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 1.27 by root, Mon Apr 9 05:09:57 2007 UTC vs.
Revision 1.70 by root, Wed Mar 19 00:44:54 2008 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 UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size 49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38 50
39#define SB do { 51#define SB do {
40#define SE } while (0) 52#define SE } while (0)
41 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
42static 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;
43 81
44///////////////////////////////////////////////////////////////////////////// 82/////////////////////////////////////////////////////////////////////////////
45// utility functions 83// utility functions
46 84
47static UV * 85inline 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) 86shrink (SV *sv)
58{ 87{
59 sv_utf8_downgrade (sv, 1); 88 sv_utf8_downgrade (sv, 1);
60 if (SvLEN (sv) > SvCUR (sv) + 1) 89 if (SvLEN (sv) > SvCUR (sv) + 1)
61 { 90 {
… …
70// 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
71// case of an error. 100// case of an error.
72// we special-case "safe" characters from U+80 .. U+7FF, 101// we special-case "safe" characters from U+80 .. U+7FF,
73// but use the very good perl function to parse anything else. 102// but use the very good perl function to parse anything else.
74// note that we never call this function for a ascii codepoints 103// note that we never call this function for a ascii codepoints
75static UV 104inline UV
76decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
77{ 106{
78 if (s[0] > 0xdf || s[0] < 0xc2) 107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
79 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
80 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
81 { 110 {
82 *clen = 2; 111 *clen = 2;
83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
… …
96typedef struct 125typedef struct
97{ 126{
98 char *cur; // SvPVX (sv) + current output position 127 char *cur; // SvPVX (sv) + current output position
99 char *end; // SvEND (sv) 128 char *end; // SvEND (sv)
100 SV *sv; // result scalar 129 SV *sv; // result scalar
101 U32 flags; // F_* 130 JSON json;
102 U32 indent; // indentation level 131 U32 indent; // indentation level
103 U32 maxdepth; // max. indentation/recursion level 132 U32 maxdepth; // max. indentation/recursion level
104} enc_t; 133} enc_t;
105 134
106static void 135inline void
107need (enc_t *enc, STRLEN len) 136need (enc_t *enc, STRLEN len)
108{ 137{
109 if (enc->cur + len >= enc->end) 138 if (expect_false (enc->cur + len >= enc->end))
110 { 139 {
111 STRLEN cur = enc->cur - SvPVX (enc->sv); 140 STRLEN cur = enc->cur - SvPVX (enc->sv);
112 SvGROW (enc->sv, cur + len + 1); 141 SvGROW (enc->sv, cur + len + 1);
113 enc->cur = SvPVX (enc->sv) + cur; 142 enc->cur = SvPVX (enc->sv) + cur;
114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
115 } 144 }
116} 145}
117 146
118static void 147inline void
119encode_ch (enc_t *enc, char ch) 148encode_ch (enc_t *enc, char ch)
120{ 149{
121 need (enc, 1); 150 need (enc, 1);
122 *enc->cur++ = ch; 151 *enc->cur++ = ch;
123} 152}
… …
131 160
132 while (str < end) 161 while (str < end)
133 { 162 {
134 unsigned char ch = *(unsigned char *)str; 163 unsigned char ch = *(unsigned char *)str;
135 164
136 if (ch >= 0x20 && ch < 0x80) // most common case 165 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
137 { 166 {
138 if (ch == '"') // but with slow exceptions 167 if (expect_false (ch == '"')) // but with slow exceptions
139 { 168 {
140 need (enc, len += 1); 169 need (enc, len += 1);
141 *enc->cur++ = '\\'; 170 *enc->cur++ = '\\';
142 *enc->cur++ = '"'; 171 *enc->cur++ = '"';
143 } 172 }
144 else if (ch == '\\') 173 else if (expect_false (ch == '\\'))
145 { 174 {
146 need (enc, len += 1); 175 need (enc, len += 1);
147 *enc->cur++ = '\\'; 176 *enc->cur++ = '\\';
148 *enc->cur++ = '\\'; 177 *enc->cur++ = '\\';
149 } 178 }
… …
167 STRLEN clen; 196 STRLEN clen;
168 UV uch; 197 UV uch;
169 198
170 if (is_utf8) 199 if (is_utf8)
171 { 200 {
172 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173 uch = decode_utf8 (str, end - str, &clen); 201 uch = decode_utf8 (str, end - str, &clen);
174 if (clen == (STRLEN)-1) 202 if (clen == (STRLEN)-1)
175 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);
176 } 204 }
177 else 205 else
… …
181 } 209 }
182 210
183 if (uch > 0x10FFFFUL) 211 if (uch > 0x10FFFFUL)
184 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);
185 213
186 if (uch < 0x80 || enc->flags & F_ASCII) 214 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
187 { 215 {
188 if (uch > 0xFFFFUL) 216 if (uch > 0xFFFFUL)
189 { 217 {
190 need (enc, len += 11); 218 need (enc, len += 11);
191 sprintf (enc->cur, "\\u%04x\\u%04x", 219 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
205 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 233 *enc->cur++ = hexdigit [(uch >> 0) & 15];
206 } 234 }
207 235
208 str += clen; 236 str += clen;
209 } 237 }
238 else if (enc->json.flags & F_LATIN1)
239 {
240 *enc->cur++ = uch;
241 str += clen;
242 }
210 else if (is_utf8) 243 else if (is_utf8)
211 { 244 {
212 need (enc, len += clen); 245 need (enc, len += clen);
213 do 246 do
214 { 247 {
… …
216 } 249 }
217 while (--clen); 250 while (--clen);
218 } 251 }
219 else 252 else
220 { 253 {
221 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 254 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
222 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 255 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
223 ++str; 256 ++str;
224 } 257 }
225 } 258 }
226 } 259 }
… …
228 261
229 --len; 262 --len;
230 } 263 }
231} 264}
232 265
233static void 266inline void
