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 1.38 by root, Mon Jun 11 03:18:07 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
9#define F_ASCII 0x00000001 14#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 16#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 17#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 18#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 19#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 20#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 21#define F_ALLOW_NONREF 0x00000100UL
17#define F_SHRINK 0x00000100 22#define F_SHRINK 0x00000200UL
23#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27
25
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
18 30
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 32#define F_DEFAULT (9UL << S_MAXDEPTH)
21 33
22#define INIT_SIZE 32 // initial scalar size to be allocated 34#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level
36
37#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
23 38
24#define SB do { 39#define SB do {
25#define SE } while (0) 40#define SE } while (0)
26 41
42#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value))
44# define inline inline
45#else
46# define expect(expr,value) (expr)
47# define inline static
48#endif
49
50#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1)
52
27static HV *json_stash; 53static HV *json_stash; // JSON::XS::
54
55/////////////////////////////////////////////////////////////////////////////
56// utility functions
57
58static UV *
59SvJSON (SV *sv)
60{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62 croak ("object is not of type JSON::XS");
63
64 return &SvUVX (SvRV (sv));
65}
66
67static void
68shrink (SV *sv)
69{
70 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1)
72 {
73#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew)
76 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif
78 }
79}
80
81// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints
86inline UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
92 {
93 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 }
96 else
97 {
98 *clen = (STRLEN)-1;
99 return (UV)-1;
100 }
101}
102
103/////////////////////////////////////////////////////////////////////////////
104// encoder
28 105
29// structure used for encoding JSON 106// structure used for encoding JSON
30typedef struct 107typedef struct
31{ 108{
32 char *cur; 109 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 110 char *end; // SvEND (sv)
35 SV *sv; 111 SV *sv; // result scalar
36 UV flags; 112 U32 flags; // F_*
37 int max_recurse; 113 U32 indent; // indentation level
38 int indent; 114 U32 maxdepth; // max. indentation/recursion level
39} enc_t; 115} enc_t;
40 116
41// structure used for decoding JSON 117inline 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) 118need (enc_t *enc, STRLEN len)
63{ 119{
64 if (enc->cur + len >= enc->end) 120 if (expect_false (enc->cur + len >= enc->end))
65 { 121 {
66 STRLEN cur = enc->cur - SvPVX (enc->sv); 122 STRLEN cur = enc->cur - SvPVX (enc->sv);
67 SvGROW (enc->sv, cur + len + 1); 123 SvGROW (enc->sv, cur + len + 1);
68 enc->cur = SvPVX (enc->sv) + cur; 124 enc->cur = SvPVX (enc->sv) + cur;
69 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
70 } 126 }
71} 127}
72 128
73static void 129inline void
74encode_ch (enc_t *enc, char ch) 130encode_ch (enc_t *enc, char ch)
75{ 131{
76 need (enc, 1); 132 need (enc, 1);
77 *enc->cur++ = ch; 133 *enc->cur++ = ch;
78} 134}
… …
86 142
87 while (str < end) 143 while (str < end)
88 { 144 {
89 unsigned char ch = *(unsigned char *)str; 145 unsigned char ch = *(unsigned char *)str;
90 146
91 if (ch >= 0x20 && ch < 0x80) // most common case 147 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
92 { 148 {
93 if (ch == '"') // but with slow exceptions 149 if (expect_false (ch == '"')) // but with slow exceptions
94 { 150 {
95 need (enc, len += 1); 151 need (enc, len += 1);
96 *enc->cur++ = '\\'; 152 *enc->cur++ = '\\';
97 *enc->cur++ = '"'; 153 *enc->cur++ = '"';
98 } 154 }
99 else if (ch == '\\') 155 else if (expect_false (ch == '\\'))
100 { 156 {
101 need (enc, len += 1); 157 need (enc, len += 1);
102 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
103 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
104 } 160 }
… …
122 STRLEN clen; 178 STRLEN clen;
123 UV uch; 179 UV uch;
124 180
125 if (is_utf8) 181 if (is_utf8)
126 { 182 {
127 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen);
128 if (clen == (STRLEN)-1) 185 if (clen == (STRLEN)-1)
129 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
130 } 187 }
131 else 188 else
132 { 189 {
133 uch = ch; 190 uch = ch;
134 clen = 1; 191 clen = 1;
135 } 192 }
136 193
137 if (uch < 0x80 || enc->flags & F_ASCII) 194 if (uch > 0x10FFFFUL)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
138 { 198 {
139 if (uch > 0xFFFFUL) 199 if (uch > 0xFFFFUL)
140 { 200 {
141 need (enc, len += 11); 201 need (enc, len += 11);
142 sprintf (enc->cur, "\\u%04x\\u%04x", 202 sprintf (enc->cur, "\\u%04x\\u%04x",
143 (uch - 0x10000) / 0x400 + 0xD800, 203 (int)((uch - 0x10000) / 0x400 + 0xD800),
