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 1.13 by root, Sat Mar 24 22:55:16 2007 UTC vs.
Revision 1.55 by root, Mon Jul 23 22:57:40 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 {
… …
58// decode an utf-8 character and return it, or (UV)-1 in 88// decode an utf-8 character and return it, or (UV)-1 in
59// case of an error. 89// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF, 90// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else. 91// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints 92// note that we never call this function for a ascii codepoints
63static UV 93inline UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{ 95{
66 if (s[0] > 0xdf || s[0] < 0xc2) 96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69 { 99 {
70 *clen = 2; 100 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 102 }
73 else 103 else
104 {
105 *clen = (STRLEN)-1;
74 return (UV)-1; 106 return (UV)-1;
107 }
75} 108}
76 109
77///////////////////////////////////////////////////////////////////////////// 110/////////////////////////////////////////////////////////////////////////////
78// encoder 111// encoder
79 112
… …
81typedef struct 114typedef struct
82{ 115{
83 char *cur; // SvPVX (sv) + current output position 116 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
85 SV *sv; // result scalar 118 SV *sv; // result scalar
86 UV flags; // F_* 119 JSON json;
87 int indent; // indentation level 120 U32 indent; // indentation level
88 int max_depth; // max. recursion level 121 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 122} enc_t;
90 123
91static void 124inline void
92need (enc_t *enc, STRLEN len) 125need (enc_t *enc, STRLEN len)
93{ 126{
94 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
95 { 128 {
96 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
97 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
98 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100 } 133 }
101} 134}
102 135
103static void 136inline void
104encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
105{ 138{
106 need (enc, 1); 139 need (enc, 1);
107 *enc->cur++ = ch; 140 *enc->cur++ = ch;
108} 141}
… …
116 149
117 while (str < end) 150 while (str < end)
118 { 151 {
119 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
120 153
121 if (ch >= 0x20 && ch < 0x80) // most common case 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122 { 155 {
123 if (ch == '"') // but with slow exceptions 156 if (expect_false (ch == '"')) // but with slow exceptions
124 { 157 {
125 need (enc, len += 1); 158 need (enc, len += 1);
126 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
127 *enc->cur++ = '"'; 160 *enc->cur++ = '"';
128 } 161 }
129 else if (ch == '\\') 162 else if (expect_false (ch == '\\'))
130 { 163 {
131 need (enc, len += 1); 164 need (enc, len += 1);
132 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
133 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
134 } 167 }
… …
152 STRLEN clen; 185 STRLEN clen;
153 UV uch; 186 UV uch;
154 187
155 if (is_utf8) 188 if (is_utf8)
156 { 189 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen); 190 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1) 191 if (clen == (STRLEN)-1)
160 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);
161 } 193 }
162 else 194 else
… …
166 } 198 }
167 199
168 if (uch > 0x10FFFFUL) 200 if (uch > 0x10FFFFUL)
169 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);
170 202
171 if (uch < 0x80 || enc->flags & F_ASCII) 203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
172 { 204 {
173 if (uch > 0xFFFFUL) 205 if (uch > 0xFFFFUL)
174 { 206 {
175 need (enc, len += 11); 207 need (enc, len += 11);
176 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
190 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
191 } 223 }
192 224
193 str += clen; 225 str += clen;
194 } 226 }
227 else if (enc->json.flags & F_LATIN1)
228 {
229 *enc->cur++ = uch;
230 str += clen;
231 }
195 else if (is_utf8) 232 else if (is_utf8)
196 { 233 {
197 need (enc, len += clen); 234 need (enc, len += clen);
198 do 235 do
199 { 236 {
… …
201 } 238 }
202 while (--clen); 239 while (--clen);
203 } 240 }
204 else 241 else
205 { 242 {
206 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
207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str; 245 ++str;
209 } 246 }
210 } 247 }
211 } 248 }
… …
213 250
214 --len; 251 --len;
215 } 252 }
216} 253}
217 254
218static void 255inline void
219encode_indent (enc_t *enc) 256encode_indent (enc_t *enc)
220{ 257{
221 if (enc->flags & F_INDENT) 258 if (enc->json.flags & F_INDENT)
222 { 259 {
