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 1.4 by root, Thu Mar 22 21:13:58 2007 UTC vs.
Revision 1.16 by root, Sun Mar 25 02:37:00 2007 UTC

… …
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 15#define F_ALLOW_NONREF 0x00000080
16#define F_SHRINK 0x00000100
17
18// F_SKIPINVALID?
19// F_EXECCODEREF?
20// F_SELFCONVERT?
17 21
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 22#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 23#define F_DEFAULT 0
20 24
21#define INIT_SIZE 32 // initial scalar size to be allocated 25#define INIT_SIZE 32 // initial scalar size to be allocated
26#define INDENT_STEP 3 // spaces per indentation level
27
28#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
29#define SHORT_STRING_LEN 512 // special-case strings of up to this size
22 30
23#define SB do { 31#define SB do {
24#define SE } while (0) 32#define SE } while (0)
25 33
26static HV *json_stash; 34static HV *json_stash; // JSON::XS::
35
36/////////////////////////////////////////////////////////////////////////////
37// utility functions
38
39static UV *
40SvJSON (SV *sv)
41{
42 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
43 croak ("object is not of type JSON::XS");
44
45 return &SvUVX (SvRV (sv));
46}
47
48static void
49shrink (SV *sv)
50{
51 sv_utf8_downgrade (sv, 1);
52 if (SvLEN (sv) > SvCUR (sv) + 1)
53 {
54#ifdef SvPV_shrink_to_cur
55 SvPV_shrink_to_cur (sv);
56#elif defined (SvPV_renew)
57 SvPV_renew (sv, SvCUR (sv) + 1);
58#endif
59 }
60}
61
62// decode an utf-8 character and return it, or (UV)-1 in
63// case of an error.
64// we special-case "safe" characters from U+80 .. U+7FF,
65// but use the very good perl function to parse anything else.
66// note that we never call this function for a ascii codepoints
67static UV
68decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
69{
70 if (s[0] > 0xdf || s[0] < 0xc2)
71 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
72 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
73 {
74 *clen = 2;
75 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
76 }
77 else
78 return (UV)-1;
79}
80
81/////////////////////////////////////////////////////////////////////////////
82// encoder
27 83
28// structure used for encoding JSON 84// structure used for encoding JSON
29typedef struct 85typedef struct
30{ 86{
31 char *cur; 87 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 88 char *end; // SvEND (sv)
34 SV *sv; 89 SV *sv; // result scalar
35 UV flags; 90 UV flags; // F_*
36 int max_recurse; 91 int indent; // indentation level
37 int indent; 92 int max_depth; // max. recursion level
38} enc_t; 93} enc_t;
39
40// structure used for decoding JSON
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58/////////////////////////////////////////////////////////////////////////////
59 94
60static void 95static void
61need (enc_t *enc, STRLEN len) 96need (enc_t *enc, STRLEN len)
62{ 97{
63 if (enc->cur + len >= enc->end) 98 if (enc->cur + len >= enc->end)
… …
85 120
86 while (str < end) 121 while (str < end)
87 { 122 {
88 unsigned char ch = *(unsigned char *)str; 123 unsigned char ch = *(unsigned char *)str;
89 124
90 if (ch == '"') 125 if (ch >= 0x20 && ch < 0x80) // most common case
91 { 126 {
127 if (ch == '"') // but with slow exceptions
128 {
92 need (enc, len += 1); 129 need (enc, len += 1);
93 *enc->cur++ = '\\'; 130 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 131 *enc->cur++ = '"';
95 ++str;
96 } 132 }
97 else if (ch == '\\') 133 else if (ch == '\\')
98 { 134 {
99 need (enc, len += 1); 135 need (enc, len += 1);
100 *enc->cur++ = '\\'; 136 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 137 *enc->cur++ = '\\';
102 ++str;
103 } 138 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 139 else
105 {
106 *enc->cur++ = ch; 140 *enc->cur++ = ch;
107 ++str; 141
108 }
109 else if (ch == '\015')
110 {
111 need (enc, len += 1);
112 *enc->cur++ = '\\';
113 *enc->cur++ = 'r';
114 ++str;
115 }
116 else if (ch == '\012')
117 {
118 need (enc, len += 1);
119 *enc->cur++ = '\\';
