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 1.1 by root, Thu Mar 22 16:40:16 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
21#define F_ALLOW_NONREF 0x00000100UL
15#define F_JSON_RPC 0x00000040 22#define F_SHRINK 0x00000200UL
23#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27
16 25
17#define F_DEFAULT 0 26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH)
18 33
19#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
20 38
21#define SB do { 39#define SB do {
22#define SE } while (0) 40#define SE } while (0)
23 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
24static 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
25 105
26// structure used for encoding JSON 106// structure used for encoding JSON
27typedef struct 107typedef struct
28{ 108{
29 char *cur; 109 char *cur; // SvPVX (sv) + current output position
30 STRLEN len; // SvLEN (sv)
31 char *end; // SvEND (sv) 110 char *end; // SvEND (sv)
32 SV *sv; 111 SV *sv; // result scalar
33 UV flags; 112 U32 flags; // F_*
34 int max_recurse; 113 U32 indent; // indentation level
35 int indent; 114 U32 maxdepth; // max. indentation/recursion level
36} enc_t; 115} enc_t;
37 116
38// structure used for decoding JSON 117inline void
39typedef struct
40{
41 char *cur;
42 char *end;
43 char *err;
44 UV flags;
45} dec_t;
46
47static UV *
48SvJSON (SV *sv)
49{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
51 croak ("object is not of type JSON::XS");
52
53 return &SvUVX (SvRV (sv));
54}
55
56/////////////////////////////////////////////////////////////////////////////
57
58static void
59need (enc_t *enc, STRLEN len) 118need (enc_t *enc, STRLEN len)
60{ 119{
61 if (enc->cur + len >= enc->end) 120 if (expect_false (enc->cur + len >= enc->end))
62 { 121 {
63 STRLEN cur = enc->cur - SvPVX (enc->sv); 122 STRLEN cur = enc->cur - SvPVX (enc->sv);
64 SvGROW (enc->sv, cur + len + 1); 123 SvGROW (enc->sv, cur + len + 1);
65 enc->cur = SvPVX (enc->sv) + cur; 124 enc->cur = SvPVX (enc->sv) + cur;
66 enc->end = SvEND (enc->sv); 125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
67 } 126 }
68} 127}
69 128
70static void 129inline void
71encode_ch (enc_t *enc, char ch) 130encode_ch (enc_t *enc, char ch)
72{ 131{
73 need (enc, 1); 132 need (enc, 1);
74 *enc->cur++ = ch; 133 *enc->cur++ = ch;
75} 134}
… …
77static void 136static void
78encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 137encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
79{ 138{
80 char *end = str + len; 139 char *end = str + len;
81 140
141 need (enc, len);
142
82 while (str < end) 143 while (str < end)
83 { 144 {
84 unsigned char ch = *(unsigned char *)str; 145 unsigned char ch = *(unsigned char *)str;
146
85 if (ch >= 0x20 && ch < 0x80) // most common case 147 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
86 { 148 {
149 if (expect_false (ch == '"')) // but with slow exceptions
150 {
151 need (enc, len += 1);
152 *enc->cur++ = '\\';
153 *enc->cur++ = '"';
154 }
155 else if (expect_false (ch == '\\'))
156 {
157 need (enc, len += 1);
158 *enc->cur++ = '\\';
159 *enc->cur++ = '\\';
160 }
161 else
87 *enc->cur++ = ch; 162 *enc->cur++ = ch;
163
88 str++; 164 ++str;
89 } 165 }
90 else 166 else
91 { 167 {
92 STRLEN clen; 168 switch (ch)
93 UV uch;
94
95 if (is_utf8)
96 { 169 {
97 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 170 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
98 if (clen < 0) 171 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
99 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 172 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
100 } 173 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
101 else 174 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
102 {
103 uch = ch;
104 clen = 1;
105 }
106 175
107 need (enc, len += 6); 176 default:
108
109 if (uch < 0xa0 || enc->flags & F_ASCII)
110 {
111 if (uch > 0xFFFFUL)
112 { 177 {
178 STRLEN clen;
179 UV uch;
180
181 if (is_utf8)
182 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 }
188 else
189 {
190 uch = ch;
113 len += 6; 191 clen = 1;
192 }
193
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))
198 {
199 if (uch > 0xFFFFUL)
200 {
114 need (enc, len += 6); 201 need (enc, len += 11);