223 int spaces = enc->indent * INDENT_STEP; 260 int spaces = enc->indent * INDENT_STEP;
224 261
225 need (enc, spaces); 262 need (enc, spaces);
226 memset (enc->cur, ' ', spaces); 263 memset (enc->cur, ' ', spaces);
227 enc->cur += spaces; 264 enc->cur += spaces;
228 } 265 }
229} 266}
230 267
231static void 268inline void
232encode_space (enc_t *enc) 269encode_space (enc_t *enc)
233{ 270{
234 need (enc, 1); 271 need (enc, 1);
235 encode_ch (enc, ' '); 272 encode_ch (enc, ' ');
236} 273}
237 274
238static void 275inline void
239encode_nl (enc_t *enc) 276encode_nl (enc_t *enc)
240{ 277{
241 if (enc->flags & F_INDENT) 278 if (enc->json.flags & F_INDENT)
242 { 279 {
243 need (enc, 1); 280 need (enc, 1);
244 encode_ch (enc, '\n'); 281 encode_ch (enc, '\n');
245 } 282 }
246} 283}
247 284
248static void 285inline void
249encode_comma (enc_t *enc) 286encode_comma (enc_t *enc)
250{ 287{
251 encode_ch (enc, ','); 288 encode_ch (enc, ',');
252 289
253 if (enc->flags & F_INDENT) 290 if (enc->json.flags & F_INDENT)
254 encode_nl (enc); 291 encode_nl (enc);
255 else if (enc->flags & F_SPACE_AFTER) 292 else if (enc->json.flags & F_SPACE_AFTER)
256 encode_space (enc); 293 encode_space (enc);
257} 294}
258 295
259static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
260 297
261static void 298static void
262encode_av (enc_t *enc, AV *av) 299encode_av (enc_t *enc, AV *av)
263{ 300{
264 int i, len = av_len (av); 301 int i, len = av_len (av);
265 302
303 if (enc->indent >= enc->maxdepth)
304 croak ("data structure too deep (hit recursion limit)");
305
266 encode_ch (enc, '['); encode_nl (enc); 306 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 307 ++enc->indent;
268 308
269 for (i = 0; i <= len; ++i) 309 for (i = 0; i <= len; ++i)
270 { 310 {
311 SV **svp = av_fetch (av, i, 0);
312
271 encode_indent (enc); 313 encode_indent (enc);
272 encode_sv (enc, *av_fetch (av, i, 0)); 314
315 if (svp)
316 encode_sv (enc, *svp);
317 else
318 encode_str (enc, "null", 4, 0);
273 319
274 if (i < len) 320 if (i < len)
275 encode_comma (enc); 321 encode_comma (enc);
276 } 322 }
277 323
… …
300 else 346 else
301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 347 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
302 348
303 encode_ch (enc, '"'); 349 encode_ch (enc, '"');
304 350
305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
306 encode_ch (enc, ':'); 352 encode_ch (enc, ':');
307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
308 encode_sv (enc, HeVAL (he)); 354 encode_sv (enc, HeVAL (he));
309} 355}
310 356
311// compare hash entries, used when all keys are bytestrings 357// compare hash entries, used when all keys are bytestrings
312static int 358static int
… …
335 381
336static void 382static void
337encode_hv (enc_t *enc, HV *hv) 383encode_hv (enc_t *enc, HV *hv)
338{ 384{
339 int count, i; 385 int count, i;
386
387 if (enc->indent >= enc->maxdepth)
388 croak ("data structure too deep (hit recursion limit)");
340 389
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 390 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 391
343 if ((count = hv_iterinit (hv))) 392 if ((count = hv_iterinit (hv)))
344 { 393 {
345 // for canonical output we have to sort by keys first 394 // for canonical output we have to sort by keys first
346 // actually, this is mostly due to the stupid so-called 395 // actually, this is mostly due to the stupid so-called
347 // security workaround added somewhere in 5.8.x. 396 // security workaround added somewhere in 5.8.x.
348 // that randomises hash orderings 397 // that randomises hash orderings
349 if (enc->flags & F_CANONICAL) 398 if (enc->json.flags & F_CANONICAL)
350 { 399 {
351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
352 int fast = 1; 400 int fast = 1;
401 HE *he;
402#if defined(__BORLANDC__) || defined(_MSC_VER)
403 HE **hes = _alloca (count * sizeof (HE));
404#else
405 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
406#endif
353 407
354 i = 0; 408 i = 0;
355 while ((he = hv_iternext (hv))) 409 while ((he = hv_iternext (hv)))
356 { 410 {
357 hes [i++] = he; 411 hes [i++] = he;
… …
392 446
393 encode_nl (enc); 447 encode_nl (enc);
394 } 448 }
395 else 449 else
396 { 450 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 451 HE *he = hv_iternext (hv);
399 452
400 for (;;) 453 for (;;)
401 { 454 {
402 encode_indent (enc); 455 encode_indent (enc);
… …
411 encode_nl (enc); 464 encode_nl (enc);
412 } 465 }
413 } 466 }
414 467
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 468 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
469}
470
471// encode objects, arrays and special \0=false and \1=true values.
