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 1.18 by root, Sun Mar 25 21:19:13 2007 UTC vs.
Revision 1.58 by root, Mon Aug 13 16:06:25 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
17#define F_MAXDEPTH 0xf8000000UL 32#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27 33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
19 37
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24 40
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (13UL << S_MAXDEPTH) 42#define F_DEFAULT (9UL << S_MAXDEPTH)
27 43
28#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 45#define INDENT_STEP 3 // spaces per indentation level
30 46
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 47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
33 48
34#define SB do { 49#define SB do {
35#define SE } while (0) 50#define SE } while (0)
36 51
52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value))
54# define inline inline
55#else
56# define expect(expr,value) (expr)
57# define inline static
58#endif
59
60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#ifdef USE_ITHREADS
64# define JSON_SLOW 1
65#else
66# define JSON_SLOW 0
67#endif
68
37static HV *json_stash; // JSON::XS:: 69static HV *json_stash, *json_boolean_stash; // JSON::XS::
70static SV *json_true, *json_false;
71
72typedef struct {
73 U32 flags;
74 SV *cb_object;
75 HV *cb_sk_object;
76} JSON;
38 77
39///////////////////////////////////////////////////////////////////////////// 78/////////////////////////////////////////////////////////////////////////////
40// utility functions 79// utility functions
41 80
42static UV * 81inline 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) 82shrink (SV *sv)
53{ 83{
54 sv_utf8_downgrade (sv, 1); 84 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1) 85 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 86 {
… …
65// decode an utf-8 character and return it, or (UV)-1 in 95// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error. 96// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 97// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 98// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 99// note that we never call this function for a ascii codepoints
70static UV 100inline UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 101decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 102{
73 if (s[0] > 0xdf || s[0] < 0xc2) 103 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 104 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 105 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 106 {
77 *clen = 2; 107 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 108 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 109 }
80 else 110 else
111 {
112 *clen = (STRLEN)-1;
81 return (UV)-1; 113 return (UV)-1;
114 }
82} 115}
83 116
84///////////////////////////////////////////////////////////////////////////// 117/////////////////////////////////////////////////////////////////////////////
85// encoder 118// encoder
86 119
… …
88typedef struct 121typedef struct
89{ 122{
90 char *cur; // SvPVX (sv) + current output position 123 char *cur; // SvPVX (sv) + current output position
91 char *end; // SvEND (sv) 124 char *end; // SvEND (sv)
92 SV *sv; // result scalar 125 SV *sv; // result scalar
93 U32 flags; // F_* 126 JSON json;
94 U32 indent; // indentation level 127 U32 indent; // indentation level
95 U32 maxdepth; // max. indentation/recursion level 128 U32 maxdepth; // max. indentation/recursion level
96} enc_t; 129} enc_t;
97 130
98static void 131inline void
99need (enc_t *enc, STRLEN len) 132need (enc_t *enc, STRLEN len)
100{ 133{
101 if (enc->cur + len >= enc->end) 134 if (expect_false (enc->cur + len >= enc->end))
102 { 135 {
103 STRLEN cur = enc->cur - SvPVX (enc->sv); 136 STRLEN cur = enc->cur - SvPVX (enc->sv);
104 SvGROW (enc->sv, cur + len + 1); 137 SvGROW (enc->sv, cur + len + 1);
105 enc->cur = SvPVX (enc->sv) + cur; 138 enc->cur = SvPVX (enc->sv) + cur;
106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 139 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107 } 140 }
108} 141}
109 142
110static void 143inline void
111encode_ch (enc_t *enc, char ch) 144encode_ch (enc_t *enc, char ch)
112{ 145{
113 need (enc, 1); 146 need (enc, 1);
114 *enc->cur++ = ch; 147 *enc->cur++ = ch;
115} 148}
… …
123 156
124 while (str < end) 157 while (str < end)
125 { 158 {
126 unsigned char ch = *(unsigned char *)str; 159 unsigned char ch = *(unsigned char *)str;
127 160
128 if (ch >= 0x20 && ch < 0x80) // most common case 161 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129 { 162 {
130 if (ch == '"') // but with slow exceptions 163 if (expect_false (ch == '"')) // but with slow exceptions
131 { 164 {
132 need (enc, len += 1); 165 need (enc, len += 1);
133 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
134 *enc->cur++ = '"'; 167 *enc->cur++ = '"';
135 } 168 }
136 else if (ch == '\\') 169 else if (expect_false (ch == '\\'))
137 { 170 {
138 need (enc, len += 1); 171 need (enc, len += 1);
139 *enc->cur++ = '\\'; 172 *enc->cur++ = '\\';
140 *enc->cur++ = '\\'; 173 *enc->cur++ = '\\';
141 } 174 }
… …
159 STRLEN clen; 192 STRLEN clen;
160 UV uch; 193 UV uch;
161 194
162 if (is_utf8) 195 if (is_utf8)
163 { 196 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen); 197 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1) 198 if (clen == (STRLEN)-1)
