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 1.7 by root, Fri Mar 23 15:57:18 2007 UTC vs.
Revision 1.44 by root, Mon Jun 25 04:08:17 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_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_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL // NYI
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33
34#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
18 35
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 36#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 37#define F_DEFAULT (9UL << S_MAXDEPTH)
21 38
22#define INIT_SIZE 32 // initial scalar size to be allocated 39#define INIT_SIZE 32 // initial scalar size to be allocated
40#define INDENT_STEP 3 // spaces per indentation level
41
42#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
23 43
24#define SB do { 44#define SB do {
25#define SE } while (0) 45#define SE } while (0)
26 46
27static HV *json_stash; 47#if __GNUC__ >= 3
48# define expect(expr,value) __builtin_expect ((expr),(value))
49# define inline inline
50#else
51# define expect(expr,value) (expr)
52# define inline static
53#endif
54
55#define expect_false(expr) expect ((expr) != 0, 0)
56#define expect_true(expr) expect ((expr) != 0, 1)
57
58static HV *json_stash, *json_boolean_stash; // JSON::XS::
59static SV *json_true, *json_false;
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
59static void
60shrink (SV *sv)
61{
62 sv_utf8_downgrade (sv, 1);
63#ifdef SvPV_shrink_to_cur
64 SvPV_shrink_to_cur (sv);
65#endif
66}
67
68/////////////////////////////////////////////////////////////////////////////
69
70static void
71need (enc_t *enc, STRLEN len) 124need (enc_t *enc, STRLEN len)
72{ 125{
73 if (enc->cur + len >= enc->end) 126 if (expect_false (enc->cur + len >= enc->end))
74 { 127 {
75 STRLEN cur = enc->cur - SvPVX (enc->sv); 128 STRLEN cur = enc->cur - SvPVX (enc->sv);
76 SvGROW (enc->sv, cur + len + 1); 129 SvGROW (enc->sv, cur + len + 1);
77 enc->cur = SvPVX (enc->sv) + cur; 130 enc->cur = SvPVX (enc->sv) + cur;
78 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
79 } 132 }
80} 133}
81 134
82static void 135inline void
83encode_ch (enc_t *enc, char ch) 136encode_ch (enc_t *enc, char ch)
84{ 137{
85 need (enc, 1); 138 need (enc, 1);
86 *enc->cur++ = ch; 139 *enc->cur++ = ch;
87} 140}
… …
95 148
96 while (str < end) 149 while (str < end)
97 { 150 {
98 unsigned char ch = *(unsigned char *)str; 151 unsigned char ch = *(unsigned char *)str;
99 152
100 if (ch >= 0x20 && ch < 0x80) // most common case 153 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
101 { 154 {
102 if (ch == '"') // but with slow exceptions 155 if (expect_false (ch == '"')) // but with slow exceptions
103 { 156 {
104 need (enc, len += 1); 157 need (enc, len += 1);
105 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
106 *enc->cur++ = '"'; 159 *enc->cur++ = '"';
107 } 160 }
108 else if (ch == '\\') 161 else if (expect_false (ch == '\\'))
109 { 162 {
110 need (enc, len += 1); 163 need (enc, len += 1);
111 *enc->cur++ = '\\'; 164 *enc->cur++ = '\\';
112 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
113 } 166 }
… …
131 STRLEN clen; 184 STRLEN clen;
132 UV uch; 185 UV uch;
133 186
134 if (is_utf8) 187 if (is_utf8)
135 { 188 {
136 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 189 uch = decode_utf8 (str, end - str, &clen);
137 if (clen == (STRLEN)-1) 190 if (clen == (STRLEN)-1)
138 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);
139 } 192 }
140 else 193 else
141 { 194 {
142 uch = ch; 195 uch = ch;
143 clen = 1; 196 clen = 1;
144 } 197 }
145 198
146 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))
147 { 203 {
148 if (uch > 0xFFFFUL) 204 if (uch > 0xFFFFUL)
149 { 205 {
150 need (enc, len += 11); 206 need (enc, len += 11);
151 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
152 (uch - 0x10000) / 0x400 + 0xD800, 208 (int)((uch - 0x10000) / 0x400 + 0xD800),
153 (uch - 0x10000) % 0x400 + 0xDC00); 209 (int)((uch - 0x10000) % 0x400 + 0xDC00));
154 enc->cur += 12; 210 enc->cur += 12;
155 } 211 }
156 else 212 else
157 { 213 {
158 static char hexdigit [16] = "0123456789abcdef"; 214 static char hexdigit [16] = "0123456789abcdef";
… …
165 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
166 } 222 }
167 223
168 str += clen; 224 str += clen;
169 } 225 }
226 else if (enc->flags & F_LATIN1)
227 {
228 *enc->cur++ = uch;
229 str += clen;
230 }
170 else if (is_utf8) 231 else if (is_utf8)
171 { 232 {
172 need (enc, len += clen); 233 need (enc, len += clen);
173 do 234 do
174 { 235 {
… …
176 } 237 }
177 while (--clen); 238 while (--clen);
178 } 239 }
179 else 240 else
180 { 241 {
181 need (enc, len += 10); // never more than 11 bytes needed 242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
