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 1.6 by root, Fri Mar 23 15:10:55 2007 UTC vs.
Revision 1.41 by root, Sat Jun 23 22:53:16 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.
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_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 27#define F_ALLOW_NONREF 0x00000100UL
17#define F_SHRINK 0x00000100 28#define F_SHRINK 0x00000200UL
29#define F_MAXDEPTH 0xf8000000UL
30#define S_MAXDEPTH 27
31
32#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
33
34// F_SELFCONVERT? <=> to_json/toJson
35// F_BLESSED? <=> { $__class__$ => }
18 36
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 37#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 38#define F_DEFAULT (9UL << S_MAXDEPTH)
21 39
22#define INIT_SIZE 32 // initial scalar size to be allocated 40#define INIT_SIZE 32 // initial scalar size to be allocated
41#define INDENT_STEP 3 // spaces per indentation level
42
43#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
23 44
24#define SB do { 45#define SB do {
25#define SE } while (0) 46#define SE } while (0)
26 47
48#if __GNUC__ >= 3
49# define expect(expr,value) __builtin_expect ((expr),(value))
50# define inline inline
51#else
52# define expect(expr,value) (expr)
53# define inline static
54#endif
55
56#define expect_false(expr) expect ((expr) != 0, 0)
57#define expect_true(expr) expect ((expr) != 0, 1)
58
27static HV *json_stash; 59static HV *json_stash; // JSON::XS::
60
61/////////////////////////////////////////////////////////////////////////////
62// utility functions
63
64static UV *
65SvJSON (SV *sv)
66{
67 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
68 croak ("object is not of type JSON::XS");
69
70 return &SvUVX (SvRV (sv));
71}
72
73static void
74shrink (SV *sv)
75{
76 sv_utf8_downgrade (sv, 1);
77 if (SvLEN (sv) > SvCUR (sv) + 1)
78 {
79#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew)
82 SvPV_renew (sv, SvCUR (sv) + 1);
83#endif
84 }
85}
86
87// decode an utf-8 character and return it, or (UV)-1 in
88// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints
92inline UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
98 {
99 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 }
102 else
103 {
104 *clen = (STRLEN)-1;
105 return (UV)-1;
106 }
107}
108
109/////////////////////////////////////////////////////////////////////////////
110// encoder
28 111
29// structure used for encoding JSON 112// structure used for encoding JSON
30typedef struct 113typedef struct
31{ 114{
32 char *cur; 115 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 116 char *end; // SvEND (sv)
35 SV *sv; 117 SV *sv; // result scalar
36 UV flags; 118 U32 flags; // F_*
37 int max_recurse; 119 U32 indent; // indentation level
38 int indent; 120 U32 maxdepth; // max. indentation/recursion level
39} enc_t; 121} enc_t;
40 122
41// structure used for decoding JSON 123inline void
42typedef struct
43{
44 char *cur;
45 char *end;
46 const char *err;
47 UV flags;
48} dec_t;
49
50static UV *
51SvJSON (SV *sv)
52{
53 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
54 croak ("object is not of type JSON::XS");
55
56 return &SvUVX (SvRV (sv));
57}
58
59/////////////////////////////////////////////////////////////////////////////
60
61static void
62need (enc_t *enc, STRLEN len) 124need (enc_t *enc, STRLEN len)
63{ 125{
64 if (enc->cur + len >= enc->end) 126 if (expect_false (enc->cur + len >= enc->end))
65 { 127 {
66 STRLEN cur = enc->cur - SvPVX (enc->sv); 128 STRLEN cur = enc->cur - SvPVX (enc->sv);
67 SvGROW (enc->sv, cur + len + 1); 129 SvGROW (enc->sv, cur + len + 1);
68 enc->cur = SvPVX (enc->sv) + cur; 130 enc->cur = SvPVX (enc->sv) + cur;
69 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
70 } 132 }
71} 133}
72 134
73static void 135inline void
74encode_ch (enc_t *enc, char ch) 136encode_ch (enc_t *enc, char ch)
75{ 137{
76 need (enc, 1); 138 need (enc, 1);
77 *enc->cur++ = ch; 139 *enc->cur++ = ch;
78} 140}
… …
86 148
87 while (str < end) 149 while (str < end)
88 { 150 {
89 unsigned char ch = *(unsigned char *)str; 151 unsigned char ch = *(unsigned char *)str;
90 152
91 if (ch >= 0x20 && ch < 0x80) // most common case 153 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
92 { 154 {
93 if (ch == '"') // but with slow exceptions 155 if (expect_false (ch == '"')) // but with slow exceptions
94 { 156 {
95 need (enc, len += 1); 157 need (enc, len += 1);
96 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
97 *enc->cur++ = '"'; 159 *enc->cur++ = '"';
98 } 160 }
99 else if (ch == '\\') 161 else if (expect_false (ch == '\\'))
100 { 162 {
101 need (enc, len += 1); 163 need (enc, len += 1);
