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 1.23 by root, Tue Apr 3 23:34:17 2007 UTC vs.
Revision 1.64 by root, Mon Aug 27 02:03:23 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include <stdio.h>
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 (12UL << 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 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
99} enc_t; 133} enc_t;
100 134
101static void 135inline void
102need (enc_t *enc, STRLEN len) 136need (enc_t *enc, STRLEN len)
103{ 137{
104 if (enc->cur + len >= enc->end) 138 if (expect_false (enc->cur + len >= enc->end))
105 { 139 {
106 STRLEN cur = enc->cur - SvPVX (enc->sv); 140 STRLEN cur = enc->cur - SvPVX (enc->sv);
107 SvGROW (enc->sv, cur + len + 1); 141 SvGROW (enc->sv, cur + len + 1);
108 enc->cur = SvPVX (enc->sv) + cur; 142 enc->cur = SvPVX (enc->sv) + cur;
109 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
110 } 144 }
111} 145}
112 146
113static void 147inline void
114encode_ch (enc_t *enc, char ch) 148encode_ch (enc_t *enc, char ch)
115{ 149{
116 need (enc, 1); 150 need (enc, 1);
117 *enc->cur++ = ch; 151 *enc->cur++ = ch;
118} 152}
… …
126 160
127 while (str < end) 161 while (str < end)
128 { 162 {
129 unsigned char ch = *(unsigned char *)str; 163 unsigned char ch = *(unsigned char *)str;
130 164
131 if (ch >= 0x20 && ch < 0x80) // most common case 165 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
132 { 166 {
133 if (ch == '"') // but with slow exceptions 167 if (expect_false (ch == '"')) // but with slow exceptions
134 { 168 {
135 need (enc, len += 1); 169 need (enc, len += 1);
136 *enc->cur++ = '\\'; 170 *enc->cur++ = '\\';
137 *enc->cur++ = '"'; 171 *enc->cur++ = '"';
138 } 172 }
139 else if (ch == '\\') 173 else if (expect_false (ch == '\\'))
140 { 174 {
141 need (enc, len += 1); 175 need (enc, len += 1);
142 *enc->cur++ = '\\'; 176 *enc->cur++ = '\\';
143 *enc->cur++ = '\\'; 177 *enc->cur++ = '\\';
144 } 178 }
… …
162 STRLEN clen; 196 STRLEN clen;
163 UV uch; 197 UV uch;
164 198
165 if (is_utf8) 199 if (is_utf8)
166 { 200 {
167 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
168 uch = decode_utf8 (str, end - str, &clen); 201 uch = decode_utf8 (str, end - str, &clen);
169 if (clen == (STRLEN)-1) 202 if (clen == (STRLEN)-1)
170 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);
171 } 204 }
172 else 205 else
… …
176 } 209 }
177 210
178 if (uch > 0x10FFFFUL) 211 if (uch > 0x10FFFFUL)
179 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);
180 213
181 if (uch < 0x80 || enc->flags & F_ASCII) 214 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
182 { 215 {
183 if (uch > 0xFFFFUL) 216 if (uch > 0xFFFFUL)
184 { 217 {
185 need (enc, len += 11); 218 need (enc, len += 11);
186 sprintf (enc->cur, "\\u%04x\\u%04x", 219 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
200 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 233 *enc->cur++ = hexdigit [(uch >> 0) & 15];
201 } 234 }
202 235
203 str += clen; 236 str += clen;
204 } 237 }
238 else if (enc->json.flags & F_LATIN1)
239 {
240 *enc->cur++ = uch;
241 str += clen;
242 }
205 else if (is_utf8) 243 else if (is_utf8)
206 { 244 {
207 need (enc, len += clen); 245 need (enc, len += clen);
208 do 246 do
209 { 247 {
… …
211 } 249 }
212 while (--clen); 250 while (--clen);
213 } 251 }
214 else 252 else
215 { 253 {
216 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
217 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 255 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
218 ++str; 256 ++str;
219 } 257 }
220 } 258 }
221 } 259 }
… …
223 261
224 --len; 262 --len;
225 } 263 }
226} 264}
227 265
228static void 266inline void
229encode_indent (enc_t *enc) 267encode_indent (enc_t *enc)
230{ 268{
231 if (enc->flags & F_INDENT) 269 if (enc->json.flags & F_INDENT)
232 { 270 {
233 int spaces = enc->indent * INDENT_STEP; 271 int spaces = enc->indent * INDENT_STEP;
234 272
235 need (enc, spaces); 273 need (enc, spaces);
236 memset (enc->cur, ' ', spaces); 274 memset (enc->cur, ' ', spaces);
237 enc->cur += spaces; 275 enc->cur += spaces;
238 } 276 }
239} 277}
240 278
241static void 279inline void
242encode_space (enc_t *enc) 280encode_space (enc_t *enc)
243{ 281{
244 need (enc, 1); 282 need (enc, 1);
245 encode_ch (enc, ' '); 283 encode_ch (enc, ' ');
246} 284}
247 285
248static void 286inline void
249encode_nl (enc_t *enc) 287encode_nl (enc_t *enc)
250{ 288{
251 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
252 { 290 {
253 need (enc, 1); 291 need (enc, 1);
254 encode_ch (enc, '\n'); 292 encode_ch (enc, '\n');
255 } 293 }
256} 294}
257 295
258static void 296inline void
259encode_comma (enc_t *enc) 297encode_comma (enc_t *enc)
260{ 298{
261 encode_ch (enc, ','); 299 encode_ch (enc, ',');
262 300
263 if (enc->flags & F_INDENT) 301 if (enc->json.flags & F_INDENT)
264 encode_nl (enc); 302 encode_nl (enc);
265 else if (enc->flags & F_SPACE_AFTER) 303 else if (enc->json.flags & F_SPACE_AFTER)
266 encode_space (enc); 304 encode_space (enc);
267} 305}
268 306
269static void encode_sv (enc_t *enc, SV *sv); 307static void encode_sv (enc_t *enc, SV *sv);
270 308
… …
279 encode_ch (enc, '['); encode_nl (enc); 317 encode_ch (enc, '['); encode_nl (enc);
280 ++enc->indent; 318 ++enc->indent;
281 319
