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 1.19 by root, Sun Mar 25 21:52:47 2007 UTC vs.
Revision 1.56 by root, Thu Jul 26 11:33:35 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
37static HV *json_stash; // JSON::XS:: 63static HV *json_stash, *json_boolean_stash; // JSON::XS::
64static SV *json_true, *json_false;
65
66typedef struct {
67 U32 flags;
68 SV *cb_object;
69 HV *cb_sk_object;
70} JSON;
38 71
39///////////////////////////////////////////////////////////////////////////// 72/////////////////////////////////////////////////////////////////////////////
40// utility functions 73// utility functions
41 74
42static UV * 75inline 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) 76shrink (SV *sv)
53{ 77{
54 sv_utf8_downgrade (sv, 1); 78 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1) 79 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 80 {
… …
65// decode an utf-8 character and return it, or (UV)-1 in 89// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error. 90// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 91// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 92// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 93// note that we never call this function for a ascii codepoints
70static UV 94inline UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 95decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 96{
73 if (s[0] > 0xdf || s[0] < 0xc2) 97 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 98 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 99 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 100 {
77 *clen = 2; 101 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 102 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 103 }
80 else 104 else
105 {
106 *clen = (STRLEN)-1;
81 return (UV)-1; 107 return (UV)-1;
108 }
82} 109}
83 110
84///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
85// encoder 112// encoder
86 113
… …
88typedef struct 115typedef struct
89{ 116{
90 char *cur; // SvPVX (sv) + current output position 117 char *cur; // SvPVX (sv) + current output position
91 char *end; // SvEND (sv) 118 char *end; // SvEND (sv)
92 SV *sv; // result scalar 119 SV *sv; // result scalar
93 U32 flags; // F_* 120 JSON json;
94 U32 indent; // indentation level 121 U32 indent; // indentation level
95 U32 maxdepth; // max. indentation/recursion level 122 U32 maxdepth; // max. indentation/recursion level
96} enc_t; 123} enc_t;
97 124
98static void 125inline void
99need (enc_t *enc, STRLEN len) 126need (enc_t *enc, STRLEN len)
100{ 127{
101 if (enc->cur + len >= enc->end) 128 if (expect_false (enc->cur + len >= enc->end))
102 { 129 {
103 STRLEN cur = enc->cur - SvPVX (enc->sv); 130 STRLEN cur = enc->cur - SvPVX (enc->sv);
104 SvGROW (enc->sv, cur + len + 1); 131 SvGROW (enc->sv, cur + len + 1);
105 enc->cur = SvPVX (enc->sv) + cur; 132 enc->cur = SvPVX (enc->sv) + cur;
106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 133 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107 } 134 }
108} 135}
109 136
110static void 137inline void
111encode_ch (enc_t *enc, char ch) 138encode_ch (enc_t *enc, char ch)
112{ 139{
113 need (enc, 1); 140 need (enc, 1);
114 *enc->cur++ = ch; 141 *enc->cur++ = ch;
115} 142}
… …
123 150
124 while (str < end) 151 while (str < end)
125 { 152 {
126 unsigned char ch = *(unsigned char *)str; 153 unsigned char ch = *(unsigned char *)str;
127 154
128 if (ch >= 0x20 && ch < 0x80) // most common case 155 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129 { 156 {
130 if (ch == '"') // but with slow exceptions 157 if (expect_false (ch == '"')) // but with slow exceptions
131 { 158 {
132 need (enc, len += 1); 159 need (enc, len += 1);
133 *enc->cur++ = '\\'; 160 *enc->cur++ = '\\';
134 *enc->cur++ = '"'; 161 *enc->cur++ = '"';
135 } 162 }
136 else if (ch == '\\') 163 else if (expect_false (ch == '\\'))
137 { 164 {
138 need (enc, len += 1); 165 need (enc, len += 1);
139 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
140 *enc->cur++ = '\\'; 167 *enc->cur++ = '\\';
141 } 168 }
… …
159 STRLEN clen; 186 STRLEN clen;
160 UV uch; 187 UV uch;
161 188
162 if (is_utf8) 189 if (is_utf8)
163 { 190 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen); 191 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1) 192 if (clen == (STRLEN)-1)
167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 193 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
168 } 194 }
169 else 195 else
… …
173 } 199 }
174 200
175 if (uch > 0x10FFFFUL) 201 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 202 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177 203
178 if (uch < 0x80 || enc->flags & F_ASCII) 204 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
179 { 205 {
180 if (uch > 0xFFFFUL) 206 if (uch > 0xFFFFUL)
181 { 207 {
182 need (enc, len += 11); 208 need (enc, len += 11);
183 sprintf (enc->cur, "\\u%04x\\u%04x", 209 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
197 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 223 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 } 224 }
199 225
200 str += clen; 226 str += clen;
201 } 227 }
228 else if (enc->json.flags & F_LATIN1)
229 {
230 *enc->cur++ = uch;
231 str += clen;
232 }
202 else if (is_utf8) 233 else if (is_utf8)
203 { 234 {
204 need (enc, len += clen); 235 need (enc, len += clen);
205 do 236 do