102 *enc->cur++ = '\\'; 164 *enc->cur++ = '\\';
103 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
104 } 166 }
… …
122 STRLEN clen; 184 STRLEN clen;
123 UV uch; 185 UV uch;
124 186
125 if (is_utf8) 187 if (is_utf8)
126 { 188 {
127 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 189 uch = decode_utf8 (str, end - str, &clen);
128 if (clen == (STRLEN)-1) 190 if (clen == (STRLEN)-1)
129 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 191 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
130 } 192 }
131 else 193 else
132 { 194 {
133 uch = ch; 195 uch = ch;
134 clen = 1; 196 clen = 1;
135 } 197 }
136 198
137 if (uch < 0x80 || enc->flags & F_ASCII) 199 if (uch > 0x10FFFFUL)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
138 { 203 {
139 if (uch > 0xFFFFUL) 204 if (uch > 0xFFFFUL)
140 { 205 {
141 need (enc, len += 11); 206 need (enc, len += 11);
142 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
143 (uch - 0x10000) / 0x400 + 0xD800, 208 (int)((uch - 0x10000) / 0x400 + 0xD800),
144 (uch - 0x10000) % 0x400 + 0xDC00); 209 (int)((uch - 0x10000) % 0x400 + 0xDC00));
145 enc->cur += 12; 210 enc->cur += 12;
146 } 211 }
147 else 212 else
148 { 213 {
149 static char hexdigit [16] = "0123456789abcdef"; 214 static char hexdigit [16] = "0123456789abcdef";
… …
156 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
157 } 222 }
158 223
159 str += clen; 224 str += clen;
160 } 225 }
226 else if (enc->flags & F_LATIN1)
227 {
228 *enc->cur++ = uch;
229 str += clen;
230 }
161 else if (is_utf8) 231 else if (is_utf8)
162 { 232 {
163 need (enc, len += clen); 233 need (enc, len += clen);
164 do 234 do
165 { 235 {
… …
167 } 237 }
168 while (--clen); 238 while (--clen);
169 } 239 }
170 else 240 else
171 { 241 {
172 need (enc, 10); // never more than 11 bytes needed 242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str; 244 ++str;
175 } 245 }
176 } 246 }
177 } 247 }
… …
179 249
180 --len; 250 --len;
181 } 251 }
182} 252}
183 253
184#define INDENT SB \ 254inline void
255encode_indent (enc_t *enc)
256{
185 if (enc->flags & F_INDENT) \ 257 if (enc->flags & F_INDENT)
186 { \ 258 {
187 int i_; \ 259 int spaces = enc->indent * INDENT_STEP;
188 need (enc, enc->indent); \
189 for (i_ = enc->indent * 3; i_--; )\
190 encode_ch (enc, ' '); \
191 } \
192 SE
193 260
194#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 261 need (enc, spaces);
195#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 262 memset (enc->cur, ' ', spaces);
196#define COMMA SB \ 263 enc->cur += spaces;
264 }
265}
266
267inline void
268encode_space (enc_t *enc)
269{
270 need (enc, 1);
197 encode_ch (enc, ','); \ 271 encode_ch (enc, ' ');
272}
273
274inline void
275encode_nl (enc_t *enc)
276{
198 if (enc->flags & F_INDENT) \ 277 if (enc->flags & F_INDENT)
199 NL; \ 278 {
279 need (enc, 1);
280 encode_ch (enc, '\n');
281 }
282}
283
284inline void
285encode_comma (enc_t *enc)
286{
287 encode_ch (enc, ',');
288
289 if (enc->flags & F_INDENT)
290 encode_nl (enc);
200 else if (enc->flags & F_SPACE_AFTER) \ 291 else if (enc->flags & F_SPACE_AFTER)
201 SPACE; \ 292 encode_space (enc);
202 SE 293}
203 294
204static void encode_sv (enc_t *enc, SV *sv); 295static void encode_sv (enc_t *enc, SV *sv);
205 296
206static void 297static void
207encode_av (enc_t *enc, AV *av) 298encode_av (enc_t *enc, AV *av)
208{ 299{
209 int i, len = av_len (av); 300 int i, len = av_len (av);
210 301
211 encode_ch (enc, '['); NL; 302 if (enc->indent >= enc->maxdepth)
303 croak ("data structure too deep (hit recursion limit)");
304
305 encode_ch (enc, '['); encode_nl (enc);
212 ++enc->indent; 306 ++enc->indent;
213 307
214 for (i = 0; i <= len; ++i) 308 for (i = 0; i <= len; ++i)
215 { 309 {
216 INDENT; 310 encode_indent (enc);
217 encode_sv (enc, *av_fetch (av, i, 0)); 311 encode_sv (enc, *av_fetch (av, i, 0));
218 312
219 if (i < len) 313 if (i < len)
220 COMMA; 314 encode_comma (enc);
221 } 315 }
222 316
223 NL; 317 encode_nl (enc);
224 318
225 --enc->indent; 319 --enc->indent;
226 INDENT; encode_ch (enc, ']'); 320 encode_indent (enc); encode_ch (enc, ']');
227} 321}
228 322
229static void 323static void
230encode_he (enc_t *enc, HE *he) 324encode_he (enc_t *enc, HE *he)
231{ 325{
… …
245 else 339 else
246 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
247 341
248 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
249 343
250 if (enc->flags & F_SPACE_BEFORE) SPACE; 344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
251 encode_ch (enc, ':'); 345 encode_ch (enc, ':');
252 if (enc->flags & F_SPACE_AFTER ) SPACE; 346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