182 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
183 ++str; 244 ++str;
184 } 245 }
185 } 246 }
186 } 247 }
… …
188 249
189 --len; 250 --len;
190 } 251 }
191} 252}
192 253
193#define INDENT SB \ 254inline void
255encode_indent (enc_t *enc)
256{
194 if (enc->flags & F_INDENT) \ 257 if (enc->flags & F_INDENT)
195 { \ 258 {
196 int i_; \ 259 int spaces = enc->indent * INDENT_STEP;
197 need (enc, enc->indent); \
198 for (i_ = enc->indent * 3; i_--; )\
199 encode_ch (enc, ' '); \
200 } \
201 SE
202 260
203#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 261 need (enc, spaces);
204#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 262 memset (enc->cur, ' ', spaces);
205#define COMMA SB \ 263 enc->cur += spaces;
264 }
265}
266
267inline void
268encode_space (enc_t *enc)
269{
270 need (enc, 1);
206 encode_ch (enc, ','); \ 271 encode_ch (enc, ' ');
272}
273
274inline void
275encode_nl (enc_t *enc)
276{
207 if (enc->flags & F_INDENT) \ 277 if (enc->flags & F_INDENT)
208 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);
209 else if (enc->flags & F_SPACE_AFTER) \ 291 else if (enc->flags & F_SPACE_AFTER)
210 SPACE; \ 292 encode_space (enc);
211 SE 293}
212 294
213static void encode_sv (enc_t *enc, SV *sv); 295static void encode_sv (enc_t *enc, SV *sv);
214 296
215static void 297static void
216encode_av (enc_t *enc, AV *av) 298encode_av (enc_t *enc, AV *av)
217{ 299{
218 int i, len = av_len (av); 300 int i, len = av_len (av);
219 301
220 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);
221 ++enc->indent; 306 ++enc->indent;
222 307
223 for (i = 0; i <= len; ++i) 308 for (i = 0; i <= len; ++i)
224 { 309 {
225 INDENT; 310 encode_indent (enc);
226 encode_sv (enc, *av_fetch (av, i, 0)); 311 encode_sv (enc, *av_fetch (av, i, 0));
227 312
228 if (i < len) 313 if (i < len)
229 COMMA; 314 encode_comma (enc);
230 } 315 }
231 316
232 NL; 317 encode_nl (enc);
233 318
234 --enc->indent; 319 --enc->indent;
235 INDENT; encode_ch (enc, ']'); 320 encode_indent (enc); encode_ch (enc, ']');
236} 321}
237 322
238static void 323static void
239encode_he (enc_t *enc, HE *he) 324encode_he (enc_t *enc, HE *he)
240{ 325{
… …
254 else 339 else
255 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
256 341
257 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
258 343
259 if (enc->flags & F_SPACE_BEFORE) SPACE; 344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
260 encode_ch (enc, ':'); 345 encode_ch (enc, ':');
261 if (enc->flags & F_SPACE_AFTER ) SPACE; 346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
262 encode_sv (enc, HeVAL (he)); 347 encode_sv (enc, HeVAL (he));
263} 348}
264 349
265// compare hash entries, used when all keys are bytestrings 350// compare hash entries, used when all keys are bytestrings
266static int 351static int
… …
272 HE *b = *(HE **)b_; 357 HE *b = *(HE **)b_;
273 358
274 STRLEN la = HeKLEN (a); 359 STRLEN la = HeKLEN (a);
275 STRLEN lb = HeKLEN (b); 360 STRLEN lb = HeKLEN (b);
276 361
277 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
278 cmp = la < lb ? -1 : la == lb ? 0 : 1; 363 cmp = la - lb;
279 364
280 return cmp; 365 return cmp;
281} 366}
282 367
283// 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
… …
290static void 375static void
291encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
292{ 377{
293 int count, i; 378 int count, i;
294 379
380 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)");
382
295 encode_ch (enc, '{'); NL; ++enc->indent; 383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
296 384
297 if ((count = hv_iterinit (hv))) 385 if ((count = hv_iterinit (hv)))
298 { 386 {
299 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
300 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
301 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
302 // that randomises hash orderings 390 // that randomises hash orderings
303 if (enc->flags & F_CANONICAL) 391 if (enc->flags & F_CANONICAL)
304 { 392 {
305 HE *he, *hes [count];
306 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
307 400
308 i = 0; 401 i = 0;
309 while ((he = hv_iternext (hv))) 402 while ((he = hv_iternext (hv)))
310 { 403 {
311 hes [i++] = he; 404 hes [i++] = he;
… …
317 410
318 if (fast) 411 if (fast)
319 qsort (hes, count, sizeof (HE *), he_cmp_fast); 412 qsort (hes, count, sizeof (HE *), he_cmp_fast);
320 else 413 else
321 { 414 {
322 // hack to disable "use bytes" 415 // hack to forcefully disable "use bytes"
323 COP *oldcop = PL_curcop, cop; 416 COP cop = *PL_curcop;
324 cop.op_private = 0; 417 cop.op_private = 0;
418
419 ENTER;
420 SAVETMPS;
421
422 SAVEVPTR (PL_curcop);
325 PL_curcop = &cop; 423 PL_curcop = &cop;
326 424
327 SAVETMPS;
