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 1.12 by root, Sat Mar 24 22:10:08 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 17
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 19#define F_DEFAULT 0
20 20
21#define INIT_SIZE 32 // initial scalar size to be allocated 21#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level
23
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#define SHORT_STRING_LEN 256 // special-case strings of up to this size
22 26
23#define SB do { 27#define SB do {
24#define SE } while (0) 28#define SE } while (0)
25 29
26static HV *json_stash; 30static HV *json_stash; // JSON::XS::
31
32/////////////////////////////////////////////////////////////////////////////
33// utility functions
34
35static UV *
36SvJSON (SV *sv)
37{
38 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
39 croak ("object is not of type JSON::XS");
40
41 return &SvUVX (SvRV (sv));
42}
43
44static void
45shrink (SV *sv)
46{
47 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1)
49 {
50#ifdef SvPV_shrink_to_cur
51 SvPV_shrink_to_cur (sv);
52#elif defined (SvPV_renew)
53 SvPV_renew (sv, SvCUR (sv) + 1);
54#endif
55 }
56}
57
58/////////////////////////////////////////////////////////////////////////////
59// encoder
27 60
28// structure used for encoding JSON 61// structure used for encoding JSON
29typedef struct 62typedef struct
30{ 63{
31 char *cur; 64 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 65 char *end; // SvEND (sv)
34 SV *sv; 66 SV *sv; // result scalar
35 UV flags; 67 UV flags; // F_*
36 int max_recurse; 68 int indent; // indentation level
37 int indent; 69 int max_depth; // max. recursion level
38} enc_t; 70} 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 71
60static void 72static void
61need (enc_t *enc, STRLEN len) 73need (enc_t *enc, STRLEN len)
62{ 74{
63 if (enc->cur + len >= enc->end) 75 if (enc->cur + len >= enc->end)
… …
85 97
86 while (str < end) 98 while (str < end)
87 { 99 {
88 unsigned char ch = *(unsigned char *)str; 100 unsigned char ch = *(unsigned char *)str;
89 101
90 if (ch == '"') 102 if (ch >= 0x20 && ch < 0x80) // most common case
91 { 103 {
104 if (ch == '"') // but with slow exceptions
105 {
92 need (enc, len += 1); 106 need (enc, len += 1);
93 *enc->cur++ = '\\'; 107 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 108 *enc->cur++ = '"';
95 ++str;
96 } 109 }
97 else if (ch == '\\') 110 else if (ch == '\\')
98 { 111 {
99 need (enc, len += 1); 112 need (enc, len += 1);
100 *enc->cur++ = '\\'; 113 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 114 *enc->cur++ = '\\';
102 ++str;
103 } 115 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 116 else
105 {
106 *enc->cur++ = ch; 117 *enc->cur++ = ch;
107 ++str; 118
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; 119 ++str;
122 } 120 }
123 else 121 else
124 { 122 {
125 STRLEN clen; 123 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 { 124 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 125 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
131 if (clen < 0) 126 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"); 127 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
133 } 128 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
134 else 129 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
135 {
136 uch = ch;
137 clen = 1;
138 }
139 130
140 if (uch < 0x80 || enc->flags & F_ASCII) 131 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 132 {
133 STRLEN clen;
134 UV uch;
135
136 if (is_utf8)
137 {
138 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
139 if (clen == (STRLEN)-1)
140 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
141 }
142 else
143 {
144 uch = ch;
145 clen = 1;
146 }
147
148 if (uch > 0x10FFFFUL)
149 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
150
151 if (uch < 0x80 || enc->flags & F_ASCII)
152 {
153 if (uch > 0xFFFFUL)
154 {
144 need (enc, len += 11); 155 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 156 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 157 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 158 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 159 enc->cur += 12;
160 }
161 else
162 {
163 static char hexdigit [16] = "0123456789abcdef";
164 need (enc, len += 5);
165 *enc->cur++ = '\\';
166 *enc->cur++ = 'u';
167 *enc->cur++ = hexdigit [ uch >> 12 ];
168 *enc->cur++ = hexdigit [(uch >> 8) & 15];
169 *enc->cur++ = hexdigit [(uch >> 4) & 15];