282 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
283 { 321 {
322 SV **svp = av_fetch (av, i, 0);
323
284 encode_indent (enc); 324 encode_indent (enc);
285 encode_sv (enc, *av_fetch (av, i, 0)); 325
326 if (svp)
327 encode_sv (enc, *svp);
328 else
329 encode_str (enc, "null", 4, 0);
286 330
287 if (i < len) 331 if (i < len)
288 encode_comma (enc); 332 encode_comma (enc);
289 } 333 }
290 334
… …
293 --enc->indent; 337 --enc->indent;
294 encode_indent (enc); encode_ch (enc, ']'); 338 encode_indent (enc); encode_ch (enc, ']');
295} 339}
296 340
297static void 341static void
298encode_he (enc_t *enc, HE *he) 342encode_hk (enc_t *enc, HE *he)
299{ 343{
300 encode_ch (enc, '"'); 344 encode_ch (enc, '"');
301 345
302 if (HeKLEN (he) == HEf_SVKEY) 346 if (HeKLEN (he) == HEf_SVKEY)
303 { 347 {
… …
313 else 357 else
314 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 358 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
315 359
316 encode_ch (enc, '"'); 360 encode_ch (enc, '"');
317 361
318 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 362 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
319 encode_ch (enc, ':'); 363 encode_ch (enc, ':');
320 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 364 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
321 encode_sv (enc, HeVAL (he));
322} 365}
323 366
324// compare hash entries, used when all keys are bytestrings 367// compare hash entries, used when all keys are bytestrings
325static int 368static int
326he_cmp_fast (const void *a_, const void *b_) 369he_cmp_fast (const void *a_, const void *b_)
… …
331 HE *b = *(HE **)b_; 374 HE *b = *(HE **)b_;
332 375
333 STRLEN la = HeKLEN (a); 376 STRLEN la = HeKLEN (a);
334 STRLEN lb = HeKLEN (b); 377 STRLEN lb = HeKLEN (b);
335 378
336 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 379 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
337 cmp = la - lb; 380 cmp = lb - la;
338 381
339 return cmp; 382 return cmp;
340} 383}
341 384
342// compare hash entries, used when some keys are sv's or utf-x 385// compare hash entries, used when some keys are sv's or utf-x
343static int 386static int
344he_cmp_slow (const void *a, const void *b) 387he_cmp_slow (const void *a, const void *b)
345{ 388{
346 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 389 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
347} 390}
348 391
349static void 392static void
350encode_hv (enc_t *enc, HV *hv) 393encode_hv (enc_t *enc, HV *hv)
351{ 394{
395 HE *he;
352 int count, i; 396 int count;
353 397
354 if (enc->indent >= enc->maxdepth) 398 if (enc->indent >= enc->maxdepth)
355 croak ("data structure too deep (hit recursion limit)"); 399 croak ("data structure too deep (hit recursion limit)");
356 400
357 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 401 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
358 402
359 if ((count = hv_iterinit (hv)))
360 {
361 // for canonical output we have to sort by keys first 403 // for canonical output we have to sort by keys first
362 // actually, this is mostly due to the stupid so-called 404 // actually, this is mostly due to the stupid so-called
363 // security workaround added somewhere in 5.8.x. 405 // security workaround added somewhere in 5.8.x.
364 // that randomises hash orderings 406 // that randomises hash orderings
365 if (enc->flags & F_CANONICAL) 407 if (enc->json.flags & F_CANONICAL)
408 {
409 int count = hv_iterinit (hv);
410
411 if (SvMAGICAL (hv))
366 { 412 {
413 // need to count by iterating. could improve by dynamically building the vector below
414 // but I don't care for the speed of this special case.
415 // note also that we will run into undefined behaviour when the two iterations
416 // do not result in the same count, something I might care for in some later release.
417
418 count = 0;
419 while (hv_iternext (hv))
420 ++count;
421
422 hv_iterinit (hv);
423 }
424
425 if (count)
426 {
427 int i, fast = 1;
428#if defined(__BORLANDC__) || defined(_MSC_VER)
429 HE **hes = _alloca (count * sizeof (HE));
430#else
367 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 431 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
368 int fast = 1; 432#endif
369 433
370 i = 0; 434 i = 0;
371 while ((he = hv_iternext (hv))) 435 while ((he = hv_iternext (hv)))
372 { 436 {
373 hes [i++] = he; 437 hes [i++] = he;
… …
395 459
396 FREETMPS; 460 FREETMPS;
397 LEAVE; 461 LEAVE;
398 } 462 }
399 463
400 for (i = 0; i < count; ++i) 464 while (count--)
401 { 465 {
402 encode_indent (enc); 466 encode_indent (enc);
467 he = hes [count];
403 encode_he (enc, hes [i]); 468 encode_hk (enc, he);
469 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
404 470
405 if (i < count - 1) 471 if (count)
406 encode_comma (enc); 472 encode_comma (enc);
407 } 473 }
408 474 }
475 }
476 else
477 {
478 if (hv_iterinit (hv) || SvMAGICAL (hv))
479 if ((he = hv_iternext (hv)))
480 for (;;)
481 {
409 encode_nl (enc); 482 encode_indent (enc);
483 encode_hk (enc, he);
484 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
485
486 if (!(he = hv_iternext (hv)))
487 break;
488
489 encode_comma (enc);
490 }
491 }
492
493 encode_nl (enc);
494
495 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
496}
497
498// encode objects, arrays and special \0=false and \1=true values.
