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 1.12 by root, Sat Mar 24 22:10:08 2007 UTC vs.
Revision 1.52 by root, Mon Jul 2 02:57:11 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
36
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 39
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 41#define F_DEFAULT (9UL << S_MAXDEPTH)
20 42
21#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
23 45
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#define SHORT_STRING_LEN 256 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
26 47
27#define SB do { 48#define SB do {
28#define SE } while (0) 49#define SE } while (0)
29 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
30static 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;
68 HV *cb_sk_object;
69} JSON;
31 70
32///////////////////////////////////////////////////////////////////////////// 71/////////////////////////////////////////////////////////////////////////////
33// utility functions 72// utility functions
34 73
35static UV * 74inline void
36SvJSON (SV *sv)
37{
38 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
39 croak ("object is not of type JSON::XS");
40
41 return &SvUVX (SvRV (sv));
42}
43
44static void
45shrink (SV *sv) 75shrink (SV *sv)
46{ 76{
47 sv_utf8_downgrade (sv, 1); 77 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 78 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 79 {
… …
53 SvPV_renew (sv, SvCUR (sv) + 1); 83 SvPV_renew (sv, SvCUR (sv) + 1);
54#endif 84#endif
55 } 85 }
56} 86}
57 87
88// decode an utf-8 character and return it, or (UV)-1 in
89// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints
93inline UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
99 {
100 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 }
103 else
104 {
105 *clen = (STRLEN)-1;
106 return (UV)-1;
107 }
108}
109
58///////////////////////////////////////////////////////////////////////////// 110/////////////////////////////////////////////////////////////////////////////
59// encoder 111// encoder
60 112
61// structure used for encoding JSON 113// structure used for encoding JSON
62typedef struct 114typedef struct
63{ 115{
64 char *cur; // SvPVX (sv) + current output position 116 char *cur; // SvPVX (sv) + current output position
65 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
66 SV *sv; // result scalar 118 SV *sv; // result scalar
67 UV flags; // F_* 119 JSON json;
68 int indent; // indentation level 120 U32 indent; // indentation level
69 int max_depth; // max. recursion level 121 U32 maxdepth; // max. indentation/recursion level
70} enc_t; 122} enc_t;
71 123
72static void 124inline void
73need (enc_t *enc, STRLEN len) 125need (enc_t *enc, STRLEN len)
74{ 126{
75 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
76 { 128 {
77 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
78 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
79 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
80 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
81 } 133 }
82} 134}
83 135
84static void 136inline void
85encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
86{ 138{
87 need (enc, 1); 139 need (enc, 1);
88 *enc->cur++ = ch; 140 *enc->cur++ = ch;
89} 141}
… …
97 149
98 while (str < end) 150 while (str < end)
99 { 151 {
100 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
101 153
102 if (ch >= 0x20 && ch < 0x80) // most common case 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
103 { 155 {
104 if (ch == '"') // but with slow exceptions 156 if (expect_false (ch == '"')) // but with slow exceptions
105 { 157 {
106 need (enc, len += 1); 158 need (enc, len += 1);
107 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
108 *enc->cur++ = '"'; 160 *enc->cur++ = '"';
109 } 161 }
110 else if (ch == '\\') 162 else if (expect_false (ch == '\\'))
111 { 163 {
112 need (enc, len += 1); 164 need (enc, len += 1);
113 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
114 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
115 } 167 }
… …
133 STRLEN clen; 185 STRLEN clen;
134 UV uch; 186 UV uch;
135 187
136 if (is_utf8) 188 if (is_utf8)
137 { 189 {
138 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 190 uch = decode_utf8 (str, end - str, &clen);
139 if (clen == (STRLEN)-1) 191 if (clen == (STRLEN)-1)
140 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 192 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
141 } 193 }
142 else 194 else
143 { 195 {
… …
146 } 198 }
147 199
148 if (uch > 0x10FFFFUL) 200 if (uch > 0x10FFFFUL)
149 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
150 202
151 if (uch < 0x80 || enc->flags & F_ASCII) 203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
152 { 204 {
153 if (uch > 0xFFFFUL) 205 if (uch > 0xFFFFUL)
154 { 206 {
155 need (enc, len += 11); 207 need (enc, len += 11);
156 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
170 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
171 } 223 }
172 224
173 str += clen; 225 str += clen;
174 } 226 }
227 else if (enc->json.flags & F_LATIN1)
228 {
229 *enc->cur++ = uch;
230 str += clen;
231 }
175 else if (is_utf8) 232 else if (is_utf8)
176 { 233 {
177 need (enc, len += clen); 234 need (enc, len += clen);
178 do 235 do
179 { 236 {
… …
181 } 238 }
182 while (--clen); 239 while (--clen);
183 } 240 }
184 else 241 else
185 { 242 {
186 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
187 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
188 ++str; 245 ++str;
189 } 246 }
190 } 247 }
191 } 248 }
… …