120 *enc->cur++ = 'n';
121 ++str; 142 ++str;
122 } 143 }
123 else 144 else
124 { 145 {
125 STRLEN clen; 146 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 { 147 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 148 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
131 if (clen < 0) 149 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
132 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 150 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
133 } 151 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
134 else 152 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
135 {
136 uch = ch;
137 clen = 1;
138 }
139 153
140 if (uch < 0x80 || enc->flags & F_ASCII) 154 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 155 {
156 STRLEN clen;
157 UV uch;
158
159 if (is_utf8)
160 {
161 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
162 uch = decode_utf8 (str, end - str, &clen);
163 if (clen == (STRLEN)-1)
164 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
165 }
166 else
167 {
168 uch = ch;
169 clen = 1;
170 }
171
172 if (uch > 0x10FFFFUL)
173 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
174
175 if (uch < 0x80 || enc->flags & F_ASCII)
176 {
177 if (uch > 0xFFFFUL)
178 {
144 need (enc, len += 11); 179 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 180 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 181 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 182 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 183 enc->cur += 12;
184 }
185 else
186 {
187 static char hexdigit [16] = "0123456789abcdef";
188 need (enc, len += 5);
189 *enc->cur++ = '\\';
190 *enc->cur++ = 'u';
191 *enc->cur++ = hexdigit [ uch >> 12 ];
192 *enc->cur++ = hexdigit [(uch >> 8) & 15];
193 *enc->cur++ = hexdigit [(uch >> 4) & 15];
194 *enc->cur++ = hexdigit [(uch >> 0) & 15];
195 }
196
197 str += clen;
198 }
199 else if (is_utf8)
200 {
201 need (enc, len += clen);
202 do
203 {
204 *enc->cur++ = *str++;
205 }
206 while (--clen);
207 }
208 else
209 {
210 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
211 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
212 ++str;
213 }
149 } 214 }
150 else
151 {
152 static char hexdigit [16] = "0123456789abcdef";
153 need (enc, len += 5);
154 *enc->cur++ = '\\';
155 *enc->cur++ = 'u';
156 *enc->cur++ = hexdigit [ uch >> 12 ];
157 *enc->cur++ = hexdigit [(uch >> 8) & 15];
158 *enc->cur++ = hexdigit [(uch >> 4) & 15];
159 *enc->cur++ = hexdigit [(uch >> 0) & 15];
160 }
161
162 str += clen;
163 }
164 else if (is_utf8)
165 {
166 need (enc, len += clen);
167 while (clen--)
168 *enc->cur++ = *str++;
169 }
170 else
171 {
172 need (enc, 10); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str;
175 } 215 }
176 } 216 }
177 217
178 --len; 218 --len;
179 } 219 }
180} 220}
181 221
182#define INDENT SB \ 222static void
223encode_indent (enc_t *enc)
224{
183 if (enc->flags & F_INDENT) \ 225 if (enc->flags & F_INDENT)
184 { \ 226 {
185 int i_; \ 227 int spaces = enc->indent * INDENT_STEP;
186 need (enc, enc->indent); \
187 for (i_ = enc->indent * 3; i_--; )\
188 encode_ch (enc, ' '); \
189 } \
190 SE
191 228
192#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 229 need (enc, spaces);
193#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 230 memset (enc->cur, ' ', spaces);
194#define COMMA SB \ 231 enc->cur += spaces;
232 }
233}
234
235static void
236encode_space (enc_t *enc)
237{
238 need (enc, 1);
195 encode_ch (enc, ','); \ 239 encode_ch (enc, ' ');
240}
241
242static void
243encode_nl (enc_t *enc)
244{
196 if (enc->flags & F_INDENT) \ 245 if (enc->flags & F_INDENT)
197 NL; \ 246 {
247 need (enc, 1);
248 encode_ch (enc, '\n');
249 }
250}
251
252static void
253encode_comma (enc_t *enc)
254{
255 encode_ch (enc, ',');
256
257 if (enc->flags & F_INDENT)
258 encode_nl (enc);
198 else if (enc->flags & F_SPACE_AFTER) \ 259 else if (enc->flags & F_SPACE_AFTER)
199 SPACE; \ 260 encode_space (enc);
200 SE 261}
201 262
202static void encode_sv (enc_t *enc, SV *sv); 263static void encode_sv (enc_t *enc, SV *sv);