167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 199 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
168 } 200 }
169 else 201 else
… …
173 } 205 }
174 206
175 if (uch > 0x10FFFFUL) 207 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 208 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177 209
178 if (uch < 0x80 || enc->flags & F_ASCII) 210 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
179 { 211 {
180 if (uch > 0xFFFFUL) 212 if (uch > 0xFFFFUL)
181 { 213 {
182 need (enc, len += 11); 214 need (enc, len += 11);
183 sprintf (enc->cur, "\\u%04x\\u%04x", 215 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
197 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 229 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 } 230 }
199 231
200 str += clen; 232 str += clen;
201 } 233 }
234 else if (enc->json.flags & F_LATIN1)
235 {
236 *enc->cur++ = uch;
237 str += clen;
238 }
202 else if (is_utf8) 239 else if (is_utf8)
203 { 240 {
204 need (enc, len += clen); 241 need (enc, len += clen);
205 do 242 do
206 { 243 {
… …
208 } 245 }
209 while (--clen); 246 while (--clen);
210 } 247 }
211 else 248 else
212 { 249 {
213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 250 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 251 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str; 252 ++str;
216 } 253 }
217 } 254 }
218 } 255 }
… …
220 257
221 --len; 258 --len;
222 } 259 }
223} 260}
224 261
225static void 262inline void
226encode_indent (enc_t *enc) 263encode_indent (enc_t *enc)
227{ 264{
228 if (enc->flags & F_INDENT) 265 if (enc->json.flags & F_INDENT)
229 { 266 {
230 int spaces = enc->indent * INDENT_STEP; 267 int spaces = enc->indent * INDENT_STEP;
231 268
232 need (enc, spaces); 269 need (enc, spaces);
233 memset (enc->cur, ' ', spaces); 270 memset (enc->cur, ' ', spaces);
234 enc->cur += spaces; 271 enc->cur += spaces;
235 } 272 }
236} 273}
237 274
238static void 275inline void
239encode_space (enc_t *enc) 276encode_space (enc_t *enc)
240{ 277{
241 need (enc, 1); 278 need (enc, 1);
242 encode_ch (enc, ' '); 279 encode_ch (enc, ' ');
243} 280}
244 281
245static void 282inline void
246encode_nl (enc_t *enc) 283encode_nl (enc_t *enc)
247{ 284{
248 if (enc->flags & F_INDENT) 285 if (enc->json.flags & F_INDENT)
249 { 286 {
250 need (enc, 1); 287 need (enc, 1);
251 encode_ch (enc, '\n'); 288 encode_ch (enc, '\n');
252 } 289 }
253} 290}
254 291
255static void 292inline void
256encode_comma (enc_t *enc) 293encode_comma (enc_t *enc)
257{ 294{
258 encode_ch (enc, ','); 295 encode_ch (enc, ',');
259 296
260 if (enc->flags & F_INDENT) 297 if (enc->json.flags & F_INDENT)
261 encode_nl (enc); 298 encode_nl (enc);
262 else if (enc->flags & F_SPACE_AFTER) 299 else if (enc->json.flags & F_SPACE_AFTER)
263 encode_space (enc); 300 encode_space (enc);
264} 301}
265 302
266static void encode_sv (enc_t *enc, SV *sv); 303static void encode_sv (enc_t *enc, SV *sv);
267 304
268static void 305static void
269encode_av (enc_t *enc, AV *av) 306encode_av (enc_t *enc, AV *av)
270{ 307{
271 int i, len = av_len (av); 308 int i, len = av_len (av);
272 309
310 if (enc->indent >= enc->maxdepth)
311 croak ("data structure too deep (hit recursion limit)");
312
273 encode_ch (enc, '['); encode_nl (enc); 313 encode_ch (enc, '['); encode_nl (enc);
274 ++enc->indent; 314 ++enc->indent;
275 315
276 for (i = 0; i <= len; ++i) 316 for (i = 0; i <= len; ++i)
277 { 317 {
318 SV **svp = av_fetch (av, i, 0);
319
278 encode_indent (enc); 320 encode_indent (enc);
279 encode_sv (enc, *av_fetch (av, i, 0)); 321
322 if (svp)
323 encode_sv (enc, *svp);
324 else
325 encode_str (enc, "null", 4, 0);
280 326
281 if (i < len) 327 if (i < len)
282 encode_comma (enc); 328 encode_comma (enc);
283 } 329 }
284 330
… …
307 else 353 else
308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 354 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
309 355
310 encode_ch (enc, '"'); 356 encode_ch (enc, '"');
311 357
312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 358 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
313 encode_ch (enc, ':'); 359 encode_ch (enc, ':');
314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 360 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
315 encode_sv (enc, HeVAL (he)); 361 encode_sv (enc, HeVAL (he));
316} 362}
317 363
318// compare hash entries, used when all keys are bytestrings 364// compare hash entries, used when all keys are bytestrings
319static int 365static int
… …
342 388
343static void 389static void
344encode_hv (enc_t *enc, HV *hv) 390encode_hv (enc_t *enc, HV *hv)
345{ 391{
346 int count, i; 392 int count, i;
393
394 if (enc->indent >= enc->maxdepth)
395 croak ("data structure too deep (hit recursion limit)");
347 396
348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 397 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
349 398
350 if ((count = hv_iterinit (hv))) 399 if ((count = hv_iterinit (hv)))
351 { 400 {
352 // for canonical output we have to sort by keys first 401 // for canonical output we have to sort by keys first
353 // actually, this is mostly due to the stupid so-called 402 // actually, this is mostly due to the stupid so-called
354 // security workaround added somewhere in 5.8.x. 403 // security workaround added somewhere in 5.8.x.
