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 1.16 by root, Sun Mar 25 02:37:00 2007 UTC vs.
Revision 1.50 by root, Mon Jul 2 00:29:38 2007 UTC

… …
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"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 27#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
17 36
18// F_SKIPINVALID? 37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
19// F_EXECCODEREF? 38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
20// F_SELFCONVERT?
21 39
22#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
23#define F_DEFAULT 0 41#define F_DEFAULT (9UL << S_MAXDEPTH)
24 42
25#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
26#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
27 45
28#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
29#define SHORT_STRING_LEN 512 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
30 47
31#define SB do { 48#define SB do {
32#define SE } while (0) 49#define SE } while (0)
33 50
51#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value))
53# define inline inline
54#else
55# define expect(expr,value) (expr)
56# define inline static
57#endif
58
59#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1)
61
34static HV *json_stash; // JSON::XS:: 62static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false;
64
65typedef struct {
66 U32 flags;
67 SV *cb_object, *cb_sk_object;
68} JSON;
35 69
36///////////////////////////////////////////////////////////////////////////// 70/////////////////////////////////////////////////////////////////////////////
37// utility functions 71// utility functions
38 72
39static UV * 73inline void
40SvJSON (SV *sv)
41{
42 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
43 croak ("object is not of type JSON::XS");
44
45 return &SvUVX (SvRV (sv));
46}
47
48static void
49shrink (SV *sv) 74shrink (SV *sv)
50{ 75{
51 sv_utf8_downgrade (sv, 1); 76 sv_utf8_downgrade (sv, 1);
52 if (SvLEN (sv) > SvCUR (sv) + 1) 77 if (SvLEN (sv) > SvCUR (sv) + 1)
53 { 78 {
… …
62// decode an utf-8 character and return it, or (UV)-1 in 87// decode an utf-8 character and return it, or (UV)-1 in
63// case of an error. 88// case of an error.
64// we special-case "safe" characters from U+80 .. U+7FF, 89// we special-case "safe" characters from U+80 .. U+7FF,
65// but use the very good perl function to parse anything else. 90// but use the very good perl function to parse anything else.
66// note that we never call this function for a ascii codepoints 91// note that we never call this function for a ascii codepoints
67static UV 92inline UV
68decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
69{ 94{
70 if (s[0] > 0xdf || s[0] < 0xc2) 95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
71 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
72 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
73 { 98 {
74 *clen = 2; 99 *clen = 2;
75 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
76 } 101 }
77 else 102 else
103 {
104 *clen = (STRLEN)-1;
78 return (UV)-1; 105 return (UV)-1;
106 }
79} 107}
80 108
81///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
82// encoder 110// encoder
83 111
… …
85typedef struct 113typedef struct
86{ 114{
87 char *cur; // SvPVX (sv) + current output position 115 char *cur; // SvPVX (sv) + current output position
88 char *end; // SvEND (sv) 116 char *end; // SvEND (sv)
89 SV *sv; // result scalar 117 SV *sv; // result scalar
90 UV flags; // F_* 118 JSON json;
91 int indent; // indentation level 119 U32 indent; // indentation level
92 int max_depth; // max. recursion level 120 U32 maxdepth; // max. indentation/recursion level
93} enc_t; 121} enc_t;
94 122
95static void 123inline void
96need (enc_t *enc, STRLEN len) 124need (enc_t *enc, STRLEN len)
97{ 125{
98 if (enc->cur + len >= enc->end) 126 if (expect_false (enc->cur + len >= enc->end))
99 { 127 {
100 STRLEN cur = enc->cur - SvPVX (enc->sv); 128 STRLEN cur = enc->cur - SvPVX (enc->sv);
101 SvGROW (enc->sv, cur + len + 1); 129 SvGROW (enc->sv, cur + len + 1);
102 enc->cur = SvPVX (enc->sv) + cur; 130 enc->cur = SvPVX (enc->sv) + cur;
103 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
104 } 132 }
105} 133}
106 134
107static void 135inline void
108encode_ch (enc_t *enc, char ch) 136encode_ch (enc_t *enc, char ch)
109{ 137{
110 need (enc, 1); 138 need (enc, 1);
111 *enc->cur++ = ch; 139 *enc->cur++ = ch;
112} 140}
… …
120 148
121 while (str < end) 149 while (str < end)
122 { 150 {
123 unsigned char ch = *(unsigned char *)str; 151 unsigned char ch = *(unsigned char *)str;
124 152
125 if (ch >= 0x20 && ch < 0x80) // most common case 153 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
126 { 154 {
127 if (ch == '"') // but with slow exceptions 155 if (expect_false (ch == '"')) // but with slow exceptions
128 { 156 {
129 need (enc, len += 1); 157 need (enc, len += 1);
130 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
131 *enc->cur++ = '"'; 159 *enc->cur++ = '"';
132 } 160 }
133 else if (ch == '\\') 161 else if (expect_false (ch == '\\'))
134 { 162 {
135 need (enc, len += 1); 163 need (enc, len += 1);
136 *enc->cur++ = '\\'; 164 *enc->cur++ = '\\';
137 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
138 } 166 }
… …
156 STRLEN clen; 184 STRLEN clen;
157 UV uch; 185 UV uch;
158 186
159 if (is_utf8) 187 if (is_utf8)
160 { 188 {
161 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
162 uch = decode_utf8 (str, end - str, &clen); 189 uch = decode_utf8 (str, end - str, &clen);
163 if (clen == (STRLEN)-1) 190 if (clen == (STRLEN)-1)