253 encode_sv (enc, HeVAL (he)); 347 encode_sv (enc, HeVAL (he));
254} 348}
255 349
256// compare hash entries, used when all keys are bytestrings 350// compare hash entries, used when all keys are bytestrings
257static int 351static int
… …
263 HE *b = *(HE **)b_; 357 HE *b = *(HE **)b_;
264 358
265 STRLEN la = HeKLEN (a); 359 STRLEN la = HeKLEN (a);
266 STRLEN lb = HeKLEN (b); 360 STRLEN lb = HeKLEN (b);
267 361
268 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
269 cmp = la < lb ? -1 : la == lb ? 0 : 1; 363 cmp = la - lb;
270 364
271 return cmp; 365 return cmp;
272} 366}
273 367
274// compare hash entries, used when some keys are sv's or utf-x 368// compare hash entries, used when some keys are sv's or utf-x
… …
281static void 375static void
282encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
283{ 377{
284 int count, i; 378 int count, i;
285 379
380 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)");
382
286 encode_ch (enc, '{'); NL; ++enc->indent; 383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
287 384
288 if ((count = hv_iterinit (hv))) 385 if ((count = hv_iterinit (hv)))
289 { 386 {
290 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
291 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
292 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
293 // that randomises hash orderings 390 // that randomises hash orderings
294 if (enc->flags & F_CANONICAL) 391 if (enc->flags & F_CANONICAL)
295 { 392 {
296 HE *he, *hes [count];
297 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
298 400
299 i = 0; 401 i = 0;
300 while ((he = hv_iternext (hv))) 402 while ((he = hv_iternext (hv)))
301 { 403 {
302 hes [i++] = he; 404 hes [i++] = he;
… …
308 410
309 if (fast) 411 if (fast)
310 qsort (hes, count, sizeof (HE *), he_cmp_fast); 412 qsort (hes, count, sizeof (HE *), he_cmp_fast);
311 else 413 else
312 { 414 {
313 // hack to disable "use bytes" 415 // hack to forcefully disable "use bytes"
314 COP *oldcop = PL_curcop, cop; 416 COP cop = *PL_curcop;
315 cop.op_private = 0; 417 cop.op_private = 0;
418
419 ENTER;
420 SAVETMPS;
421
422 SAVEVPTR (PL_curcop);
316 PL_curcop = &cop; 423 PL_curcop = &cop;
317 424
318 SAVETMPS;
319 qsort (hes, count, sizeof (HE *), he_cmp_slow); 425 qsort (hes, count, sizeof (HE *), he_cmp_slow);
426
320 FREETMPS; 427 FREETMPS;
321 428 LEAVE;
322 PL_curcop = oldcop;
323 } 429 }
324 430
325 for (i = 0; i < count; ++i) 431 for (i = 0; i < count; ++i)
326 { 432 {
327 INDENT; 433 encode_indent (enc);
328 encode_he (enc, hes [i]); 434 encode_he (enc, hes [i]);
329 435
330 if (i < count - 1) 436 if (i < count - 1)
331 COMMA; 437 encode_comma (enc);
332 } 438 }
333 439
334 NL; 440 encode_nl (enc);
335 } 441 }
336 else 442 else
337 { 443 {
338 SV *sv;
339 HE *he = hv_iternext (hv); 444 HE *he = hv_iternext (hv);
340 445
341 for (;;) 446 for (;;)
342 { 447 {
343 INDENT; 448 encode_indent (enc);
344 encode_he (enc, he); 449 encode_he (enc, he);
345 450
346 if (!(he = hv_iternext (hv))) 451 if (!(he = hv_iternext (hv)))
347 break; 452 break;
348 453
349 COMMA; 454 encode_comma (enc);
350 } 455 }
351 456
352 NL; 457 encode_nl (enc);
353 } 458 }
354 } 459 }
355 460
356 --enc->indent; INDENT; 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 (svt == SVt_PVHV)
474 encode_hv (enc, (HV *)sv);
475 else if (svt == SVt_PVAV)
476 encode_av (enc, (AV *)sv);
477 else if (svt < SVt_PVAV)
478 {
479 if (SvNIOK (sv) && SvIV (sv) == 0)
480 encode_str (enc, "false", 5, 0);
481 else if (SvNIOK (sv) && SvIV (sv) == 1)
482 encode_str (enc, "true", 4, 0);
483 else
484 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
486 }
487 else
488 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
489 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
357} 490}
358 491
359static void 492static void
360encode_sv (enc_t *enc, SV *sv) 493encode_sv (enc_t *enc, SV *sv)
361{ 494{
… …
369 encode_str (enc, str, len, SvUTF8 (sv)); 502 encode_str (enc, str, len, SvUTF8 (sv));
370 encode_ch (enc, '"'); 503 encode_ch (enc, '"');
371 } 504 }
372 else if (SvNOKp (sv)) 505 else if (SvNOKp (sv))
373 { 506 {
507 // trust that perl will do the right thing w.r.t. JSON syntax.
374 need (enc, NV_DIG + 32); 508 need (enc, NV_DIG + 32);
375 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 509 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
376 enc->cur += strlen (enc->cur); 510 enc->cur += strlen (enc->cur);
377 } 511 }
378 else if (SvIOKp (sv)) 512 else if (SvIOKp (sv))
379 { 513 {
380 need (enc, 64); 514 // we assume we can always read an IV as a UV
515 if (SvUV (sv) & ~(UV)0x7fff)
516 {
517 // large integer, use the (rather slow) snprintf way.