234encode_indent (enc_t *enc) 267encode_indent (enc_t *enc)
235{ 268{
236 if (enc->flags & F_INDENT) 269 if (enc->json.flags & F_INDENT)
237 { 270 {
238 int spaces = enc->indent * INDENT_STEP; 271 int spaces = enc->indent * INDENT_STEP;
239 272
240 need (enc, spaces); 273 need (enc, spaces);
241 memset (enc->cur, ' ', spaces); 274 memset (enc->cur, ' ', spaces);
242 enc->cur += spaces; 275 enc->cur += spaces;
243 } 276 }
244} 277}
245 278
246static void 279inline void
247encode_space (enc_t *enc) 280encode_space (enc_t *enc)
248{ 281{
249 need (enc, 1); 282 need (enc, 1);
250 encode_ch (enc, ' '); 283 encode_ch (enc, ' ');
251} 284}
252 285
253static void 286inline void
254encode_nl (enc_t *enc) 287encode_nl (enc_t *enc)
255{ 288{
256 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
257 { 290 {
258 need (enc, 1); 291 need (enc, 1);
259 encode_ch (enc, '\n'); 292 encode_ch (enc, '\n');
260 } 293 }
261} 294}
262 295
263static void 296inline void
264encode_comma (enc_t *enc) 297encode_comma (enc_t *enc)
265{ 298{
266 encode_ch (enc, ','); 299 encode_ch (enc, ',');
267 300
268 if (enc->flags & F_INDENT) 301 if (enc->json.flags & F_INDENT)
269 encode_nl (enc); 302 encode_nl (enc);
270 else if (enc->flags & F_SPACE_AFTER) 303 else if (enc->json.flags & F_SPACE_AFTER)
271 encode_space (enc); 304 encode_space (enc);
272} 305}
273 306
274static void encode_sv (enc_t *enc, SV *sv); 307static void encode_sv (enc_t *enc, SV *sv);
275 308
… …
279 int i, len = av_len (av); 312 int i, len = av_len (av);
280 313
281 if (enc->indent >= enc->maxdepth) 314 if (enc->indent >= enc->maxdepth)
282 croak ("data structure too deep (hit recursion limit)"); 315 croak ("data structure too deep (hit recursion limit)");
283 316
284 encode_ch (enc, '['); encode_nl (enc); 317 encode_ch (enc, '[');
285 ++enc->indent; 318
319 if (len >= 0)
320 {
321 encode_nl (enc); ++enc->indent;
286 322
287 for (i = 0; i <= len; ++i) 323 for (i = 0; i <= len; ++i)
288 { 324 {
325 SV **svp = av_fetch (av, i, 0);
326
289 encode_indent (enc); 327 encode_indent (enc);
290 encode_sv (enc, *av_fetch (av, i, 0));
291 328
329 if (svp)
330 encode_sv (enc, *svp);
331 else
332 encode_str (enc, "null", 4, 0);
333
292 if (i < len) 334 if (i < len)
293 encode_comma (enc); 335 encode_comma (enc);
294 } 336 }
295 337
338 encode_nl (enc); --enc->indent; encode_indent (enc);
339 }
340
296 encode_nl (enc); 341 encode_ch (enc, ']');
297
298 --enc->indent;
299 encode_indent (enc); encode_ch (enc, ']');
300} 342}
301 343
302static void 344static void
303encode_he (enc_t *enc, HE *he) 345encode_hk (enc_t *enc, HE *he)
304{ 346{
305 encode_ch (enc, '"'); 347 encode_ch (enc, '"');
306 348
307 if (HeKLEN (he) == HEf_SVKEY) 349 if (HeKLEN (he) == HEf_SVKEY)
308 { 350 {
… …
318 else 360 else
319 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 361 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
320 362
321 encode_ch (enc, '"'); 363 encode_ch (enc, '"');
322 364
323 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 365 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
324 encode_ch (enc, ':'); 366 encode_ch (enc, ':');
325 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 367 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
326 encode_sv (enc, HeVAL (he));
327} 368}
328 369
329// compare hash entries, used when all keys are bytestrings 370// compare hash entries, used when all keys are bytestrings
330static int 371static int
331he_cmp_fast (const void *a_, const void *b_) 372he_cmp_fast (const void *a_, const void *b_)
… …
336 HE *b = *(HE **)b_; 377 HE *b = *(HE **)b_;
337 378
338 STRLEN la = HeKLEN (a); 379 STRLEN la = HeKLEN (a);
339 STRLEN lb = HeKLEN (b); 380 STRLEN lb = HeKLEN (b);
340 381
341 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 382 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
342 cmp = la - lb; 383 cmp = lb - la;
343 384
344 return cmp; 385 return cmp;
345} 386}
346 387
347// compare hash entries, used when some keys are sv's or utf-x 388// compare hash entries, used when some keys are sv's or utf-x
348static int 389static int
349he_cmp_slow (const void *a, const void *b) 390he_cmp_slow (const void *a, const void *b)
350{ 391{
351 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 392 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
352} 393}
353 394
354static void 395static void
355encode_hv (enc_t *enc, HV *hv) 396encode_hv (enc_t *enc, HV *hv)
356{ 397{
398 HE *he;
357 int count, i; 399 int count;
358 400
359 if (enc->indent >= enc->maxdepth) 401 if (enc->indent >= enc->maxdepth)
360 croak ("data structure too deep (hit recursion limit)"); 402 croak ("data structure too deep (hit recursion limit)");
361 403
362 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 404 encode_ch (enc, '{');
363 405
364 if ((count = hv_iterinit (hv)))
365 {
366 // for canonical output we have to sort by keys first 406 // for canonical output we have to sort by keys first
367 // actually, this is mostly due to the stupid so-called 407 // actually, this is mostly due to the stupid so-called
368 // security workaround added somewhere in 5.8.x. 408 // security workaround added somewhere in 5.8.x.
369 // that randomises hash orderings 409 // that randomises hash orderings
370 if (enc->flags & F_CANONICAL) 410 if (enc->json.flags & F_CANONICAL)
411 {
412 int count = hv_iterinit (hv);
413
414 if (SvMAGICAL (hv))
371 { 415 {
416 // need to count by iterating. could improve by dynamically building the vector below
417 // but I don't care for the speed of this special case.
418 // note also that we will run into undefined behaviour when the two iterations
419 // do not result in the same count, something I might care for in some later release.