164 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 191 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
165 } 192 }
166 else 193 else
… …
170 } 197 }
171 198
172 if (uch > 0x10FFFFUL) 199 if (uch > 0x10FFFFUL)
173 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
174 201
175 if (uch < 0x80 || enc->flags & F_ASCII) 202 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
176 { 203 {
177 if (uch > 0xFFFFUL) 204 if (uch > 0xFFFFUL)
178 { 205 {
179 need (enc, len += 11); 206 need (enc, len += 11);
180 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
194 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
195 } 222 }
196 223
197 str += clen; 224 str += clen;
198 } 225 }
226 else if (enc->json.flags & F_LATIN1)
227 {
228 *enc->cur++ = uch;
229 str += clen;
230 }
199 else if (is_utf8) 231 else if (is_utf8)
200 { 232 {
201 need (enc, len += clen); 233 need (enc, len += clen);
202 do 234 do
203 { 235 {
… …
205 } 237 }
206 while (--clen); 238 while (--clen);
207 } 239 }
208 else 240 else
209 { 241 {
210 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
211 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
212 ++str; 244 ++str;
213 } 245 }
214 } 246 }
215 } 247 }
… …
217 249
218 --len; 250 --len;
219 } 251 }
220} 252}
221 253
222static void 254inline void
223encode_indent (enc_t *enc) 255encode_indent (enc_t *enc)
224{ 256{
225 if (enc->flags & F_INDENT) 257 if (enc->json.flags & F_INDENT)
226 { 258 {
227 int spaces = enc->indent * INDENT_STEP; 259 int spaces = enc->indent * INDENT_STEP;
228 260
229 need (enc, spaces); 261 need (enc, spaces);
230 memset (enc->cur, ' ', spaces); 262 memset (enc->cur, ' ', spaces);
231 enc->cur += spaces; 263 enc->cur += spaces;
232 } 264 }
233} 265}
234 266
235static void 267inline void
236encode_space (enc_t *enc) 268encode_space (enc_t *enc)
237{ 269{
238 need (enc, 1); 270 need (enc, 1);
239 encode_ch (enc, ' '); 271 encode_ch (enc, ' ');
240} 272}
241 273
242static void 274inline void
243encode_nl (enc_t *enc) 275encode_nl (enc_t *enc)
244{ 276{
245 if (enc->flags & F_INDENT) 277 if (enc->json.flags & F_INDENT)
246 { 278 {
247 need (enc, 1); 279 need (enc, 1);
248 encode_ch (enc, '\n'); 280 encode_ch (enc, '\n');
249 } 281 }
250} 282}
251 283
252static void 284inline void
253encode_comma (enc_t *enc) 285encode_comma (enc_t *enc)
254{ 286{
255 encode_ch (enc, ','); 287 encode_ch (enc, ',');
256 288
257 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
258 encode_nl (enc); 290 encode_nl (enc);
259 else if (enc->flags & F_SPACE_AFTER) 291 else if (enc->json.flags & F_SPACE_AFTER)
260 encode_space (enc); 292 encode_space (enc);
261} 293}
262 294
263static void encode_sv (enc_t *enc, SV *sv); 295static void encode_sv (enc_t *enc, SV *sv);
264 296
265static void 297static void
266encode_av (enc_t *enc, AV *av) 298encode_av (enc_t *enc, AV *av)
267{ 299{
268 int i, len = av_len (av); 300 int i, len = av_len (av);
301
302 if (enc->indent >= enc->maxdepth)
303 croak ("data structure too deep (hit recursion limit)");
269 304
270 encode_ch (enc, '['); encode_nl (enc); 305 encode_ch (enc, '['); encode_nl (enc);
271 ++enc->indent; 306 ++enc->indent;
272 307
273 for (i = 0; i <= len; ++i) 308 for (i = 0; i <= len; ++i)
… …
304 else 339 else
305 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
306 341
307 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
308 343
309 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 344 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
310 encode_ch (enc, ':'); 345 encode_ch (enc, ':');
311 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 346 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
312 encode_sv (enc, HeVAL (he)); 347 encode_sv (enc, HeVAL (he));
313} 348}
314 349
315// compare hash entries, used when all keys are bytestrings 350// compare hash entries, used when all keys are bytestrings
316static int 351static int
… …
339 374
340static void 375static void
341encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
342{ 377{
343 int count, i; 378 int count, i;
379
380 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)");
344 382
345 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
346 384
347 if ((count = hv_iterinit (hv))) 385 if ((count = hv_iterinit (hv)))
348 { 386 {
349 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
350 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
351 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
352 // that randomises hash orderings 390 // that randomises hash orderings
353 if (enc->flags & F_CANONICAL) 391 if (enc->json.flags & F_CANONICAL)
354 { 392 {
355 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
356 int fast = 1; 393 int fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE));
397#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif
357 400
358 i = 0; 401 i = 0;
359 while ((he = hv_iternext (hv))) 402 while ((he = hv_iternext (hv)))
360 { 403 {
361 hes [i++] = he; 404 hes [i++] = he;
… …
396 439
397 encode_nl (enc); 440 encode_nl (enc);
398 } 441 }
399 else 442 else
400 { 443 {
401 SV *sv;
402 HE *he = hv_iternext (hv); 444 HE *he = hv_iternext (hv);
403 445
404 for (;;) 446 for (;;)
405 { 447 {
406 encode_indent (enc); 448 encode_indent (enc);
… …
415 encode_nl (enc); 457 encode_nl (enc);
416 } 458 }
417 } 459 }
418 460
419 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 461 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462}
463
464// encode objects, arrays and special \0=false and \1=true values.