115 sprintf (enc->cur, "\\u%04x\\u%04x", 202 sprintf (enc->cur, "\\u%04x\\u%04x",
116 (uch - 0x10000) / 0x400 + 0xD800, 203 (int)((uch - 0x10000) / 0x400 + 0xD800),
117 (uch - 0x10000) % 0x400 + 0xDC00); 204 (int)((uch - 0x10000) % 0x400 + 0xDC00));
118 enc->cur += 12; 205 enc->cur += 12;
206 }
207 else
208 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5);
211 *enc->cur++ = '\\';
212 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15];
217 }
218
219 str += clen;
119 } 220 }
221 else if (enc->flags & F_LATIN1)
222 {
223 *enc->cur++ = uch;
224 str += clen;
225 }
226 else if (is_utf8)
227 {
228 need (enc, len += clen);
229 do
230 {
231 *enc->cur++ = *str++;
232 }
233 while (--clen);
234 }
120 else 235 else
121 { 236 {
122 sprintf (enc->cur, "\\u%04x", uch); 237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
123 enc->cur += 6; 238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
239 ++str;
240 }
124 } 241 }
125 } 242 }
126 else if (is_utf8)
127 {
128 memcpy (enc->cur, str, clen);
129 enc->cur += clen;
130 }
131 else
132 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
133
134 str += clen;
135 } 243 }
136 244
137 --len; 245 --len;
138 } 246 }
139} 247}
140 248
141#define INDENT SB \ 249inline void
250encode_indent (enc_t *enc)
251{
142 if (enc->flags & F_INDENT) \ 252 if (enc->flags & F_INDENT)
143 { \ 253 {
144 int i_; \ 254 int spaces = enc->indent * INDENT_STEP;
145 need (enc, enc->indent); \
146 for (i_ = enc->indent * 3; i_--; )\
147 encode_ch (enc, ' '); \
148 } \
149 SE
150 255
151#define SPACE SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, ' '); } SE 256 need (enc, spaces);
152#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 257 memset (enc->cur, ' ', spaces);
153#define COMMA SB \ 258 enc->cur += spaces;
259 }
260}
261
262inline void
263encode_space (enc_t *enc)
264{
265 need (enc, 1);
154 encode_ch (enc, ','); \ 266 encode_ch (enc, ' ');
267}
268
269inline void
270encode_nl (enc_t *enc)
271{
155 if (enc->flags & F_INDENT) \ 272 if (enc->flags & F_INDENT)
156 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);
157 else if (enc->flags & F_SPACE_AFTER) \ 286 else if (enc->flags & F_SPACE_AFTER)
158 SPACE; \ 287 encode_space (enc);
159 SE 288}
160 289
161static void encode_sv (enc_t *enc, SV *sv); 290static void encode_sv (enc_t *enc, SV *sv);
162 291
163static void 292static void
164encode_av (enc_t *enc, AV *av) 293encode_av (enc_t *enc, AV *av)
165{ 294{
166 int i, len = av_len (av); 295 int i, len = av_len (av);
167 296
168 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);
169 ++enc->indent; 301 ++enc->indent;
170 302
171 for (i = 0; i <= len; ++i) 303 for (i = 0; i <= len; ++i)
172 { 304 {
173 INDENT; 305 encode_indent (enc);
174 encode_sv (enc, *av_fetch (av, i, 0)); 306 encode_sv (enc, *av_fetch (av, i, 0));
175 307
176 if (i < len) 308 if (i < len)
177 COMMA; 309 encode_comma (enc);
178 } 310 }
179 311
180 NL; 312 encode_nl (enc);
181 313
182 --enc->indent; 314 --enc->indent;
183 INDENT; encode_ch (enc, ']'); 315 encode_indent (enc); encode_ch (enc, ']');
184} 316}
185 317
186static void 318static void
187encode_he (enc_t *enc, HE *he) 319encode_he (enc_t *enc, HE *he)
188{ 320{
… …
190 322
191 if (HeKLEN (he) == HEf_SVKEY) 323 if (HeKLEN (he) == HEf_SVKEY)
192 { 324 {
193 SV *sv = HeSVKEY (he); 325 SV *sv = HeSVKEY (he);
194 STRLEN len; 326 STRLEN len;
327 char *str;
328
329 SvGETMAGIC (sv);
195 char *str = SvPV (sv, len); 330 str = SvPV (sv, len);
196 331
197 encode_str (enc, str, len, SvUTF8 (sv)); 332 encode_str (enc, str, len, SvUTF8 (sv));
198 } 333 }
199 else 334 else
200 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
201 336
202 encode_ch (enc, '"'); 337 encode_ch (enc, '"');
203 338
204 if (enc->flags & F_SPACE_BEFORE) SPACE; 339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
205 encode_ch (enc, ':'); 340 encode_ch (enc, ':');
206 if (enc->flags & F_SPACE_AFTER ) SPACE; 341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
207 encode_sv (enc, HeVAL (he)); 342 encode_sv (enc, HeVAL (he));
208} 343}
209 344
210// compare hash entries, used when all keys are bytestrings 345// compare hash entries, used when all keys are bytestrings
211static int 346static int
… …
217 HE *b = *(HE **)b_; 352 HE *b = *(HE **)b_;