518 need (enc, sizeof (UV) * 3);
381 enc->cur += 519 enc->cur +=
382 SvIsUV(sv) 520 SvIsUV(sv)
383 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 521 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
384 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 522 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
523 }
524 else
525 {
526 // optimise the "small number case"
527 // code will likely be branchless and use only a single multiplication
528 I32 i = SvIV (sv);
529 U32 u;
530 char digit, nz = 0;
531
532 need (enc, 6);
533
534 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
535 u = i < 0 ? -i : i;
536
537 // convert to 4.28 fixed-point representation
538 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
539
540 // now output digit by digit, each time masking out the integer part
541 // and multiplying by 5 while moving the decimal point one to the right,
542 // resulting in a net multiplication by 10.
543 // we always write the digit to memory but conditionally increment
544 // the pointer, to ease the usage of conditional move instructions.
545 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
546 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
547 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
548 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
549 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
550 }
385 } 551 }
386 else if (SvROK (sv)) 552 else if (SvROK (sv))
387 { 553 encode_rv (enc, SvRV (sv));
388 if (!--enc->max_recurse)
389 croak ("data structure too deep (hit recursion limit)");
390
391 sv = SvRV (sv);
392
393 switch (SvTYPE (sv))
394 {
395 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
396 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
397
398 default:
399 croak ("JSON can only represent references to arrays or hashes");
400 }
401 }
402 else if (!SvOK (sv)) 554 else if (!SvOK (sv))
403 encode_str (enc, "null", 4, 0); 555 encode_str (enc, "null", 4, 0);
404 else 556 else
405 croak ("encountered perl type that JSON cannot handle"); 557 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
558 SvPV_nolen (sv), SvFLAGS (sv));
406} 559}
407 560
408static SV * 561static SV *
409encode_json (SV *scalar, UV flags) 562encode_json (SV *scalar, U32 flags)
410{ 563{
564 enc_t enc;
565
411 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 566 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
412 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 567 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
413 568
414 enc_t enc;
415 enc.flags = flags; 569 enc.flags = flags;
416 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 570 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
417 enc.cur = SvPVX (enc.sv); 571 enc.cur = SvPVX (enc.sv);
418 enc.end = SvEND (enc.sv); 572 enc.end = SvEND (enc.sv);
419 enc.max_recurse = 0;
420 enc.indent = 0; 573 enc.indent = 0;
574 enc.maxdepth = DEC_DEPTH (flags);
421 575
422 SvPOK_only (enc.sv); 576 SvPOK_only (enc.sv);
423 encode_sv (&enc, scalar); 577 encode_sv (&enc, scalar);
424 578
579 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
580 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
581
425 if (!(flags & (F_ASCII | F_UTF8))) 582 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
426 SvUTF8_on (enc.sv); 583 SvUTF8_on (enc.sv);
427 584
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429
430#ifdef SvPV_shrink_to_cur
431 if (enc.flags & F_SHRINK) 585 if (enc.flags & F_SHRINK)
432 SvPV_shrink_to_cur (enc.sv); 586 shrink (enc.sv);
433#endif 587
434 return enc.sv; 588 return enc.sv;
435} 589}
436 590
437///////////////////////////////////////////////////////////////////////////// 591/////////////////////////////////////////////////////////////////////////////
592// decoder
438 593
439#define WS \ 594// structure used for decoding JSON
595typedef struct
596{
597 char *cur; // current parser pointer
598 char *end; // end of input string
599 const char *err; // parse error, if != 0
600 U32 flags; // F_*
601 U32 depth; // recursion depth
602 U32 maxdepth; // recursion depth limit
603} dec_t;
604
605inline void
606decode_ws (dec_t *dec)
607{
440 for (;;) \ 608 for (;;)
441 { \ 609 {
442 char ch = *dec->cur; \ 610 char ch = *dec->cur;
611
443 if (ch > 0x20 \ 612 if (ch > 0x20
444 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 613 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
445 break; \ 614 break;
615
446 ++dec->cur; \ 616 ++dec->cur;
447 } 617 }
618}
448 619
449#define ERR(reason) SB dec->err = reason; goto fail; SE 620#define ERR(reason) SB dec->err = reason; goto fail; SE
621
450#define EXPECT_CH(ch) SB \ 622#define EXPECT_CH(ch) SB \
