (rv, "JSON::Literal"))
430 for (;;)
431 {
432 encode_indent (enc);
433 encode_he (enc, he);
434
435 if (!(he = hv_iternext (hv)))
436 break;
437
438 encode_comma (enc);
439 } 522 {
523 // not yet
524 }
525#endif
526 if (enc->json.flags & F_CONV_BLESSED)
527 {
528 // we re-bless the reference to get overload and other niceties right
529 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
440 530
531 if (to_json)
532 {
533 dSP;
534
535 ENTER; SAVETMPS; PUSHMARK (SP);
536 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
537
538 // calling with G_SCALAR ensures that we always get a 1 return value
539 PUTBACK;
540 call_sv ((SV *)GvCV (to_json), G_SCALAR);
541 SPAGAIN;
542
543 // catch this surprisingly common error
544 if (SvROK (TOPs) && SvRV (TOPs) == sv)
545 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
546
547 sv = POPs;
548 PUTBACK;
549
441 encode_nl (enc); 550 encode_sv (enc, sv);
551
552 FREETMPS; LEAVE;
553 }
554 else if (enc->json.flags & F_ALLOW_BLESSED)
555 encode_str (enc, "null", 4, 0);
556 else
557 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
558 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
559 }
560 else if (enc->json.flags & F_ALLOW_BLESSED)
561 encode_str (enc, "null", 4, 0);
562 else
563 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
564 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
442 } 565 }
443 } 566 }
444
445 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
446}
447
448// encode objects, arrays and special \0=false and \1=true values.
449static void
450encode_rv (enc_t *enc, SV *sv)
451{
452 svtype svt;
453
454 SvGETMAGIC (sv);
455 svt = SvTYPE (sv);
456
457 if (svt == SVt_PVHV) 567 else if (svt == SVt_PVHV)
458 encode_hv (enc, (HV *)sv); 568 encode_hv (enc, (HV *)sv);
459 else if (svt == SVt_PVAV) 569 else if (svt == SVt_PVAV)
460 encode_av (enc, (AV *)sv); 570 encode_av (enc, (AV *)sv);
461 else if (svt < SVt_PVAV) 571 else if (svt < SVt_PVAV)
462 { 572 {
463 if (SvNIOK (sv) && SvIV (sv) == 0) 573 STRLEN len = 0;
574 char *pv = svt ? SvPV (sv, len) : 0;
575
576 if (len == 1 && *pv == '1')
577 encode_str (enc, "true", 4, 0);
578 else if (len == 1 && *pv == '0')
464 encode_str (enc, "false", 5, 0); 579 encode_str (enc, "false", 5, 0);
465 else if (SvNIOK (sv) && SvIV (sv) == 1)
466 encode_str (enc, "true", 4, 0);
467 else 580 else
468 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 581 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
469 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 582 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
470 } 583 }
471 else 584 else
… …
486 encode_str (enc, str, len, SvUTF8 (sv)); 599 encode_str (enc, str, len, SvUTF8 (sv));
487 encode_ch (enc, '"'); 600 encode_ch (enc, '"');
488 } 601 }
489 else if (SvNOKp (sv)) 602 else if (SvNOKp (sv))
490 { 603 {
604 // trust that perl will do the right thing w.r.t. JSON syntax.
491 need (enc, NV_DIG + 32); 605 need (enc, NV_DIG + 32);
492 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 606 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
493 enc->cur += strlen (enc->cur); 607 enc->cur += strlen (enc->cur);
494 } 608 }
495 else if (SvIOKp (sv)) 609 else if (SvIOKp (sv))
496 { 610 {
497 need (enc, 64); 611 // we assume we can always read an IV as a UV
612 if (SvUV (sv) & ~(UV)0x7fff)
613 {
614 // large integer, use the (rather slow) snprintf way.