218 353
219 STRLEN la = HeKLEN (a); 354 STRLEN la = HeKLEN (a);
220 STRLEN lb = HeKLEN (b); 355 STRLEN lb = HeKLEN (b);
221 356
222 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
223 cmp = la < lb ? -1 : la == lb ? 0 : 1; 358 cmp = la - lb;
224 359
225 return cmp; 360 return cmp;
226} 361}
227 362
228// 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
… …
235static void 370static void
236encode_hv (enc_t *enc, HV *hv) 371encode_hv (enc_t *enc, HV *hv)
237{ 372{
238 int count, i; 373 int count, i;
239 374
375 if (enc->indent >= enc->maxdepth)
376 croak ("data structure too deep (hit recursion limit)");
377
240 encode_ch (enc, '{'); NL; ++enc->indent; 378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
241 379
242 if ((count = hv_iterinit (hv))) 380 if ((count = hv_iterinit (hv)))
243 { 381 {
244 // for canonical output we have to sort by keys first 382 // for canonical output we have to sort by keys first
245 // actually, this is mostly due to the stupid so-called 383 // actually, this is mostly due to the stupid so-called
246 // security workaround added somewhere in 5.8.x. 384 // security workaround added somewhere in 5.8.x.
247 // that randomises hash orderings 385 // that randomises hash orderings
248 if (enc->flags & F_CANONICAL) 386 if (enc->flags & F_CANONICAL)
249 { 387 {
250 HE *he, *hes [count];
251 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
252 395
253 i = 0; 396 i = 0;
254 while ((he = hv_iternext (hv))) 397 while ((he = hv_iternext (hv)))
255 { 398 {
256 hes [i++] = he; 399 hes [i++] = he;
… …
262 405
263 if (fast) 406 if (fast)
264 qsort (hes, count, sizeof (HE *), he_cmp_fast); 407 qsort (hes, count, sizeof (HE *), he_cmp_fast);
265 else 408 else
266 { 409 {
267 // hack to disable "use bytes" 410 // hack to forcefully disable "use bytes"
268 COP *oldcop = PL_curcop, cop; 411 COP cop = *PL_curcop;
269 cop.op_private = 0; 412 cop.op_private = 0;
413
414 ENTER;
415 SAVETMPS;
416
417 SAVEVPTR (PL_curcop);
270 PL_curcop = &cop; 418 PL_curcop = &cop;
271 419
272 SAVETMPS;
273 qsort (hes, count, sizeof (HE *), he_cmp_slow); 420 qsort (hes, count, sizeof (HE *), he_cmp_slow);
421
274 FREETMPS; 422 FREETMPS;
275 423 LEAVE;
276 PL_curcop = oldcop;
277 } 424 }
278 425
279 for (i = 0; i < count; ++i) 426 for (i = 0; i < count; ++i)
280 { 427 {
281 INDENT; 428 encode_indent (enc);
282 encode_he (enc, hes [i]); 429 encode_he (enc, hes [i]);
283 430
284 if (i < count - 1) 431 if (i < count - 1)
285 COMMA; 432 encode_comma (enc);
286 } 433 }
287 434
288 NL; 435 encode_nl (enc);
289 } 436 }
290 else 437 else
291 { 438 {
292 SV *sv;
293 HE *he = hv_iternext (hv); 439 HE *he = hv_iternext (hv);
294 440
295 for (;;) 441 for (;;)
296 { 442 {
297 INDENT; 443 encode_indent (enc);
298 encode_he (enc, he); 444 encode_he (enc, he);
299 445
300 if (!(he = hv_iternext (hv))) 446 if (!(he = hv_iternext (hv)))
301 break; 447 break;
302 448
303 COMMA; 449 encode_comma (enc);
304 } 450 }
305 451
306 NL; 452 encode_nl (enc);
307 } 453 }
308 } 454 }
309 455
310 --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))));
311} 485}
312 486
313static void 487static void
314encode_sv (enc_t *enc, SV *sv) 488encode_sv (enc_t *enc, SV *sv)
315{ 489{
490 SvGETMAGIC (sv);
491
316 if (SvPOKp (sv)) 492 if (SvPOKp (sv))
317 { 493 {
318 STRLEN len; 494 STRLEN len;
319 char *str = SvPV (sv, len); 495 char *str = SvPV (sv, len);
320 encode_ch (enc, '"'); 496 encode_ch (enc, '"');
… …
327 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
328 enc->cur += strlen (enc->cur); 504 enc->cur += strlen (enc->cur);
329 } 505 }
330 else if (SvIOKp (sv)) 506 else if (SvIOKp (sv))
331 { 507 {
332 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);
333 enc->cur += 512 enc->cur +=
334 SvIsUV(sv) 513 SvIsUV(sv)
335 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 514 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
336 : 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 }
337 } 540 }
338 else if (SvROK (sv)) 541 else if (SvROK (sv))
339 { 542 encode_rv (enc, SvRV (sv));
340 if (!--enc->max_recurse)
341 croak ("data structure too deep (hit recursion limit)");
342
343 sv = SvRV (sv);
344
345 switch (SvTYPE (sv))
346 {
347 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