206 { 237 {
… …
208 } 239 }
209 while (--clen); 240 while (--clen);
210 } 241 }
211 else 242 else
212 { 243 {
213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 244 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 245 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str; 246 ++str;
216 } 247 }
217 } 248 }
218 } 249 }
… …
220 251
221 --len; 252 --len;
222 } 253 }
223} 254}
224 255
225static void 256inline void
226encode_indent (enc_t *enc) 257encode_indent (enc_t *enc)
227{ 258{
228 if (enc->flags & F_INDENT) 259 if (enc->json.flags & F_INDENT)
229 { 260 {
230 int spaces = enc->indent * INDENT_STEP; 261 int spaces = enc->indent * INDENT_STEP;
231 262
232 need (enc, spaces); 263 need (enc, spaces);
233 memset (enc->cur, ' ', spaces); 264 memset (enc->cur, ' ', spaces);
234 enc->cur += spaces; 265 enc->cur += spaces;
235 } 266 }
236} 267}
237 268
238static void 269inline void
239encode_space (enc_t *enc) 270encode_space (enc_t *enc)
240{ 271{
241 need (enc, 1); 272 need (enc, 1);
242 encode_ch (enc, ' '); 273 encode_ch (enc, ' ');
243} 274}
244 275
245static void 276inline void
246encode_nl (enc_t *enc) 277encode_nl (enc_t *enc)
247{ 278{
248 if (enc->flags & F_INDENT) 279 if (enc->json.flags & F_INDENT)
249 { 280 {
250 need (enc, 1); 281 need (enc, 1);
251 encode_ch (enc, '\n'); 282 encode_ch (enc, '\n');
252 } 283 }
253} 284}
254 285
255static void 286inline void
256encode_comma (enc_t *enc) 287encode_comma (enc_t *enc)
257{ 288{
258 encode_ch (enc, ','); 289 encode_ch (enc, ',');
259 290
260 if (enc->flags & F_INDENT) 291 if (enc->json.flags & F_INDENT)
261 encode_nl (enc); 292 encode_nl (enc);
262 else if (enc->flags & F_SPACE_AFTER) 293 else if (enc->json.flags & F_SPACE_AFTER)
263 encode_space (enc); 294 encode_space (enc);
264} 295}
265 296
266static void encode_sv (enc_t *enc, SV *sv); 297static void encode_sv (enc_t *enc, SV *sv);
267 298
268static void 299static void
269encode_av (enc_t *enc, AV *av) 300encode_av (enc_t *enc, AV *av)
270{ 301{
271 int i, len = av_len (av); 302 int i, len = av_len (av);
272 303
304 if (enc->indent >= enc->maxdepth)
305 croak ("data structure too deep (hit recursion limit)");
306
273 encode_ch (enc, '['); encode_nl (enc); 307 encode_ch (enc, '['); encode_nl (enc);
274 ++enc->indent; 308 ++enc->indent;
275 309
276 for (i = 0; i <= len; ++i) 310 for (i = 0; i <= len; ++i)
277 { 311 {
312 SV **svp = av_fetch (av, i, 0);
313
278 encode_indent (enc); 314 encode_indent (enc);
279 encode_sv (enc, *av_fetch (av, i, 0)); 315
316 if (svp)
317 encode_sv (enc, *svp);
318 else
319 encode_str (enc, "null", 4, 0);
280 320
281 if (i < len) 321 if (i < len)
282 encode_comma (enc); 322 encode_comma (enc);
283 } 323 }
284 324
… …
307 else 347 else
308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 348 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
309 349
310 encode_ch (enc, '"'); 350 encode_ch (enc, '"');
311 351
312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 352 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
313 encode_ch (enc, ':'); 353 encode_ch (enc, ':');
314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 354 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
315 encode_sv (enc, HeVAL (he)); 355 encode_sv (enc, HeVAL (he));
316} 356}
317 357
318// compare hash entries, used when all keys are bytestrings 358// compare hash entries, used when all keys are bytestrings
319static int 359static int
… …
342 382
343static void 383static void
344encode_hv (enc_t *enc, HV *hv) 384encode_hv (enc_t *enc, HV *hv)
345{ 385{
346 int count, i; 386 int count, i;
387
388 if (enc->indent >= enc->maxdepth)
389 croak ("data structure too deep (hit recursion limit)");
347 390
348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 391 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
349 392
350 if ((count = hv_iterinit (hv))) 393 if ((count = hv_iterinit (hv)))
351 { 394 {
352 // for canonical output we have to sort by keys first 395 // for canonical output we have to sort by keys first
353 // actually, this is mostly due to the stupid so-called 396 // actually, this is mostly due to the stupid so-called
354 // security workaround added somewhere in 5.8.x. 397 // security workaround added somewhere in 5.8.x.
355 // that randomises hash orderings 398 // that randomises hash orderings
356 if (enc->flags & F_CANONICAL) 399 if (enc->json.flags & F_CANONICAL)
357 { 400 {
358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
359 int fast = 1; 401 int fast = 1;
402 HE *he;
403#if defined(__BORLANDC__) || defined(_MSC_VER)
404 HE **hes = _alloca (count * sizeof (HE));
405#else
406 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
407#endif
360 408
361 i = 0; 409 i = 0;
362 while ((he = hv_iternext (hv))) 410 while ((he = hv_iternext (hv)))
363 { 411 {
364 hes [i++] = he; 412 hes [i++] = he;
… …
399 447
400 encode_nl (enc); 448 encode_nl (enc);
401 } 449 }
402 else 450 else
403 { 451 {
404 SV *sv;
405 HE *he = hv_iternext (hv); 452 HE *he = hv_iternext (hv);
406 453
407 for (;;) 454 for (;;)
408 { 455 {
409 encode_indent (enc); 456 encode_indent (enc);
… …
418 encode_nl (enc); 465 encode_nl (enc);
419 } 466 }
420 } 467 }
421 468
422 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 469 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
470}
471
472// encode objects, arrays and special \0=false and \1=true values.