472static void
473encode_rv (enc_t *enc, SV *sv)
474{
475 svtype svt;
476
477 SvGETMAGIC (sv);
478 svt = SvTYPE (sv);
479
480 if (expect_false (SvOBJECT (sv)))
481 {
482 if (SvSTASH (sv) == json_boolean_stash)
483 {
484 if (SvIV (sv))
485 encode_str (enc, "true", 4, 0);
486 else
487 encode_str (enc, "false", 5, 0);
488 }
489 else
490 {
491#if 0
492 if (0 && sv_derived_from (rv, "JSON::Literal"))
493 {
494 // not yet
495 }
496#endif
497 if (enc->json.flags & F_CONV_BLESSED)
498 {
499 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
501
502 if (to_json)
503 {
504 int count;
505 dSP;
506
507 ENTER; SAVETMPS; PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509
510 // calling with G_SCALAR ensures that we always get a 1 return value
511 PUTBACK;
512 call_sv ((SV *)GvCV (to_json), G_SCALAR);
513 SPAGAIN;
514
515 // catch this surprisingly common error
516 if (SvROK (TOPs) && SvRV (TOPs) == sv)
517 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
518
519 sv = POPs;
520 PUTBACK;
521
522 encode_sv (enc, sv);
523
524 FREETMPS; LEAVE;
525 }
526 else if (enc->json.flags & F_ALLOW_BLESSED)
527 encode_str (enc, "null", 4, 0);
528 else
529 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
530 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
531 }
532 else if (enc->json.flags & F_ALLOW_BLESSED)
533 encode_str (enc, "null", 4, 0);
534 else
535 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
536 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
537 }
538 }
539 else if (svt == SVt_PVHV)
540 encode_hv (enc, (HV *)sv);
541 else if (svt == SVt_PVAV)
542 encode_av (enc, (AV *)sv);
543 else if (svt < SVt_PVAV)
544 {
545 STRLEN len = 0;
546 char *pv = svt ? SvPV (sv, len) : 0;
547
548 if (len == 1 && *pv == '1')
549 encode_str (enc, "true", 4, 0);
550 else if (len == 1 && *pv == '0')
551 encode_str (enc, "false", 5, 0);
552 else
553 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
554 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
555 }
556 else
557 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
558 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 559}
417 560
418static void 561static void
419encode_sv (enc_t *enc, SV *sv) 562encode_sv (enc_t *enc, SV *sv)
420{ 563{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 571 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 572 encode_ch (enc, '"');
430 } 573 }
431 else if (SvNOKp (sv)) 574 else if (SvNOKp (sv))
432 { 575 {
576 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 577 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 578 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 579 enc->cur += strlen (enc->cur);
436 } 580 }
437 else if (SvIOKp (sv)) 581 else if (SvIOKp (sv))
438 { 582 {
439 need (enc, 64); 583 // we assume we can always read an IV as a UV
584 if (SvUV (sv) & ~(UV)0x7fff)
585 {
586 // large integer, use the (rather slow) snprintf way.
587 need (enc, sizeof (UV) * 3);
440 enc->cur += 588 enc->cur +=
441 SvIsUV(sv) 589 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 590 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 591 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
592 }
593 else
594 {
595 // optimise the "small number case"
596 // code will likely be branchless and use only a single multiplication
597 I32 i = SvIV (sv);
598 U32 u;
599 char digit, nz = 0;
600
601 need (enc, 6);
602
603 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
604 u = i < 0 ? -i : i;
605
606 // convert to 4.28 fixed-point representation
607 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
608
609 // now output digit by digit, each time masking out the integer part
610 // and multiplying by 5 while moving the decimal point one to the right,
611 // resulting in a net multiplication by 10.
612 // we always write the digit to memory but conditionally increment
613 // the pointer, to ease the usage of conditional move instructions.