615 need (enc, sizeof (UV) * 3);
498 enc->cur += 616 enc->cur +=
499 SvIsUV(sv) 617 SvIsUV(sv)
500 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 618 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
501 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 619 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
620 }
621 else
622 {
623 // optimise the "small number case"
624 // code will likely be branchless and use only a single multiplication
625 I32 i = SvIV (sv);
626 U32 u;
627 char digit, nz = 0;
628
629 need (enc, 6);
630
631 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
632 u = i < 0 ? -i : i;
633
634 // convert to 4.28 fixed-point representation
635 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
636
637 // now output digit by digit, each time masking out the integer part
638 // and multiplying by 5 while moving the decimal point one to the right,
639 // resulting in a net multiplication by 10.
640 // we always write the digit to memory but conditionally increment
641 // the pointer, to ease the usage of conditional move instructions.
642 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
643 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
644 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
645 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
646 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
647 }
502 } 648 }
503 else if (SvROK (sv)) 649 else if (SvROK (sv))
504 encode_rv (enc, SvRV (sv)); 650 encode_rv (enc, SvRV (sv));
505 else if (!SvOK (sv)) 651 else if (!SvOK (sv))
506 encode_str (enc, "null", 4, 0); 652 encode_str (enc, "null", 4, 0);
… …
508 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 654 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
509 SvPV_nolen (sv), SvFLAGS (sv)); 655 SvPV_nolen (sv), SvFLAGS (sv));
510} 656}
511 657
512static SV * 658static SV *
513encode_json (SV *scalar, U32 flags) 659encode_json (SV *scalar, JSON *json)
514{ 660{
515 enc_t enc; 661 enc_t enc;
516 662
517 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 663 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
518 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 664 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
519 665
520 enc.flags = flags; 666 enc.json = *json;
521 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 667 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
522 enc.cur = SvPVX (enc.sv); 668 enc.cur = SvPVX (enc.sv);
523 enc.end = SvEND (enc.sv); 669 enc.end = SvEND (enc.sv);
524 enc.indent = 0; 670 enc.indent = 0;
525 enc.maxdepth = DEC_DEPTH (flags); 671 enc.maxdepth = DEC_DEPTH (enc.json.flags);
526 672
527 SvPOK_only (enc.sv); 673 SvPOK_only (enc.sv);
528 encode_sv (&enc, scalar); 674 encode_sv (&enc, scalar);
529 675
530 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 676 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
531 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 677 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
532 678
533 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 679 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
534 SvUTF8_on (enc.sv); 680 SvUTF8_on (enc.sv);
535 681
536 if (enc.flags & F_SHRINK) 682 if (enc.json.flags & F_SHRINK)
537 shrink (enc.sv); 683 shrink (enc.sv);
538 684
539 return enc.sv; 685 return enc.sv;
540} 686}
541 687
… …
546typedef struct 692typedef struct
547{ 693{
548 char *cur; // current parser pointer 694 char *cur; // current parser pointer
549 char *end; // end of input string 695 char *end; // end of input string
550 const char *err; // parse error, if != 0 696 const char *err; // parse error, if != 0
551 U32 flags; // F_* 697 JSON json;
552 U32 depth; // recursion depth 698 U32 depth; // recursion depth
553 U32 maxdepth; // recursion depth limit 699 U32 maxdepth; // recursion depth limit
554} dec_t; 700} dec_t;
555 701
556static void 702inline void
557decode_ws (dec_t *dec) 703decode_ws (dec_t *dec)
558{ 704{
559 for (;;) 705 for (;;)
560 { 706 {