170 *enc->cur++ = hexdigit [(uch >> 0) & 15];
171 }
172
173 str += clen;
174 }
175 else if (is_utf8)
176 {
177 need (enc, len += clen);
178 do
179 {
180 *enc->cur++ = *str++;
181 }
182 while (--clen);
183 }
184 else
185 {
186 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
187 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
188 ++str;
189 }
149 } 190 }
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 } 191 }
176 } 192 }
177 193
178 --len; 194 --len;
179 } 195 }
180} 196}
181 197
182#define INDENT SB \ 198static void
199encode_indent (enc_t *enc)
200{
183 if (enc->flags & F_INDENT) \ 201 if (enc->flags & F_INDENT)
184 { \ 202 {
185 int i_; \ 203 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
191 204
192#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 205 need (enc, spaces);
193#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 206 memset (enc->cur, ' ', spaces);
194#define COMMA SB \ 207 enc->cur += spaces;
208 }
209}
210
211static void
212encode_space (enc_t *enc)
213{
214 need (enc, 1);
195 encode_ch (enc, ','); \ 215 encode_ch (enc, ' ');
216}
217
218static void
219encode_nl (enc_t *enc)
220{
196 if (enc->flags & F_INDENT) \ 221 if (enc->flags & F_INDENT)
197 NL; \ 222 {
223 need (enc, 1);
224 encode_ch (enc, '\n');
225 }
226}
227
228static void
229encode_comma (enc_t *enc)
230{
231 encode_ch (enc, ',');
232
233 if (enc->flags & F_INDENT)
234 encode_nl (enc);
198 else if (enc->flags & F_SPACE_AFTER) \ 235 else if (enc->flags & F_SPACE_AFTER)
199 SPACE; \ 236 encode_space (enc);
200 SE 237}
201 238
202static void encode_sv (enc_t *enc, SV *sv); 239static void encode_sv (enc_t *enc, SV *sv);
203 240
204static void 241static void
205encode_av (enc_t *enc, AV *av) 242encode_av (enc_t *enc, AV *av)
206{ 243{
207 int i, len = av_len (av); 244 int i, len = av_len (av);
208 245
209 encode_ch (enc, '['); NL; 246 encode_ch (enc, '['); encode_nl (enc);
210 ++enc->indent; 247 ++enc->indent;
211 248
212 for (i = 0; i <= len; ++i) 249 for (i = 0; i <= len; ++i)
213 { 250 {
214 INDENT; 251 encode_indent (enc);
215 encode_sv (enc, *av_fetch (av, i, 0)); 252 encode_sv (enc, *av_fetch (av, i, 0));
216 253
217 if (i < len) 254 if (i < len)
218 COMMA; 255 encode_comma (enc);
219 } 256 }
220 257
221 NL; 258 encode_nl (enc);
222 259
223 --enc->indent; 260 --enc->indent;
224 INDENT; encode_ch (enc, ']'); 261 encode_indent (enc); encode_ch (enc, ']');
225} 262}
226 263
227static void 264static void
228encode_he (enc_t *enc, HE *he) 265encode_he (enc_t *enc, HE *he)
229{ 266{
… …
243 else 280 else
244 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 281 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
245 282
246 encode_ch (enc, '"'); 283 encode_ch (enc, '"');
247 284
248 if (enc->flags & F_SPACE_BEFORE) SPACE; 285 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
249 encode_ch (enc, ':'); 286 encode_ch (enc, ':');
250 if (enc->flags & F_SPACE_AFTER ) SPACE; 287 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
251 encode_sv (enc, HeVAL (he)); 288 encode_sv (enc, HeVAL (he));
252} 289}
253 290
254// compare hash entries, used when all keys are bytestrings 291// compare hash entries, used when all keys are bytestrings
255static int 292static int
… …
261 HE *b = *(HE **)b_; 298 HE *b = *(HE **)b_;
262 299
263 STRLEN la = HeKLEN (a); 300 STRLEN la = HeKLEN (a);
264 STRLEN lb = HeKLEN (b); 301 STRLEN lb = HeKLEN (b);
265 302
266 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 303 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
267 cmp = la < lb ? -1 : la == lb ? 0 : 1; 304 cmp = la - lb;
268 305
269 return cmp; 306 return cmp;
270} 307}
271 308
272// compare hash entries, used when some keys are sv's or utf-x 309// compare hash entries, used when some keys are sv's or utf-x
… …
279static void 316static void
280encode_hv (enc_t *enc, HV *hv) 317encode_hv (enc_t *enc, HV *hv)
281{ 318{
282 int count, i; 319 int count, i;
283 320
284 encode_ch (enc, '{'); NL; ++enc->indent; 321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
285 322
286 if ((count = hv_iterinit (hv))) 323 if ((count = hv_iterinit (hv)))
287 { 324 {
288 // for canonical output we have to sort by keys first 325 // for canonical output we have to sort by keys first
289 // actually, this is mostly due to the stupid so-called 326 // actually, this is mostly due to the stupid so-called
290 // security workaround added somewhere in 5.8.x. 327 // security workaround added somewhere in 5.8.x.
