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 1.29 by root, Wed May 9 15:34:04 2007 UTC vs.
Revision 1.63 by root, Mon Aug 27 01:49:01 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
22#define F_LATIN1 0x00000002UL
15#define F_UTF8 0x00000002UL 23#define F_UTF8 0x00000004UL
16#define F_INDENT 0x00000004UL 24#define F_INDENT 0x00000008UL
17#define F_CANONICAL 0x00000008UL 25#define F_CANONICAL 0x00000010UL
18#define F_SPACE_BEFORE 0x00000010UL 26#define F_SPACE_BEFORE 0x00000020UL
19#define F_SPACE_AFTER 0x00000020UL 27#define F_SPACE_AFTER 0x00000040UL
20#define F_ALLOW_NONREF 0x00000080UL 28#define F_ALLOW_NONREF 0x00000100UL
21#define F_SHRINK 0x00000100UL 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL
33
22#define F_MAXDEPTH 0xf8000000UL 34#define F_MAXDEPTH 0xf8000000UL
23#define S_MAXDEPTH 27 35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
24 39
25#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
26 41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
27// F_SELFCONVERT? <=> to_json/toJson
28// F_BLESSED? <=> { $__class__$ => }
29 42
30#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
31#define F_DEFAULT (9UL << S_MAXDEPTH) 44#define F_DEFAULT (9UL << S_MAXDEPTH)
32 45
33#define INIT_SIZE 32 // initial scalar size to be allocated 46#define INIT_SIZE 32 // initial scalar size to be allocated
34#define INDENT_STEP 3 // spaces per indentation level 47#define INDENT_STEP 3 // spaces per indentation level
35 48
36#define SHORT_STRING_LEN 512 // special-case strings of up to this size 49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
37 50
38#define SB do { 51#define SB do {
39#define SE } while (0) 52#define SE } while (0)
40 53
54#if __GNUC__ >= 3
55# define expect(expr,value) __builtin_expect ((expr),(value))
56# define inline inline
57#else
58# define expect(expr,value) (expr)
59# define inline static
60#endif
61
62#define expect_false(expr) expect ((expr) != 0, 0)
63#define expect_true(expr) expect ((expr) != 0, 1)
64
65#ifdef USE_ITHREADS
66# define JSON_SLOW 1
67# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
68#else
69# define JSON_SLOW 0
70# define JSON_STASH json_stash
71#endif
72
41static HV *json_stash; // JSON::XS:: 73static HV *json_stash, *json_boolean_stash; // JSON::XS::
74static SV *json_true, *json_false;
75
76typedef struct {
77 U32 flags;
78 SV *cb_object;
79 HV *cb_sk_object;
80} JSON;
42 81
43///////////////////////////////////////////////////////////////////////////// 82/////////////////////////////////////////////////////////////////////////////
44// utility functions 83// utility functions
45 84
46static UV * 85inline void
47SvJSON (SV *sv)
48{
49 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
50 croak ("object is not of type JSON::XS");
51
52 return &SvUVX (SvRV (sv));
53}
54
55static void
56shrink (SV *sv) 86shrink (SV *sv)
57{ 87{
58 sv_utf8_downgrade (sv, 1); 88 sv_utf8_downgrade (sv, 1);
59 if (SvLEN (sv) > SvCUR (sv) + 1) 89 if (SvLEN (sv) > SvCUR (sv) + 1)
60 { 90 {
… …
69// decode an utf-8 character and return it, or (UV)-1 in 99// decode an utf-8 character and return it, or (UV)-1 in
70// case of an error. 100// case of an error.
71// we special-case "safe" characters from U+80 .. U+7FF, 101// we special-case "safe" characters from U+80 .. U+7FF,
72// but use the very good perl function to parse anything else. 102// but use the very good perl function to parse anything else.
73// note that we never call this function for a ascii codepoints 103// note that we never call this function for a ascii codepoints
74static UV 104inline UV
75decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
76{ 106{
77 if (s[0] > 0xdf || s[0] < 0xc2) 107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
78 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
79 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
80 { 110 {
81 *clen = 2; 111 *clen = 2;
82 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
… …
95typedef struct 125typedef struct
96{ 126{
97 char *cur; // SvPVX (sv) + current output position 127 char *cur; // SvPVX (sv) + current output position
98 char *end; // SvEND (sv) 128 char *end; // SvEND (sv)
99 SV *sv; // result scalar 129 SV *sv; // result scalar
100 U32 flags; // F_* 130 JSON json;
101 U32 indent; // indentation level 131 U32 indent; // indentation level
102 U32 maxdepth; // max. indentation/recursion level 132 U32 maxdepth; // max. indentation/recursion level
103} enc_t; 133} enc_t;
104 134
105static void 135inline void
106need (enc_t *enc, STRLEN len) 136need (enc_t *enc, STRLEN len)
107{ 137{
108 if (enc->cur + len >= enc->end) 138 if (expect_false (enc->cur + len >= enc->end))
109 { 139 {
110 STRLEN cur = enc->cur - SvPVX (enc->sv); 140 STRLEN cur = enc->cur - SvPVX (enc->sv);
111 SvGROW (enc->sv, cur + len + 1); 141 SvGROW (enc->sv, cur + len + 1);
112 enc->cur = SvPVX (enc->sv) + cur; 142 enc->cur = SvPVX (enc->sv) + cur;
113 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
114 } 144 }
115} 145}
116 146
117static void 147inline void
118encode_ch (enc_t *enc, char ch) 148encode_ch (enc_t *enc, char ch)
119{ 149{
120 need (enc, 1); 150 need (enc, 1);
121 *enc->cur++ = ch; 151 *enc->cur++ = ch;
122} 152}
… …
130 160
131 while (str < end) 161 while (str < end)
132 { 162 {
133 unsigned char ch = *(unsigned char *)str; 163 unsigned char ch = *(unsigned char *)str;
134 164
135 if (ch >= 0x20 && ch < 0x80) // most common case 165 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
136 { 166 {
137 if (ch == '"') // but with slow exceptions 167 if (expect_false (ch == '"')) // but with slow exceptions
138 { 168 {
139 need (enc, len += 1); 169 need (enc, len += 1);
140 *enc->cur++ = '\\'; 170 *enc->cur++ = '\\';
141 *enc->cur++ = '"'; 171 *enc->cur++ = '"';
142 } 172 }
143 else if (ch == '\\') 173 else if (expect_false (ch == '\\'))
144 { 174 {
145 need (enc, len += 1); 175 need (enc, len += 1);
146 *enc->cur++ = '\\'; 176 *enc->cur++ = '\\';
147 *enc->cur++ = '\\'; 177 *enc->cur++ = '\\';
148 } 178 }
… …
166 STRLEN clen; 196 STRLEN clen;
167 UV uch; 197 UV uch;
168 198
169 if (is_utf8) 199 if (is_utf8)
170 { 200 {
171 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