203 264
204static void 265static void
205encode_av (enc_t *enc, AV *av) 266encode_av (enc_t *enc, AV *av)
206{ 267{
207 int i, len = av_len (av); 268 int i, len = av_len (av);
208 269
209 encode_ch (enc, '['); NL; 270 encode_ch (enc, '['); encode_nl (enc);
210 ++enc->indent; 271 ++enc->indent;
211 272
212 for (i = 0; i <= len; ++i) 273 for (i = 0; i <= len; ++i)
213 { 274 {
214 INDENT; 275 encode_indent (enc);
215 encode_sv (enc, *av_fetch (av, i, 0)); 276 encode_sv (enc, *av_fetch (av, i, 0));
216 277
217 if (i < len) 278 if (i < len)
218 COMMA; 279 encode_comma (enc);
219 } 280 }
220 281
221 NL; 282 encode_nl (enc);
222 283
223 --enc->indent; 284 --enc->indent;
224 INDENT; encode_ch (enc, ']'); 285 encode_indent (enc); encode_ch (enc, ']');
225} 286}
226 287
227static void 288static void
228encode_he (enc_t *enc, HE *he) 289encode_he (enc_t *enc, HE *he)
229{ 290{
… …
243 else 304 else
244 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 305 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
245 306
246 encode_ch (enc, '"'); 307 encode_ch (enc, '"');
247 308
248 if (enc->flags & F_SPACE_BEFORE) SPACE; 309 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
249 encode_ch (enc, ':'); 310 encode_ch (enc, ':');
250 if (enc->flags & F_SPACE_AFTER ) SPACE; 311 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
251 encode_sv (enc, HeVAL (he)); 312 encode_sv (enc, HeVAL (he));
252} 313}
253 314
254// compare hash entries, used when all keys are bytestrings 315// compare hash entries, used when all keys are bytestrings
255static int 316static int
… …
261 HE *b = *(HE **)b_; 322 HE *b = *(HE **)b_;
262 323
263 STRLEN la = HeKLEN (a); 324 STRLEN la = HeKLEN (a);
264 STRLEN lb = HeKLEN (b); 325 STRLEN lb = HeKLEN (b);
265 326
266 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 327 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
267 cmp = la < lb ? -1 : la == lb ? 0 : 1; 328 cmp = la - lb;
268 329
269 return cmp; 330 return cmp;
270} 331}
271 332
272// compare hash entries, used when some keys are sv's or utf-x 333// compare hash entries, used when some keys are sv's or utf-x
… …
279static void 340static void
280encode_hv (enc_t *enc, HV *hv) 341encode_hv (enc_t *enc, HV *hv)
281{ 342{
282 int count, i; 343 int count, i;
283 344
284 encode_ch (enc, '{'); NL; ++enc->indent; 345 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
285 346
286 if ((count = hv_iterinit (hv))) 347 if ((count = hv_iterinit (hv)))
287 { 348 {
288 // for canonical output we have to sort by keys first 349 // for canonical output we have to sort by keys first
289 // actually, this is mostly due to the stupid so-called 350 // actually, this is mostly due to the stupid so-called
290 // security workaround added somewhere in 5.8.x. 351 // security workaround added somewhere in 5.8.x.
291 // that randomises hash orderings 352 // that randomises hash orderings
292 if (enc->flags & F_CANONICAL) 353 if (enc->flags & F_CANONICAL)
293 { 354 {
294 HE *he, *hes [count]; 355 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
295 int fast = 1; 356 int fast = 1;
296 357
297 i = 0; 358 i = 0;
298 while ((he = hv_iternext (hv))) 359 while ((he = hv_iternext (hv)))
299 { 360 {
… …
306 367
307 if (fast) 368 if (fast)
308 qsort (hes, count, sizeof (HE *), he_cmp_fast); 369 qsort (hes, count, sizeof (HE *), he_cmp_fast);
309 else 370 else
310 { 371 {
311 // hack to disable "use bytes" 372 // hack to forcefully disable "use bytes"
312 COP *oldcop = PL_curcop, cop; 373 COP cop = *PL_curcop;
313 cop.op_private = 0; 374 cop.op_private = 0;
375
376 ENTER;
377 SAVETMPS;
378
379 SAVEVPTR (PL_curcop);
314 PL_curcop = &cop; 380 PL_curcop = &cop;
315 381
316 SAVETMPS;
317 qsort (hes, count, sizeof (HE *), he_cmp_slow); 382 qsort (hes, count, sizeof (HE *), he_cmp_slow);
383
318 FREETMPS; 384 FREETMPS;
319 385 LEAVE;
320 PL_curcop = oldcop;
321 } 386 }
322 387
323 for (i = 0; i < count; ++i) 388 for (i = 0; i < count; ++i)
324 { 389 {
325 INDENT; 390 encode_indent (enc);