499static void
500encode_rv (enc_t *enc, SV *sv)
501{
502 svtype svt;
503
504 SvGETMAGIC (sv);
505 svt = SvTYPE (sv);
506
507 if (expect_false (SvOBJECT (sv)))
508 {
509 HV *stash = !JSON_SLOW || json_boolean_stash
510 ? json_boolean_stash
511 : gv_stashpv ("JSON::XS::Boolean", 1);
512
513 if (SvSTASH (sv) == stash)
514 {
515 if (SvIV (sv))
516 encode_str (enc, "true", 4, 0);
517 else
518 encode_str (enc, "false", 5, 0);
410 } 519 }
411 else 520 else
412 { 521 {
413 SV *sv; 522#if 0
414 HE *he = hv_iternext (hv); 523 if (0 && sv_derived_from (rv, "JSON::Literal"))
415
416 for (;;)
417 {
418 encode_indent (enc);
419 encode_he (enc, he);
420
421 if (!(he = hv_iternext (hv)))
422 break;
423
424 encode_comma (enc);
425 } 524 {
525 // not yet
526 }
527#endif
528 if (enc->json.flags & F_CONV_BLESSED)
529 {
530 // we re-bless the reference to get overload and other niceties right
531 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
426 532
533 if (to_json)
534 {
535 dSP;
536
537 ENTER; SAVETMPS; PUSHMARK (SP);
538 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
539
540 // calling with G_SCALAR ensures that we always get a 1 return value
541 PUTBACK;
542 call_sv ((SV *)GvCV (to_json), G_SCALAR);
543 SPAGAIN;
544
545 // catch this surprisingly common error
546 if (SvROK (TOPs) && SvRV (TOPs) == sv)
547 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
548
549 sv = POPs;
550 PUTBACK;
551
427 encode_nl (enc); 552 encode_sv (enc, sv);
553
554 FREETMPS; LEAVE;
555 }
556 else if (enc->json.flags & F_ALLOW_BLESSED)
557 encode_str (enc, "null", 4, 0);
558 else
559 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
560 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
561 }
562 else if (enc->json.flags & F_ALLOW_BLESSED)
563 encode_str (enc, "null", 4, 0);
564 else
565 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
566 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
428 } 567 }
429 } 568 }
430
431 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
432}
433
434// encode objects, arrays and special \0=false and \1=true values.
435static void
436encode_rv (enc_t *enc, SV *sv)
437{
438 SvGETMAGIC (sv);
439
440 svtype svt = SvTYPE (sv);
441
442 if (svt == SVt_PVHV) 569 else if (svt == SVt_PVHV)
443 encode_hv (enc, (HV *)sv); 570 encode_hv (enc, (HV *)sv);
444 else if (svt == SVt_PVAV) 571 else if (svt == SVt_PVAV)
445 encode_av (enc, (AV *)sv); 572 encode_av (enc, (AV *)sv);
446 else if (svt < SVt_PVAV) 573 else if (svt < SVt_PVAV)
447 { 574 {
448 if (SvNIOK (sv) && SvIV (sv) == 0) 575 STRLEN len = 0;
576 char *pv = svt ? SvPV (sv, len) : 0;
577
578 if (len == 1 && *pv == '1')
579 encode_str (enc, "true", 4, 0);
580 else if (len == 1 && *pv == '0')
449 encode_str (enc, "false", 5, 0); 581 encode_str (enc, "false", 5, 0);
450 else if (SvNIOK (sv) && SvIV (sv) == 1)
451 encode_str (enc, "true", 4, 0);
452 else 582 else
453 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 583 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
454 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 584 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
455 } 585 }
456 else 586 else
… …
471 encode_str (enc, str, len, SvUTF8 (sv)); 601 encode_str (enc, str, len, SvUTF8 (sv));
472 encode_ch (enc, '"'); 602 encode_ch (enc, '"');
473 } 603 }
474 else if (SvNOKp (sv)) 604 else if (SvNOKp (sv))
475 { 605 {
606 // trust that perl will do the right thing w.r.t. JSON syntax.
476 need (enc, NV_DIG + 32); 607 need (enc, NV_DIG + 32);
477 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 608 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
478 enc->cur += strlen (enc->cur); 609 enc->cur += strlen (enc->cur);
479 } 610 }
480 else if (SvIOKp (sv)) 611 else if (SvIOKp (sv))
481 { 612 {
482 need (enc, 64); 613 // we assume we can always read an IV as a UV
614 if (SvUV (sv) & ~(UV)0x7fff)
615 {
616 // large integer, use the (rather slow) snprintf way.
617 need (enc, sizeof (UV) * 3);
483 enc->cur += 618 enc->cur +=
484 SvIsUV(sv) 619 SvIsUV(sv)
485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 620 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 621 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
622 }
623 else
624 {
625 // optimise the "small number case"
626 // code will likely be branchless and use only a single multiplication
627 I32 i = SvIV (sv);
628 U32 u;
629 char digit, nz = 0;
630
631 need (enc, 6);
632
633 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
634 u = i < 0 ? -i : i;
635
636 // convert to 4.28 fixed-point representation
637 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
638
639 // now output digit by digit, each time masking out the integer part
640 // and multiplying by 5 while moving the decimal point one to the right,
641 // resulting in a net multiplication by 10.
642 // we always write the digit to memory but conditionally increment
643 // the pointer, to ease the usage of conditional move instructions.