420
421 count = 0;
422 while (hv_iternext (hv))
423 ++count;
424
425 hv_iterinit (hv);
426 }
427
428 if (count)
429 {
430 int i, fast = 1;
431#if defined(__BORLANDC__) || defined(_MSC_VER)
432 HE **hes = _alloca (count * sizeof (HE));
433#else
372 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 434 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
373 int fast = 1; 435#endif
374 436
375 i = 0; 437 i = 0;
376 while ((he = hv_iternext (hv))) 438 while ((he = hv_iternext (hv)))
377 { 439 {
378 hes [i++] = he; 440 hes [i++] = he;
… …
400 462
401 FREETMPS; 463 FREETMPS;
402 LEAVE; 464 LEAVE;
403 } 465 }
404 466
405 for (i = 0; i < count; ++i) 467 encode_nl (enc); ++enc->indent;
468
469 while (count--)
406 { 470 {
407 encode_indent (enc); 471 encode_indent (enc);
472 he = hes [count];
408 encode_he (enc, hes [i]); 473 encode_hk (enc, he);
474 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
409 475
410 if (i < count - 1) 476 if (count)
411 encode_comma (enc); 477 encode_comma (enc);
412 } 478 }
413 479
480 encode_nl (enc); --enc->indent; encode_indent (enc);
481 }
482 }
483 else
484 {
485 if (hv_iterinit (hv) || SvMAGICAL (hv))
486 if ((he = hv_iternext (hv)))
487 {
488 encode_nl (enc); ++enc->indent;
489
490 for (;;)
491 {
414 encode_nl (enc); 492 encode_indent (enc);
493 encode_hk (enc, he);
494 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
495
496 if (!(he = hv_iternext (hv)))
497 break;
498
499 encode_comma (enc);
500 }
501
502 encode_nl (enc); --enc->indent; encode_indent (enc);
503 }
504 }
505
506 encode_ch (enc, '}');
507}
508
509// encode objects, arrays and special \0=false and \1=true values.
510static void
511encode_rv (enc_t *enc, SV *sv)
512{
513 svtype svt;
514
515 SvGETMAGIC (sv);
516 svt = SvTYPE (sv);
517
518 if (expect_false (SvOBJECT (sv)))
519 {
520 HV *stash = !JSON_SLOW || json_boolean_stash
521 ? json_boolean_stash
522 : gv_stashpv ("JSON::XS::Boolean", 1);
523
524 if (SvSTASH (sv) == stash)
525 {
526 if (SvIV (sv))
527 encode_str (enc, "true", 4, 0);
528 else
529 encode_str (enc, "false", 5, 0);
415 } 530 }
416 else 531 else
417 { 532 {
418 HE *he = hv_iternext (hv); 533#if 0
419 534 if (0 && sv_derived_from (rv, "JSON::Literal"))
420 for (;;)
421 { 535 {
422 encode_indent (enc); 536 // not yet
423 encode_he (enc, he);
424
425 if (!(he = hv_iternext (hv)))
426 break;
427
428 encode_comma (enc);
429 } 537 }
538#endif
539 if (enc->json.flags & F_CONV_BLESSED)
540 {
541 // we re-bless the reference to get overload and other niceties right
542 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
430 543
544 if (to_json)
545 {
546 dSP;
547
548 ENTER; SAVETMPS; PUSHMARK (SP);
549 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
550
551 // calling with G_SCALAR ensures that we always get a 1 return value
552 PUTBACK;
553 call_sv ((SV *)GvCV (to_json), G_SCALAR);
554 SPAGAIN;
555
556 // catch this surprisingly common error
557 if (SvROK (TOPs) && SvRV (TOPs) == sv)
558 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
559
560 sv = POPs;
561 PUTBACK;
562
431 encode_nl (enc); 563 encode_sv (enc, sv);
564
565 FREETMPS; LEAVE;
566 }
567 else if (enc->json.flags & F_ALLOW_BLESSED)
568 encode_str (enc, "null", 4, 0);
569 else
570 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
571 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
572 }
573 else if (enc->json.flags & F_ALLOW_BLESSED)
574 encode_str (enc, "null", 4, 0);
575 else
576 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
577 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
432 } 578 }
433 } 579 }
434
435 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
436}
437
438// encode objects, arrays and special \0=false and \1=true values.
439static void
440encode_rv (enc_t *enc, SV *sv)
441{
442 svtype svt;
443
444 SvGETMAGIC (sv);
445 svt = SvTYPE (sv);
446
447 if (svt == SVt_PVHV) 580 else if (svt == SVt_PVHV)
448 encode_hv (enc, (HV *)sv); 581 encode_hv (enc, (HV *)sv);
449 else if (svt == SVt_PVAV) 582 else if (svt == SVt_PVAV)
450 encode_av (enc, (AV *)sv); 583 encode_av (enc, (AV *)sv);
451 else if (svt < SVt_PVAV) 584 else if (svt < SVt_PVAV)
452 { 585 {
453 if (SvNIOK (sv) && SvIV (sv) == 0) 586 STRLEN len = 0;
587 char *pv = svt ? SvPV (sv, len) : 0;
588
589 if (len == 1 && *pv == '1')
590 encode_str (enc, "true", 4, 0);
591 else if (len == 1 && *pv == '0')
454 encode_str (enc, "false", 5, 0); 592 encode_str (enc, "false", 5, 0);
455 else if (SvNIOK (sv) && SvIV (sv) == 1)
456 encode_str (enc, "true", 4, 0);
457 else 593 else
458 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 594 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
459 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 595 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
460 } 596 }
461 else 597 else
… …
476 encode_str (enc, str, len, SvUTF8 (sv)); 612 encode_str (enc, str, len, SvUTF8 (sv));
477 encode_ch (enc, '"'); 613 encode_ch (enc, '"');
478 } 614 }
479 else if (SvNOKp (sv)) 615 else if (SvNOKp (sv))
480 { 616 {
617 // trust that perl will do the right thing w.r.t. JSON syntax.
481 need (enc, NV_DIG + 32); 618 need (enc, NV_DIG + 32);
482 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 619 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
483 enc->cur += strlen (enc->cur); 620 enc->cur += strlen (enc->cur);
484 } 621 }
485 else if (SvIOKp (sv)) 622 else if (SvIOKp (sv))
486 { 623 {
487 need (enc, 64); 624 // we assume we can always read an IV as a UV
625 if (SvUV (sv) & ~(UV)0x7fff)
626 {
627 // large integer, use the (rather slow) snprintf way.