561 char ch = *dec->cur; 707 char ch = *dec->cur;
… …
587decode_4hex (dec_t *dec) 733decode_4hex (dec_t *dec)
588{ 734{
589 signed char d1, d2, d3, d4; 735 signed char d1, d2, d3, d4;
590 unsigned char *cur = (unsigned char *)dec->cur; 736 unsigned char *cur = (unsigned char *)dec->cur;
591 737
592 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 738 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
593 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 739 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
594 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 740 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
595 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 741 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
596 742
597 dec->cur += 4; 743 dec->cur += 4;
598 744
599 return ((UV)d1) << 12 745 return ((UV)d1) << 12
600 | ((UV)d2) << 8 746 | ((UV)d2) << 8
… …
608static SV * 754static SV *
609decode_str (dec_t *dec) 755decode_str (dec_t *dec)
610{ 756{
611 SV *sv = 0; 757 SV *sv = 0;
612 int utf8 = 0; 758 int utf8 = 0;
759 char *dec_cur = dec->cur;
613 760
614 do 761 do
615 { 762 {
616 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 763 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
617 char *cur = buf; 764 char *cur = buf;
618 765
619 do 766 do
620 { 767 {
621 unsigned char ch = *(unsigned char *)dec->cur++; 768 unsigned char ch = *(unsigned char *)dec_cur++;
622 769
623 if (ch == '"') 770 if (expect_false (ch == '"'))
624 { 771 {
625 --dec->cur; 772 --dec_cur;
626 break; 773 break;
627 } 774 }
628 else if (ch == '\\') 775 else if (expect_false (ch == '\\'))
629 { 776 {
630 switch (*dec->cur) 777 switch (*dec_cur)
631 { 778 {
632 case '\\': 779 case '\\':
633 case '/': 780 case '/':
634 case '"': *cur++ = *dec->cur++; break; 781 case '"': *cur++ = *dec_cur++; break;
635 782
636 case 'b': ++dec->cur; *cur++ = '\010'; break; 783 case 'b': ++dec_cur; *cur++ = '\010'; break;
637 case 't': ++dec->cur; *cur++ = '\011'; break; 784 case 't': ++dec_cur; *cur++ = '\011'; break;
638 case 'n': ++dec->cur; *cur++ = '\012'; break; 785 case 'n': ++dec_cur; *cur++ = '\012'; break;
639 case 'f': ++dec->cur; *cur++ = '\014'; break; 786 case 'f': ++dec_cur; *cur++ = '\014'; break;
640 case 'r': ++dec->cur; *cur++ = '\015'; break; 787 case 'r': ++dec_cur; *cur++ = '\015'; break;
641 788
642 case 'u': 789 case 'u':
643 { 790 {
644 UV lo, hi; 791 UV lo, hi;
645 ++dec->cur; 792 ++dec_cur;
646 793
794 dec->cur = dec_cur;
647 hi = decode_4hex (dec); 795 hi = decode_4hex (dec);
796 dec_cur = dec->cur;
648 if (hi == (UV)-1) 797 if (hi == (UV)-1)
649 goto fail; 798 goto fail;
650 799
651 // possibly a surrogate pair 800 // possibly a surrogate pair
652 if (hi >= 0xd800) 801 if (hi >= 0xd800)
653 if (hi < 0xdc00) 802 if (hi < 0xdc00)
654 { 803 {
655 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 804 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
656 ERR ("missing low surrogate character in surrogate pair"); 805 ERR ("missing low surrogate character in surrogate pair");
657 806
658 dec->cur += 2; 807 dec_cur += 2;
659 808
809 dec->cur = dec_cur;
660 lo = decode_4hex (dec); 810 lo = decode_4hex (dec);
811 dec_cur = dec->cur;
661 if (lo == (UV)-1) 812 if (lo == (UV)-1)
662 goto fail; 813 goto fail;
663 814
664 if (lo < 0xdc00 || lo >= 0xe000) 815 if (lo < 0xdc00 || lo >= 0xe000)
665 ERR ("surrogate pair expected"); 816 ERR ("surrogate pair expected");
… …
679 *cur++ = hi; 830 *cur++ = hi;
680 } 831 }
681 break; 832 break;
682 833
683 default: 834 default:
684 --dec->cur; 835 --dec_cur;
685 ERR ("illegal backslash escape sequence in string"); 836 ERR ("illegal backslash escape sequence in string");
686 } 837 }
687 } 838 }
688 else if (ch >= 0x20 && ch <= 0x7f) 839 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
689 *cur++ = ch; 840 *cur++ = ch;
690 else if (ch >= 0x80) 841 else if (ch >= 0x80)
691 { 842 {
692 STRLEN clen; 843 STRLEN clen;
693 UV uch; 844 UV uch;
694 845
695 --dec->cur; 846 --dec_cur;
696 847
