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 1.25 by root, Fri Apr 6 20:37:22 2007 UTC vs.
Revision 1.71 by root, Wed Mar 19 03:17:38 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>
9#include <float.h>
8 10
11#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h
13#endif
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
9#define F_ASCII 0x00000001UL 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002UL 23#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004UL 24#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008UL 25#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010UL 26#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020UL 27#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080UL 28#define F_ALLOW_NONREF 0x00000100UL
16#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
17#define F_MAXDEPTH 0xf8000000UL 34#define F_MAXDEPTH 0xf8000000UL
18#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
19 39
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24 42
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (9UL << S_MAXDEPTH) 44#define F_DEFAULT (9UL << S_MAXDEPTH)
27 45
28#define INIT_SIZE 32 // initial scalar size to be allocated 46#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 47#define INDENT_STEP 3 // spaces per indentation level
30 48
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#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
33 50
34#define SB do { 51#define SB do {
35#define SE } while (0) 52#define SE } while (0)
36 53
54#if __GNUC__ >= 3
55# define expect(expr,value) __builtin_expect ((expr), (value))
56# define INLINE static 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
37static 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;
38 81
39///////////////////////////////////////////////////////////////////////////// 82/////////////////////////////////////////////////////////////////////////////
40// utility functions 83// utility functions
41 84
42static UV * 85INLINE void
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv) 86shrink (SV *sv)
53{ 87{
54 sv_utf8_downgrade (sv, 1); 88 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1) 89 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 90 {
… …
65// 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
66// case of an error. 100// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 101// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 102// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 103// note that we never call this function for a ascii codepoints
70static UV 104INLINE UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 106{
73 if (s[0] > 0xdf || s[0] < 0xc2) 107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 110 {
77 *clen = 2; 111 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
… …
91typedef struct 125typedef struct
92{ 126{
93 char *cur; // SvPVX (sv) + current output position 127 char *cur; // SvPVX (sv) + current output position
94 char *end; // SvEND (sv) 128 char *end; // SvEND (sv)
95 SV *sv; // result scalar 129 SV *sv; // result scalar
96 U32 flags; // F_* 130 JSON json;
97 U32 indent; // indentation level 131 U32 indent; // indentation level
98 U32 maxdepth; // max. indentation/recursion level 132 U32 maxdepth; // max. indentation/recursion level
133 UV limit; // escape character values >= this value when encoding
99} enc_t; 134} enc_t;
100 135
101static void 136INLINE void
102need (enc_t *enc, STRLEN len) 137need (enc_t *enc, STRLEN len)
103{ 138{
104 if (enc->cur + len >= enc->end) 139 if (expect_false (enc->cur + len >= enc->end))
105 { 140 {
106 STRLEN cur = enc->cur - SvPVX (enc->sv); 141 STRLEN cur = enc->cur - SvPVX (enc->sv);
107 SvGROW (enc->sv, cur + len + 1); 142 SvGROW (enc->sv, cur + len + 1);
108 enc->cur = SvPVX (enc->sv) + cur; 143 enc->cur = SvPVX (enc->sv) + cur;
109 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 144 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
110 } 145 }
111} 146}
112 147
113static void 148INLINE void
114encode_ch (enc_t *enc, char ch) 149encode_ch (enc_t *enc, char ch)
115{ 150{
116 need (enc, 1); 151 need (enc, 1);
117 *enc->cur++ = ch; 152 *enc->cur++ = ch;
118} 153}
… …
126 161
127 while (str < end) 162 while (str < end)
128 { 163 {
129 unsigned char ch = *(unsigned char *)str; 164 unsigned char ch = *(unsigned char *)str;
130 165
131 if (ch >= 0x20 && ch < 0x80) // most common case 166 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
132 { 167 {
133 if (ch == '"') // but with slow exceptions 168 if (expect_false (ch == '"')) // but with slow exceptions
134 { 169 {
135 need (enc, len += 1); 170 need (enc, len += 1);
136 *enc->cur++ = '\\'; 171 *enc->cur++ = '\\';
137 *enc->cur++ = '"'; 172 *enc->cur++ = '"';
138 } 173 }
139 else if (ch == '\\') 174 else if (expect_false (ch == '\\'))
140 { 175 {
141 need (enc, len += 1); 176 need (enc, len += 1);
142 *enc->cur++ = '\\'; 177 *enc->cur++ = '\\';
143 *enc->cur++ = '\\'; 178 *enc->cur++ = '\\';
144 } 179 }
… …
162 STRLEN clen; 197 STRLEN clen;
163 UV uch; 198 UV uch;
164 199
165 if (is_utf8) 200 if (is_utf8)
166 { 201 {
167 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
168 uch = decode_utf8 (str, end - str, &clen); 202 uch = decode_utf8 (str, end - str, &clen);
169 if (clen == (STRLEN)-1) 203 if (clen == (STRLEN)-1)
170 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 204 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
171 } 205 }
172 else 206 else
173 { 207 {
174 uch = ch; 208 uch = ch;
175 clen = 1; 209 clen = 1;
176 } 210 }
177 211
178 if (uch > 0x10FFFFUL)
179 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
180
181 if (uch < 0x80 || enc->flags & F_ASCII) 212 if (uch < 0x20 || uch >= enc->limit)
182 { 213 {
183 if (uch > 0xFFFFUL) 214 if (uch > 0xFFFFUL)
184 { 215 {
216 if (uch > 0x10FFFFUL)
217 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
218
185 need (enc, len += 11); 219 need (enc, len += 11);
186 sprintf (enc->cur, "\\u%04x\\u%04x", 220 sprintf (enc->cur, "\\u%04x\\u%04x",
187 (int)((uch - 0x10000) / 0x400 + 0xD800), 221 (int)((uch - 0x10000) / 0x400 + 0xD800),
188 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 222 (int)((uch - 0x10000) % 0x400 + 0xDC00));
189 enc->cur += 12; 223 enc->cur += 12;
… …
200 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 234 *enc->cur++ = hexdigit [(uch >> 0) & 15];