614 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
615 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
616 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
617 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
618 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
619 }
444 } 620 }
445 else if (SvROK (sv)) 621 else if (SvROK (sv))
446 { 622 encode_rv (enc, SvRV (sv));
447 SV *rv = SvRV (sv);
448
449 if (enc->indent >= enc->max_depth)
450 croak ("data structure too deep (hit recursion limit)");
451
452 switch (SvTYPE (rv))
453 {
454 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
455 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
456
457 default:
458 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
459 SvPV_nolen (sv));
460 }
461 }
462 else if (!SvOK (sv)) 623 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 624 encode_str (enc, "null", 4, 0);
464 else 625 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 626 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 627 SvPV_nolen (sv), SvFLAGS (sv));
467} 628}
468 629
469static SV * 630static SV *
470encode_json (SV *scalar, UV flags) 631encode_json (SV *scalar, JSON *json)
471{ 632{
633 enc_t enc;
634
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 635 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 636 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 637
475 enc_t enc; 638 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 639 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 640 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 641 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 642 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 643 enc.maxdepth = DEC_DEPTH (enc.json.flags);
482 644
483 SvPOK_only (enc.sv); 645 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 646 encode_sv (&enc, scalar);
485 647
648 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
649 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
650
486 if (!(flags & (F_ASCII | F_UTF8))) 651 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 652 SvUTF8_on (enc.sv);
488 653
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 654 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 655 shrink (enc.sv);
493 656
494 return enc.sv; 657 return enc.sv;
495} 658}
496 659
… …
501typedef struct 664typedef struct
502{ 665{
503 char *cur; // current parser pointer 666 char *cur; // current parser pointer
504 char *end; // end of input string 667 char *end; // end of input string
505 const char *err; // parse error, if != 0 668 const char *err; // parse error, if != 0
506 UV flags; // F_* 669 JSON json;
670 U32 depth; // recursion depth
671 U32 maxdepth; // recursion depth limit
507} dec_t; 672} dec_t;
508 673
509static void 674inline void
510decode_ws (dec_t *dec) 675decode_ws (dec_t *dec)
511{ 676{
512 for (;;) 677 for (;;)
513 { 678 {
514 char ch = *dec->cur; 679 char ch = *dec->cur;
… …
520 ++dec->cur; 685 ++dec->cur;
521 } 686 }
522} 687}
523 688
524#define ERR(reason) SB dec->err = reason; goto fail; SE 689#define ERR(reason) SB dec->err = reason; goto fail; SE
690
525#define EXPECT_CH(ch) SB \ 691#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 692 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 693 ERR (# ch " expected"); \
528 ++dec->cur; \ 694 ++dec->cur; \
529 SE 695 SE
530 696
697#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
698#define DEC_DEC_DEPTH --dec->depth
699
531static SV *decode_sv (dec_t *dec); 700static SV *decode_sv (dec_t *dec);
532 701
533static signed char decode_hexdigit[256]; 702static signed char decode_hexdigit[256];
534 703
535static UV 704static UV
536decode_4hex (dec_t *dec) 705decode_4hex (dec_t *dec)
537{ 706{
538 signed char d1, d2, d3, d4; 707 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 708 unsigned char *cur = (unsigned char *)dec->cur;
540 709
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 710 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
542 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 711 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
543 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 712 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
544 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 713 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 714
546 dec->cur += 4; 715 dec->cur += 4;
547 716
548 return ((UV)d1) << 12 717 return ((UV)d1) << 12
549 | ((UV)d2) << 8 718 | ((UV)d2) << 8
… …
557static SV * 726static SV *
558decode_str (dec_t *dec) 727decode_str (dec_t *dec)
559{ 728{
560 SV *sv = 0; 729 SV *sv = 0;
561 int utf8 = 0; 730 int utf8 = 0;
731 char *dec_cur = dec->cur;
562 732
563 do 733 do
564 { 734 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 735 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 736 char *cur = buf;
567 737
568 do 738 do
569 { 739 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 740 unsigned char ch = *(unsigned char *)dec_cur++;
571 741
572 if (ch == '"') 742 if (expect_false (ch == '"'))
573 { 743 {
574 --dec->cur; 744 --dec_cur;
575 break; 745 break;
576 } 746 }
577 else if (ch == '\\') 747 else if (expect_false (ch == '\\'))
578 { 748 {
579 switch (*dec->cur) 749 switch (*dec_cur)
580 { 750 {
581 case '\\': 751 case '\\':
582 case '/': 752 case '/':
583 case '"': *cur++ = *dec->cur++; break; 753 case '"': *cur++ = *dec_cur++; break;
584 754
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 755 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 756 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 757 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 758 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 759 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 760
591 case 'u': 761 case 'u':
592 { 762 {
593 UV lo, hi; 763 UV lo, hi;
594 ++dec->cur; 764 ++dec_cur;
595 765
766 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 767 hi = decode_4hex (dec);
768 dec_cur = dec->cur;
597 if (hi == (UV)-1) 769 if (hi == (UV)-1)