193 250
194 --len; 251 --len;
195 } 252 }
196} 253}
197 254
198static void 255inline void
199encode_indent (enc_t *enc) 256encode_indent (enc_t *enc)
200{ 257{
201 if (enc->flags & F_INDENT) 258 if (enc->json.flags & F_INDENT)
202 { 259 {
203 int spaces = enc->indent * INDENT_STEP; 260 int spaces = enc->indent * INDENT_STEP;
204 261
205 need (enc, spaces); 262 need (enc, spaces);
206 memset (enc->cur, ' ', spaces); 263 memset (enc->cur, ' ', spaces);
207 enc->cur += spaces; 264 enc->cur += spaces;
208 } 265 }
209} 266}
210 267
211static void 268inline void
212encode_space (enc_t *enc) 269encode_space (enc_t *enc)
213{ 270{
214 need (enc, 1); 271 need (enc, 1);
215 encode_ch (enc, ' '); 272 encode_ch (enc, ' ');
216} 273}
217 274
218static void 275inline void
219encode_nl (enc_t *enc) 276encode_nl (enc_t *enc)
220{ 277{
221 if (enc->flags & F_INDENT) 278 if (enc->json.flags & F_INDENT)
222 { 279 {
223 need (enc, 1); 280 need (enc, 1);
224 encode_ch (enc, '\n'); 281 encode_ch (enc, '\n');
225 } 282 }
226} 283}
227 284
228static void 285inline void
229encode_comma (enc_t *enc) 286encode_comma (enc_t *enc)
230{ 287{
231 encode_ch (enc, ','); 288 encode_ch (enc, ',');
232 289
233 if (enc->flags & F_INDENT) 290 if (enc->json.flags & F_INDENT)
234 encode_nl (enc); 291 encode_nl (enc);
235 else if (enc->flags & F_SPACE_AFTER) 292 else if (enc->json.flags & F_SPACE_AFTER)
236 encode_space (enc); 293 encode_space (enc);
237} 294}
238 295
239static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
240 297
241static void 298static void
242encode_av (enc_t *enc, AV *av) 299encode_av (enc_t *enc, AV *av)
243{ 300{
244 int i, len = av_len (av); 301 int i, len = av_len (av);
302
303 if (enc->indent >= enc->maxdepth)
304 croak ("data structure too deep (hit recursion limit)");
245 305
246 encode_ch (enc, '['); encode_nl (enc); 306 encode_ch (enc, '['); encode_nl (enc);
247 ++enc->indent; 307 ++enc->indent;
248 308
249 for (i = 0; i <= len; ++i) 309 for (i = 0; i <= len; ++i)
… …
280 else 340 else
281 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 341 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
282 342
283 encode_ch (enc, '"'); 343 encode_ch (enc, '"');
284 344
285 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 345 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
286 encode_ch (enc, ':'); 346 encode_ch (enc, ':');
287 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 347 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
288 encode_sv (enc, HeVAL (he)); 348 encode_sv (enc, HeVAL (he));
289} 349}
290 350
291// compare hash entries, used when all keys are bytestrings 351// compare hash entries, used when all keys are bytestrings
292static int 352static int
… …
315 375
316static void 376static void
317encode_hv (enc_t *enc, HV *hv) 377encode_hv (enc_t *enc, HV *hv)
318{ 378{
319 int count, i; 379 int count, i;
380
381 if (enc->indent >= enc->maxdepth)
382 croak ("data structure too deep (hit recursion limit)");
320 383
321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 384 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
322 385
323 if ((count = hv_iterinit (hv))) 386 if ((count = hv_iterinit (hv)))
324 { 387 {
325 // for canonical output we have to sort by keys first 388 // for canonical output we have to sort by keys first
326 // actually, this is mostly due to the stupid so-called 389 // actually, this is mostly due to the stupid so-called
327 // security workaround added somewhere in 5.8.x. 390 // security workaround added somewhere in 5.8.x.
328 // that randomises hash orderings 391 // that randomises hash orderings
329 if (enc->flags & F_CANONICAL) 392 if (enc->json.flags & F_CANONICAL)
330 { 393 {
331 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
332 int fast = 1; 394 int fast = 1;
395 HE *he;
396#if defined(__BORLANDC__) || defined(_MSC_VER)
397 HE **hes = _alloca (count * sizeof (HE));
398#else
399 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
400#endif
333 401
334 i = 0; 402 i = 0;
335 while ((he = hv_iternext (hv))) 403 while ((he = hv_iternext (hv)))
336 { 404 {
337 hes [i++] = he; 405 hes [i++] = he;
… …
372 440
373 encode_nl (enc); 441 encode_nl (enc);
374 } 442 }
375 else 443 else
376 { 444 {
377 SV *sv;
378 HE *he = hv_iternext (hv); 445 HE *he = hv_iternext (hv);
379 446
380 for (;;) 447 for (;;)
381 { 448 {
382 encode_indent (enc); 449 encode_indent (enc);
… …
391 encode_nl (enc); 458 encode_nl (enc);
392 } 459 }
393 } 460 }
394 461
395 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 462 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
463}
464
465// encode objects, arrays and special \0=false and \1=true values.
466static void
467encode_rv (enc_t *enc, SV *sv)
468{
469 svtype svt;
470
471 SvGETMAGIC (sv);
472 svt = SvTYPE (sv);
473
474 if (expect_false (SvOBJECT (sv)))
475 {
476 if (SvSTASH (sv) == json_boolean_stash)
477 {
478 if (SvIV (sv))
479 encode_str (enc, "true", 4, 0);
480 else
481 encode_str (enc, "false", 5, 0);
482 }
483 else
484 {
485#if 0
486 if (0 && sv_derived_from (rv, "JSON::Literal"))
487 {
488 // not yet
489 }
490#endif
491 if (enc->json.flags & F_CONV_BLESSED)
492 {
493 // we re-bless the reference to get overload and other niceties right
494 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
495
496 if (to_json)
497 {
498 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
499 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
500
501 // calling with G_SCALAR ensures that we always get a 1 reutrn value
502 // check anyways.