326 encode_he (enc, hes [i]); 391 encode_he (enc, hes [i]);
327 392
328 if (i < count - 1) 393 if (i < count - 1)
329 COMMA; 394 encode_comma (enc);
330 } 395 }
331 396
332 NL; 397 encode_nl (enc);
333 } 398 }
334 else 399 else
335 { 400 {
336 SV *sv; 401 SV *sv;
337 HE *he = hv_iternext (hv); 402 HE *he = hv_iternext (hv);
338 403
339 for (;;) 404 for (;;)
340 { 405 {
341 INDENT; 406 encode_indent (enc);
342 encode_he (enc, he); 407 encode_he (enc, he);
343 408
344 if (!(he = hv_iternext (hv))) 409 if (!(he = hv_iternext (hv)))
345 break; 410 break;
346 411
347 COMMA; 412 encode_comma (enc);
348 } 413 }
349 414
350 NL; 415 encode_nl (enc);
351 } 416 }
352 } 417 }
353 418
354 --enc->indent; INDENT; encode_ch (enc, '}'); 419 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
355} 420}
356 421
357static void 422static void
358encode_sv (enc_t *enc, SV *sv) 423encode_sv (enc_t *enc, SV *sv)
359{ 424{
… …
381 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 446 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
382 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 447 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
383 } 448 }
384 else if (SvROK (sv)) 449 else if (SvROK (sv))
385 { 450 {
386 if (!--enc->max_recurse) 451 SV *rv = SvRV (sv);
452
453 if (enc->indent >= enc->max_depth)
387 croak ("data structure too deep (hit recursion limit)"); 454 croak ("data structure too deep (hit recursion limit)");
388 455
389 sv = SvRV (sv);
390
391 switch (SvTYPE (sv)) 456 switch (SvTYPE (rv))
392 { 457 {
393 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 458 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
394 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 459 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
395 460
396 default: 461 default:
397 croak ("JSON can only represent references to arrays or hashes"); 462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv));
398 } 464 }
399 } 465 }
400 else if (!SvOK (sv)) 466 else if (!SvOK (sv))
401 encode_str (enc, "null", 4, 0); 467 encode_str (enc, "null", 4, 0);
402 else 468 else
403 croak ("encountered perl type that JSON cannot handle"); 469 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
470 SvPV_nolen (sv), SvFLAGS (sv));
404} 471}
405 472
406static SV * 473static SV *
407encode_json (SV *scalar, UV flags) 474encode_json (SV *scalar, UV flags)
408{ 475{
409 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 476 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
410 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 477 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
411 478
412 enc_t enc; 479 enc_t enc;
413 enc.flags = flags; 480 enc.flags = flags;
414 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 481 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
415 enc.cur = SvPVX (enc.sv); 482 enc.cur = SvPVX (enc.sv);
416 enc.end = SvEND (enc.sv); 483 enc.end = SvEND (enc.sv);
417 enc.max_recurse = 0;
418 enc.indent = 0; 484 enc.indent = 0;
485 enc.max_depth = 0x7fffffffUL;
419 486
420 SvPOK_only (enc.sv); 487 SvPOK_only (enc.sv);
421 encode_sv (&enc, scalar); 488 encode_sv (&enc, scalar);
422 489
423 if (!(flags & (F_ASCII | F_UTF8))) 490 if (!(flags & (F_ASCII | F_UTF8)))
424 SvUTF8_on (enc.sv); 491 SvUTF8_on (enc.sv);
425 492
426 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 493 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
494
495 if (enc.flags & F_SHRINK)
496 shrink (enc.sv);
497
427 return enc.sv; 498 return enc.sv;
428} 499}
429 500
430///////////////////////////////////////////////////////////////////////////// 501/////////////////////////////////////////////////////////////////////////////
502// decoder
431 503
432#define WS \ 504// structure used for decoding JSON
505typedef struct
506{
507 char *cur; // current parser pointer
508 char *end; // end of input string
509 const char *err; // parse error, if != 0
510 UV flags; // F_*
511} dec_t;
512
513static void
514decode_ws (dec_t *dec)
515{
433 for (;;) \ 516 for (;;)
