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 1.20 by root, Sun Mar 25 21:53:19 2007 UTC vs.
Revision 1.59 by root, Mon Aug 13 16:14:20 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# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else
67# define JSON_SLOW 0
68# define JSON_STASH json_stash
69#endif
70
37static HV *json_stash; // JSON::XS:: 71static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false;
73
74typedef struct {
75 U32 flags;
76 SV *cb_object;
77 HV *cb_sk_object;
78} JSON;
38 79
39///////////////////////////////////////////////////////////////////////////// 80/////////////////////////////////////////////////////////////////////////////
40// utility functions 81// utility functions
41 82
42static UV * 83inline 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) 84shrink (SV *sv)
53{ 85{
54 sv_utf8_downgrade (sv, 1); 86 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1) 87 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 88 {
… …
65// decode an utf-8 character and return it, or (UV)-1 in 97// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error. 98// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 99// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 100// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 101// note that we never call this function for a ascii codepoints
70static UV 102inline UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 104{
73 if (s[0] > 0xdf || s[0] < 0xc2) 105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 108 {
77 *clen = 2; 109 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 111 }
80 else 112 else
113 {
114 *clen = (STRLEN)-1;
81 return (UV)-1; 115 return (UV)-1;
116 }
82} 117}
83 118
84///////////////////////////////////////////////////////////////////////////// 119/////////////////////////////////////////////////////////////////////////////
85// encoder 120// encoder
86 121
… …
88typedef struct 123typedef struct
89{ 124{
90 char *cur; // SvPVX (sv) + current output position 125 char *cur; // SvPVX (sv) + current output position
91 char *end; // SvEND (sv) 126 char *end; // SvEND (sv)
92 SV *sv; // result scalar 127 SV *sv; // result scalar
93 U32 flags; // F_* 128 JSON json;
94 U32 indent; // indentation level 129 U32 indent; // indentation level
95 U32 maxdepth; // max. indentation/recursion level 130 U32 maxdepth; // max. indentation/recursion level
96} enc_t; 131} enc_t;
97 132
98static void 133inline void
99need (enc_t *enc, STRLEN len) 134need (enc_t *enc, STRLEN len)
100{ 135{
101 if (enc->cur + len >= enc->end) 136 if (expect_false (enc->cur + len >= enc->end))
102 { 137 {
103 STRLEN cur = enc->cur - SvPVX (enc->sv); 138 STRLEN cur = enc->cur - SvPVX (enc->sv);
104 SvGROW (enc->sv, cur + len + 1); 139 SvGROW (enc->sv, cur + len + 1);
105 enc->cur = SvPVX (enc->sv) + cur; 140 enc->cur = SvPVX (enc->sv) + cur;
106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107 } 142 }
108} 143}
109 144
110static void 145inline void
111encode_ch (enc_t *enc, char ch) 146encode_ch (enc_t *enc, char ch)
112{ 147{
113 need (enc, 1); 148 need (enc, 1);
114 *enc->cur++ = ch; 149 *enc->cur++ = ch;
115} 150}
… …
123 158
124 while (str < end) 159 while (str < end)
125 { 160 {
126 unsigned char ch = *(unsigned char *)str; 161 unsigned char ch = *(unsigned char *)str;
127 162
128 if (ch >= 0x20 && ch < 0x80) // most common case 163 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129 { 164 {
130 if (ch == '"') // but with slow exceptions 165 if (expect_false (ch == '"')) // but with slow exceptions
131 { 166 {
132 need (enc, len += 1); 167 need (enc, len += 1);
133 *enc->cur++ = '\\'; 168 *enc->cur++ = '\\';
134 *enc->cur++ = '"'; 169 *enc->cur++ = '"';
135 } 170 }
136 else if (ch == '\\') 171 else if (expect_false (ch == '\\'))
137 { 172 {
138 need (enc, len += 1); 173 need (enc, len += 1);
139 *enc->cur++ = '\\'; 174 *enc->cur++ = '\\';
140 *enc->cur++ = '\\'; 175 *enc->cur++ = '\\';
141 } 176 }
… …
159 STRLEN clen; 194 STRLEN clen;
160 UV uch; 195 UV uch;
161 196
162 if (is_utf8) 197 if (is_utf8)
163 { 198 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen); 199 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1) 200 if (clen == (STRLEN)-1)
167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 201 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
168 } 202 }
169 else 203 else
… …
173 } 207 }
174 208
175 if (uch > 0x10FFFFUL) 209 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177 211
178 if (uch < 0x80 || enc->flags & F_ASCII) 212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
179 { 213 {
180 if (uch > 0xFFFFUL) 214 if (uch > 0xFFFFUL)
181 { 215 {
182 need (enc, len += 11); 216 need (enc, len += 11);
183 sprintf (enc->cur, "\\u%04x\\u%04x", 217 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
197 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 231 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 } 232 }
199 233
200 str += clen; 234 str += clen;
201 } 235 }
236 else if (enc->json.flags & F_LATIN1)
237 {
238 *enc->cur++ = uch;
239 str += clen;
240 }
202 else if (is_utf8) 241 else if (is_utf8)
203 { 242 {
204 need (enc, len += clen); 243 need (enc, len += clen);
205 do 244 do
206 { 245 {
… …
208 } 247 }