355 // that randomises hash orderings 404 // that randomises hash orderings
356 if (enc->flags & F_CANONICAL) 405 if (enc->json.flags & F_CANONICAL)
357 { 406 {
358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
359 int fast = 1; 407 int fast = 1;
408 HE *he;
409#if defined(__BORLANDC__) || defined(_MSC_VER)
410 HE **hes = _alloca (count * sizeof (HE));
411#else
412 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
413#endif
360 414
361 i = 0; 415 i = 0;
362 while ((he = hv_iternext (hv))) 416 while ((he = hv_iternext (hv)))
363 { 417 {
364 hes [i++] = he; 418 hes [i++] = he;
… …
399 453
400 encode_nl (enc); 454 encode_nl (enc);
401 } 455 }
402 else 456 else
403 { 457 {
404 SV *sv;
405 HE *he = hv_iternext (hv); 458 HE *he = hv_iternext (hv);
406 459
407 for (;;) 460 for (;;)
408 { 461 {
409 encode_indent (enc); 462 encode_indent (enc);
… …
418 encode_nl (enc); 471 encode_nl (enc);
419 } 472 }
420 } 473 }
421 474
422 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 475 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
476}
477
478// encode objects, arrays and special \0=false and \1=true values.
479static void
480encode_rv (enc_t *enc, SV *sv)
481{
482 svtype svt;
483
484 SvGETMAGIC (sv);
485 svt = SvTYPE (sv);
486
487 if (expect_false (SvOBJECT (sv)))
488 {
489 HV *stash = !JSON_SLOW || json_boolean_stash
490 ? json_boolean_stash
491 : gv_stashpv ("JSON::XS::Boolean", 1);
492
493 if (SvSTASH (sv) == stash)
494 {
495 if (SvIV (sv))
496 encode_str (enc, "true", 4, 0);
497 else
498 encode_str (enc, "false", 5, 0);
499 }
500 else
501 {
502#if 0
503 if (0 && sv_derived_from (rv, "JSON::Literal"))
504 {
505 // not yet
506 }
507#endif
508 if (enc->json.flags & F_CONV_BLESSED)
509 {
510 // we re-bless the reference to get overload and other niceties right
511 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
512
513 if (to_json)
514 {
515 dSP;
516
517 ENTER; SAVETMPS; PUSHMARK (SP);
518 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
519
520 // calling with G_SCALAR ensures that we always get a 1 return value
521 PUTBACK;
522 call_sv ((SV *)GvCV (to_json), G_SCALAR);
523 SPAGAIN;
524
525 // catch this surprisingly common error
526 if (SvROK (TOPs) && SvRV (TOPs) == sv)
527 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
528
529 sv = POPs;
530 PUTBACK;
531
532 encode_sv (enc, sv);
533
534 FREETMPS; LEAVE;
535 }
536 else if (enc->json.flags & F_ALLOW_BLESSED)
537 encode_str (enc, "null", 4, 0);
538 else
539 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
540 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
541 }
542 else if (enc->json.flags & F_ALLOW_BLESSED)
543 encode_str (enc, "null", 4, 0);
544 else
545 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547 }
548 }
549 else if (svt == SVt_PVHV)
550 encode_hv (enc, (HV *)sv);
551 else if (svt == SVt_PVAV)
552 encode_av (enc, (AV *)sv);
553 else if (svt < SVt_PVAV)
554 {
555 STRLEN len = 0;
556 char *pv = svt ? SvPV (sv, len) : 0;
557
558 if (len == 1 && *pv == '1')
559 encode_str (enc, "true", 4, 0);
560 else if (len == 1 && *pv == '0')
561 encode_str (enc, "false", 5, 0);
562 else
563 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
564 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
565 }
566 else
567 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
568 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
423} 569}
424 570
425static void 571static void
426encode_sv (enc_t *enc, SV *sv) 572encode_sv (enc_t *enc, SV *sv)
427{ 573{
… …
435 encode_str (enc, str, len, SvUTF8 (sv)); 581 encode_str (enc, str, len, SvUTF8 (sv));
436 encode_ch (enc, '"'); 582 encode_ch (enc, '"');
437 } 583 }
438 else if (SvNOKp (sv)) 584 else if (SvNOKp (sv))
439 { 585 {
586 // trust that perl will do the right thing w.r.t. JSON syntax.
440 need (enc, NV_DIG + 32); 587 need (enc, NV_DIG + 32);
441 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 588 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
442 enc->cur += strlen (enc->cur); 589 enc->cur += strlen (enc->cur);
443 } 590 }
444 else if (SvIOKp (sv)) 591 else if (SvIOKp (sv))
445 { 592 {
446 need (enc, 64); 593 // we assume we can always read an IV as a UV
594 if (SvUV (sv) & ~(UV)0x7fff)
595 {
596 // large integer, use the (rather slow) snprintf way.
597 need (enc, sizeof (UV) * 3);
447 enc->cur += 598 enc->cur +=
448 SvIsUV(sv) 599 SvIsUV(sv)
449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 600 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 601 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
602 }
603 else
604 {
605 // optimise the "small number case"
606 // code will likely be branchless and use only a single multiplication
607 I32 i = SvIV (sv);
608 U32 u;
609 char digit, nz = 0;
610
611 need (enc, 6);
612
613 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
614 u = i < 0 ? -i : i;
615
616 // convert to 4.28 fixed-point representation
617 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
618
619 // now output digit by digit, each time masking out the integer part
620 // and multiplying by 5 while moving the decimal point one to the right,
621 // resulting in a net multiplication by 10.
622 // we always write the digit to memory but conditionally increment
623 // the pointer, to ease the usage of conditional move instructions.