473static void
474encode_rv (enc_t *enc, SV *sv)
475{
476 svtype svt;
477
478 SvGETMAGIC (sv);
479 svt = SvTYPE (sv);
480
481 if (expect_false (SvOBJECT (sv)))
482 {
483 if (SvSTASH (sv) == json_boolean_stash)
484 {
485 if (SvIV (sv))
486 encode_str (enc, "true", 4, 0);
487 else
488 encode_str (enc, "false", 5, 0);
489 }
490 else
491 {
492#if 0
493 if (0 && sv_derived_from (rv, "JSON::Literal"))
494 {
495 // not yet
496 }
497#endif
498 if (enc->json.flags & F_CONV_BLESSED)
499 {
500 // we re-bless the reference to get overload and other niceties right
501 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
502
503 if (to_json)
504 {
505 int count;
506 dSP;
507
508 ENTER; SAVETMPS; PUSHMARK (SP);
509 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
510
511 // calling with G_SCALAR ensures that we always get a 1 return value
512 PUTBACK;
513 call_sv ((SV *)GvCV (to_json), G_SCALAR);
514 SPAGAIN;
515
516 // catch this surprisingly common error
517 if (SvROK (TOPs) && SvRV (TOPs) == sv)
518 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
519
520 sv = POPs;
521 PUTBACK;
522
523 encode_sv (enc, sv);
524
525 FREETMPS; LEAVE;
526 }
527 else if (enc->json.flags & F_ALLOW_BLESSED)
528 encode_str (enc, "null", 4, 0);
529 else
530 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
531 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
532 }
533 else if (enc->json.flags & F_ALLOW_BLESSED)
534 encode_str (enc, "null", 4, 0);
535 else
536 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
537 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
538 }
539 }
540 else if (svt == SVt_PVHV)
541 encode_hv (enc, (HV *)sv);
542 else if (svt == SVt_PVAV)
543 encode_av (enc, (AV *)sv);
544 else if (svt < SVt_PVAV)
545 {
546 STRLEN len = 0;
547 char *pv = svt ? SvPV (sv, len) : 0;
548
549 if (len == 1 && *pv == '1')
550 encode_str (enc, "true", 4, 0);
551 else if (len == 1 && *pv == '0')
552 encode_str (enc, "false", 5, 0);
553 else
554 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
555 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
556 }
557 else
558 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
559 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
423} 560}
424 561
425static void 562static void
426encode_sv (enc_t *enc, SV *sv) 563encode_sv (enc_t *enc, SV *sv)
427{ 564{
… …
435 encode_str (enc, str, len, SvUTF8 (sv)); 572 encode_str (enc, str, len, SvUTF8 (sv));
436 encode_ch (enc, '"'); 573 encode_ch (enc, '"');
437 } 574 }
438 else if (SvNOKp (sv)) 575 else if (SvNOKp (sv))
439 { 576 {
577 // trust that perl will do the right thing w.r.t. JSON syntax.
440 need (enc, NV_DIG + 32); 578 need (enc, NV_DIG + 32);
441 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 579 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
442 enc->cur += strlen (enc->cur); 580 enc->cur += strlen (enc->cur);
443 } 581 }
444 else if (SvIOKp (sv)) 582 else if (SvIOKp (sv))
445 { 583 {
446 need (enc, 64); 584 // we assume we can always read an IV as a UV
585 if (SvUV (sv) & ~(UV)0x7fff)
586 {
587 // large integer, use the (rather slow) snprintf way.
588 need (enc, sizeof (UV) * 3);
447 enc->cur += 589 enc->cur +=
448 SvIsUV(sv) 590 SvIsUV(sv)
449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 591 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 592 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
593 }
594 else
595 {
596 // optimise the "small number case"
597 // code will likely be branchless and use only a single multiplication
598 I32 i = SvIV (sv);
599 U32 u;
600 char digit, nz = 0;
601
602 need (enc, 6);
603
604 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
605 u = i < 0 ? -i : i;
606
607 // convert to 4.28 fixed-point representation
608 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
609
610 // now output digit by digit, each time masking out the integer part
611 // and multiplying by 5 while moving the decimal point one to the right,
612 // resulting in a net multiplication by 10.
613 // we always write the digit to memory but conditionally increment
614 // the pointer, to ease the usage of conditional move instructions.