503 PUTBACK;
504 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
505 SPAGAIN;
506
507 encode_sv (enc, POPs);
508
509 FREETMPS; LEAVE;
510 }
511 else if (enc->json.flags & F_ALLOW_BLESSED)
512 encode_str (enc, "null", 4, 0);
513 else
514 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
515 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
516 }
517 else if (enc->json.flags & F_ALLOW_BLESSED)
518 encode_str (enc, "null", 4, 0);
519 else
520 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
522 }
523 }
524 else if (svt == SVt_PVHV)
525 encode_hv (enc, (HV *)sv);
526 else if (svt == SVt_PVAV)
527 encode_av (enc, (AV *)sv);
528 else if (svt < SVt_PVAV)
529 {
530 STRLEN len = 0;
531 char *pv = svt ? SvPV (sv, len) : 0;
532
533 if (len == 1 && *pv == '1')
534 encode_str (enc, "true", 4, 0);
535 else if (len == 1 && *pv == '0')
536 encode_str (enc, "false", 5, 0);
537 else
538 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540 }
541 else
542 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
543 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
396} 544}
397 545
398static void 546static void
399encode_sv (enc_t *enc, SV *sv) 547encode_sv (enc_t *enc, SV *sv)
400{ 548{
… …
408 encode_str (enc, str, len, SvUTF8 (sv)); 556 encode_str (enc, str, len, SvUTF8 (sv));
409 encode_ch (enc, '"'); 557 encode_ch (enc, '"');
410 } 558 }
411 else if (SvNOKp (sv)) 559 else if (SvNOKp (sv))
412 { 560 {
561 // trust that perl will do the right thing w.r.t. JSON syntax.
413 need (enc, NV_DIG + 32); 562 need (enc, NV_DIG + 32);
414 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 563 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
415 enc->cur += strlen (enc->cur); 564 enc->cur += strlen (enc->cur);
416 } 565 }
417 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
418 { 567 {
419 need (enc, 64); 568 // we assume we can always read an IV as a UV
569 if (SvUV (sv) & ~(UV)0x7fff)
570 {
571 // large integer, use the (rather slow) snprintf way.
572 need (enc, sizeof (UV) * 3);
420 enc->cur += 573 enc->cur +=
421 SvIsUV(sv) 574 SvIsUV(sv)
422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 575 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 576 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
577 }
578 else
579 {
580 // optimise the "small number case"
581 // code will likely be branchless and use only a single multiplication
582 I32 i = SvIV (sv);
583 U32 u;
584 char digit, nz = 0;
585
586 need (enc, 6);
587
588 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
589 u = i < 0 ? -i : i;
590
591 // convert to 4.28 fixed-point representation
592 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
593
594 // now output digit by digit, each time masking out the integer part
595 // and multiplying by 5 while moving the decimal point one to the right,
596 // resulting in a net multiplication by 10.
597 // we always write the digit to memory but conditionally increment
598 // the pointer, to ease the usage of conditional move instructions.
599 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
600 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
601 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
602 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
603 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
604 }
424 } 605 }
425 else if (SvROK (sv)) 606 else if (SvROK (sv))
426 { 607 encode_rv (enc, SvRV (sv));
427 SV *rv = SvRV (sv);
428
429 if (enc->indent >= enc->max_depth)
430 croak ("data structure too deep (hit recursion limit)");
431
432 switch (SvTYPE (rv))
433 {
434 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
435 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
436
437 default:
438 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
439 SvPV_nolen (sv));
440 }
441 }
442 else if (!SvOK (sv)) 608 else if (!SvOK (sv))
443 encode_str (enc, "null", 4, 0); 609 encode_str (enc, "null", 4, 0);
444 else 610 else
445 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 611 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
446 SvPV_nolen (sv), SvFLAGS (sv)); 612 SvPV_nolen (sv), SvFLAGS (sv));
447} 613}
448 614
449static SV * 615static SV *
450encode_json (SV *scalar, UV flags) 616encode_json (SV *scalar, JSON *json)
451{ 617{
618 enc_t enc;
619
452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 620 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
453 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 621 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
454 622
455 enc_t enc; 623 enc.json = *json;
456 enc.flags = flags;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 624 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 625 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 626 enc.end = SvEND (enc.sv);
460 enc.indent = 0; 627 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL; 628 enc.maxdepth = DEC_DEPTH (enc.json.flags);
462 629
463 SvPOK_only (enc.sv); 630 SvPOK_only (enc.sv);
464 encode_sv (&enc, scalar); 631 encode_sv (&enc, scalar);
465 632
633 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
634 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
635
466 if (!(flags & (F_ASCII | F_UTF8))) 636 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