644 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
645 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
646 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
647 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
648 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
649 }
487 } 650 }
488 else if (SvROK (sv)) 651 else if (SvROK (sv))
489 encode_rv (enc, SvRV (sv)); 652 encode_rv (enc, SvRV (sv));
490 else if (!SvOK (sv)) 653 else if (!SvOK (sv))
491 encode_str (enc, "null", 4, 0); 654 encode_str (enc, "null", 4, 0);
… …
493 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 656 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
494 SvPV_nolen (sv), SvFLAGS (sv)); 657 SvPV_nolen (sv), SvFLAGS (sv));
495} 658}
496 659
497static SV * 660static SV *
498encode_json (SV *scalar, U32 flags) 661encode_json (SV *scalar, JSON *json)
499{ 662{
663 enc_t enc;
664
500 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 665 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
501 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 666 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
502 667
503 enc_t enc; 668 enc.json = *json;
504 enc.flags = flags;
505 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 669 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
506 enc.cur = SvPVX (enc.sv); 670 enc.cur = SvPVX (enc.sv);
507 enc.end = SvEND (enc.sv); 671 enc.end = SvEND (enc.sv);
508 enc.indent = 0; 672 enc.indent = 0;
509 enc.maxdepth = DEC_DEPTH (flags); 673 enc.maxdepth = DEC_DEPTH (enc.json.flags);
510 674
511 SvPOK_only (enc.sv); 675 SvPOK_only (enc.sv);
512 encode_sv (&enc, scalar); 676 encode_sv (&enc, scalar);
513 677
678 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
679 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
680
514 if (!(flags & (F_ASCII | F_UTF8))) 681 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
515 SvUTF8_on (enc.sv); 682 SvUTF8_on (enc.sv);
516 683
517 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
518
519 if (enc.flags & F_SHRINK) 684 if (enc.json.flags & F_SHRINK)
520 shrink (enc.sv); 685 shrink (enc.sv);
521 686
522 return enc.sv; 687 return enc.sv;
523} 688}
524 689
… …
529typedef struct 694typedef struct
530{ 695{
531 char *cur; // current parser pointer 696 char *cur; // current parser pointer
532 char *end; // end of input string 697 char *end; // end of input string
533 const char *err; // parse error, if != 0 698 const char *err; // parse error, if != 0
534 U32 flags; // F_* 699 JSON json;
535 U32 depth; // recursion depth 700 U32 depth; // recursion depth
536 U32 maxdepth; // recursion depth limit 701 U32 maxdepth; // recursion depth limit
537} dec_t; 702} dec_t;
538 703
539static void 704inline void
705decode_comment (dec_t *dec)
706{
707 // only '#'-style comments allowed a.t.m.
708
709 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
710 ++dec->cur;
711}
712
713inline void
540decode_ws (dec_t *dec) 714decode_ws (dec_t *dec)
541{ 715{
542 for (;;) 716 for (;;)
543 { 717 {
544 char ch = *dec->cur; 718 char ch = *dec->cur;
545 719
546 if (ch > 0x20 720 if (ch > 0x20)
721 {
722 if (expect_false (ch == '#'))
723 {
724 if (dec->json.flags & F_RELAXED)
725 decode_comment (dec);
726 else
727 break;
728 }
729 else
730 break;
731 }
547 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 732 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
548 break; 733 break; // parse error, but let higher level handle it, gives better error messages
549 734
550 ++dec->cur; 735 ++dec->cur;
551 } 736 }
552} 737}
553 738
… …
570decode_4hex (dec_t *dec) 755decode_4hex (dec_t *dec)
571{ 756{
572 signed char d1, d2, d3, d4; 757 signed char d1, d2, d3, d4;
573 unsigned char *cur = (unsigned char *)dec->cur; 758 unsigned char *cur = (unsigned char *)dec->cur;
574 759
575 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 760 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
576 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 761 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
577 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 762 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
578 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 763 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
579 764
580 dec->cur += 4; 765 dec->cur += 4;
581 766
582 return ((UV)d1) << 12 767 return ((UV)d1) << 12
583 | ((UV)d2) << 8 768 | ((UV)d2) << 8
… …
591static SV * 776static SV *
592decode_str (dec_t *dec) 777decode_str (dec_t *dec)
593{ 778{
594 SV *sv = 0; 779 SV *sv = 0;
595 int utf8 = 0; 780 int utf8 = 0;
781 char *dec_cur = dec->cur;
596 782
597 do 783 do
598 { 784 {
599 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 785 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
600 char *cur = buf; 786 char *cur = buf;
601 787
602 do 788 do
603 { 789 {
604 unsigned char ch = *(unsigned char *)dec->cur++; 790 unsigned char ch = *(unsigned char *)dec_cur++;
605 791
606 if (ch == '"') 792 if (expect_false (ch == '"'))
607 { 793 {
608 --dec->cur; 794 --dec_cur;
609 break; 795 break;
610 } 796 }
611 else if (ch == '\\') 797 else if (expect_false (ch == '\\'))
612 { 798 {
613 switch (*dec->cur) 799 switch (*dec_cur)
614 { 800 {
615 case '\\': 801 case '\\':
616 case '/': 802 case '/':
617 case '"': *cur++ = *dec->cur++; break; 803 case '"': *cur++ = *dec_cur++; break;
618 804
619 case 'b': ++dec->cur; *cur++ = '\010'; break; 805 case 'b': ++dec_cur; *cur++ = '\010'; break;
620 case 't': ++dec->cur; *cur++ = '\011'; break; 806 case 't': ++dec_cur; *cur++ = '\011'; break;
621 case 'n': ++dec->cur; *cur++ = '\012'; break; 807 case 'n': ++dec_cur; *cur++ = '\012'; break;
622 case 'f': ++dec->cur; *cur++ = '\014'; break; 808 case 'f': ++dec_cur; *cur++ = '\014'; break;
623 case 'r': ++dec->cur; *cur++ = '\015'; break; 809 case 'r': ++dec_cur; *cur++ = '\015'; break;
624 810
625 case 'u': 811 case 'u':
626 { 812 {
627 UV lo, hi; 813 UV lo, hi;
628 ++dec->cur; 814 ++dec_cur;
629 815
816 dec->cur = dec_cur;
630 hi = decode_4hex (dec); 817 hi = decode_4hex (dec);
818 dec_cur = dec->cur;
631 if (hi == (UV)-1) 819 if (hi == (UV)-1)
632 goto fail; 820 goto fail;
633 821
634 // possibly a surrogate pair 822 // possibly a surrogate pair