615 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
616 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
617 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
618 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
619 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
620 }
451 } 621 }
452 else if (SvROK (sv)) 622 else if (SvROK (sv))
453 { 623 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)) 624 else if (!SvOK (sv))
470 encode_str (enc, "null", 4, 0); 625 encode_str (enc, "null", 4, 0);
471 else 626 else
472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 627 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv)); 628 SvPV_nolen (sv), SvFLAGS (sv));
474} 629}
475 630
476static SV * 631static SV *
477encode_json (SV *scalar, U32 flags) 632encode_json (SV *scalar, JSON *json)
478{ 633{
634 enc_t enc;
635
479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 636 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 637 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
481 638
482 enc_t enc; 639 enc.json = *json;
483 enc.flags = flags;
484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 640 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
485 enc.cur = SvPVX (enc.sv); 641 enc.cur = SvPVX (enc.sv);
486 enc.end = SvEND (enc.sv); 642 enc.end = SvEND (enc.sv);
487 enc.indent = 0; 643 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags); 644 enc.maxdepth = DEC_DEPTH (enc.json.flags);
489 645
490 SvPOK_only (enc.sv); 646 SvPOK_only (enc.sv);
491 encode_sv (&enc, scalar); 647 encode_sv (&enc, scalar);
492 648
649 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
650 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
651
493 if (!(flags & (F_ASCII | F_UTF8))) 652 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
494 SvUTF8_on (enc.sv); 653 SvUTF8_on (enc.sv);
495 654
496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497
498 if (enc.flags & F_SHRINK) 655 if (enc.json.flags & F_SHRINK)
499 shrink (enc.sv); 656 shrink (enc.sv);
500 657
501 return enc.sv; 658 return enc.sv;
502} 659}
503 660
… …
508typedef struct 665typedef struct
509{ 666{
510 char *cur; // current parser pointer 667 char *cur; // current parser pointer
511 char *end; // end of input string 668 char *end; // end of input string
512 const char *err; // parse error, if != 0 669 const char *err; // parse error, if != 0
513 U32 flags; // F_* 670 JSON json;
514 U32 depth; // recursion depth 671 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit 672 U32 maxdepth; // recursion depth limit
516} dec_t; 673} dec_t;
517 674
518static void 675inline void
519decode_ws (dec_t *dec) 676decode_ws (dec_t *dec)
520{ 677{
521 for (;;) 678 for (;;)
522 { 679 {
523 char ch = *dec->cur; 680 char ch = *dec->cur;
… …
549decode_4hex (dec_t *dec) 706decode_4hex (dec_t *dec)
550{ 707{
551 signed char d1, d2, d3, d4; 708 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur; 709 unsigned char *cur = (unsigned char *)dec->cur;
553 710
554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 711 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"); 712 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"); 713 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"); 714 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
558 715
559 dec->cur += 4; 716 dec->cur += 4;
560 717
561 return ((UV)d1) << 12 718 return ((UV)d1) << 12
562 | ((UV)d2) << 8 719 | ((UV)d2) << 8
… …
570static SV * 727static SV *
571decode_str (dec_t *dec) 728decode_str (dec_t *dec)
572{ 729{
573 SV *sv = 0; 730 SV *sv = 0;
574 int utf8 = 0; 731 int utf8 = 0;
732 char *dec_cur = dec->cur;
575 733
576 do 734 do
577 { 735 {
578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 736 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
579 char *cur = buf; 737 char *cur = buf;
580 738
581 do 739 do
582 { 740 {
583 unsigned char ch = *(unsigned char *)dec->cur++; 741 unsigned char ch = *(unsigned char *)dec_cur++;
584 742
585 if (ch == '"') 743 if (expect_false (ch == '"'))
586 { 744 {
587 --dec->cur; 745 --dec_cur;
588 break; 746 break;
589 } 747 }
590 else if (ch == '\\') 748 else if (expect_false (ch == '\\'))
591 { 749 {
592 switch (*dec->cur) 750 switch (*dec_cur)
593 { 751 {
594 case '\\': 752 case '\\':
595 case '/': 753 case '/':
596 case '"': *cur++ = *dec->cur++; break; 754 case '"': *cur++ = *dec_cur++; break;
597 755
598 case 'b': ++dec->cur; *cur++ = '\010'; break; 756 case 'b': ++dec_cur; *cur++ = '\010'; break;
599 case 't': ++dec->cur; *cur++ = '\011'; break; 757 case 't': ++dec_cur; *cur++ = '\011'; break;
600 case 'n': ++dec->cur; *cur++ = '\012'; break; 758 case 'n': ++dec_cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break; 759 case 'f': ++dec_cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break; 760 case 'r': ++dec_cur; *cur++ = '\015'; break;
603 761
604 case 'u': 762 case 'u':
605 { 763 {
606 UV lo, hi; 764 UV lo, hi;
607 ++dec->cur; 765 ++dec_cur;
608 766
767 dec->cur = dec_cur;
609 hi = decode_4hex (dec); 768 hi = decode_4hex (dec);
769 dec_cur = dec->cur;
610 if (hi == (UV)-1) 770 if (hi == (UV)-1)
611 goto fail; 771 goto fail;
612 772
613 // possibly a surrogate pair 773 // possibly a surrogate pair
614 if (hi >= 0xd800) 774 if (hi >= 0xd800)
615 if (hi < 0xdc00) 775 if (hi < 0xdc00)
616 { 776 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 777 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair"); 778 ERR ("missing low surrogate character in surrogate pair");