328 qsort (hes, count, sizeof (HE *), he_cmp_slow); 425 qsort (hes, count, sizeof (HE *), he_cmp_slow);
426
329 FREETMPS; 427 FREETMPS;
330 428 LEAVE;
331 PL_curcop = oldcop;
332 } 429 }
333 430
334 for (i = 0; i < count; ++i) 431 for (i = 0; i < count; ++i)
335 { 432 {
336 INDENT; 433 encode_indent (enc);
337 encode_he (enc, hes [i]); 434 encode_he (enc, hes [i]);
338 435
339 if (i < count - 1) 436 if (i < count - 1)
340 COMMA; 437 encode_comma (enc);
341 } 438 }
342 439
343 NL; 440 encode_nl (enc);
344 } 441 }
345 else 442 else
346 { 443 {
347 SV *sv;
348 HE *he = hv_iternext (hv); 444 HE *he = hv_iternext (hv);
349 445
350 for (;;) 446 for (;;)
351 { 447 {
352 INDENT; 448 encode_indent (enc);
353 encode_he (enc, he); 449 encode_he (enc, he);
354 450
355 if (!(he = hv_iternext (hv))) 451 if (!(he = hv_iternext (hv)))
356 break; 452 break;
357 453
358 COMMA; 454 encode_comma (enc);
359 } 455 }
360 456
361 NL; 457 encode_nl (enc);
362 } 458 }
363 } 459 }
364 460
365 --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 (expect_false (SvOBJECT (sv)))
474 {
475 if (SvSTASH (sv) == json_boolean_stash)
476 {
477 if (SvIV (sv) == 0)
478 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 }
482 else
483 {
484#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 {
487 // not yet
488 }
489#endif
490 if (enc->flags & F_CONV_BLESSED)
491 {
492 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
494
495 if (to_json)
496 {
497 dSP;
498 ENTER;
499 SAVETMPS;
500 PUSHMARK (SP);
501 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
502
503 // calling with G_SCALAR ensures that we always get a 1 reutrn value
504 // check anyways.
505 PUTBACK;
506 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
507 SPAGAIN;
508
509 encode_sv (enc, POPs);
510
511 FREETMPS;
512 LEAVE;
513 }
514 else if (enc->flags & F_ALLOW_BLESSED)
515 encode_str (enc, "null", 4, 0);
516 else
517 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
518 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
519 }
520 else if (enc->flags & F_ALLOW_BLESSED)
521 encode_str (enc, "null", 4, 0);
522 else
523 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
524 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
525 }
526 }
527 else if (svt == SVt_PVHV)
528 encode_hv (enc, (HV *)sv);
529 else if (svt == SVt_PVAV)
530 encode_av (enc, (AV *)sv);
531 else if (svt < SVt_PVAV)
532 {
533 if (SvNIOK (sv) && SvIV (sv) == 0)
534 encode_str (enc, "false", 5, 0);
535 else if (SvNIOK (sv) && SvIV (sv) == 1)
536 encode_str (enc, "true", 4, 0);
537 else
538 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540 }
541 else
542 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
543 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
366} 544}
367 545
368static void 546static void
369encode_sv (enc_t *enc, SV *sv) 547encode_sv (enc_t *enc, SV *sv)
370{ 548{
… …
378 encode_str (enc, str, len, SvUTF8 (sv)); 556 encode_str (enc, str, len, SvUTF8 (sv));
379 encode_ch (enc, '"'); 557 encode_ch (enc, '"');
380 } 558 }
381 else if (SvNOKp (sv)) 559 else if (SvNOKp (sv))
382 { 560 {
561 // trust that perl will do the right thing w.r.t. JSON syntax.
383 need (enc, NV_DIG + 32); 562 need (enc, NV_DIG + 32);
384 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 563 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
385 enc->cur += strlen (enc->cur); 564 enc->cur += strlen (enc->cur);
386 } 565 }
387 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
388 { 567 {
389 need (enc, 64); 568 // we assume we can always read an IV as a UV
569 if (SvUV (sv) & ~(UV)0x7fff)
570 {
571 // large integer, use the (rather slow) snprintf way.
572 need (enc, sizeof (UV) * 3);
390 enc->cur += 573 enc->cur +=
391 SvIsUV(sv) 574 SvIsUV(sv)
392 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 575 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
393 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 576 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
577 }
578 else
579 {
580 // optimise the "small number case"
581 // code will likely be branchless and use only a single multiplication
582 I32 i = SvIV (sv);
583 U32 u;
584 char digit, nz = 0;
585
586 need (enc, 6);
587
588 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
589 u = i < 0 ? -i : i;
590
591 // convert to 4.28 fixed-point representation
592 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
593
594 // now output digit by digit, each time masking out the integer part
595 // and multiplying by 5 while moving the decimal point one to the right,
596 // resulting in a net multiplication by 10.
597 // we always write the digit to memory but conditionally increment
598 // the pointer, to ease the usage of conditional move instructions.