291 // that randomises hash orderings 328 // that randomises hash orderings
292 if (enc->flags & F_CANONICAL) 329 if (enc->flags & F_CANONICAL)
293 { 330 {
294 HE *he, *hes [count]; 331 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
295 int fast = 1; 332 int fast = 1;
296 333
297 i = 0; 334 i = 0;
298 while ((he = hv_iternext (hv))) 335 while ((he = hv_iternext (hv)))
299 { 336 {
… …
306 343
307 if (fast) 344 if (fast)
308 qsort (hes, count, sizeof (HE *), he_cmp_fast); 345 qsort (hes, count, sizeof (HE *), he_cmp_fast);
309 else 346 else
310 { 347 {
311 // hack to disable "use bytes" 348 // hack to forcefully disable "use bytes"
312 COP *oldcop = PL_curcop, cop; 349 COP cop = *PL_curcop;
313 cop.op_private = 0; 350 cop.op_private = 0;
351
352 ENTER;
353 SAVETMPS;
354
355 SAVEVPTR (PL_curcop);
314 PL_curcop = &cop; 356 PL_curcop = &cop;
315 357
316 SAVETMPS;
317 qsort (hes, count, sizeof (HE *), he_cmp_slow); 358 qsort (hes, count, sizeof (HE *), he_cmp_slow);
359
318 FREETMPS; 360 FREETMPS;
319 361 LEAVE;
320 PL_curcop = oldcop;
321 } 362 }
322 363
323 for (i = 0; i < count; ++i) 364 for (i = 0; i < count; ++i)
324 { 365 {
325 INDENT; 366 encode_indent (enc);
326 encode_he (enc, hes [i]); 367 encode_he (enc, hes [i]);
327 368
328 if (i < count - 1) 369 if (i < count - 1)
329 COMMA; 370 encode_comma (enc);
330 } 371 }
331 372
332 NL; 373 encode_nl (enc);
333 } 374 }
334 else 375 else
335 { 376 {
336 SV *sv; 377 SV *sv;
337 HE *he = hv_iternext (hv); 378 HE *he = hv_iternext (hv);
338 379
339 for (;;) 380 for (;;)
340 { 381 {
341 INDENT; 382 encode_indent (enc);
342 encode_he (enc, he); 383 encode_he (enc, he);
343 384
344 if (!(he = hv_iternext (hv))) 385 if (!(he = hv_iternext (hv)))
345 break; 386 break;
346 387
347 COMMA; 388 encode_comma (enc);
348 } 389 }
349 390
350 NL; 391 encode_nl (enc);
351 } 392 }
352 } 393 }
353 394
354 --enc->indent; INDENT; encode_ch (enc, '}'); 395 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
355} 396}
356 397
357static void 398static void
358encode_sv (enc_t *enc, SV *sv) 399encode_sv (enc_t *enc, SV *sv)
359{ 400{
… …
381 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
382 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
383 } 424 }
384 else if (SvROK (sv)) 425 else if (SvROK (sv))
385 { 426 {
386 if (!--enc->max_recurse) 427 SV *rv = SvRV (sv);
428
429 if (enc->indent >= enc->max_depth)
387 croak ("data structure too deep (hit recursion limit)"); 430 croak ("data structure too deep (hit recursion limit)");
388 431
389 sv = SvRV (sv);
390
391 switch (SvTYPE (sv)) 432 switch (SvTYPE (rv))
392 { 433 {
393 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 434 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
394 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 435 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
395 436
396 default: 437 default:
397 croak ("JSON can only represent references to arrays or hashes"); 438 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
439 SvPV_nolen (sv));
398 } 440 }
399 } 441 }
400 else if (!SvOK (sv)) 442 else if (!SvOK (sv))
401 encode_str (enc, "null", 4, 0); 443 encode_str (enc, "null", 4, 0);
402 else 444 else