201 } 235 }
202 236
203 str += clen; 237 str += clen;
204 } 238 }
239 else if (enc->json.flags & F_LATIN1)
240 {
241 *enc->cur++ = uch;
242 str += clen;
243 }
205 else if (is_utf8) 244 else if (is_utf8)
206 { 245 {
207 need (enc, len += clen); 246 need (enc, len += clen);
208 do 247 do
209 { 248 {
… …
211 } 250 }
212 while (--clen); 251 while (--clen);
213 } 252 }
214 else 253 else
215 { 254 {
216 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 255 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
217 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 256 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
218 ++str; 257 ++str;
219 } 258 }
220 } 259 }
221 } 260 }
… …
223 262
224 --len; 263 --len;
225 } 264 }
226} 265}
227 266
228static void 267INLINE void
229encode_indent (enc_t *enc) 268encode_indent (enc_t *enc)
230{ 269{
231 if (enc->flags & F_INDENT) 270 if (enc->json.flags & F_INDENT)
232 { 271 {
233 int spaces = enc->indent * INDENT_STEP; 272 int spaces = enc->indent * INDENT_STEP;
234 273
235 need (enc, spaces); 274 need (enc, spaces);
236 memset (enc->cur, ' ', spaces); 275 memset (enc->cur, ' ', spaces);
237 enc->cur += spaces; 276 enc->cur += spaces;
238 } 277 }
239} 278}
240 279
241static void 280INLINE void
242encode_space (enc_t *enc) 281encode_space (enc_t *enc)
243{ 282{
244 need (enc, 1); 283 need (enc, 1);
245 encode_ch (enc, ' '); 284 encode_ch (enc, ' ');
246} 285}
247 286
248static void 287INLINE void
249encode_nl (enc_t *enc) 288encode_nl (enc_t *enc)
250{ 289{
251 if (enc->flags & F_INDENT) 290 if (enc->json.flags & F_INDENT)
252 { 291 {
253 need (enc, 1); 292 need (enc, 1);
254 encode_ch (enc, '\n'); 293 encode_ch (enc, '\n');
255 } 294 }
256} 295}
257 296
258static void 297INLINE void
259encode_comma (enc_t *enc) 298encode_comma (enc_t *enc)
260{ 299{
261 encode_ch (enc, ','); 300 encode_ch (enc, ',');
262 301
263 if (enc->flags & F_INDENT) 302 if (enc->json.flags & F_INDENT)
264 encode_nl (enc); 303 encode_nl (enc);
265 else if (enc->flags & F_SPACE_AFTER) 304 else if (enc->json.flags & F_SPACE_AFTER)
266 encode_space (enc); 305 encode_space (enc);
267} 306}
268 307
269static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
270 309
… …
274 int i, len = av_len (av); 313 int i, len = av_len (av);
275 314
276 if (enc->indent >= enc->maxdepth) 315 if (enc->indent >= enc->maxdepth)
277 croak ("data structure too deep (hit recursion limit)"); 316 croak ("data structure too deep (hit recursion limit)");
278 317
279 encode_ch (enc, '['); encode_nl (enc); 318 encode_ch (enc, '[');
280 ++enc->indent; 319
320 if (len >= 0)
321 {
322 encode_nl (enc); ++enc->indent;
281 323
282 for (i = 0; i <= len; ++i) 324 for (i = 0; i <= len; ++i)
283 { 325 {
326 SV **svp = av_fetch (av, i, 0);
327
284 encode_indent (enc); 328 encode_indent (enc);
285 encode_sv (enc, *av_fetch (av, i, 0));
286 329
330 if (svp)
331 encode_sv (enc, *svp);
332 else
333 encode_str (enc, "null", 4, 0);
334
287 if (i < len) 335 if (i < len)
288 encode_comma (enc); 336 encode_comma (enc);
289 } 337 }
290 338
339 encode_nl (enc); --enc->indent; encode_indent (enc);
340 }
341
291 encode_nl (enc); 342 encode_ch (enc, ']');
292
293 --enc->indent;
294 encode_indent (enc); encode_ch (enc, ']');
295} 343}
296 344
297static void 345static void
298encode_he (enc_t *enc, HE *he) 346encode_hk (enc_t *enc, HE *he)
299{ 347{
300 encode_ch (enc, '"'); 348 encode_ch (enc, '"');
301 349
302 if (HeKLEN (he) == HEf_SVKEY) 350 if (HeKLEN (he) == HEf_SVKEY)
303 { 351 {
… …
313 else 361 else
314 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 362 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
315 363
316 encode_ch (enc, '"'); 364 encode_ch (enc, '"');
317 365
318 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 366 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
319 encode_ch (enc, ':'); 367 encode_ch (enc, ':');
320 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 368 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
321 encode_sv (enc, HeVAL (he));
322} 369}
323 370
324// compare hash entries, used when all keys are bytestrings 371// compare hash entries, used when all keys are bytestrings
325static int 372static int
326he_cmp_fast (const void *a_, const void *b_) 373he_cmp_fast (const void *a_, const void *b_)
… …
331 HE *b = *(HE **)b_; 378 HE *b = *(HE **)b_;
332 379
333 STRLEN la = HeKLEN (a); 380 STRLEN la = HeKLEN (a);
334 STRLEN lb = HeKLEN (b); 381 STRLEN lb = HeKLEN (b);
335 382
336 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 383 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
337 cmp = la - lb; 384 cmp = lb - la;
338 385
339 return cmp; 386 return cmp;
340} 387}
341 388
342// compare hash entries, used when some keys are sv's or utf-x 389// compare hash entries, used when some keys are sv's or utf-x
343static int 390static int
344he_cmp_slow (const void *a, const void *b) 391he_cmp_slow (const void *a, const void *b)
345{ 392{
346 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 393 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
347} 394}
348 395
349static void 396static void
350encode_hv (enc_t *enc, HV *hv) 397encode_hv (enc_t *enc, HV *hv)
351{ 398{
399 HE *he;
352 int count, i; 400 int count;
353 401
354 if (enc->indent >= enc->maxdepth) 402 if (enc->indent >= enc->maxdepth)
355 croak ("data structure too deep (hit recursion limit)"); 403 croak ("data structure too deep (hit recursion limit)");
356 404
357 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 405 encode_ch (enc, '{');
358 406
359 if ((count = hv_iterinit (hv)))
360 {
361 // for canonical output we have to sort by keys first 407 // for canonical output we have to sort by keys first
362 // actually, this is mostly due to the stupid so-called 408 // actually, this is mostly due to the stupid so-called
363 // security workaround added somewhere in 5.8.x. 409 // security workaround added somewhere in 5.8.x.
364 // that randomises hash orderings 410 // that randomises hash orderings
365 if (enc->flags & F_CANONICAL) 411 if (enc->json.flags & F_CANONICAL)
412 {
413 int count = hv_iterinit (hv);
414
415 if (SvMAGICAL (hv))
366 { 416 {
417 // need to count by iterating. could improve by dynamically building the vector below
418 // but I don't care for the speed of this special case.