624 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
625 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
626 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
627 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
628 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
629 }
451 } 630 }
452 else if (SvROK (sv)) 631 else if (SvROK (sv))
453 { 632 encode_rv (enc, SvRV (sv));
454 SV *rv = SvRV (sv);
455
456 if (enc->indent >= enc->maxdepth)
457 croak ("data structure too deep (hit recursion limit)");
458
459 switch (SvTYPE (rv))
460 {
461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
462 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
463
464 default:
465 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
466 SvPV_nolen (sv));
467 }
468 }
469 else if (!SvOK (sv)) 633 else if (!SvOK (sv))
470 encode_str (enc, "null", 4, 0); 634 encode_str (enc, "null", 4, 0);
471 else 635 else
472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 636 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv)); 637 SvPV_nolen (sv), SvFLAGS (sv));
474} 638}
475 639
476static SV * 640static SV *
477encode_json (SV *scalar, U32 flags) 641encode_json (SV *scalar, JSON *json)
478{ 642{
643 enc_t enc;
644
479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 645 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 646 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
481 647
482 enc_t enc; 648 enc.json = *json;
483 enc.flags = flags;
484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 649 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
485 enc.cur = SvPVX (enc.sv); 650 enc.cur = SvPVX (enc.sv);
486 enc.end = SvEND (enc.sv); 651 enc.end = SvEND (enc.sv);
487 enc.indent = 0; 652 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags); 653 enc.maxdepth = DEC_DEPTH (enc.json.flags);
489 654
490 SvPOK_only (enc.sv); 655 SvPOK_only (enc.sv);
491 encode_sv (&enc, scalar); 656 encode_sv (&enc, scalar);
492 657
658 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
659 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
660
493 if (!(flags & (F_ASCII | F_UTF8))) 661 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
494 SvUTF8_on (enc.sv); 662 SvUTF8_on (enc.sv);
495 663
496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497
498 if (enc.flags & F_SHRINK) 664 if (enc.json.flags & F_SHRINK)
499 shrink (enc.sv); 665 shrink (enc.sv);
500 666
501 return enc.sv; 667 return enc.sv;
502} 668}
503 669
… …
508typedef struct 674typedef struct
509{ 675{
510 char *cur; // current parser pointer 676 char *cur; // current parser pointer
511 char *end; // end of input string 677 char *end; // end of input string
512 const char *err; // parse error, if != 0 678 const char *err; // parse error, if != 0
513 U32 flags; // F_* 679 JSON json;
514 U32 depth; // recursion depth 680 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit 681 U32 maxdepth; // recursion depth limit
516} dec_t; 682} dec_t;
517 683
518static void 684inline void
519decode_ws (dec_t *dec) 685decode_ws (dec_t *dec)
520{ 686{
521 for (;;) 687 for (;;)
522 { 688 {
523 char ch = *dec->cur; 689 char ch = *dec->cur;
… …
549decode_4hex (dec_t *dec) 715decode_4hex (dec_t *dec)
550{ 716{
551 signed char d1, d2, d3, d4; 717 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur; 718 unsigned char *cur = (unsigned char *)dec->cur;
553 719
554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 720 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 721 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 722 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 723 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
558 724
559 dec->cur += 4; 725 dec->cur += 4;
560 726
561 return ((UV)d1) << 12 727 return ((UV)d1) << 12
562 | ((UV)d2) << 8 728 | ((UV)d2) << 8
… …
570static SV * 736static SV *
571decode_str (dec_t *dec) 737decode_str (dec_t *dec)
572{ 738{
573 SV *sv = 0; 739 SV *sv = 0;
574 int utf8 = 0; 740 int utf8 = 0;
741 char *dec_cur = dec->cur;
575 742
576 do 743 do
577 { 744 {
578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 745 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
579 char *cur = buf; 746 char *cur = buf;
580 747
581 do 748 do
582 { 749 {
583 unsigned char ch = *(unsigned char *)dec->cur++; 750 unsigned char ch = *(unsigned char *)dec_cur++;
584 751
585 if (ch == '"') 752 if (expect_false (ch == '"'))
586 { 753 {
587 --dec->cur; 754 --dec_cur;
588 break; 755 break;
589 } 756 }
590 else if (ch == '\\') 757 else if (expect_false (ch == '\\'))
591 { 758 {
592 switch (*dec->cur) 759 switch (*dec_cur)
593 { 760 {
594 case '\\': 761 case '\\':
595 case '/': 762 case '/':
596 case '"': *cur++ = *dec->cur++; break; 763 case '"': *cur++ = *dec_cur++; break;
597 764
598 case 'b': ++dec->cur; *cur++ = '\010'; break; 765 case 'b': ++dec_cur; *cur++ = '\010'; break;
599 case 't': ++dec->cur; *cur++ = '\011'; break; 766 case 't': ++dec_cur; *cur++ = '\011'; break;
600 case 'n': ++dec->cur; *cur++ = '\012'; break; 767 case 'n': ++dec_cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break; 768 case 'f': ++dec_cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break; 769 case 'r': ++dec_cur; *cur++ = '\015'; break;
603 770
604 case 'u': 771 case 'u':
605 { 772 {
606 UV lo, hi; 773 UV lo, hi;
607 ++dec->cur; 774 ++dec_cur;
608 775
776 dec->cur = dec_cur;
609 hi = decode_4hex (dec); 777 hi = decode_4hex (dec);
778 dec_cur = dec->cur;
610 if (hi == (UV)-1) 779 if (hi == (UV)-1)
611 goto fail; 780 goto fail;
612 781
613 // possibly a surrogate pair 782 // possibly a surrogate pair
614 if (hi >= 0xd800) 783 if (hi >= 0xd800)