172 uch = decode_utf8 (str, end - str, &clen); 201 uch = decode_utf8 (str, end - str, &clen);
173 if (clen == (STRLEN)-1) 202 if (clen == (STRLEN)-1)
174 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 203 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
175 } 204 }
176 else 205 else
… …
180 } 209 }
181 210
182 if (uch > 0x10FFFFUL) 211 if (uch > 0x10FFFFUL)
183 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 212 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
184 213
185 if (uch < 0x80 || enc->flags & F_ASCII) 214 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
186 { 215 {
187 if (uch > 0xFFFFUL) 216 if (uch > 0xFFFFUL)
188 { 217 {
189 need (enc, len += 11); 218 need (enc, len += 11);
190 sprintf (enc->cur, "\\u%04x\\u%04x", 219 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
204 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 233 *enc->cur++ = hexdigit [(uch >> 0) & 15];
205 } 234 }
206 235
207 str += clen; 236 str += clen;
208 } 237 }
238 else if (enc->json.flags & F_LATIN1)
239 {
240 *enc->cur++ = uch;
241 str += clen;
242 }
209 else if (is_utf8) 243 else if (is_utf8)
210 { 244 {
211 need (enc, len += clen); 245 need (enc, len += clen);
212 do 246 do
213 { 247 {
… …
227 261
228 --len; 262 --len;
229 } 263 }
230} 264}
231 265
232static void 266inline void
233encode_indent (enc_t *enc) 267encode_indent (enc_t *enc)
234{ 268{
235 if (enc->flags & F_INDENT) 269 if (enc->json.flags & F_INDENT)
236 { 270 {
237 int spaces = enc->indent * INDENT_STEP; 271 int spaces = enc->indent * INDENT_STEP;
238 272
239 need (enc, spaces); 273 need (enc, spaces);
240 memset (enc->cur, ' ', spaces); 274 memset (enc->cur, ' ', spaces);
241 enc->cur += spaces; 275 enc->cur += spaces;
242 } 276 }
243} 277}
244 278
245static void 279inline void
246encode_space (enc_t *enc) 280encode_space (enc_t *enc)
247{ 281{
248 need (enc, 1); 282 need (enc, 1);
249 encode_ch (enc, ' '); 283 encode_ch (enc, ' ');
250} 284}
251 285
252static void 286inline void
253encode_nl (enc_t *enc) 287encode_nl (enc_t *enc)
254{ 288{
255 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
256 { 290 {
257 need (enc, 1); 291 need (enc, 1);
258 encode_ch (enc, '\n'); 292 encode_ch (enc, '\n');
259 } 293 }
260} 294}
261 295
262static void 296inline void
263encode_comma (enc_t *enc) 297encode_comma (enc_t *enc)
264{ 298{
265 encode_ch (enc, ','); 299 encode_ch (enc, ',');
266 300
267 if (enc->flags & F_INDENT) 301 if (enc->json.flags & F_INDENT)
268 encode_nl (enc); 302 encode_nl (enc);
269 else if (enc->flags & F_SPACE_AFTER) 303 else if (enc->json.flags & F_SPACE_AFTER)
270 encode_space (enc); 304 encode_space (enc);
271} 305}
272 306
273static void encode_sv (enc_t *enc, SV *sv); 307static void encode_sv (enc_t *enc, SV *sv);
274 308
… …
283 encode_ch (enc, '['); encode_nl (enc); 317 encode_ch (enc, '['); encode_nl (enc);
284 ++enc->indent; 318 ++enc->indent;
285 319
286 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
287 { 321 {
322 SV **svp = av_fetch (av, i, 0);
323
288 encode_indent (enc); 324 encode_indent (enc);
289 encode_sv (enc, *av_fetch (av, i, 0)); 325
326 if (svp)
327 encode_sv (enc, *svp);
328 else
329 encode_str (enc, "null", 4, 0);
290 330
291 if (i < len) 331 if (i < len)
292 encode_comma (enc); 332 encode_comma (enc);
293 } 333 }
294 334
… …
297 --enc->indent; 337 --enc->indent;
298 encode_indent (enc); encode_ch (enc, ']'); 338 encode_indent (enc); encode_ch (enc, ']');
299} 339}
300 340
301static void 341static void
302encode_he (enc_t *enc, HE *he) 342encode_hk (enc_t *enc, HE *he)
303{ 343{
304 encode_ch (enc, '"'); 344 encode_ch (enc, '"');
305 345
306 if (HeKLEN (he) == HEf_SVKEY) 346 if (HeKLEN (he) == HEf_SVKEY)
307 { 347 {
… …
317 else 357 else
318 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 358 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
319 359
320 encode_ch (enc, '"'); 360 encode_ch (enc, '"');
321 361
322 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 362 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
323 encode_ch (enc, ':'); 363 encode_ch (enc, ':');
324 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 364 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
325 encode_sv (enc, HeVAL (he));
326} 365}
327 366
328// compare hash entries, used when all keys are bytestrings 367// compare hash entries, used when all keys are bytestrings
329static int 368static int
330he_cmp_fast (const void *a_, const void *b_) 369he_cmp_fast (const void *a_, const void *b_)
… …
335 HE *b = *(HE **)b_; 374 HE *b = *(HE **)b_;
336 375
337 STRLEN la = HeKLEN (a); 376 STRLEN la = HeKLEN (a);
338 STRLEN lb = HeKLEN (b); 377 STRLEN lb = HeKLEN (b);
339 378
340 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 379 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
341 cmp = la - lb; 380 cmp = lb - la;
342 381
343 return cmp; 382 return cmp;
344} 383}
345 384
346// compare hash entries, used when some keys are sv's or utf-x 385// compare hash entries, used when some keys are sv's or utf-x
347static int 386static int
348he_cmp_slow (const void *a, const void *b) 387he_cmp_slow (const void *a, const void *b)
349{ 388{
350 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 389 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
351} 390}
352 391
353static void 392static void
354encode_hv (enc_t *enc, HV *hv) 393encode_hv (enc_t *enc, HV *hv)
355{ 394{
395 HE *he;
356 int count, i; 396 int count;
357 397
358 if (enc->indent >= enc->maxdepth) 398 if (enc->indent >= enc->maxdepth)
359 croak ("data structure too deep (hit recursion limit)"); 399 croak ("data structure too deep (hit recursion limit)");
360 400
361 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 401 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
362 402
363 if ((count = hv_iterinit (hv)))
364 {
365 // for canonical output we have to sort by keys first 403 // for canonical output we have to sort by keys first
366 // actually, this is mostly due to the stupid so-called 404 // actually, this is mostly due to the stupid so-called
367 // security workaround added somewhere in 5.8.x. 405 // security workaround added somewhere in 5.8.x.