599 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
600 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
601 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
602 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
603 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
604 }
394 } 605 }
395 else if (SvROK (sv)) 606 else if (SvROK (sv))
396 { 607 encode_rv (enc, SvRV (sv));
397 if (!--enc->max_recurse)
398 croak ("data structure too deep (hit recursion limit)");
399
400 sv = SvRV (sv);
401
402 switch (SvTYPE (sv))
403 {
404 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
405 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
406
407 default:
408 croak ("JSON can only represent references to arrays or hashes");
409 }
410 }
411 else if (!SvOK (sv)) 608 else if (!SvOK (sv))
412 encode_str (enc, "null", 4, 0); 609 encode_str (enc, "null", 4, 0);
413 else 610 else
414 croak ("encountered perl type that JSON cannot handle"); 611 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
612 SvPV_nolen (sv), SvFLAGS (sv));
415} 613}
416 614
417static SV * 615static SV *
418encode_json (SV *scalar, UV flags) 616encode_json (SV *scalar, U32 flags)
419{ 617{
618 enc_t enc;
619
420 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 620 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
421 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 621 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
422 622
423 enc_t enc;
424 enc.flags = flags; 623 enc.flags = flags;
425 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 624 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
426 enc.cur = SvPVX (enc.sv); 625 enc.cur = SvPVX (enc.sv);
427 enc.end = SvEND (enc.sv); 626 enc.end = SvEND (enc.sv);
428 enc.max_recurse = 0;
429 enc.indent = 0; 627 enc.indent = 0;
628 enc.maxdepth = DEC_DEPTH (flags);
430 629
431 SvPOK_only (enc.sv); 630 SvPOK_only (enc.sv);
432 encode_sv (&enc, scalar); 631 encode_sv (&enc, scalar);
433 632
633 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
634 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
635
434 if (!(flags & (F_ASCII | F_UTF8))) 636 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
435 SvUTF8_on (enc.sv); 637 SvUTF8_on (enc.sv);
436
437 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
438 638
439 if (enc.flags & F_SHRINK) 639 if (enc.flags & F_SHRINK)
440 shrink (enc.sv); 640 shrink (enc.sv);
441 641
442 return enc.sv; 642 return enc.sv;
443} 643}
444 644
445///////////////////////////////////////////////////////////////////////////// 645/////////////////////////////////////////////////////////////////////////////
646// decoder
446 647
447#define WS \ 648// structure used for decoding JSON
649typedef struct
650{
651 char *cur; // current parser pointer
652 char *end; // end of input string
653 const char *err; // parse error, if != 0
654 U32 flags; // F_*
655 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit
657} dec_t;
658
659inline void
660decode_ws (dec_t *dec)
661{
448 for (;;) \ 662 for (;;)
449 { \ 663 {
450 char ch = *dec->cur; \ 664 char ch = *dec->cur;
665
451 if (ch > 0x20 \ 666 if (ch > 0x20
452 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 667 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
453 break; \ 668 break;
669
454 ++dec->cur; \ 670 ++dec->cur;
455 } 671 }
672}
456 673
457#define ERR(reason) SB dec->err = reason; goto fail; SE 674#define ERR(reason) SB dec->err = reason; goto fail; SE
675
458#define EXPECT_CH(ch) SB \ 676#define EXPECT_CH(ch) SB \
459 if (*dec->cur != ch) \ 677 if (*dec->cur != ch) \
460 ERR (# ch " expected"); \ 678 ERR (# ch " expected"); \
461 ++dec->cur; \ 679 ++dec->cur; \
462 SE 680 SE
463 681
682#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
683#define DEC_DEC_DEPTH --dec->depth
684
464static SV *decode_sv (dec_t *dec); 685static SV *decode_sv (dec_t *dec);
465 686
466static signed char decode_hexdigit[256]; 687static signed char decode_hexdigit[256];
467 688
468static UV 689static UV
469decode_4hex (dec_t *dec) 690decode_4hex (dec_t *dec)
470{ 691{
471 signed char d1, d2, d3, d4; 692 signed char d1, d2, d3, d4;
693 unsigned char *cur = (unsigned char *)dec->cur;
472 694
473 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 695 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
474 if (d1 < 0) ERR ("four hexadecimal digits expected"); 696 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
475 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 697 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
476 if (d2 < 0) ERR ("four hexadecimal digits expected"); 698 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
477 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
478 if (d3 < 0) ERR ("four hexadecimal digits expected");
479 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
480 if (d4 < 0) ERR ("four hexadecimal digits expected");
481 699
482 dec->cur += 4; 700 dec->cur += 4;
483 701
484 return ((UV)d1) << 12 702 return ((UV)d1) << 12
485 | ((UV)d2) << 8 703 | ((UV)d2) << 8
… …
488 706
489fail: 707fail:
490 return (UV)-1; 708 return (UV)-1;
491} 709}
492 710
493#define APPEND_GROW(n) SB \
494 if (cur + (n) >= end) \
495 { \
496 STRLEN ofs = cur - SvPVX (sv); \
497 SvGROW (sv, ofs + (n) + 1); \
498 cur = SvPVX (sv) + ofs; \
499 end = SvEND (sv); \
500 } \
501 SE
502
503#define APPEND_CH(ch) SB \
504 APPEND_GROW (1); \
505 *cur++ = (ch); \
506 SE
507
508static SV * 711static SV *
509decode_str (dec_t *dec) 712decode_str (dec_t *dec)
510{ 713{
511 SV *sv = NEWSV (0,2); 714 SV *sv = 0;
512 int utf8 = 0; 715 int utf8 = 0;
513 char *cur = SvPVX (sv); 716 char *dec_cur = dec->cur;
514 char *end = SvEND (sv);
515 717
516 for (;;) 718 do
517 { 719 {
518 unsigned char ch = *(unsigned char *)dec->cur; 720 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
721 char *cur = buf;
519 722
520 if (ch == '"') 723 do
521 break;
522 else if (ch == '\\')
523 { 724 {
524 switch (*++dec->cur) 725 unsigned char ch = *(unsigned char *)dec_cur++;
726
727 if (expect_false (ch == '"'))
525 { 728 {
526 case '\\': 729 --dec_cur;
527 case '/': 730 break;
528 case '"': APPEND_CH (*dec->cur++); break; 731 }
529 732 else if (expect_false (ch == '\\'))
530 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 733 {
531 case 't': APPEND_CH ('\011'); ++dec->cur; break; 734 switch (*dec_cur)
532 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
533 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
534 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
535
536 case 'u':
537 { 735 {
538 UV lo, hi; 736 case '\\':
539 ++dec->cur; 737 case '/':
738 case '"': *cur++ = *dec_cur++; break;
540 739
541 hi = decode_4hex (dec); 740 case 'b': ++dec_cur; *cur++ = '\010'; break;
542 if (hi == (UV)-1) 741 case 't': ++dec_cur; *cur++ = '\011'; break;
543 goto fail; 742 case 'n': ++dec_cur; *cur++ = '\012'; break;
743 case 'f': ++dec_cur; *cur++ = '\014'; break;
744 case 'r': ++dec_cur; *cur++ = '\015'; break;
544 745
545 // possibly a surrogate pair 746 case 'u':
546 if (hi >= 0xd800 && hi < 0xdc00)
547 { 747 {
548 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 748 UV lo, hi;
549 ERR ("missing low surrogate character in surrogate pair"); 749 ++dec_cur;
550 750
551 dec->cur += 2; 751 dec->cur = dec_cur;
552
553 lo = decode_4hex (dec); 752 hi = decode_4hex (dec);
753 dec_cur = dec->cur;
554 if (lo == (UV)-1) 754 if (hi == (UV)-1)
555 goto fail; 755 goto fail;
556 756
757 // possibly a surrogate pair
758 if (hi >= 0xd800)
759 if (hi < 0xdc00)
760 {
761 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
762 ERR ("missing low surrogate character in surrogate pair");
763
764 dec_cur += 2;
765
766 dec->cur = dec_cur;
767 lo = decode_4hex (dec);
768 dec_cur = dec->cur;
769 if (lo == (UV)-1)
770 goto fail;
771
557 if (lo < 0xdc00 || lo >= 0xe000) 772 if (lo < 0xdc00 || lo >= 0xe000)
558 ERR ("surrogate pair expected"); 773 ERR ("surrogate pair expected");
559 774
560 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 775 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
776 }
777 else if (hi < 0xe000)
778 ERR ("missing high surrogate character in surrogate pair");
779
780 if (hi >= 0x80)
781 {
782 utf8 = 1;
783
784 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
785 }
786 else
787 *cur++ = hi;
561 } 788 }
562 else if (hi >= 0xdc00 && hi < 0xe000)
563 ERR ("missing high surrogate character in surrogate pair");
564
565 if (hi >= 0x80)
566 { 789 break;
567 utf8 = 1;
568 790
569 APPEND_GROW (4); // at most 4 bytes for 21 bits
570 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
571 }
572 else 791 default:
573 APPEND_CH (hi); 792 --dec_cur;
793 ERR ("illegal backslash escape sequence in string");
574 } 794 }
575 break; 795 }
796 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
797 *cur++ = ch;
798 else if (ch >= 0x80)
799 {
800 STRLEN clen;
801 UV uch;
576 802
577 default:
578 --dec->cur; 803 --dec_cur;
579 ERR ("illegal backslash escape sequence in string"); 804
805 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
806 if (clen == (STRLEN)-1)
807 ERR ("malformed UTF-8 character in JSON string");
808
809 do
810 *cur++ = *dec_cur++;
811 while (--clen);
812
813 utf8 = 1;
814 }
815 else
816 {
817 --dec_cur;
818
819 if (!ch)
820 ERR ("unexpected end of string while parsing JSON string");
821 else
822 ERR ("invalid character encountered while parsing JSON string");
580 } 823 }
581 } 824 }
582 else if (ch >= 0x20 && ch <= 0x7f) 825 while (cur < buf + SHORT_STRING_LEN);
583 APPEND_CH (*dec->cur++); 826
584 else if (ch >= 0x80)
585 { 827 {
586 STRLEN clen; 828 STRLEN len = cur - buf;
587 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
588 if (clen == (STRLEN)-1)
589 ERR ("malformed UTF-8 character in JSON string");
590 829
591 APPEND_GROW (clen); 830 if (sv)
592 do
593 { 831 {
594 *cur++ = *dec->cur++; 832 SvGROW (sv, SvCUR (sv) + len + 1);
833 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
834 SvCUR_set (sv, SvCUR (sv) + len);
595 } 835 }
596 while (--clen);
597
598 utf8 = 1;
599 }
600 else if (dec->cur == dec->end)
601 ERR ("unexpected end of string while parsing json string");
602 else 836 else