619 779
620 dec->cur += 2; 780 dec_cur += 2;
621 781
782 dec->cur = dec_cur;
622 lo = decode_4hex (dec); 783 lo = decode_4hex (dec);
784 dec_cur = dec->cur;
623 if (lo == (UV)-1) 785 if (lo == (UV)-1)
624 goto fail; 786 goto fail;
625 787
626 if (lo < 0xdc00 || lo >= 0xe000) 788 if (lo < 0xdc00 || lo >= 0xe000)
627 ERR ("surrogate pair expected"); 789 ERR ("surrogate pair expected");
… …
641 *cur++ = hi; 803 *cur++ = hi;
642 } 804 }
643 break; 805 break;
644 806
645 default: 807 default:
646 --dec->cur; 808 --dec_cur;
647 ERR ("illegal backslash escape sequence in string"); 809 ERR ("illegal backslash escape sequence in string");
648 } 810 }
649 } 811 }
650 else if (ch >= 0x20 && ch <= 0x7f) 812 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
651 *cur++ = ch; 813 *cur++ = ch;
652 else if (ch >= 0x80) 814 else if (ch >= 0x80)
653 { 815 {
654 --dec->cur;
655
656 STRLEN clen; 816 STRLEN clen;
817 UV uch;
818
819 --dec_cur;
820
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 821 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
658 if (clen == (STRLEN)-1) 822 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string"); 823 ERR ("malformed UTF-8 character in JSON string");
660 824
661 do 825 do
662 {
663 *cur++ = *dec->cur++; 826 *cur++ = *dec_cur++;
664 }
665 while (--clen); 827 while (--clen);
666 828
667 utf8 = 1; 829 utf8 = 1;
668 } 830 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else 831 else
832 {
833 --dec_cur;
834
835 if (!ch)
836 ERR ("unexpected end of string while parsing JSON string");
837 else
672 ERR ("invalid character encountered"); 838 ERR ("invalid character encountered while parsing JSON string");
673 839 }
674 } 840 }
675 while (cur < buf + SHORT_STRING_LEN); 841 while (cur < buf + SHORT_STRING_LEN);
676 842
843 {
677 STRLEN len = cur - buf; 844 STRLEN len = cur - buf;
678 845
679 if (sv) 846 if (sv)
680 { 847 {
681 SvGROW (sv, SvCUR (sv) + len + 1); 848 SvGROW (sv, SvCUR (sv) + len + 1);
682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 849 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
683 SvCUR_set (sv, SvCUR (sv) + len); 850 SvCUR_set (sv, SvCUR (sv) + len);
684 } 851 }
685 else 852 else
686 sv = newSVpvn (buf, len); 853 sv = newSVpvn (buf, len);
687 } 854 }
855 }
688 while (*dec->cur != '"'); 856 while (*dec_cur != '"');
689 857
690 ++dec->cur; 858 ++dec_cur;
691 859
692 if (sv) 860 if (sv)
693 { 861 {
694 SvPOK_only (sv); 862 SvPOK_only (sv);
695 *SvEND (sv) = 0; 863 *SvEND (sv) = 0;
… …
698 SvUTF8_on (sv); 866 SvUTF8_on (sv);
699 } 867 }
700 else 868 else
701 sv = newSVpvn ("", 0); 869 sv = newSVpvn ("", 0);
702 870
871 dec->cur = dec_cur;
703 return sv; 872 return sv;
704 873
705fail: 874fail:
875 dec->cur = dec_cur;
706 return 0; 876 return 0;
707} 877}
708 878
709static SV * 879static SV *
710decode_num (dec_t *dec) 880decode_num (dec_t *dec)
… …
768 is_nv = 1; 938 is_nv = 1;
769 } 939 }
770 940
771 if (!is_nv) 941 if (!is_nv)
772 { 942 {
773 UV uv; 943 int len = dec->cur - start;
774 int numtype = grok_number (start, dec->cur - start, &uv); 944
775 if (numtype & IS_NUMBER_IN_UV) 945 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
776 if (numtype & IS_NUMBER_NEG) 946 if (*start == '-')
947 switch (len)
777 { 948 {
778 if (uv < (UV)IV_MIN) 949 case 2: return newSViv (-( start [1] - '0' * 1));
779 return newSViv (-(IV)uv); 950 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
951 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
952 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
780 } 953 }
954 else
955 switch (len)
956 {
957 case 1: return newSViv ( start [0] - '0' * 1);
958 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
959 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
960 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
961 }
962
963 {
964 UV uv;
965 int numtype = grok_number (start, len, &uv);
966 if (numtype & IS_NUMBER_IN_UV)
967 if (numtype & IS_NUMBER_NEG)
968 {
969 if (uv < (UV)IV_MIN)
970 return newSViv (-(IV)uv);
971 }
781 else 972 else
782 return newSVuv (uv); 973 return newSVuv (uv);
783 } 974 }
784 975
976 len -= *start == '-' ? 1 : 0;
977
978 // does not fit into IV or UV, try NV
979 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
980 #if defined (LDBL_DIG)
981 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
982 #endif
983 )
984 // fits into NV without loss of precision
985 return newSVnv (Atof (start));
986
987 // everything else fails, convert it to a string
988 return newSVpvn (start, dec->cur - start);
989 }
990
991 // loss of precision here
785 return newSVnv (Atof (start)); 992 return newSVnv (Atof (start));
786 993
787fail: 994fail:
788 return 0; 995 return 0;
789} 996}
… …
833} 1040}
834 1041
835static SV * 1042static SV *
836decode_hv (dec_t *dec) 1043decode_hv (dec_t *dec)
837{ 1044{
1045 SV *sv;
838 HV *hv = newHV (); 1046 HV *hv = newHV ();
839 1047
840 DEC_INC_DEPTH; 1048 DEC_INC_DEPTH;
841 decode_ws (dec); 1049 decode_ws (dec);
842 1050
843 if (*dec->cur == '}') 1051 if (*dec->cur == '}')
844 ++dec->cur; 1052 ++dec->cur;
845 else 1053 else
846 for (;;) 1054 for (;;)
847 { 1055 {
848 SV *key, *value;
849
850 decode_ws (dec); EXPECT_CH ('"'); 1056 decode_ws (dec); EXPECT_CH ('"');
851 1057
852 key = decode_str (dec); 1058 // heuristic: assume that
853 if (!key) 1059 // a) decode_str + hv_store_ent are abysmally slow.