144 (uch - 0x10000) % 0x400 + 0xDC00); 204 (int)((uch - 0x10000) % 0x400 + 0xDC00));
145 enc->cur += 12; 205 enc->cur += 12;
146 } 206 }
147 else 207 else
148 { 208 {
149 static char hexdigit [16] = "0123456789abcdef"; 209 static char hexdigit [16] = "0123456789abcdef";
… …
156 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 216 *enc->cur++ = hexdigit [(uch >> 0) & 15];
157 } 217 }
158 218
159 str += clen; 219 str += clen;
160 } 220 }
221 else if (enc->flags & F_LATIN1)
222 {
223 *enc->cur++ = uch;
224 str += clen;
225 }
161 else if (is_utf8) 226 else if (is_utf8)
162 { 227 {
163 need (enc, len += clen); 228 need (enc, len += clen);
164 do 229 do
165 { 230 {
… …
167 } 232 }
168 while (--clen); 233 while (--clen);
169 } 234 }
170 else 235 else
171 { 236 {
172 need (enc, 10); // never more than 11 bytes needed 237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str; 239 ++str;
175 } 240 }
176 } 241 }
177 } 242 }
… …
179 244
180 --len; 245 --len;
181 } 246 }
182} 247}
183 248
184#define INDENT SB \ 249inline void
250encode_indent (enc_t *enc)
251{
185 if (enc->flags & F_INDENT) \ 252 if (enc->flags & F_INDENT)
186 { \ 253 {
187 int i_; \ 254 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
193 255
194#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 256 need (enc, spaces);
195#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 257 memset (enc->cur, ' ', spaces);
196#define COMMA SB \ 258 enc->cur += spaces;
259 }
260}
261
262inline void
263encode_space (enc_t *enc)
264{
265 need (enc, 1);
197 encode_ch (enc, ','); \ 266 encode_ch (enc, ' ');
267}
268
269inline void
270encode_nl (enc_t *enc)
271{
198 if (enc->flags & F_INDENT) \ 272 if (enc->flags & F_INDENT)
199 NL; \ 273 {
274 need (enc, 1);
275 encode_ch (enc, '\n');
276 }
277}
278
279inline void
280encode_comma (enc_t *enc)
281{
282 encode_ch (enc, ',');
283
284 if (enc->flags & F_INDENT)
285 encode_nl (enc);
200 else if (enc->flags & F_SPACE_AFTER) \ 286 else if (enc->flags & F_SPACE_AFTER)
201 SPACE; \ 287 encode_space (enc);
202 SE 288}
203 289
204static void encode_sv (enc_t *enc, SV *sv); 290static void encode_sv (enc_t *enc, SV *sv);
205 291
206static void 292static void
207encode_av (enc_t *enc, AV *av) 293encode_av (enc_t *enc, AV *av)
208{ 294{
209 int i, len = av_len (av); 295 int i, len = av_len (av);
210 296
211 encode_ch (enc, '['); NL; 297 if (enc->indent >= enc->maxdepth)
298 croak ("data structure too deep (hit recursion limit)");
299
300 encode_ch (enc, '['); encode_nl (enc);
212 ++enc->indent; 301 ++enc->indent;
213 302
214 for (i = 0; i <= len; ++i) 303 for (i = 0; i <= len; ++i)
215 { 304 {
216 INDENT; 305 encode_indent (enc);
217 encode_sv (enc, *av_fetch (av, i, 0)); 306 encode_sv (enc, *av_fetch (av, i, 0));
218 307
219 if (i < len) 308 if (i < len)
220 COMMA; 309 encode_comma (enc);
221 } 310 }
222 311
223 NL; 312 encode_nl (enc);
224 313
225 --enc->indent; 314 --enc->indent;
226 INDENT; encode_ch (enc, ']'); 315 encode_indent (enc); encode_ch (enc, ']');
227} 316}
228 317
229static void 318static void
230encode_he (enc_t *enc, HE *he) 319encode_he (enc_t *enc, HE *he)
231{ 320{
… …
245 else 334 else
246 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
247 336
248 encode_ch (enc, '"'); 337 encode_ch (enc, '"');
249 338
250 if (enc->flags & F_SPACE_BEFORE) SPACE; 339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
251 encode_ch (enc, ':'); 340 encode_ch (enc, ':');
252 if (enc->flags & F_SPACE_AFTER ) SPACE; 341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
253 encode_sv (enc, HeVAL (he)); 342 encode_sv (enc, HeVAL (he));
254} 343}
255 344
256// compare hash entries, used when all keys are bytestrings 345// compare hash entries, used when all keys are bytestrings
257static int 346static int
… …
263 HE *b = *(HE **)b_; 352 HE *b = *(HE **)b_;
264 353
265 STRLEN la = HeKLEN (a); 354 STRLEN la = HeKLEN (a);
266 STRLEN lb = HeKLEN (b); 355 STRLEN lb = HeKLEN (b);
267 356
268 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
269 cmp = la < lb ? -1 : la == lb ? 0 : 1; 358 cmp = la - lb;
270 359
271 return cmp; 360 return cmp;
272} 361}
273 362
274// compare hash entries, used when some keys are sv's or utf-x 363// compare hash entries, used when some keys are sv's or utf-x
… …
281static void 370static void
282encode_hv (enc_t *enc, HV *hv) 371encode_hv (enc_t *enc, HV *hv)
283{ 372{
284 int count, i; 373 int count, i;
285 374
375 if (enc->indent >= enc->maxdepth)
376 croak ("data structure too deep (hit recursion limit)");
377
286 encode_ch (enc, '{'); NL; ++enc->indent; 378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
287 379
288 if ((count = hv_iterinit (hv))) 380 if ((count = hv_iterinit (hv)))
289 { 381 {
290 // for canonical output we have to sort by keys first 382 // for canonical output we have to sort by keys first
291 // actually, this is mostly due to the stupid so-called 383 // actually, this is mostly due to the stupid so-called
292 // security workaround added somewhere in 5.8.x. 384 // security workaround added somewhere in 5.8.x.