403 croak ("encountered perl type that JSON cannot handle"); 445 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
446 SvPV_nolen (sv), SvFLAGS (sv));
404} 447}
405 448
406static SV * 449static SV *
407encode_json (SV *scalar, UV flags) 450encode_json (SV *scalar, UV flags)
408{ 451{
409 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
410 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 453 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
411 454
412 enc_t enc; 455 enc_t enc;
413 enc.flags = flags; 456 enc.flags = flags;
414 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
415 enc.cur = SvPVX (enc.sv); 458 enc.cur = SvPVX (enc.sv);
416 enc.end = SvEND (enc.sv); 459 enc.end = SvEND (enc.sv);
417 enc.max_recurse = 0;
418 enc.indent = 0; 460 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL;
419 462
420 SvPOK_only (enc.sv); 463 SvPOK_only (enc.sv);
421 encode_sv (&enc, scalar); 464 encode_sv (&enc, scalar);
422 465
423 if (!(flags & (F_ASCII | F_UTF8))) 466 if (!(flags & (F_ASCII | F_UTF8)))
424 SvUTF8_on (enc.sv); 467 SvUTF8_on (enc.sv);
425 468
426 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 469 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
470
471 if (enc.flags & F_SHRINK)
472 shrink (enc.sv);
473
427 return enc.sv; 474 return enc.sv;
428} 475}
429 476
430///////////////////////////////////////////////////////////////////////////// 477/////////////////////////////////////////////////////////////////////////////
478// decoder
431 479
432#define WS \ 480// structure used for decoding JSON
481typedef struct
482{
483 char *cur; // current parser pointer
484 char *end; // end of input string
485 const char *err; // parse error, if != 0
486 UV flags; // F_*
487} dec_t;
488
489static void
490decode_ws (dec_t *dec)
491{
433 for (;;) \ 492 for (;;)
434 { \ 493 {
435 char ch = *dec->cur; \ 494 char ch = *dec->cur;
495
436 if (ch > 0x20 \ 496 if (ch > 0x20
437 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 497 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
438 break; \ 498 break;
499
439 ++dec->cur; \ 500 ++dec->cur;
440 } 501 }
502}
441 503
442#define ERR(reason) SB dec->err = reason; goto fail; SE 504#define ERR(reason) SB dec->err = reason; goto fail; SE
443#define EXPECT_CH(ch) SB \ 505#define EXPECT_CH(ch) SB \
444 if (*dec->cur != ch) \ 506 if (*dec->cur != ch) \
445 ERR (# ch " expected"); \ 507 ERR (# ch " expected"); \
… …
452 514
453static UV 515static UV
454decode_4hex (dec_t *dec) 516decode_4hex (dec_t *dec)
455{ 517{
456 signed char d1, d2, d3, d4; 518 signed char d1, d2, d3, d4;
519 unsigned char *cur = (unsigned char *)dec->cur;
457 520
458 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
459 if (d1 < 0) ERR ("four hexadecimal digits expected"); 521 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"); 522 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"); 523 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"); 524 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
466 525
467 dec->cur += 4; 526 dec->cur += 4;
468 527
469 return ((UV)d1) << 12 528 return ((UV)d1) << 12
470 | ((UV)d2) << 8 529 | ((UV)d2) << 8
… …
473 532
474fail: 533fail:
475 return (UV)-1; 534 return (UV)-1;
476} 535}
477 536
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 * 537static SV *