348 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
349
350 default:
351 croak ("JSON can only represent references to arrays or hashes");
352 }
353 }
354 else if (!SvOK (sv)) 543 else if (!SvOK (sv))
355 encode_str (enc, "null", 4, 0); 544 encode_str (enc, "null", 4, 0);
356 else 545 else
357 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));
358} 548}
359 549
360static SV * 550static SV *
361encode_json (SV *scalar, UV flags) 551encode_json (SV *scalar, U32 flags)
362{ 552{
363 enc_t enc; 553 enc_t enc;
554
555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557
364 enc.flags = flags; 558 enc.flags = flags;
365 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
366 enc.cur = SvPVX (enc.sv); 560 enc.cur = SvPVX (enc.sv);
367 enc.end = SvEND (enc.sv); 561 enc.end = SvEND (enc.sv);
368 enc.max_recurse = 0;
369 enc.indent = 0; 562 enc.indent = 0;
563 enc.maxdepth = DEC_DEPTH (flags);
370 564
371 SvPOK_only (enc.sv); 565 SvPOK_only (enc.sv);
372 encode_sv (&enc, scalar); 566 encode_sv (&enc, scalar);
373 567
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
374 if (!(flags & (F_ASCII | F_UTF8))) 571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
375 SvUTF8_on (enc.sv); 572 SvUTF8_on (enc.sv);
376 573
377 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 574 if (enc.flags & F_SHRINK)
575 shrink (enc.sv);
576
378 return enc.sv; 577 return enc.sv;
379} 578}
380 579
381///////////////////////////////////////////////////////////////////////////// 580/////////////////////////////////////////////////////////////////////////////
581// decoder
382 582
383#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{
384 for (;;) \ 597 for (;;)
385 { \ 598 {
386 char ch = *dec->cur; \ 599 char ch = *dec->cur;
600
387 if (ch > 0x20 \ 601 if (ch > 0x20
388 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 602 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
389 break; \ 603 break;
604
390 ++dec->cur; \ 605 ++dec->cur;
391 } 606 }
607}
392 608
393#define ERR(reason) SB dec->err = reason; goto fail; SE 609#define ERR(reason) SB dec->err = reason; goto fail; SE
610
394#define EXPECT_CH(ch) SB \ 611#define EXPECT_CH(ch) SB \
395 if (*dec->cur != ch) \ 612 if (*dec->cur != ch) \
396 ERR (# ch " expected"); \ 613 ERR (# ch " expected"); \
397 ++dec->cur; \ 614 ++dec->cur; \
398 SE 615 SE
399 616
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
400static SV *decode_sv (dec_t *dec); 620static SV *decode_sv (dec_t *dec);
401
402#define APPEND_CH(ch) SB \
403 SvGROW (sv, cur + 1 + 1); \
404 SvPVX (sv)[cur++] = (ch); \
405 SE
406 621
407static signed char decode_hexdigit[256]; 622static signed char decode_hexdigit[256];
408 623
409static UV 624static UV
410decode_4hex (dec_t *dec) 625decode_4hex (dec_t *dec)
411{ 626{
412 signed char d1, d2, d3, d4; 627 signed char d1, d2, d3, d4;
628 unsigned char *cur = (unsigned char *)dec->cur;
413 629
414 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");
415 if (d1 < 0) ERR ("four hexadecimal digits expected"); 631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected");
416 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected");
417 if (d2 < 0) ERR ("four hexadecimal digits expected"); 633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected");
418 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
419 if (d3 < 0) ERR ("four hexadecimal digits expected");
420 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
421 if (d4 < 0) ERR ("four hexadecimal digits expected");
422 634
423 dec->cur += 4; 635 dec->cur += 4;
424 636
425 return ((UV)d1) << 12 637 return ((UV)d1) << 12
426 | ((UV)d2) << 8 638 | ((UV)d2) << 8
… …
432} 644}
433 645
434static SV * 646static SV *
435decode_str (dec_t *dec) 647decode_str (dec_t *dec)
436{ 648{
437 SV *sv = NEWSV (0,2); 649 SV *sv = 0;
438 STRLEN cur = 0;
439 int utf8 = 0; 650 int utf8 = 0;
651 char *dec_cur = dec->cur;
440 652
441 for (;;) 653 do
442 { 654 {
443 unsigned char ch = *(unsigned char *)dec->cur; 655 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
656 char *cur = buf;
444 657
445 if (ch == '"') 658 do
446 break;
447 else if (ch == '\\')
448 { 659 {
449 switch (*++dec->cur) 660 unsigned char ch = *(unsigned char *)dec_cur++;
661