628 need (enc, sizeof (UV) * 3);
488 enc->cur += 629 enc->cur +=
489 SvIsUV(sv) 630 SvIsUV(sv)
490 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 631 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
491 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 632 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
633 }
634 else
635 {
636 // optimise the "small number case"
637 // code will likely be branchless and use only a single multiplication
638 I32 i = SvIV (sv);
639 U32 u;
640 char digit, nz = 0;
641
642 need (enc, 6);
643
644 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
645 u = i < 0 ? -i : i;
646
647 // convert to 4.28 fixed-point representation
648 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
649
650 // now output digit by digit, each time masking out the integer part
651 // and multiplying by 5 while moving the decimal point one to the right,
652 // resulting in a net multiplication by 10.
653 // we always write the digit to memory but conditionally increment
654 // the pointer, to ease the usage of conditional move instructions.
655 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
656 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
657 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
658 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
659 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
660 }
492 } 661 }
493 else if (SvROK (sv)) 662 else if (SvROK (sv))
494 encode_rv (enc, SvRV (sv)); 663 encode_rv (enc, SvRV (sv));
495 else if (!SvOK (sv)) 664 else if (!SvOK (sv))
496 encode_str (enc, "null", 4, 0); 665 encode_str (enc, "null", 4, 0);
… …
498 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 667 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
499 SvPV_nolen (sv), SvFLAGS (sv)); 668 SvPV_nolen (sv), SvFLAGS (sv));
500} 669}
501 670
502static SV * 671static SV *
503encode_json (SV *scalar, U32 flags) 672encode_json (SV *scalar, JSON *json)
504{ 673{
505 enc_t enc; 674 enc_t enc;
506 675
507 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 676 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
508 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 677 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
509 678
510 enc.flags = flags; 679 enc.json = *json;
511 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 680 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
512 enc.cur = SvPVX (enc.sv); 681 enc.cur = SvPVX (enc.sv);
513 enc.end = SvEND (enc.sv); 682 enc.end = SvEND (enc.sv);
514 enc.indent = 0; 683 enc.indent = 0;
515 enc.maxdepth = DEC_DEPTH (flags); 684 enc.maxdepth = DEC_DEPTH (enc.json.flags);
516 685
517 SvPOK_only (enc.sv); 686 SvPOK_only (enc.sv);
518 encode_sv (&enc, scalar); 687 encode_sv (&enc, scalar);
519 688
520 if (!(flags & (F_ASCII | F_UTF8)))
521 SvUTF8_on (enc.sv);
522
523 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 689 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
524 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 690 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
525 691
692 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
693 SvUTF8_on (enc.sv);
694
526 if (enc.flags & F_SHRINK) 695 if (enc.json.flags & F_SHRINK)
527 shrink (enc.sv); 696 shrink (enc.sv);
528 697
529 return enc.sv; 698 return enc.sv;
530} 699}
531 700
… …
536typedef struct 705typedef struct
537{ 706{
538 char *cur; // current parser pointer 707 char *cur; // current parser pointer
539 char *end; // end of input string 708 char *end; // end of input string
540 const char *err; // parse error, if != 0 709 const char *err; // parse error, if != 0
541 U32 flags; // F_* 710 JSON json;
542 U32 depth; // recursion depth 711 U32 depth; // recursion depth
543 U32 maxdepth; // recursion depth limit 712 U32 maxdepth; // recursion depth limit
544} dec_t; 713} dec_t;
545 714
546static void 715inline void
716decode_comment (dec_t *dec)
717{
718 // only '#'-style comments allowed a.t.m.
719
720 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
721 ++dec->cur;
722}
723
724inline void
547decode_ws (dec_t *dec) 725decode_ws (dec_t *dec)
548{ 726{
549 for (;;) 727 for (;;)
550 { 728 {
551 char ch = *dec->cur; 729 char ch = *dec->cur;
552 730
553 if (ch > 0x20 731 if (ch > 0x20)
732 {
733 if (expect_false (ch == '#'))
734 {
735 if (dec->json.flags & F_RELAXED)
736 decode_comment (dec);
737 else
738 break;
739 }
740 else
741 break;
742 }
554 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 743 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
555 break; 744 break; // parse error, but let higher level handle it, gives better error messages
556 745
557 ++dec->cur; 746 ++dec->cur;
558 } 747 }
559} 748}
560 749
… …
577decode_4hex (dec_t *dec) 766decode_4hex (dec_t *dec)
578{ 767{
579 signed char d1, d2, d3, d4; 768 signed char d1, d2, d3, d4;
580 unsigned char *cur = (unsigned char *)dec->cur; 769 unsigned char *cur = (unsigned char *)dec->cur;
581 770
582 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 771 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
583 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 772 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
584 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 773 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
585 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 774 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
586 775
587 dec->cur += 4; 776 dec->cur += 4;
588 777
589 return ((UV)d1) << 12 778 return ((UV)d1) << 12
590 | ((UV)d2) << 8 779 | ((UV)d2) << 8
… …
598static SV * 787static SV *
599decode_str (dec_t *dec) 788decode_str (dec_t *dec)
600{ 789{
601 SV *sv = 0; 790 SV *sv = 0;
602 int utf8 = 0; 791 int utf8 = 0;
792 char *dec_cur = dec->cur;
603 793
604 do 794 do
605 { 795 {
606 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 796 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
607 char *cur = buf; 797 char *cur = buf;
608 798
609 do 799 do
610 { 800 {
611 unsigned char ch = *(unsigned char *)dec->cur++; 801 unsigned char ch = *(unsigned char *)dec_cur++;
612 802
613 if (ch == '"') 803 if (expect_false (ch == '"'))
614 { 804 {
615 --dec->cur; 805 --dec_cur;
616 break; 806 break;
617 } 807 }
618 else if (ch == '\\') 808 else if (expect_false (ch == '\\'))
619 { 809 {
620 switch (*dec->cur) 810 switch (*dec_cur)
621 { 811 {
622 case '\\': 812 case '\\':
623 case '/': 813 case '/':
624 case '"': *cur++ = *dec->cur++; break; 814 case '"': *cur++ = *dec_cur++; break;
625 815
626 case 'b': ++dec->cur; *cur++ = '\010'; break; 816 case 'b': ++dec_cur; *cur++ = '\010'; break;