419 // note also that we will run into undefined behaviour when the two iterations
420 // do not result in the same count, something I might care for in some later release.
421
422 count = 0;
423 while (hv_iternext (hv))
424 ++count;
425
426 hv_iterinit (hv);
427 }
428
429 if (count)
430 {
431 int i, fast = 1;
432#if defined(__BORLANDC__) || defined(_MSC_VER)
433 HE **hes = _alloca (count * sizeof (HE));
434#else
367 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 435 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
368 int fast = 1; 436#endif
369 437
370 i = 0; 438 i = 0;
371 while ((he = hv_iternext (hv))) 439 while ((he = hv_iternext (hv)))
372 { 440 {
373 hes [i++] = he; 441 hes [i++] = he;
… …
395 463
396 FREETMPS; 464 FREETMPS;
397 LEAVE; 465 LEAVE;
398 } 466 }
399 467
400 for (i = 0; i < count; ++i) 468 encode_nl (enc); ++enc->indent;
469
470 while (count--)
401 { 471 {
402 encode_indent (enc); 472 encode_indent (enc);
473 he = hes [count];
403 encode_he (enc, hes [i]); 474 encode_hk (enc, he);
475 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
404 476
405 if (i < count - 1) 477 if (count)
406 encode_comma (enc); 478 encode_comma (enc);
407 } 479 }
408 480
481 encode_nl (enc); --enc->indent; encode_indent (enc);
482 }
483 }
484 else
485 {
486 if (hv_iterinit (hv) || SvMAGICAL (hv))
487 if ((he = hv_iternext (hv)))
488 {
489 encode_nl (enc); ++enc->indent;
490
491 for (;;)
492 {
409 encode_nl (enc); 493 encode_indent (enc);
494 encode_hk (enc, he);
495 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
496
497 if (!(he = hv_iternext (hv)))
498 break;
499
500 encode_comma (enc);
501 }
502
503 encode_nl (enc); --enc->indent; encode_indent (enc);
504 }
505 }
506
507 encode_ch (enc, '}');
508}
509
510// encode objects, arrays and special \0=false and \1=true values.
511static void
512encode_rv (enc_t *enc, SV *sv)
513{
514 svtype svt;
515
516 SvGETMAGIC (sv);
517 svt = SvTYPE (sv);
518
519 if (expect_false (SvOBJECT (sv)))
520 {
521 HV *stash = !JSON_SLOW || json_boolean_stash
522 ? json_boolean_stash
523 : gv_stashpv ("JSON::XS::Boolean", 1);
524
525 if (SvSTASH (sv) == stash)
526 {
527 if (SvIV (sv))
528 encode_str (enc, "true", 4, 0);
529 else
530 encode_str (enc, "false", 5, 0);
410 } 531 }
411 else 532 else
412 { 533 {
413 HE *he = hv_iternext (hv); 534#if 0
414 535 if (0 && sv_derived_from (rv, "JSON::Literal"))
415 for (;;)
416 { 536 {
417 encode_indent (enc); 537 // not yet
418 encode_he (enc, he);
419
420 if (!(he = hv_iternext (hv)))
421 break;
422
423 encode_comma (enc);
424 } 538 }
539#endif
540 if (enc->json.flags & F_CONV_BLESSED)
541 {
542 // we re-bless the reference to get overload and other niceties right
543 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
425 544
545 if (to_json)
546 {
547 dSP;
548
549 ENTER; SAVETMPS; PUSHMARK (SP);
550 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
551
552 // calling with G_SCALAR ensures that we always get a 1 return value
553 PUTBACK;
554 call_sv ((SV *)GvCV (to_json), G_SCALAR);
555 SPAGAIN;
556
557 // catch this surprisingly common error
558 if (SvROK (TOPs) && SvRV (TOPs) == sv)
559 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
560
561 sv = POPs;
562 PUTBACK;
563
426 encode_nl (enc); 564 encode_sv (enc, sv);
565
566 FREETMPS; LEAVE;
567 }
568 else if (enc->json.flags & F_ALLOW_BLESSED)
569 encode_str (enc, "null", 4, 0);
570 else
571 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
572 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
573 }
574 else if (enc->json.flags & F_ALLOW_BLESSED)
575 encode_str (enc, "null", 4, 0);
576 else
577 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
578 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
427 } 579 }
428 } 580 }
429
430 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
431}
432
433// encode objects, arrays and special \0=false and \1=true values.
434static void
435encode_rv (enc_t *enc, SV *sv)
436{
437 svtype svt;
438
439 SvGETMAGIC (sv);
440 svt = SvTYPE (sv);
441
442 if (svt == SVt_PVHV) 581 else if (svt == SVt_PVHV)
443 encode_hv (enc, (HV *)sv); 582 encode_hv (enc, (HV *)sv);
444 else if (svt == SVt_PVAV) 583 else if (svt == SVt_PVAV)
445 encode_av (enc, (AV *)sv); 584 encode_av (enc, (AV *)sv);
446 else if (svt < SVt_PVAV) 585 else if (svt < SVt_PVAV)
447 { 586 {
448 if (SvNIOK (sv) && SvIV (sv) == 0) 587 STRLEN len = 0;
588 char *pv = svt ? SvPV (sv, len) : 0;
589
590 if (len == 1 && *pv == '1')
591 encode_str (enc, "true", 4, 0);
592 else if (len == 1 && *pv == '0')
449 encode_str (enc, "false", 5, 0); 593 encode_str (enc, "false", 5, 0);
450 else if (SvNIOK (sv) && SvIV (sv) == 1)
451 encode_str (enc, "true", 4, 0);
452 else 594 else
453 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 595 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
454 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 596 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
455 } 597 }
456 else 598 else
… …
471 encode_str (enc, str, len, SvUTF8 (sv)); 613 encode_str (enc, str, len, SvUTF8 (sv));
472 encode_ch (enc, '"'); 614 encode_ch (enc, '"');
473 } 615 }
474 else if (SvNOKp (sv)) 616 else if (SvNOKp (sv))
475 { 617 {
618 // trust that perl will do the right thing w.r.t. JSON syntax.