465static void
466encode_rv (enc_t *enc, SV *sv)
467{
468 svtype svt;
469
470 SvGETMAGIC (sv);
471 svt = SvTYPE (sv);
472
473 if (expect_false (SvOBJECT (sv)))
474 {
475 if (SvSTASH (sv) == json_boolean_stash)
476 {
477 if (SvIV (sv) == 0)
478 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 }
482 else
483 {
484#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 {
487 // not yet
488 }
489#endif
490 if (enc->json.flags & F_CONV_BLESSED)
491 {
492 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
494
495 if (to_json)
496 {
497 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
498 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
499
500 // calling with G_SCALAR ensures that we always get a 1 reutrn value
501 // check anyways.
502 PUTBACK;
503 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
504 SPAGAIN;
505
506 encode_sv (enc, POPs);
507
508 FREETMPS; LEAVE;
509 }
510 else if (enc->json.flags & F_ALLOW_BLESSED)
511 encode_str (enc, "null", 4, 0);
512 else
513 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
514 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
515 }
516 else if (enc->json.flags & F_ALLOW_BLESSED)
517 encode_str (enc, "null", 4, 0);
518 else
519 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
520 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
521 }
522 }
523 else if (svt == SVt_PVHV)
524 encode_hv (enc, (HV *)sv);
525 else if (svt == SVt_PVAV)
526 encode_av (enc, (AV *)sv);
527 else if (svt < SVt_PVAV)
528 {
529 STRLEN len = 0;
530 char *pv = svt ? SvPV (sv, len) : 0;
531
532 if (len == 1 && *pv == '0')
533 encode_str (enc, "false", 5, 0);
534 else if (len == 1 && *pv == '1')
535 encode_str (enc, "true", 4, 0);
536 else
537 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
538 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
539 }
540 else
541 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
420} 543}
421 544
422static void 545static void
423encode_sv (enc_t *enc, SV *sv) 546encode_sv (enc_t *enc, SV *sv)
424{ 547{
… …
432 encode_str (enc, str, len, SvUTF8 (sv)); 555 encode_str (enc, str, len, SvUTF8 (sv));
433 encode_ch (enc, '"'); 556 encode_ch (enc, '"');
434 } 557 }
435 else if (SvNOKp (sv)) 558 else if (SvNOKp (sv))
436 { 559 {
560 // trust that perl will do the right thing w.r.t. JSON syntax.
437 need (enc, NV_DIG + 32); 561 need (enc, NV_DIG + 32);
438 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 562 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
439 enc->cur += strlen (enc->cur); 563 enc->cur += strlen (enc->cur);
440 } 564 }
441 else if (SvIOKp (sv)) 565 else if (SvIOKp (sv))
442 { 566 {
443 need (enc, 64); 567 // we assume we can always read an IV as a UV
568 if (SvUV (sv) & ~(UV)0x7fff)
569 {
570 // large integer, use the (rather slow) snprintf way.
571 need (enc, sizeof (UV) * 3);
444 enc->cur += 572 enc->cur +=
445 SvIsUV(sv) 573 SvIsUV(sv)
446 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 574 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
447 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 575 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
576 }
577 else
578 {
579 // optimise the "small number case"
580 // code will likely be branchless and use only a single multiplication
581 I32 i = SvIV (sv);
582 U32 u;
583 char digit, nz = 0;
584
585 need (enc, 6);
586
587 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
588 u = i < 0 ? -i : i;
589
590 // convert to 4.28 fixed-point representation
591 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
592
593 // now output digit by digit, each time masking out the integer part
594 // and multiplying by 5 while moving the decimal point one to the right,
595 // resulting in a net multiplication by 10.
596 // we always write the digit to memory but conditionally increment
597 // the pointer, to ease the usage of conditional move instructions.
598 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
599 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
600 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
601 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
602 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
603 }
448 } 604 }
449 else if (SvROK (sv)) 605 else if (SvROK (sv))
450 { 606 encode_rv (enc, SvRV (sv));
451 SV *rv = SvRV (sv);
452
453 if (enc->indent >= enc->max_depth)
454 croak ("data structure too deep (hit recursion limit)");
455
456 switch (SvTYPE (rv))
457 {
458 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
459 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
460
461 default:
462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv));
464 }
465 }
466 else if (!SvOK (sv)) 607 else if (!SvOK (sv))
467 encode_str (enc, "null", 4, 0); 608 encode_str (enc, "null", 4, 0);
468 else 609 else
469 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 610 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