627 case 't': ++dec->cur; *cur++ = '\011'; break; 817 case 't': ++dec_cur; *cur++ = '\011'; break;
628 case 'n': ++dec->cur; *cur++ = '\012'; break; 818 case 'n': ++dec_cur; *cur++ = '\012'; break;
629 case 'f': ++dec->cur; *cur++ = '\014'; break; 819 case 'f': ++dec_cur; *cur++ = '\014'; break;
630 case 'r': ++dec->cur; *cur++ = '\015'; break; 820 case 'r': ++dec_cur; *cur++ = '\015'; break;
631 821
632 case 'u': 822 case 'u':
633 { 823 {
634 UV lo, hi; 824 UV lo, hi;
635 ++dec->cur; 825 ++dec_cur;
636 826
827 dec->cur = dec_cur;
637 hi = decode_4hex (dec); 828 hi = decode_4hex (dec);
829 dec_cur = dec->cur;
638 if (hi == (UV)-1) 830 if (hi == (UV)-1)
639 goto fail; 831 goto fail;
640 832
641 // possibly a surrogate pair 833 // possibly a surrogate pair
642 if (hi >= 0xd800) 834 if (hi >= 0xd800)
643 if (hi < 0xdc00) 835 if (hi < 0xdc00)
644 { 836 {
645 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 837 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
646 ERR ("missing low surrogate character in surrogate pair"); 838 ERR ("missing low surrogate character in surrogate pair");
647 839
648 dec->cur += 2; 840 dec_cur += 2;
649 841
842 dec->cur = dec_cur;
650 lo = decode_4hex (dec); 843 lo = decode_4hex (dec);
844 dec_cur = dec->cur;
651 if (lo == (UV)-1) 845 if (lo == (UV)-1)
652 goto fail; 846 goto fail;
653 847
654 if (lo < 0xdc00 || lo >= 0xe000) 848 if (lo < 0xdc00 || lo >= 0xe000)
655 ERR ("surrogate pair expected"); 849 ERR ("surrogate pair expected");
… …
669 *cur++ = hi; 863 *cur++ = hi;
670 } 864 }
671 break; 865 break;
672 866
673 default: 867 default:
674 --dec->cur; 868 --dec_cur;
675 ERR ("illegal backslash escape sequence in string"); 869 ERR ("illegal backslash escape sequence in string");
676 } 870 }
677 } 871 }
678 else if (ch >= 0x20 && ch <= 0x7f) 872 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
679 *cur++ = ch; 873 *cur++ = ch;
680 else if (ch >= 0x80) 874 else if (ch >= 0x80)
681 { 875 {
682 STRLEN clen; 876 STRLEN clen;
683 UV uch; 877 UV uch;
684 878
685 --dec->cur; 879 --dec_cur;
686 880
687 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 881 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
688 if (clen == (STRLEN)-1) 882 if (clen == (STRLEN)-1)
689 ERR ("malformed UTF-8 character in JSON string"); 883 ERR ("malformed UTF-8 character in JSON string");
690 884
691 do 885 do
692 *cur++ = *dec->cur++; 886 *cur++ = *dec_cur++;
693 while (--clen); 887 while (--clen);
694 888
695 utf8 = 1; 889 utf8 = 1;
696 } 890 }
697 else 891 else
698 { 892 {
699 --dec->cur; 893 --dec_cur;
700 894
701 if (!ch) 895 if (!ch)
702 ERR ("unexpected end of string while parsing JSON string"); 896 ERR ("unexpected end of string while parsing JSON string");
703 else 897 else
704 ERR ("invalid character encountered while parsing JSON string"); 898 ERR ("invalid character encountered while parsing JSON string");
… …
717 } 911 }
718 else 912 else
719 sv = newSVpvn (buf, len); 913 sv = newSVpvn (buf, len);
720 } 914 }
721 } 915 }
722 while (*dec->cur != '"'); 916 while (*dec_cur != '"');
723 917
724 ++dec->cur; 918 ++dec_cur;
725 919
726 if (sv) 920 if (sv)
727 { 921 {
728 SvPOK_only (sv); 922 SvPOK_only (sv);
729 *SvEND (sv) = 0; 923 *SvEND (sv) = 0;
… …
732 SvUTF8_on (sv); 926 SvUTF8_on (sv);
733 } 927 }
734 else 928 else
735 sv = newSVpvn ("", 0); 929 sv = newSVpvn ("", 0);
736 930
931 dec->cur = dec_cur;
737 return sv; 932 return sv;
738 933
739fail: 934fail:
935 dec->cur = dec_cur;
740 return 0; 936 return 0;
741} 937}
742 938
743static SV * 939static SV *
744decode_num (dec_t *dec) 940decode_num (dec_t *dec)
… …
802 is_nv = 1; 998 is_nv = 1;
803 } 999 }
804 1000
805 if (!is_nv) 1001 if (!is_nv)
806 { 1002 {
807 UV uv; 1003 int len = dec->cur - start;
808 int numtype = grok_number (start, dec->cur - start, &uv); 1004
809 if (numtype & IS_NUMBER_IN_UV) 1005 // special case the rather common 1..5-digit-int case
810 if (numtype & IS_NUMBER_NEG) 1006 if (*start == '-')
1007 switch (len)
811 { 1008 {
812 if (uv < (UV)IV_MIN) 1009 case 2: return newSViv (-( start [1] - '0' * 1));
813 return newSViv (-(IV)uv); 1010 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1011 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1012 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1013 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
814 } 1014 }
1015 else
1016 switch (len)
1017 {
1018 case 1: return newSViv ( start [0] - '0' * 1);
1019 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1020 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1021 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1022 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1023 }
1024
1025 {
1026 UV uv;
1027 int numtype = grok_number (start, len, &uv);
1028 if (numtype & IS_NUMBER_IN_UV)
1029 if (numtype & IS_NUMBER_NEG)
1030 {
1031 if (uv < (UV)IV_MIN)
1032 return newSViv (-(IV)uv);
1033 }
815 else 1034 else
816 return newSVuv (uv); 1035 return newSVuv (uv);
817 } 1036 }
818 1037
1038 len -= *start == '-' ? 1 : 0;
1039
1040 // does not fit into IV or UV, try NV
1041 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1042 #if defined (LDBL_DIG)
1043 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1044 #endif
1045 )
1046 // fits into NV without loss of precision
1047 return newSVnv (Atof (start));
1048
1049 // everything else fails, convert it to a string
1050 return newSVpvn (start, dec->cur - start);
1051 }
1052
1053 // loss of precision here
819 return newSVnv (Atof (start)); 1054 return newSVnv (Atof (start));
820 1055
821fail: 1056fail:
822 return 0; 1057 return 0;
823} 1058}
… …
853 1088
854 if (*dec->cur != ',') 1089 if (*dec->cur != ',')
855 ERR (", or ] expected while parsing array"); 1090 ERR (", or ] expected while parsing array");
856 1091
857 ++dec->cur; 1092 ++dec->cur;
1093
1094 decode_ws (dec);
1095
1096 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1097 {
1098 ++dec->cur;
1099 break;
1100 }
858 } 1101 }
859 1102
860 DEC_DEC_DEPTH; 1103 DEC_DEC_DEPTH;
861 return newRV_noinc ((SV *)av); 1104 return newRV_noinc ((SV *)av);
862 1105
… …
867} 1110}
868 1111
869static SV * 1112static SV *
870decode_hv (dec_t *dec) 1113decode_hv (dec_t *dec)
871{ 1114{
1115 SV *sv;
872 HV *hv = newHV (); 1116 HV *hv = newHV ();
873 1117
874 DEC_INC_DEPTH; 1118 DEC_INC_DEPTH;
875 decode_ws (dec); 1119 decode_ws (dec);
876 1120
877 if (*dec->cur == '}') 1121 if (*dec->cur == '}')
878 ++dec->cur; 1122 ++dec->cur;
879 else 1123 else
880 for (;;) 1124 for (;;)
881 { 1125 {
1126 EXPECT_CH ('"');
1127
1128 // heuristic: assume that
1129 // a) decode_str + hv_store_ent are abysmally slow.