476 need (enc, NV_DIG + 32); 619 need (enc, NV_DIG + 32);
477 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 620 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
478 enc->cur += strlen (enc->cur); 621 enc->cur += strlen (enc->cur);
479 } 622 }
480 else if (SvIOKp (sv)) 623 else if (SvIOKp (sv))
481 { 624 {
482 need (enc, 64); 625 // we assume we can always read an IV as a UV
626 if (SvUV (sv) & ~(UV)0x7fff)
627 {
628 // large integer, use the (rather slow) snprintf way.
629 need (enc, sizeof (UV) * 3);
483 enc->cur += 630 enc->cur +=
484 SvIsUV(sv) 631 SvIsUV(sv)
485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 632 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 633 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
634 }
635 else
636 {
637 // optimise the "small number case"
638 // code will likely be branchless and use only a single multiplication
639 I32 i = SvIV (sv);
640 U32 u;
641 char digit, nz = 0;
642
643 need (enc, 6);
644
645 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
646 u = i < 0 ? -i : i;
647
648 // convert to 4.28 fixed-point representation
649 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
650
651 // now output digit by digit, each time masking out the integer part
652 // and multiplying by 5 while moving the decimal point one to the right,
653 // resulting in a net multiplication by 10.
654 // we always write the digit to memory but conditionally increment
655 // the pointer, to ease the usage of conditional move instructions.
656 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
657 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
658 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
659 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
660 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
661 }
487 } 662 }
488 else if (SvROK (sv)) 663 else if (SvROK (sv))
489 encode_rv (enc, SvRV (sv)); 664 encode_rv (enc, SvRV (sv));
490 else if (!SvOK (sv)) 665 else if (!SvOK (sv))
491 encode_str (enc, "null", 4, 0); 666 encode_str (enc, "null", 4, 0);
… …
493 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 668 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
494 SvPV_nolen (sv), SvFLAGS (sv)); 669 SvPV_nolen (sv), SvFLAGS (sv));
495} 670}
496 671
497static SV * 672static SV *
498encode_json (SV *scalar, U32 flags) 673encode_json (SV *scalar, JSON *json)
499{ 674{
500 enc_t enc; 675 enc_t enc;
501 676
502 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 677 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
503 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 678 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
504 679
505 enc.flags = flags; 680 enc.json = *json;
506 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 681 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
507 enc.cur = SvPVX (enc.sv); 682 enc.cur = SvPVX (enc.sv);
508 enc.end = SvEND (enc.sv); 683 enc.end = SvEND (enc.sv);
509 enc.indent = 0; 684 enc.indent = 0;
510 enc.maxdepth = DEC_DEPTH (flags); 685 enc.maxdepth = DEC_DEPTH (enc.json.flags);
686 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
687 : enc.json.flags & F_LATIN1 ? 0x000100UL
688 : 0x10FFFFUL;
511 689
512 SvPOK_only (enc.sv); 690 SvPOK_only (enc.sv);
513 encode_sv (&enc, scalar); 691 encode_sv (&enc, scalar);
514 692
693 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
694 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
695
515 if (!(flags & (F_ASCII | F_UTF8))) 696 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
516 SvUTF8_on (enc.sv); 697 SvUTF8_on (enc.sv);
517 698
518 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
519
520 if (enc.flags & F_SHRINK) 699 if (enc.json.flags & F_SHRINK)
521 shrink (enc.sv); 700 shrink (enc.sv);
522 701
523 return enc.sv; 702 return enc.sv;
524} 703}
525 704
… …
530typedef struct 709typedef struct
531{ 710{
532 char *cur; // current parser pointer 711 char *cur; // current parser pointer
533 char *end; // end of input string 712 char *end; // end of input string
534 const char *err; // parse error, if != 0 713 const char *err; // parse error, if != 0
535 U32 flags; // F_* 714 JSON json;
536 U32 depth; // recursion depth 715 U32 depth; // recursion depth
537 U32 maxdepth; // recursion depth limit 716 U32 maxdepth; // recursion depth limit
538} dec_t; 717} dec_t;
539 718
540static void 719INLINE void
720decode_comment (dec_t *dec)
721{
722 // only '#'-style comments allowed a.t.m.
723
724 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
725 ++dec->cur;
726}
727
728INLINE void
541decode_ws (dec_t *dec) 729decode_ws (dec_t *dec)
542{ 730{
543 for (;;) 731 for (;;)
544 { 732 {
545 char ch = *dec->cur; 733 char ch = *dec->cur;
546 734
547 if (ch > 0x20 735 if (ch > 0x20)
736 {
737 if (expect_false (ch == '#'))
738 {
739 if (dec->json.flags & F_RELAXED)
740 decode_comment (dec);
741 else
742 break;
743 }
744 else
745 break;
746 }
548 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 747 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
549 break; 748 break; // parse error, but let higher level handle it, gives better error messages
550 749
551 ++dec->cur; 750 ++dec->cur;
552 } 751 }
553} 752}
554 753
… …
571decode_4hex (dec_t *dec) 770decode_4hex (dec_t *dec)
572{ 771{
573 signed char d1, d2, d3, d4; 772 signed char d1, d2, d3, d4;
574 unsigned char *cur = (unsigned char *)dec->cur; 773 unsigned char *cur = (unsigned char *)dec->cur;
575 774
576 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 775 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
577 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 776 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
578 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 777 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