598 goto fail; 770 goto fail;
599 771
600 // possibly a surrogate pair 772 // possibly a surrogate pair
601 if (hi >= 0xd800) 773 if (hi >= 0xd800)
602 if (hi < 0xdc00) 774 if (hi < 0xdc00)
603 { 775 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 776 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 777 ERR ("missing low surrogate character in surrogate pair");
606 778
607 dec->cur += 2; 779 dec_cur += 2;
608 780
781 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 782 lo = decode_4hex (dec);
783 dec_cur = dec->cur;
610 if (lo == (UV)-1) 784 if (lo == (UV)-1)
611 goto fail; 785 goto fail;
612 786
613 if (lo < 0xdc00 || lo >= 0xe000) 787 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 788 ERR ("surrogate pair expected");
… …
628 *cur++ = hi; 802 *cur++ = hi;
629 } 803 }
630 break; 804 break;
631 805
632 default: 806 default:
633 --dec->cur; 807 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 808 ERR ("illegal backslash escape sequence in string");
635 } 809 }
636 } 810 }
637 else if (ch >= 0x20 && ch <= 0x7f) 811 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638 *cur++ = ch; 812 *cur++ = ch;
639 else if (ch >= 0x80) 813 else if (ch >= 0x80)
640 { 814 {
641 --dec->cur;
642
643 STRLEN clen; 815 STRLEN clen;
816 UV uch;
817
818 --dec_cur;
819
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 820 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 821 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 822 ERR ("malformed UTF-8 character in JSON string");
647 823
648 do 824 do
649 {
650 *cur++ = *dec->cur++; 825 *cur++ = *dec_cur++;
651 }
652 while (--clen); 826 while (--clen);
653 827
654 utf8 = 1; 828 utf8 = 1;
655 } 829 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 830 else
831 {
832 --dec_cur;
833
834 if (!ch)
835 ERR ("unexpected end of string while parsing JSON string");
836 else
659 ERR ("invalid character encountered"); 837 ERR ("invalid character encountered while parsing JSON string");
660 838 }
661 } 839 }
662 while (cur < buf + SHORT_STRING_LEN); 840 while (cur < buf + SHORT_STRING_LEN);
663 841
842 {
664 STRLEN len = cur - buf; 843 STRLEN len = cur - buf;
665 844
666 if (sv) 845 if (sv)
667 { 846 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 847 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 848 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 849 SvCUR_set (sv, SvCUR (sv) + len);
671 } 850 }
672 else 851 else
673 sv = newSVpvn (buf, len); 852 sv = newSVpvn (buf, len);
674 } 853 }
854 }
675 while (*dec->cur != '"'); 855 while (*dec_cur != '"');
676 856
677 ++dec->cur; 857 ++dec_cur;
678 858
679 if (sv) 859 if (sv)
680 { 860 {
681 SvPOK_only (sv); 861 SvPOK_only (sv);
682 *SvEND (sv) = 0; 862 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 865 SvUTF8_on (sv);
686 } 866 }
687 else 867 else
688 sv = newSVpvn ("", 0); 868 sv = newSVpvn ("", 0);
689 869
870 dec->cur = dec_cur;
690 return sv; 871 return sv;
691 872
692fail: 873fail:
874 dec->cur = dec_cur;
693 return 0; 875 return 0;
694} 876}
695 877
696static SV * 878static SV *
697decode_num (dec_t *dec) 879decode_num (dec_t *dec)
… …
755 is_nv = 1; 937 is_nv = 1;
756 } 938 }
757 939
758 if (!is_nv) 940 if (!is_nv)
759 { 941 {
760 UV uv; 942 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
761 int numtype = grok_number (start, dec->cur - start, &uv); 943 if (*start == '-')
762 if (numtype & IS_NUMBER_IN_UV) 944 switch (dec->cur - start)
763 if (numtype & IS_NUMBER_NEG)
764 { 945 {
765 if (uv < (UV)IV_MIN) 946 case 2: return newSViv (-( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 947 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
948 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
949 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767 } 950 }
951 else
952 switch (dec->cur - start)
953 {
954 case 1: return newSViv ( start [0] - '0' * 1);
955 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
956 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
957 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
958 }
959
960 {
961 UV uv;
962 int numtype = grok_number (start, dec->cur - start, &uv);
963 if (numtype & IS_NUMBER_IN_UV)
964 if (numtype & IS_NUMBER_NEG)
965 {
966 if (uv < (UV)IV_MIN)
967 return newSViv (-(IV)uv);
968 }
768 else 969 else
769 return newSVuv (uv); 970 return newSVuv (uv);
971
972 // here would likely be the place for bigint support
770 } 973 }
974 }
771 975
976 // if we ever support bigint or bigfloat, this is the place for bigfloat
772 return newSVnv (Atof (start)); 977 return newSVnv (Atof (start));
773 978
774fail: 979fail:
775 return 0; 980 return 0;
776} 981}
… …
778static SV * 983static SV *
779decode_av (dec_t *dec) 984decode_av (dec_t *dec)
780{ 985{
781 AV *av = newAV (); 986 AV *av = newAV ();
782 987
988 DEC_INC_DEPTH;
783 decode_ws (dec); 989 decode_ws (dec);
990
784 if (*dec->cur == ']') 991 if (*dec->cur == ']')
785 ++dec->cur; 992 ++dec->cur;
786 else 993 else
787 for (;;) 994 for (;;)
788 { 995 {
… …
806 ERR (", or ] expected while parsing array"); 1013 ERR (", or ] expected while parsing array");
807 1014
808 ++dec->cur; 1015 ++dec->cur;
809 } 1016 }
810 1017
1018 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1019 return newRV_noinc ((SV *)av);
812 1020
813fail: 1021fail:
814 SvREFCNT_dec (av); 1022 SvREFCNT_dec (av);
1023 DEC_DEC_DEPTH;
815 return 0; 1024 return 0;
816} 1025}
817 1026
818static SV * 1027static SV *
819decode_hv (dec_t *dec) 1028decode_hv (dec_t *dec)
820{ 1029{
1030 SV *sv;
821 HV *hv = newHV (); 1031 HV *hv = newHV ();
822 1032
1033 DEC_INC_DEPTH;
823 decode_ws (dec); 1034 decode_ws (dec);
1035
824 if (*dec->cur == '}') 1036 if (*dec->cur == '}')
825 ++dec->cur; 1037 ++dec->cur;
826 else 1038 else
827 for (;;) 1039 for (;;)
828 { 1040 {
829 SV *key, *value;
830
831 decode_ws (dec); EXPECT_CH ('"'); 1041 decode_ws (dec); EXPECT_CH ('"');
832 1042
833 key = decode_str (dec); 1043 // heuristic: assume that
834 if (!key) 1044 // a) decode_str + hv_store_ent are abysmally slow.