467 SvUTF8_on (enc.sv); 637 SvUTF8_on (enc.sv);
468 638
469 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
470
471 if (enc.flags & F_SHRINK) 639 if (enc.json.flags & F_SHRINK)
472 shrink (enc.sv); 640 shrink (enc.sv);
473 641
474 return enc.sv; 642 return enc.sv;
475} 643}
476 644
… …
481typedef struct 649typedef struct
482{ 650{
483 char *cur; // current parser pointer 651 char *cur; // current parser pointer
484 char *end; // end of input string 652 char *end; // end of input string
485 const char *err; // parse error, if != 0 653 const char *err; // parse error, if != 0
486 UV flags; // F_* 654 JSON json;
655 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit
487} dec_t; 657} dec_t;
488 658
489static void 659inline void
490decode_ws (dec_t *dec) 660decode_ws (dec_t *dec)
491{ 661{
492 for (;;) 662 for (;;)
493 { 663 {
494 char ch = *dec->cur; 664 char ch = *dec->cur;
… …
500 ++dec->cur; 670 ++dec->cur;
501 } 671 }
502} 672}
503 673
504#define ERR(reason) SB dec->err = reason; goto fail; SE 674#define ERR(reason) SB dec->err = reason; goto fail; SE
675
505#define EXPECT_CH(ch) SB \ 676#define EXPECT_CH(ch) SB \
506 if (*dec->cur != ch) \ 677 if (*dec->cur != ch) \
507 ERR (# ch " expected"); \ 678 ERR (# ch " expected"); \
508 ++dec->cur; \ 679 ++dec->cur; \
509 SE 680 SE
510 681
682#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
683#define DEC_DEC_DEPTH --dec->depth
684
511static SV *decode_sv (dec_t *dec); 685static SV *decode_sv (dec_t *dec);
512 686
513static signed char decode_hexdigit[256]; 687static signed char decode_hexdigit[256];
514 688
515static UV 689static UV
516decode_4hex (dec_t *dec) 690decode_4hex (dec_t *dec)
517{ 691{
518 signed char d1, d2, d3, d4; 692 signed char d1, d2, d3, d4;
519 unsigned char *cur = (unsigned char *)dec->cur; 693 unsigned char *cur = (unsigned char *)dec->cur;
520 694
521 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 695 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
522 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 696 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
523 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 697 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
524 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 698 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
525 699
526 dec->cur += 4; 700 dec->cur += 4;
527 701
528 return ((UV)d1) << 12 702 return ((UV)d1) << 12
529 | ((UV)d2) << 8 703 | ((UV)d2) << 8
… …
537static SV * 711static SV *
538decode_str (dec_t *dec) 712decode_str (dec_t *dec)
539{ 713{
540 SV *sv = 0; 714 SV *sv = 0;
541 int utf8 = 0; 715 int utf8 = 0;
716 char *dec_cur = dec->cur;
542 717
543 do 718 do
544 { 719 {
545 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 720 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
546 char *cur = buf; 721 char *cur = buf;
547 722
548 do 723 do
549 { 724 {
550 unsigned char ch = *(unsigned char *)dec->cur++; 725 unsigned char ch = *(unsigned char *)dec_cur++;
551 726
552 if (ch == '"') 727 if (expect_false (ch == '"'))
553 { 728 {
554 --dec->cur; 729 --dec_cur;
555 break; 730 break;
556 } 731 }
557 else if (ch == '\\') 732 else if (expect_false (ch == '\\'))
558 { 733 {
559 switch (*dec->cur) 734 switch (*dec_cur)
560 { 735 {
561 case '\\': 736 case '\\':
562 case '/': 737 case '/':
563 case '"': *cur++ = *dec->cur++; break; 738 case '"': *cur++ = *dec_cur++; break;
564 739
565 case 'b': ++dec->cur; *cur++ = '\010'; break; 740 case 'b': ++dec_cur; *cur++ = '\010'; break;
566 case 't': ++dec->cur; *cur++ = '\011'; break; 741 case 't': ++dec_cur; *cur++ = '\011'; break;
567 case 'n': ++dec->cur; *cur++ = '\012'; break; 742 case 'n': ++dec_cur; *cur++ = '\012'; break;
568 case 'f': ++dec->cur; *cur++ = '\014'; break; 743 case 'f': ++dec_cur; *cur++ = '\014'; break;
569 case 'r': ++dec->cur; *cur++ = '\015'; break; 744 case 'r': ++dec_cur; *cur++ = '\015'; break;
570 745
571 case 'u': 746 case 'u':
572 { 747 {
573 UV lo, hi; 748 UV lo, hi;
574 ++dec->cur; 749 ++dec_cur;
575 750
751 dec->cur = dec_cur;
576 hi = decode_4hex (dec); 752 hi = decode_4hex (dec);
753 dec_cur = dec->cur;
577 if (hi == (UV)-1) 754 if (hi == (UV)-1)
578 goto fail; 755 goto fail;
579 756
580 // possibly a surrogate pair 757 // possibly a surrogate pair
581 if (hi >= 0xd800) 758 if (hi >= 0xd800)
582 if (hi < 0xdc00) 759 if (hi < 0xdc00)
583 { 760 {
584 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 761 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
585 ERR ("missing low surrogate character in surrogate pair"); 762 ERR ("missing low surrogate character in surrogate pair");
586 763
587 dec->cur += 2; 764 dec_cur += 2;
588 765
766 dec->cur = dec_cur;
589 lo = decode_4hex (dec); 767 lo = decode_4hex (dec);
768 dec_cur = dec->cur;
590 if (lo == (UV)-1) 769 if (lo == (UV)-1)
591 goto fail; 770 goto fail;
592 771
593 if (lo < 0xdc00 || lo >= 0xe000) 772 if (lo < 0xdc00 || lo >= 0xe000)
594 ERR ("surrogate pair expected"); 773 ERR ("surrogate pair expected");
… …
608 *cur++ = hi; 787 *cur++ = hi;
609 } 788 }
610 break; 789 break;
611 790
612 default: 791 default:
613 --dec->cur; 792 --dec_cur;