615 if (hi < 0xdc00) 784 if (hi < 0xdc00)
616 { 785 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 786 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair"); 787 ERR ("missing low surrogate character in surrogate pair");
619 788
620 dec->cur += 2; 789 dec_cur += 2;
621 790
791 dec->cur = dec_cur;
622 lo = decode_4hex (dec); 792 lo = decode_4hex (dec);
793 dec_cur = dec->cur;
623 if (lo == (UV)-1) 794 if (lo == (UV)-1)
624 goto fail; 795 goto fail;
625 796
626 if (lo < 0xdc00 || lo >= 0xe000) 797 if (lo < 0xdc00 || lo >= 0xe000)
627 ERR ("surrogate pair expected"); 798 ERR ("surrogate pair expected");
… …
641 *cur++ = hi; 812 *cur++ = hi;
642 } 813 }
643 break; 814 break;
644 815
645 default: 816 default:
646 --dec->cur; 817 --dec_cur;
647 ERR ("illegal backslash escape sequence in string"); 818 ERR ("illegal backslash escape sequence in string");
648 } 819 }
649 } 820 }
650 else if (ch >= 0x20 && ch <= 0x7f) 821 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
651 *cur++ = ch; 822 *cur++ = ch;
652 else if (ch >= 0x80) 823 else if (ch >= 0x80)
653 { 824 {
654 --dec->cur;
655
656 STRLEN clen; 825 STRLEN clen;
826 UV uch;
827
828 --dec_cur;
829
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 830 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
658 if (clen == (STRLEN)-1) 831 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string"); 832 ERR ("malformed UTF-8 character in JSON string");
660 833
661 do 834 do
662 {
663 *cur++ = *dec->cur++; 835 *cur++ = *dec_cur++;
664 }
665 while (--clen); 836 while (--clen);
666 837
667 utf8 = 1; 838 utf8 = 1;
668 } 839 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else 840 else
841 {
842 --dec_cur;
843
844 if (!ch)
845 ERR ("unexpected end of string while parsing JSON string");
846 else
672 ERR ("invalid character encountered"); 847 ERR ("invalid character encountered while parsing JSON string");
673 848 }
674 } 849 }
675 while (cur < buf + SHORT_STRING_LEN); 850 while (cur < buf + SHORT_STRING_LEN);
676 851
852 {
677 STRLEN len = cur - buf; 853 STRLEN len = cur - buf;
678 854
679 if (sv) 855 if (sv)
680 { 856 {
681 SvGROW (sv, SvCUR (sv) + len + 1); 857 SvGROW (sv, SvCUR (sv) + len + 1);
682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 858 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
683 SvCUR_set (sv, SvCUR (sv) + len); 859 SvCUR_set (sv, SvCUR (sv) + len);
684 } 860 }
685 else 861 else
686 sv = newSVpvn (buf, len); 862 sv = newSVpvn (buf, len);
687 } 863 }
864 }
688 while (*dec->cur != '"'); 865 while (*dec_cur != '"');
689 866
690 ++dec->cur; 867 ++dec_cur;
691 868
692 if (sv) 869 if (sv)
693 { 870 {
694 SvPOK_only (sv); 871 SvPOK_only (sv);
695 *SvEND (sv) = 0; 872 *SvEND (sv) = 0;
… …
698 SvUTF8_on (sv); 875 SvUTF8_on (sv);
699 } 876 }
700 else 877 else
701 sv = newSVpvn ("", 0); 878 sv = newSVpvn ("", 0);
702 879
880 dec->cur = dec_cur;
703 return sv; 881 return sv;
704 882
705fail: 883fail:
884 dec->cur = dec_cur;
706 return 0; 885 return 0;
707} 886}
708 887
709static SV * 888static SV *
710decode_num (dec_t *dec) 889decode_num (dec_t *dec)
… …
768 is_nv = 1; 947 is_nv = 1;
769 } 948 }
770 949
771 if (!is_nv) 950 if (!is_nv)
772 { 951 {
773 UV uv; 952 int len = dec->cur - start;
774 int numtype = grok_number (start, dec->cur - start, &uv); 953
775 if (numtype & IS_NUMBER_IN_UV) 954 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
776 if (numtype & IS_NUMBER_NEG) 955 if (*start == '-')
956 switch (len)
777 { 957 {
778 if (uv < (UV)IV_MIN) 958 case 2: return newSViv (-( start [1] - '0' * 1));
779 return newSViv (-(IV)uv); 959 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
960 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
961 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
780 } 962 }
963 else
964 switch (len)
965 {
966 case 1: return newSViv ( start [0] - '0' * 1);
967 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
968 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
969 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
970 }
971
972 {
973 UV uv;
974 int numtype = grok_number (start, len, &uv);
975 if (numtype & IS_NUMBER_IN_UV)
976 if (numtype & IS_NUMBER_NEG)
977 {
978 if (uv < (UV)IV_MIN)
979 return newSViv (-(IV)uv);
980 }
781 else 981 else
782 return newSVuv (uv); 982 return newSVuv (uv);
783 } 983 }
784 984
985 len -= *start == '-' ? 1 : 0;
986
987 // does not fit into IV or UV, try NV
988 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
989 #if defined (LDBL_DIG)
990 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
991 #endif
992 )
993 // fits into NV without loss of precision
994 return newSVnv (Atof (start));
995
996 // everything else fails, convert it to a string
997 return newSVpvn (start, dec->cur - start);
998 }
999
1000 // loss of precision here
785 return newSVnv (Atof (start)); 1001 return newSVnv (Atof (start));
786 1002
787fail: 1003fail:
788 return 0; 1004 return 0;
789} 1005}
… …
833} 1049}
834 1050
835static SV * 1051static SV *
836decode_hv (dec_t *dec) 1052decode_hv (dec_t *dec)
837{ 1053{
1054 SV *sv;
838 HV *hv = newHV (); 1055 HV *hv = newHV ();
839 1056
840 DEC_INC_DEPTH; 1057 DEC_INC_DEPTH;
841 decode_ws (dec); 1058 decode_ws (dec);
842 1059
843 if (*dec->cur == '}') 1060 if (*dec->cur == '}')
844 ++dec->cur; 1061 ++dec->cur;
845 else 1062 else
846 for (;;) 1063 for (;;)
847 { 1064 {
848 SV *key, *value;
849
850 decode_ws (dec); EXPECT_CH ('"'); 1065 decode_ws (dec); EXPECT_CH ('"');
851 1066
852 key = decode_str (dec); 1067 // heuristic: assume that
853 if (!key) 1068 // a) decode_str + hv_store_ent are abysmally slow.