368 // that randomises hash orderings 406 // that randomises hash orderings
369 if (enc->flags & F_CANONICAL) 407 if (enc->json.flags & F_CANONICAL)
408 {
409 int count = hv_iterinit (hv);
410
411 if (SvMAGICAL (hv))
370 { 412 {
413 // need to count by iterating. could improve by dynamically building the vector below
414 // but I don't care for the speed of this special case.
415 // note also that we will run into undefined behaviour when the two iterations
416 // do not result in the same count, something I might care for in some later release.
417
418 count = 0;
419 while (hv_iternext (hv))
420 ++count;
421
422 hv_iterinit (hv);
423 }
424
425 if (count)
426 {
427 int i, fast = 1;
428#if defined(__BORLANDC__) || defined(_MSC_VER)
429 HE **hes = _alloca (count * sizeof (HE));
430#else
371 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 431 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
372 int fast = 1; 432#endif
373 433
374 i = 0; 434 i = 0;
375 while ((he = hv_iternext (hv))) 435 while ((he = hv_iternext (hv)))
376 { 436 {
377 hes [i++] = he; 437 hes [i++] = he;
… …
399 459
400 FREETMPS; 460 FREETMPS;
401 LEAVE; 461 LEAVE;
402 } 462 }
403 463
404 for (i = 0; i < count; ++i) 464 while (count--)
405 { 465 {
406 encode_indent (enc); 466 encode_indent (enc);
467 he = hes [count];
407 encode_he (enc, hes [i]); 468 encode_hk (enc, he);
469 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
408 470
409 if (i < count - 1) 471 if (count)
410 encode_comma (enc); 472 encode_comma (enc);
411 } 473 }
412 474 }
475 }
476 else
477 {
478 if (hv_iterinit (hv) || SvMAGICAL (hv))
479 if ((he = hv_iternext (hv)))
480 for (;;)
481 {
413 encode_nl (enc); 482 encode_indent (enc);
483 encode_hk (enc, he);
484 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
485
486 if (!(he = hv_iternext (hv)))
487 break;
488
489 encode_comma (enc);
490 }
491 }
492
493 encode_nl (enc);
494
495 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
496}
497
498// encode objects, arrays and special \0=false and \1=true values.
499static void
500encode_rv (enc_t *enc, SV *sv)
501{
502 svtype svt;
503
504 SvGETMAGIC (sv);
505 svt = SvTYPE (sv);
506
507 if (expect_false (SvOBJECT (sv)))
508 {
509 HV *stash = !JSON_SLOW || json_boolean_stash
510 ? json_boolean_stash
511 : gv_stashpv ("JSON::XS::Boolean", 1);
512
513 if (SvSTASH (sv) == stash)
514 {
515 if (SvIV (sv))
516 encode_str (enc, "true", 4, 0);
517 else
518 encode_str (enc, "false", 5, 0);
414 } 519 }
415 else 520 else
416 { 521 {
417 HE *he = hv_iternext (hv); 522#if 0
418 523 if (0 && sv_derived_from (rv, "JSON::Literal"))
419 for (;;)
420 {
421 encode_indent (enc);
422 encode_he (enc, he);
423
424 if (!(he = hv_iternext (hv)))
425 break;
426
427 encode_comma (enc);
428 } 524 {
525 // not yet
526 }
527#endif
528 if (enc->json.flags & F_CONV_BLESSED)
529 {
530 // we re-bless the reference to get overload and other niceties right
531 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
429 532
533 if (to_json)
534 {
535 dSP;
536
537 ENTER; SAVETMPS; PUSHMARK (SP);
538 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
539
540 // calling with G_SCALAR ensures that we always get a 1 return value
541 PUTBACK;
542 call_sv ((SV *)GvCV (to_json), G_SCALAR);
543 SPAGAIN;
544
545 // catch this surprisingly common error
546 if (SvROK (TOPs) && SvRV (TOPs) == sv)
547 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
548
549 sv = POPs;
550 PUTBACK;
551
430 encode_nl (enc); 552 encode_sv (enc, sv);
553
554 FREETMPS; LEAVE;
555 }
556 else if (enc->json.flags & F_ALLOW_BLESSED)
557 encode_str (enc, "null", 4, 0);
558 else
559 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
560 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
561 }
562 else if (enc->json.flags & F_ALLOW_BLESSED)
563 encode_str (enc, "null", 4, 0);
564 else
565 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
566 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
431 } 567 }
432 } 568 }
433
434 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
435}
436
437// encode objects, arrays and special \0=false and \1=true values.