293 // that randomises hash orderings 385 // that randomises hash orderings
294 if (enc->flags & F_CANONICAL) 386 if (enc->flags & F_CANONICAL)
295 { 387 {
296 HE *he, *hes [count];
297 int fast = 1; 388 int fast = 1;
389 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE));
392#else
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
394#endif
298 395
299 i = 0; 396 i = 0;
300 while ((he = hv_iternext (hv))) 397 while ((he = hv_iternext (hv)))
301 { 398 {
302 hes [i++] = he; 399 hes [i++] = he;
… …
308 405
309 if (fast) 406 if (fast)
310 qsort (hes, count, sizeof (HE *), he_cmp_fast); 407 qsort (hes, count, sizeof (HE *), he_cmp_fast);
311 else 408 else
312 { 409 {
313 // hack to disable "use bytes" 410 // hack to forcefully disable "use bytes"
314 COP *oldcop = PL_curcop, cop; 411 COP cop = *PL_curcop;
315 cop.op_private = 0; 412 cop.op_private = 0;
413
414 ENTER;
415 SAVETMPS;
416
417 SAVEVPTR (PL_curcop);
316 PL_curcop = &cop; 418 PL_curcop = &cop;
317 419
318 SAVETMPS;
319 qsort (hes, count, sizeof (HE *), he_cmp_slow); 420 qsort (hes, count, sizeof (HE *), he_cmp_slow);
421
320 FREETMPS; 422 FREETMPS;
321 423 LEAVE;
322 PL_curcop = oldcop;
323 } 424 }
324 425
325 for (i = 0; i < count; ++i) 426 for (i = 0; i < count; ++i)
326 { 427 {
327 INDENT; 428 encode_indent (enc);
328 encode_he (enc, hes [i]); 429 encode_he (enc, hes [i]);
329 430
330 if (i < count - 1) 431 if (i < count - 1)
331 COMMA; 432 encode_comma (enc);
332 } 433 }
333 434
334 NL; 435 encode_nl (enc);
335 } 436 }
336 else 437 else
337 { 438 {
338 SV *sv;
339 HE *he = hv_iternext (hv); 439 HE *he = hv_iternext (hv);
340 440
341 for (;;) 441 for (;;)
342 { 442 {
343 INDENT; 443 encode_indent (enc);
344 encode_he (enc, he); 444 encode_he (enc, he);
345 445
346 if (!(he = hv_iternext (hv))) 446 if (!(he = hv_iternext (hv)))
347 break; 447 break;
348 448
349 COMMA; 449 encode_comma (enc);
350 } 450 }
351 451
352 NL; 452 encode_nl (enc);
353 } 453 }
354 } 454 }
355 455
356 --enc->indent; INDENT; encode_ch (enc, '}'); 456 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
457}
458
459// encode objects, arrays and special \0=false and \1=true values.