451 if (*dec->cur != ch) \ 623 if (*dec->cur != ch) \
452 ERR (# ch " expected"); \ 624 ERR (# ch " expected"); \
453 ++dec->cur; \ 625 ++dec->cur; \
454 SE 626 SE
455 627
628#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
629#define DEC_DEC_DEPTH --dec->depth
630
456static SV *decode_sv (dec_t *dec); 631static SV *decode_sv (dec_t *dec);
457 632
458static signed char decode_hexdigit[256]; 633static signed char decode_hexdigit[256];
459 634
460static UV 635static UV
461decode_4hex (dec_t *dec) 636decode_4hex (dec_t *dec)
462{ 637{
463 signed char d1, d2, d3, d4; 638 signed char d1, d2, d3, d4;
639 unsigned char *cur = (unsigned char *)dec->cur;
464 640
465 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 641 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
466 if (d1 < 0) ERR ("four hexadecimal digits expected"); 642 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
467 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 643 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
468 if (d2 < 0) ERR ("four hexadecimal digits expected"); 644 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
469 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
470 if (d3 < 0) ERR ("four hexadecimal digits expected");
471 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
472 if (d4 < 0) ERR ("four hexadecimal digits expected");
473 645
474 dec->cur += 4; 646 dec->cur += 4;
475 647
476 return ((UV)d1) << 12 648 return ((UV)d1) << 12
477 | ((UV)d2) << 8 649 | ((UV)d2) << 8
… …
480 652
481fail: 653fail:
482 return (UV)-1; 654 return (UV)-1;
483} 655}
484 656
485#define APPEND_GROW(n) SB \
486 if (cur + (n) >= end) \
487 { \
488 STRLEN ofs = cur - SvPVX (sv); \
489 SvGROW (sv, ofs + (n) + 1); \
490 cur = SvPVX (sv) + ofs; \
491 end = SvEND (sv); \
492 } \
493 SE
494
495#define APPEND_CH(ch) SB \
496 APPEND_GROW (1); \
497 *cur++ = (ch); \
498 SE
499
500static SV * 657static SV *
501decode_str (dec_t *dec) 658decode_str (dec_t *dec)
502{ 659{
503 SV *sv = NEWSV (0,2); 660 SV *sv = 0;
504 int utf8 = 0; 661 int utf8 = 0;
505 char *cur = SvPVX (sv); 662 char *dec_cur = dec->cur;
506 char *end = SvEND (sv);
507 663
508 for (;;) 664 do
509 { 665 {
510 unsigned char ch = *(unsigned char *)dec->cur; 666 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
667 char *cur = buf;
511 668
512 if (ch == '"') 669 do
513 break;
514 else if (ch == '\\')
515 { 670 {
516 switch (*++dec->cur) 671 unsigned char ch = *(unsigned char *)dec_cur++;
672
673 if (expect_false (ch == '"'))
517 { 674 {
518 case '\\': 675 --dec_cur;
519 case '/': 676 break;
520 case '"': APPEND_CH (*dec->cur++); break; 677 }
521 678 else if (expect_false (ch == '\\'))
522 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 679 {
523 case 't': APPEND_CH ('\011'); ++dec->cur; break; 680 switch (*dec_cur)
524 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
525 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
526 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
527
528 case 'u':
529 { 681 {
530 UV lo, hi; 682 case '\\':
531 ++dec->cur; 683 case '/':
684 case '"': *cur++ = *dec_cur++; break;
532 685
533 hi = decode_4hex (dec); 686 case 'b': ++dec_cur; *cur++ = '\010'; break;
534 if (hi == (UV)-1) 687 case 't': ++dec_cur; *cur++ = '\011'; break;
535 goto fail; 688 case 'n': ++dec_cur; *cur++ = '\012'; break;
689 case 'f': ++dec_cur; *cur++ = '\014'; break;
690 case 'r': ++dec_cur; *cur++ = '\015'; break;
536 691
537 // possibly a surrogate pair 692 case 'u':
538 if (hi >= 0xd800 && hi < 0xdc00)
539 { 693 {
540 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 694 UV lo, hi;
541 ERR ("missing low surrogate character in surrogate pair"); 695 ++dec_cur;
542 696
543 dec->cur += 2; 697 dec->cur = dec_cur;
544
545 lo = decode_4hex (dec); 698 hi = decode_4hex (dec);
699 dec_cur = dec->cur;
546 if (lo == (UV)-1) 700 if (hi == (UV)-1)
547 goto fail; 701 goto fail;
548 702
703 // possibly a surrogate pair
704 if (hi >= 0xd800)
705 if (hi < 0xdc00)
706 {
707 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
708 ERR ("missing low surrogate character in surrogate pair");
709
710 dec_cur += 2;
711
712 dec->cur = dec_cur;
713 lo = decode_4hex (dec);
714 dec_cur = dec->cur;
715 if (lo == (UV)-1)
716 goto fail;
717
549 if (lo < 0xdc00 || lo >= 0xe000) 718 if (lo < 0xdc00 || lo >= 0xe000)
550 ERR ("surrogate pair expected"); 719 ERR ("surrogate pair expected");
551 720
552 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 721 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
722 }
723 else if (hi < 0xe000)
724 ERR ("missing high surrogate character in surrogate pair");