434 { \ 517 {
435 char ch = *dec->cur; \ 518 char ch = *dec->cur;
519
436 if (ch > 0x20 \ 520 if (ch > 0x20
437 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 521 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
438 break; \ 522 break;
523
439 ++dec->cur; \ 524 ++dec->cur;
440 } 525 }
526}
441 527
442#define ERR(reason) SB dec->err = reason; goto fail; SE 528#define ERR(reason) SB dec->err = reason; goto fail; SE
443#define EXPECT_CH(ch) SB \ 529#define EXPECT_CH(ch) SB \
444 if (*dec->cur != ch) \ 530 if (*dec->cur != ch) \
445 ERR (# ch " expected"); \ 531 ERR (# ch " expected"); \
… …
452 538
453static UV 539static UV
454decode_4hex (dec_t *dec) 540decode_4hex (dec_t *dec)
455{ 541{
456 signed char d1, d2, d3, d4; 542 signed char d1, d2, d3, d4;
543 unsigned char *cur = (unsigned char *)dec->cur;
457 544
458 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
459 if (d1 < 0) ERR ("four hexadecimal digits expected"); 545 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
460 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
461 if (d2 < 0) ERR ("four hexadecimal digits expected"); 546 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
462 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
463 if (d3 < 0) ERR ("four hexadecimal digits expected"); 547 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
464 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
465 if (d4 < 0) ERR ("four hexadecimal digits expected"); 548 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
466 549
467 dec->cur += 4; 550 dec->cur += 4;
468 551
469 return ((UV)d1) << 12 552 return ((UV)d1) << 12
470 | ((UV)d2) << 8 553 | ((UV)d2) << 8
… …
473 556
474fail: 557fail:
475 return (UV)-1; 558 return (UV)-1;
476} 559}
477 560
478#define APPEND_GROW(n) SB \
479 if (cur + (n) >= end) \
480 { \
481 STRLEN ofs = cur - SvPVX (sv); \
482 SvGROW (sv, ofs + (n) + 1); \
483 cur = SvPVX (sv) + ofs; \
484 end = SvEND (sv); \
485 } \
486 SE
487
488#define APPEND_CH(ch) SB \
489 APPEND_GROW (1); \
490 *cur++ = (ch); \
491 SE
492
493static SV * 561static SV *
494decode_str (dec_t *dec) 562decode_str (dec_t *dec)
495{ 563{
496 SV *sv = NEWSV (0,2); 564 SV *sv = 0;
497 int utf8 = 0; 565 int utf8 = 0;
498 char *cur = SvPVX (sv);
499 char *end = SvEND (sv);
500 566
501 for (;;) 567 do
502 { 568 {
503 unsigned char ch = *(unsigned char *)dec->cur; 569 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
570 char *cur = buf;
504 571
505 if (ch == '"') 572 do
506 break;
507 else if (ch == '\\')
508 { 573 {
509 switch (*++dec->cur) 574 unsigned char ch = *(unsigned char *)dec->cur++;
575
576 if (ch == '"')
510 { 577 {
511 case '\\': 578 --dec->cur;
512 case '/': 579 break;
513 case '"': APPEND_CH (*dec->cur++); break; 580 }
514 581 else if (ch == '\\')
515 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 582 {
516 case 't': APPEND_CH ('\011'); ++dec->cur; break; 583 switch (*dec->cur)
517 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
518 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
519 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
520
521 case 'u':
522 { 584 {
523 UV lo, hi; 585 case '\\':
524 ++dec->cur; 586 case '/':
587 case '"': *cur++ = *dec->cur++; break;
525 588
526 hi = decode_4hex (dec); 589 case 'b': ++dec->cur; *cur++ = '\010'; break;
527 if (hi == (UV)-1) 590 case 't': ++dec->cur; *cur++ = '\011'; break;
528 goto fail; 591 case 'n': ++dec->cur; *cur++ = '\012'; break;
592 case 'f': ++dec->cur; *cur++ = '\014'; break;
593 case 'r': ++dec->cur; *cur++ = '\015'; break;
529 594
530 // possibly a surrogate pair 595 case 'u':
531 if (hi >= 0xd800 && hi < 0xdc00)
532 { 596 {
533 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 597 UV lo, hi;
534 ERR ("illegal surrogate character");
535
536 dec->cur += 2; 598 ++dec->cur;
537 599
538 lo = decode_4hex (dec); 600 hi = decode_4hex (dec);
539 if (lo == (UV)-1) 601 if (hi == (UV)-1)
540 goto fail; 602 goto fail;
541 603
604 // possibly a surrogate pair
605 if (hi >= 0xd800)
606 if (hi < 0xdc00)