662 if (expect_false (ch == '"'))
450 { 663 {
451 case '\\': 664 --dec_cur;
452 case '/': 665 break;
453 case '"': APPEND_CH (*dec->cur++); break; 666 }
454 667 else if (expect_false (ch == '\\'))
455 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 668 {
456 case 't': APPEND_CH ('\011'); ++dec->cur; break; 669 switch (*dec_cur)
457 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
458 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
459 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
460
461 case 'u':
462 { 670 {
463 UV lo, hi; 671 case '\\':
464 ++dec->cur; 672 case '/':
673 case '"': *cur++ = *dec_cur++; break;
465 674
466 hi = decode_4hex (dec); 675 case 'b': ++dec_cur; *cur++ = '\010'; break;
467 if (hi == (UV)-1) 676 case 't': ++dec_cur; *cur++ = '\011'; break;
468 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;
469 680
470 // possibly a surrogate pair 681 case 'u':
471 if (hi >= 0xd800 && hi < 0xdc00)
472 { 682 {
473 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 683 UV lo, hi;
474 ERR ("illegal surrogate character"); 684 ++dec_cur;
475 685
476 dec->cur += 2; 686 dec->cur = dec_cur;
477
478 lo = decode_4hex (dec); 687 hi = decode_4hex (dec);
688 dec_cur = dec->cur;
479 if (lo == (UV)-1) 689 if (hi == (UV)-1)
480 goto fail; 690 goto fail;
481 691
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
482 if (lo < 0xdc00 || lo >= 0xe000) 707 if (lo < 0xdc00 || lo >= 0xe000)
483 ERR ("surrogate pair expected"); 708 ERR ("surrogate pair expected");
484 709
485 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;
486 } 723 }
487 else if (lo >= 0xdc00 && lo < 0xe000)
488 ERR ("illegal surrogate character");
489
490 if (hi >= 0x80)
491 { 724 break;
492 utf8 = 1;
493 725
494 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
495 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
496 }
497 else 726 default:
498 APPEND_CH (hi); 727 --dec_cur;
728 ERR ("illegal backslash escape sequence in string");
499 } 729 }
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;
737
738 --dec_cur;
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");
500 break; 756 else
757 ERR ("invalid character encountered while parsing JSON string");
501 } 758 }
502 } 759 }
503 else if (ch >= 0x20 && ch <= 0x7f) 760 while (cur < buf + SHORT_STRING_LEN);
504 APPEND_CH (*dec->cur++); 761
505 else if (ch >= 0x80) 762 {
763 STRLEN len = cur - buf;
764
765 if (sv)
506 { 766 {
507 STRLEN clen; 767 SvGROW (sv, SvCUR (sv) + len + 1);
508 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 768 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
509 if (clen < 0) 769 SvCUR_set (sv, SvCUR (sv) + len);
510 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
511
512 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits
513 memcpy (SvPVX (sv) + cur, dec->cur, clen);
514 dec->cur += clen;
515 } 770 }
516 else 771 else
517 ERR ("invalid character encountered"); 772 sv = newSVpvn (buf, len);
518 } 773 }
774 }
775 while (*dec_cur != '"');
519 776
520 ++dec->cur; 777 ++dec_cur;
521 778
779 if (sv)
780 {
522 SvPOK_only (sv); 781 SvPOK_only (sv);
523
524 SvCUR_set (sv, cur);
525 *SvEND (sv) = 0; 782 *SvEND (sv) = 0;
526 783
527 if (utf8) 784 if (utf8)
528 SvUTF8_on (sv); 785 SvUTF8_on (sv);
786 }
787 else
788 sv = newSVpvn ("", 0);
529 789
790 dec->cur = dec_cur;
530 return sv; 791 return sv;
531 792
532fail: 793fail:
533 SvREFCNT_dec (sv); 794 dec->cur = dec_cur;
534 return 0; 795 return 0;
535} 796}
536 797
537static SV * 798static SV *
538decode_num (dec_t *dec) 799decode_num (dec_t *dec)
… …
548 { 809 {
549 ++dec->cur; 810 ++dec->cur;
550 if (*dec->cur >= '0' && *dec->cur <= '9') 811 if (*dec->cur >= '0' && *dec->cur <= '9')
551 ERR ("malformed number (leading zero must not be followed by another digit)"); 812 ERR ("malformed number (leading zero must not be followed by another digit)");
552 } 813 }
553 814 else if (*dec->cur < '0' || *dec->cur > '9')
554 // int 815 ERR ("malformed number (no digits after initial minus)");
816 else
817 do
818 {
819 ++dec->cur;
820 }
555 while (*dec->cur >= '0' && *dec->cur <= '9') 821 while (*dec->cur >= '0' && *dec->cur <= '9');
556 ++dec->cur;
557 822
558 // [frac] 823 // [frac]
559 if (*dec->cur == '.') 824 if (*dec->cur == '.')