614 ERR ("illegal backslash escape sequence in string"); 793 ERR ("illegal backslash escape sequence in string");
615 } 794 }
616 } 795 }
617 else if (ch >= 0x20 && ch <= 0x7f) 796 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
618 *cur++ = ch; 797 *cur++ = ch;
619 else if (ch >= 0x80) 798 else if (ch >= 0x80)
620 { 799 {
621 --dec->cur;
622
623 STRLEN clen; 800 STRLEN clen;
624 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 801 UV uch;
802
803 --dec_cur;
804
805 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
625 if (clen == (STRLEN)-1) 806 if (clen == (STRLEN)-1)
626 ERR ("malformed UTF-8 character in JSON string"); 807 ERR ("malformed UTF-8 character in JSON string");
627 808
628 do 809 do
629 {
630 *cur++ = *dec->cur++; 810 *cur++ = *dec_cur++;
631 }
632 while (--clen); 811 while (--clen);
633 812
634 utf8 = 1; 813 utf8 = 1;
635 } 814 }
636 else if (!ch)
637 ERR ("unexpected end of string while parsing json string");
638 else 815 else
816 {
817 --dec_cur;
818
819 if (!ch)
820 ERR ("unexpected end of string while parsing JSON string");
821 else
639 ERR ("invalid character encountered"); 822 ERR ("invalid character encountered while parsing JSON string");
640 823 }
641 } 824 }
642 while (cur < buf + SHORT_STRING_LEN); 825 while (cur < buf + SHORT_STRING_LEN);
643 826
827 {
644 STRLEN len = cur - buf; 828 STRLEN len = cur - buf;
645 829
646 if (sv) 830 if (sv)
647 { 831 {
648 SvGROW (sv, SvCUR (sv) + len + 1); 832 SvGROW (sv, SvCUR (sv) + len + 1);
649 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 833 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
650 SvCUR_set (sv, SvCUR (sv) + len); 834 SvCUR_set (sv, SvCUR (sv) + len);
651 } 835 }
652 else 836 else
653 sv = newSVpvn (buf, len); 837 sv = newSVpvn (buf, len);
654 } 838 }
839 }
655 while (*dec->cur != '"'); 840 while (*dec_cur != '"');
656 841
657 ++dec->cur; 842 ++dec_cur;
658 843
659 if (sv) 844 if (sv)
660 { 845 {
661 SvPOK_only (sv); 846 SvPOK_only (sv);
662 *SvEND (sv) = 0; 847 *SvEND (sv) = 0;
… …
665 SvUTF8_on (sv); 850 SvUTF8_on (sv);
666 } 851 }
667 else 852 else
668 sv = newSVpvn ("", 0); 853 sv = newSVpvn ("", 0);
669 854
855 dec->cur = dec_cur;
670 return sv; 856 return sv;
671 857
672fail: 858fail:
859 dec->cur = dec_cur;
673 return 0; 860 return 0;
674} 861}
675 862
676static SV * 863static SV *
677decode_num (dec_t *dec) 864decode_num (dec_t *dec)
… …
735 is_nv = 1; 922 is_nv = 1;
736 } 923 }
737 924
738 if (!is_nv) 925 if (!is_nv)
739 { 926 {
740 UV uv; 927 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
741 int numtype = grok_number (start, dec->cur - start, &uv); 928 if (*start == '-')
742 if (numtype & IS_NUMBER_IN_UV) 929 switch (dec->cur - start)
743 if (numtype & IS_NUMBER_NEG)
744 { 930 {
745 if (uv < (UV)IV_MIN) 931 case 2: return newSViv (-( start [1] - '0' * 1));
746 return newSViv (-(IV)uv); 932 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
933 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
934 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
747 } 935 }
936 else
937 switch (dec->cur - start)
938 {
939 case 1: return newSViv ( start [0] - '0' * 1);
940 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
941 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
942 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
943 }
944
945 {
946 UV uv;
947 int numtype = grok_number (start, dec->cur - start, &uv);
948 if (numtype & IS_NUMBER_IN_UV)
949 if (numtype & IS_NUMBER_NEG)
950 {
951 if (uv < (UV)IV_MIN)
952 return newSViv (-(IV)uv);
953 }
748 else 954 else
749 return newSVuv (uv); 955 return newSVuv (uv);
956
957 // here would likely be the place for bigint support
750 } 958 }
959 }
751 960
961 // if we ever support bigint or bigfloat, this is the place for bigfloat
752 return newSVnv (Atof (start)); 962 return newSVnv (Atof (start));
753 963
754fail: 964fail:
755 return 0; 965 return 0;
756} 966}
… …
758static SV * 968static SV *
759decode_av (dec_t *dec) 969decode_av (dec_t *dec)
760{ 970{
761 AV *av = newAV (); 971 AV *av = newAV ();
762 972
973 DEC_INC_DEPTH;
763 decode_ws (dec); 974 decode_ws (dec);
975
764 if (*dec->cur == ']') 976 if (*dec->cur == ']')
765 ++dec->cur; 977 ++dec->cur;
766 else 978 else
767 for (;;) 979 for (;;)
768 { 980 {
… …
786 ERR (", or ] expected while parsing array"); 998 ERR (", or ] expected while parsing array");
787 999
788 ++dec->cur; 1000 ++dec->cur;
789 } 1001 }
790 1002
1003 DEC_DEC_DEPTH;
791 return newRV_noinc ((SV *)av); 1004 return newRV_noinc ((SV *)av);
792 1005
793fail: 1006fail:
794 SvREFCNT_dec (av); 1007 SvREFCNT_dec (av);
1008 DEC_DEC_DEPTH;
795 return 0; 1009 return 0;
796} 1010}
797 1011
798static SV * 1012static SV *
799decode_hv (dec_t *dec) 1013decode_hv (dec_t *dec)
800{ 1014{
1015 SV *sv;
801 HV *hv = newHV (); 1016 HV *hv = newHV ();
802 1017
1018 DEC_INC_DEPTH;
803 decode_ws (dec); 1019 decode_ws (dec);
1020
804 if (*dec->cur == '}') 1021 if (*dec->cur == '}')
805 ++dec->cur; 1022 ++dec->cur;
806 else 1023 else
807 for (;;) 1024 for (;;)
808 { 1025 {
809 SV *key, *value;
810
811 decode_ws (dec); EXPECT_CH ('"'); 1026 decode_ws (dec); EXPECT_CH ('"');
812 1027
813 key = decode_str (dec); 1028 // heuristic: assume that
814 if (!key) 1029 // a) decode_str + hv_store_ent are abysmally slow.