494decode_str (dec_t *dec) 538decode_str (dec_t *dec)
495{ 539{
496 SV *sv = NEWSV (0,2); 540 SV *sv = 0;
497 int utf8 = 0; 541 int utf8 = 0;
498 char *cur = SvPVX (sv);
499 char *end = SvEND (sv);
500 542
501 for (;;) 543 do
502 { 544 {
503 unsigned char ch = *(unsigned char *)dec->cur; 545 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
546 char *cur = buf;
504 547
505 if (ch == '"') 548 do
506 break;
507 else if (ch == '\\')
508 { 549 {
509 switch (*++dec->cur) 550 unsigned char ch = *(unsigned char *)dec->cur++;
551
552 if (ch == '"')
510 { 553 {
511 case '\\': 554 --dec->cur;
512 case '/': 555 break;
513 case '"': APPEND_CH (*dec->cur++); break; 556 }
514 557 else if (ch == '\\')
515 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 558 {
516 case 't': APPEND_CH ('\011'); ++dec->cur; break; 559 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 { 560 {
523 UV lo, hi; 561 case '\\':
524 ++dec->cur; 562 case '/':
563 case '"': *cur++ = *dec->cur++; break;
525 564
526 hi = decode_4hex (dec); 565 case 'b': ++dec->cur; *cur++ = '\010'; break;
527 if (hi == (UV)-1) 566 case 't': ++dec->cur; *cur++ = '\011'; break;
528 goto fail; 567 case 'n': ++dec->cur; *cur++ = '\012'; break;
568 case 'f': ++dec->cur; *cur++ = '\014'; break;
569 case 'r': ++dec->cur; *cur++ = '\015'; break;
529 570
530 // possibly a surrogate pair 571 case 'u':
531 if (hi >= 0xd800 && hi < 0xdc00)
532 { 572 {
533 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 573 UV lo, hi;
534 ERR ("illegal surrogate character");
535
536 dec->cur += 2; 574 ++dec->cur;
537 575
538 lo = decode_4hex (dec); 576 hi = decode_4hex (dec);
539 if (lo == (UV)-1) 577 if (hi == (UV)-1)
540 goto fail; 578 goto fail;
541 579
580 // possibly a surrogate pair
581 if (hi >= 0xd800)
582 if (hi < 0xdc00)
583 {
584 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
585 ERR ("missing low surrogate character in surrogate pair");
586
587 dec->cur += 2;
588
589 lo = decode_4hex (dec);
590 if (lo == (UV)-1)
591 goto fail;
592
542 if (lo < 0xdc00 || lo >= 0xe000) 593 if (lo < 0xdc00 || lo >= 0xe000)
543 ERR ("surrogate pair expected"); 594 ERR ("surrogate pair expected");
544 595
545 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 596 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
597 }
598 else if (hi < 0xe000)
599 ERR ("missing high surrogate character in surrogate pair");
600
601 if (hi >= 0x80)
602 {
603 utf8 = 1;
604
605 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
606 }
607 else
608 *cur++ = hi;
546 } 609 }
547 else if (lo >= 0xdc00 && lo < 0xe000)
548 ERR ("illegal surrogate character");
549
550 if (hi >= 0x80)
551 { 610 break;
552 utf8 = 1;
553 611
554 APPEND_GROW (4); // at most 4 bytes for 21 bits
555 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
556 }
557 else 612 default:
558 APPEND_CH (hi); 613 --dec->cur;
614 ERR ("illegal backslash escape sequence in string");
559 } 615 }
560 break;
561 } 616 }
617 else if (ch >= 0x20 && ch <= 0x7f)
618 *cur++ = ch;
619 else if (ch >= 0x80)
620 {
621 --dec->cur;
622
623 STRLEN clen;
624 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
625 if (clen == (STRLEN)-1)
626 ERR ("malformed UTF-8 character in JSON string");
627
628 do
629 {