635 if (hi >= 0xd800) 823 if (hi >= 0xd800)
636 if (hi < 0xdc00) 824 if (hi < 0xdc00)
637 { 825 {
638 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 826 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
639 ERR ("missing low surrogate character in surrogate pair"); 827 ERR ("missing low surrogate character in surrogate pair");
640 828
641 dec->cur += 2; 829 dec_cur += 2;
642 830
831 dec->cur = dec_cur;
643 lo = decode_4hex (dec); 832 lo = decode_4hex (dec);
833 dec_cur = dec->cur;
644 if (lo == (UV)-1) 834 if (lo == (UV)-1)
645 goto fail; 835 goto fail;
646 836
647 if (lo < 0xdc00 || lo >= 0xe000) 837 if (lo < 0xdc00 || lo >= 0xe000)
648 ERR ("surrogate pair expected"); 838 ERR ("surrogate pair expected");
… …
662 *cur++ = hi; 852 *cur++ = hi;
663 } 853 }
664 break; 854 break;
665 855
666 default: 856 default:
667 --dec->cur; 857 --dec_cur;
668 ERR ("illegal backslash escape sequence in string"); 858 ERR ("illegal backslash escape sequence in string");
669 } 859 }
670 } 860 }
671 else if (ch >= 0x20 && ch <= 0x7f) 861 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
672 *cur++ = ch; 862 *cur++ = ch;
673 else if (ch >= 0x80) 863 else if (ch >= 0x80)
674 { 864 {
675 --dec->cur;
676
677 STRLEN clen; 865 STRLEN clen;
866 UV uch;
867
868 --dec_cur;
869
678 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 870 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
679 if (clen == (STRLEN)-1) 871 if (clen == (STRLEN)-1)
680 ERR ("malformed UTF-8 character in JSON string"); 872 ERR ("malformed UTF-8 character in JSON string");
681 873
682 do 874 do
683 *cur++ = *dec->cur++; 875 *cur++ = *dec_cur++;
684 while (--clen); 876 while (--clen);
685 877
686 utf8 = 1; 878 utf8 = 1;
687 } 879 }
688 else 880 else
689 { 881 {
690 --dec->cur; 882 --dec_cur;
691 883
692 if (!ch) 884 if (!ch)
693 ERR ("unexpected end of string while parsing json string"); 885 ERR ("unexpected end of string while parsing JSON string");
694 else 886 else
695 ERR ("invalid character encountered while parsing json string"); 887 ERR ("invalid character encountered while parsing JSON string");
696 } 888 }
697 } 889 }
698 while (cur < buf + SHORT_STRING_LEN); 890 while (cur < buf + SHORT_STRING_LEN);
699 891
892 {
700 STRLEN len = cur - buf; 893 STRLEN len = cur - buf;
701 894
702 if (sv) 895 if (sv)
703 { 896 {
704 SvGROW (sv, SvCUR (sv) + len + 1); 897 SvGROW (sv, SvCUR (sv) + len + 1);
705 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 898 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
706 SvCUR_set (sv, SvCUR (sv) + len); 899 SvCUR_set (sv, SvCUR (sv) + len);
707 } 900 }
708 else 901 else
709 sv = newSVpvn (buf, len); 902 sv = newSVpvn (buf, len);
710 } 903 }
904 }
711 while (*dec->cur != '"'); 905 while (*dec_cur != '"');
712 906
713 ++dec->cur; 907 ++dec_cur;
714 908
715 if (sv) 909 if (sv)
716 { 910 {
717 SvPOK_only (sv); 911 SvPOK_only (sv);
718 *SvEND (sv) = 0; 912 *SvEND (sv) = 0;
… …
721 SvUTF8_on (sv); 915 SvUTF8_on (sv);
722 } 916 }
723 else 917 else
724 sv = newSVpvn ("", 0); 918 sv = newSVpvn ("", 0);
725 919
920 dec->cur = dec_cur;
726 return sv; 921 return sv;
727 922
728fail: 923fail:
924 dec->cur = dec_cur;
729 return 0; 925 return 0;
730} 926}
731 927
732static SV * 928static SV *
733decode_num (dec_t *dec) 929decode_num (dec_t *dec)
… …
791 is_nv = 1; 987 is_nv = 1;
792 } 988 }
793 989
794 if (!is_nv) 990 if (!is_nv)
795 { 991 {
796 UV uv; 992 int len = dec->cur - start;
797 int numtype = grok_number (start, dec->cur - start, &uv); 993
798 if (numtype & IS_NUMBER_IN_UV) 994 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
799 if (numtype & IS_NUMBER_NEG) 995 if (*start == '-')
996 switch (len)
800 { 997 {
801 if (uv < (UV)IV_MIN) 998 case 2: return newSViv (-( start [1] - '0' * 1));
802 return newSViv (-(IV)uv); 999 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1000 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1001 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
803 } 1002 }
1003 else
1004 switch (len)
1005 {
1006 case 1: return newSViv ( start [0] - '0' * 1);
1007 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1008 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1009 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1010 }
1011
1012 {
1013 UV uv;
1014 int numtype = grok_number (start, len, &uv);
1015 if (numtype & IS_NUMBER_IN_UV)
1016 if (numtype & IS_NUMBER_NEG)
1017 {
1018 if (uv < (UV)IV_MIN)
1019 return newSViv (-(IV)uv);
1020 }
804 else 1021 else
805 return newSVuv (uv); 1022 return newSVuv (uv);
806 } 1023 }
807 1024
1025 len -= *start == '-' ? 1 : 0;
1026
1027 // does not fit into IV or UV, try NV
1028 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1029 #if defined (LDBL_DIG)
1030 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1031 #endif
1032 )
1033 // fits into NV without loss of precision
1034 return newSVnv (Atof (start));
1035
1036 // everything else fails, convert it to a string
1037 return newSVpvn (start, dec->cur - start);
1038 }
1039
1040 // loss of precision here
808 return newSVnv (Atof (start)); 1041 return newSVnv (Atof (start));
809 1042
810fail: 1043fail:
811 return 0; 1044 return 0;
812} 1045}
… …
842 1075
843 if (*dec->cur != ',') 1076 if (*dec->cur != ',')
844 ERR (", or ] expected while parsing array"); 1077 ERR (", or ] expected while parsing array");
845 1078
846 ++dec->cur; 1079 ++dec->cur;
1080
1081 decode_ws (dec);
1082
1083 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1084 {
1085 ++dec->cur;
1086 break;
1087 }
847 } 1088 }
848 1089
849 DEC_DEC_DEPTH; 1090 DEC_DEC_DEPTH;
850 return newRV_noinc ((SV *)av); 1091 return newRV_noinc ((SV *)av);
851 1092
… …
856} 1097}
857 1098
858static SV * 1099static SV *
859decode_hv (dec_t *dec) 1100decode_hv (dec_t *dec)
860{ 1101{
1102 SV *sv;
861 HV *hv = newHV (); 1103 HV *hv = newHV ();
862 1104
863 DEC_INC_DEPTH; 1105 DEC_INC_DEPTH;
864 decode_ws (dec); 1106 decode_ws (dec);
865 1107
866 if (*dec->cur == '}') 1108 if (*dec->cur == '}')
867 ++dec->cur; 1109 ++dec->cur;
868 else 1110 else
869 for (;;) 1111 for (;;)
870 { 1112 {
1113 EXPECT_CH ('"');
1114
1115 // heuristic: assume that
1116 // a) decode_str + hv_store_ent are abysmally slow.