470 SvPV_nolen (sv), SvFLAGS (sv)); 611 SvPV_nolen (sv), SvFLAGS (sv));
471} 612}
472 613
473static SV * 614static SV *
474encode_json (SV *scalar, UV flags) 615encode_json (SV *scalar, JSON *json)
475{ 616{
617 enc_t enc;
618
476 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 619 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
477 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 620 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
478 621
479 enc_t enc; 622 enc.json = *json;
480 enc.flags = flags;
481 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 623 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
482 enc.cur = SvPVX (enc.sv); 624 enc.cur = SvPVX (enc.sv);
483 enc.end = SvEND (enc.sv); 625 enc.end = SvEND (enc.sv);
484 enc.indent = 0; 626 enc.indent = 0;
485 enc.max_depth = 0x7fffffffUL; 627 enc.maxdepth = DEC_DEPTH (enc.json.flags);
486 628
487 SvPOK_only (enc.sv); 629 SvPOK_only (enc.sv);
488 encode_sv (&enc, scalar); 630 encode_sv (&enc, scalar);
489 631
632 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
633 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
634
490 if (!(flags & (F_ASCII | F_UTF8))) 635 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
491 SvUTF8_on (enc.sv); 636 SvUTF8_on (enc.sv);
492 637
493 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
494
495 if (enc.flags & F_SHRINK) 638 if (enc.json.flags & F_SHRINK)
496 shrink (enc.sv); 639 shrink (enc.sv);
497 640
498 return enc.sv; 641 return enc.sv;
499} 642}
500 643
… …
505typedef struct 648typedef struct
506{ 649{
507 char *cur; // current parser pointer 650 char *cur; // current parser pointer
508 char *end; // end of input string 651 char *end; // end of input string
509 const char *err; // parse error, if != 0 652 const char *err; // parse error, if != 0
510 UV flags; // F_* 653 JSON json;
654 U32 depth; // recursion depth
655 U32 maxdepth; // recursion depth limit
511} dec_t; 656} dec_t;
512 657
513static void 658inline void
514decode_ws (dec_t *dec) 659decode_ws (dec_t *dec)
515{ 660{
516 for (;;) 661 for (;;)
517 { 662 {
518 char ch = *dec->cur; 663 char ch = *dec->cur;
… …
524 ++dec->cur; 669 ++dec->cur;
525 } 670 }
526} 671}
527 672
528#define ERR(reason) SB dec->err = reason; goto fail; SE 673#define ERR(reason) SB dec->err = reason; goto fail; SE
674
529#define EXPECT_CH(ch) SB \ 675#define EXPECT_CH(ch) SB \
530 if (*dec->cur != ch) \ 676 if (*dec->cur != ch) \
531 ERR (# ch " expected"); \ 677 ERR (# ch " expected"); \
532 ++dec->cur; \ 678 ++dec->cur; \
533 SE 679 SE
534 680
681#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
682#define DEC_DEC_DEPTH --dec->depth
683
535static SV *decode_sv (dec_t *dec); 684static SV *decode_sv (dec_t *dec);
536 685
537static signed char decode_hexdigit[256]; 686static signed char decode_hexdigit[256];
538 687
539static UV 688static UV
540decode_4hex (dec_t *dec) 689decode_4hex (dec_t *dec)
541{ 690{
542 signed char d1, d2, d3, d4; 691 signed char d1, d2, d3, d4;
543 unsigned char *cur = (unsigned char *)dec->cur; 692 unsigned char *cur = (unsigned char *)dec->cur;
544 693
545 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 694 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
546 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 695 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
547 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 696 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
548 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 697 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
549 698
550 dec->cur += 4; 699 dec->cur += 4;
551 700
552 return ((UV)d1) << 12 701 return ((UV)d1) << 12
553 | ((UV)d2) << 8 702 | ((UV)d2) << 8
… …
561static SV * 710static SV *
562decode_str (dec_t *dec) 711decode_str (dec_t *dec)
563{ 712{
564 SV *sv = 0; 713 SV *sv = 0;
565 int utf8 = 0; 714 int utf8 = 0;
715 char *dec_cur = dec->cur;
566 716
567 do 717 do
568 { 718 {
569 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 719 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
570 char *cur = buf; 720 char *cur = buf;
571 721
572 do 722 do
573 { 723 {
574 unsigned char ch = *(unsigned char *)dec->cur++; 724 unsigned char ch = *(unsigned char *)dec_cur++;
575 725
576 if (ch == '"') 726 if (expect_false (ch == '"'))
577 { 727 {
578 --dec->cur; 728 --dec_cur;
579 break; 729 break;
580 } 730 }
581 else if (ch == '\\') 731 else if (expect_false (ch == '\\'))
582 { 732 {
583 switch (*dec->cur) 733 switch (*dec_cur)
584 { 734 {
585 case '\\': 735 case '\\':
586 case '/': 736 case '/':
587 case '"': *cur++ = *dec->cur++; break; 737 case '"': *cur++ = *dec_cur++; break;
588 738
589 case 'b': ++dec->cur; *cur++ = '\010'; break; 739 case 'b': ++dec_cur; *cur++ = '\010'; break;
590 case 't': ++dec->cur; *cur++ = '\011'; break; 740 case 't': ++dec_cur; *cur++ = '\011'; break;