438static void
439encode_rv (enc_t *enc, SV *sv)
440{
441 svtype svt;
442
443 SvGETMAGIC (sv);
444 svt = SvTYPE (sv);
445
446 if (svt == SVt_PVHV) 569 else if (svt == SVt_PVHV)
447 encode_hv (enc, (HV *)sv); 570 encode_hv (enc, (HV *)sv);
448 else if (svt == SVt_PVAV) 571 else if (svt == SVt_PVAV)
449 encode_av (enc, (AV *)sv); 572 encode_av (enc, (AV *)sv);
450 else if (svt < SVt_PVAV) 573 else if (svt < SVt_PVAV)
451 { 574 {
452 if (SvNIOK (sv) && SvIV (sv) == 0) 575 STRLEN len = 0;
576 char *pv = svt ? SvPV (sv, len) : 0;
577
578 if (len == 1 && *pv == '1')
579 encode_str (enc, "true", 4, 0);
580 else if (len == 1 && *pv == '0')
453 encode_str (enc, "false", 5, 0); 581 encode_str (enc, "false", 5, 0);
454 else if (SvNIOK (sv) && SvIV (sv) == 1)
455 encode_str (enc, "true", 4, 0);
456 else 582 else
457 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 583 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
458 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 584 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
459 } 585 }
460 else 586 else
… …
475 encode_str (enc, str, len, SvUTF8 (sv)); 601 encode_str (enc, str, len, SvUTF8 (sv));
476 encode_ch (enc, '"'); 602 encode_ch (enc, '"');
477 } 603 }
478 else if (SvNOKp (sv)) 604 else if (SvNOKp (sv))
479 { 605 {
606 // trust that perl will do the right thing w.r.t. JSON syntax.
480 need (enc, NV_DIG + 32); 607 need (enc, NV_DIG + 32);
481 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 608 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
482 enc->cur += strlen (enc->cur); 609 enc->cur += strlen (enc->cur);
483 } 610 }
484 else if (SvIOKp (sv)) 611 else if (SvIOKp (sv))
485 { 612 {
486 need (enc, 64); 613 // we assume we can always read an IV as a UV
614 if (SvUV (sv) & ~(UV)0x7fff)
615 {
616 // large integer, use the (rather slow) snprintf way.
617 need (enc, sizeof (UV) * 3);
487 enc->cur += 618 enc->cur +=
488 SvIsUV(sv) 619 SvIsUV(sv)
489 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 620 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
490 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 621 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
622 }
623 else
624 {
625 // optimise the "small number case"
626 // code will likely be branchless and use only a single multiplication
627 I32 i = SvIV (sv);
628 U32 u;
629 char digit, nz = 0;
630
631 need (enc, 6);
632
633 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
634 u = i < 0 ? -i : i;
635
636 // convert to 4.28 fixed-point representation
637 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
638
639 // now output digit by digit, each time masking out the integer part
640 // and multiplying by 5 while moving the decimal point one to the right,
641 // resulting in a net multiplication by 10.
642 // we always write the digit to memory but conditionally increment
643 // the pointer, to ease the usage of conditional move instructions.
644 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
645 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
646 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
647 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
648 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
649 }
491 } 650 }
492 else if (SvROK (sv)) 651 else if (SvROK (sv))
493 encode_rv (enc, SvRV (sv)); 652 encode_rv (enc, SvRV (sv));
494 else if (!SvOK (sv)) 653 else if (!SvOK (sv))
495 encode_str (enc, "null", 4, 0); 654 encode_str (enc, "null", 4, 0);
… …
497 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 656 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
498 SvPV_nolen (sv), SvFLAGS (sv)); 657 SvPV_nolen (sv), SvFLAGS (sv));
499} 658}
500 659
501static SV * 660static SV *
502encode_json (SV *scalar, U32 flags) 661encode_json (SV *scalar, JSON *json)
503{ 662{
504 enc_t enc; 663 enc_t enc;
505 664
506 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 665 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
507 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 666 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
508 667
509 enc.flags = flags; 668 enc.json = *json;
510 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 669 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
511 enc.cur = SvPVX (enc.sv); 670 enc.cur = SvPVX (enc.sv);
512 enc.end = SvEND (enc.sv); 671 enc.end = SvEND (enc.sv);
513 enc.indent = 0; 672 enc.indent = 0;
514 enc.maxdepth = DEC_DEPTH (flags); 673 enc.maxdepth = DEC_DEPTH (enc.json.flags);
515 674
516 SvPOK_only (enc.sv); 675 SvPOK_only (enc.sv);
517 encode_sv (&enc, scalar); 676 encode_sv (&enc, scalar);
518 677
519 if (!(flags & (F_ASCII | F_UTF8)))
520 SvUTF8_on (enc.sv);
521
522 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 678 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
523 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 679 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
524 680
681 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
682 SvUTF8_on (enc.sv);
683
525 if (enc.flags & F_SHRINK) 684 if (enc.json.flags & F_SHRINK)
526 shrink (enc.sv); 685 shrink (enc.sv);
527 686
528 return enc.sv; 687 return enc.sv;
529} 688}
530 689
… …
535typedef struct 694typedef struct
536{ 695{
537 char *cur; // current parser pointer 696 char *cur; // current parser pointer
538 char *end; // end of input string 697 char *end; // end of input string
539 const char *err; // parse error, if != 0 698 const char *err; // parse error, if != 0
540 U32 flags; // F_* 699 JSON json;
541 U32 depth; // recursion depth 700 U32 depth; // recursion depth
542 U32 maxdepth; // recursion depth limit 701 U32 maxdepth; // recursion depth limit