815 goto fail; 1030 // b) most hash keys are short, simple ascii text.
1031 // => try to "fast-match" such strings to avoid
1032 // the overhead of decode_str + hv_store_ent.
1033 {
1034 SV *value;
1035 char *p = dec->cur;
1036 char *e = p + 24; // only try up to 24 bytes
816 1037
817 decode_ws (dec); EXPECT_CH (':'); 1038 for (;;)
818
819 value = decode_sv (dec);
820 if (!value)
821 { 1039 {
1040 // the >= 0x80 is true on most architectures
1041 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1042 {
1043 // slow path, back up and use decode_str
1044 SV *key = decode_str (dec);
1045 if (!key)
1046 goto fail;
1047
1048 decode_ws (dec); EXPECT_CH (':');
1049
1050 value = decode_sv (dec);
1051 if (!value)
1052 {
1053 SvREFCNT_dec (key);
1054 goto fail;
1055 }
1056
1057 hv_store_ent (hv, key, value, 0);
822 SvREFCNT_dec (key); 1058 SvREFCNT_dec (key);
1059
1060 break;
1061 }
1062 else if (*p == '"')
1063 {
1064 // fast path, got a simple key
1065 char *key = dec->cur;
1066 int len = p - key;
1067 dec->cur = p + 1;
1068
1069 decode_ws (dec); EXPECT_CH (':');
1070
1071 value = decode_sv (dec);
1072 if (!value)
823 goto fail; 1073 goto fail;
1074
1075 hv_store (hv, key, len, value, 0);
1076
1077 break;
1078 }
1079
1080 ++p;
824 } 1081 }
825 1082 }
826 //TODO: optimise
827 hv_store_ent (hv, key, value, 0);
828 1083
829 decode_ws (dec); 1084 decode_ws (dec);
830 1085
831 if (*dec->cur == '}') 1086 if (*dec->cur == '}')
832 { 1087 {
… …
838 ERR (", or } expected while parsing object/hash"); 1093 ERR (", or } expected while parsing object/hash");
839 1094
840 ++dec->cur; 1095 ++dec->cur;
841 } 1096 }
842 1097
1098 DEC_DEC_DEPTH;
843 return newRV_noinc ((SV *)hv); 1099 sv = newRV_noinc ((SV *)hv);
1100
1101 // check filter callbacks
1102 if (dec->json.flags & F_HOOK)
1103 {
1104 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1105 {
1106 HE *cb, *he;
1107
1108 hv_iterinit (hv);
1109 he = hv_iternext (hv);
1110 hv_iterinit (hv);
1111
1112 // the next line creates a mortal sv each time its called.
1113 // might want to optimise this for common cases.
1114 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1115
1116 if (cb)
1117 {
1118 int count;
1119 ENTER; SAVETMPS;
1120
1121 dSP; PUSHMARK (SP);
1122 XPUSHs (HeVAL (he));
1123
1124 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1125
1126 if (count == 1)
1127 {
1128 sv = newSVsv (POPs);
1129 FREETMPS; LEAVE;
1130 return sv;
1131 }
1132
1133 FREETMPS; LEAVE;
1134 }
1135 }
1136
1137 if (dec->json.cb_object)
1138 {
1139 int count;
1140 ENTER; SAVETMPS;
1141
1142 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
1143 XPUSHs (sv_2mortal (sv));
1144
1145 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1146
1147 if (count == 1)
1148 {
1149 sv = newSVsv (POPs);
1150 FREETMPS; LEAVE;
1151 return sv;
1152 }
1153
1154 SvREFCNT_inc (sv);
1155 FREETMPS; LEAVE;
1156 }
1157 }
1158
1159 return sv;
844 1160
845fail: 1161fail:
846 SvREFCNT_dec (hv); 1162 SvREFCNT_dec (hv);
1163 DEC_DEC_DEPTH;
847 return 0; 1164 return 0;
848} 1165}
849 1166
850static SV * 1167static SV *
851decode_sv (dec_t *dec) 1168decode_sv (dec_t *dec)
852{ 1169{
853 decode_ws (dec); 1170 decode_ws (dec);
1171
1172 // the beauty of JSON: you need exactly one character lookahead
1173 // to parse anything.