603 ERR ("invalid character encountered"); 837 sv = newSVpvn (buf, len);
604 } 838 }
839 }
840 while (*dec_cur != '"');
605 841
606 ++dec->cur; 842 ++dec_cur;
607 843
608 SvCUR_set (sv, cur - SvPVX (sv)); 844 if (sv)
609 845 {
610 SvPOK_only (sv); 846 SvPOK_only (sv);
611 *SvEND (sv) = 0; 847 *SvEND (sv) = 0;
612 848
613 if (utf8) 849 if (utf8)
614 SvUTF8_on (sv); 850 SvUTF8_on (sv);
851 }
852 else
853 sv = newSVpvn ("", 0);
615 854
616 if (dec->flags & F_SHRINK) 855 dec->cur = dec_cur;
617 shrink (sv);
618
619 return sv; 856 return sv;
620 857
621fail: 858fail:
622 SvREFCNT_dec (sv); 859 dec->cur = dec_cur;
623 return 0; 860 return 0;
624} 861}
625 862
626static SV * 863static SV *
627decode_num (dec_t *dec) 864decode_num (dec_t *dec)
… …
685 is_nv = 1; 922 is_nv = 1;
686 } 923 }
687 924
688 if (!is_nv) 925 if (!is_nv)
689 { 926 {
690 UV uv; 927 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
691 int numtype = grok_number (start, dec->cur - start, &uv); 928 if (*start == '-')
692 if (numtype & IS_NUMBER_IN_UV) 929 switch (dec->cur - start)
693 if (numtype & IS_NUMBER_NEG)
694 { 930 {
695 if (uv < (UV)IV_MIN) 931 case 2: return newSViv (-( start [1] - '0' * 1));
696 return newSViv (-(IV)uv); 932 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
933 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
934 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
697 } 935 }
936 else
937 switch (dec->cur - start)
938 {
939 case 1: return newSViv ( start [0] - '0' * 1);
940 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
941 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
942 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
943 }
944
945 {
946 UV uv;
947 int numtype = grok_number (start, dec->cur - start, &uv);
948 if (numtype & IS_NUMBER_IN_UV)
949 if (numtype & IS_NUMBER_NEG)
950 {
951 if (uv < (UV)IV_MIN)
952 return newSViv (-(IV)uv);
953 }
698 else 954 else
699 return newSVuv (uv); 955 return newSVuv (uv);
956
957 // here would likely be the place for bigint support
700 } 958 }
959 }
701 960
961 // if we ever support bigint or bigfloat, this is the place for bigfloat
702 return newSVnv (Atof (start)); 962 return newSVnv (Atof (start));
703 963
704fail: 964fail:
705 return 0; 965 return 0;
706} 966}
… …
708static SV * 968static SV *
709decode_av (dec_t *dec) 969decode_av (dec_t *dec)
710{ 970{
711 AV *av = newAV (); 971 AV *av = newAV ();
712 972
713 WS; 973 DEC_INC_DEPTH;
974 decode_ws (dec);
975
714 if (*dec->cur == ']') 976 if (*dec->cur == ']')
715 ++dec->cur; 977 ++dec->cur;
716 else 978 else
717 for (;;) 979 for (;;)
718 { 980 {
… …
722 if (!value) 984 if (!value)
723 goto fail; 985 goto fail;
724 986
725 av_push (av, value); 987 av_push (av, value);
726 988
727 WS; 989 decode_ws (dec);
728 990
729 if (*dec->cur == ']') 991 if (*dec->cur == ']')
730 { 992 {
731 ++dec->cur; 993 ++dec->cur;
732 break; 994 break;
… …
736 ERR (", or ] expected while parsing array"); 998 ERR (", or ] expected while parsing array");
737 999
738 ++dec->cur; 1000 ++dec->cur;
739 } 1001 }
740 1002
1003 DEC_DEC_DEPTH;
741 return newRV_noinc ((SV *)av); 1004 return newRV_noinc ((SV *)av);
742 1005
743fail: 1006fail:
744 SvREFCNT_dec (av); 1007 SvREFCNT_dec (av);
1008 DEC_DEC_DEPTH;
745 return 0; 1009 return 0;
746} 1010}
747 1011
748static SV * 1012static SV *
749decode_hv (dec_t *dec) 1013decode_hv (dec_t *dec)
750{ 1014{
751 HV *hv = newHV (); 1015 HV *hv = newHV ();
752 1016
753 WS; 1017 DEC_INC_DEPTH;
1018 decode_ws (dec);
1019
754 if (*dec->cur == '}') 1020 if (*dec->cur == '}')
755 ++dec->cur; 1021 ++dec->cur;
756 else 1022 else
757 for (;;) 1023 for (;;)
758 { 1024 {
759 SV *key, *value; 1025 SV *key, *value;
760 1026
761 WS; EXPECT_CH ('"'); 1027 decode_ws (dec); EXPECT_CH ('"');
762 1028
763 key = decode_str (dec); 1029 key = decode_str (dec);
764 if (!key) 1030 if (!key)
765 goto fail; 1031 goto fail;
766 1032
767 WS; EXPECT_CH (':'); 1033 decode_ws (dec); EXPECT_CH (':');
768 1034
769 value = decode_sv (dec); 1035 value = decode_sv (dec);
770 if (!value) 1036 if (!value)
771 { 1037 {
772 SvREFCNT_dec (key); 1038 SvREFCNT_dec (key);
773 goto fail; 1039 goto fail;
774 } 1040 }
775 1041
776 //TODO: optimise
777 hv_store_ent (hv, key, value, 0); 1042 hv_store_ent (hv, key, value, 0);
1043 SvREFCNT_dec (key);
778 1044
779 WS; 1045 decode_ws (dec);
780 1046
781 if (*dec->cur == '}') 1047 if (*dec->cur == '}')
782 { 1048 {
783 ++dec->cur; 1049 ++dec->cur;
784 break; 1050 break;
… …
788 ERR (", or } expected while parsing object/hash"); 1054 ERR (", or } expected while parsing object/hash");
789 1055
790 ++dec->cur; 1056 ++dec->cur;
791 } 1057 }
792 1058
1059 DEC_DEC_DEPTH;
793 return newRV_noinc ((SV *)hv); 1060 return newRV_noinc ((SV *)hv);
794 1061
795fail: 1062fail:
796 SvREFCNT_dec (hv); 1063 SvREFCNT_dec (hv);
1064 DEC_DEC_DEPTH;
797 return 0; 1065 return 0;
798} 1066}
799 1067
800static SV * 1068static SV *
801decode_sv (dec_t *dec) 1069decode_sv (dec_t *dec)
802{ 1070{
803 WS; 1071 decode_ws (dec);
1072
1073 // the beauty of JSON: you need exactly one character lookahead
1074 // to parse anything.