725
726 if (hi >= 0x80)
727 {
728 utf8 = 1;
729
730 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
731 }
732 else
733 *cur++ = hi;
553 } 734 }
554 else if (hi >= 0xdc00 && hi < 0xe000)
555 ERR ("missing high surrogate character in surrogate pair");
556
557 if (hi >= 0x80)
558 { 735 break;
559 utf8 = 1;
560 736
561 APPEND_GROW (4); // at most 4 bytes for 21 bits
562 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
563 }
564 else 737 default:
565 APPEND_CH (hi); 738 --dec_cur;
739 ERR ("illegal backslash escape sequence in string");
566 } 740 }
567 break; 741 }
742 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
743 *cur++ = ch;
744 else if (ch >= 0x80)
745 {
746 STRLEN clen;
747 UV uch;
568 748
569 default:
570 --dec->cur; 749 --dec_cur;
571 ERR ("illegal backslash escape sequence in string"); 750
751 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
752 if (clen == (STRLEN)-1)
753 ERR ("malformed UTF-8 character in JSON string");
754
755 do
756 *cur++ = *dec_cur++;
757 while (--clen);
758
759 utf8 = 1;
760 }
761 else
762 {
763 --dec_cur;
764
765 if (!ch)
766 ERR ("unexpected end of string while parsing JSON string");
767 else
768 ERR ("invalid character encountered while parsing JSON string");
572 } 769 }
573 } 770 }
574 else if (ch >= 0x20 && ch <= 0x7f) 771 while (cur < buf + SHORT_STRING_LEN);
575 APPEND_CH (*dec->cur++); 772
576 else if (ch >= 0x80)
577 { 773 {
578 STRLEN clen; 774 STRLEN len = cur - buf;
579 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
580 if (clen == (STRLEN)-1)
581 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
582 775
583 APPEND_GROW (clen); 776 if (sv)
584 do
585 { 777 {
586 *cur++ = *dec->cur++; 778 SvGROW (sv, SvCUR (sv) + len + 1);
779 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
780 SvCUR_set (sv, SvCUR (sv) + len);
587 } 781 }
588 while (--clen);
589
590 utf8 = 1;
591 }
592 else if (dec->cur == dec->end)
593 ERR ("unexpected end of string while parsing json string");
594 else 782 else
595 ERR ("invalid character encountered"); 783 sv = newSVpvn (buf, len);
596 } 784 }
785 }
786 while (*dec_cur != '"');
597 787
598 ++dec->cur; 788 ++dec_cur;
599 789
600 SvCUR_set (sv, cur - SvPVX (sv)); 790 if (sv)
601 791 {
602 SvPOK_only (sv); 792 SvPOK_only (sv);
603 *SvEND (sv) = 0; 793 *SvEND (sv) = 0;
604 794
605 if (utf8) 795 if (utf8)
606 SvUTF8_on (sv); 796 SvUTF8_on (sv);
797 }
798 else
799 sv = newSVpvn ("", 0);
607 800
608#ifdef SvPV_shrink_to_cur 801 dec->cur = dec_cur;
609 if (dec->flags & F_SHRINK)
610 SvPV_shrink_to_cur (sv);
611#endif
612
613 return sv; 802 return sv;
614 803
615fail: 804fail:
616 SvREFCNT_dec (sv); 805 dec->cur = dec_cur;
617 return 0; 806 return 0;
618} 807}
619 808
620static SV * 809static SV *
621decode_num (dec_t *dec) 810decode_num (dec_t *dec)
… …
679 is_nv = 1; 868 is_nv = 1;
680 } 869 }
681 870
682 if (!is_nv) 871 if (!is_nv)
683 { 872 {
684 UV uv; 873 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
685 int numtype = grok_number (start, dec->cur - start, &uv); 874 if (*start == '-')
686 if (numtype & IS_NUMBER_IN_UV) 875 switch (dec->cur - start)
687 if (numtype & IS_NUMBER_NEG)
688 { 876 {
689 if (uv < (UV)IV_MIN) 877 case 2: return newSViv (-( start [1] - '0' * 1));
690 return newSViv (-(IV)uv); 878 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
879 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
880 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
691 } 881 }
882 else
883 switch (dec->cur - start)
884 {
885 case 1: return newSViv ( start [0] - '0' * 1);
886 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
887 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
888 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
889 }
890
891 {
892 UV uv;
893 int numtype = grok_number (start, dec->cur - start, &uv);
894 if (numtype & IS_NUMBER_IN_UV)
895 if (numtype & IS_NUMBER_NEG)
896 {
897 if (uv < (UV)IV_MIN)
898 return newSViv (-(IV)uv);
899 }
692 else 900 else
693 return newSVuv (uv); 901 return newSVuv (uv);
902
903 // here would likely be the place for bigint support
694 } 904 }
905 }
695 906
907 // if we ever support bigint or bigfloat, this is the place for bigfloat
696 return newSVnv (Atof (start)); 908 return newSVnv (Atof (start));
697 909
698fail: 910fail:
699 return 0; 911 return 0;
700} 912}
… …
702static SV * 914static SV *
703decode_av (dec_t *dec) 915decode_av (dec_t *dec)
704{ 916{
705 AV *av = newAV (); 917 AV *av = newAV ();
706 918
707 WS; 919 DEC_INC_DEPTH;
920 decode_ws (dec);
921
708 if (*dec->cur == ']') 922 if (*dec->cur == ']')
709 ++dec->cur; 923 ++dec->cur;
710 else 924 else
711 for (;;) 925 for (;;)