697 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 848 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
698 if (clen == (STRLEN)-1) 849 if (clen == (STRLEN)-1)
699 ERR ("malformed UTF-8 character in JSON string"); 850 ERR ("malformed UTF-8 character in JSON string");
700 851
701 do 852 do
702 *cur++ = *dec->cur++; 853 *cur++ = *dec_cur++;
703 while (--clen); 854 while (--clen);
704 855
705 utf8 = 1; 856 utf8 = 1;
706 } 857 }
707 else 858 else
708 { 859 {
709 --dec->cur; 860 --dec_cur;
710 861
711 if (!ch) 862 if (!ch)
712 ERR ("unexpected end of string while parsing JSON string"); 863 ERR ("unexpected end of string while parsing JSON string");
713 else 864 else
714 ERR ("invalid character encountered while parsing JSON string"); 865 ERR ("invalid character encountered while parsing JSON string");
… …
727 } 878 }
728 else 879 else
729 sv = newSVpvn (buf, len); 880 sv = newSVpvn (buf, len);
730 } 881 }
731 } 882 }
732 while (*dec->cur != '"'); 883 while (*dec_cur != '"');
733 884
734 ++dec->cur; 885 ++dec_cur;
735 886
736 if (sv) 887 if (sv)
737 { 888 {
738 SvPOK_only (sv); 889 SvPOK_only (sv);
739 *SvEND (sv) = 0; 890 *SvEND (sv) = 0;
… …
742 SvUTF8_on (sv); 893 SvUTF8_on (sv);
743 } 894 }
744 else 895 else
745 sv = newSVpvn ("", 0); 896 sv = newSVpvn ("", 0);
746 897
898 dec->cur = dec_cur;
747 return sv; 899 return sv;
748 900
749fail: 901fail:
902 dec->cur = dec_cur;
750 return 0; 903 return 0;
751} 904}
752 905
753static SV * 906static SV *
754decode_num (dec_t *dec) 907decode_num (dec_t *dec)
… …
812 is_nv = 1; 965 is_nv = 1;
813 } 966 }
814 967
815 if (!is_nv) 968 if (!is_nv)
816 { 969 {
817 UV uv; 970 int len = dec->cur - start;
818 int numtype = grok_number (start, dec->cur - start, &uv); 971
819 if (numtype & IS_NUMBER_IN_UV) 972 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
820 if (numtype & IS_NUMBER_NEG) 973 if (*start == '-')
974 switch (len)
821 { 975 {
822 if (uv < (UV)IV_MIN) 976 case 2: return newSViv (-( start [1] - '0' * 1));
823 return newSViv (-(IV)uv); 977 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
978 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
979 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
824 } 980 }
981 else
982 switch (len)
983 {
984 case 1: return newSViv ( start [0] - '0' * 1);
985 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
986 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
987 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
988 }
989
990 {
991 UV uv;
992 int numtype = grok_number (start, len, &uv);
993 if (numtype & IS_NUMBER_IN_UV)
994 if (numtype & IS_NUMBER_NEG)
995 {
996 if (uv < (UV)IV_MIN)
997 return newSViv (-(IV)uv);
998 }
825 else 999 else
826 return newSVuv (uv); 1000 return newSVuv (uv);
827 } 1001 }
828 1002
1003 len -= *start == '-' ? 1 : 0;
1004
1005 // does not fit into IV or UV, try NV
1006 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1007 #if defined (LDBL_DIG)
1008 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1009 #endif
1010 )
1011 // fits into NV without loss of precision
1012 return newSVnv (Atof (start));
1013
1014 // everything else fails, convert it to a string
1015 return newSVpvn (start, dec->cur - start);
1016 }
1017
1018 // loss of precision here
829 return newSVnv (Atof (start)); 1019 return newSVnv (Atof (start));
830 1020
831fail: 1021fail:
832 return 0; 1022 return 0;
833} 1023}
… …
877} 1067}
878 1068
879static SV * 1069static SV *
880decode_hv (dec_t *dec) 1070decode_hv (dec_t *dec)
881{ 1071{
1072 SV *sv;
882 HV *hv = newHV (); 1073 HV *hv = newHV ();
883 1074
884 DEC_INC_DEPTH; 1075 DEC_INC_DEPTH;
885 decode_ws (dec); 1076 decode_ws (dec);
886 1077
887 if (*dec->cur == '}') 1078 if (*dec->cur == '}')
888 ++dec->cur; 1079 ++dec->cur;
889 else 1080 else
890 for (;;) 1081 for (;;)
891 { 1082 {
892 SV *key, *value;
893
894 decode_ws (dec); EXPECT_CH ('"'); 1083 decode_ws (dec); EXPECT_CH ('"');
895 1084
896 key = decode_str (dec); 1085 // heuristic: assume that
897 if (!key) 1086 // a) decode_str + hv_store_ent are abysmally slow.