1130 // b) most hash keys are short, simple ascii text.
1131 // => try to "fast-match" such strings to avoid
1132 // the overhead of decode_str + hv_store_ent.
1133 {
882 SV *key, *value; 1134 SV *value;
1135 char *p = dec->cur;
1136 char *e = p + 24; // only try up to 24 bytes
883 1137
884 decode_ws (dec); EXPECT_CH ('"'); 1138 for (;;)
885
886 key = decode_str (dec);
887 if (!key)
888 goto fail;
889
890 decode_ws (dec); EXPECT_CH (':');
891
892 value = decode_sv (dec);
893 if (!value)
894 { 1139 {
1140 // the >= 0x80 is true on most architectures
1141 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1142 {
1143 // slow path, back up and use decode_str
1144 SV *key = decode_str (dec);
1145 if (!key)
1146 goto fail;
1147
1148 decode_ws (dec); EXPECT_CH (':');
1149
1150 decode_ws (dec);
1151 value = decode_sv (dec);
1152 if (!value)
1153 {
1154 SvREFCNT_dec (key);
1155 goto fail;
1156 }
1157
1158 hv_store_ent (hv, key, value, 0);
895 SvREFCNT_dec (key); 1159 SvREFCNT_dec (key);
1160
1161 break;
1162 }
1163 else if (*p == '"')
1164 {
1165 // fast path, got a simple key
1166 char *key = dec->cur;
1167 int len = p - key;
1168 dec->cur = p + 1;
1169
1170 decode_ws (dec); EXPECT_CH (':');
1171
1172 decode_ws (dec);
1173 value = decode_sv (dec);
1174 if (!value)
896 goto fail; 1175 goto fail;
1176
1177 hv_store (hv, key, len, value, 0);
1178
1179 break;
1180 }
1181
1182 ++p;
897 } 1183 }
898 1184 }
899 hv_store_ent (hv, key, value, 0);
900 SvREFCNT_dec (key);
901 1185
902 decode_ws (dec); 1186 decode_ws (dec);
903 1187
904 if (*dec->cur == '}') 1188 if (*dec->cur == '}')
905 { 1189 {
… …
909 1193
910 if (*dec->cur != ',') 1194 if (*dec->cur != ',')
911 ERR (", or } expected while parsing object/hash"); 1195 ERR (", or } expected while parsing object/hash");
912 1196
913 ++dec->cur; 1197 ++dec->cur;
1198
1199 decode_ws (dec);
1200
1201 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1202 {
1203 ++dec->cur;
1204 break;
1205 }
914 } 1206 }
915 1207
916 DEC_DEC_DEPTH; 1208 DEC_DEC_DEPTH;
917 return newRV_noinc ((SV *)hv); 1209 sv = newRV_noinc ((SV *)hv);
1210
1211 // check filter callbacks
1212 if (dec->json.flags & F_HOOK)
1213 {
1214 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1215 {
1216 HE *cb, *he;
1217
1218 hv_iterinit (hv);
1219 he = hv_iternext (hv);
1220 hv_iterinit (hv);
1221
1222 // the next line creates a mortal sv each time its called.
1223 // might want to optimise this for common cases.
1224 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1225
1226 if (cb)
1227 {
1228 dSP;
1229 int count;
1230
1231 ENTER; SAVETMPS; PUSHMARK (SP);
1232 XPUSHs (HeVAL (he));
1233
1234 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1235
1236 if (count == 1)
1237 {
1238 sv = newSVsv (POPs);
1239 FREETMPS; LEAVE;
1240 return sv;
1241 }
1242
1243 FREETMPS; LEAVE;
1244 }
1245 }
1246
1247 if (dec->json.cb_object)
1248 {
1249 dSP;
1250 int count;
1251
1252 ENTER; SAVETMPS; PUSHMARK (SP);
1253 XPUSHs (sv_2mortal (sv));
1254
1255 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1256
1257 if (count == 1)
1258 {
1259 sv = newSVsv (POPs);
1260 FREETMPS; LEAVE;
1261 return sv;
1262 }
1263
1264 SvREFCNT_inc (sv);
1265 FREETMPS; LEAVE;
1266 }
1267 }
1268
1269 return sv;
918 1270
919fail: 1271fail:
920 SvREFCNT_dec (hv); 1272 SvREFCNT_dec (hv);
921 DEC_DEC_DEPTH; 1273 DEC_DEC_DEPTH;
922 return 0; 1274 return 0;
923} 1275}
924 1276
925static SV * 1277static SV *
926decode_sv (dec_t *dec) 1278decode_sv (dec_t *dec)
927{ 1279{
928 decode_ws (dec); 1280 // the beauty of JSON: you need exactly one character lookahead
1281 // to parse anything.