712 { 926 {
… …
716 if (!value) 930 if (!value)
717 goto fail; 931 goto fail;
718 932
719 av_push (av, value); 933 av_push (av, value);
720 934
721 WS; 935 decode_ws (dec);
722 936
723 if (*dec->cur == ']') 937 if (*dec->cur == ']')
724 { 938 {
725 ++dec->cur; 939 ++dec->cur;
726 break; 940 break;
… …
730 ERR (", or ] expected while parsing array"); 944 ERR (", or ] expected while parsing array");
731 945
732 ++dec->cur; 946 ++dec->cur;
733 } 947 }
734 948
949 DEC_DEC_DEPTH;
735 return newRV_noinc ((SV *)av); 950 return newRV_noinc ((SV *)av);
736 951
737fail: 952fail:
738 SvREFCNT_dec (av); 953 SvREFCNT_dec (av);
954 DEC_DEC_DEPTH;
739 return 0; 955 return 0;
740} 956}
741 957
742static SV * 958static SV *
743decode_hv (dec_t *dec) 959decode_hv (dec_t *dec)
744{ 960{
745 HV *hv = newHV (); 961 HV *hv = newHV ();
746 962
747 WS; 963 DEC_INC_DEPTH;
964 decode_ws (dec);
965
748 if (*dec->cur == '}') 966 if (*dec->cur == '}')
749 ++dec->cur; 967 ++dec->cur;
750 else 968 else
751 for (;;) 969 for (;;)
752 { 970 {
753 SV *key, *value; 971 SV *key, *value;
754 972
755 WS; EXPECT_CH ('"'); 973 decode_ws (dec); EXPECT_CH ('"');
756 974
757 key = decode_str (dec); 975 key = decode_str (dec);
758 if (!key) 976 if (!key)
759 goto fail; 977 goto fail;
760 978
761 WS; EXPECT_CH (':'); 979 decode_ws (dec); EXPECT_CH (':');
762 980
763 value = decode_sv (dec); 981 value = decode_sv (dec);
764 if (!value) 982 if (!value)
765 { 983 {
766 SvREFCNT_dec (key); 984 SvREFCNT_dec (key);
767 goto fail; 985 goto fail;
768 } 986 }
769 987
770 //TODO: optimise
771 hv_store_ent (hv, key, value, 0); 988 hv_store_ent (hv, key, value, 0);
989 SvREFCNT_dec (key);
772 990
773 WS; 991 decode_ws (dec);
774 992
775 if (*dec->cur == '}') 993 if (*dec->cur == '}')
776 { 994 {
777 ++dec->cur; 995 ++dec->cur;
778 break; 996 break;
… …
782 ERR (", or } expected while parsing object/hash"); 1000 ERR (", or } expected while parsing object/hash");
783 1001
784 ++dec->cur; 1002 ++dec->cur;
785 } 1003 }
786 1004
1005 DEC_DEC_DEPTH;
787 return newRV_noinc ((SV *)hv); 1006 return newRV_noinc ((SV *)hv);
788 1007
789fail: 1008fail:
790 SvREFCNT_dec (hv); 1009 SvREFCNT_dec (hv);
1010 DEC_DEC_DEPTH;
791 return 0; 1011 return 0;
792} 1012}
793 1013
794static SV * 1014static SV *
795decode_sv (dec_t *dec) 1015decode_sv (dec_t *dec)
796{ 1016{
797 WS; 1017 decode_ws (dec);
1018
1019 // the beauty of JSON: you need exactly one character lookahead
1020 // to parse anything.
798 switch (*dec->cur) 1021 switch (*dec->cur)
799 { 1022 {
800 case '"': ++dec->cur; return decode_str (dec); 1023 case '"': ++dec->cur; return decode_str (dec);
801 case '[': ++dec->cur; return decode_av (dec); 1024 case '[': ++dec->cur; return decode_av (dec);
802 case '{': ++dec->cur; return decode_hv (dec); 1025 case '{': ++dec->cur; return decode_hv (dec);
… …
838 ERR ("'null' expected"); 1061 ERR ("'null' expected");
839 1062
840 break; 1063 break;
841 1064
842 default: 1065 default:
843 ERR ("malformed json string"); 1066 ERR ("malformed JSON string, neither array, object, number, string or atom");
844 break; 1067 break;
845 } 1068 }
846 1069
847fail: 1070fail:
848 return 0; 1071 return 0;
849} 1072}
850 1073
851static SV * 1074static SV *
852decode_json (SV *string, UV flags) 1075decode_json (SV *string, U32 flags, UV *offset_return)
853{ 1076{
1077 dec_t dec;
1078 UV offset;
854 SV *sv; 1079 SV *sv;
1080
1081 SvGETMAGIC (string);
1082 SvUPGRADE (string, SVt_PV);
855 1083
856 if (flags & F_UTF8) 1084 if (flags & F_UTF8)
857 sv_utf8_downgrade (string, 0); 1085 sv_utf8_downgrade (string, 0);
858 else 1086 else
859 sv_utf8_upgrade (string); 1087 sv_utf8_upgrade (string);
860 1088
861 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1089 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
862 1090
863 dec_t dec;
864 dec.flags = flags; 1091 dec.flags = flags;
865 dec.cur = SvPVX (string); 1092 dec.cur = SvPVX (string);
866 dec.end = SvEND (string); 1093 dec.end = SvEND (string);
867 dec.err = 0; 1094 dec.err = 0;
1095 dec.depth = 0;
1096 dec.maxdepth = DEC_DEPTH (dec.flags);
868 1097
1098 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
869 sv = decode_sv (&dec); 1099 sv = decode_sv (&dec);
870 1100
1101 if (!(offset_return || !sv))
1102 {
1103 // check for trailing garbage
1104 decode_ws (&dec);
1105
1106 if (*dec.cur)
1107 {
1108 dec.err = "garbage after JSON object";
1109 SvREFCNT_dec (sv);
1110 sv = 0;
1111 }
1112 }
1113
1114 if (offset_return || !sv)
1115 {
1116 offset = dec.flags & F_UTF8
1117 ? dec.cur - SvPVX (string)
1118 : utf8_distance (dec.cur, SvPVX (string));
1119
1120 if (offset_return)
1121 *offset_return = offset;