543} dec_t; 702} dec_t;
544 703
545static void 704inline void
546decode_ws (dec_t *dec) 705decode_ws (dec_t *dec)
547{ 706{
548 for (;;) 707 for (;;)
549 { 708 {
550 char ch = *dec->cur; 709 char ch = *dec->cur;
551 710
552 if (ch > 0x20 711 if (ch > 0x20
553 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 712 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
554 break; 713 break;
714
715 if (ch == '#' && dec->json.flags & F_RELAXED)
716 ++dec->cur;
555 717
556 ++dec->cur; 718 ++dec->cur;
557 } 719 }
558} 720}
559 721
… …
576decode_4hex (dec_t *dec) 738decode_4hex (dec_t *dec)
577{ 739{
578 signed char d1, d2, d3, d4; 740 signed char d1, d2, d3, d4;
579 unsigned char *cur = (unsigned char *)dec->cur; 741 unsigned char *cur = (unsigned char *)dec->cur;
580 742
581 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 743 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
582 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 744 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
583 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 745 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
584 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 746 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
585 747
586 dec->cur += 4; 748 dec->cur += 4;
587 749
588 return ((UV)d1) << 12 750 return ((UV)d1) << 12
589 | ((UV)d2) << 8 751 | ((UV)d2) << 8
… …
597static SV * 759static SV *
598decode_str (dec_t *dec) 760decode_str (dec_t *dec)
599{ 761{
600 SV *sv = 0; 762 SV *sv = 0;
601 int utf8 = 0; 763 int utf8 = 0;
764 char *dec_cur = dec->cur;
602 765
603 do 766 do
604 { 767 {
605 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 768 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
606 char *cur = buf; 769 char *cur = buf;
607 770
608 do 771 do
609 { 772 {
610 unsigned char ch = *(unsigned char *)dec->cur++; 773 unsigned char ch = *(unsigned char *)dec_cur++;
611 774
612 if (ch == '"') 775 if (expect_false (ch == '"'))
613 { 776 {
614 --dec->cur; 777 --dec_cur;
615 break; 778 break;
616 } 779 }
617 else if (ch == '\\') 780 else if (expect_false (ch == '\\'))
618 { 781 {
619 switch (*dec->cur) 782 switch (*dec_cur)
620 { 783 {
621 case '\\': 784 case '\\':
622 case '/': 785 case '/':
623 case '"': *cur++ = *dec->cur++; break; 786 case '"': *cur++ = *dec_cur++; break;
624 787
625 case 'b': ++dec->cur; *cur++ = '\010'; break; 788 case 'b': ++dec_cur; *cur++ = '\010'; break;
626 case 't': ++dec->cur; *cur++ = '\011'; break; 789 case 't': ++dec_cur; *cur++ = '\011'; break;
627 case 'n': ++dec->cur; *cur++ = '\012'; break; 790 case 'n': ++dec_cur; *cur++ = '\012'; break;
628 case 'f': ++dec->cur; *cur++ = '\014'; break; 791 case 'f': ++dec_cur; *cur++ = '\014'; break;
629 case 'r': ++dec->cur; *cur++ = '\015'; break; 792 case 'r': ++dec_cur; *cur++ = '\015'; break;
630 793
631 case 'u': 794 case 'u':
632 { 795 {
633 UV lo, hi; 796 UV lo, hi;
634 ++dec->cur; 797 ++dec_cur;
635 798
799 dec->cur = dec_cur;
636 hi = decode_4hex (dec); 800 hi = decode_4hex (dec);
801 dec_cur = dec->cur;
637 if (hi == (UV)-1) 802 if (hi == (UV)-1)
638 goto fail; 803 goto fail;
639 804
640 // possibly a surrogate pair 805 // possibly a surrogate pair
641 if (hi >= 0xd800) 806 if (hi >= 0xd800)
642 if (hi < 0xdc00) 807 if (hi < 0xdc00)
643 { 808 {
644 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 809 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
645 ERR ("missing low surrogate character in surrogate pair"); 810 ERR ("missing low surrogate character in surrogate pair");
646 811
647 dec->cur += 2; 812 dec_cur += 2;
648 813
814 dec->cur = dec_cur;
649 lo = decode_4hex (dec); 815 lo = decode_4hex (dec);
816 dec_cur = dec->cur;
650 if (lo == (UV)-1) 817 if (lo == (UV)-1)
651 goto fail; 818 goto fail;
652 819
653 if (lo < 0xdc00 || lo >= 0xe000) 820 if (lo < 0xdc00 || lo >= 0xe000)
654 ERR ("surrogate pair expected"); 821 ERR ("surrogate pair expected");
… …
668 *cur++ = hi; 835 *cur++ = hi;
669 } 836 }
670 break; 837 break;
671 838
672 default: 839 default:
673 --dec->cur; 840 --dec_cur;
674 ERR ("illegal backslash escape sequence in string"); 841 ERR ("illegal backslash escape sequence in string");
675 } 842 }
676 } 843 }
677 else if (ch >= 0x20 && ch <= 0x7f) 844 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
678 *cur++ = ch; 845 *cur++ = ch;
679 else if (ch >= 0x80) 846 else if (ch >= 0x80)
680 { 847 {
681 STRLEN clen; 848 STRLEN clen;
682 UV uch; 849 UV uch;
683 850
684 --dec->cur; 851 --dec_cur;
685 852
686 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 853 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
687 if (clen == (STRLEN)-1) 854 if (clen == (STRLEN)-1)
688 ERR ("malformed UTF-8 character in JSON string"); 855 ERR ("malformed UTF-8 character in JSON string");
689 856
690 do 857 do
691 *cur++ = *dec->cur++; 858 *cur++ = *dec_cur++;
692 while (--clen); 859 while (--clen);
693 860
694 utf8 = 1; 861 utf8 = 1;
695 } 862 }
696 else 863 else
697 { 864 {
698 --dec->cur; 865 --dec_cur;
699 866
700 if (!ch) 867 if (!ch)
701 ERR ("unexpected end of string while parsing JSON string"); 868 ERR ("unexpected end of string while parsing JSON string");
702 else 869 else
703 ERR ("invalid character encountered while parsing JSON string"); 870 ERR ("invalid character encountered while parsing JSON string");
… …
716 } 883 }
717 else 884 else
718 sv = newSVpvn (buf, len); 885 sv = newSVpvn (buf, len);
719 } 886 }
720 } 887 }
721 while (*dec->cur != '"'); 888 while (*dec_cur != '"');
722 889
723 ++dec->cur; 890 ++dec_cur;
724 891
725 if (sv) 892 if (sv)
726 { 893 {
727 SvPOK_only (sv); 894 SvPOK_only (sv);
728 *SvEND (sv) = 0; 895 *SvEND (sv) = 0;
… …