929 switch (*dec->cur) 1282 switch (*dec->cur)
930 { 1283 {
931 case '"': ++dec->cur; return decode_str (dec); 1284 case '"': ++dec->cur; return decode_str (dec);
932 case '[': ++dec->cur; return decode_av (dec); 1285 case '[': ++dec->cur; return decode_av (dec);
933 case '{': ++dec->cur; return decode_hv (dec); 1286 case '{': ++dec->cur; return decode_hv (dec);
… …
939 1292
940 case 't': 1293 case 't':
941 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1294 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
942 { 1295 {
943 dec->cur += 4; 1296 dec->cur += 4;
944 return newSViv (1); 1297#if JSON_SLOW
1298 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1299#endif
1300 return SvREFCNT_inc (json_true);
945 } 1301 }
946 else 1302 else
947 ERR ("'true' expected"); 1303 ERR ("'true' expected");
948 1304
949 break; 1305 break;
950 1306
951 case 'f': 1307 case 'f':
952 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1308 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
953 { 1309 {
954 dec->cur += 5; 1310 dec->cur += 5;
955 return newSViv (0); 1311#if JSON_SLOW
1312 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1313#endif
1314 return SvREFCNT_inc (json_false);
956 } 1315 }
957 else 1316 else
958 ERR ("'false' expected"); 1317 ERR ("'false' expected");
959 1318
960 break; 1319 break;
… …
978fail: 1337fail:
979 return 0; 1338 return 0;
980} 1339}
981 1340
982static SV * 1341static SV *
983decode_json (SV *string, U32 flags) 1342decode_json (SV *string, JSON *json, UV *offset_return)
984{ 1343{
985 dec_t dec; 1344 dec_t dec;
1345 UV offset;
986 SV *sv; 1346 SV *sv;
987 1347
1348 SvGETMAGIC (string);
988 SvUPGRADE (string, SVt_PV); 1349 SvUPGRADE (string, SVt_PV);
989 1350
1351 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1352 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1353 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1354
990 if (flags & F_UTF8) 1355 if (json->flags & F_UTF8)
991 sv_utf8_downgrade (string, 0); 1356 sv_utf8_downgrade (string, 0);
992 else 1357 else
993 sv_utf8_upgrade (string); 1358 sv_utf8_upgrade (string);
994 1359
995 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1360 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
996 1361
997 dec.flags = flags; 1362 dec.json = *json;
998 dec.cur = SvPVX (string); 1363 dec.cur = SvPVX (string);
999 dec.end = SvEND (string); 1364 dec.end = SvEND (string);
1000 dec.err = 0; 1365 dec.err = 0;
1001 dec.depth = 0; 1366 dec.depth = 0;
1002 dec.maxdepth = DEC_DEPTH (dec.flags); 1367 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1003 1368
1369 if (dec.json.cb_object || dec.json.cb_sk_object)
1370 dec.json.flags |= F_HOOK;
1371
1004 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1372 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1373
1374 decode_ws (&dec);
1005 sv = decode_sv (&dec); 1375 sv = decode_sv (&dec);
1006 1376
1377 if (!(offset_return || !sv))
1378 {
1379 // check for trailing garbage
1380 decode_ws (&dec);
1381
1382 if (*dec.cur)
1383 {
1384 dec.err = "garbage after JSON object";
1385 SvREFCNT_dec (sv);
1386 sv = 0;
1387 }
1388 }
1389
1390 if (offset_return || !sv)
1391 {
1392 offset = dec.json.flags & F_UTF8
1393 ? dec.cur - SvPVX (string)
1394 : utf8_distance (dec.cur, SvPVX (string));
1395
1396 if (offset_return)
1397 *offset_return = offset;
1398 }
1399
1007 if (!sv) 1400 if (!sv)
1008 { 1401 {
1009 IV offset = dec.flags & F_UTF8
1010 ? dec.cur - SvPVX (string)
1011 : utf8_distance (dec.cur, SvPVX (string));
1012 SV *uni = sv_newmortal (); 1402 SV *uni = sv_newmortal ();
1013 1403
1014 // horrible hack to silence warning inside pv_uni_display 1404 // horrible hack to silence warning inside pv_uni_display
1015 COP cop = *PL_curcop; 1405 COP cop = *PL_curcop;
1016 cop.cop_warnings = pWARN_NONE; 1406 cop.cop_warnings = pWARN_NONE;
… …
1026 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1416 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1027 } 1417 }
1028 1418
1029 sv = sv_2mortal (sv); 1419 sv = sv_2mortal (sv);
1030 1420
1031 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1421 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)"); 1422 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1033 1423
1034 return sv; 1424 return sv;
1035} 1425}
1036 1426
… …
1040MODULE = JSON::XS PACKAGE = JSON::XS 1430MODULE = JSON::XS PACKAGE = JSON::XS
1041 1431
1042BOOT: 1432BOOT:
1043{ 1433{
1044 int i; 1434 int i;
1045
1046 memset (decode_hexdigit, 0xff, 256);
1047 1435
1048 for (i = 0; i < 256; ++i) 1436 for (i = 0; i < 256; ++i)
1049 decode_hexdigit [i] = 1437 decode_hexdigit [i] =
1050 i >= '0' && i <= '9' ? i - '0' 1438 i >= '0' && i <= '9' ? i - '0'
1051 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1439 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1052 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1440 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1053 : -1; 1441 : -1;
1054 1442
1055 json_stash = gv_stashpv ("JSON::XS", 1); 1443 json_stash = gv_stashpv ("JSON::XS" , 1);
1444 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1445
1446 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1447 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1056} 1448}
1057 1449
1058PROTOTYPES: DISABLE 1450PROTOTYPES: DISABLE
1059 1451
1060SV *new (char *dummy) 1452void CLONE (...)