607 {
608 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
609 ERR ("missing low surrogate character in surrogate pair");
610
611 dec->cur += 2;
612
613 lo = decode_4hex (dec);
614 if (lo == (UV)-1)
615 goto fail;
616
542 if (lo < 0xdc00 || lo >= 0xe000) 617 if (lo < 0xdc00 || lo >= 0xe000)
543 ERR ("surrogate pair expected"); 618 ERR ("surrogate pair expected");
544 619
545 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 620 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
621 }
622 else if (hi < 0xe000)
623 ERR ("missing high surrogate character in surrogate pair");
624
625 if (hi >= 0x80)
626 {
627 utf8 = 1;
628
629 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
630 }
631 else
632 *cur++ = hi;
546 } 633 }
547 else if (lo >= 0xdc00 && lo < 0xe000)
548 ERR ("illegal surrogate character");
549
550 if (hi >= 0x80)
551 { 634 break;
552 utf8 = 1;
553 635
554 APPEND_GROW (4); // at most 4 bytes for 21 bits
555 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
556 }
557 else 636 default:
558 APPEND_CH (hi); 637 --dec->cur;
638 ERR ("illegal backslash escape sequence in string");
559 } 639 }
560 break;
561 } 640 }
641 else if (ch >= 0x20 && ch <= 0x7f)
642 *cur++ = ch;
643 else if (ch >= 0x80)
644 {
645 --dec->cur;
646
647 STRLEN clen;
648 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
649 if (clen == (STRLEN)-1)
650 ERR ("malformed UTF-8 character in JSON string");
651
652 do
653 {
654 *cur++ = *dec->cur++;
655 }
656 while (--clen);
657
658 utf8 = 1;
659 }
660 else if (!ch)
661 ERR ("unexpected end of string while parsing json string");
662 else
663 ERR ("invalid character encountered");
664
562 } 665 }
563 else if (ch >= 0x20 && ch <= 0x7f) 666 while (cur < buf + SHORT_STRING_LEN);
564 APPEND_CH (*dec->cur++); 667
565 else if (ch >= 0x80) 668 STRLEN len = cur - buf;
669
670 if (sv)
566 { 671 {
567 STRLEN clen; 672 SvGROW (sv, SvCUR (sv) + len + 1);
568 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 673 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
569 if (clen < 0) 674 SvCUR_set (sv, SvCUR (sv) + len);
570 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
571
572 APPEND_GROW (clen);
573 memcpy (cur, dec->cur, clen);
574 cur += clen;
575 dec->cur += clen;
576 } 675 }
577 else 676 else
578 ERR ("invalid character encountered"); 677 sv = newSVpvn (buf, len);
579 } 678 }
679 while (*dec->cur != '"');
580 680
581 ++dec->cur; 681 ++dec->cur;
582 682
583 SvCUR_set (sv, cur - SvPVX (sv)); 683 if (sv)
584 684 {
585 SvPOK_only (sv); 685 SvPOK_only (sv);
586 *SvEND (sv) = 0; 686 *SvEND (sv) = 0;
587 687
588 if (utf8) 688 if (utf8)
589 SvUTF8_on (sv); 689 SvUTF8_on (sv);
690 }
691 else
692 sv = newSVpvn ("", 0);
590 693
591 return sv; 694 return sv;
592 695
593fail: 696fail:
594 SvREFCNT_dec (sv);
595 return 0; 697 return 0;
596} 698}
597 699
598static SV * 700static SV *
599decode_num (dec_t *dec) 701decode_num (dec_t *dec)
… …
609 { 711 {
610 ++dec->cur; 712 ++dec->cur;
611 if (*dec->cur >= '0' && *dec->cur <= '9') 713 if (*dec->cur >= '0' && *dec->cur <= '9')
612 ERR ("malformed number (leading zero must not be followed by another digit)"); 714 ERR ("malformed number (leading zero must not be followed by another digit)");
613 } 715 }
614 716 else if (*dec->cur < '0' || *dec->cur > '9')
615 // int 717 ERR ("malformed number (no digits after initial minus)");
718 else
719 do
720 {
721 ++dec->cur;
722 }
616 while (*dec->cur >= '0' && *dec->cur <= '9') 723 while (*dec->cur >= '0' && *dec->cur <= '9');
617 ++dec->cur;
618 724
619 // [frac] 725 // [frac]
620 if (*dec->cur == '.') 726 if (*dec->cur == '.')
621 { 727 {
622 is_nv = 1; 728 ++dec->cur;
729
730 if (*dec->cur < '0' || *dec->cur > '9')
731 ERR ("malformed number (no digits after decimal point)");
623 732
624 do 733 do
625 { 734 {
626 ++dec->cur; 735 ++dec->cur;
627 } 736 }
628 while (*dec->cur >= '0' && *dec->cur <= '9'); 737 while (*dec->cur >= '0' && *dec->cur <= '9');
738
739 is_nv = 1;
629 } 740 }
630 741
631 // [exp] 742 // [exp]