1117 // b) most hash keys are short, simple ascii text.
1118 // => try to "fast-match" such strings to avoid
1119 // the overhead of decode_str + hv_store_ent.
1120 {
871 SV *key, *value; 1121 SV *value;
1122 char *p = dec->cur;
1123 char *e = p + 24; // only try up to 24 bytes
872 1124
873 decode_ws (dec); EXPECT_CH ('"'); 1125 for (;;)
874
875 key = decode_str (dec);
876 if (!key)
877 goto fail;
878
879 decode_ws (dec); EXPECT_CH (':');
880
881 value = decode_sv (dec);
882 if (!value)
883 { 1126 {
1127 // the >= 0x80 is true on most architectures
1128 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1129 {
1130 // slow path, back up and use decode_str
1131 SV *key = decode_str (dec);
1132 if (!key)
1133 goto fail;
1134
1135 decode_ws (dec); EXPECT_CH (':');
1136
1137 decode_ws (dec);
1138 value = decode_sv (dec);
1139 if (!value)
1140 {
1141 SvREFCNT_dec (key);
1142 goto fail;
1143 }
1144
1145 hv_store_ent (hv, key, value, 0);
884 SvREFCNT_dec (key); 1146 SvREFCNT_dec (key);
1147
1148 break;
1149 }
1150 else if (*p == '"')
1151 {
1152 // fast path, got a simple key
1153 char *key = dec->cur;
1154 int len = p - key;
1155 dec->cur = p + 1;
1156
1157 decode_ws (dec); EXPECT_CH (':');
1158
1159 decode_ws (dec);
1160 value = decode_sv (dec);
1161 if (!value)
885 goto fail; 1162 goto fail;
1163
1164 hv_store (hv, key, len, value, 0);
1165
1166 break;
1167 }
1168
1169 ++p;
886 } 1170 }
887 1171 }
888 hv_store_ent (hv, key, value, 0);
889 SvREFCNT_dec (key);
890 1172
891 decode_ws (dec); 1173 decode_ws (dec);
892 1174
893 if (*dec->cur == '}') 1175 if (*dec->cur == '}')
894 { 1176 {
… …
898 1180
899 if (*dec->cur != ',') 1181 if (*dec->cur != ',')
900 ERR (", or } expected while parsing object/hash"); 1182 ERR (", or } expected while parsing object/hash");
901 1183
902 ++dec->cur; 1184 ++dec->cur;
1185
1186 decode_ws (dec);
1187
1188 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1189 {
1190 ++dec->cur;
1191 break;
1192 }
903 } 1193 }
904 1194
905 DEC_DEC_DEPTH; 1195 DEC_DEC_DEPTH;
906 return newRV_noinc ((SV *)hv); 1196 sv = newRV_noinc ((SV *)hv);
1197
1198 // check filter callbacks
1199 if (dec->json.flags & F_HOOK)
1200 {
1201 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1202 {
1203 HE *cb, *he;
1204
1205 hv_iterinit (hv);
1206 he = hv_iternext (hv);
1207 hv_iterinit (hv);
1208
1209 // the next line creates a mortal sv each time its called.
1210 // might want to optimise this for common cases.
1211 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1212
1213 if (cb)
1214 {
1215 dSP;
1216 int count;
1217
1218 ENTER; SAVETMPS; PUSHMARK (SP);
1219 XPUSHs (HeVAL (he));
1220
1221 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1222
1223 if (count == 1)
1224 {
1225 sv = newSVsv (POPs);
1226 FREETMPS; LEAVE;
1227 return sv;
1228 }
1229
1230 FREETMPS; LEAVE;
1231 }
1232 }
1233
1234 if (dec->json.cb_object)
1235 {
1236 dSP;
1237 int count;
1238
1239 ENTER; SAVETMPS; PUSHMARK (SP);
1240 XPUSHs (sv_2mortal (sv));
1241
1242 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1243
1244 if (count == 1)
1245 {
1246 sv = newSVsv (POPs);
1247 FREETMPS; LEAVE;
1248 return sv;
1249 }
1250
1251 SvREFCNT_inc (sv);
1252 FREETMPS; LEAVE;
1253 }
1254 }
1255
1256 return sv;
907 1257
908fail: 1258fail:
909 SvREFCNT_dec (hv); 1259 SvREFCNT_dec (hv);
910 DEC_DEC_DEPTH; 1260 DEC_DEC_DEPTH;
911 return 0; 1261 return 0;
912} 1262}
913 1263
914static SV * 1264static SV *
915decode_sv (dec_t *dec) 1265decode_sv (dec_t *dec)
916{ 1266{
917 decode_ws (dec); 1267 // the beauty of JSON: you need exactly one character lookahead
1268 // to parse anything.