560 { 825 {
561 is_nv = 1; 826 ++dec->cur;
827
828 if (*dec->cur < '0' || *dec->cur > '9')
829 ERR ("malformed number (no digits after decimal point)");
562 830
563 do 831 do
564 { 832 {
565 ++dec->cur; 833 ++dec->cur;
566 } 834 }
567 while (*dec->cur >= '0' && *dec->cur <= '9'); 835 while (*dec->cur >= '0' && *dec->cur <= '9');
836
837 is_nv = 1;
568 } 838 }
569 839
570 // [exp] 840 // [exp]
571 if (*dec->cur == 'e' || *dec->cur == 'E') 841 if (*dec->cur == 'e' || *dec->cur == 'E')
572 { 842 {
573 is_nv = 1;
574
575 ++dec->cur; 843 ++dec->cur;
844
576 if (*dec->cur == '-' || *dec->cur == '+') 845 if (*dec->cur == '-' || *dec->cur == '+')
577 ++dec->cur; 846 ++dec->cur;
578 847
848 if (*dec->cur < '0' || *dec->cur > '9')
849 ERR ("malformed number (no digits after exp sign)");
850
851 do
852 {
853 ++dec->cur;
854 }
579 while (*dec->cur >= '0' && *dec->cur <= '9') 855 while (*dec->cur >= '0' && *dec->cur <= '9');
580 ++dec->cur; 856
857 is_nv = 1;
581 } 858 }
582 859
583 if (!is_nv) 860 if (!is_nv)
584 { 861 {
585 UV uv; 862 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
586 int numtype = grok_number (start, dec->cur - start, &uv); 863 if (*start == '-')
587 if (numtype & IS_NUMBER_IN_UV) 864 switch (dec->cur - start)
588 if (numtype & IS_NUMBER_NEG)
589 { 865 {
590 if (uv < (UV)IV_MIN) 866 case 2: return newSViv (-( start [1] - '0' ));
591 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));
592 } 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 }
593 else 889 else
594 return newSVuv (uv); 890 return newSVuv (uv);
891 }
595 } 892 }
596 893
597 return newSVnv (Atof (start)); 894 return newSVnv (Atof (start));
598 895
599fail: 896fail:
… …
603static SV * 900static SV *
604decode_av (dec_t *dec) 901decode_av (dec_t *dec)
605{ 902{
606 AV *av = newAV (); 903 AV *av = newAV ();
607 904
905 DEC_INC_DEPTH;
906 decode_ws (dec);
907
908 if (*dec->cur == ']')
909 ++dec->cur;
910 else
608 for (;;) 911 for (;;)
609 { 912 {
610 SV *value; 913 SV *value;
611 914
612 value = decode_sv (dec); 915 value = decode_sv (dec);
613 if (!value) 916 if (!value)
614 goto fail; 917 goto fail;
615 918
616 av_push (av, value); 919 av_push (av, value);
617 920
618 WS; 921 decode_ws (dec);
619 922
620 if (*dec->cur == ']') 923 if (*dec->cur == ']')
621 { 924 {
622 ++dec->cur; 925 ++dec->cur;
623 break; 926 break;
927 }
624 } 928
625
626 if (*dec->cur != ',') 929 if (*dec->cur != ',')
627 ERR (", or ] expected while parsing array"); 930 ERR (", or ] expected while parsing array");
628 931
629 ++dec->cur; 932 ++dec->cur;
630 } 933 }
631 934
935 DEC_DEC_DEPTH;
632 return newRV_noinc ((SV *)av); 936 return newRV_noinc ((SV *)av);
633 937
634fail: 938fail:
635 SvREFCNT_dec (av); 939 SvREFCNT_dec (av);
940 DEC_DEC_DEPTH;
636 return 0; 941 return 0;
637} 942}
638 943
639static SV * 944static SV *
640decode_hv (dec_t *dec) 945decode_hv (dec_t *dec)
641{ 946{
642 HV *hv = newHV (); 947 HV *hv = newHV ();
643 948
949 DEC_INC_DEPTH;
950 decode_ws (dec);
951
952 if (*dec->cur == '}')
953 ++dec->cur;
954 else
644 for (;;) 955 for (;;)
645 { 956 {
646 SV *key, *value; 957 SV *key, *value;
647 958
648 WS; EXPECT_CH ('"'); 959 decode_ws (dec); EXPECT_CH ('"');
649 960
650 key = decode_str (dec); 961 key = decode_str (dec);
651 if (!key) 962 if (!key)
652 goto fail;
653
654 WS; EXPECT_CH (':');
655
656 value = decode_sv (dec);
657 if (!value)
658 {
659 SvREFCNT_dec (key);
660 goto fail; 963 goto fail;
964
965 decode_ws (dec); EXPECT_CH (':');