854 goto fail; 1069 // b) most hash keys are short, simple ascii text.
1070 // => try to "fast-match" such strings to avoid
1071 // the overhead of decode_str + hv_store_ent.
1072 {
1073 SV *value;
1074 char *p = dec->cur;
1075 char *e = p + 24; // only try up to 24 bytes
855 1076
856 decode_ws (dec); EXPECT_CH (':'); 1077 for (;;)
857
858 value = decode_sv (dec);
859 if (!value)
860 { 1078 {
1079 // the >= 0x80 is true on most architectures
1080 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1081 {
1082 // slow path, back up and use decode_str
1083 SV *key = decode_str (dec);
1084 if (!key)
1085 goto fail;
1086
1087 decode_ws (dec); EXPECT_CH (':');
1088
1089 value = decode_sv (dec);
1090 if (!value)
1091 {
1092 SvREFCNT_dec (key);
1093 goto fail;
1094 }
1095
1096 hv_store_ent (hv, key, value, 0);
861 SvREFCNT_dec (key); 1097 SvREFCNT_dec (key);
1098
1099 break;
1100 }
1101 else if (*p == '"')
1102 {
1103 // fast path, got a simple key
1104 char *key = dec->cur;
1105 int len = p - key;
1106 dec->cur = p + 1;
1107
1108 decode_ws (dec); EXPECT_CH (':');
1109
1110 value = decode_sv (dec);
1111 if (!value)
862 goto fail; 1112 goto fail;
1113
1114 hv_store (hv, key, len, value, 0);
1115
1116 break;
1117 }
1118
1119 ++p;
863 } 1120 }
864 1121 }
865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
867 1122
868 decode_ws (dec); 1123 decode_ws (dec);
869 1124
870 if (*dec->cur == '}') 1125 if (*dec->cur == '}')
871 { 1126 {
… …
878 1133
879 ++dec->cur; 1134 ++dec->cur;
880 } 1135 }
881 1136
882 DEC_DEC_DEPTH; 1137 DEC_DEC_DEPTH;
883 return newRV_noinc ((SV *)hv); 1138 sv = newRV_noinc ((SV *)hv);
1139
1140 // check filter callbacks
1141 if (dec->json.flags & F_HOOK)
1142 {
1143 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1144 {
1145 HE *cb, *he;
1146
1147 hv_iterinit (hv);
1148 he = hv_iternext (hv);
1149 hv_iterinit (hv);
1150
1151 // the next line creates a mortal sv each time its called.
1152 // might want to optimise this for common cases.
1153 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1154
1155 if (cb)
1156 {
1157 dSP;
1158 int count;
1159
1160 ENTER; SAVETMPS; PUSHMARK (SP);
1161 XPUSHs (HeVAL (he));
1162
1163 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1164
1165 if (count == 1)
1166 {
1167 sv = newSVsv (POPs);
1168 FREETMPS; LEAVE;
1169 return sv;
1170 }
1171
1172 FREETMPS; LEAVE;
1173 }
1174 }
1175
1176 if (dec->json.cb_object)
1177 {
1178 dSP;
1179 int count;
1180
1181 ENTER; SAVETMPS; PUSHMARK (SP);
1182 XPUSHs (sv_2mortal (sv));
1183
1184 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1185
1186 if (count == 1)
1187 {
1188 sv = newSVsv (POPs);
1189 FREETMPS; LEAVE;
1190 return sv;
1191 }
1192
1193 SvREFCNT_inc (sv);
1194 FREETMPS; LEAVE;
1195 }
1196 }
1197
1198 return sv;
884 1199
885fail: 1200fail:
886 SvREFCNT_dec (hv); 1201 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH; 1202 DEC_DEC_DEPTH;
888 return 0; 1203 return 0;
… …
890 1205
891static SV * 1206static SV *
892decode_sv (dec_t *dec) 1207decode_sv (dec_t *dec)
893{ 1208{
894 decode_ws (dec); 1209 decode_ws (dec);
1210
1211 // the beauty of JSON: you need exactly one character lookahead
1212 // to parse anything.