591 case 'n': ++dec->cur; *cur++ = '\012'; break; 741 case 'n': ++dec_cur; *cur++ = '\012'; break;
592 case 'f': ++dec->cur; *cur++ = '\014'; break; 742 case 'f': ++dec_cur; *cur++ = '\014'; break;
593 case 'r': ++dec->cur; *cur++ = '\015'; break; 743 case 'r': ++dec_cur; *cur++ = '\015'; break;
594 744
595 case 'u': 745 case 'u':
596 { 746 {
597 UV lo, hi; 747 UV lo, hi;
598 ++dec->cur; 748 ++dec_cur;
599 749
750 dec->cur = dec_cur;
600 hi = decode_4hex (dec); 751 hi = decode_4hex (dec);
752 dec_cur = dec->cur;
601 if (hi == (UV)-1) 753 if (hi == (UV)-1)
602 goto fail; 754 goto fail;
603 755
604 // possibly a surrogate pair 756 // possibly a surrogate pair
605 if (hi >= 0xd800) 757 if (hi >= 0xd800)
606 if (hi < 0xdc00) 758 if (hi < 0xdc00)
607 { 759 {
608 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 760 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
609 ERR ("missing low surrogate character in surrogate pair"); 761 ERR ("missing low surrogate character in surrogate pair");
610 762
611 dec->cur += 2; 763 dec_cur += 2;
612 764
765 dec->cur = dec_cur;
613 lo = decode_4hex (dec); 766 lo = decode_4hex (dec);
767 dec_cur = dec->cur;
614 if (lo == (UV)-1) 768 if (lo == (UV)-1)
615 goto fail; 769 goto fail;
616 770
617 if (lo < 0xdc00 || lo >= 0xe000) 771 if (lo < 0xdc00 || lo >= 0xe000)
618 ERR ("surrogate pair expected"); 772 ERR ("surrogate pair expected");
… …
632 *cur++ = hi; 786 *cur++ = hi;
633 } 787 }
634 break; 788 break;
635 789
636 default: 790 default:
637 --dec->cur; 791 --dec_cur;
638 ERR ("illegal backslash escape sequence in string"); 792 ERR ("illegal backslash escape sequence in string");
639 } 793 }
640 } 794 }
641 else if (ch >= 0x20 && ch <= 0x7f) 795 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
642 *cur++ = ch; 796 *cur++ = ch;
643 else if (ch >= 0x80) 797 else if (ch >= 0x80)
644 { 798 {
645 --dec->cur;
646
647 STRLEN clen; 799 STRLEN clen;
800 UV uch;
801
802 --dec_cur;
803
648 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 804 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
649 if (clen == (STRLEN)-1) 805 if (clen == (STRLEN)-1)
650 ERR ("malformed UTF-8 character in JSON string"); 806 ERR ("malformed UTF-8 character in JSON string");
651 807
652 do 808 do
653 {
654 *cur++ = *dec->cur++; 809 *cur++ = *dec_cur++;
655 }
656 while (--clen); 810 while (--clen);
657 811
658 utf8 = 1; 812 utf8 = 1;
659 } 813 }
660 else if (!ch)
661 ERR ("unexpected end of string while parsing json string");
662 else 814 else
815 {
816 --dec_cur;
817
818 if (!ch)
819 ERR ("unexpected end of string while parsing JSON string");
820 else
663 ERR ("invalid character encountered"); 821 ERR ("invalid character encountered while parsing JSON string");
664 822 }
665 } 823 }
666 while (cur < buf + SHORT_STRING_LEN); 824 while (cur < buf + SHORT_STRING_LEN);
667 825
826 {
668 STRLEN len = cur - buf; 827 STRLEN len = cur - buf;
669 828
670 if (sv) 829 if (sv)
671 { 830 {
672 SvGROW (sv, SvCUR (sv) + len + 1); 831 SvGROW (sv, SvCUR (sv) + len + 1);
673 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 832 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
674 SvCUR_set (sv, SvCUR (sv) + len); 833 SvCUR_set (sv, SvCUR (sv) + len);
675 } 834 }
676 else 835 else
677 sv = newSVpvn (buf, len); 836 sv = newSVpvn (buf, len);
678 } 837 }
838 }
679 while (*dec->cur != '"'); 839 while (*dec_cur != '"');
680 840
681 ++dec->cur; 841 ++dec_cur;
682 842
683 if (sv) 843 if (sv)
684 { 844 {
685 SvPOK_only (sv); 845 SvPOK_only (sv);
686 *SvEND (sv) = 0; 846 *SvEND (sv) = 0;
… …
689 SvUTF8_on (sv); 849 SvUTF8_on (sv);
690 } 850 }
691 else 851 else
692 sv = newSVpvn ("", 0); 852 sv = newSVpvn ("", 0);
693 853
854 dec->cur = dec_cur;
694 return sv; 855 return sv;
695 856
696fail: 857fail:
858 dec->cur = dec_cur;
697 return 0; 859 return 0;
698} 860}
699 861
700static SV * 862static SV *
701decode_num (dec_t *dec) 863decode_num (dec_t *dec)
… …
759 is_nv = 1; 921 is_nv = 1;
760 } 922 }
761 923
762 if (!is_nv) 924 if (!is_nv)
763 { 925 {
764 UV uv; 926 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
765 int numtype = grok_number (start, dec->cur - start, &uv); 927 if (*start == '-')
766 if (numtype & IS_NUMBER_IN_UV) 928 switch (dec->cur - start)
767 if (numtype & IS_NUMBER_NEG)
768 { 929 {
769 if (uv < (UV)IV_MIN) 930 case 2: return newSViv (-( start [1] - '0' * 1));
770 return newSViv (-(IV)uv); 931 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
932 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
933 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
771 } 934 }
935 else
936 switch (dec->cur - start)
937 {
938 case 1: return newSViv ( start [0] - '0' * 1);
939 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
940 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
941 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