854 goto fail; 1060 // b) most hash keys are short, simple ascii text.
1061 // => try to "fast-match" such strings to avoid
1062 // the overhead of decode_str + hv_store_ent.
1063 {
1064 SV *value;
1065 char *p = dec->cur;
1066 char *e = p + 24; // only try up to 24 bytes
855 1067
856 decode_ws (dec); EXPECT_CH (':'); 1068 for (;;)
857
858 value = decode_sv (dec);
859 if (!value)
860 { 1069 {
1070 // the >= 0x80 is true on most architectures
1071 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1072 {
1073 // slow path, back up and use decode_str
1074 SV *key = decode_str (dec);
1075 if (!key)
1076 goto fail;
1077
1078 decode_ws (dec); EXPECT_CH (':');
1079
1080 value = decode_sv (dec);
1081 if (!value)
1082 {
1083 SvREFCNT_dec (key);
1084 goto fail;
1085 }
1086
1087 hv_store_ent (hv, key, value, 0);
861 SvREFCNT_dec (key); 1088 SvREFCNT_dec (key);
1089
1090 break;
1091 }
1092 else if (*p == '"')
1093 {
1094 // fast path, got a simple key
1095 char *key = dec->cur;
1096 int len = p - key;
1097 dec->cur = p + 1;
1098
1099 decode_ws (dec); EXPECT_CH (':');
1100
1101 value = decode_sv (dec);
1102 if (!value)
862 goto fail; 1103 goto fail;
1104
1105 hv_store (hv, key, len, value, 0);
1106
1107 break;
1108 }
1109
1110 ++p;
863 } 1111 }
864 1112 }
865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
867 1113
868 decode_ws (dec); 1114 decode_ws (dec);
869 1115
870 if (*dec->cur == '}') 1116 if (*dec->cur == '}')
871 { 1117 {
… …
878 1124
879 ++dec->cur; 1125 ++dec->cur;
880 } 1126 }
881 1127
882 DEC_DEC_DEPTH; 1128 DEC_DEC_DEPTH;
883 return newRV_noinc ((SV *)hv); 1129 sv = newRV_noinc ((SV *)hv);
1130
1131 // check filter callbacks
1132 if (dec->json.flags & F_HOOK)
1133 {
1134 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1135 {
1136 HE *cb, *he;
1137
1138 hv_iterinit (hv);
1139 he = hv_iternext (hv);
1140 hv_iterinit (hv);
1141
1142 // the next line creates a mortal sv each time its called.
1143 // might want to optimise this for common cases.
1144 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1145
1146 if (cb)
1147 {
1148 dSP;
1149 int count;
1150
1151 ENTER; SAVETMPS; PUSHMARK (SP);
1152 XPUSHs (HeVAL (he));
1153
1154 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1155
1156 if (count == 1)
1157 {
1158 sv = newSVsv (POPs);
1159 FREETMPS; LEAVE;
1160 return sv;
1161 }
1162
1163 FREETMPS; LEAVE;
1164 }
1165 }
1166
1167 if (dec->json.cb_object)
1168 {
1169 dSP;
1170 int count;
1171
1172 ENTER; SAVETMPS; PUSHMARK (SP);
1173 XPUSHs (sv_2mortal (sv));
1174
1175 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1176
1177 if (count == 1)
1178 {
1179 sv = newSVsv (POPs);
1180 FREETMPS; LEAVE;
1181 return sv;
1182 }
1183
1184 SvREFCNT_inc (sv);
1185 FREETMPS; LEAVE;
1186 }
1187 }
1188
1189 return sv;
884 1190
885fail: 1191fail:
886 SvREFCNT_dec (hv); 1192 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH; 1193 DEC_DEC_DEPTH;
888 return 0; 1194 return 0;
… …
890 1196
891static SV * 1197static SV *
892decode_sv (dec_t *dec) 1198decode_sv (dec_t *dec)
893{ 1199{
894 decode_ws (dec); 1200 decode_ws (dec);
1201
1202 // the beauty of JSON: you need exactly one character lookahead
1203 // to parse anything.