1061 CODE: 1453 CODE:
1062 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1454 json_stash = 0;
1455 json_boolean_stash = 0;
1456
1457void new (char *klass)
1458 PPCODE:
1459{
1460 SV *pv = NEWSV (0, sizeof (JSON));
1461 SvPOK_only (pv);
1462 Zero (SvPVX (pv), 1, JSON);
1463 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1464 XPUSHs (sv_2mortal (sv_bless (
1465 newRV_noinc (pv),
1466 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1467 )));
1468}
1469
1470void ascii (JSON *self, int enable = 1)
1471 ALIAS:
1472 ascii = F_ASCII
1473 latin1 = F_LATIN1
1474 utf8 = F_UTF8
1475 indent = F_INDENT
1476 canonical = F_CANONICAL
1477 space_before = F_SPACE_BEFORE
1478 space_after = F_SPACE_AFTER
1479 pretty = F_PRETTY
1480 allow_nonref = F_ALLOW_NONREF
1481 shrink = F_SHRINK
1482 allow_blessed = F_ALLOW_BLESSED
1483 convert_blessed = F_CONV_BLESSED
1484 relaxed = F_RELAXED
1485 PPCODE:
1486{
1487 if (enable)
1488 self->flags |= ix;
1489 else
1490 self->flags &= ~ix;
1491
1492 XPUSHs (ST (0));
1493}
1494
1495void get_ascii (JSON *self)
1496 ALIAS:
1497 get_ascii = F_ASCII
1498 get_latin1 = F_LATIN1
1499 get_utf8 = F_UTF8
1500 get_indent = F_INDENT
1501 get_canonical = F_CANONICAL
1502 get_space_before = F_SPACE_BEFORE
1503 get_space_after = F_SPACE_AFTER
1504 get_allow_nonref = F_ALLOW_NONREF
1505 get_shrink = F_SHRINK
1506 get_allow_blessed = F_ALLOW_BLESSED
1507 get_convert_blessed = F_CONV_BLESSED
1508 get_relaxed = F_RELAXED
1509 PPCODE:
1510 XPUSHs (boolSV (self->flags & ix));
1511
1512void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1513 PPCODE:
1514{
1515 UV log2 = 0;
1516
1517 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1518
1519 while ((1UL << log2) < max_depth)
1520 ++log2;
1521
1522 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1523
1524 XPUSHs (ST (0));
1525}
1526
1527U32 get_max_depth (JSON *self)
1528 CODE:
1529 RETVAL = DEC_DEPTH (self->flags);
1063 OUTPUT: 1530 OUTPUT:
1064 RETVAL 1531 RETVAL
1065 1532
1066SV *ascii (SV *self, int enable = 1) 1533void max_size (JSON *self, UV max_size = 0)
1067 ALIAS: 1534 PPCODE:
1068 ascii = F_ASCII 1535{
1069 utf8 = F_UTF8 1536 UV log2 = 0;
1070 indent = F_INDENT 1537
1071 canonical = F_CANONICAL 1538 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1072 space_before = F_SPACE_BEFORE 1539 if (max_size == 1) max_size = 2;
1073 space_after = F_SPACE_AFTER 1540
1074 pretty = F_PRETTY 1541 while ((1UL << log2) < max_size)
1075 allow_nonref = F_ALLOW_NONREF 1542 ++log2;
1076 shrink = F_SHRINK 1543
1544 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1545
1546 XPUSHs (ST (0));
1547}
1548
1549int get_max_size (JSON *self)
1077 CODE: 1550 CODE:
1078{ 1551 RETVAL = DEC_SIZE (self->flags);
1079 UV *uv = SvJSON (self);
1080 if (enable)
1081 *uv |= ix;
1082 else
1083 *uv &= ~ix;
1084
1085 RETVAL = newSVsv (self);
1086}
1087 OUTPUT: 1552 OUTPUT:
1088 RETVAL 1553 RETVAL
1089 1554
1090SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1555void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1556 PPCODE:
1557{
1558 SvREFCNT_dec (self->cb_object);
1559 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1560
1561 XPUSHs (ST (0));
1562}
1563
1564void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1565 PPCODE:
1566{
1567 if (!self->cb_sk_object)
1568 self->cb_sk_object = newHV ();
1569
1570 if (SvOK (cb))
1571 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1572 else
1573 {
1574 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1575
1576 if (!HvKEYS (self->cb_sk_object))
1577 {
1578 SvREFCNT_dec (self->cb_sk_object);
1579 self->cb_sk_object = 0;
1580 }
1581 }
1582
1583 XPUSHs (ST (0));
1584}
1585
1586void encode (JSON *self, SV *scalar)
1587 PPCODE:
1588 XPUSHs (encode_json (scalar, self));
1589
1590void decode (JSON *self, SV *jsonstr)
1591 PPCODE:
1592 XPUSHs (decode_json (jsonstr, self, 0));
1593
1594void decode_prefix (JSON *self, SV *jsonstr)
1595 PPCODE:
1596{
1597 UV offset;
1598 EXTEND (SP, 2);
1599 PUSHs (decode_json (jsonstr, self, &offset));
1600 PUSHs (sv_2mortal (newSVuv (offset)));
1601}
1602
1603void DESTROY (JSON *self)
1091 CODE: 1604 CODE:
1092{ 1605 SvREFCNT_dec (self->cb_sk_object);
1093 UV *uv = SvJSON (self); 1606 SvREFCNT_dec (self->cb_object);
1094 UV log2 = 0;
1095 1607
1096 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1608PROTOTYPES: ENABLE
1097 1609
1098 while ((1UL << log2) < max_depth)
1099 ++log2;
1100
1101 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1102
1103 RETVAL = newSVsv (self);
1104}
1105 OUTPUT:
1106 RETVAL
1107
1108void encode (SV *self, SV *scalar) 1610void encode_json (SV *scalar)
1109 PPCODE: 1611 PPCODE:
1612{
1613 JSON json = { F_DEFAULT | F_UTF8 };
1110 XPUSHs (encode_json (scalar, *SvJSON (self))); 1614 XPUSHs (encode_json (scalar, &json));
1615}
1111 1616
1112void decode (SV *self, SV *jsonstr) 1617void decode_json (SV *jsonstr)
1113 PPCODE: 1618 PPCODE:
1619{
1620 JSON json = { F_DEFAULT | F_UTF8 };
1114 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1621 XPUSHs (decode_json (jsonstr, &json, 0));
1622}
1115 1623
1116PROTOTYPES: ENABLE
1117
1118void to_json (SV *scalar)
1119 ALIAS:
1120 objToJson = 0
1121 PPCODE:
1122 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1123
1124void from_json (SV *jsonstr)
1125 ALIAS:
1126 jsonToObj = 0
1127 PPCODE:
1128 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1129

Diff Legend

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