579 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 778 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
580 779
581 dec->cur += 4; 780 dec->cur += 4;
582 781
583 return ((UV)d1) << 12 782 return ((UV)d1) << 12
584 | ((UV)d2) << 8 783 | ((UV)d2) << 8
… …
592static SV * 791static SV *
593decode_str (dec_t *dec) 792decode_str (dec_t *dec)
594{ 793{
595 SV *sv = 0; 794 SV *sv = 0;
596 int utf8 = 0; 795 int utf8 = 0;
796 char *dec_cur = dec->cur;
597 797
598 do 798 do
599 { 799 {
600 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 800 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
601 char *cur = buf; 801 char *cur = buf;
602 802
603 do 803 do
604 { 804 {
605 unsigned char ch = *(unsigned char *)dec->cur++; 805 unsigned char ch = *(unsigned char *)dec_cur++;
606 806
607 if (ch == '"') 807 if (expect_false (ch == '"'))
608 { 808 {
609 --dec->cur; 809 --dec_cur;
610 break; 810 break;
611 } 811 }
612 else if (ch == '\\') 812 else if (expect_false (ch == '\\'))
613 { 813 {
614 switch (*dec->cur) 814 switch (*dec_cur)
615 { 815 {
616 case '\\': 816 case '\\':
617 case '/': 817 case '/':
618 case '"': *cur++ = *dec->cur++; break; 818 case '"': *cur++ = *dec_cur++; break;
619 819
620 case 'b': ++dec->cur; *cur++ = '\010'; break; 820 case 'b': ++dec_cur; *cur++ = '\010'; break;
621 case 't': ++dec->cur; *cur++ = '\011'; break; 821 case 't': ++dec_cur; *cur++ = '\011'; break;
622 case 'n': ++dec->cur; *cur++ = '\012'; break; 822 case 'n': ++dec_cur; *cur++ = '\012'; break;
623 case 'f': ++dec->cur; *cur++ = '\014'; break; 823 case 'f': ++dec_cur; *cur++ = '\014'; break;
624 case 'r': ++dec->cur; *cur++ = '\015'; break; 824 case 'r': ++dec_cur; *cur++ = '\015'; break;
625 825
626 case 'u': 826 case 'u':
627 { 827 {
628 UV lo, hi; 828 UV lo, hi;
629 ++dec->cur; 829 ++dec_cur;
630 830
831 dec->cur = dec_cur;
631 hi = decode_4hex (dec); 832 hi = decode_4hex (dec);
833 dec_cur = dec->cur;
632 if (hi == (UV)-1) 834 if (hi == (UV)-1)
633 goto fail; 835 goto fail;
634 836
635 // possibly a surrogate pair 837 // possibly a surrogate pair
636 if (hi >= 0xd800) 838 if (hi >= 0xd800)
637 if (hi < 0xdc00) 839 if (hi < 0xdc00)
638 { 840 {
639 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 841 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
640 ERR ("missing low surrogate character in surrogate pair"); 842 ERR ("missing low surrogate character in surrogate pair");
641 843
642 dec->cur += 2; 844 dec_cur += 2;
643 845
846 dec->cur = dec_cur;
644 lo = decode_4hex (dec); 847 lo = decode_4hex (dec);
848 dec_cur = dec->cur;
645 if (lo == (UV)-1) 849 if (lo == (UV)-1)
646 goto fail; 850 goto fail;
647 851
648 if (lo < 0xdc00 || lo >= 0xe000) 852 if (lo < 0xdc00 || lo >= 0xe000)
649 ERR ("surrogate pair expected"); 853 ERR ("surrogate pair expected");
… …
663 *cur++ = hi; 867 *cur++ = hi;
664 } 868 }
665 break; 869 break;
666 870
667 default: 871 default:
668 --dec->cur; 872 --dec_cur;
669 ERR ("illegal backslash escape sequence in string"); 873 ERR ("illegal backslash escape sequence in string");
670 } 874 }
671 } 875 }
672 else if (ch >= 0x20 && ch <= 0x7f) 876 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
673 *cur++ = ch; 877 *cur++ = ch;
674 else if (ch >= 0x80) 878 else if (ch >= 0x80)
675 { 879 {
676 STRLEN clen; 880 STRLEN clen;
677 UV uch; 881 UV uch;
678 882
679 --dec->cur; 883 --dec_cur;
680 884
681 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 885 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
682 if (clen == (STRLEN)-1) 886 if (clen == (STRLEN)-1)
683 ERR ("malformed UTF-8 character in JSON string"); 887 ERR ("malformed UTF-8 character in JSON string");
684 888
685 do 889 do
686 *cur++ = *dec->cur++; 890 *cur++ = *dec_cur++;
687 while (--clen); 891 while (--clen);
688 892
689 utf8 = 1; 893 utf8 = 1;
690 } 894 }
691 else 895 else
692 { 896 {
693 --dec->cur; 897 --dec_cur;
694 898
695 if (!ch) 899 if (!ch)
696 ERR ("unexpected end of string while parsing JSON string"); 900 ERR ("unexpected end of string while parsing JSON string");
697 else 901 else
698 ERR ("invalid character encountered while parsing JSON string"); 902 ERR ("invalid character encountered while parsing JSON string");
… …
711 } 915 }
712 else 916 else
713 sv = newSVpvn (buf, len); 917 sv = newSVpvn (buf, len);
714 } 918 }
715 } 919 }
716 while (*dec->cur != '"'); 920 while (*dec_cur != '"');
717 921
718 ++dec->cur; 922 ++dec_cur;
719 923
720 if (sv) 924 if (sv)
721 { 925 {
722 SvPOK_only (sv); 926 SvPOK_only (sv);
723 *SvEND (sv) = 0; 927 *SvEND (sv) = 0;
… …
726 SvUTF8_on (sv); 930 SvUTF8_on (sv);
727 } 931 }
728 else 932 else
729 sv = newSVpvn ("", 0); 933 sv = newSVpvn ("", 0);
730 934
935 dec->cur = dec_cur;
731 return sv; 936 return sv;
732 937
733fail: 938fail:
939 dec->cur = dec_cur;
734 return 0; 940 return 0;
735} 941}
736 942
737static SV * 943static SV *
738decode_num (dec_t *dec) 944decode_num (dec_t *dec)
… …
796 is_nv = 1; 1002 is_nv = 1;
797 } 1003 }
798 1004
799 if (!is_nv) 1005 if (!is_nv)
800 { 1006 {
801 UV uv; 1007 int len = dec->cur - start;
802 int numtype = grok_number (start, dec->cur - start, &uv); 1008
803 if (numtype & IS_NUMBER_IN_UV) 1009 // special case the rather common 1..5-digit-int case
804 if (numtype & IS_NUMBER_NEG) 1010 if (*start == '-')
1011 switch (len)
805 { 1012 {
806 if (uv < (UV)IV_MIN) 1013 case 2: return newSViv (-( start [1] - '0' * 1));
807 return newSViv (-(IV)uv); 1014 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1015 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1016 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1017 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
808 } 1018 }
1019 else
1020 switch (len)
1021 {
1022 case 1: return newSViv ( start [0] - '0' * 1);