460static void
461encode_rv (enc_t *enc, SV *sv)
462{
463 svtype svt;
464
465 SvGETMAGIC (sv);
466 svt = SvTYPE (sv);
467
468 if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV)
473 {
474 if (SvNIOK (sv) && SvIV (sv) == 0)
475 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1)
477 encode_str (enc, "true", 4, 0);
478 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 }
482 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
357} 485}
358 486
359static void 487static void
360encode_sv (enc_t *enc, SV *sv) 488encode_sv (enc_t *enc, SV *sv)
361{ 489{
… …
375 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
376 enc->cur += strlen (enc->cur); 504 enc->cur += strlen (enc->cur);
377 } 505 }
378 else if (SvIOKp (sv)) 506 else if (SvIOKp (sv))
379 { 507 {
380 need (enc, 64); 508 // we assume we can always read an IV as a UV
509 if (SvUV (sv) & ~(UV)0x7fff)
510 {
511 need (enc, sizeof (UV) * 3);
381 enc->cur += 512 enc->cur +=
382 SvIsUV(sv) 513 SvIsUV(sv)
383 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 514 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
384 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 515 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
516 }
517 else
518 {
519 // optimise the "small number case"
520 // code will likely be branchless and use only a single multiplication
521 I32 i = SvIV (sv);
522 U32 u;
523
524 need (enc, 6);
525
526 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
527 u = i < 0 ? -i : i;
528
529 // convert to 4.28 fixed-point representation
530 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531
532 char digit, nz = 0;
533
534 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
535 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
536 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
537 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
538 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1;
539 }
385 } 540 }
386 else if (SvROK (sv)) 541 else if (SvROK (sv))
387 { 542 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)) 543 else if (!SvOK (sv))
403 encode_str (enc, "null", 4, 0); 544 encode_str (enc, "null", 4, 0);
404 else 545 else
405 croak ("encountered perl type that JSON cannot handle"); 546 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
547 SvPV_nolen (sv), SvFLAGS (sv));
406} 548}
407 549
408static SV * 550static SV *
409encode_json (SV *scalar, UV flags) 551encode_json (SV *scalar, U32 flags)
410{ 552{
553 enc_t enc;
554
411 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
412 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
413 557
414 enc_t enc;
415 enc.flags = flags; 558 enc.flags = flags;
416 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
417 enc.cur = SvPVX (enc.sv); 560 enc.cur = SvPVX (enc.sv);
418 enc.end = SvEND (enc.sv); 561 enc.end = SvEND (enc.sv);
419 enc.max_recurse = 0;
420 enc.indent = 0; 562 enc.indent = 0;
563 enc.maxdepth = DEC_DEPTH (flags);
421 564
422 SvPOK_only (enc.sv); 565 SvPOK_only (enc.sv);
423 encode_sv (&enc, scalar); 566 encode_sv (&enc, scalar);
424 567
568 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570
425 if (!(flags & (F_ASCII | F_UTF8))) 571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
426 SvUTF8_on (enc.sv); 572 SvUTF8_on (enc.sv);
427 573
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429
430#ifdef SvPV_shrink_to_cur
431 if (enc.flags & F_SHRINK) 574 if (enc.flags & F_SHRINK)
432 SvPV_shrink_to_cur (enc.sv); 575 shrink (enc.sv);
433#endif 576
434 return enc.sv; 577 return enc.sv;
435} 578}
436 579
437///////////////////////////////////////////////////////////////////////////// 580/////////////////////////////////////////////////////////////////////////////
581// decoder
438 582
439#define WS \ 583// structure used for decoding JSON
584typedef struct
585{
586 char *cur; // current parser pointer
587 char *end; // end of input string
588 const char *err; // parse error, if != 0
589 U32 flags; // F_*
590 U32 depth; // recursion depth
591 U32 maxdepth; // recursion depth limit
592} dec_t;
593
594inline void
595decode_ws (dec_t *dec)
596{
440 for (;;) \ 597 for (;;)
441 { \ 598 {
442 char ch = *dec->cur; \ 599 char ch = *dec->cur;
600
443 if (ch > 0x20 \ 601 if (ch > 0x20
444 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 602 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
445 break; \ 603 break;