854 switch (*dec->cur) 1174 switch (*dec->cur)
855 { 1175 {
856 case '"': ++dec->cur; return decode_str (dec); 1176 case '"': ++dec->cur; return decode_str (dec);
857 case '[': ++dec->cur; return decode_av (dec); 1177 case '[': ++dec->cur; return decode_av (dec);
858 case '{': ++dec->cur; return decode_hv (dec); 1178 case '{': ++dec->cur; return decode_hv (dec);
… …
864 1184
865 case 't': 1185 case 't':
866 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1186 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
867 { 1187 {
868 dec->cur += 4; 1188 dec->cur += 4;
869 return newSViv (1); 1189 return SvREFCNT_inc (json_true);
870 } 1190 }
871 else 1191 else
872 ERR ("'true' expected"); 1192 ERR ("'true' expected");
873 1193
874 break; 1194 break;
875 1195
876 case 'f': 1196 case 'f':
877 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1197 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
878 { 1198 {
879 dec->cur += 5; 1199 dec->cur += 5;
880 return newSViv (0); 1200 return SvREFCNT_inc (json_false);
881 } 1201 }
882 else 1202 else
883 ERR ("'false' expected"); 1203 ERR ("'false' expected");
884 1204
885 break; 1205 break;
… …
894 ERR ("'null' expected"); 1214 ERR ("'null' expected");
895 1215
896 break; 1216 break;
897 1217
898 default: 1218 default:
899 ERR ("malformed json string, neither array, object, number, string or atom"); 1219 ERR ("malformed JSON string, neither array, object, number, string or atom");
900 break; 1220 break;
901 } 1221 }
902 1222
903fail: 1223fail:
904 return 0; 1224 return 0;
905} 1225}
906 1226
907static SV * 1227static SV *
908decode_json (SV *string, UV flags) 1228decode_json (SV *string, JSON *json, UV *offset_return)
909{ 1229{
1230 dec_t dec;
1231 UV offset;
910 SV *sv; 1232 SV *sv;
911 1233
1234 SvGETMAGIC (string);
1235 SvUPGRADE (string, SVt_PV);
1236
1237 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1238 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1239 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1240
912 if (flags & F_UTF8) 1241 if (json->flags & F_UTF8)
913 sv_utf8_downgrade (string, 0); 1242 sv_utf8_downgrade (string, 0);
914 else 1243 else
915 sv_utf8_upgrade (string); 1244 sv_utf8_upgrade (string);
916 1245
917 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1246 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
918 1247
919 dec_t dec; 1248 dec.json = *json;
920 dec.flags = flags;
921 dec.cur = SvPVX (string); 1249 dec.cur = SvPVX (string);
922 dec.end = SvEND (string); 1250 dec.end = SvEND (string);
923 dec.err = 0; 1251 dec.err = 0;
1252 dec.depth = 0;
1253 dec.maxdepth = DEC_DEPTH (dec.json.flags);
924 1254
1255 if (dec.json.cb_object || dec.json.cb_sk_object)
1256 dec.json.flags |= F_HOOK;
1257
1258 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
925 sv = decode_sv (&dec); 1259 sv = decode_sv (&dec);
926 1260
1261 if (!(offset_return || !sv))
1262 {
1263 // check for trailing garbage
1264 decode_ws (&dec);
1265
1266 if (*dec.cur)
1267 {
1268 dec.err = "garbage after JSON object";
1269 SvREFCNT_dec (sv);
1270 sv = 0;
1271 }
1272 }
1273
1274 if (offset_return || !sv)
1275 {
1276 offset = dec.json.flags & F_UTF8
1277 ? dec.cur - SvPVX (string)
1278 : utf8_distance (dec.cur, SvPVX (string));
1279
1280 if (offset_return)
1281 *offset_return = offset;
1282 }
1283
927 if (!sv) 1284 if (!sv)
928 { 1285 {
929 IV offset = dec.flags & F_UTF8
930 ? dec.cur - SvPVX (string)
931 : utf8_distance (dec.cur, SvPVX (string));
932 SV *uni = sv_newmortal (); 1286 SV *uni = sv_newmortal ();
933 1287
934 // horrible hack to silence warning inside pv_uni_display 1288 // horrible hack to silence warning inside pv_uni_display
935 COP cop = *PL_curcop; 1289 COP cop = *PL_curcop;
936 cop.cop_warnings = pWARN_NONE; 1290 cop.cop_warnings = pWARN_NONE;
… …
938 SAVEVPTR (PL_curcop); 1292 SAVEVPTR (PL_curcop);
939 PL_curcop = &cop; 1293 PL_curcop = &cop;
940 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1294 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
941 LEAVE; 1295 LEAVE;
942 1296
943 croak ("%s, at character offset %d (%s)", 1297 croak ("%s, at character offset %d [\"%s\"]",
944 dec.err, 1298 dec.err,
945 (int)offset, 1299 (int)offset,
946 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1300 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
947 } 1301 }
948 1302
949 sv = sv_2mortal (sv); 1303 sv = sv_2mortal (sv);
950 1304
951 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1305 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
952 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1306 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
953 1307
954 return sv; 1308 return sv;
955} 1309}
956 1310
… …
961 1315
962BOOT: 1316BOOT:
963{ 1317{
964 int i; 1318 int i;
965 1319
966 memset (decode_hexdigit, 0xff, 256);