1023 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1024 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1025 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1026 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1027 }
1028
1029 {
1030 UV uv;
1031 int numtype = grok_number (start, len, &uv);
1032 if (numtype & IS_NUMBER_IN_UV)
1033 if (numtype & IS_NUMBER_NEG)
1034 {
1035 if (uv < (UV)IV_MIN)
1036 return newSViv (-(IV)uv);
1037 }
809 else 1038 else
810 return newSVuv (uv); 1039 return newSVuv (uv);
811 } 1040 }
812 1041
1042 len -= *start == '-' ? 1 : 0;
1043
1044 // does not fit into IV or UV, try NV
1045 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1046 #if defined (LDBL_DIG)
1047 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1048 #endif
1049 )
1050 // fits into NV without loss of precision
1051 return newSVnv (Atof (start));
1052
1053 // everything else fails, convert it to a string
1054 return newSVpvn (start, dec->cur - start);
1055 }
1056
1057 // loss of precision here
813 return newSVnv (Atof (start)); 1058 return newSVnv (Atof (start));
814 1059
815fail: 1060fail:
816 return 0; 1061 return 0;
817} 1062}
… …
847 1092
848 if (*dec->cur != ',') 1093 if (*dec->cur != ',')
849 ERR (", or ] expected while parsing array"); 1094 ERR (", or ] expected while parsing array");
850 1095
851 ++dec->cur; 1096 ++dec->cur;
1097
1098 decode_ws (dec);
1099
1100 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1101 {
1102 ++dec->cur;
1103 break;
1104 }
852 } 1105 }
853 1106
854 DEC_DEC_DEPTH; 1107 DEC_DEC_DEPTH;
855 return newRV_noinc ((SV *)av); 1108 return newRV_noinc ((SV *)av);
856 1109
… …
861} 1114}
862 1115
863static SV * 1116static SV *
864decode_hv (dec_t *dec) 1117decode_hv (dec_t *dec)
865{ 1118{
1119 SV *sv;
866 HV *hv = newHV (); 1120 HV *hv = newHV ();
867 1121
868 DEC_INC_DEPTH; 1122 DEC_INC_DEPTH;
869 decode_ws (dec); 1123 decode_ws (dec);
870 1124
871 if (*dec->cur == '}') 1125 if (*dec->cur == '}')
872 ++dec->cur; 1126 ++dec->cur;
873 else 1127 else
874 for (;;) 1128 for (;;)
875 { 1129 {
1130 EXPECT_CH ('"');
1131
1132 // heuristic: assume that
1133 // a) decode_str + hv_store_ent are abysmally slow.
1134 // b) most hash keys are short, simple ascii text.
1135 // => try to "fast-match" such strings to avoid
1136 // the overhead of decode_str + hv_store_ent.
1137 {
876 SV *key, *value; 1138 SV *value;
1139 char *p = dec->cur;
1140 char *e = p + 24; // only try up to 24 bytes
877 1141
878 decode_ws (dec); EXPECT_CH ('"'); 1142 for (;;)
879
880 key = decode_str (dec);
881 if (!key)
882 goto fail;
883
884 decode_ws (dec); EXPECT_CH (':');
885
886 value = decode_sv (dec);
887 if (!value)
888 { 1143 {
1144 // the >= 0x80 is true on most architectures
1145 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1146 {
1147 // slow path, back up and use decode_str
1148 SV *key = decode_str (dec);
1149 if (!key)
1150 goto fail;
1151
1152 decode_ws (dec); EXPECT_CH (':');
1153
1154 decode_ws (dec);
1155 value = decode_sv (dec);
1156 if (!value)
1157 {
1158 SvREFCNT_dec (key);
1159 goto fail;
1160 }
1161
1162 hv_store_ent (hv, key, value, 0);
889 SvREFCNT_dec (key); 1163 SvREFCNT_dec (key);
1164
1165 break;
1166 }
1167 else if (*p == '"')
1168 {
1169 // fast path, got a simple key
1170 char *key = dec->cur;
1171 int len = p - key;
1172 dec->cur = p + 1;
1173
1174 decode_ws (dec); EXPECT_CH (':');
1175
1176 decode_ws (dec);
1177 value = decode_sv (dec);
1178 if (!value)
890 goto fail; 1179 goto fail;
1180
1181 hv_store (hv, key, len, value, 0);
1182
1183 break;
1184 }
1185
1186 ++p;
891 } 1187 }
892 1188 }
893 hv_store_ent (hv, key, value, 0);
894 SvREFCNT_dec (key);
895 1189
896 decode_ws (dec); 1190 decode_ws (dec);
897 1191
898 if (*dec->cur == '}') 1192 if (*dec->cur == '}')
899 { 1193 {
… …
903 1197
904 if (*dec->cur != ',') 1198 if (*dec->cur != ',')
905 ERR (", or } expected while parsing object/hash"); 1199 ERR (", or } expected while parsing object/hash");
906 1200
907 ++dec->cur; 1201 ++dec->cur;
1202
1203 decode_ws (dec);
1204
1205 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1206 {
1207 ++dec->cur;
1208 break;
1209 }
908 } 1210 }
909 1211
910 DEC_DEC_DEPTH; 1212 DEC_DEC_DEPTH;
911 return newRV_noinc ((SV *)hv); 1213 sv = newRV_noinc ((SV *)hv);
1214
1215 // check filter callbacks
1216 if (dec->json.flags & F_HOOK)
1217 {
1218 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1219 {
1220 HE *cb, *he;
1221
1222 hv_iterinit (hv);
1223 he = hv_iternext (hv);
1224 hv_iterinit (hv);
1225
1226 // the next line creates a mortal sv each time its called.
1227 // might want to optimise this for common cases.
1228 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1229
1230 if (cb)
1231 {
1232 dSP;
1233 int count;
1234
1235 ENTER; SAVETMPS; PUSHMARK (SP);
1236 XPUSHs (HeVAL (he));
1237
1238 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1239
1240 if (count == 1)
1241 {
1242 sv = newSVsv (POPs);
1243 FREETMPS; LEAVE;
1244 return sv;
1245 }
1246
1247 FREETMPS; LEAVE;
1248 }
1249 }
1250
1251 if (dec->json.cb_object)
1252 {
1253 dSP;
1254 int count;
1255
1256 ENTER; SAVETMPS; PUSHMARK (SP);
1257 XPUSHs (sv_2mortal (sv));
1258
1259 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1260
1261 if (count == 1)
1262 {
1263 sv = newSVsv (POPs);
1264 FREETMPS; LEAVE;
1265 return sv;
1266 }
1267
1268 SvREFCNT_inc (sv);
1269 FREETMPS; LEAVE;
1270 }
1271 }
1272
1273 return sv;
912 1274
913fail: 1275fail:
914 SvREFCNT_dec (hv); 1276 SvREFCNT_dec (hv);
915 DEC_DEC_DEPTH; 1277 DEC_DEC_DEPTH;
916 return 0; 1278 return 0;
917} 1279}
918 1280
919static SV * 1281static SV *
920decode_sv (dec_t *dec) 1282decode_sv (dec_t *dec)
921{ 1283{
922 decode_ws (dec); 1284 // the beauty of JSON: you need exactly one character lookahead
1285 // to parse anything.