604
446 ++dec->cur; \ 605 ++dec->cur;
447 } 606 }
607}
448 608
449#define ERR(reason) SB dec->err = reason; goto fail; SE 609#define ERR(reason) SB dec->err = reason; goto fail; SE
610
450#define EXPECT_CH(ch) SB \ 611#define EXPECT_CH(ch) SB \
451 if (*dec->cur != ch) \ 612 if (*dec->cur != ch) \
452 ERR (# ch " expected"); \ 613 ERR (# ch " expected"); \
453 ++dec->cur; \ 614 ++dec->cur; \
454 SE 615 SE
455 616
617#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
618#define DEC_DEC_DEPTH --dec->depth
619
456static SV *decode_sv (dec_t *dec); 620static SV *decode_sv (dec_t *dec);
457 621
458static signed char decode_hexdigit[256]; 622static signed char decode_hexdigit[256];
459 623
460static UV 624static UV
461decode_4hex (dec_t *dec) 625decode_4hex (dec_t *dec)
462{ 626{
463 signed char d1, d2, d3, d4; 627 signed char d1, d2, d3, d4;
628 unsigned char *cur = (unsigned char *)dec->cur;
464 629
465 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");
466 if (d1 < 0) ERR ("four hexadecimal digits expected"); 631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected");
467 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected");
468 if (d2 < 0) ERR ("four hexadecimal digits expected"); 633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("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");
473 634
474 dec->cur += 4; 635 dec->cur += 4;
475 636
476 return ((UV)d1) << 12 637 return ((UV)d1) << 12
477 | ((UV)d2) << 8 638 | ((UV)d2) << 8
… …
480 641
481fail: 642fail:
482 return (UV)-1; 643 return (UV)-1;
483} 644}
484 645
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 * 646static SV *
501decode_str (dec_t *dec) 647decode_str (dec_t *dec)
502{ 648{
503 SV *sv = NEWSV (0,2); 649 SV *sv = 0;
504 int utf8 = 0; 650 int utf8 = 0;
505 char *cur = SvPVX (sv); 651 char *dec_cur = dec->cur;
506 char *end = SvEND (sv);
507 652
508 for (;;) 653 do
509 { 654 {
510 unsigned char ch = *(unsigned char *)dec->cur; 655 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
656 char *cur = buf;
511 657
512 if (ch == '"') 658 do
513 break;
514 else if (ch == '\\')
515 { 659 {
516 switch (*++dec->cur) 660 unsigned char ch = *(unsigned char *)dec_cur++;
661
662 if (expect_false (ch == '"'))
517 { 663 {
518 case '\\': 664 --dec_cur;
519 case '/': 665 break;
520 case '"': APPEND_CH (*dec->cur++); break; 666 }
521 667 else if (expect_false (ch == '\\'))
522 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 668 {
523 case 't': APPEND_CH ('\011'); ++dec->cur; break; 669 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 { 670 {
530 UV lo, hi; 671 case '\\':
531 ++dec->cur; 672 case '/':
673 case '"': *cur++ = *dec_cur++; break;
532 674
533 hi = decode_4hex (dec); 675 case 'b': ++dec_cur; *cur++ = '\010'; break;
534 if (hi == (UV)-1) 676 case 't': ++dec_cur; *cur++ = '\011'; break;
535 goto fail; 677 case 'n': ++dec_cur; *cur++ = '\012'; break;
678 case 'f': ++dec_cur; *cur++ = '\014'; break;
679 case 'r': ++dec_cur; *cur++ = '\015'; break;
536 680
537 // possibly a surrogate pair 681 case 'u':
538 if (hi >= 0xd800 && hi < 0xdc00)
539 { 682 {
540 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 683 UV lo, hi;
541 ERR ("missing low surrogate character in surrogate pair"); 684 ++dec_cur;
542 685
543 dec->cur += 2; 686 dec->cur = dec_cur;
544
545 lo = decode_4hex (dec); 687 hi = decode_4hex (dec);
688 dec_cur = dec->cur;
546 if (lo == (UV)-1) 689 if (hi == (UV)-1)
547 goto fail; 690 goto fail;
548 691
692 // possibly a surrogate pair
693 if (hi >= 0xd800)
694 if (hi < 0xdc00)
695 {
696 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
697 ERR ("missing low surrogate character in surrogate pair");
698
699 dec_cur += 2;
700
701 dec->cur = dec_cur;
702 lo = decode_4hex (dec);
703 dec_cur = dec->cur;
704 if (lo == (UV)-1)
705 goto fail;
706
549 if (lo < 0xdc00 || lo >= 0xe000) 707 if (lo < 0xdc00 || lo >= 0xe000)
550 ERR ("surrogate pair expected"); 708 ERR ("surrogate pair expected");
551 709
552 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 710 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
711 }
712 else if (hi < 0xe000)
713 ERR ("missing high surrogate character in surrogate pair");
714
715 if (hi >= 0x80)
716 {
717 utf8 = 1;
718
719 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
720 }
721 else
722 *cur++ = hi;
553 } 723 }
554 else if (hi >= 0xdc00 && hi < 0xe000)