209 while (--clen); 248 while (--clen);
210 } 249 }
211 else 250 else
212 { 251 {
213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 252 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str; 254 ++str;
216 } 255 }
217 } 256 }
218 } 257 }
… …
220 259
221 --len; 260 --len;
222 } 261 }
223} 262}
224 263
225static void 264inline void
226encode_indent (enc_t *enc) 265encode_indent (enc_t *enc)
227{ 266{
228 if (enc->flags & F_INDENT) 267 if (enc->json.flags & F_INDENT)
229 { 268 {
230 int spaces = enc->indent * INDENT_STEP; 269 int spaces = enc->indent * INDENT_STEP;
231 270
232 need (enc, spaces); 271 need (enc, spaces);
233 memset (enc->cur, ' ', spaces); 272 memset (enc->cur, ' ', spaces);
234 enc->cur += spaces; 273 enc->cur += spaces;
235 } 274 }
236} 275}
237 276
238static void 277inline void
239encode_space (enc_t *enc) 278encode_space (enc_t *enc)
240{ 279{
241 need (enc, 1); 280 need (enc, 1);
242 encode_ch (enc, ' '); 281 encode_ch (enc, ' ');
243} 282}
244 283
245static void 284inline void
246encode_nl (enc_t *enc) 285encode_nl (enc_t *enc)
247{ 286{
248 if (enc->flags & F_INDENT) 287 if (enc->json.flags & F_INDENT)
249 { 288 {
250 need (enc, 1); 289 need (enc, 1);
251 encode_ch (enc, '\n'); 290 encode_ch (enc, '\n');
252 } 291 }
253} 292}
254 293
255static void 294inline void
256encode_comma (enc_t *enc) 295encode_comma (enc_t *enc)
257{ 296{
258 encode_ch (enc, ','); 297 encode_ch (enc, ',');
259 298
260 if (enc->flags & F_INDENT) 299 if (enc->json.flags & F_INDENT)
261 encode_nl (enc); 300 encode_nl (enc);
262 else if (enc->flags & F_SPACE_AFTER) 301 else if (enc->json.flags & F_SPACE_AFTER)
263 encode_space (enc); 302 encode_space (enc);
264} 303}
265 304
266static void encode_sv (enc_t *enc, SV *sv); 305static void encode_sv (enc_t *enc, SV *sv);
267 306
268static void 307static void
269encode_av (enc_t *enc, AV *av) 308encode_av (enc_t *enc, AV *av)
270{ 309{
271 int i, len = av_len (av); 310 int i, len = av_len (av);
272 311
312 if (enc->indent >= enc->maxdepth)
313 croak ("data structure too deep (hit recursion limit)");
314
273 encode_ch (enc, '['); encode_nl (enc); 315 encode_ch (enc, '['); encode_nl (enc);
274 ++enc->indent; 316 ++enc->indent;
275 317
276 for (i = 0; i <= len; ++i) 318 for (i = 0; i <= len; ++i)
277 { 319 {
320 SV **svp = av_fetch (av, i, 0);
321
278 encode_indent (enc); 322 encode_indent (enc);
279 encode_sv (enc, *av_fetch (av, i, 0)); 323
324 if (svp)
325 encode_sv (enc, *svp);
326 else
327 encode_str (enc, "null", 4, 0);
280 328
281 if (i < len) 329 if (i < len)
282 encode_comma (enc); 330 encode_comma (enc);
283 } 331 }
284 332
… …
307 else 355 else
308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 356 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
309 357
310 encode_ch (enc, '"'); 358 encode_ch (enc, '"');
311 359
312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 360 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
313 encode_ch (enc, ':'); 361 encode_ch (enc, ':');
314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 362 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
315 encode_sv (enc, HeVAL (he)); 363 encode_sv (enc, HeVAL (he));
316} 364}
317 365
318// compare hash entries, used when all keys are bytestrings 366// compare hash entries, used when all keys are bytestrings
319static int 367static int
… …
342 390
343static void 391static void
344encode_hv (enc_t *enc, HV *hv) 392encode_hv (enc_t *enc, HV *hv)
345{ 393{
346 int count, i; 394 int count, i;
395
396 if (enc->indent >= enc->maxdepth)
397 croak ("data structure too deep (hit recursion limit)");
347 398
348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
349 400
350 if ((count = hv_iterinit (hv))) 401 if ((count = hv_iterinit (hv)))
351 { 402 {
352 // for canonical output we have to sort by keys first 403 // for canonical output we have to sort by keys first
353 // actually, this is mostly due to the stupid so-called 404 // actually, this is mostly due to the stupid so-called
354 // security workaround added somewhere in 5.8.x. 405 // security workaround added somewhere in 5.8.x.
355 // that randomises hash orderings 406 // that randomises hash orderings
356 if (enc->flags & F_CANONICAL) 407 if (enc->json.flags & F_CANONICAL)
357 { 408 {
358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
359 int fast = 1; 409 int fast = 1;
410 HE *he;
411#if defined(__BORLANDC__) || defined(_MSC_VER)
412 HE **hes = _alloca (count * sizeof (HE));
413#else
414 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
415#endif
360 416
361 i = 0; 417 i = 0;
362 while ((he = hv_iternext (hv))) 418 while ((he = hv_iternext (hv)))
363 { 419 {
364 hes [i++] = he; 420 hes [i++] = he;
… …
399 455
400 encode_nl (enc); 456 encode_nl (enc);
401 } 457 }
402 else 458 else
403 { 459 {
404 SV *sv;
405 HE *he = hv_iternext (hv); 460 HE *he = hv_iternext (hv);
406 461
407 for (;;) 462 for (;;)
408 { 463 {
409 encode_indent (enc); 464 encode_indent (enc);
… …
418 encode_nl (enc); 473 encode_nl (enc);
419 } 474 }
420 } 475 }
421 476
422 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 477 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
478}
479
480// encode objects, arrays and special \0=false and \1=true values.