923 switch (*dec->cur) 1286 switch (*dec->cur)
924 { 1287 {
925 case '"': ++dec->cur; return decode_str (dec); 1288 case '"': ++dec->cur; return decode_str (dec);
926 case '[': ++dec->cur; return decode_av (dec); 1289 case '[': ++dec->cur; return decode_av (dec);
927 case '{': ++dec->cur; return decode_hv (dec); 1290 case '{': ++dec->cur; return decode_hv (dec);
… …
933 1296
934 case 't': 1297 case 't':
935 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1298 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
936 { 1299 {
937 dec->cur += 4; 1300 dec->cur += 4;
938 return newSViv (1); 1301#if JSON_SLOW
1302 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1303#endif
1304 return SvREFCNT_inc (json_true);
939 } 1305 }
940 else 1306 else
941 ERR ("'true' expected"); 1307 ERR ("'true' expected");
942 1308
943 break; 1309 break;
944 1310
945 case 'f': 1311 case 'f':
946 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1312 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
947 { 1313 {
948 dec->cur += 5; 1314 dec->cur += 5;
949 return newSViv (0); 1315#if JSON_SLOW
1316 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1317#endif
1318 return SvREFCNT_inc (json_false);
950 } 1319 }
951 else 1320 else
952 ERR ("'false' expected"); 1321 ERR ("'false' expected");
953 1322
954 break; 1323 break;
… …
972fail: 1341fail:
973 return 0; 1342 return 0;
974} 1343}
975 1344
976static SV * 1345static SV *
977decode_json (SV *string, U32 flags) 1346decode_json (SV *string, JSON *json, UV *offset_return)
978{ 1347{
979 dec_t dec; 1348 dec_t dec;
1349 UV offset;
980 SV *sv; 1350 SV *sv;
981 1351
1352 SvGETMAGIC (string);
982 SvUPGRADE (string, SVt_PV); 1353 SvUPGRADE (string, SVt_PV);
983 1354
1355 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1356 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1357 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1358
984 if (flags & F_UTF8) 1359 if (json->flags & F_UTF8)
985 sv_utf8_downgrade (string, 0); 1360 sv_utf8_downgrade (string, 0);
986 else 1361 else
987 sv_utf8_upgrade (string); 1362 sv_utf8_upgrade (string);
988 1363
989 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1364 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
990 1365
991 dec.flags = flags; 1366 dec.json = *json;
992 dec.cur = SvPVX (string); 1367 dec.cur = SvPVX (string);
993 dec.end = SvEND (string); 1368 dec.end = SvEND (string);
994 dec.err = 0; 1369 dec.err = 0;
995 dec.depth = 0; 1370 dec.depth = 0;
996 dec.maxdepth = DEC_DEPTH (dec.flags); 1371 dec.maxdepth = DEC_DEPTH (dec.json.flags);
997 1372
1373 if (dec.json.cb_object || dec.json.cb_sk_object)
1374 dec.json.flags |= F_HOOK;
1375
998 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1376 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1377
1378 decode_ws (&dec);
999 sv = decode_sv (&dec); 1379 sv = decode_sv (&dec);
1000 1380
1381 if (!(offset_return || !sv))
1382 {
1383 // check for trailing garbage
1384 decode_ws (&dec);
1385
1386 if (*dec.cur)
1387 {
1388 dec.err = "garbage after JSON object";
1389 SvREFCNT_dec (sv);
1390 sv = 0;
1391 }
1392 }
1393
1394 if (offset_return || !sv)
1395 {
1396 offset = dec.json.flags & F_UTF8
1397 ? dec.cur - SvPVX (string)
1398 : utf8_distance (dec.cur, SvPVX (string));
1399
1400 if (offset_return)
1401 *offset_return = offset;
1402 }
1403
1001 if (!sv) 1404 if (!sv)
1002 { 1405 {
1003 IV offset = dec.flags & F_UTF8
1004 ? dec.cur - SvPVX (string)
1005 : utf8_distance (dec.cur, SvPVX (string));
1006 SV *uni = sv_newmortal (); 1406 SV *uni = sv_newmortal ();
1007 1407
1008 // horrible hack to silence warning inside pv_uni_display 1408 // horrible hack to silence warning inside pv_uni_display
1009 COP cop = *PL_curcop; 1409 COP cop = *PL_curcop;
1010 cop.cop_warnings = pWARN_NONE; 1410 cop.cop_warnings = pWARN_NONE;
… …
1020 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1420 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1021 } 1421 }
1022 1422
1023 sv = sv_2mortal (sv); 1423 sv = sv_2mortal (sv);
1024 1424
1025 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1425 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1026 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1426 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1027 1427
1028 return sv; 1428 return sv;
1029} 1429}
1030 1430
… …
1034MODULE = JSON::XS PACKAGE = JSON::XS 1434MODULE = JSON::XS PACKAGE = JSON::XS
1035 1435
1036BOOT: 1436BOOT:
1037{ 1437{
1038 int i; 1438 int i;
1039
1040 memset (decode_hexdigit, 0xff, 256);
1041 1439
1042 for (i = 0; i < 256; ++i) 1440 for (i = 0; i < 256; ++i)
1043 decode_hexdigit [i] = 1441 decode_hexdigit [i] =
1044 i >= '0' && i <= '9' ? i - '0' 1442 i >= '0' && i <= '9' ? i - '0'
1045 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1443 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1046 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1444 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1047 : -1; 1445 : -1;
1048 1446
1049 json_stash = gv_stashpv ("JSON::XS", 1); 1447 json_stash = gv_stashpv ("JSON::XS" , 1);
1448 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1449
1450 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1451 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1050} 1452}
1051 1453
1052PROTOTYPES: DISABLE 1454PROTOTYPES: DISABLE
1053 1455
1054SV *new (char *dummy) 1456void CLONE (...)