555 ERR ("missing high surrogate character in surrogate pair");
556
557 if (hi >= 0x80)
558 { 724 break;
559 utf8 = 1;
560 725
561 APPEND_GROW (4); // at most 4 bytes for 21 bits
562 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
563 }
564 else 726 default:
565 APPEND_CH (hi); 727 --dec_cur;
728 ERR ("illegal backslash escape sequence in string");
566 } 729 }
567 break; 730 }
731 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
732 *cur++ = ch;
733 else if (ch >= 0x80)
734 {
735 STRLEN clen;
736 UV uch;
568 737
569 default:
570 --dec->cur; 738 --dec_cur;
571 ERR ("illegal backslash escape sequence in string"); 739
740 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
741 if (clen == (STRLEN)-1)
742 ERR ("malformed UTF-8 character in JSON string");
743
744 do
745 *cur++ = *dec_cur++;
746 while (--clen);
747
748 utf8 = 1;
749 }
750 else
751 {
752 --dec_cur;
753
754 if (!ch)
755 ERR ("unexpected end of string while parsing JSON string");
756 else
757 ERR ("invalid character encountered while parsing JSON string");
572 } 758 }
573 } 759 }
574 else if (ch >= 0x20 && ch <= 0x7f) 760 while (cur < buf + SHORT_STRING_LEN);
575 APPEND_CH (*dec->cur++); 761
576 else if (ch >= 0x80)
577 { 762 {
578 STRLEN clen; 763 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");
582 764
583 APPEND_GROW (clen); 765 if (sv)
584 do
585 { 766 {
586 *cur++ = *dec->cur++; 767 SvGROW (sv, SvCUR (sv) + len + 1);
768 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
769 SvCUR_set (sv, SvCUR (sv) + len);
587 } 770 }
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 771 else
595 ERR ("invalid character encountered"); 772 sv = newSVpvn (buf, len);
596 } 773 }
774 }
775 while (*dec_cur != '"');
597 776
598 ++dec->cur; 777 ++dec_cur;
599 778
600 SvCUR_set (sv, cur - SvPVX (sv)); 779 if (sv)
601 780 {
602 SvPOK_only (sv); 781 SvPOK_only (sv);
603 *SvEND (sv) = 0; 782 *SvEND (sv) = 0;
604 783
605 if (utf8) 784 if (utf8)
606 SvUTF8_on (sv); 785 SvUTF8_on (sv);
786 }
787 else
788 sv = newSVpvn ("", 0);
607 789
608#ifdef SvPV_shrink_to_cur 790 dec->cur = dec_cur;
609 if (dec->flags & F_SHRINK)
610 SvPV_shrink_to_cur (sv);
611#endif
612
613 return sv; 791 return sv;
614 792
615fail: 793fail:
616 SvREFCNT_dec (sv); 794 dec->cur = dec_cur;
617 return 0; 795 return 0;
618} 796}
619 797
620static SV * 798static SV *
621decode_num (dec_t *dec) 799decode_num (dec_t *dec)
… …
679 is_nv = 1; 857 is_nv = 1;
680 } 858 }
681 859
682 if (!is_nv) 860 if (!is_nv)
683 { 861 {
684 UV uv; 862 // 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); 863 if (*start == '-')
686 if (numtype & IS_NUMBER_IN_UV) 864 switch (dec->cur - start)
687 if (numtype & IS_NUMBER_NEG)
688 { 865 {
689 if (uv < (UV)IV_MIN) 866 case 2: return newSViv (-( start [1] - '0' ));
690 return newSViv (-(IV)uv); 867 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
868 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
869 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
691 } 870 }
871 else
872 switch (dec->cur - start)
873 {
874 case 1: return newSViv ( start [0] - '0' );
875 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
876 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
877 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
878 }
879
880 {
881 UV uv;
882 int numtype = grok_number (start, dec->cur - start, &uv);
883 if (numtype & IS_NUMBER_IN_UV)
884 if (numtype & IS_NUMBER_NEG)
885 {
886 if (uv < (UV)IV_MIN)
887 return newSViv (-(IV)uv);
888 }
692 else 889 else
693 return newSVuv (uv); 890 return newSVuv (uv);
891 }
694 } 892 }
695 893
696 return newSVnv (Atof (start)); 894 return newSVnv (Atof (start));
697 895
698fail: 896fail:
… …
702static SV * 900static SV *
703decode_av (dec_t *dec) 901decode_av (dec_t *dec)
704{ 902{
705 AV *av = newAV (); 903 AV *av = newAV ();
706 904
707 WS; 905 DEC_INC_DEPTH;
906 decode_ws (dec);
907
708 if (*dec->cur == ']') 908 if (*dec->cur == ']')
709 ++dec->cur; 909 ++dec->cur;
710 else 910 else
711 for (;;) 911 for (;;)
712 { 912 {
… …
716 if (!value) 916 if (!value)
717 goto fail; 917 goto fail;
718 918
719 av_push (av, value); 919 av_push (av, value);
720 920
721 WS; 921 decode_ws (dec);
722 922
723 if (*dec->cur == ']') 923 if (*dec->cur == ']')
724 { 924 {
725 ++dec->cur; 925 ++dec->cur;
726 break; 926 break;
… …
730 ERR (", or ] expected while parsing array"); 930 ERR (", or ] expected while parsing array");