966
967 value = decode_sv (dec);
968 if (!value)
969 {
970 SvREFCNT_dec (key);
971 goto fail;
661 } 972 }
662 973
663 //TODO: optimise
664 hv_store_ent (hv, key, value, 0); 974 hv_store_ent (hv, key, value, 0);
975 SvREFCNT_dec (key);
665 976
666 WS; 977 decode_ws (dec);
667 978
668 if (*dec->cur == '}') 979 if (*dec->cur == '}')
669 { 980 {
670 ++dec->cur; 981 ++dec->cur;
671 break; 982 break;
672 } 983 }
673 984
674 if (*dec->cur != ',') 985 if (*dec->cur != ',')
675 ERR (", or } expected while parsing object/hash"); 986 ERR (", or } expected while parsing object/hash");
676 987
677 ++dec->cur; 988 ++dec->cur;
678 } 989 }
679 990
991 DEC_DEC_DEPTH;
680 return newRV_noinc ((SV *)hv); 992 return newRV_noinc ((SV *)hv);
681 993
682fail: 994fail:
683 SvREFCNT_dec (hv); 995 SvREFCNT_dec (hv);
996 DEC_DEC_DEPTH;
684 return 0; 997 return 0;
685} 998}
686 999
687static SV * 1000static SV *
688decode_sv (dec_t *dec) 1001decode_sv (dec_t *dec)
689{ 1002{
690 WS; 1003 decode_ws (dec);
691 switch (*dec->cur) 1004 switch (*dec->cur)
692 { 1005 {
693 case '"': ++dec->cur; return decode_str (dec); 1006 case '"': ++dec->cur; return decode_str (dec);
694 case '[': ++dec->cur; return decode_av (dec); 1007 case '[': ++dec->cur; return decode_av (dec);
695 case '{': ++dec->cur; return decode_hv (dec); 1008 case '{': ++dec->cur; return decode_hv (dec);
… …
723 1036
724 case 'n': 1037 case 'n':
725 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1038 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
726 { 1039 {
727 dec->cur += 4; 1040 dec->cur += 4;
728 return newSViv (1); 1041 return newSVsv (&PL_sv_undef);
729 } 1042 }
730 else 1043 else
731 ERR ("'null' expected"); 1044 ERR ("'null' expected");
732 1045
733 break; 1046 break;
734 1047
735 default: 1048 default:
736 ERR ("malformed json string"); 1049 ERR ("malformed JSON string, neither array, object, number, string or atom");
737 break; 1050 break;
738 } 1051 }
739 1052
740fail: 1053fail:
741 return 0; 1054 return 0;
742} 1055}
743 1056
744static SV * 1057static SV *
745decode_json (SV *string, UV flags) 1058decode_json (SV *string, U32 flags, UV *offset_return)
746{ 1059{
1060 dec_t dec;
1061 UV offset;
747 SV *sv; 1062 SV *sv;
748 1063
1064 SvGETMAGIC (string);
1065 SvUPGRADE (string, SVt_PV);
1066
749 if (!(flags & F_UTF8)) 1067 if (flags & F_UTF8)
1068 sv_utf8_downgrade (string, 0);
1069 else
750 sv_utf8_upgrade (string); 1070 sv_utf8_upgrade (string);
751 1071
752 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1072 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
753 1073
754 dec_t dec;
755 dec.flags = flags; 1074 dec.flags = flags;
756 dec.cur = SvPVX (string); 1075 dec.cur = SvPVX (string);
757 dec.end = SvEND (string); 1076 dec.end = SvEND (string);
758 dec.err = 0; 1077 dec.err = 0;
1078 dec.depth = 0;
1079 dec.maxdepth = DEC_DEPTH (dec.flags);
759 1080
760 *dec.end = 1; // invalid anywhere 1081 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
761 sv = decode_sv (&dec); 1082 sv = decode_sv (&dec);
762 *dec.end = 0; 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 }
763 1106
764 if (!sv) 1107 if (!sv)
765 { 1108 {
766 IV offset = utf8_distance (dec.cur, SvPVX (string));
767 SV *uni = sv_newmortal (); 1109 SV *uni = sv_newmortal ();
768 1110
1111 // horrible hack to silence warning inside pv_uni_display
1112 COP cop = *PL_curcop;
1113 cop.cop_warnings = pWARN_NONE;
1114 ENTER;
1115 SAVEVPTR (PL_curcop);
1116 PL_curcop = &cop;