942 }
943
944 {
945 UV uv;
946 int numtype = grok_number (start, dec->cur - start, &uv);
947 if (numtype & IS_NUMBER_IN_UV)
948 if (numtype & IS_NUMBER_NEG)
949 {
950 if (uv < (UV)IV_MIN)
951 return newSViv (-(IV)uv);
952 }
772 else 953 else
773 return newSVuv (uv); 954 return newSVuv (uv);
955
956 // here would likely be the place for bigint support
774 } 957 }
958 }
775 959
960 // if we ever support bigint or bigfloat, this is the place for bigfloat
776 return newSVnv (Atof (start)); 961 return newSVnv (Atof (start));
777 962
778fail: 963fail:
779 return 0; 964 return 0;
780} 965}
… …
782static SV * 967static SV *
783decode_av (dec_t *dec) 968decode_av (dec_t *dec)
784{ 969{
785 AV *av = newAV (); 970 AV *av = newAV ();
786 971
972 DEC_INC_DEPTH;
787 decode_ws (dec); 973 decode_ws (dec);
974
788 if (*dec->cur == ']') 975 if (*dec->cur == ']')
789 ++dec->cur; 976 ++dec->cur;
790 else 977 else
791 for (;;) 978 for (;;)
792 { 979 {
… …
810 ERR (", or ] expected while parsing array"); 997 ERR (", or ] expected while parsing array");
811 998
812 ++dec->cur; 999 ++dec->cur;
813 } 1000 }
814 1001
1002 DEC_DEC_DEPTH;
815 return newRV_noinc ((SV *)av); 1003 return newRV_noinc ((SV *)av);
816 1004
817fail: 1005fail:
818 SvREFCNT_dec (av); 1006 SvREFCNT_dec (av);
1007 DEC_DEC_DEPTH;
819 return 0; 1008 return 0;
820} 1009}
821 1010
822static SV * 1011static SV *
823decode_hv (dec_t *dec) 1012decode_hv (dec_t *dec)
824{ 1013{
1014 SV *sv;
825 HV *hv = newHV (); 1015 HV *hv = newHV ();
826 1016
1017 DEC_INC_DEPTH;
827 decode_ws (dec); 1018 decode_ws (dec);
1019
828 if (*dec->cur == '}') 1020 if (*dec->cur == '}')
829 ++dec->cur; 1021 ++dec->cur;
830 else 1022 else
831 for (;;) 1023 for (;;)
832 { 1024 {
833 SV *key, *value;
834
835 decode_ws (dec); EXPECT_CH ('"'); 1025 decode_ws (dec); EXPECT_CH ('"');
836 1026
837 key = decode_str (dec); 1027 // heuristic: assume that
838 if (!key) 1028 // a) decode_str + hv_store_ent are abysmally slow.
839 goto fail; 1029 // b) most hash keys are short, simple ascii text.
1030 // => try to "fast-match" such strings to avoid
1031 // the overhead of decode_str + hv_store_ent.
1032 {
1033 SV *value;
1034 char *p = dec->cur;
1035 char *e = p + 24; // only try up to 24 bytes
840 1036
841 decode_ws (dec); EXPECT_CH (':'); 1037 for (;;)
842
843 value = decode_sv (dec);
844 if (!value)
845 { 1038 {
1039 // the >= 0x80 is true on most architectures
1040 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1041 {
1042 // slow path, back up and use decode_str
1043 SV *key = decode_str (dec);
1044 if (!key)
1045 goto fail;
1046
1047 decode_ws (dec); EXPECT_CH (':');
1048
1049 value = decode_sv (dec);
1050 if (!value)
1051 {
1052 SvREFCNT_dec (key);
1053 goto fail;
1054 }
1055
1056 hv_store_ent (hv, key, value, 0);
846 SvREFCNT_dec (key); 1057 SvREFCNT_dec (key);
1058
1059 break;
1060 }
1061 else if (*p == '"')
1062 {
1063 // fast path, got a simple key
1064 char *key = dec->cur;
1065 int len = p - key;
1066 dec->cur = p + 1;
1067
1068 decode_ws (dec); EXPECT_CH (':');
1069
1070 value = decode_sv (dec);
1071 if (!value)
847 goto fail; 1072 goto fail;
1073
1074 hv_store (hv, key, len, value, 0);
1075
1076 break;
1077 }
1078
1079 ++p;
848 } 1080 }
849 1081 }
850 //TODO: optimise
851 hv_store_ent (hv, key, value, 0);
852 1082
853 decode_ws (dec); 1083 decode_ws (dec);
854 1084
855 if (*dec->cur == '}') 1085 if (*dec->cur == '}')
856 { 1086 {
… …
862 ERR (", or } expected while parsing object/hash"); 1092 ERR (", or } expected while parsing object/hash");
863 1093
864 ++dec->cur; 1094 ++dec->cur;
865 } 1095 }
866 1096
1097 DEC_DEC_DEPTH;
1098 sv = newRV_noinc ((SV *)hv);
1099
1100 // check filter callbacks
1101 if (dec->json.flags & F_HOOK)
1102 {
1103 ENTER; SAVETMPS;
1104
1105 if (HvKEYS (hv) == 1 && dec->json.cb_sk_object)
1106 {
1107 int count;
1108
1109 dSP; PUSHMARK (SP);
1110 XPUSHs (sv_2mortal (sv));
1111
1112 PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN;
1113
1114 if (count == 1)
1115 sv = newSVsv (POPs);
1116 else
1117 SvREFCNT_inc (sv);
1118 }
1119
1120 if (dec->json.cb_object)
1121 {
1122 int count;
1123
1124 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
1125 XPUSHs (sv_2mortal (sv));
1126
1127 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1128
1129 if (count == 1)
1130 sv = newSVsv (POPs);
1131 else
1132 SvREFCNT_inc (sv);
1133 }
1134
1135 FREETMPS; LEAVE;
1136 }
1137
867 return newRV_noinc ((SV *)hv); 1138 return newRV_noinc ((SV *)hv);
868 1139
869fail: 1140fail:
870 SvREFCNT_dec (hv); 1141 SvREFCNT_dec (hv);
1142 DEC_DEC_DEPTH;
871 return 0; 1143 return 0;
872} 1144}
873 1145
874static SV * 1146static SV *
875decode_sv (dec_t *dec) 1147decode_sv (dec_t *dec)
876{ 1148{
877 decode_ws (dec); 1149 decode_ws (dec);
1150
1151 // the beauty of JSON: you need exactly one character lookahead
1152 // to parse anything.