918 switch (*dec->cur) 1269 switch (*dec->cur)
919 { 1270 {
920 case '"': ++dec->cur; return decode_str (dec); 1271 case '"': ++dec->cur; return decode_str (dec);
921 case '[': ++dec->cur; return decode_av (dec); 1272 case '[': ++dec->cur; return decode_av (dec);
922 case '{': ++dec->cur; return decode_hv (dec); 1273 case '{': ++dec->cur; return decode_hv (dec);
… …
928 1279
929 case 't': 1280 case 't':
930 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1281 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
931 { 1282 {
932 dec->cur += 4; 1283 dec->cur += 4;
933 return newSViv (1); 1284#if JSON_SLOW
1285 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1286#endif
1287 return SvREFCNT_inc (json_true);
934 } 1288 }
935 else 1289 else
936 ERR ("'true' expected"); 1290 ERR ("'true' expected");
937 1291
938 break; 1292 break;
939 1293
940 case 'f': 1294 case 'f':
941 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1295 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
942 { 1296 {
943 dec->cur += 5; 1297 dec->cur += 5;
944 return newSViv (0); 1298#if JSON_SLOW
1299 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1300#endif
1301 return SvREFCNT_inc (json_false);
945 } 1302 }
946 else 1303 else
947 ERR ("'false' expected"); 1304 ERR ("'false' expected");
948 1305
949 break; 1306 break;
… …
958 ERR ("'null' expected"); 1315 ERR ("'null' expected");
959 1316
960 break; 1317 break;
961 1318
962 default: 1319 default:
963 ERR ("malformed json string, neither array, object, number, string or atom"); 1320 ERR ("malformed JSON string, neither array, object, number, string or atom");
964 break; 1321 break;
965 } 1322 }
966 1323
967fail: 1324fail:
968 return 0; 1325 return 0;
969} 1326}
970 1327
971static SV * 1328static SV *
972decode_json (SV *string, U32 flags) 1329decode_json (SV *string, JSON *json, UV *offset_return)
973{ 1330{
1331 dec_t dec;
1332 UV offset;
974 SV *sv; 1333 SV *sv;
975 1334
1335 SvGETMAGIC (string);
976 SvUPGRADE (string, SVt_PV); 1336 SvUPGRADE (string, SVt_PV);
977 1337
1338 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1339 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1340 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1341
978 if (flags & F_UTF8) 1342 if (json->flags & F_UTF8)
979 sv_utf8_downgrade (string, 0); 1343 sv_utf8_downgrade (string, 0);
980 else 1344 else
981 sv_utf8_upgrade (string); 1345 sv_utf8_upgrade (string);
982 1346
983 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1347 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
984 1348
985 dec_t dec; 1349 dec.json = *json;
986 dec.flags = flags;
987 dec.cur = SvPVX (string); 1350 dec.cur = SvPVX (string);
988 dec.end = SvEND (string); 1351 dec.end = SvEND (string);
989 dec.err = 0; 1352 dec.err = 0;
990 dec.depth = 0; 1353 dec.depth = 0;
991 dec.maxdepth = DEC_DEPTH (dec.flags); 1354 dec.maxdepth = DEC_DEPTH (dec.json.flags);
992 1355
1356 if (dec.json.cb_object || dec.json.cb_sk_object)
1357 dec.json.flags |= F_HOOK;
1358
993 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1359 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1360
1361 decode_ws (&dec);
994 sv = decode_sv (&dec); 1362 sv = decode_sv (&dec);
995 1363
1364 if (!(offset_return || !sv))
1365 {
1366 // check for trailing garbage
1367 decode_ws (&dec);
1368
1369 if (*dec.cur)
1370 {
1371 dec.err = "garbage after JSON object";
1372 SvREFCNT_dec (sv);
1373 sv = 0;
1374 }
1375 }
1376
1377 if (offset_return || !sv)
1378 {
1379 offset = dec.json.flags & F_UTF8
1380 ? dec.cur - SvPVX (string)
1381 : utf8_distance (dec.cur, SvPVX (string));
1382
1383 if (offset_return)
1384 *offset_return = offset;
1385 }
1386
996 if (!sv) 1387 if (!sv)
997 { 1388 {
998 IV offset = dec.flags & F_UTF8
999 ? dec.cur - SvPVX (string)
1000 : utf8_distance (dec.cur, SvPVX (string));
1001 SV *uni = sv_newmortal (); 1389 SV *uni = sv_newmortal ();
1002 1390
1003 // horrible hack to silence warning inside pv_uni_display 1391 // horrible hack to silence warning inside pv_uni_display
1004 COP cop = *PL_curcop; 1392 COP cop = *PL_curcop;
1005 cop.cop_warnings = pWARN_NONE; 1393 cop.cop_warnings = pWARN_NONE;
… …
1015 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1403 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1016 } 1404 }
1017 1405
1018 sv = sv_2mortal (sv); 1406 sv = sv_2mortal (sv);
1019 1407
1020 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1408 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1021 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1409 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1022 1410
1023 return sv; 1411 return sv;
1024} 1412}
1025 1413
… …
1029MODULE = JSON::XS PACKAGE = JSON::XS 1417MODULE = JSON::XS PACKAGE = JSON::XS
1030 1418
1031BOOT: 1419BOOT:
1032{ 1420{
1033 int i; 1421 int i;
1034
1035 memset (decode_hexdigit, 0xff, 256);
1036 1422
1037 for (i = 0; i < 256; ++i) 1423 for (i = 0; i < 256; ++i)
1038 decode_hexdigit [i] = 1424 decode_hexdigit [i] =
1039 i >= '0' && i <= '9' ? i - '0' 1425 i >= '0' && i <= '9' ? i - '0'
1040 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1426 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1041 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1427 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1042 : -1; 1428 : -1;
1043 1429
1044 json_stash = gv_stashpv ("JSON::XS", 1); 1430 json_stash = gv_stashpv ("JSON::XS" , 1);
1431 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1432
1433 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1434 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1045} 1435}
1046 1436
1047PROTOTYPES: DISABLE 1437PROTOTYPES: DISABLE
1048 1438
1049SV *new (char *dummy) 1439void CLONE (...)