804 switch (*dec->cur) 1075 switch (*dec->cur)
805 { 1076 {
806 case '"': ++dec->cur; return decode_str (dec); 1077 case '"': ++dec->cur; return decode_str (dec);
807 case '[': ++dec->cur; return decode_av (dec); 1078 case '[': ++dec->cur; return decode_av (dec);
808 case '{': ++dec->cur; return decode_hv (dec); 1079 case '{': ++dec->cur; return decode_hv (dec);
… …
814 1085
815 case 't': 1086 case 't':
816 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1087 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
817 { 1088 {
818 dec->cur += 4; 1089 dec->cur += 4;
819 return newSViv (1); 1090 return SvREFCNT_inc (json_true);
820 } 1091 }
821 else 1092 else
822 ERR ("'true' expected"); 1093 ERR ("'true' expected");
823 1094
824 break; 1095 break;
825 1096
826 case 'f': 1097 case 'f':
827 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1098 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
828 { 1099 {
829 dec->cur += 5; 1100 dec->cur += 5;
830 return newSViv (0); 1101 return SvREFCNT_inc (json_false);
831 } 1102 }
832 else 1103 else
833 ERR ("'false' expected"); 1104 ERR ("'false' expected");
834 1105
835 break; 1106 break;
… …
844 ERR ("'null' expected"); 1115 ERR ("'null' expected");
845 1116
846 break; 1117 break;
847 1118
848 default: 1119 default:
849 ERR ("malformed json string, neither array, object, number, string or atom"); 1120 ERR ("malformed JSON string, neither array, object, number, string or atom");
850 break; 1121 break;
851 } 1122 }
852 1123
853fail: 1124fail:
854 return 0; 1125 return 0;
855} 1126}
856 1127
857static SV * 1128static SV *
858decode_json (SV *string, UV flags) 1129decode_json (SV *string, U32 flags, UV *offset_return)
859{ 1130{
1131 dec_t dec;
1132 UV offset;
860 SV *sv; 1133 SV *sv;
1134
1135 SvGETMAGIC (string);
1136 SvUPGRADE (string, SVt_PV);
861 1137
862 if (flags & F_UTF8) 1138 if (flags & F_UTF8)
863 sv_utf8_downgrade (string, 0); 1139 sv_utf8_downgrade (string, 0);
864 else 1140 else
865 sv_utf8_upgrade (string); 1141 sv_utf8_upgrade (string);
866 1142
867 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1143 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
868 1144
869 dec_t dec;
870 dec.flags = flags; 1145 dec.flags = flags;
871 dec.cur = SvPVX (string); 1146 dec.cur = SvPVX (string);
872 dec.end = SvEND (string); 1147 dec.end = SvEND (string);
873 dec.err = 0; 1148 dec.err = 0;
1149 dec.depth = 0;
1150 dec.maxdepth = DEC_DEPTH (dec.flags);
874 1151
1152 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
875 sv = decode_sv (&dec); 1153 sv = decode_sv (&dec);
876 1154
1155 if (!(offset_return || !sv))
1156 {
1157 // check for trailing garbage
1158 decode_ws (&dec);
1159
1160 if (*dec.cur)
1161 {
1162 dec.err = "garbage after JSON object";
1163 SvREFCNT_dec (sv);
1164 sv = 0;
1165 }
1166 }
1167
1168 if (offset_return || !sv)
1169 {
1170 offset = dec.flags & F_UTF8
1171 ? dec.cur - SvPVX (string)
1172 : utf8_distance (dec.cur, SvPVX (string));
1173
1174 if (offset_return)
1175 *offset_return = offset;
1176 }
1177
877 if (!sv) 1178 if (!sv)
878 { 1179 {
879 IV offset = dec.flags & F_UTF8
880 ? dec.cur - SvPVX (string)
881 : utf8_distance (dec.cur, SvPVX (string));
882 SV *uni = sv_newmortal (); 1180 SV *uni = sv_newmortal ();
1181
883 // horrible hack to silence warning inside pv_uni_display 1182 // horrible hack to silence warning inside pv_uni_display
884 COP cop; 1183 COP cop = *PL_curcop;
885 memset (&cop, 0, sizeof (cop));
886 cop.cop_warnings = pWARN_NONE; 1184 cop.cop_warnings = pWARN_NONE;
1185 ENTER;
887 SAVEVPTR (PL_curcop); 1186 SAVEVPTR (PL_curcop);
888 PL_curcop = &cop; 1187 PL_curcop = &cop;
889
890 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1188 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1189 LEAVE;
1190
891 croak ("%s, at character offset %d (%s)", 1191 croak ("%s, at character offset %d [\"%s\"]",
892 dec.err, 1192 dec.err,
893 (int)offset, 1193 (int)offset,
894 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1194 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
895 } 1195 }
896 1196
897 sv = sv_2mortal (sv); 1197 sv = sv_2mortal (sv);