769 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
770 croak ("%s, at character %d (%s)", 1120 croak ("%s, at character offset %d [\"%s\"]",
771 dec.err, 1121 dec.err,
772 (int)offset, 1122 (int)offset,
773 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1123 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
774 } 1124 }
775 1125
776 sv_dump (sv);//D
777 return sv_2mortal (sv); 1126 sv = sv_2mortal (sv);
1127
1128 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
1129 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1130
1131 return sv;
778} 1132}
1133
1134/////////////////////////////////////////////////////////////////////////////
1135// XS interface functions
779 1136
780MODULE = JSON::XS PACKAGE = JSON::XS 1137MODULE = JSON::XS PACKAGE = JSON::XS
781 1138
782BOOT: 1139BOOT:
783{ 1140{
784 int i; 1141 int i;
785 1142
786 memset (decode_hexdigit, 0xff, 256); 1143 memset (decode_hexdigit, 0xff, 256);
1144
787 for (i = 10; i--; ) 1145 for (i = 0; i < 256; ++i)
788 decode_hexdigit ['0' + i] = i; 1146 decode_hexdigit [i] =
789 1147 i >= '0' && i <= '9' ? i - '0'
790 for (i = 6; --i; ) 1148 : i >= 'a' && i <= 'f' ? i - 'a' + 10
791 { 1149 : i >= 'A' && i <= 'F' ? i - 'A' + 10
792 decode_hexdigit ['a' + i] = 10 + i; 1150 : -1;
793 decode_hexdigit ['A' + i] = 10 + i;
794 }
795 1151
796 json_stash = gv_stashpv ("JSON::XS", 1); 1152 json_stash = gv_stashpv ("JSON::XS", 1);
797} 1153}
1154
1155PROTOTYPES: DISABLE
798 1156
799SV *new (char *dummy) 1157SV *new (char *dummy)
800 CODE: 1158 CODE:
801 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1159 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
802 OUTPUT: 1160 OUTPUT:
803 RETVAL 1161 RETVAL
804 1162
805SV *ascii (SV *self, int enable) 1163SV *ascii (SV *self, int enable = 1)
806 ALIAS: 1164 ALIAS:
807 ascii = F_ASCII 1165 ascii = F_ASCII
1166 latin1 = F_LATIN1
808 utf8 = F_UTF8 1167 utf8 = F_UTF8
809 indent = F_INDENT 1168 indent = F_INDENT
810 canonical = F_CANONICAL 1169 canonical = F_CANONICAL
811 space_before = F_SPACE_BEFORE 1170 space_before = F_SPACE_BEFORE
812 space_after = F_SPACE_AFTER 1171 space_after = F_SPACE_AFTER
813 json_rpc = F_JSON_RPC 1172 pretty = F_PRETTY
1173 allow_nonref = F_ALLOW_NONREF
1174 shrink = F_SHRINK
814 CODE: 1175 CODE:
815{ 1176{
816 UV *uv = SvJSON (self); 1177 UV *uv = SvJSON (self);
817 if (enable) 1178 if (enable)
818 *uv |= ix; 1179 *uv |= ix;
… …
822 RETVAL = newSVsv (self); 1183 RETVAL = newSVsv (self);
823} 1184}
824 OUTPUT: 1185 OUTPUT:
825 RETVAL 1186 RETVAL
826 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
827void encode (SV *self, SV *scalar) 1206void encode (SV *self, SV *scalar)
828 PPCODE: 1207 PPCODE:
829 XPUSHs (encode_json (scalar, *SvJSON (self))); 1208 XPUSHs (encode_json (scalar, *SvJSON (self)));
830 1209
831void decode (SV *self, SV *jsondata) 1210void decode (SV *self, SV *jsonstr)
832 PPCODE: 1211 PPCODE:
833 XPUSHs (decode_json (jsondata, *SvJSON (self))); 1212 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
834 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}
1222
1223PROTOTYPES: ENABLE
1224
1225void to_json (SV *scalar)
1226 ALIAS:
1227 objToJson = 0
1228 PPCODE:
1229 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1230
1231void from_json (SV *jsonstr)
1232 ALIAS:
1233 jsonToObj = 0
1234 PPCODE:
1235 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1236

Diff Legend

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