481static void
482encode_rv (enc_t *enc, SV *sv)
483{
484 svtype svt;
485
486 SvGETMAGIC (sv);
487 svt = SvTYPE (sv);
488
489 if (expect_false (SvOBJECT (sv)))
490 {
491 HV *stash = !JSON_SLOW || json_boolean_stash
492 ? json_boolean_stash
493 : gv_stashpv ("JSON::XS::Boolean", 1);
494
495 if (SvSTASH (sv) == stash)
496 {
497 if (SvIV (sv))
498 encode_str (enc, "true", 4, 0);
499 else
500 encode_str (enc, "false", 5, 0);
501 }
502 else
503 {
504#if 0
505 if (0 && sv_derived_from (rv, "JSON::Literal"))
506 {
507 // not yet
508 }
509#endif
510 if (enc->json.flags & F_CONV_BLESSED)
511 {
512 // we re-bless the reference to get overload and other niceties right
513 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
514
515 if (to_json)
516 {
517 dSP;
518
519 ENTER; SAVETMPS; PUSHMARK (SP);
520 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
521
522 // calling with G_SCALAR ensures that we always get a 1 return value
523 PUTBACK;
524 call_sv ((SV *)GvCV (to_json), G_SCALAR);
525 SPAGAIN;
526
527 // catch this surprisingly common error
528 if (SvROK (TOPs) && SvRV (TOPs) == sv)
529 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
530
531 sv = POPs;
532 PUTBACK;
533
534 encode_sv (enc, sv);
535
536 FREETMPS; LEAVE;
537 }
538 else if (enc->json.flags & F_ALLOW_BLESSED)
539 encode_str (enc, "null", 4, 0);
540 else
541 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
543 }
544 else if (enc->json.flags & F_ALLOW_BLESSED)
545 encode_str (enc, "null", 4, 0);
546 else
547 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
548 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
549 }
550 }
551 else if (svt == SVt_PVHV)
552 encode_hv (enc, (HV *)sv);
553 else if (svt == SVt_PVAV)
554 encode_av (enc, (AV *)sv);
555 else if (svt < SVt_PVAV)
556 {
557 STRLEN len = 0;
558 char *pv = svt ? SvPV (sv, len) : 0;
559
560 if (len == 1 && *pv == '1')
561 encode_str (enc, "true", 4, 0);
562 else if (len == 1 && *pv == '0')
563 encode_str (enc, "false", 5, 0);
564 else
565 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
566 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
567 }
568 else
569 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
570 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
423} 571}
424 572
425static void 573static void
426encode_sv (enc_t *enc, SV *sv) 574encode_sv (enc_t *enc, SV *sv)
427{ 575{
… …
435 encode_str (enc, str, len, SvUTF8 (sv)); 583 encode_str (enc, str, len, SvUTF8 (sv));
436 encode_ch (enc, '"'); 584 encode_ch (enc, '"');
437 } 585 }
438 else if (SvNOKp (sv)) 586 else if (SvNOKp (sv))
439 { 587 {
588 // trust that perl will do the right thing w.r.t. JSON syntax.
440 need (enc, NV_DIG + 32); 589 need (enc, NV_DIG + 32);
441 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 590 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
442 enc->cur += strlen (enc->cur); 591 enc->cur += strlen (enc->cur);
443 } 592 }
444 else if (SvIOKp (sv)) 593 else if (SvIOKp (sv))
445 { 594 {
446 need (enc, 64); 595 // we assume we can always read an IV as a UV
596 if (SvUV (sv) & ~(UV)0x7fff)
597 {
598 // large integer, use the (rather slow) snprintf way.
599 need (enc, sizeof (UV) * 3);
447 enc->cur += 600 enc->cur +=
448 SvIsUV(sv) 601 SvIsUV(sv)
449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 602 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 603 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
604 }
605 else
606 {
607 // optimise the "small number case"
608 // code will likely be branchless and use only a single multiplication
609 I32 i = SvIV (sv);
610 U32 u;
611 char digit, nz = 0;
612
613 need (enc, 6);
614
615 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
616 u = i < 0 ? -i : i;
617
618 // convert to 4.28 fixed-point representation
619 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
620
621 // now output digit by digit, each time masking out the integer part
622 // and multiplying by 5 while moving the decimal point one to the right,
623 // resulting in a net multiplication by 10.
624 // we always write the digit to memory but conditionally increment
625 // the pointer, to ease the usage of conditional move instructions.