630 *cur++ = *dec->cur++;
631 }
632 while (--clen);
633
634 utf8 = 1;
635 }
636 else if (!ch)
637 ERR ("unexpected end of string while parsing json string");
638 else
639 ERR ("invalid character encountered");
640
562 } 641 }
563 else if (ch >= 0x20 && ch <= 0x7f) 642 while (cur < buf + SHORT_STRING_LEN);
564 APPEND_CH (*dec->cur++); 643
565 else if (ch >= 0x80) 644 STRLEN len = cur - buf;
645
646 if (sv)
566 { 647 {
567 STRLEN clen; 648 SvGROW (sv, SvCUR (sv) + len + 1);
568 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 649 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
569 if (clen < 0) 650 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 } 651 }
577 else 652 else
578 ERR ("invalid character encountered"); 653 sv = newSVpvn (buf, len);
579 } 654 }
655 while (*dec->cur != '"');
580 656
581 ++dec->cur; 657 ++dec->cur;
582 658
583 SvCUR_set (sv, cur - SvPVX (sv)); 659 if (sv)
584 660 {
585 SvPOK_only (sv); 661 SvPOK_only (sv);
586 *SvEND (sv) = 0; 662 *SvEND (sv) = 0;
587 663
588 if (utf8) 664 if (utf8)
589 SvUTF8_on (sv); 665 SvUTF8_on (sv);
666 }
667 else
668 sv = newSVpvn ("", 0);
590 669
591 return sv; 670 return sv;
592 671
593fail: 672fail:
594 SvREFCNT_dec (sv);
595 return 0; 673 return 0;
596} 674}
597 675
598static SV * 676static SV *
599decode_num (dec_t *dec) 677decode_num (dec_t *dec)
… …
609 { 687 {
610 ++dec->cur; 688 ++dec->cur;
611 if (*dec->cur >= '0' && *dec->cur <= '9') 689 if (*dec->cur >= '0' && *dec->cur <= '9')
612 ERR ("malformed number (leading zero must not be followed by another digit)"); 690 ERR ("malformed number (leading zero must not be followed by another digit)");
613 } 691 }
614 692 else if (*dec->cur < '0' || *dec->cur > '9')
615 // int 693 ERR ("malformed number (no digits after initial minus)");
694 else
695 do
696 {
697 ++dec->cur;
698 }
616 while (*dec->cur >= '0' && *dec->cur <= '9') 699 while (*dec->cur >= '0' && *dec->cur <= '9');
617 ++dec->cur;
618 700
619 // [frac] 701 // [frac]
620 if (*dec->cur == '.') 702 if (*dec->cur == '.')
621 { 703 {
622 is_nv = 1; 704 ++dec->cur;
705
706 if (*dec->cur < '0' || *dec->cur > '9')
707 ERR ("malformed number (no digits after decimal point)");
623 708
624 do 709 do
625 { 710 {
626 ++dec->cur; 711 ++dec->cur;
627 } 712 }
628 while (*dec->cur >= '0' && *dec->cur <= '9'); 713 while (*dec->cur >= '0' && *dec->cur <= '9');
714
715 is_nv = 1;
629 } 716 }
630 717
631 // [exp] 718 // [exp]
632 if (*dec->cur == 'e' || *dec->cur == 'E') 719 if (*dec->cur == 'e' || *dec->cur == 'E')
633 { 720 {
634 is_nv = 1;
635
636 ++dec->cur; 721 ++dec->cur;
722
637 if (*dec->cur == '-' || *dec->cur == '+') 723 if (*dec->cur == '-' || *dec->cur == '+')
638 ++dec->cur; 724 ++dec->cur;
639 725
726 if (*dec->cur < '0' || *dec->cur > '9')
727 ERR ("malformed number (no digits after exp sign)");
728
729 do
730 {
731 ++dec->cur;
732 }
640 while (*dec->cur >= '0' && *dec->cur <= '9') 733 while (*dec->cur >= '0' && *dec->cur <= '9');
641 ++dec->cur; 734
735 is_nv = 1;
642 } 736 }
643 737
644 if (!is_nv) 738 if (!is_nv)
645 { 739 {
646 UV uv; 740 UV uv;
… …
664static SV * 758static SV *
665decode_av (dec_t *dec) 759decode_av (dec_t *dec)
666{ 760{
667 AV *av = newAV (); 761 AV *av = newAV ();
668 762
763 decode_ws (dec);