895 switch (*dec->cur) 1213 switch (*dec->cur)
896 { 1214 {
897 case '"': ++dec->cur; return decode_str (dec); 1215 case '"': ++dec->cur; return decode_str (dec);
898 case '[': ++dec->cur; return decode_av (dec); 1216 case '[': ++dec->cur; return decode_av (dec);
899 case '{': ++dec->cur; return decode_hv (dec); 1217 case '{': ++dec->cur; return decode_hv (dec);
… …
905 1223
906 case 't': 1224 case 't':
907 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1225 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
908 { 1226 {
909 dec->cur += 4; 1227 dec->cur += 4;
910 return newSViv (1); 1228 return SvREFCNT_inc (json_true);
911 } 1229 }
912 else 1230 else
913 ERR ("'true' expected"); 1231 ERR ("'true' expected");
914 1232
915 break; 1233 break;
916 1234
917 case 'f': 1235 case 'f':
918 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1236 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
919 { 1237 {
920 dec->cur += 5; 1238 dec->cur += 5;
921 return newSViv (0); 1239 return SvREFCNT_inc (json_false);
922 } 1240 }
923 else 1241 else
924 ERR ("'false' expected"); 1242 ERR ("'false' expected");
925 1243
926 break; 1244 break;
… …
935 ERR ("'null' expected"); 1253 ERR ("'null' expected");
936 1254
937 break; 1255 break;
938 1256
939 default: 1257 default:
940 ERR ("malformed json string, neither array, object, number, string or atom"); 1258 ERR ("malformed JSON string, neither array, object, number, string or atom");
941 break; 1259 break;
942 } 1260 }
943 1261
944fail: 1262fail:
945 return 0; 1263 return 0;
946} 1264}
947 1265
948static SV * 1266static SV *
949decode_json (SV *string, U32 flags) 1267decode_json (SV *string, JSON *json, UV *offset_return)
950{ 1268{
1269 dec_t dec;
1270 UV offset;
951 SV *sv; 1271 SV *sv;
952 1272
1273 SvGETMAGIC (string);
1274 SvUPGRADE (string, SVt_PV);
1275
1276 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1277 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1278 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1279
953 if (flags & F_UTF8) 1280 if (json->flags & F_UTF8)
954 sv_utf8_downgrade (string, 0); 1281 sv_utf8_downgrade (string, 0);
955 else 1282 else
956 sv_utf8_upgrade (string); 1283 sv_utf8_upgrade (string);
957 1284
958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1285 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
959 1286
960 dec_t dec; 1287 dec.json = *json;
961 dec.flags = flags;
962 dec.cur = SvPVX (string); 1288 dec.cur = SvPVX (string);
963 dec.end = SvEND (string); 1289 dec.end = SvEND (string);
964 dec.err = 0; 1290 dec.err = 0;
965 dec.depth = 0; 1291 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags); 1292 dec.maxdepth = DEC_DEPTH (dec.json.flags);
967 1293
968 *SvEND (sv) = 0; // this shou[ld basically be a nop, too 1294 if (dec.json.cb_object || dec.json.cb_sk_object)
1295 dec.json.flags |= F_HOOK;
1296
1297 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
969 sv = decode_sv (&dec); 1298 sv = decode_sv (&dec);
970 1299
1300 if (!(offset_return || !sv))
1301 {
1302 // check for trailing garbage
1303 decode_ws (&dec);
1304
1305 if (*dec.cur)
1306 {
1307 dec.err = "garbage after JSON object";
1308 SvREFCNT_dec (sv);
1309 sv = 0;
1310 }
1311 }
1312
1313 if (offset_return || !sv)
1314 {
1315 offset = dec.json.flags & F_UTF8
1316 ? dec.cur - SvPVX (string)
1317 : utf8_distance (dec.cur, SvPVX (string));
1318
1319 if (offset_return)
1320 *offset_return = offset;
1321 }
1322
971 if (!sv) 1323 if (!sv)
972 { 1324 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
975 : utf8_distance (dec.cur, SvPVX (string));
976 SV *uni = sv_newmortal (); 1325 SV *uni = sv_newmortal ();
977 1326
978 // horrible hack to silence warning inside pv_uni_display 1327 // horrible hack to silence warning inside pv_uni_display
979 COP cop = *PL_curcop; 1328 COP cop = *PL_curcop;
980 cop.cop_warnings = pWARN_NONE; 1329 cop.cop_warnings = pWARN_NONE;
… …
982 SAVEVPTR (PL_curcop); 1331 SAVEVPTR (PL_curcop);
983 PL_curcop = &cop; 1332 PL_curcop = &cop;
984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1333 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE; 1334 LEAVE;
986 1335
987 croak ("%s, at character offset %d (%s)", 1336 croak ("%s, at character offset %d [\"%s\"]",
988 dec.err, 1337 dec.err,
989 (int)offset, 1338 (int)offset,
990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1339 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
991 } 1340 }
992 1341
993 sv = sv_2mortal (sv); 1342 sv = sv_2mortal (sv);
994 1343
995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1344 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
996 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1345 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
997 1346
998 return sv; 1347 return sv;
999} 1348}
1000 1349
… …
1004MODULE = JSON::XS PACKAGE = JSON::XS 1353MODULE = JSON::XS PACKAGE = JSON::XS
1005 1354
1006BOOT: 1355BOOT:
1007{ 1356{
1008 int i; 1357 int i;
1009
1010 memset (decode_hexdigit, 0xff, 256);
1011 1358
1012 for (i = 0; i < 256; ++i) 1359 for (i = 0; i < 256; ++i)
1013 decode_hexdigit [i] = 1360 decode_hexdigit [i] =
1014 i >= '0' && i <= '9' ? i - '0' 1361 i >= '0' && i <= '9' ? i - '0'
1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1362 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1363 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1017 : -1; 1364 : -1;
1018 1365
1019 json_stash = gv_stashpv ("JSON::XS", 1); 1366 json_stash = gv_stashpv ("JSON::XS" , 1);
1367 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1368
1369 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1370 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1020} 1371}
1021 1372
1022PROTOTYPES: DISABLE 1373PROTOTYPES: DISABLE
1023 1374
1024SV *new (char *dummy) 1375void CLONE (...)