878 switch (*dec->cur) 1153 switch (*dec->cur)
879 { 1154 {
880 case '"': ++dec->cur; return decode_str (dec); 1155 case '"': ++dec->cur; return decode_str (dec);
881 case '[': ++dec->cur; return decode_av (dec); 1156 case '[': ++dec->cur; return decode_av (dec);
882 case '{': ++dec->cur; return decode_hv (dec); 1157 case '{': ++dec->cur; return decode_hv (dec);
… …
888 1163
889 case 't': 1164 case 't':
890 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1165 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
891 { 1166 {
892 dec->cur += 4; 1167 dec->cur += 4;
893 return newSViv (1); 1168 return SvREFCNT_inc (json_true);
894 } 1169 }
895 else 1170 else
896 ERR ("'true' expected"); 1171 ERR ("'true' expected");
897 1172
898 break; 1173 break;
899 1174
900 case 'f': 1175 case 'f':
901 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1176 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
902 { 1177 {
903 dec->cur += 5; 1178 dec->cur += 5;
904 return newSViv (0); 1179 return SvREFCNT_inc (json_false);
905 } 1180 }
906 else 1181 else
907 ERR ("'false' expected"); 1182 ERR ("'false' expected");
908 1183
909 break; 1184 break;
… …
918 ERR ("'null' expected"); 1193 ERR ("'null' expected");
919 1194
920 break; 1195 break;
921 1196
922 default: 1197 default:
923 ERR ("malformed json string, neither array, object, number, string or atom"); 1198 ERR ("malformed JSON string, neither array, object, number, string or atom");
924 break; 1199 break;
925 } 1200 }
926 1201
927fail: 1202fail:
928 return 0; 1203 return 0;
929} 1204}
930 1205
931static SV * 1206static SV *
932decode_json (SV *string, UV flags) 1207decode_json (SV *string, JSON *json, UV *offset_return)
933{ 1208{
1209 dec_t dec;
1210 UV offset;
934 SV *sv; 1211 SV *sv;
935 1212
1213 SvGETMAGIC (string);
1214 SvUPGRADE (string, SVt_PV);
1215
1216 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1217 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1218 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1219
936 if (flags & F_UTF8) 1220 if (json->flags & F_UTF8)
937 sv_utf8_downgrade (string, 0); 1221 sv_utf8_downgrade (string, 0);
938 else 1222 else
939 sv_utf8_upgrade (string); 1223 sv_utf8_upgrade (string);
940 1224
941 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1225 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
942 1226
943 dec_t dec; 1227 dec.json = *json;
944 dec.flags = flags;
945 dec.cur = SvPVX (string); 1228 dec.cur = SvPVX (string);
946 dec.end = SvEND (string); 1229 dec.end = SvEND (string);
947 dec.err = 0; 1230 dec.err = 0;
1231 dec.depth = 0;
1232 dec.maxdepth = DEC_DEPTH (dec.json.flags);
948 1233
1234 if (dec.json.cb_object || dec.json.cb_sk_object)
1235 dec.json.flags |= F_HOOK;
1236
1237 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
949 sv = decode_sv (&dec); 1238 sv = decode_sv (&dec);
950 1239
1240 if (!(offset_return || !sv))
1241 {
1242 // check for trailing garbage
1243 decode_ws (&dec);
1244
1245 if (*dec.cur)
1246 {
1247 dec.err = "garbage after JSON object";
1248 SvREFCNT_dec (sv);
1249 sv = 0;
1250 }
1251 }
1252
1253 if (offset_return || !sv)
1254 {
1255 offset = dec.json.flags & F_UTF8
1256 ? dec.cur - SvPVX (string)
1257 : utf8_distance (dec.cur, SvPVX (string));
1258
1259 if (offset_return)
1260 *offset_return = offset;
1261 }
1262
951 if (!sv) 1263 if (!sv)
952 { 1264 {
953 IV offset = dec.flags & F_UTF8
954 ? dec.cur - SvPVX (string)
955 : utf8_distance (dec.cur, SvPVX (string));
956 SV *uni = sv_newmortal (); 1265 SV *uni = sv_newmortal ();
957 1266
958 // horrible hack to silence warning inside pv_uni_display 1267 // horrible hack to silence warning inside pv_uni_display
959 COP cop = *PL_curcop; 1268 COP cop = *PL_curcop;
960 cop.cop_warnings = pWARN_NONE; 1269 cop.cop_warnings = pWARN_NONE;
… …
962 SAVEVPTR (PL_curcop); 1271 SAVEVPTR (PL_curcop);
963 PL_curcop = &cop; 1272 PL_curcop = &cop;
964 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1273 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
965 LEAVE; 1274 LEAVE;
966 1275
967 croak ("%s, at character offset %d (%s)", 1276 croak ("%s, at character offset %d [\"%s\"]",
968 dec.err, 1277 dec.err,
969 (int)offset, 1278 (int)offset,
970 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1279 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
971 } 1280 }
972 1281
973 sv = sv_2mortal (sv); 1282 sv = sv_2mortal (sv);
974 1283
975 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1284 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
976 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1285 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