632 if (*dec->cur == 'e' || *dec->cur == 'E') 743 if (*dec->cur == 'e' || *dec->cur == 'E')
633 { 744 {
634 is_nv = 1;
635
636 ++dec->cur; 745 ++dec->cur;
746
637 if (*dec->cur == '-' || *dec->cur == '+') 747 if (*dec->cur == '-' || *dec->cur == '+')
638 ++dec->cur; 748 ++dec->cur;
639 749
750 if (*dec->cur < '0' || *dec->cur > '9')
751 ERR ("malformed number (no digits after exp sign)");
752
753 do
754 {
755 ++dec->cur;
756 }
640 while (*dec->cur >= '0' && *dec->cur <= '9') 757 while (*dec->cur >= '0' && *dec->cur <= '9');
641 ++dec->cur; 758
759 is_nv = 1;
642 } 760 }
643 761
644 if (!is_nv) 762 if (!is_nv)
645 { 763 {
646 UV uv; 764 UV uv;
… …
664static SV * 782static SV *
665decode_av (dec_t *dec) 783decode_av (dec_t *dec)
666{ 784{
667 AV *av = newAV (); 785 AV *av = newAV ();
668 786
787 decode_ws (dec);
788 if (*dec->cur == ']')
789 ++dec->cur;
790 else
669 for (;;) 791 for (;;)
670 { 792 {
671 SV *value; 793 SV *value;
672 794
673 value = decode_sv (dec); 795 value = decode_sv (dec);
674 if (!value) 796 if (!value)
675 goto fail; 797 goto fail;
676 798
677 av_push (av, value); 799 av_push (av, value);
678 800
679 WS; 801 decode_ws (dec);
680 802
681 if (*dec->cur == ']') 803 if (*dec->cur == ']')
682 { 804 {
683 ++dec->cur; 805 ++dec->cur;
684 break; 806 break;
807 }
685 } 808
686
687 if (*dec->cur != ',') 809 if (*dec->cur != ',')
688 ERR (", or ] expected while parsing array"); 810 ERR (", or ] expected while parsing array");
689 811
690 ++dec->cur; 812 ++dec->cur;
691 } 813 }
692 814
693 return newRV_noinc ((SV *)av); 815 return newRV_noinc ((SV *)av);
694 816
695fail: 817fail:
696 SvREFCNT_dec (av); 818 SvREFCNT_dec (av);
… …
700static SV * 822static SV *
701decode_hv (dec_t *dec) 823decode_hv (dec_t *dec)
702{ 824{
703 HV *hv = newHV (); 825 HV *hv = newHV ();
704 826
827 decode_ws (dec);
828 if (*dec->cur == '}')
829 ++dec->cur;
830 else
705 for (;;) 831 for (;;)
706 { 832 {
707 SV *key, *value; 833 SV *key, *value;
708 834
709 WS; EXPECT_CH ('"'); 835 decode_ws (dec); EXPECT_CH ('"');
710 836
711 key = decode_str (dec); 837 key = decode_str (dec);
712 if (!key) 838 if (!key)
713 goto fail;
714
715 WS; EXPECT_CH (':');
716
717 value = decode_sv (dec);
718 if (!value)
719 {
720 SvREFCNT_dec (key);
721 goto fail; 839 goto fail;
840
841 decode_ws (dec); EXPECT_CH (':');
842
843 value = decode_sv (dec);
844 if (!value)
845 {
846 SvREFCNT_dec (key);
847 goto fail;
722 } 848 }
723 849
724 //TODO: optimise 850 //TODO: optimise
725 hv_store_ent (hv, key, value, 0); 851 hv_store_ent (hv, key, value, 0);
726 852
727 WS; 853 decode_ws (dec);
728 854
729 if (*dec->cur == '}') 855 if (*dec->cur == '}')
730 { 856 {
731 ++dec->cur; 857 ++dec->cur;
732 break; 858 break;
733 } 859 }
734 860
735 if (*dec->cur != ',') 861 if (*dec->cur != ',')
736 ERR (", or } expected while parsing object/hash"); 862 ERR (", or } expected while parsing object/hash");
737 863
738 ++dec->cur; 864 ++dec->cur;
739 } 865 }
740 866
741 return newRV_noinc ((SV *)hv); 867 return newRV_noinc ((SV *)hv);
742 868
743fail: 869fail:
744 SvREFCNT_dec (hv); 870 SvREFCNT_dec (hv);
… …
746} 872}
747 873
748static SV * 874static SV *
749decode_sv (dec_t *dec) 875decode_sv (dec_t *dec)
750{ 876{
751 WS; 877 decode_ws (dec);
752 switch (*dec->cur) 878 switch (*dec->cur)
753 { 879 {
754 case '"': ++dec->cur; return decode_str (dec); 880 case '"': ++dec->cur; return decode_str (dec);
755 case '[': ++dec->cur; return decode_av (dec); 881 case '[': ++dec->cur; return decode_av (dec);
756 case '{': ++dec->cur; return decode_hv (dec); 882 case '{': ++dec->cur; return decode_hv (dec);
… …
784 910
785 case 'n': 911 case 'n':
786 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 912 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
787 { 913 {
788 dec->cur += 4; 914 dec->cur += 4;
789 return newSViv (1); 915 return newSVsv (&PL_sv_undef);
790 } 916 }
791 else 917 else
792 ERR ("'null' expected"); 918 ERR ("'null' expected");
793 919
794 break; 920 break;
795 921
796 default: 922 default:
797 ERR ("malformed json string"); 923 ERR ("malformed json string, neither array, object, number, string or atom");
798 break; 924 break;
799 } 925 }
800 926
801fail: 927fail:
802 return 0; 928 return 0;
… …
805static SV * 931static SV *
806decode_json (SV *string, UV flags) 932decode_json (SV *string, UV flags)
807{ 933{
808 SV *sv; 934 SV *sv;
809 935
810 if (!(flags & F_UTF8)) 936 if (flags & F_UTF8)
937 sv_utf8_downgrade (string, 0);
938 else
811 sv_utf8_upgrade (string); 939 sv_utf8_upgrade (string);
812 940
813 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 941 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
814 942
815 dec_t dec; 943 dec_t dec;
816 dec.flags = flags; 944 dec.flags = flags;
817 dec.cur = SvPVX (string); 945 dec.cur = SvPVX (string);
818 dec.end = SvEND (string); 946 dec.end = SvEND (string);
819 dec.err = 0; 947 dec.err = 0;
820 948
821 *dec.end = 1; // invalid anywhere
822 sv = decode_sv (&dec); 949 sv = decode_sv (&dec);
823 *dec.end = 0;
824 950
825 if (!sv) 951 if (!sv)
826 { 952 {
953 IV offset = dec.flags & F_UTF8
954 ? dec.cur - SvPVX (string)
827 IV offset = utf8_distance (dec.cur, SvPVX (string)); 955 : utf8_distance (dec.cur, SvPVX (string));
828 SV *uni = sv_newmortal (); 956 SV *uni = sv_newmortal ();
829 957
958 // horrible hack to silence warning inside pv_uni_display
959 COP cop = *PL_curcop;
960 cop.cop_warnings = pWARN_NONE;
961 ENTER;
962 SAVEVPTR (PL_curcop);
963 PL_curcop = &cop;
830 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 964 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
965 LEAVE;
966
831 croak ("%s, at character %d (%s)", 967 croak ("%s, at character offset %d (%s)",
832 dec.err, 968 dec.err,
833 (int)offset, 969 (int)offset,
834 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 970 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
835 } 971 }
836 972
837 sv = sv_2mortal (sv); 973 sv = sv_2mortal (sv);
838 974
839 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 975 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
840 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 976 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
841 977
842 return sv; 978 return sv;
843} 979}
980
981/////////////////////////////////////////////////////////////////////////////
982// XS interface functions
844 983
845MODULE = JSON::XS PACKAGE = JSON::XS 984MODULE = JSON::XS PACKAGE = JSON::XS
846 985
847BOOT: 986BOOT:
848{ 987{
… …
867 CODE: 1006 CODE:
868 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1007 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
869 OUTPUT: 1008 OUTPUT:
870 RETVAL 1009 RETVAL
871 1010
872SV *ascii (SV *self, int enable) 1011SV *ascii (SV *self, int enable = 1)
873 ALIAS: 1012 ALIAS:
874 ascii = F_ASCII 1013 ascii = F_ASCII
875 utf8 = F_UTF8 1014 utf8 = F_UTF8
876 indent = F_INDENT 1015 indent = F_INDENT
877 canonical = F_CANONICAL 1016 canonical = F_CANONICAL
878 space_before = F_SPACE_BEFORE 1017 space_before = F_SPACE_BEFORE
879 space_after = F_SPACE_AFTER 1018 space_after = F_SPACE_AFTER
880 json_rpc = F_JSON_RPC
881 pretty = F_PRETTY 1019 pretty = F_PRETTY
882 allow_nonref = F_ALLOW_NONREF 1020 allow_nonref = F_ALLOW_NONREF
1021 shrink = F_SHRINK
883 CODE: 1022 CODE:
884{ 1023{
885 UV *uv = SvJSON (self); 1024 UV *uv = SvJSON (self);
886 if (enable) 1025 if (enable)
887 *uv |= ix; 1026 *uv |= ix;
… …
902 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1041 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
903 1042
904PROTOTYPES: ENABLE 1043PROTOTYPES: ENABLE
905 1044
906void to_json (SV *scalar) 1045void to_json (SV *scalar)
1046 ALIAS:
1047 objToJson = 0
907 PPCODE: 1048 PPCODE:
908 XPUSHs (encode_json (scalar, F_UTF8)); 1049 XPUSHs (encode_json (scalar, F_UTF8));
909 1050
910void from_json (SV *jsonstr) 1051void from_json (SV *jsonstr)
1052 ALIAS:
1053 jsonToObj = 0
911 PPCODE: 1054 PPCODE:
912 XPUSHs (decode_json (jsonstr, F_UTF8)); 1055 XPUSHs (decode_json (jsonstr, F_UTF8));
913 1056

Diff Legend

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