895 switch (*dec->cur) 1204 switch (*dec->cur)
896 { 1205 {
897 case '"': ++dec->cur; return decode_str (dec); 1206 case '"': ++dec->cur; return decode_str (dec);
898 case '[': ++dec->cur; return decode_av (dec); 1207 case '[': ++dec->cur; return decode_av (dec);
899 case '{': ++dec->cur; return decode_hv (dec); 1208 case '{': ++dec->cur; return decode_hv (dec);
… …
905 1214
906 case 't': 1215 case 't':
907 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1216 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
908 { 1217 {
909 dec->cur += 4; 1218 dec->cur += 4;
910 return newSViv (1); 1219 return SvREFCNT_inc (json_true);
911 } 1220 }
912 else 1221 else
913 ERR ("'true' expected"); 1222 ERR ("'true' expected");
914 1223
915 break; 1224 break;
916 1225
917 case 'f': 1226 case 'f':
918 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1227 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
919 { 1228 {
920 dec->cur += 5; 1229 dec->cur += 5;
921 return newSViv (0); 1230 return SvREFCNT_inc (json_false);
922 } 1231 }
923 else 1232 else
924 ERR ("'false' expected"); 1233 ERR ("'false' expected");
925 1234
926 break; 1235 break;
… …
935 ERR ("'null' expected"); 1244 ERR ("'null' expected");
936 1245
937 break; 1246 break;
938 1247
939 default: 1248 default:
940 ERR ("malformed json string, neither array, object, number, string or atom"); 1249 ERR ("malformed JSON string, neither array, object, number, string or atom");
941 break; 1250 break;
942 } 1251 }
943 1252
944fail: 1253fail:
945 return 0; 1254 return 0;
946} 1255}
947 1256
948static SV * 1257static SV *
949decode_json (SV *string, U32 flags) 1258decode_json (SV *string, JSON *json, UV *offset_return)
950{ 1259{
1260 dec_t dec;
1261 UV offset;
951 SV *sv; 1262 SV *sv;
952 1263
1264 SvGETMAGIC (string);
1265 SvUPGRADE (string, SVt_PV);
1266
1267 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1268 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1269 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1270
953 if (flags & F_UTF8) 1271 if (json->flags & F_UTF8)
954 sv_utf8_downgrade (string, 0); 1272 sv_utf8_downgrade (string, 0);
955 else 1273 else
956 sv_utf8_upgrade (string); 1274 sv_utf8_upgrade (string);
957 1275
958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1276 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
959 1277
960 dec_t dec; 1278 dec.json = *json;
961 dec.flags = flags;
962 dec.cur = SvPVX (string); 1279 dec.cur = SvPVX (string);
963 dec.end = SvEND (string); 1280 dec.end = SvEND (string);
964 dec.err = 0; 1281 dec.err = 0;
965 dec.depth = 0; 1282 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags); 1283 dec.maxdepth = DEC_DEPTH (dec.json.flags);
967 1284
968 *SvEND (string) = 0; // this should basically be a nop, too 1285 if (dec.json.cb_object || dec.json.cb_sk_object)
1286 dec.json.flags |= F_HOOK;
1287
1288 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
969 sv = decode_sv (&dec); 1289 sv = decode_sv (&dec);
970 1290
1291 if (!(offset_return || !sv))
1292 {
1293 // check for trailing garbage
1294 decode_ws (&dec);
1295
1296 if (*dec.cur)
1297 {
1298 dec.err = "garbage after JSON object";
1299 SvREFCNT_dec (sv);
1300 sv = 0;
1301 }
1302 }
1303
1304 if (offset_return || !sv)
1305 {
1306 offset = dec.json.flags & F_UTF8
1307 ? dec.cur - SvPVX (string)
1308 : utf8_distance (dec.cur, SvPVX (string));
1309
1310 if (offset_return)
1311 *offset_return = offset;
1312 }
1313
971 if (!sv) 1314 if (!sv)
972 { 1315 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
975 : utf8_distance (dec.cur, SvPVX (string));
976 SV *uni = sv_newmortal (); 1316 SV *uni = sv_newmortal ();
977 1317
978 // horrible hack to silence warning inside pv_uni_display 1318 // horrible hack to silence warning inside pv_uni_display
979 COP cop = *PL_curcop; 1319 COP cop = *PL_curcop;
980 cop.cop_warnings = pWARN_NONE; 1320 cop.cop_warnings = pWARN_NONE;
… …
982 SAVEVPTR (PL_curcop); 1322 SAVEVPTR (PL_curcop);
983 PL_curcop = &cop; 1323 PL_curcop = &cop;
984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1324 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE; 1325 LEAVE;
986 1326
987 croak ("%s, at character offset %d (%s)", 1327 croak ("%s, at character offset %d [\"%s\"]",
988 dec.err, 1328 dec.err,
989 (int)offset, 1329 (int)offset,
990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1330 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
991 } 1331 }
992 1332
993 sv = sv_2mortal (sv); 1333 sv = sv_2mortal (sv);
994 1334
995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1335 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)"); 1336 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
997 1337
998 return sv; 1338 return sv;
999} 1339}
1000 1340
… …
1004MODULE = JSON::XS PACKAGE = JSON::XS 1344MODULE = JSON::XS PACKAGE = JSON::XS
1005 1345
1006BOOT: 1346BOOT:
1007{ 1347{
1008 int i; 1348 int i;
1009
1010 memset (decode_hexdigit, 0xff, 256);
1011 1349
1012 for (i = 0; i < 256; ++i) 1350 for (i = 0; i < 256; ++i)
1013 decode_hexdigit [i] = 1351 decode_hexdigit [i] =
1014 i >= '0' && i <= '9' ? i - '0' 1352 i >= '0' && i <= '9' ? i - '0'