1122 }
1123
871 if (!sv) 1124 if (!sv)
872 { 1125 {
873 IV offset = utf8_distance (dec.cur, SvPVX (string));
874 SV *uni = sv_newmortal (); 1126 SV *uni = sv_newmortal ();
1127
875 // horrible hack to silence warning inside pv_uni_display 1128 // horrible hack to silence warning inside pv_uni_display
876 COP cop; 1129 COP cop = *PL_curcop;
877 memset (&cop, 0, sizeof (cop));
878 cop.cop_warnings = pWARN_NONE; 1130 cop.cop_warnings = pWARN_NONE;
1131 ENTER;
879 SAVEVPTR (PL_curcop); 1132 SAVEVPTR (PL_curcop);
880 PL_curcop = &cop; 1133 PL_curcop = &cop;
881
882 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1134 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1135 LEAVE;
1136
883 croak ("%s, at character offset %d (%s)", 1137 croak ("%s, at character offset %d [\"%s\"]",
884 dec.err, 1138 dec.err,
885 (int)offset, 1139 (int)offset,
886 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1140 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
887 } 1141 }
888 1142
889 sv = sv_2mortal (sv); 1143 sv = sv_2mortal (sv);
890 1144
891 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1145 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
892 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1146 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
893 1147
894 return sv; 1148 return sv;
895} 1149}
896 1150
1151/////////////////////////////////////////////////////////////////////////////
1152// XS interface functions
1153
897MODULE = JSON::XS PACKAGE = JSON::XS 1154MODULE = JSON::XS PACKAGE = JSON::XS
898 1155
899BOOT: 1156BOOT:
900{ 1157{
901 int i; 1158 int i;
902 1159
903 memset (decode_hexdigit, 0xff, 256);
904 for (i = 10; i--; ) 1160 for (i = 0; i < 256; ++i)
905 decode_hexdigit ['0' + i] = i; 1161 decode_hexdigit [i] =
906 1162 i >= '0' && i <= '9' ? i - '0'
907 for (i = 7; i--; ) 1163 : i >= 'a' && i <= 'f' ? i - 'a' + 10
908 { 1164 : i >= 'A' && i <= 'F' ? i - 'A' + 10
909 decode_hexdigit ['a' + i] = 10 + i; 1165 : -1;
910 decode_hexdigit ['A' + i] = 10 + i;
911 }
912 1166
913 json_stash = gv_stashpv ("JSON::XS", 1); 1167 json_stash = gv_stashpv ("JSON::XS", 1);
914} 1168}
915 1169
916PROTOTYPES: DISABLE 1170PROTOTYPES: DISABLE
… …
922 RETVAL 1176 RETVAL
923 1177
924SV *ascii (SV *self, int enable = 1) 1178SV *ascii (SV *self, int enable = 1)
925 ALIAS: 1179 ALIAS:
926 ascii = F_ASCII 1180 ascii = F_ASCII
1181 latin1 = F_LATIN1
927 utf8 = F_UTF8 1182 utf8 = F_UTF8
928 indent = F_INDENT 1183 indent = F_INDENT
929 canonical = F_CANONICAL 1184 canonical = F_CANONICAL
930 space_before = F_SPACE_BEFORE 1185 space_before = F_SPACE_BEFORE
931 space_after = F_SPACE_AFTER 1186 space_after = F_SPACE_AFTER
932 json_rpc = F_JSON_RPC
933 pretty = F_PRETTY 1187 pretty = F_PRETTY
934 allow_nonref = F_ALLOW_NONREF 1188 allow_nonref = F_ALLOW_NONREF
935 shrink = F_SHRINK 1189 shrink = F_SHRINK
936 CODE: 1190 CODE:
937{ 1191{
… …
944 RETVAL = newSVsv (self); 1198 RETVAL = newSVsv (self);
945} 1199}
946 OUTPUT: 1200 OUTPUT:
947 RETVAL 1201 RETVAL
948 1202
1203SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1204 CODE:
1205{
1206 UV *uv = SvJSON (self);
1207 UV log2 = 0;
1208
1209 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1210
1211 while ((1UL << log2) < max_depth)
1212 ++log2;
1213
1214 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1215
1216 RETVAL = newSVsv (self);
1217}
1218 OUTPUT:
1219 RETVAL
1220
949void encode (SV *self, SV *scalar) 1221void encode (SV *self, SV *scalar)
950 PPCODE: 1222 PPCODE:
951 XPUSHs (encode_json (scalar, *SvJSON (self))); 1223 XPUSHs (encode_json (scalar, *SvJSON (self)));
952 1224
953void decode (SV *self, SV *jsonstr) 1225void decode (SV *self, SV *jsonstr)
954 PPCODE: 1226 PPCODE:
955 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1227 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1228
1229void decode_prefix (SV *self, SV *jsonstr)
1230 PPCODE:
1231{
1232 UV offset;
1233 EXTEND (SP, 2);
1234 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1235 PUSHs (sv_2mortal (newSVuv (offset)));
1236}
956 1237
957PROTOTYPES: ENABLE 1238PROTOTYPES: ENABLE
958 1239
959void to_json (SV *scalar) 1240void to_json (SV *scalar)
1241 ALIAS:
1242 objToJson = 0
960 PPCODE: 1243 PPCODE:
961 XPUSHs (encode_json (scalar, F_UTF8)); 1244 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
962 1245
963void from_json (SV *jsonstr) 1246void from_json (SV *jsonstr)
1247 ALIAS:
1248 jsonToObj = 0
964 PPCODE: 1249 PPCODE:
965 XPUSHs (decode_json (jsonstr, F_UTF8)); 1250 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
966 1251

Diff Legend

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