835 goto fail; 1045 // b) most hash keys are short, simple ascii text.
1046 // => try to "fast-match" such strings to avoid
1047 // the overhead of decode_str + hv_store_ent.
1048 {
1049 SV *value;
1050 char *p = dec->cur;
1051 char *e = p + 24; // only try up to 24 bytes
836 1052
837 decode_ws (dec); EXPECT_CH (':'); 1053 for (;;)
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1054 {
1055 // the >= 0x80 is true on most architectures
1056 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1057 {
1058 // slow path, back up and use decode_str
1059 SV *key = decode_str (dec);
1060 if (!key)
1061 goto fail;
1062
1063 decode_ws (dec); EXPECT_CH (':');
1064
1065 value = decode_sv (dec);
1066 if (!value)
1067 {
1068 SvREFCNT_dec (key);
1069 goto fail;
1070 }
1071
1072 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1073 SvREFCNT_dec (key);
1074
1075 break;
1076 }
1077 else if (*p == '"')
1078 {
1079 // fast path, got a simple key
1080 char *key = dec->cur;
1081 int len = p - key;
1082 dec->cur = p + 1;
1083
1084 decode_ws (dec); EXPECT_CH (':');
1085
1086 value = decode_sv (dec);
1087 if (!value)
843 goto fail; 1088 goto fail;
1089
1090 hv_store (hv, key, len, value, 0);
1091
1092 break;
1093 }
1094
1095 ++p;
844 } 1096 }
845 1097 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1098
849 decode_ws (dec); 1099 decode_ws (dec);
850 1100
851 if (*dec->cur == '}') 1101 if (*dec->cur == '}')
852 { 1102 {
… …
858 ERR (", or } expected while parsing object/hash"); 1108 ERR (", or } expected while parsing object/hash");
859 1109
860 ++dec->cur; 1110 ++dec->cur;
861 } 1111 }
862 1112
1113 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 1114 sv = newRV_noinc ((SV *)hv);
1115
1116 // check filter callbacks
1117 if (dec->json.flags & F_HOOK)
1118 {
1119 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1120 {
1121 HE *cb, *he;
1122
1123 hv_iterinit (hv);
1124 he = hv_iternext (hv);
1125 hv_iterinit (hv);
1126
1127 // the next line creates a mortal sv each time its called.
1128 // might want to optimise this for common cases.
1129 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1130
1131 if (cb)
1132 {
1133 dSP;
1134 int count;
1135
1136 ENTER; SAVETMPS; PUSHMARK (SP);
1137 XPUSHs (HeVAL (he));
1138
1139 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1140
1141 if (count == 1)
1142 {
1143 sv = newSVsv (POPs);
1144 FREETMPS; LEAVE;
1145 return sv;
1146 }
1147
1148 FREETMPS; LEAVE;
1149 }
1150 }
1151
1152 if (dec->json.cb_object)
1153 {
1154 dSP;
1155 int count;
1156
1157 ENTER; SAVETMPS; PUSHMARK (SP);
1158 XPUSHs (sv_2mortal (sv));
1159
1160 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1161
1162 if (count == 1)
1163 {
1164 sv = newSVsv (POPs);
1165 FREETMPS; LEAVE;
1166 return sv;
1167 }
1168
1169 SvREFCNT_inc (sv);
1170 FREETMPS; LEAVE;
1171 }
1172 }
1173
1174 return sv;
864 1175
865fail: 1176fail:
866 SvREFCNT_dec (hv); 1177 SvREFCNT_dec (hv);
1178 DEC_DEC_DEPTH;
867 return 0; 1179 return 0;
868} 1180}
869 1181
870static SV * 1182static SV *
871decode_sv (dec_t *dec) 1183decode_sv (dec_t *dec)
872{ 1184{
873 decode_ws (dec); 1185 decode_ws (dec);
1186
1187 // the beauty of JSON: you need exactly one character lookahead
1188 // to parse anything.