898 goto fail; 1087 // b) most hash keys are short, simple ascii text.
1088 // => try to "fast-match" such strings to avoid
1089 // the overhead of decode_str + hv_store_ent.
1090 {
1091 SV *value;
1092 char *p = dec->cur;
1093 char *e = p + 24; // only try up to 24 bytes
899 1094
900 decode_ws (dec); EXPECT_CH (':'); 1095 for (;;)
901
902 value = decode_sv (dec);
903 if (!value)
904 { 1096 {
1097 // the >= 0x80 is true on most architectures
1098 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1099 {
1100 // slow path, back up and use decode_str
1101 SV *key = decode_str (dec);
1102 if (!key)
1103 goto fail;
1104
1105 decode_ws (dec); EXPECT_CH (':');
1106
1107 value = decode_sv (dec);
1108 if (!value)
1109 {
1110 SvREFCNT_dec (key);
1111 goto fail;
1112 }
1113
1114 hv_store_ent (hv, key, value, 0);
905 SvREFCNT_dec (key); 1115 SvREFCNT_dec (key);
1116
1117 break;
1118 }
1119 else if (*p == '"')
1120 {
1121 // fast path, got a simple key
1122 char *key = dec->cur;
1123 int len = p - key;
1124 dec->cur = p + 1;
1125
1126 decode_ws (dec); EXPECT_CH (':');
1127
1128 value = decode_sv (dec);
1129 if (!value)
906 goto fail; 1130 goto fail;
1131
1132 hv_store (hv, key, len, value, 0);
1133
1134 break;
1135 }
1136
1137 ++p;
907 } 1138 }
908 1139 }
909 hv_store_ent (hv, key, value, 0);
910 SvREFCNT_dec (key);
911 1140
912 decode_ws (dec); 1141 decode_ws (dec);
913 1142
914 if (*dec->cur == '}') 1143 if (*dec->cur == '}')
915 { 1144 {
… …
922 1151
923 ++dec->cur; 1152 ++dec->cur;
924 } 1153 }
925 1154
926 DEC_DEC_DEPTH; 1155 DEC_DEC_DEPTH;
927 return newRV_noinc ((SV *)hv); 1156 sv = newRV_noinc ((SV *)hv);
1157
1158 // check filter callbacks
1159 if (dec->json.flags & F_HOOK)
1160 {
1161 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1162 {
1163 HE *cb, *he;
1164
1165 hv_iterinit (hv);
1166 he = hv_iternext (hv);
1167 hv_iterinit (hv);
1168
1169 // the next line creates a mortal sv each time its called.
1170 // might want to optimise this for common cases.