764 if (*dec->cur == ']')
765 ++dec->cur;
766 else
669 for (;;) 767 for (;;)
670 { 768 {
671 SV *value; 769 SV *value;
672 770
673 value = decode_sv (dec); 771 value = decode_sv (dec);
674 if (!value) 772 if (!value)
675 goto fail; 773 goto fail;
676 774
677 av_push (av, value); 775 av_push (av, value);
678 776
679 WS; 777 decode_ws (dec);
680 778
681 if (*dec->cur == ']') 779 if (*dec->cur == ']')
682 { 780 {
683 ++dec->cur; 781 ++dec->cur;
684 break; 782 break;
783 }
685 } 784
686
687 if (*dec->cur != ',') 785 if (*dec->cur != ',')
688 ERR (", or ] expected while parsing array"); 786 ERR (", or ] expected while parsing array");
689 787
690 ++dec->cur; 788 ++dec->cur;
691 } 789 }
692 790
693 return newRV_noinc ((SV *)av); 791 return newRV_noinc ((SV *)av);
694 792
695fail: 793fail:
696 SvREFCNT_dec (av); 794 SvREFCNT_dec (av);
… …
700static SV * 798static SV *
701decode_hv (dec_t *dec) 799decode_hv (dec_t *dec)
702{ 800{
703 HV *hv = newHV (); 801 HV *hv = newHV ();
704 802
803 decode_ws (dec);
804 if (*dec->cur == '}')
805 ++dec->cur;
806 else
705 for (;;) 807 for (;;)
706 { 808 {
707 SV *key, *value; 809 SV *key, *value;
708 810
709 WS; EXPECT_CH ('"'); 811 decode_ws (dec); EXPECT_CH ('"');
710 812
711 key = decode_str (dec); 813 key = decode_str (dec);
712 if (!key) 814 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; 815 goto fail;
816
817 decode_ws (dec); EXPECT_CH (':');
818
819 value = decode_sv (dec);
820 if (!value)
821 {
822 SvREFCNT_dec (key);
823 goto fail;
722 } 824 }
723 825
724 //TODO: optimise 826 //TODO: optimise
725 hv_store_ent (hv, key, value, 0); 827 hv_store_ent (hv, key, value, 0);
726 828
727 WS; 829 decode_ws (dec);
728 830
729 if (*dec->cur == '}') 831 if (*dec->cur == '}')
730 { 832 {
731 ++dec->cur; 833 ++dec->cur;
732 break; 834 break;
733 } 835 }
734 836
735 if (*dec->cur != ',') 837 if (*dec->cur != ',')
736 ERR (", or } expected while parsing object/hash"); 838 ERR (", or } expected while parsing object/hash");
737 839
738 ++dec->cur; 840 ++dec->cur;
739 } 841 }
740 842
741 return newRV_noinc ((SV *)hv); 843 return newRV_noinc ((SV *)hv);
742 844
743fail: 845fail:
744 SvREFCNT_dec (hv); 846 SvREFCNT_dec (hv);
… …
746} 848}
747 849
748static SV * 850static SV *
749decode_sv (dec_t *dec) 851decode_sv (dec_t *dec)
750{ 852{
751 WS; 853 decode_ws (dec);
752 switch (*dec->cur) 854 switch (*dec->cur)
753 { 855 {
754 case '"': ++dec->cur; return decode_str (dec); 856 case '"': ++dec->cur; return decode_str (dec);
755 case '[': ++dec->cur; return decode_av (dec); 857 case '[': ++dec->cur; return decode_av (dec);
756 case '{': ++dec->cur; return decode_hv (dec); 858 case '{': ++dec->cur; return decode_hv (dec);
… …
784 886
785 case 'n': 887 case 'n':
786 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 888 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
787 { 889 {
788 dec->cur += 4; 890 dec->cur += 4;
789 return newSViv (1); 891 return newSVsv (&PL_sv_undef);
790 } 892 }
791 else 893 else
792 ERR ("'null' expected"); 894 ERR ("'null' expected");
793 895
794 break; 896 break;
795 897
796 default: 898 default:
797 ERR ("malformed json string"); 899 ERR ("malformed json string, neither array, object, number, string or atom");
798 break; 900 break;