731 931
732 ++dec->cur; 932 ++dec->cur;
733 } 933 }
734 934
935 DEC_DEC_DEPTH;
735 return newRV_noinc ((SV *)av); 936 return newRV_noinc ((SV *)av);
736 937
737fail: 938fail:
738 SvREFCNT_dec (av); 939 SvREFCNT_dec (av);
940 DEC_DEC_DEPTH;
739 return 0; 941 return 0;
740} 942}
741 943
742static SV * 944static SV *
743decode_hv (dec_t *dec) 945decode_hv (dec_t *dec)
744{ 946{
745 HV *hv = newHV (); 947 HV *hv = newHV ();
746 948
747 WS; 949 DEC_INC_DEPTH;
950 decode_ws (dec);
951
748 if (*dec->cur == '}') 952 if (*dec->cur == '}')
749 ++dec->cur; 953 ++dec->cur;
750 else 954 else
751 for (;;) 955 for (;;)
752 { 956 {
753 SV *key, *value; 957 SV *key, *value;
754 958
755 WS; EXPECT_CH ('"'); 959 decode_ws (dec); EXPECT_CH ('"');
756 960
757 key = decode_str (dec); 961 key = decode_str (dec);
758 if (!key) 962 if (!key)
759 goto fail; 963 goto fail;
760 964
761 WS; EXPECT_CH (':'); 965 decode_ws (dec); EXPECT_CH (':');
762 966
763 value = decode_sv (dec); 967 value = decode_sv (dec);
764 if (!value) 968 if (!value)
765 { 969 {
766 SvREFCNT_dec (key); 970 SvREFCNT_dec (key);
767 goto fail; 971 goto fail;
768 } 972 }
769 973
770 //TODO: optimise
771 hv_store_ent (hv, key, value, 0); 974 hv_store_ent (hv, key, value, 0);
975 SvREFCNT_dec (key);
772 976
773 WS; 977 decode_ws (dec);
774 978
775 if (*dec->cur == '}') 979 if (*dec->cur == '}')
776 { 980 {
777 ++dec->cur; 981 ++dec->cur;
778 break; 982 break;
… …
782 ERR (", or } expected while parsing object/hash"); 986 ERR (", or } expected while parsing object/hash");
783 987
784 ++dec->cur; 988 ++dec->cur;
785 } 989 }
786 990
991 DEC_DEC_DEPTH;
787 return newRV_noinc ((SV *)hv); 992 return newRV_noinc ((SV *)hv);
788 993
789fail: 994fail:
790 SvREFCNT_dec (hv); 995 SvREFCNT_dec (hv);
996 DEC_DEC_DEPTH;
791 return 0; 997 return 0;
792} 998}
793 999
794static SV * 1000static SV *
795decode_sv (dec_t *dec) 1001decode_sv (dec_t *dec)
796{ 1002{
797 WS; 1003 decode_ws (dec);
798 switch (*dec->cur) 1004 switch (*dec->cur)
799 { 1005 {
800 case '"': ++dec->cur; return decode_str (dec); 1006 case '"': ++dec->cur; return decode_str (dec);
801 case '[': ++dec->cur; return decode_av (dec); 1007 case '[': ++dec->cur; return decode_av (dec);
802 case '{': ++dec->cur; return decode_hv (dec); 1008 case '{': ++dec->cur; return decode_hv (dec);
… …
838 ERR ("'null' expected"); 1044 ERR ("'null' expected");
839 1045
840 break; 1046 break;
841 1047
842 default: 1048 default:
843 ERR ("malformed json string"); 1049 ERR ("malformed JSON string, neither array, object, number, string or atom");
844 break; 1050 break;
845 } 1051 }
846 1052
847fail: 1053fail:
848 return 0; 1054 return 0;
849} 1055}
850 1056
851static SV * 1057static SV *
852decode_json (SV *string, UV flags) 1058decode_json (SV *string, U32 flags, UV *offset_return)
853{ 1059{
1060 dec_t dec;
1061 UV offset;
854 SV *sv; 1062 SV *sv;
1063
1064 SvGETMAGIC (string);
1065 SvUPGRADE (string, SVt_PV);
855 1066
856 if (flags & F_UTF8) 1067 if (flags & F_UTF8)
857 sv_utf8_downgrade (string, 0); 1068 sv_utf8_downgrade (string, 0);
858 else 1069 else
859 sv_utf8_upgrade (string); 1070 sv_utf8_upgrade (string);
860 1071
861 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1072 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
862 1073
863 dec_t dec;
864 dec.flags = flags; 1074 dec.flags = flags;
865 dec.cur = SvPVX (string); 1075 dec.cur = SvPVX (string);
866 dec.end = SvEND (string); 1076 dec.end = SvEND (string);
867 dec.err = 0; 1077 dec.err = 0;
1078 dec.depth = 0;
1079 dec.maxdepth = DEC_DEPTH (dec.flags);
868 1080
1081 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
869 sv = decode_sv (&dec); 1082 sv = decode_sv (&dec);
870 1083
1084 if (!(offset_return || !sv))
1085 {
1086 // check for trailing garbage
1087 decode_ws (&dec);
1088
1089 if (*dec.cur)
1090 {
1091 dec.err = "garbage after JSON object";
1092 SvREFCNT_dec (sv);
1093 sv = 0;
1094 }
1095 }
1096
1097 if (offset_return || !sv)
1098 {
1099 offset = dec.flags & F_UTF8
1100 ? dec.cur - SvPVX (string)
1101 : utf8_distance (dec.cur, SvPVX (string));
1102
1103 if (offset_return)
1104 *offset_return = offset;
1105 }
1106
871 if (!sv) 1107 if (!sv)
872 { 1108 {
873 IV offset = utf8_distance (dec.cur, SvPVX (string));
874 SV *uni = sv_newmortal (); 1109 SV *uni = sv_newmortal ();
1110
875 // horrible hack to silence warning inside pv_uni_display 1111 // horrible hack to silence warning inside pv_uni_display