731 SvUTF8_on (sv); 898 SvUTF8_on (sv);
732 } 899 }
733 else 900 else
734 sv = newSVpvn ("", 0); 901 sv = newSVpvn ("", 0);
735 902
903 dec->cur = dec_cur;
736 return sv; 904 return sv;
737 905
738fail: 906fail:
907 dec->cur = dec_cur;
739 return 0; 908 return 0;
740} 909}
741 910
742static SV * 911static SV *
743decode_num (dec_t *dec) 912decode_num (dec_t *dec)
… …
801 is_nv = 1; 970 is_nv = 1;
802 } 971 }
803 972
804 if (!is_nv) 973 if (!is_nv)
805 { 974 {
806 UV uv; 975 int len = dec->cur - start;
807 int numtype = grok_number (start, dec->cur - start, &uv); 976
808 if (numtype & IS_NUMBER_IN_UV) 977 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
809 if (numtype & IS_NUMBER_NEG) 978 if (*start == '-')
979 switch (len)
810 { 980 {
811 if (uv < (UV)IV_MIN) 981 case 2: return newSViv (-( start [1] - '0' * 1));
812 return newSViv (-(IV)uv); 982 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
983 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
984 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
813 } 985 }
986 else
987 switch (len)
988 {
989 case 1: return newSViv ( start [0] - '0' * 1);
990 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
991 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
992 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
993 }
994
995 {
996 UV uv;
997 int numtype = grok_number (start, len, &uv);
998 if (numtype & IS_NUMBER_IN_UV)
999 if (numtype & IS_NUMBER_NEG)
1000 {
1001 if (uv < (UV)IV_MIN)
1002 return newSViv (-(IV)uv);
1003 }
814 else 1004 else
815 return newSVuv (uv); 1005 return newSVuv (uv);
816 } 1006 }
817 1007
1008 len -= *start == '-' ? 1 : 0;
1009
1010 // does not fit into IV or UV, try NV
1011 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1012 #if defined (LDBL_DIG)
1013 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1014 #endif
1015 )
1016 // fits into NV without loss of precision
1017 return newSVnv (Atof (start));
1018
1019 // everything else fails, convert it to a string
1020 return newSVpvn (start, dec->cur - start);
1021 }
1022
1023 // loss of precision here
818 return newSVnv (Atof (start)); 1024 return newSVnv (Atof (start));
819 1025
820fail: 1026fail:
821 return 0; 1027 return 0;
822} 1028}
… …
852 1058
853 if (*dec->cur != ',') 1059 if (*dec->cur != ',')
854 ERR (", or ] expected while parsing array"); 1060 ERR (", or ] expected while parsing array");
855 1061
856 ++dec->cur; 1062 ++dec->cur;
1063
1064 decode_ws (dec);
1065
1066 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1067 {
1068 ++dec->cur;
1069 break;
1070 }
857 } 1071 }
858 1072
859 DEC_DEC_DEPTH; 1073 DEC_DEC_DEPTH;
860 return newRV_noinc ((SV *)av); 1074 return newRV_noinc ((SV *)av);
861 1075
… …
866} 1080}
867 1081
868static SV * 1082static SV *
869decode_hv (dec_t *dec) 1083decode_hv (dec_t *dec)
870{ 1084{
1085 SV *sv;
871 HV *hv = newHV (); 1086 HV *hv = newHV ();
872 1087
873 DEC_INC_DEPTH; 1088 DEC_INC_DEPTH;
874 decode_ws (dec); 1089 decode_ws (dec);
875 1090
876 if (*dec->cur == '}') 1091 if (*dec->cur == '}')
877 ++dec->cur; 1092 ++dec->cur;
878 else 1093 else
879 for (;;) 1094 for (;;)
880 { 1095 {
1096 EXPECT_CH ('"');
1097
1098 // heuristic: assume that
1099 // a) decode_str + hv_store_ent are abysmally slow.
1100 // b) most hash keys are short, simple ascii text.
1101 // => try to "fast-match" such strings to avoid
1102 // the overhead of decode_str + hv_store_ent.
1103 {
881 SV *key, *value; 1104 SV *value;
1105 char *p = dec->cur;
1106 char *e = p + 24; // only try up to 24 bytes
882 1107
883 decode_ws (dec); EXPECT_CH ('"'); 1108 for (;;)
884
885 key = decode_str (dec);
886 if (!key)
887 goto fail;
888
889 decode_ws (dec); EXPECT_CH (':');
890
891 value = decode_sv (dec);
892 if (!value)
893 { 1109 {
1110 // the >= 0x80 is true on most architectures
1111 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1112 {
1113 // slow path, back up and use decode_str
1114 SV *key = decode_str (dec);
1115 if (!key)
1116 goto fail;
1117
1118 decode_ws (dec); EXPECT_CH (':');
1119
1120 decode_ws (dec);
1121 value = decode_sv (dec);
1122 if (!value)
1123 {
1124 SvREFCNT_dec (key);
1125 goto fail;
1126 }
1127
1128 hv_store_ent (hv, key, value, 0);
894 SvREFCNT_dec (key); 1129 SvREFCNT_dec (key);
1130
1131 break;
1132 }
1133 else if (*p == '"')
1134 {
1135 // fast path, got a simple key
1136 char *key = dec->cur;
1137 int len = p - key;
1138 dec->cur = p + 1;
1139
1140 decode_ws (dec); EXPECT_CH (':');
1141
1142 decode_ws (dec);
1143 value = decode_sv (dec);
1144 if (!value)
895 goto fail; 1145 goto fail;
1146
1147 hv_store (hv, key, len, value, 0);
1148
1149 break;
1150 }
1151
1152 ++p;
896 } 1153 }
897 1154 }
898 hv_store_ent (hv, key, value, 0);
899 SvREFCNT_dec (key);
900 1155
901 decode_ws (dec); 1156 decode_ws (dec);
902 1157
903 if (*dec->cur == '}') 1158 if (*dec->cur == '}')
904 { 1159 {
… …
908 1163
909 if (*dec->cur != ',') 1164 if (*dec->cur != ',')
910 ERR (", or } expected while parsing object/hash"); 1165 ERR (", or } expected while parsing object/hash");
911 1166
912 ++dec->cur; 1167 ++dec->cur;
1168
1169 decode_ws (dec);
1170
1171 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1172 {
1173 ++dec->cur;
1174 break;
1175 }
913 } 1176 }
914 1177
915 DEC_DEC_DEPTH; 1178 DEC_DEC_DEPTH;
916 return newRV_noinc ((SV *)hv); 1179 sv = newRV_noinc ((SV *)hv);
1180
1181 // check filter callbacks
1182 if (dec->json.flags & F_HOOK)
1183 {
1184 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1185 {
1186 HE *cb, *he;
1187
1188 hv_iterinit (hv);
1189 he = hv_iternext (hv);
1190 hv_iterinit (hv);
1191
1192 // the next line creates a mortal sv each time its called.
1193 // might want to optimise this for common cases.