626 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
627 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
628 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
629 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
630 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
631 }
451 } 632 }
452 else if (SvROK (sv)) 633 else if (SvROK (sv))
453 { 634 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)) 635 else if (!SvOK (sv))
470 encode_str (enc, "null", 4, 0); 636 encode_str (enc, "null", 4, 0);
471 else 637 else
472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 638 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv)); 639 SvPV_nolen (sv), SvFLAGS (sv));
474} 640}
475 641
476static SV * 642static SV *
477encode_json (SV *scalar, U32 flags) 643encode_json (SV *scalar, JSON *json)
478{ 644{
645 enc_t enc;
646
479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 647 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 648 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
481 649
482 enc_t enc; 650 enc.json = *json;
483 enc.flags = flags;
484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 651 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
485 enc.cur = SvPVX (enc.sv); 652 enc.cur = SvPVX (enc.sv);
486 enc.end = SvEND (enc.sv); 653 enc.end = SvEND (enc.sv);
487 enc.indent = 0; 654 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags); 655 enc.maxdepth = DEC_DEPTH (enc.json.flags);
489 656
490 SvPOK_only (enc.sv); 657 SvPOK_only (enc.sv);
491 encode_sv (&enc, scalar); 658 encode_sv (&enc, scalar);
492 659
660 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
661 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
662
493 if (!(flags & (F_ASCII | F_UTF8))) 663 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
494 SvUTF8_on (enc.sv); 664 SvUTF8_on (enc.sv);
495 665
496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497
498 if (enc.flags & F_SHRINK) 666 if (enc.json.flags & F_SHRINK)
499 shrink (enc.sv); 667 shrink (enc.sv);
500 668
501 return enc.sv; 669 return enc.sv;
502} 670}
503 671
… …
508typedef struct 676typedef struct
509{ 677{
510 char *cur; // current parser pointer 678 char *cur; // current parser pointer
511 char *end; // end of input string 679 char *end; // end of input string
512 const char *err; // parse error, if != 0 680 const char *err; // parse error, if != 0
513 U32 flags; // F_* 681 JSON json;
514 U32 depth; // recursion depth 682 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit 683 U32 maxdepth; // recursion depth limit
516} dec_t; 684} dec_t;
517 685
518static void 686inline void
519decode_ws (dec_t *dec) 687decode_ws (dec_t *dec)
520{ 688{
521 for (;;) 689 for (;;)
522 { 690 {
523 char ch = *dec->cur; 691 char ch = *dec->cur;
… …
549decode_4hex (dec_t *dec) 717decode_4hex (dec_t *dec)
550{ 718{
551 signed char d1, d2, d3, d4; 719 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur; 720 unsigned char *cur = (unsigned char *)dec->cur;
553 721
554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 722 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"); 723 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"); 724 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"); 725 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
558 726
559 dec->cur += 4; 727 dec->cur += 4;
560 728
561 return ((UV)d1) << 12 729 return ((UV)d1) << 12
562 | ((UV)d2) << 8 730 | ((UV)d2) << 8
… …
570static SV * 738static SV *
571decode_str (dec_t *dec) 739decode_str (dec_t *dec)
572{ 740{
573 SV *sv = 0; 741 SV *sv = 0;
574 int utf8 = 0; 742 int utf8 = 0;
743 char *dec_cur = dec->cur;
575 744
576 do 745 do
577 { 746 {
578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 747 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
579 char *cur = buf; 748 char *cur = buf;
580 749
581 do 750 do
582 { 751 {
583 unsigned char ch = *(unsigned char *)dec->cur++; 752 unsigned char ch = *(unsigned char *)dec_cur++;
584 753
585 if (ch == '"') 754 if (expect_false (ch == '"'))
586 { 755 {
587 --dec->cur; 756 --dec_cur;
588 break; 757 break;
589 } 758 }
590 else if (ch == '\\') 759 else if (expect_false (ch == '\\'))
591 { 760 {
592 switch (*dec->cur) 761 switch (*dec_cur)
593 { 762 {
594 case '\\': 763 case '\\':
595 case '/': 764 case '/':
596 case '"': *cur++ = *dec->cur++; break; 765 case '"': *cur++ = *dec_cur++; break;
597 766
598 case 'b': ++dec->cur; *cur++ = '\010'; break; 767 case 'b': ++dec_cur; *cur++ = '\010'; break;
599 case 't': ++dec->cur; *cur++ = '\011'; break; 768 case 't': ++dec_cur; *cur++ = '\011'; break;
600 case 'n': ++dec->cur; *cur++ = '\012'; break; 769 case 'n': ++dec_cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break; 770 case 'f': ++dec_cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break; 771 case 'r': ++dec_cur; *cur++ = '\015'; break;
603 772
604 case 'u': 773 case 'u':
605 { 774 {
606 UV lo, hi; 775 UV lo, hi;
607 ++dec->cur; 776 ++dec_cur;
608 777
778 dec->cur = dec_cur;
609 hi = decode_4hex (dec); 779 hi = decode_4hex (dec);
780 dec_cur = dec->cur;
610 if (hi == (UV)-1) 781 if (hi == (UV)-1)
611 goto fail; 782 goto fail;
612 783
613 // possibly a surrogate pair 784 // possibly a surrogate pair
614 if (hi >= 0xd800) 785 if (hi >= 0xd800)
615 if (hi < 0xdc00) 786 if (hi < 0xdc00)
616 { 787 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 788 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair"); 789 ERR ("missing low surrogate character in surrogate pair");