1171 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1172
1173 if (cb)
1174 {
1175 dSP;
1176 int count;
1177
1178 ENTER; SAVETMPS; PUSHMARK (SP);
1179 XPUSHs (HeVAL (he));
1180
1181 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1182
1183 if (count == 1)
1184 {
1185 sv = newSVsv (POPs);
1186 FREETMPS; LEAVE;
1187 return sv;
1188 }
1189
1190 FREETMPS; LEAVE;
1191 }
1192 }
1193
1194 if (dec->json.cb_object)
1195 {
1196 dSP;
1197 int count;
1198
1199 ENTER; SAVETMPS; PUSHMARK (SP);
1200 XPUSHs (sv_2mortal (sv));
1201
1202 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1203
1204 if (count == 1)
1205 {
1206 sv = newSVsv (POPs);
1207 FREETMPS; LEAVE;
1208 return sv;
1209 }
1210
1211 SvREFCNT_inc (sv);
1212 FREETMPS; LEAVE;
1213 }
1214 }
1215
1216 return sv;
928 1217
929fail: 1218fail:
930 SvREFCNT_dec (hv); 1219 SvREFCNT_dec (hv);
931 DEC_DEC_DEPTH; 1220 DEC_DEC_DEPTH;
932 return 0; 1221 return 0;
… …
934 1223
935static SV * 1224static SV *
936decode_sv (dec_t *dec) 1225decode_sv (dec_t *dec)
937{ 1226{
938 decode_ws (dec); 1227 decode_ws (dec);
1228
1229 // the beauty of JSON: you need exactly one character lookahead
1230 // to parse anything.
939 switch (*dec->cur) 1231 switch (*dec->cur)
940 { 1232 {
941 case '"': ++dec->cur; return decode_str (dec); 1233 case '"': ++dec->cur; return decode_str (dec);
942 case '[': ++dec->cur; return decode_av (dec); 1234 case '[': ++dec->cur; return decode_av (dec);
943 case '{': ++dec->cur; return decode_hv (dec); 1235 case '{': ++dec->cur; return decode_hv (dec);
… …
949 1241
950 case 't': 1242 case 't':
951 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1243 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
952 { 1244 {
953 dec->cur += 4; 1245 dec->cur += 4;
954 return newSViv (1); 1246#if JSON_SLOW
1247 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1248#endif
1249 return SvREFCNT_inc (json_true);
955 } 1250 }
956 else 1251 else
957 ERR ("'true' expected"); 1252 ERR ("'true' expected");
958 1253
959 break; 1254 break;
960 1255
961 case 'f': 1256 case 'f':
962 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1257 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
963 { 1258 {
964 dec->cur += 5; 1259 dec->cur += 5;
965 return newSViv (0); 1260#if JSON_SLOW
1261 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1262#endif
1263 return SvREFCNT_inc (json_false);
966 } 1264 }
967 else 1265 else
968 ERR ("'false' expected"); 1266 ERR ("'false' expected");
969 1267
970 break; 1268 break;
… …
988fail: 1286fail:
989 return 0; 1287 return 0;
990} 1288}
991 1289
992static SV * 1290static SV *
993decode_json (SV *string, U32 flags, UV *offset_return) 1291decode_json (SV *string, JSON *json, UV *offset_return)
994{ 1292{
995 dec_t dec; 1293 dec_t dec;
996 UV offset; 1294 UV offset;
997 SV *sv; 1295 SV *sv;
998 1296
999 SvGETMAGIC (string); 1297 SvGETMAGIC (string);
1000 SvUPGRADE (string, SVt_PV); 1298 SvUPGRADE (string, SVt_PV);
1001 1299
1300 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1301 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1302 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1303
1002 if (flags & F_UTF8) 1304 if (json->flags & F_UTF8)
1003 sv_utf8_downgrade (string, 0); 1305 sv_utf8_downgrade (string, 0);
1004 else 1306 else
1005 sv_utf8_upgrade (string); 1307 sv_utf8_upgrade (string);
1006 1308
1007 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1309 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1008 1310
1009 dec.flags = flags; 1311 dec.json = *json;
1010 dec.cur = SvPVX (string); 1312 dec.cur = SvPVX (string);
1011 dec.end = SvEND (string); 1313 dec.end = SvEND (string);
1012 dec.err = 0; 1314 dec.err = 0;