1055 CODE: 1457 CODE:
1056 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1458 json_stash = 0;
1459 json_boolean_stash = 0;
1460
1461void new (char *klass)
1462 PPCODE:
1463{
1464 SV *pv = NEWSV (0, sizeof (JSON));
1465 SvPOK_only (pv);
1466 Zero (SvPVX (pv), 1, JSON);
1467 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1468 XPUSHs (sv_2mortal (sv_bless (
1469 newRV_noinc (pv),
1470 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1471 )));
1472}
1473
1474void ascii (JSON *self, int enable = 1)
1475 ALIAS:
1476 ascii = F_ASCII
1477 latin1 = F_LATIN1
1478 utf8 = F_UTF8
1479 indent = F_INDENT
1480 canonical = F_CANONICAL
1481 space_before = F_SPACE_BEFORE
1482 space_after = F_SPACE_AFTER
1483 pretty = F_PRETTY
1484 allow_nonref = F_ALLOW_NONREF
1485 shrink = F_SHRINK
1486 allow_blessed = F_ALLOW_BLESSED
1487 convert_blessed = F_CONV_BLESSED
1488 relaxed = F_RELAXED
1489 PPCODE:
1490{
1491 if (enable)
1492 self->flags |= ix;
1493 else
1494 self->flags &= ~ix;
1495
1496 XPUSHs (ST (0));
1497}
1498
1499void get_ascii (JSON *self)
1500 ALIAS:
1501 get_ascii = F_ASCII
1502 get_latin1 = F_LATIN1
1503 get_utf8 = F_UTF8
1504 get_indent = F_INDENT
1505 get_canonical = F_CANONICAL
1506 get_space_before = F_SPACE_BEFORE
1507 get_space_after = F_SPACE_AFTER
1508 get_allow_nonref = F_ALLOW_NONREF
1509 get_shrink = F_SHRINK
1510 get_allow_blessed = F_ALLOW_BLESSED
1511 get_convert_blessed = F_CONV_BLESSED
1512 get_relaxed = F_RELAXED
1513 PPCODE:
1514 XPUSHs (boolSV (self->flags & ix));
1515
1516void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1517 PPCODE:
1518{
1519 UV log2 = 0;
1520
1521 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1522
1523 while ((1UL << log2) < max_depth)
1524 ++log2;
1525
1526 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1527
1528 XPUSHs (ST (0));
1529}
1530
1531U32 get_max_depth (JSON *self)
1532 CODE:
1533 RETVAL = DEC_DEPTH (self->flags);
1057 OUTPUT: 1534 OUTPUT:
1058 RETVAL 1535 RETVAL
1059 1536
1060SV *ascii (SV *self, int enable = 1) 1537void max_size (JSON *self, UV max_size = 0)
1061 ALIAS: 1538 PPCODE:
1062 ascii = F_ASCII 1539{
1063 utf8 = F_UTF8 1540 UV log2 = 0;
1064 indent = F_INDENT 1541
1065 canonical = F_CANONICAL 1542 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1066 space_before = F_SPACE_BEFORE 1543 if (max_size == 1) max_size = 2;
1067 space_after = F_SPACE_AFTER 1544
1068 pretty = F_PRETTY 1545 while ((1UL << log2) < max_size)
1069 allow_nonref = F_ALLOW_NONREF 1546 ++log2;
1070 shrink = F_SHRINK 1547
1548 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1549
1550 XPUSHs (ST (0));
1551}
1552
1553int get_max_size (JSON *self)
1071 CODE: 1554 CODE:
1072{ 1555 RETVAL = DEC_SIZE (self->flags);
1073 UV *uv = SvJSON (self);
1074 if (enable)
1075 *uv |= ix;
1076 else
1077 *uv &= ~ix;
1078
1079 RETVAL = newSVsv (self);
1080}
1081 OUTPUT: 1556 OUTPUT:
1082 RETVAL 1557 RETVAL
1083 1558
1084SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1559void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1560 PPCODE:
1561{
1562 SvREFCNT_dec (self->cb_object);
1563 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1564
1565 XPUSHs (ST (0));
1566}
1567
1568void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1569 PPCODE:
1570{
1571 if (!self->cb_sk_object)
1572 self->cb_sk_object = newHV ();
1573
1574 if (SvOK (cb))
1575 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1576 else
1577 {
1578 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1579
1580 if (!HvKEYS (self->cb_sk_object))
1581 {
1582 SvREFCNT_dec (self->cb_sk_object);
1583 self->cb_sk_object = 0;
1584 }
1585 }
1586
1587 XPUSHs (ST (0));
1588}
1589
1590void encode (JSON *self, SV *scalar)
1591 PPCODE:
1592 XPUSHs (encode_json (scalar, self));
1593
1594void decode (JSON *self, SV *jsonstr)
1595 PPCODE:
1596 XPUSHs (decode_json (jsonstr, self, 0));
1597
1598void decode_prefix (JSON *self, SV *jsonstr)
1599 PPCODE:
1600{
1601 UV offset;
1602 EXTEND (SP, 2);
1603 PUSHs (decode_json (jsonstr, self, &offset));
1604 PUSHs (sv_2mortal (newSVuv (offset)));
1605}
1606
1607void DESTROY (JSON *self)
1085 CODE: 1608 CODE:
1086{ 1609 SvREFCNT_dec (self->cb_sk_object);
1087 UV *uv = SvJSON (self); 1610 SvREFCNT_dec (self->cb_object);
1088 UV log2 = 0;
1089 1611
1090 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1612PROTOTYPES: ENABLE
1091 1613
1092 while ((1UL << log2) < max_depth)
1093 ++log2;
1094
1095 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1096
1097 RETVAL = newSVsv (self);
1098}
1099 OUTPUT:
1100 RETVAL
1101
1102void encode (SV *self, SV *scalar) 1614void encode_json (SV *scalar)
1103 PPCODE: 1615 PPCODE:
1616{
1617 JSON json = { F_DEFAULT | F_UTF8 };
1104 XPUSHs (encode_json (scalar, *SvJSON (self))); 1618 XPUSHs (encode_json (scalar, &json));
1619}
1105 1620
1106void decode (SV *self, SV *jsonstr) 1621void decode_json (SV *jsonstr)
1107 PPCODE: 1622 PPCODE:
1623{
1624 JSON json = { F_DEFAULT | F_UTF8 };
1108 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1625 XPUSHs (decode_json (jsonstr, &json, 0));
1626}
1109 1627
1110PROTOTYPES: ENABLE
1111
1112void to_json (SV *scalar)
1113 ALIAS:
1114 objToJson = 0
1115 PPCODE:
1116 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1117
1118void from_json (SV *jsonstr)
1119 ALIAS:
1120 jsonToObj = 0
1121 PPCODE:
1122 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1123

Diff Legend

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