619 790
620 dec->cur += 2; 791 dec_cur += 2;
621 792
793 dec->cur = dec_cur;
622 lo = decode_4hex (dec); 794 lo = decode_4hex (dec);
795 dec_cur = dec->cur;
623 if (lo == (UV)-1) 796 if (lo == (UV)-1)
624 goto fail; 797 goto fail;
625 798
626 if (lo < 0xdc00 || lo >= 0xe000) 799 if (lo < 0xdc00 || lo >= 0xe000)
627 ERR ("surrogate pair expected"); 800 ERR ("surrogate pair expected");
… …
641 *cur++ = hi; 814 *cur++ = hi;
642 } 815 }
643 break; 816 break;
644 817
645 default: 818 default:
646 --dec->cur; 819 --dec_cur;
647 ERR ("illegal backslash escape sequence in string"); 820 ERR ("illegal backslash escape sequence in string");
648 } 821 }
649 } 822 }
650 else if (ch >= 0x20 && ch <= 0x7f) 823 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
651 *cur++ = ch; 824 *cur++ = ch;
652 else if (ch >= 0x80) 825 else if (ch >= 0x80)
653 { 826 {
654 --dec->cur;
655
656 STRLEN clen; 827 STRLEN clen;
828 UV uch;
829
830 --dec_cur;
831
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 832 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
658 if (clen == (STRLEN)-1) 833 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string"); 834 ERR ("malformed UTF-8 character in JSON string");
660 835
661 do 836 do
662 {
663 *cur++ = *dec->cur++; 837 *cur++ = *dec_cur++;
664 }
665 while (--clen); 838 while (--clen);
666 839
667 utf8 = 1; 840 utf8 = 1;
668 } 841 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else 842 else
843 {
844 --dec_cur;
845
846 if (!ch)
847 ERR ("unexpected end of string while parsing JSON string");
848 else
672 ERR ("invalid character encountered"); 849 ERR ("invalid character encountered while parsing JSON string");
673 850 }
674 } 851 }
675 while (cur < buf + SHORT_STRING_LEN); 852 while (cur < buf + SHORT_STRING_LEN);
676 853
854 {
677 STRLEN len = cur - buf; 855 STRLEN len = cur - buf;
678 856
679 if (sv) 857 if (sv)
680 { 858 {
681 SvGROW (sv, SvCUR (sv) + len + 1); 859 SvGROW (sv, SvCUR (sv) + len + 1);
682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 860 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
683 SvCUR_set (sv, SvCUR (sv) + len); 861 SvCUR_set (sv, SvCUR (sv) + len);
684 } 862 }
685 else 863 else
686 sv = newSVpvn (buf, len); 864 sv = newSVpvn (buf, len);
687 } 865 }
866 }
688 while (*dec->cur != '"'); 867 while (*dec_cur != '"');
689 868
690 ++dec->cur; 869 ++dec_cur;
691 870
692 if (sv) 871 if (sv)
693 { 872 {
694 SvPOK_only (sv); 873 SvPOK_only (sv);
695 *SvEND (sv) = 0; 874 *SvEND (sv) = 0;
… …
698 SvUTF8_on (sv); 877 SvUTF8_on (sv);
699 } 878 }
700 else 879 else
701 sv = newSVpvn ("", 0); 880 sv = newSVpvn ("", 0);
702 881
882 dec->cur = dec_cur;
703 return sv; 883 return sv;
704 884
705fail: 885fail:
886 dec->cur = dec_cur;
706 return 0; 887 return 0;
707} 888}
708 889
709static SV * 890static SV *
710decode_num (dec_t *dec) 891decode_num (dec_t *dec)
… …
768 is_nv = 1; 949 is_nv = 1;
769 } 950 }
770 951
771 if (!is_nv) 952 if (!is_nv)
772 { 953 {
773 UV uv; 954 int len = dec->cur - start;
774 int numtype = grok_number (start, dec->cur - start, &uv); 955
775 if (numtype & IS_NUMBER_IN_UV) 956 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
776 if (numtype & IS_NUMBER_NEG) 957 if (*start == '-')
958 switch (len)
777 { 959 {
778 if (uv < (UV)IV_MIN) 960 case 2: return newSViv (-( start [1] - '0' * 1));
779 return newSViv (-(IV)uv); 961 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
962 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
963 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
780 } 964 }
965 else
966 switch (len)
967 {
968 case 1: return newSViv ( start [0] - '0' * 1);
969 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
970 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
971 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
972 }
973
974 {
975 UV uv;
976 int numtype = grok_number (start, len, &uv);
977 if (numtype & IS_NUMBER_IN_UV)
978 if (numtype & IS_NUMBER_NEG)
979 {
980 if (uv < (UV)IV_MIN)
981 return newSViv (-(IV)uv);
982 }
781 else 983 else
782 return newSVuv (uv); 984 return newSVuv (uv);
783 } 985 }
784 986
987 len -= *start == '-' ? 1 : 0;
988
989 // does not fit into IV or UV, try NV
990 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
991 #if defined (LDBL_DIG)
992 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
993 #endif
994 )
995 // fits into NV without loss of precision
996 return newSVnv (Atof (start));
997
998 // everything else fails, convert it to a string
999 return newSVpvn (start, dec->cur - start);
1000 }
1001
1002 // loss of precision here
785 return newSVnv (Atof (start)); 1003 return newSVnv (Atof (start));
786 1004
787fail: 1005fail:
788 return 0; 1006 return 0;
789} 1007}
… …
833} 1051}
834 1052
835static SV * 1053static SV *
836decode_hv (dec_t *dec) 1054decode_hv (dec_t *dec)
837{ 1055{
1056 SV *sv;
838 HV *hv = newHV (); 1057 HV *hv = newHV ();
839 1058
840 DEC_INC_DEPTH; 1059 DEC_INC_DEPTH;
841 decode_ws (dec); 1060 decode_ws (dec);
842 1061
843 if (*dec->cur == '}') 1062 if (*dec->cur == '}')
844 ++dec->cur; 1063 ++dec->cur;
845 else 1064 else
846 for (;;) 1065 for (;;)
847 { 1066 {
848 SV *key, *value;
849
850 decode_ws (dec); EXPECT_CH ('"'); 1067 decode_ws (dec); EXPECT_CH ('"');
851 1068
852 key = decode_str (dec); 1069 // heuristic: assume that
853 if (!key) 1070 // a) decode_str + hv_store_ent are abysmally slow.