1194 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1195
1196 if (cb)
1197 {
1198 dSP;
1199 int count;
1200
1201 ENTER; SAVETMPS; PUSHMARK (SP);
1202 XPUSHs (HeVAL (he));
1203
1204 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1205
1206 if (count == 1)
1207 {
1208 sv = newSVsv (POPs);
1209 FREETMPS; LEAVE;
1210 return sv;
1211 }
1212
1213 FREETMPS; LEAVE;
1214 }
1215 }
1216
1217 if (dec->json.cb_object)
1218 {
1219 dSP;
1220 int count;
1221
1222 ENTER; SAVETMPS; PUSHMARK (SP);
1223 XPUSHs (sv_2mortal (sv));
1224
1225 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1226
1227 if (count == 1)
1228 {
1229 sv = newSVsv (POPs);
1230 FREETMPS; LEAVE;
1231 return sv;
1232 }
1233
1234 SvREFCNT_inc (sv);
1235 FREETMPS; LEAVE;
1236 }
1237 }
1238
1239 return sv;
917 1240
918fail: 1241fail:
919 SvREFCNT_dec (hv); 1242 SvREFCNT_dec (hv);
920 DEC_DEC_DEPTH; 1243 DEC_DEC_DEPTH;
921 return 0; 1244 return 0;
922} 1245}
923 1246
924static SV * 1247static SV *
925decode_sv (dec_t *dec) 1248decode_sv (dec_t *dec)
926{ 1249{
927 decode_ws (dec); 1250 // the beauty of JSON: you need exactly one character lookahead
1251 // to parse anything.
928 switch (*dec->cur) 1252 switch (*dec->cur)
929 { 1253 {
930 case '"': ++dec->cur; return decode_str (dec); 1254 case '"': ++dec->cur; return decode_str (dec);
931 case '[': ++dec->cur; return decode_av (dec); 1255 case '[': ++dec->cur; return decode_av (dec);
932 case '{': ++dec->cur; return decode_hv (dec); 1256 case '{': ++dec->cur; return decode_hv (dec);
… …
938 1262
939 case 't': 1263 case 't':
940 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1264 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
941 { 1265 {
942 dec->cur += 4; 1266 dec->cur += 4;
943 return newSViv (1); 1267#if JSON_SLOW
1268 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1269#endif
1270 return SvREFCNT_inc (json_true);
944 } 1271 }
945 else 1272 else
946 ERR ("'true' expected"); 1273 ERR ("'true' expected");
947 1274
948 break; 1275 break;
949 1276
950 case 'f': 1277 case 'f':
951 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1278 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
952 { 1279 {
953 dec->cur += 5; 1280 dec->cur += 5;
954 return newSViv (0); 1281#if JSON_SLOW
1282 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1283#endif
1284 return SvREFCNT_inc (json_false);
955 } 1285 }
956 else 1286 else
957 ERR ("'false' expected"); 1287 ERR ("'false' expected");
958 1288
959 break; 1289 break;
… …
977fail: 1307fail:
978 return 0; 1308 return 0;
979} 1309}
980 1310
981static SV * 1311static SV *
982decode_json (SV *string, U32 flags) 1312decode_json (SV *string, JSON *json, UV *offset_return)
983{ 1313{
984 dec_t dec; 1314 dec_t dec;
1315 UV offset;
985 SV *sv; 1316 SV *sv;
986 1317
987 SvGETMAGIC (string); 1318 SvGETMAGIC (string);
988 SvUPGRADE (string, SVt_PV); 1319 SvUPGRADE (string, SVt_PV);
989 1320
1321 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1322 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1323 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1324
990 if (flags & F_UTF8) 1325 if (json->flags & F_UTF8)
991 sv_utf8_downgrade (string, 0); 1326 sv_utf8_downgrade (string, 0);
992 else 1327 else
993 sv_utf8_upgrade (string); 1328 sv_utf8_upgrade (string);
994 1329
995 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1330 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
996 1331
997 dec.flags = flags; 1332 dec.json = *json;
998 dec.cur = SvPVX (string); 1333 dec.cur = SvPVX (string);
999 dec.end = SvEND (string); 1334 dec.end = SvEND (string);
1000 dec.err = 0; 1335 dec.err = 0;
1001 dec.depth = 0; 1336 dec.depth = 0;
1002 dec.maxdepth = DEC_DEPTH (dec.flags); 1337 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1003 1338
1339 if (dec.json.cb_object || dec.json.cb_sk_object)
1340 dec.json.flags |= F_HOOK;
1341
1004 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1342 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1343
1344 decode_ws (&dec);
1005 sv = decode_sv (&dec); 1345 sv = decode_sv (&dec);
1006 1346
1347 if (!(offset_return || !sv))
1348 {
1349 // check for trailing garbage
1350 decode_ws (&dec);
1351
1352 if (*dec.cur)
1353 {
1354 dec.err = "garbage after JSON object";
1355 SvREFCNT_dec (sv);
1356 sv = 0;
1357 }
1358 }
1359
1360 if (offset_return || !sv)
1361 {
1362 offset = dec.json.flags & F_UTF8
1363 ? dec.cur - SvPVX (string)
1364 : utf8_distance (dec.cur, SvPVX (string));
1365
1366 if (offset_return)
1367 *offset_return = offset;
1368 }
1369
1007 if (!sv) 1370 if (!sv)
1008 { 1371 {
1009 IV offset = dec.flags & F_UTF8
1010 ? dec.cur - SvPVX (string)
1011 : utf8_distance (dec.cur, SvPVX (string));
1012 SV *uni = sv_newmortal (); 1372 SV *uni = sv_newmortal ();
1013 1373
1014 // horrible hack to silence warning inside pv_uni_display 1374 // horrible hack to silence warning inside pv_uni_display
1015 COP cop = *PL_curcop; 1375 COP cop = *PL_curcop;
1016 cop.cop_warnings = pWARN_NONE; 1376 cop.cop_warnings = pWARN_NONE;
… …
1026 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1386 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1027 } 1387 }
1028 1388
1029 sv = sv_2mortal (sv); 1389 sv = sv_2mortal (sv);
1030 1390
1031 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1391 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1032 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1392 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1033 1393
1034 return sv; 1394 return sv;
1035} 1395}
1036 1396
… …
1040MODULE = JSON::XS PACKAGE = JSON::XS 1400MODULE = JSON::XS PACKAGE = JSON::XS
1041 1401
1042BOOT: 1402BOOT:
1043{ 1403{
1044 int i; 1404 int i;
1045
1046 memset (decode_hexdigit, 0xff, 256);
1047 1405
1048 for (i = 0; i < 256; ++i) 1406 for (i = 0; i < 256; ++i)
1049 decode_hexdigit [i] = 1407 decode_hexdigit [i] =
1050 i >= '0' && i <= '9' ? i - '0' 1408 i >= '0' && i <= '9' ? i - '0'
1051 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1409 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1052 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1410 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1053 : -1; 1411 : -1;
1054 1412
1055 json_stash = gv_stashpv ("JSON::XS", 1); 1413 json_stash = gv_stashpv ("JSON::XS" , 1);
1414 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1415
1416 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1417 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1056} 1418}
1057 1419
1058PROTOTYPES: DISABLE 1420PROTOTYPES: DISABLE
1059 1421
1060SV *new (char *dummy) 1422void CLONE (...)