874 switch (*dec->cur) 1189 switch (*dec->cur)
875 { 1190 {
876 case '"': ++dec->cur; return decode_str (dec); 1191 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1192 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1193 case '{': ++dec->cur; return decode_hv (dec);
… …
884 1199
885 case 't': 1200 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1201 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1202 {
888 dec->cur += 4; 1203 dec->cur += 4;
889 return newSViv (1); 1204 return SvREFCNT_inc (json_true);
890 } 1205 }
891 else 1206 else
892 ERR ("'true' expected"); 1207 ERR ("'true' expected");
893 1208
894 break; 1209 break;
895 1210
896 case 'f': 1211 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1212 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1213 {
899 dec->cur += 5; 1214 dec->cur += 5;
900 return newSViv (0); 1215 return SvREFCNT_inc (json_false);
901 } 1216 }
902 else 1217 else
903 ERR ("'false' expected"); 1218 ERR ("'false' expected");
904 1219
905 break; 1220 break;
… …
914 ERR ("'null' expected"); 1229 ERR ("'null' expected");
915 1230
916 break; 1231 break;
917 1232
918 default: 1233 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1234 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1235 break;
921 } 1236 }
922 1237
923fail: 1238fail:
924 return 0; 1239 return 0;
925} 1240}
926 1241
927static SV * 1242static SV *
928decode_json (SV *string, UV flags) 1243decode_json (SV *string, JSON *json, UV *offset_return)
929{ 1244{
1245 dec_t dec;
1246 UV offset;
930 SV *sv; 1247 SV *sv;
931 1248
1249 SvGETMAGIC (string);
1250 SvUPGRADE (string, SVt_PV);
1251
1252 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1253 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1254 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1255
932 if (flags & F_UTF8) 1256 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1257 sv_utf8_downgrade (string, 0);
934 else 1258 else
935 sv_utf8_upgrade (string); 1259 sv_utf8_upgrade (string);
936 1260
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1261 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1262
939 dec_t dec; 1263 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1264 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1265 dec.end = SvEND (string);
943 dec.err = 0; 1266 dec.err = 0;
1267 dec.depth = 0;
1268 dec.maxdepth = DEC_DEPTH (dec.json.flags);
944 1269
1270 if (dec.json.cb_object || dec.json.cb_sk_object)
1271 dec.json.flags |= F_HOOK;
1272
1273 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
945 sv = decode_sv (&dec); 1274 sv = decode_sv (&dec);
946 1275
1276 if (!(offset_return || !sv))
1277 {
1278 // check for trailing garbage
1279 decode_ws (&dec);
1280
1281 if (*dec.cur)
1282 {
1283 dec.err = "garbage after JSON object";
1284 SvREFCNT_dec (sv);
1285 sv = 0;
1286 }
1287 }
1288
1289 if (offset_return || !sv)
1290 {
1291 offset = dec.json.flags & F_UTF8
1292 ? dec.cur - SvPVX (string)
1293 : utf8_distance (dec.cur, SvPVX (string));
1294
1295 if (offset_return)
1296 *offset_return = offset;
1297 }
1298
947 if (!sv) 1299 if (!sv)
948 { 1300 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1301 SV *uni = sv_newmortal ();
953 1302
954 // horrible hack to silence warning inside pv_uni_display 1303 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1304 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1305 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1307 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1308 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1309 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1310 LEAVE;
962 1311
963 croak ("%s, at character offset %d (%s)", 1312 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1313 dec.err,
965 (int)offset, 1314 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1315 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1316 }
968 1317
969 sv = sv_2mortal (sv); 1318 sv = sv_2mortal (sv);
970 1319
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1320 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
972 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1321 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1322
974 return sv; 1323 return sv;
975} 1324}
976 1325
… …
981 1330
982BOOT: 1331BOOT:
983{ 1332{
984 int i; 1333 int i;
985 1334
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1335 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1336 decode_hexdigit [i] =
1337 i >= '0' && i <= '9' ? i - '0'
1338 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1339 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1340 : -1;