1013 dec.depth = 0; 1315 dec.depth = 0;
1014 dec.maxdepth = DEC_DEPTH (dec.flags); 1316 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1317
1318 if (dec.json.cb_object || dec.json.cb_sk_object)
1319 dec.json.flags |= F_HOOK;
1015 1320
1016 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1321 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1017 sv = decode_sv (&dec); 1322 sv = decode_sv (&dec);
1018 1323
1019 if (!(offset_return || !sv)) 1324 if (!(offset_return || !sv))
… …
1029 } 1334 }
1030 } 1335 }
1031 1336
1032 if (offset_return || !sv) 1337 if (offset_return || !sv)
1033 { 1338 {
1034 offset = dec.flags & F_UTF8 1339 offset = dec.json.flags & F_UTF8
1035 ? dec.cur - SvPVX (string) 1340 ? dec.cur - SvPVX (string)
1036 : utf8_distance (dec.cur, SvPVX (string)); 1341 : utf8_distance (dec.cur, SvPVX (string));
1037 1342
1038 if (offset_return) 1343 if (offset_return)
1039 *offset_return = offset; 1344 *offset_return = offset;
… …
1058 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1363 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1059 } 1364 }
1060 1365
1061 sv = sv_2mortal (sv); 1366 sv = sv_2mortal (sv);
1062 1367
1063 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1368 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1064 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1369 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1065 1370
1066 return sv; 1371 return sv;
1067} 1372}
1068 1373
… …
1072MODULE = JSON::XS PACKAGE = JSON::XS 1377MODULE = JSON::XS PACKAGE = JSON::XS
1073 1378
1074BOOT: 1379BOOT:
1075{ 1380{
1076 int i; 1381 int i;
1077
1078 memset (decode_hexdigit, 0xff, 256);
1079 1382
1080 for (i = 0; i < 256; ++i) 1383 for (i = 0; i < 256; ++i)
1081 decode_hexdigit [i] = 1384 decode_hexdigit [i] =
1082 i >= '0' && i <= '9' ? i - '0' 1385 i >= '0' && i <= '9' ? i - '0'
1083 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1386 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1084 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1387 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1085 : -1; 1388 : -1;
1086 1389
1087 json_stash = gv_stashpv ("JSON::XS", 1); 1390 json_stash = gv_stashpv ("JSON::XS" , 1);
1391 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1392
1393 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1394 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1088} 1395}
1089 1396
1090PROTOTYPES: DISABLE 1397PROTOTYPES: DISABLE
1091 1398
1092SV *new (char *dummy) 1399void CLONE (...)
1093 CODE: 1400 CODE:
1094 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1401 json_stash = 0;
1095 OUTPUT: 1402 json_boolean_stash = 0;
1096 RETVAL
1097 1403
1404void new (char *klass)
1405 PPCODE:
1406{
1407 SV *pv = NEWSV (0, sizeof (JSON));
1408 SvPOK_only (pv);
1409 Zero (SvPVX (pv), 1, JSON);
1410 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1411 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1412}
1413
1098SV *ascii (SV *self, int enable = 1) 1414void ascii (JSON *self, int enable = 1)
1099 ALIAS: 1415 ALIAS:
1100 ascii = F_ASCII 1416 ascii = F_ASCII
1101 latin1 = F_LATIN1 1417 latin1 = F_LATIN1
1102 utf8 = F_UTF8 1418 utf8 = F_UTF8
1103 indent = F_INDENT 1419 indent = F_INDENT
1104 canonical = F_CANONICAL 1420 canonical = F_CANONICAL
1105 space_before = F_SPACE_BEFORE 1421 space_before = F_SPACE_BEFORE
1106 space_after = F_SPACE_AFTER 1422 space_after = F_SPACE_AFTER
1107 pretty = F_PRETTY 1423 pretty = F_PRETTY
1108 allow_nonref = F_ALLOW_NONREF 1424 allow_nonref = F_ALLOW_NONREF
1109 shrink = F_SHRINK 1425 shrink = F_SHRINK
1426 allow_blessed = F_ALLOW_BLESSED
1427 convert_blessed = F_CONV_BLESSED
1110 CODE: 1428 PPCODE:
1111{ 1429{