854 goto fail; 1071 // b) most hash keys are short, simple ascii text.
1072 // => try to "fast-match" such strings to avoid
1073 // the overhead of decode_str + hv_store_ent.
1074 {
1075 SV *value;
1076 char *p = dec->cur;
1077 char *e = p + 24; // only try up to 24 bytes
855 1078
856 decode_ws (dec); EXPECT_CH (':'); 1079 for (;;)
857
858 value = decode_sv (dec);
859 if (!value)
860 { 1080 {
1081 // the >= 0x80 is true on most architectures
1082 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1083 {
1084 // slow path, back up and use decode_str
1085 SV *key = decode_str (dec);
1086 if (!key)
1087 goto fail;
1088
1089 decode_ws (dec); EXPECT_CH (':');
1090
1091 value = decode_sv (dec);
1092 if (!value)
1093 {
1094 SvREFCNT_dec (key);
1095 goto fail;
1096 }
1097
1098 hv_store_ent (hv, key, value, 0);
861 SvREFCNT_dec (key); 1099 SvREFCNT_dec (key);
1100
1101 break;
1102 }
1103 else if (*p == '"')
1104 {
1105 // fast path, got a simple key
1106 char *key = dec->cur;
1107 int len = p - key;
1108 dec->cur = p + 1;
1109
1110 decode_ws (dec); EXPECT_CH (':');
1111
1112 value = decode_sv (dec);
1113 if (!value)
862 goto fail; 1114 goto fail;
1115
1116 hv_store (hv, key, len, value, 0);
1117
1118 break;
1119 }
1120
1121 ++p;
863 } 1122 }
864 1123 }
865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
867 1124
868 decode_ws (dec); 1125 decode_ws (dec);
869 1126
870 if (*dec->cur == '}') 1127 if (*dec->cur == '}')
871 { 1128 {
… …
878 1135
879 ++dec->cur; 1136 ++dec->cur;
880 } 1137 }
881 1138
882 DEC_DEC_DEPTH; 1139 DEC_DEC_DEPTH;
883 return newRV_noinc ((SV *)hv); 1140 sv = newRV_noinc ((SV *)hv);
1141
1142 // check filter callbacks
1143 if (dec->json.flags & F_HOOK)
1144 {
1145 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1146 {
1147 HE *cb, *he;
1148
1149 hv_iterinit (hv);
1150 he = hv_iternext (hv);
1151 hv_iterinit (hv);
1152
1153 // the next line creates a mortal sv each time its called.
1154 // might want to optimise this for common cases.
1155 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1156
1157 if (cb)
1158 {
1159 dSP;
1160 int count;
1161
1162 ENTER; SAVETMPS; PUSHMARK (SP);
1163 XPUSHs (HeVAL (he));
1164
1165 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1166
1167 if (count == 1)
1168 {
1169 sv = newSVsv (POPs);
1170 FREETMPS; LEAVE;
1171 return sv;
1172 }
1173
1174 FREETMPS; LEAVE;
1175 }
1176 }
1177
1178 if (dec->json.cb_object)
1179 {
1180 dSP;
1181 int count;
1182
1183 ENTER; SAVETMPS; PUSHMARK (SP);
1184 XPUSHs (sv_2mortal (sv));
1185
1186 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1187
1188 if (count == 1)
1189 {
1190 sv = newSVsv (POPs);
1191 FREETMPS; LEAVE;
1192 return sv;
1193 }
1194
1195 SvREFCNT_inc (sv);
1196 FREETMPS; LEAVE;
1197 }
1198 }
1199
1200 return sv;
884 1201
885fail: 1202fail:
886 SvREFCNT_dec (hv); 1203 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH; 1204 DEC_DEC_DEPTH;
888 return 0; 1205 return 0;
… …
890 1207
891static SV * 1208static SV *
892decode_sv (dec_t *dec) 1209decode_sv (dec_t *dec)
893{ 1210{
894 decode_ws (dec); 1211 decode_ws (dec);
1212
1213 // the beauty of JSON: you need exactly one character lookahead
1214 // to parse anything.