989 1341
990 for (i = 7; i--; ) 1342 json_stash = gv_stashpv ("JSON::XS" , 1);
1343 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1344
1345 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1346 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1347}
1348
1349PROTOTYPES: DISABLE
1350
1351void new (char *klass)
1352 PPCODE:
1353{
1354 SV *pv = NEWSV (0, sizeof (JSON));
1355 SvPOK_only (pv);
1356 Zero (SvPVX (pv), 1, JSON);
1357 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1358 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1359}
1360
1361void ascii (JSON *self, int enable = 1)
1362 ALIAS:
1363 ascii = F_ASCII
1364 latin1 = F_LATIN1
1365 utf8 = F_UTF8
1366 indent = F_INDENT
1367 canonical = F_CANONICAL
1368 space_before = F_SPACE_BEFORE
1369 space_after = F_SPACE_AFTER
1370 pretty = F_PRETTY
1371 allow_nonref = F_ALLOW_NONREF
1372 shrink = F_SHRINK
1373 allow_blessed = F_ALLOW_BLESSED
1374 convert_blessed = F_CONV_BLESSED
1375 PPCODE:
1376{
1377 if (enable)
1378 self->flags |= ix;
1379 else
1380 self->flags &= ~ix;
1381
1382 XPUSHs (ST (0));
1383}
1384
1385void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1386 PPCODE:
1387{
1388 UV log2 = 0;
1389
1390 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1391
1392 while ((1UL << log2) < max_depth)
1393 ++log2;
1394
1395 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1396
1397 XPUSHs (ST (0));
1398}
1399
1400void max_size (JSON *self, UV max_size = 0)
1401 PPCODE:
1402{
1403 UV log2 = 0;
1404
1405 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1406 if (max_size == 1) max_size = 2;
1407
1408 while ((1UL << log2) < max_size)
1409 ++log2;
1410
1411 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1412
1413 XPUSHs (ST (0));
1414}
1415
1416void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1417 PPCODE:
1418{
1419 SvREFCNT_dec (self->cb_object);
1420 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1421
1422 XPUSHs (ST (0));
1423}
1424
1425void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1426 PPCODE:
1427{
1428 if (!self->cb_sk_object)
1429 self->cb_sk_object = newHV ();
1430
1431 if (SvOK (cb))
1432 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1433 else
991 { 1434 {
992 decode_hexdigit ['a' + i] = 10 + i; 1435 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
993 decode_hexdigit ['A' + i] = 10 + i; 1436
1437 if (!HvKEYS (self->cb_sk_object))
1438 {
1439 SvREFCNT_dec (self->cb_sk_object);
1440 self->cb_sk_object = 0;
1441 }
994 } 1442 }
995 1443
996 json_stash = gv_stashpv ("JSON::XS", 1); 1444 XPUSHs (ST (0));
997} 1445}
998 1446
999PROTOTYPES: DISABLE 1447void encode (JSON *self, SV *scalar)
1448 PPCODE:
1449 XPUSHs (encode_json (scalar, self));
1000 1450
1001SV *new (char *dummy) 1451void decode (JSON *self, SV *jsonstr)
1452 PPCODE:
1453 XPUSHs (decode_json (jsonstr, self, 0));
1454
1455void decode_prefix (JSON *self, SV *jsonstr)
1456 PPCODE:
1457{
1458 UV offset;
1459 EXTEND (SP, 2);
1460 PUSHs (decode_json (jsonstr, self, &offset));
1461 PUSHs (sv_2mortal (newSVuv (offset)));
1462}
1463
1464void DESTROY (JSON *self)
1002 CODE: 1465 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1466 SvREFCNT_dec (self->cb_sk_object);
1004 OUTPUT: 1467 SvREFCNT_dec (self->cb_object);
1005 RETVAL
1006
1007SV *ascii (SV *self, int enable = 1)
1008 ALIAS:
1009 ascii = F_ASCII
1010 utf8 = F_UTF8
1011 indent = F_INDENT
1012 canonical = F_CANONICAL
1013 space_before = F_SPACE_BEFORE
1014 space_after = F_SPACE_AFTER
1015 pretty = F_PRETTY
1016 allow_nonref = F_ALLOW_NONREF
1017 shrink = F_SHRINK
1018 CODE:
1019{
1020 UV *uv = SvJSON (self);
1021 if (enable)
1022 *uv |= ix;
1023 else
1024 *uv &= ~ix;
1025
1026 RETVAL = newSVsv (self);
1027}
1028 OUTPUT:
1029 RETVAL
1030
1031void encode (SV *self, SV *scalar)
1032 PPCODE:
1033 XPUSHs (encode_json (scalar, *SvJSON (self)));
1034
1035void decode (SV *self, SV *jsonstr)
1036 PPCODE:
1037 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
1038 1468
1039PROTOTYPES: ENABLE 1469PROTOTYPES: ENABLE
1040 1470
1041void to_json (SV *scalar) 1471void to_json (SV *scalar)
1042 PPCODE: 1472 PPCODE:
1473{
1474 JSON json = { F_DEFAULT | F_UTF8 };
1043 XPUSHs (encode_json (scalar, F_UTF8)); 1475 XPUSHs (encode_json (scalar, &json));
1476}
1044 1477
1045void from_json (SV *jsonstr) 1478void from_json (SV *jsonstr)
1046 PPCODE: 1479 PPCODE:
1480{
1481 JSON json = { F_DEFAULT | F_UTF8 };
1047 XPUSHs (decode_json (jsonstr, F_UTF8)); 1482 XPUSHs (decode_json (jsonstr, &json, 0));
1483}
1048 1484

Diff Legend

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