799 } 901 }
800 902
801fail: 903fail:
802 return 0; 904 return 0;
… …
805static SV * 907static SV *
806decode_json (SV *string, UV flags) 908decode_json (SV *string, UV flags)
807{ 909{
808 SV *sv; 910 SV *sv;
809 911
810 if (!(flags & F_UTF8)) 912 if (flags & F_UTF8)
913 sv_utf8_downgrade (string, 0);
914 else
811 sv_utf8_upgrade (string); 915 sv_utf8_upgrade (string);
812 916
813 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 917 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
814 918
815 dec_t dec; 919 dec_t dec;
816 dec.flags = flags; 920 dec.flags = flags;
817 dec.cur = SvPVX (string); 921 dec.cur = SvPVX (string);
818 dec.end = SvEND (string); 922 dec.end = SvEND (string);
819 dec.err = 0; 923 dec.err = 0;
820 924
821 *dec.end = 1; // invalid anywhere
822 sv = decode_sv (&dec); 925 sv = decode_sv (&dec);
823 *dec.end = 0;
824 926
825 if (!sv) 927 if (!sv)
826 { 928 {
929 IV offset = dec.flags & F_UTF8
930 ? dec.cur - SvPVX (string)
827 IV offset = utf8_distance (dec.cur, SvPVX (string)); 931 : utf8_distance (dec.cur, SvPVX (string));
828 SV *uni = sv_newmortal (); 932 SV *uni = sv_newmortal ();
829 933
934 // horrible hack to silence warning inside pv_uni_display
935 COP cop = *PL_curcop;
936 cop.cop_warnings = pWARN_NONE;
937 ENTER;
938 SAVEVPTR (PL_curcop);
939 PL_curcop = &cop;
830 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 940 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
941 LEAVE;
942
831 croak ("%s, at character %d (%s)", 943 croak ("%s, at character offset %d (%s)",
832 dec.err, 944 dec.err,
833 (int)offset, 945 (int)offset,
834 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 946 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
835 } 947 }
836 948
837 sv = sv_2mortal (sv); 949 sv = sv_2mortal (sv);
838 950
839 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 951 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)"); 952 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
841 953
842 return sv; 954 return sv;
843} 955}
956
957/////////////////////////////////////////////////////////////////////////////
958// XS interface functions
844 959
845MODULE = JSON::XS PACKAGE = JSON::XS 960MODULE = JSON::XS PACKAGE = JSON::XS
846 961
847BOOT: 962BOOT:
848{ 963{
… …
867 CODE: 982 CODE:
868 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 983 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
869 OUTPUT: 984 OUTPUT:
870 RETVAL 985 RETVAL
871 986
872SV *ascii (SV *self, int enable) 987SV *ascii (SV *self, int enable = 1)
873 ALIAS: 988 ALIAS:
874 ascii = F_ASCII 989 ascii = F_ASCII
875 utf8 = F_UTF8 990 utf8 = F_UTF8
876 indent = F_INDENT 991 indent = F_INDENT
877 canonical = F_CANONICAL 992 canonical = F_CANONICAL
878 space_before = F_SPACE_BEFORE 993 space_before = F_SPACE_BEFORE
879 space_after = F_SPACE_AFTER 994 space_after = F_SPACE_AFTER
880 json_rpc = F_JSON_RPC
881 pretty = F_PRETTY 995 pretty = F_PRETTY
882 allow_nonref = F_ALLOW_NONREF 996 allow_nonref = F_ALLOW_NONREF
997 shrink = F_SHRINK
883 CODE: 998 CODE:
884{ 999{
885 UV *uv = SvJSON (self); 1000 UV *uv = SvJSON (self);
886 if (enable) 1001 if (enable)
887 *uv |= ix; 1002 *uv |= ix;

Diff Legend

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