895 switch (*dec->cur) 1215 switch (*dec->cur)
896 { 1216 {
897 case '"': ++dec->cur; return decode_str (dec); 1217 case '"': ++dec->cur; return decode_str (dec);
898 case '[': ++dec->cur; return decode_av (dec); 1218 case '[': ++dec->cur; return decode_av (dec);
899 case '{': ++dec->cur; return decode_hv (dec); 1219 case '{': ++dec->cur; return decode_hv (dec);
… …
905 1225
906 case 't': 1226 case 't':
907 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1227 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
908 { 1228 {
909 dec->cur += 4; 1229 dec->cur += 4;
910 return newSViv (1); 1230 return SvREFCNT_inc (json_true);
911 } 1231 }
912 else 1232 else
913 ERR ("'true' expected"); 1233 ERR ("'true' expected");
914 1234
915 break; 1235 break;
916 1236
917 case 'f': 1237 case 'f':
918 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1238 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
919 { 1239 {
920 dec->cur += 5; 1240 dec->cur += 5;
921 return newSViv (0); 1241 return SvREFCNT_inc (json_false);
922 } 1242 }
923 else 1243 else
924 ERR ("'false' expected"); 1244 ERR ("'false' expected");
925 1245
926 break; 1246 break;
… …
935 ERR ("'null' expected"); 1255 ERR ("'null' expected");
936 1256
937 break; 1257 break;
938 1258
939 default: 1259 default:
940 ERR ("malformed json string, neither array, object, number, string or atom"); 1260 ERR ("malformed JSON string, neither array, object, number, string or atom");
941 break; 1261 break;
942 } 1262 }
943 1263
944fail: 1264fail:
945 return 0; 1265 return 0;
946} 1266}
947 1267
948static SV * 1268static SV *
949decode_json (SV *string, U32 flags) 1269decode_json (SV *string, JSON *json, UV *offset_return)
950{ 1270{
1271 dec_t dec;
1272 UV offset;
951 SV *sv; 1273 SV *sv;
952 1274
1275 SvGETMAGIC (string);
1276 SvUPGRADE (string, SVt_PV);
1277
1278 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1279 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1280 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1281
953 if (flags & F_UTF8) 1282 if (json->flags & F_UTF8)
954 sv_utf8_downgrade (string, 0); 1283 sv_utf8_downgrade (string, 0);
955 else 1284 else
956 sv_utf8_upgrade (string); 1285 sv_utf8_upgrade (string);
957 1286
958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1287 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
959 1288
960 dec_t dec; 1289 dec.json = *json;
961 dec.flags = flags;
962 dec.cur = SvPVX (string); 1290 dec.cur = SvPVX (string);
963 dec.end = SvEND (string); 1291 dec.end = SvEND (string);
964 dec.err = 0; 1292 dec.err = 0;
965 dec.depth = 0; 1293 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags); 1294 dec.maxdepth = DEC_DEPTH (dec.json.flags);
967 1295
1296 if (dec.json.cb_object || dec.json.cb_sk_object)
1297 dec.json.flags |= F_HOOK;
1298
968 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1299 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
969 sv = decode_sv (&dec); 1300 sv = decode_sv (&dec);
970 1301
1302 if (!(offset_return || !sv))
1303 {
1304 // check for trailing garbage
1305 decode_ws (&dec);
1306
1307 if (*dec.cur)
1308 {
1309 dec.err = "garbage after JSON object";
1310 SvREFCNT_dec (sv);
1311 sv = 0;
1312 }
1313 }
1314
1315 if (offset_return || !sv)
1316 {
1317 offset = dec.json.flags & F_UTF8
1318 ? dec.cur - SvPVX (string)
1319 : utf8_distance (dec.cur, SvPVX (string));
1320
1321 if (offset_return)
1322 *offset_return = offset;
1323 }
1324
971 if (!sv) 1325 if (!sv)
972 { 1326 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
975 : utf8_distance (dec.cur, SvPVX (string));
976 SV *uni = sv_newmortal (); 1327 SV *uni = sv_newmortal ();
977 1328
978 // horrible hack to silence warning inside pv_uni_display 1329 // horrible hack to silence warning inside pv_uni_display
979 COP cop = *PL_curcop; 1330 COP cop = *PL_curcop;
980 cop.cop_warnings = pWARN_NONE; 1331 cop.cop_warnings = pWARN_NONE;
… …
982 SAVEVPTR (PL_curcop); 1333 SAVEVPTR (PL_curcop);
983 PL_curcop = &cop; 1334 PL_curcop = &cop;
984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1335 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE; 1336 LEAVE;
986 1337
987 croak ("%s, at character offset %d (%s)", 1338 croak ("%s, at character offset %d [\"%s\"]",
988 dec.err, 1339 dec.err,
989 (int)offset, 1340 (int)offset,
990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1341 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
991 } 1342 }
992 1343
993 sv = sv_2mortal (sv); 1344 sv = sv_2mortal (sv);
994 1345
995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1346 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)"); 1347 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
997 1348
998 return sv; 1349 return sv;
999} 1350}
1000 1351
… …
1004MODULE = JSON::XS PACKAGE = JSON::XS 1355MODULE = JSON::XS PACKAGE = JSON::XS
1005 1356
1006BOOT: 1357BOOT:
1007{ 1358{
1008 int i; 1359 int i;
1009
1010 memset (decode_hexdigit, 0xff, 256);
1011 1360
1012 for (i = 0; i < 256; ++i) 1361 for (i = 0; i < 256; ++i)
1013 decode_hexdigit [i] = 1362 decode_hexdigit [i] =
1014 i >= '0' && i <= '9' ? i - '0' 1363 i >= '0' && i <= '9' ? i - '0'
1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1364 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1365 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1017 : -1; 1366 : -1;
1018 1367
1019 json_stash = gv_stashpv ("JSON::XS", 1); 1368 json_stash = gv_stashpv ("JSON::XS" , 1);
1369 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1370
1371 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1372 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1020} 1373}
1021 1374
1022PROTOTYPES: DISABLE 1375PROTOTYPES: DISABLE
1023 1376
1024SV *new (char *dummy) 1377void CLONE (...)