876 COP cop; 1112 COP cop = *PL_curcop;
877 memset (&cop, 0, sizeof (cop));
878 cop.cop_warnings = pWARN_NONE; 1113 cop.cop_warnings = pWARN_NONE;
1114 ENTER;
879 SAVEVPTR (PL_curcop); 1115 SAVEVPTR (PL_curcop);
880 PL_curcop = &cop; 1116 PL_curcop = &cop;
881
882 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1117 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1118 LEAVE;
1119
883 croak ("%s, at character offset %d (%s)", 1120 croak ("%s, at character offset %d [\"%s\"]",
884 dec.err, 1121 dec.err,
885 (int)offset, 1122 (int)offset,
886 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1123 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
887 } 1124 }
888 1125
889 sv = sv_2mortal (sv); 1126 sv = sv_2mortal (sv);
890 1127
891 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1128 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)"); 1129 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
893 1130
894 return sv; 1131 return sv;
895} 1132}
896 1133
1134/////////////////////////////////////////////////////////////////////////////
1135// XS interface functions
1136
897MODULE = JSON::XS PACKAGE = JSON::XS 1137MODULE = JSON::XS PACKAGE = JSON::XS
898 1138
899BOOT: 1139BOOT:
900{ 1140{
901 int i; 1141 int i;
902 1142
903 memset (decode_hexdigit, 0xff, 256); 1143 memset (decode_hexdigit, 0xff, 256);
1144
904 for (i = 10; i--; ) 1145 for (i = 0; i < 256; ++i)
905 decode_hexdigit ['0' + i] = i; 1146 decode_hexdigit [i] =
906 1147 i >= '0' && i <= '9' ? i - '0'
907 for (i = 7; i--; ) 1148 : i >= 'a' && i <= 'f' ? i - 'a' + 10
908 { 1149 : i >= 'A' && i <= 'F' ? i - 'A' + 10
909 decode_hexdigit ['a' + i] = 10 + i; 1150 : -1;
910 decode_hexdigit ['A' + i] = 10 + i;
911 }
912 1151
913 json_stash = gv_stashpv ("JSON::XS", 1); 1152 json_stash = gv_stashpv ("JSON::XS", 1);
914} 1153}
915 1154
916PROTOTYPES: DISABLE 1155PROTOTYPES: DISABLE
… …
922 RETVAL 1161 RETVAL
923 1162
924SV *ascii (SV *self, int enable = 1) 1163SV *ascii (SV *self, int enable = 1)
925 ALIAS: 1164 ALIAS:
926 ascii = F_ASCII 1165 ascii = F_ASCII
1166 latin1 = F_LATIN1
927 utf8 = F_UTF8 1167 utf8 = F_UTF8
928 indent = F_INDENT 1168 indent = F_INDENT
929 canonical = F_CANONICAL 1169 canonical = F_CANONICAL
930 space_before = F_SPACE_BEFORE 1170 space_before = F_SPACE_BEFORE
931 space_after = F_SPACE_AFTER 1171 space_after = F_SPACE_AFTER
932 json_rpc = F_JSON_RPC
933 pretty = F_PRETTY 1172 pretty = F_PRETTY
934 allow_nonref = F_ALLOW_NONREF 1173 allow_nonref = F_ALLOW_NONREF
935 shrink = F_SHRINK 1174 shrink = F_SHRINK
936 CODE: 1175 CODE:
937{ 1176{
… …
944 RETVAL = newSVsv (self); 1183 RETVAL = newSVsv (self);
945} 1184}
946 OUTPUT: 1185 OUTPUT:
947 RETVAL 1186 RETVAL
948 1187
1188SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1189 CODE:
1190{
1191 UV *uv = SvJSON (self);
1192 UV log2 = 0;
1193
1194 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1195
1196 while ((1UL << log2) < max_depth)
1197 ++log2;
1198
1199 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1200
1201 RETVAL = newSVsv (self);
1202}
1203 OUTPUT:
1204 RETVAL
1205
949void encode (SV *self, SV *scalar) 1206void encode (SV *self, SV *scalar)
950 PPCODE: 1207 PPCODE:
951 XPUSHs (encode_json (scalar, *SvJSON (self))); 1208 XPUSHs (encode_json (scalar, *SvJSON (self)));
952 1209
953void decode (SV *self, SV *jsonstr) 1210void decode (SV *self, SV *jsonstr)
954 PPCODE: 1211 PPCODE:
955 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1212 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1213
1214void decode_prefix (SV *self, SV *jsonstr)
1215 PPCODE:
1216{
1217 UV offset;
1218 EXTEND (SP, 2);
1219 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1220 PUSHs (sv_2mortal (newSVuv (offset)));
1221}
956 1222
957PROTOTYPES: ENABLE 1223PROTOTYPES: ENABLE
958 1224
959void to_json (SV *scalar) 1225void to_json (SV *scalar)
1226 ALIAS:
1227 objToJson = 0
960 PPCODE: 1228 PPCODE:
961 XPUSHs (encode_json (scalar, F_UTF8)); 1229 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
962 1230
963void from_json (SV *jsonstr) 1231void from_json (SV *jsonstr)
1232 ALIAS:
1233 jsonToObj = 0
964 PPCODE: 1234 PPCODE:
965 XPUSHs (decode_json (jsonstr, F_UTF8)); 1235 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
966 1236

Diff Legend

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