898 1198
899 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1199 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
900 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1200 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
901 1201
902 return sv; 1202 return sv;
903} 1203}
904 1204
1205/////////////////////////////////////////////////////////////////////////////
1206// XS interface functions
1207
905MODULE = JSON::XS PACKAGE = JSON::XS 1208MODULE = JSON::XS PACKAGE = JSON::XS
906 1209
907BOOT: 1210BOOT:
908{ 1211{
909 int i; 1212 int i;
910 1213
911 memset (decode_hexdigit, 0xff, 256);
912 for (i = 10; i--; ) 1214 for (i = 0; i < 256; ++i)
913 decode_hexdigit ['0' + i] = i; 1215 decode_hexdigit [i] =
1216 i >= '0' && i <= '9' ? i - '0'
1217 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1218 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1219 : -1;
914 1220
915 for (i = 7; i--; )
916 {
917 decode_hexdigit ['a' + i] = 10 + i;
918 decode_hexdigit ['A' + i] = 10 + i;
919 }
920
921 json_stash = gv_stashpv ("JSON::XS", 1); 1221 json_stash = gv_stashpv ("JSON::XS" , 1);
1222 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1223
1224 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1225 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
922} 1226}
923 1227
924PROTOTYPES: DISABLE 1228PROTOTYPES: DISABLE
925 1229
926SV *new (char *dummy) 1230SV *new (char *dummy)
… …
929 OUTPUT: 1233 OUTPUT:
930 RETVAL 1234 RETVAL
931 1235
932SV *ascii (SV *self, int enable = 1) 1236SV *ascii (SV *self, int enable = 1)
933 ALIAS: 1237 ALIAS:
934 ascii = F_ASCII 1238 ascii = F_ASCII
1239 latin1 = F_LATIN1
935 utf8 = F_UTF8 1240 utf8 = F_UTF8
936 indent = F_INDENT 1241 indent = F_INDENT
937 canonical = F_CANONICAL 1242 canonical = F_CANONICAL
938 space_before = F_SPACE_BEFORE 1243 space_before = F_SPACE_BEFORE
939 space_after = F_SPACE_AFTER 1244 space_after = F_SPACE_AFTER
940 json_rpc = F_JSON_RPC
941 pretty = F_PRETTY 1245 pretty = F_PRETTY
942 allow_nonref = F_ALLOW_NONREF 1246 allow_nonref = F_ALLOW_NONREF
943 shrink = F_SHRINK 1247 shrink = F_SHRINK
1248 allow_blessed = F_ALLOW_BLESSED
1249 convert_blessed = F_CONV_BLESSED
944 CODE: 1250 CODE:
945{ 1251{
946 UV *uv = SvJSON (self); 1252 UV *uv = SvJSON (self);
947 if (enable) 1253 if (enable)
948 *uv |= ix; 1254 *uv |= ix;
… …
952 RETVAL = newSVsv (self); 1258 RETVAL = newSVsv (self);
953} 1259}
954 OUTPUT: 1260 OUTPUT:
955 RETVAL 1261 RETVAL
956 1262
1263SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1264 CODE:
1265{
1266 UV *uv = SvJSON (self);
1267 UV log2 = 0;
1268
1269 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1270
1271 while ((1UL << log2) < max_depth)
1272 ++log2;
1273
1274 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1275
1276 RETVAL = newSVsv (self);
1277}
1278 OUTPUT:
1279 RETVAL
1280
957void encode (SV *self, SV *scalar) 1281void encode (SV *self, SV *scalar)
958 PPCODE: 1282 PPCODE:
959 XPUSHs (encode_json (scalar, *SvJSON (self))); 1283 XPUSHs (encode_json (scalar, *SvJSON (self)));
960 1284
961void decode (SV *self, SV *jsonstr) 1285void decode (SV *self, SV *jsonstr)
962 PPCODE: 1286 PPCODE:
963 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1287 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1288
1289void decode_prefix (SV *self, SV *jsonstr)
1290 PPCODE:
1291{
1292 UV offset;
1293 EXTEND (SP, 2);
1294 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1295 PUSHs (sv_2mortal (newSVuv (offset)));
1296}
964 1297
965PROTOTYPES: ENABLE 1298PROTOTYPES: ENABLE
966 1299
967void to_json (SV *scalar) 1300void to_json (SV *scalar)
1301 ALIAS:
1302 objToJson = 0
968 PPCODE: 1303 PPCODE:
969 XPUSHs (encode_json (scalar, F_UTF8)); 1304 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
970 1305
971void from_json (SV *jsonstr) 1306void from_json (SV *jsonstr)
1307 ALIAS:
1308 jsonToObj = 0
972 PPCODE: 1309 PPCODE:
973 XPUSHs (decode_json (jsonstr, F_UTF8)); 1310 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
974 1311

Diff Legend

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