1025 CODE: 1378 CODE:
1026 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1379 json_stash = 0;
1027 OUTPUT: 1380 json_boolean_stash = 0;
1028 RETVAL
1029 1381
1382void new (char *klass)
1383 PPCODE:
1384{
1385 SV *pv = NEWSV (0, sizeof (JSON));
1386 SvPOK_only (pv);
1387 Zero (SvPVX (pv), 1, JSON);
1388 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1389 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1390}
1391
1030SV *ascii (SV *self, int enable = 1) 1392void ascii (JSON *self, int enable = 1)
1031 ALIAS: 1393 ALIAS:
1032 ascii = F_ASCII 1394 ascii = F_ASCII
1395 latin1 = F_LATIN1
1033 utf8 = F_UTF8 1396 utf8 = F_UTF8
1034 indent = F_INDENT 1397 indent = F_INDENT
1035 canonical = F_CANONICAL 1398 canonical = F_CANONICAL
1036 space_before = F_SPACE_BEFORE 1399 space_before = F_SPACE_BEFORE
1037 space_after = F_SPACE_AFTER 1400 space_after = F_SPACE_AFTER
1038 pretty = F_PRETTY 1401 pretty = F_PRETTY
1039 allow_nonref = F_ALLOW_NONREF 1402 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK 1403 shrink = F_SHRINK
1404 allow_blessed = F_ALLOW_BLESSED
1405 convert_blessed = F_CONV_BLESSED
1041 CODE: 1406 PPCODE:
1042{ 1407{
1043 UV *uv = SvJSON (self);
1044 if (enable) 1408 if (enable)
1045 *uv |= ix; 1409 self->flags |= ix;
1046 else 1410 else
1047 *uv &= ~ix; 1411 self->flags &= ~ix;
1048 1412
1049 RETVAL = newSVsv (self); 1413 XPUSHs (ST (0));
1050} 1414}
1051 OUTPUT:
1052 RETVAL
1053 1415
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1416void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1055 CODE: 1417 PPCODE:
1056{ 1418{
1057 UV *uv = SvJSON (self);
1058 UV log2 = 0; 1419 UV log2 = 0;
1059 1420
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1421 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1061 1422
1062 while ((1UL << log2) < max_depth) 1423 while ((1UL << log2) < max_depth)
1063 ++log2; 1424 ++log2;
1064 1425
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1426 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066 1427
1067 RETVAL = newSVsv (self); 1428 XPUSHs (ST (0));
1068} 1429}
1069 OUTPUT:
1070 RETVAL
1071 1430
1072void encode (SV *self, SV *scalar) 1431void max_size (JSON *self, UV max_size = 0)
1073 PPCODE: 1432 PPCODE:
1074 XPUSHs (encode_json (scalar, *SvJSON (self))); 1433{
1434 UV log2 = 0;
1075 1435
1076void decode (SV *self, SV *jsonstr) 1436 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1437 if (max_size == 1) max_size = 2;
1438
1439 while ((1UL << log2) < max_size)
1440 ++log2;
1441
1442 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1443
1444 XPUSHs (ST (0));
1445}
1446
1447void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1448 PPCODE:
1449{
1450 SvREFCNT_dec (self->cb_object);
1451 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1452
1453 XPUSHs (ST (0));
1454}
1455
1456void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1457 PPCODE:
1458{
1459 if (!self->cb_sk_object)
1460 self->cb_sk_object = newHV ();
1461
1462 if (SvOK (cb))
1463 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1464 else
1465 {
1466 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1467
1468 if (!HvKEYS (self->cb_sk_object))
1469 {
1470 SvREFCNT_dec (self->cb_sk_object);
1471 self->cb_sk_object = 0;
1472 }
1473 }
1474
1475 XPUSHs (ST (0));
1476}
1477
1478void encode (JSON *self, SV *scalar)
1479 PPCODE:
1480 XPUSHs (encode_json (scalar, self));
1481
1482void decode (JSON *self, SV *jsonstr)
1483 PPCODE:
1078 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1484 XPUSHs (decode_json (jsonstr, self, 0));
1485
1486void decode_prefix (JSON *self, SV *jsonstr)
1487 PPCODE:
1488{
1489 UV offset;
1490 EXTEND (SP, 2);
1491 PUSHs (decode_json (jsonstr, self, &offset));
1492 PUSHs (sv_2mortal (newSVuv (offset)));
1493}
1494
1495void DESTROY (JSON *self)
1496 CODE:
1497 SvREFCNT_dec (self->cb_sk_object);
1498 SvREFCNT_dec (self->cb_object);
1079 1499
1080PROTOTYPES: ENABLE 1500PROTOTYPES: ENABLE
1081 1501
1082void to_json (SV *scalar) 1502void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
1085 PPCODE: 1503 PPCODE:
1504{
1505 JSON json = { F_DEFAULT | F_UTF8 };
1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1506 XPUSHs (encode_json (scalar, &json));
1507}
1087 1508
1088void from_json (SV *jsonstr) 1509void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
1091 PPCODE: 1510 PPCODE:
1511{
1512 JSON json = { F_DEFAULT | F_UTF8 };
1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1513 XPUSHs (decode_json (jsonstr, &json, 0));
1514}
1093 1515

Diff Legend

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