1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1353 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1354 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1017 : -1; 1355 : -1;
1018 1356
1019 json_stash = gv_stashpv ("JSON::XS", 1); 1357 json_stash = gv_stashpv ("JSON::XS" , 1);
1358 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1359
1360 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1361 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1020} 1362}
1021 1363
1022PROTOTYPES: DISABLE 1364PROTOTYPES: DISABLE
1023 1365
1024SV *new (char *dummy) 1366void new (char *klass)
1025 CODE: 1367 PPCODE:
1026 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1368{
1027 OUTPUT: 1369 SV *pv = NEWSV (0, sizeof (JSON));
1028 RETVAL 1370 SvPOK_only (pv);
1371 Zero (SvPVX (pv), 1, JSON);
1372 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1373 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1374}
1029 1375
1030SV *ascii (SV *self, int enable = 1) 1376void ascii (JSON *self, int enable = 1)
1031 ALIAS: 1377 ALIAS:
1032 ascii = F_ASCII 1378 ascii = F_ASCII
1379 latin1 = F_LATIN1
1033 utf8 = F_UTF8 1380 utf8 = F_UTF8
1034 indent = F_INDENT 1381 indent = F_INDENT
1035 canonical = F_CANONICAL 1382 canonical = F_CANONICAL
1036 space_before = F_SPACE_BEFORE 1383 space_before = F_SPACE_BEFORE
1037 space_after = F_SPACE_AFTER 1384 space_after = F_SPACE_AFTER
1038 pretty = F_PRETTY 1385 pretty = F_PRETTY
1039 allow_nonref = F_ALLOW_NONREF 1386 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK 1387 shrink = F_SHRINK
1388 allow_blessed = F_ALLOW_BLESSED
1389 convert_blessed = F_CONV_BLESSED
1041 CODE: 1390 PPCODE:
1042{ 1391{
1043 UV *uv = SvJSON (self);
1044 if (enable) 1392 if (enable)
1045 *uv |= ix; 1393 self->flags |= ix;
1046 else 1394 else
1047 *uv &= ~ix; 1395 self->flags &= ~ix;
1048 1396
1049 RETVAL = newSVsv (self); 1397 XPUSHs (ST (0));
1050} 1398}
1051 OUTPUT:
1052 RETVAL
1053 1399
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1400void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1055 CODE: 1401 PPCODE:
1056{ 1402{
1057 UV *uv = SvJSON (self);
1058 UV log2 = 0; 1403 UV log2 = 0;
1059 1404
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1405 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1061 1406
1062 while ((1UL << log2) < max_depth) 1407 while ((1UL << log2) < max_depth)
1063 ++log2; 1408 ++log2;
1064 1409
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1410 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066 1411
1067 RETVAL = newSVsv (self); 1412 XPUSHs (ST (0));
1068} 1413}
1069 OUTPUT:
1070 RETVAL
1071 1414
1072void encode (SV *self, SV *scalar) 1415void max_size (JSON *self, UV max_size = 0)
1073 PPCODE: 1416 PPCODE:
1074 XPUSHs (encode_json (scalar, *SvJSON (self))); 1417{
1418 UV log2 = 0;
1075 1419
1076void decode (SV *self, SV *jsonstr) 1420 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1421 if (max_size == 1) max_size = 2;
1422
1423 while ((1UL << log2) < max_size)
1424 ++log2;
1425
1426 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1427
1428 XPUSHs (ST (0));
1429}
1430
1431void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1432 PPCODE:
1433{
1434 SvREFCNT_dec (self->cb_object);
1435 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1436
1437 XPUSHs (ST (0));
1438}
1439
1440void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1441 PPCODE:
1442{
1443 if (!self->cb_sk_object)
1444 self->cb_sk_object = newHV ();
1445
1446 if (SvOK (cb))
1447 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1448 else
1449 {
1450 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1451
1452 if (!HvKEYS (self->cb_sk_object))
1453 {
1454 SvREFCNT_dec (self->cb_sk_object);
1455 self->cb_sk_object = 0;
1456 }
1457 }
1458
1459 XPUSHs (ST (0));
1460}
1461
1462void encode (JSON *self, SV *scalar)
1463 PPCODE:
1464 XPUSHs (encode_json (scalar, self));
1465
1466void decode (JSON *self, SV *jsonstr)
1467 PPCODE:
1078 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1468 XPUSHs (decode_json (jsonstr, self, 0));
1469
1470void decode_prefix (JSON *self, SV *jsonstr)
1471 PPCODE:
1472{
1473 UV offset;
1474 EXTEND (SP, 2);
1475 PUSHs (decode_json (jsonstr, self, &offset));
1476 PUSHs (sv_2mortal (newSVuv (offset)));
1477}
1478
1479void DESTROY (JSON *self)
1480 CODE:
1481 SvREFCNT_dec (self->cb_sk_object);
1482 SvREFCNT_dec (self->cb_object);
1079 1483
1080PROTOTYPES: ENABLE 1484PROTOTYPES: ENABLE
1081 1485
1082void to_json (SV *scalar) 1486void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
1085 PPCODE: 1487 PPCODE:
1488{
1489 JSON json = { F_DEFAULT | F_UTF8 };
1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1490 XPUSHs (encode_json (scalar, &json));
1491}
1087 1492
1088void from_json (SV *jsonstr) 1493void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
1091 PPCODE: 1494 PPCODE:
1495{
1496 JSON json = { F_DEFAULT | F_UTF8 };
1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1497 XPUSHs (decode_json (jsonstr, &json, 0));
1498}
1093 1499

Diff Legend

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