1050 CODE: 1440 CODE:
1051 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1441 json_stash = 0;
1052 OUTPUT: 1442 json_boolean_stash = 0;
1053 RETVAL
1054 1443
1444void new (char *klass)
1445 PPCODE:
1446{
1447 SV *pv = NEWSV (0, sizeof (JSON));
1448 SvPOK_only (pv);
1449 Zero (SvPVX (pv), 1, JSON);
1450 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1451 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1452}
1453
1055SV *ascii (SV *self, int enable = 1) 1454void ascii (JSON *self, int enable = 1)
1056 ALIAS: 1455 ALIAS:
1057 ascii = F_ASCII 1456 ascii = F_ASCII
1457 latin1 = F_LATIN1
1058 utf8 = F_UTF8 1458 utf8 = F_UTF8
1059 indent = F_INDENT 1459 indent = F_INDENT
1060 canonical = F_CANONICAL 1460 canonical = F_CANONICAL
1061 space_before = F_SPACE_BEFORE 1461 space_before = F_SPACE_BEFORE
1062 space_after = F_SPACE_AFTER 1462 space_after = F_SPACE_AFTER
1063 pretty = F_PRETTY 1463 pretty = F_PRETTY
1064 allow_nonref = F_ALLOW_NONREF 1464 allow_nonref = F_ALLOW_NONREF
1065 shrink = F_SHRINK 1465 shrink = F_SHRINK
1466 allow_blessed = F_ALLOW_BLESSED
1467 convert_blessed = F_CONV_BLESSED
1468 relaxed = F_RELAXED
1066 CODE: 1469 PPCODE:
1067{ 1470{
1068 UV *uv = SvJSON (self);
1069 if (enable) 1471 if (enable)
1070 *uv |= ix; 1472 self->flags |= ix;
1071 else 1473 else
1072 *uv &= ~ix; 1474 self->flags &= ~ix;
1073 1475
1074 RETVAL = newSVsv (self); 1476 XPUSHs (ST (0));
1075} 1477}
1076 OUTPUT:
1077 RETVAL
1078 1478
1079SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1479void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1080 CODE: 1480 PPCODE:
1081{ 1481{
1082 UV *uv = SvJSON (self);
1083 UV log2 = 0; 1482 UV log2 = 0;
1084 1483
1085 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1484 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1086 1485
1087 while ((1UL << log2) < max_depth) 1486 while ((1UL << log2) < max_depth)
1088 ++log2; 1487 ++log2;
1089 1488
1090 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1489 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1091 1490
1092 RETVAL = newSVsv (self); 1491 XPUSHs (ST (0));
1093} 1492}
1094 OUTPUT:
1095 RETVAL
1096 1493
1097void encode (SV *self, SV *scalar) 1494void max_size (JSON *self, UV max_size = 0)
1098 PPCODE: 1495 PPCODE:
1099 XPUSHs (encode_json (scalar, *SvJSON (self))); 1496{
1497 UV log2 = 0;
1100 1498
1101void decode (SV *self, SV *jsonstr) 1499 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1500 if (max_size == 1) max_size = 2;
1501
1502 while ((1UL << log2) < max_size)
1503 ++log2;
1504
1505 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1506
1507 XPUSHs (ST (0));
1508}
1509
1510void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1102 PPCODE: 1511 PPCODE:
1512{
1513 SvREFCNT_dec (self->cb_object);
1514 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1515
1516 XPUSHs (ST (0));
1517}
1518
1519void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1520 PPCODE:
1521{
1522 if (!self->cb_sk_object)
1523 self->cb_sk_object = newHV ();
1524
1525 if (SvOK (cb))
1526 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1527 else
1528 {
1529 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1530
1531 if (!HvKEYS (self->cb_sk_object))
1532 {
1533 SvREFCNT_dec (self->cb_sk_object);
1534 self->cb_sk_object = 0;
1535 }
1536 }
1537
1538 XPUSHs (ST (0));
1539}
1540
1541void encode (JSON *self, SV *scalar)
1542 PPCODE:
1543 XPUSHs (encode_json (scalar, self));
1544
1545void decode (JSON *self, SV *jsonstr)
1546 PPCODE:
1103 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1547 XPUSHs (decode_json (jsonstr, self, 0));
1548
1549void decode_prefix (JSON *self, SV *jsonstr)
1550 PPCODE:
1551{
1552 UV offset;
1553 EXTEND (SP, 2);
1554 PUSHs (decode_json (jsonstr, self, &offset));
1555 PUSHs (sv_2mortal (newSVuv (offset)));
1556}
1557
1558void DESTROY (JSON *self)
1559 CODE:
1560 SvREFCNT_dec (self->cb_sk_object);
1561 SvREFCNT_dec (self->cb_object);
1104 1562
1105PROTOTYPES: ENABLE 1563PROTOTYPES: ENABLE
1106 1564
1107void to_json (SV *scalar) 1565void to_json (SV *scalar)
1108 ALIAS:
1109 objToJson = 0
1110 PPCODE: 1566 PPCODE:
1567{
1568 JSON json = { F_DEFAULT | F_UTF8 };
1111 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1569 XPUSHs (encode_json (scalar, &json));
1570}
1112 1571
1113void from_json (SV *jsonstr) 1572void from_json (SV *jsonstr)
1114 ALIAS:
1115 jsonToObj = 0
1116 PPCODE: 1573 PPCODE:
1574{
1575 JSON json = { F_DEFAULT | F_UTF8 };
1117 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1576 XPUSHs (decode_json (jsonstr, &json, 0));
1577}
1118 1578

Diff Legend

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