1061 CODE: 1423 CODE:
1062 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1424 json_stash = 0;
1063 OUTPUT: 1425 json_boolean_stash = 0;
1064 RETVAL
1065 1426
1427void new (char *klass)
1428 PPCODE:
1429{
1430 SV *pv = NEWSV (0, sizeof (JSON));
1431 SvPOK_only (pv);
1432 Zero (SvPVX (pv), 1, JSON);
1433 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1434 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1435}
1436
1066SV *ascii (SV *self, int enable = 1) 1437void ascii (JSON *self, int enable = 1)
1067 ALIAS: 1438 ALIAS:
1068 ascii = F_ASCII 1439 ascii = F_ASCII
1440 latin1 = F_LATIN1
1069 utf8 = F_UTF8 1441 utf8 = F_UTF8
1070 indent = F_INDENT 1442 indent = F_INDENT
1071 canonical = F_CANONICAL 1443 canonical = F_CANONICAL
1072 space_before = F_SPACE_BEFORE 1444 space_before = F_SPACE_BEFORE
1073 space_after = F_SPACE_AFTER 1445 space_after = F_SPACE_AFTER
1074 pretty = F_PRETTY 1446 pretty = F_PRETTY
1075 allow_nonref = F_ALLOW_NONREF 1447 allow_nonref = F_ALLOW_NONREF
1076 shrink = F_SHRINK 1448 shrink = F_SHRINK
1449 allow_blessed = F_ALLOW_BLESSED
1450 convert_blessed = F_CONV_BLESSED
1451 relaxed = F_RELAXED
1077 CODE: 1452 PPCODE:
1078{ 1453{
1079 UV *uv = SvJSON (self);
1080 if (enable) 1454 if (enable)
1081 *uv |= ix; 1455 self->flags |= ix;
1082 else 1456 else
1083 *uv &= ~ix; 1457 self->flags &= ~ix;
1084 1458
1085 RETVAL = newSVsv (self); 1459 XPUSHs (ST (0));
1086} 1460}
1087 OUTPUT:
1088 RETVAL
1089 1461
1090SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1462void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1091 CODE: 1463 PPCODE:
1092{ 1464{
1093 UV *uv = SvJSON (self);
1094 UV log2 = 0; 1465 UV log2 = 0;
1095 1466
1096 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1467 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1097 1468
1098 while ((1UL << log2) < max_depth) 1469 while ((1UL << log2) < max_depth)
1099 ++log2; 1470 ++log2;
1100 1471
1101 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1472 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1102 1473
1103 RETVAL = newSVsv (self); 1474 XPUSHs (ST (0));
1104} 1475}
1105 OUTPUT:
1106 RETVAL
1107 1476
1108void encode (SV *self, SV *scalar) 1477void max_size (JSON *self, UV max_size = 0)
1109 PPCODE: 1478 PPCODE:
1110 XPUSHs (encode_json (scalar, *SvJSON (self))); 1479{
1480 UV log2 = 0;
1111 1481
1112void decode (SV *self, SV *jsonstr) 1482 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1483 if (max_size == 1) max_size = 2;
1484
1485 while ((1UL << log2) < max_size)
1486 ++log2;
1487
1488 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1489
1490 XPUSHs (ST (0));
1491}
1492
1493void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1113 PPCODE: 1494 PPCODE:
1495{
1496 SvREFCNT_dec (self->cb_object);
1497 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1498
1499 XPUSHs (ST (0));
1500}
1501
1502void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1503 PPCODE:
1504{
1505 if (!self->cb_sk_object)
1506 self->cb_sk_object = newHV ();
1507
1508 if (SvOK (cb))
1509 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1510 else
1511 {
1512 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1513
1514 if (!HvKEYS (self->cb_sk_object))
1515 {
1516 SvREFCNT_dec (self->cb_sk_object);
1517 self->cb_sk_object = 0;
1518 }
1519 }
1520
1521 XPUSHs (ST (0));
1522}
1523
1524void encode (JSON *self, SV *scalar)
1525 PPCODE:
1526 XPUSHs (encode_json (scalar, self));
1527
1528void decode (JSON *self, SV *jsonstr)
1529 PPCODE:
1114 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1530 XPUSHs (decode_json (jsonstr, self, 0));
1531
1532void decode_prefix (JSON *self, SV *jsonstr)
1533 PPCODE:
1534{
1535 UV offset;
1536 EXTEND (SP, 2);
1537 PUSHs (decode_json (jsonstr, self, &offset));
1538 PUSHs (sv_2mortal (newSVuv (offset)));
1539}
1540
1541void DESTROY (JSON *self)
1542 CODE:
1543 SvREFCNT_dec (self->cb_sk_object);
1544 SvREFCNT_dec (self->cb_object);
1115 1545
1116PROTOTYPES: ENABLE 1546PROTOTYPES: ENABLE
1117 1547
1118void to_json (SV *scalar) 1548void to_json (SV *scalar)
1119 ALIAS:
1120 objToJson = 0
1121 PPCODE: 1549 PPCODE:
1550{
1551 JSON json = { F_DEFAULT | F_UTF8 };
1122 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1552 XPUSHs (encode_json (scalar, &json));
1553}
1123 1554
1124void from_json (SV *jsonstr) 1555void from_json (SV *jsonstr)
1125 ALIAS:
1126 jsonToObj = 0
1127 PPCODE: 1556 PPCODE:
1557{
1558 JSON json = { F_DEFAULT | F_UTF8 };
1128 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1559 XPUSHs (decode_json (jsonstr, &json, 0));
1560}
1129 1561

Diff Legend

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