977 1286
978 return sv; 1287 return sv;
979} 1288}
980 1289
… …
985 1294
986BOOT: 1295BOOT:
987{ 1296{
988 int i; 1297 int i;
989 1298
990 memset (decode_hexdigit, 0xff, 256);
991 for (i = 10; i--; ) 1299 for (i = 0; i < 256; ++i)
992 decode_hexdigit ['0' + i] = i; 1300 decode_hexdigit [i] =
1301 i >= '0' && i <= '9' ? i - '0'
1302 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1303 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1304 : -1;
993 1305
994 for (i = 7; i--; ) 1306 json_stash = gv_stashpv ("JSON::XS" , 1);
1307 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1308
1309 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1310 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1311}
1312
1313PROTOTYPES: DISABLE
1314
1315void new (char *klass)
1316 PPCODE:
1317{
1318 SV *pv = NEWSV (0, sizeof (JSON));
1319 SvPOK_only (pv);
1320 Zero (SvPVX (pv), 1, JSON);
1321 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1322 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1323}
1324
1325void ascii (JSON *self, int enable = 1)
1326 ALIAS:
1327 ascii = F_ASCII
1328 latin1 = F_LATIN1
1329 utf8 = F_UTF8
1330 indent = F_INDENT
1331 canonical = F_CANONICAL
1332 space_before = F_SPACE_BEFORE
1333 space_after = F_SPACE_AFTER
1334 pretty = F_PRETTY
1335 allow_nonref = F_ALLOW_NONREF
1336 shrink = F_SHRINK
1337 allow_blessed = F_ALLOW_BLESSED
1338 convert_blessed = F_CONV_BLESSED
1339 PPCODE:
1340{
1341 if (enable)
1342 self->flags |= ix;
1343 else
1344 self->flags &= ~ix;
1345
1346 XPUSHs (ST (0));
1347}
1348
1349void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1350 PPCODE:
1351{
1352 UV log2 = 0;
1353
1354 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1355
1356 while ((1UL << log2) < max_depth)
1357 ++log2;
1358
1359 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1360
1361 XPUSHs (ST (0));
1362}
1363
1364void max_size (JSON *self, UV max_size = 0)
1365 PPCODE:
1366{
1367 UV log2 = 0;
1368
1369 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1370 if (max_size == 1) max_size = 2;
1371
1372 while ((1UL << log2) < max_size)
1373 ++log2;
1374
1375 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1376
1377 XPUSHs (ST (0));
1378}
1379
1380void filter_json_objects (JSON *self, SV *cb = &PL_sv_undef)
1381 ALIAS:
1382 filter_sk_json_objects = 1
1383 PPCODE:
1384{
1385 if (!SvOK (cb))
1386 cb = 0;
1387
1388 switch (ix)
995 { 1389 {
996 decode_hexdigit ['a' + i] = 10 + i; 1390 case 0: self->cb_object = cb; break;
997 decode_hexdigit ['A' + i] = 10 + i; 1391 case 1: self->cb_sk_object = cb; break;
998 } 1392 }
999 1393
1000 json_stash = gv_stashpv ("JSON::XS", 1); 1394 XPUSHs (ST (0));
1001} 1395}
1002 1396
1003PROTOTYPES: DISABLE
1004
1005SV *new (char *dummy)
1006 CODE:
1007 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
1008 OUTPUT:
1009 RETVAL
1010
1011SV *ascii (SV *self, int enable = 1)
1012 ALIAS:
1013 ascii = F_ASCII
1014 utf8 = F_UTF8
1015 indent = F_INDENT
1016 canonical = F_CANONICAL
1017 space_before = F_SPACE_BEFORE
1018 space_after = F_SPACE_AFTER
1019 pretty = F_PRETTY
1020 allow_nonref = F_ALLOW_NONREF
1021 shrink = F_SHRINK
1022 CODE:
1023{
1024 UV *uv = SvJSON (self);
1025 if (enable)
1026 *uv |= ix;
1027 else
1028 *uv &= ~ix;
1029
1030 RETVAL = newSVsv (self);
1031}
1032 OUTPUT:
1033 RETVAL
1034
1035void encode (SV *self, SV *scalar) 1397void encode (JSON *self, SV *scalar)
1036 PPCODE: 1398 PPCODE:
1037 XPUSHs (encode_json (scalar, *SvJSON (self))); 1399 XPUSHs (encode_json (scalar, self));
1038 1400
1039void decode (SV *self, SV *jsonstr) 1401void decode (JSON *self, SV *jsonstr)
1040 PPCODE: 1402 PPCODE:
1041 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1403 XPUSHs (decode_json (jsonstr, self, 0));
1404
1405void decode_prefix (JSON *self, SV *jsonstr)
1406 PPCODE:
1407{
1408 UV offset;
1409 EXTEND (SP, 2);
1410 PUSHs (decode_json (jsonstr, self, &offset));
1411 PUSHs (sv_2mortal (newSVuv (offset)));
1412}
1042 1413
1043PROTOTYPES: ENABLE 1414PROTOTYPES: ENABLE
1044 1415
1045void to_json (SV *scalar) 1416void to_json (SV *scalar)
1046 ALIAS:
1047 objToJson = 0
1048 PPCODE: 1417 PPCODE:
1418{
1419 JSON json = { F_DEFAULT | F_UTF8 };
1049 XPUSHs (encode_json (scalar, F_UTF8)); 1420 XPUSHs (encode_json (scalar, &json));
1421}
1050 1422
1051void from_json (SV *jsonstr) 1423void from_json (SV *jsonstr)
1052 ALIAS:
1053 jsonToObj = 0
1054 PPCODE: 1424 PPCODE:
1425{
1426 JSON json = { F_DEFAULT | F_UTF8 };
1055 XPUSHs (decode_json (jsonstr, F_UTF8)); 1427 XPUSHs (decode_json (jsonstr, &json, 0));
1428}
1056 1429

Diff Legend

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