1025 CODE: 1376 CODE:
1026 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1377 json_stash = 0;
1027 OUTPUT: 1378 json_boolean_stash = 0;
1028 RETVAL
1029 1379
1380void new (char *klass)
1381 PPCODE:
1382{
1383 HV *stash = !JSON_SLOW || json_stash
1384 ? json_stash
1385 : gv_stashpv ("JSON::XS", 1);
1386 SV *pv = NEWSV (0, sizeof (JSON));
1387 SvPOK_only (pv);
1388 Zero (SvPVX (pv), 1, JSON);
1389 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1390 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), stash)));
1391}
1392
1030SV *ascii (SV *self, int enable = 1) 1393void ascii (JSON *self, int enable = 1)
1031 ALIAS: 1394 ALIAS:
1032 ascii = F_ASCII 1395 ascii = F_ASCII
1396 latin1 = F_LATIN1
1033 utf8 = F_UTF8 1397 utf8 = F_UTF8
1034 indent = F_INDENT 1398 indent = F_INDENT
1035 canonical = F_CANONICAL 1399 canonical = F_CANONICAL
1036 space_before = F_SPACE_BEFORE 1400 space_before = F_SPACE_BEFORE
1037 space_after = F_SPACE_AFTER 1401 space_after = F_SPACE_AFTER
1038 pretty = F_PRETTY 1402 pretty = F_PRETTY
1039 allow_nonref = F_ALLOW_NONREF 1403 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK 1404 shrink = F_SHRINK
1405 allow_blessed = F_ALLOW_BLESSED
1406 convert_blessed = F_CONV_BLESSED
1041 CODE: 1407 PPCODE:
1042{ 1408{
1043 UV *uv = SvJSON (self);
1044 if (enable) 1409 if (enable)
1045 *uv |= ix; 1410 self->flags |= ix;
1046 else 1411 else
1047 *uv &= ~ix; 1412 self->flags &= ~ix;
1048 1413
1049 RETVAL = newSVsv (self); 1414 XPUSHs (ST (0));
1050} 1415}
1051 OUTPUT:
1052 RETVAL
1053 1416
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1417void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1055 CODE: 1418 PPCODE:
1056{ 1419{
1057 UV *uv = SvJSON (self);
1058 UV log2 = 0; 1420 UV log2 = 0;
1059 1421
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1422 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1061 1423
1062 while ((1UL << log2) < max_depth) 1424 while ((1UL << log2) < max_depth)
1063 ++log2; 1425 ++log2;
1064 1426
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1427 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066 1428
1067 RETVAL = newSVsv (self); 1429 XPUSHs (ST (0));
1068} 1430}
1069 OUTPUT:
1070 RETVAL
1071 1431
1072void encode (SV *self, SV *scalar) 1432void max_size (JSON *self, UV max_size = 0)
1073 PPCODE: 1433 PPCODE:
1074 XPUSHs (encode_json (scalar, *SvJSON (self))); 1434{
1435 UV log2 = 0;
1075 1436
1076void decode (SV *self, SV *jsonstr) 1437 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1438 if (max_size == 1) max_size = 2;
1439
1440 while ((1UL << log2) < max_size)
1441 ++log2;
1442
1443 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1444
1445 XPUSHs (ST (0));
1446}
1447
1448void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1449 PPCODE:
1450{
1451 SvREFCNT_dec (self->cb_object);
1452 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1453
1454 XPUSHs (ST (0));
1455}
1456
1457void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1458 PPCODE:
1459{
1460 if (!self->cb_sk_object)
1461 self->cb_sk_object = newHV ();
1462
1463 if (SvOK (cb))
1464 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1465 else
1466 {
1467 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1468
1469 if (!HvKEYS (self->cb_sk_object))
1470 {
1471 SvREFCNT_dec (self->cb_sk_object);
1472 self->cb_sk_object = 0;
1473 }
1474 }
1475
1476 XPUSHs (ST (0));
1477}
1478
1479void encode (JSON *self, SV *scalar)
1480 PPCODE:
1481 XPUSHs (encode_json (scalar, self));
1482
1483void decode (JSON *self, SV *jsonstr)
1484 PPCODE:
1078 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1485 XPUSHs (decode_json (jsonstr, self, 0));
1486
1487void decode_prefix (JSON *self, SV *jsonstr)
1488 PPCODE:
1489{
1490 UV offset;
1491 EXTEND (SP, 2);
1492 PUSHs (decode_json (jsonstr, self, &offset));
1493 PUSHs (sv_2mortal (newSVuv (offset)));
1494}
1495
1496void DESTROY (JSON *self)
1497 CODE:
1498 SvREFCNT_dec (self->cb_sk_object);
1499 SvREFCNT_dec (self->cb_object);
1079 1500
1080PROTOTYPES: ENABLE 1501PROTOTYPES: ENABLE
1081 1502
1082void to_json (SV *scalar) 1503void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
1085 PPCODE: 1504 PPCODE:
1505{
1506 JSON json = { F_DEFAULT | F_UTF8 };
1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1507 XPUSHs (encode_json (scalar, &json));
1508}
1087 1509
1088void from_json (SV *jsonstr) 1510void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
1091 PPCODE: 1511 PPCODE:
1512{
1513 JSON json = { F_DEFAULT | F_UTF8 };
1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1514 XPUSHs (decode_json (jsonstr, &json, 0));
1515}
1093 1516

Diff Legend

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