1112 UV *uv = SvJSON (self);
1113 if (enable) 1430 if (enable)
1114 *uv |= ix; 1431 self->flags |= ix;
1115 else 1432 else
1116 *uv &= ~ix; 1433 self->flags &= ~ix;
1117 1434
1118 RETVAL = newSVsv (self); 1435 XPUSHs (ST (0));
1119} 1436}
1120 OUTPUT:
1121 RETVAL
1122 1437
1123SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1438void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1124 CODE: 1439 PPCODE:
1125{ 1440{
1126 UV *uv = SvJSON (self);
1127 UV log2 = 0; 1441 UV log2 = 0;
1128 1442
1129 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1443 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1130 1444
1131 while ((1UL << log2) < max_depth) 1445 while ((1UL << log2) < max_depth)
1132 ++log2; 1446 ++log2;
1133 1447
1134 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1448 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1135 1449
1136 RETVAL = newSVsv (self); 1450 XPUSHs (ST (0));
1137} 1451}
1138 OUTPUT:
1139 RETVAL
1140 1452
1141void encode (SV *self, SV *scalar) 1453void max_size (JSON *self, UV max_size = 0)
1142 PPCODE: 1454 PPCODE:
1143 XPUSHs (encode_json (scalar, *SvJSON (self))); 1455{
1456 UV log2 = 0;
1144 1457
1145void decode (SV *self, SV *jsonstr) 1458 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1459 if (max_size == 1) max_size = 2;
1460
1461 while ((1UL << log2) < max_size)
1462 ++log2;
1463
1464 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1465
1466 XPUSHs (ST (0));
1467}
1468
1469void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1146 PPCODE: 1470 PPCODE:
1471{
1472 SvREFCNT_dec (self->cb_object);
1473 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1474
1475 XPUSHs (ST (0));
1476}
1477
1478void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1479 PPCODE:
1480{
1481 if (!self->cb_sk_object)
1482 self->cb_sk_object = newHV ();
1483
1484 if (SvOK (cb))
1485 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1486 else
1487 {
1488 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1489
1490 if (!HvKEYS (self->cb_sk_object))
1491 {
1492 SvREFCNT_dec (self->cb_sk_object);
1493 self->cb_sk_object = 0;
1494 }
1495 }
1496
1497 XPUSHs (ST (0));
1498}
1499
1500void encode (JSON *self, SV *scalar)
1501 PPCODE:
1502 XPUSHs (encode_json (scalar, self));
1503
1504void decode (JSON *self, SV *jsonstr)
1505 PPCODE:
1147 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1506 XPUSHs (decode_json (jsonstr, self, 0));
1148 1507
1149void decode_prefix (SV *self, SV *jsonstr) 1508void decode_prefix (JSON *self, SV *jsonstr)
1150 PPCODE: 1509 PPCODE:
1151{ 1510{
1152 UV offset; 1511 UV offset;
1153 EXTEND (SP, 2); 1512 EXTEND (SP, 2);
1154 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1513 PUSHs (decode_json (jsonstr, self, &offset));
1155 PUSHs (sv_2mortal (newSVuv (offset))); 1514 PUSHs (sv_2mortal (newSVuv (offset)));
1156} 1515}
1157 1516
1517void DESTROY (JSON *self)
1518 CODE:
1519 SvREFCNT_dec (self->cb_sk_object);
1520 SvREFCNT_dec (self->cb_object);
1521
1158PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
1159 1523
1160void to_json (SV *scalar) 1524void to_json (SV *scalar)
1161 ALIAS:
1162 objToJson = 0
1163 PPCODE: 1525 PPCODE:
1526{
1527 JSON json = { F_DEFAULT | F_UTF8 };
1164 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1528 XPUSHs (encode_json (scalar, &json));
1529}
1165 1530
1166void from_json (SV *jsonstr) 1531void from_json (SV *jsonstr)
1167 ALIAS:
1168 jsonToObj = 0
1169 PPCODE: 1532 PPCODE:
1533{
1534 JSON json = { F_DEFAULT | F_UTF8 };
1170 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1535 XPUSHs (decode_json (jsonstr, &json, 0));
1536}
1171 1537

Diff Legend

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