967 for (i = 10; i--; ) 1320 for (i = 0; i < 256; ++i)
968 decode_hexdigit ['0' + i] = i; 1321 decode_hexdigit [i] =
1322 i >= '0' && i <= '9' ? i - '0'
1323 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1324 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1325 : -1;
969 1326
970 for (i = 7; i--; ) 1327 json_stash = gv_stashpv ("JSON::XS" , 1);
1328 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1329
1330 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1331 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1332}
1333
1334PROTOTYPES: DISABLE
1335
1336void new (char *klass)
1337 PPCODE:
1338{
1339 SV *pv = NEWSV (0, sizeof (JSON));
1340 SvPOK_only (pv);
1341 Zero (SvPVX (pv), 1, JSON);
1342 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1343 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1344}
1345
1346void ascii (JSON *self, int enable = 1)
1347 ALIAS:
1348 ascii = F_ASCII
1349 latin1 = F_LATIN1
1350 utf8 = F_UTF8
1351 indent = F_INDENT
1352 canonical = F_CANONICAL
1353 space_before = F_SPACE_BEFORE
1354 space_after = F_SPACE_AFTER
1355 pretty = F_PRETTY
1356 allow_nonref = F_ALLOW_NONREF
1357 shrink = F_SHRINK
1358 allow_blessed = F_ALLOW_BLESSED
1359 convert_blessed = F_CONV_BLESSED
1360 PPCODE:
1361{
1362 if (enable)
1363 self->flags |= ix;
1364 else
1365 self->flags &= ~ix;
1366
1367 XPUSHs (ST (0));
1368}
1369
1370void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1371 PPCODE:
1372{
1373 UV log2 = 0;
1374
1375 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1376
1377 while ((1UL << log2) < max_depth)
1378 ++log2;
1379
1380 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1381
1382 XPUSHs (ST (0));
1383}
1384
1385void max_size (JSON *self, UV max_size = 0)
1386 PPCODE:
1387{
1388 UV log2 = 0;
1389
1390 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1391 if (max_size == 1) max_size = 2;
1392
1393 while ((1UL << log2) < max_size)
1394 ++log2;
1395
1396 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1397
1398 XPUSHs (ST (0));
1399}
1400
1401void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1402 PPCODE:
1403{
1404 SvREFCNT_dec (self->cb_object);
1405 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1406
1407 XPUSHs (ST (0));
1408}
1409
1410void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1411 PPCODE:
1412{
1413 if (!self->cb_sk_object)
1414 self->cb_sk_object = newHV ();
1415
1416 if (SvOK (cb))
1417 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1418 else
971 { 1419 {
972 decode_hexdigit ['a' + i] = 10 + i; 1420 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
973 decode_hexdigit ['A' + i] = 10 + i; 1421
1422 if (!HvKEYS (self->cb_sk_object))
1423 {
1424 SvREFCNT_dec (self->cb_sk_object);
1425 self->cb_sk_object = 0;
1426 }
974 } 1427 }
975 1428
976 json_stash = gv_stashpv ("JSON::XS", 1); 1429 XPUSHs (ST (0));
977} 1430}
978 1431
979PROTOTYPES: DISABLE 1432void encode (JSON *self, SV *scalar)
1433 PPCODE:
1434 XPUSHs (encode_json (scalar, self));
980 1435
981SV *new (char *dummy) 1436void decode (JSON *self, SV *jsonstr)
1437 PPCODE:
1438 XPUSHs (decode_json (jsonstr, self, 0));
1439
1440void decode_prefix (JSON *self, SV *jsonstr)
1441 PPCODE:
1442{
1443 UV offset;
1444 EXTEND (SP, 2);
1445 PUSHs (decode_json (jsonstr, self, &offset));
1446 PUSHs (sv_2mortal (newSVuv (offset)));
1447}
1448
1449void DESTROY (JSON *self)
982 CODE: 1450 CODE:
983 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1451 SvREFCNT_dec (self->cb_sk_object);
984 OUTPUT: 1452 SvREFCNT_dec (self->cb_object);
985 RETVAL
986
987SV *ascii (SV *self, int enable = 1)
988 ALIAS:
989 ascii = F_ASCII
990 utf8 = F_UTF8
991 indent = F_INDENT
992 canonical = F_CANONICAL
993 space_before = F_SPACE_BEFORE
994 space_after = F_SPACE_AFTER
995 pretty = F_PRETTY
996 allow_nonref = F_ALLOW_NONREF
997 shrink = F_SHRINK
998 CODE:
999{
1000 UV *uv = SvJSON (self);
1001 if (enable)
1002 *uv |= ix;
1003 else
1004 *uv &= ~ix;
1005
1006 RETVAL = newSVsv (self);
1007}
1008 OUTPUT:
1009 RETVAL
1010
1011void encode (SV *self, SV *scalar)
1012 PPCODE:
1013 XPUSHs (encode_json (scalar, *SvJSON (self)));
1014
1015void decode (SV *self, SV *jsonstr)
1016 PPCODE:
1017 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
1018 1453
1019PROTOTYPES: ENABLE 1454PROTOTYPES: ENABLE
1020 1455
1021void to_json (SV *scalar) 1456void to_json (SV *scalar)
1022 PPCODE: 1457 PPCODE:
1458{
1459 JSON json = { F_DEFAULT | F_UTF8 };
1023 XPUSHs (encode_json (scalar, F_UTF8)); 1460 XPUSHs (encode_json (scalar, &json));
1461}
1024 1462
1025void from_json (SV *jsonstr) 1463void from_json (SV *jsonstr)
1026 PPCODE: 1464 PPCODE:
1465{
1466 JSON json = { F_DEFAULT | F_UTF8 };
1027 XPUSHs (decode_json (jsonstr, F_UTF8)); 1467 XPUSHs (decode_json (jsonstr, &json, 0));
1468}
1028 1469

Diff Legend

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