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 1.5 by root, Thu Mar 22 23:24:18 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 == (STRLEN)-1) 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 do
168 {
169 *enc->cur++ = *str++;
170 }
171 while (--clen);
172 }
173 else
174 {
175 need (enc, 10); // never more than 11 bytes needed
176 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
177 ++str;
178 } 191 }
179 } 192 }
180 193
181 --len; 194 --len;
182 } 195 }
183} 196}
184 197
185#define INDENT SB \ 198static void
199encode_indent (enc_t *enc)
200{
186 if (enc->flags & F_INDENT) \ 201 if (enc->flags & F_INDENT)
187 { \ 202 {
188 int i_; \ 203 int spaces = enc->indent * INDENT_STEP;
189 need (enc, enc->indent); \
190 for (i_ = enc->indent * 3; i_--; )\
191 encode_ch (enc, ' '); \
192 } \
193 SE
194 204
195#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 205 need (enc, spaces);
196#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 206 memset (enc->cur, ' ', spaces);
197#define COMMA SB \ 207 enc->cur += spaces;
208 }
209}
210
211static void
212encode_space (enc_t *enc)
213{
214 need (enc, 1);
198 encode_ch (enc, ','); \ 215 encode_ch (enc, ' ');
216}
217
218static void
219encode_nl (enc_t *enc)
220{
199 if (enc->flags & F_INDENT) \ 221 if (enc->flags & F_INDENT)
200 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);
201 else if (enc->flags & F_SPACE_AFTER) \ 235 else if (enc->flags & F_SPACE_AFTER)
202 SPACE; \ 236 encode_space (enc);
203 SE 237}
204 238
205static void encode_sv (enc_t *enc, SV *sv); 239static void encode_sv (enc_t *enc, SV *sv);
206 240
207static void 241static void
208encode_av (enc_t *enc, AV *av) 242encode_av (enc_t *enc, AV *av)
209{ 243{
210 int i, len = av_len (av); 244 int i, len = av_len (av);
211 245
212 encode_ch (enc, '['); NL; 246 encode_ch (enc, '['); encode_nl (enc);
213 ++enc->indent; 247 ++enc->indent;
214 248
215 for (i = 0; i <= len; ++i) 249 for (i = 0; i <= len; ++i)
216 { 250 {
217 INDENT; 251 encode_indent (enc);
218 encode_sv (enc, *av_fetch (av, i, 0)); 252 encode_sv (enc, *av_fetch (av, i, 0));
219 253
220 if (i < len) 254 if (i < len)
221 COMMA; 255 encode_comma (enc);
222 } 256 }
223 257
224 NL; 258 encode_nl (enc);
225 259
226 --enc->indent; 260 --enc->indent;
227 INDENT; encode_ch (enc, ']'); 261 encode_indent (enc); encode_ch (enc, ']');
228} 262}
229 263
230static void 264static void
231encode_he (enc_t *enc, HE *he) 265encode_he (enc_t *enc, HE *he)
232{ 266{
… …
246 else 280 else
247 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 281 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
248 282
249 encode_ch (enc, '"'); 283 encode_ch (enc, '"');
250 284
251 if (enc->flags & F_SPACE_BEFORE) SPACE; 285 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
252 encode_ch (enc, ':'); 286 encode_ch (enc, ':');
253 if (enc->flags & F_SPACE_AFTER ) SPACE; 287 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
254 encode_sv (enc, HeVAL (he)); 288 encode_sv (enc, HeVAL (he));
255} 289}
256 290
257// compare hash entries, used when all keys are bytestrings 291// compare hash entries, used when all keys are bytestrings
258static int 292static int
… …
264 HE *b = *(HE **)b_; 298 HE *b = *(HE **)b_;
265 299
266 STRLEN la = HeKLEN (a); 300 STRLEN la = HeKLEN (a);
267 STRLEN lb = HeKLEN (b); 301 STRLEN lb = HeKLEN (b);
268 302
269 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 303 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
270 cmp = la < lb ? -1 : la == lb ? 0 : 1; 304 cmp = la - lb;
271 305
272 return cmp; 306 return cmp;
273} 307}
274 308
275// 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
… …
282static void 316static void
283encode_hv (enc_t *enc, HV *hv) 317encode_hv (enc_t *enc, HV *hv)
284{ 318{
285 int count, i; 319 int count, i;
286 320
287 encode_ch (enc, '{'); NL; ++enc->indent; 321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
288 322
289 if ((count = hv_iterinit (hv))) 323 if ((count = hv_iterinit (hv)))
290 { 324 {
291 // for canonical output we have to sort by keys first 325 // for canonical output we have to sort by keys first
292 // actually, this is mostly due to the stupid so-called 326 // actually, this is mostly due to the stupid so-called
293 // security workaround added somewhere in 5.8.x. 327 // security workaround added somewhere in 5.8.x.
294 // that randomises hash orderings 328 // that randomises hash orderings
295 if (enc->flags & F_CANONICAL) 329 if (enc->flags & F_CANONICAL)
296 { 330 {
297 HE *he, *hes [count]; 331 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
298 int fast = 1; 332 int fast = 1;
299 333
300 i = 0; 334 i = 0;
301 while ((he = hv_iternext (hv))) 335 while ((he = hv_iternext (hv)))
302 { 336 {
… …
309 343
310 if (fast) 344 if (fast)
311 qsort (hes, count, sizeof (HE *), he_cmp_fast); 345 qsort (hes, count, sizeof (HE *), he_cmp_fast);
312 else 346 else
313 { 347 {
314 // hack to disable "use bytes" 348 // hack to forcefully disable "use bytes"
315 COP *oldcop = PL_curcop, cop; 349 COP cop = *PL_curcop;
316 cop.op_private = 0; 350 cop.op_private = 0;
351
352 ENTER;
353 SAVETMPS;
354
355 SAVEVPTR (PL_curcop);
317 PL_curcop = &cop; 356 PL_curcop = &cop;
318 357
319 SAVETMPS;
320 qsort (hes, count, sizeof (HE *), he_cmp_slow); 358 qsort (hes, count, sizeof (HE *), he_cmp_slow);
359
321 FREETMPS; 360 FREETMPS;
322 361 LEAVE;
323 PL_curcop = oldcop;
324 } 362 }
325 363
326 for (i = 0; i < count; ++i) 364 for (i = 0; i < count; ++i)
327 { 365 {
328 INDENT; 366 encode_indent (enc);
329 encode_he (enc, hes [i]); 367 encode_he (enc, hes [i]);
330 368
331 if (i < count - 1) 369 if (i < count - 1)
332 COMMA; 370 encode_comma (enc);
333 } 371 }
334 372
335 NL; 373 encode_nl (enc);
336 } 374 }
337 else 375 else
338 { 376 {
339 SV *sv; 377 SV *sv;
340 HE *he = hv_iternext (hv); 378 HE *he = hv_iternext (hv);
341 379
342 for (;;) 380 for (;;)
343 { 381 {
344 INDENT; 382 encode_indent (enc);
345 encode_he (enc, he); 383 encode_he (enc, he);
346 384
347 if (!(he = hv_iternext (hv))) 385 if (!(he = hv_iternext (hv)))
348 break; 386 break;
349 387
350 COMMA; 388 encode_comma (enc);
351 } 389 }
352 390
353 NL; 391 encode_nl (enc);
354 } 392 }
355 } 393 }
356 394
357 --enc->indent; INDENT; encode_ch (enc, '}'); 395 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
358} 396}
359 397
360static void 398static void
361encode_sv (enc_t *enc, SV *sv) 399encode_sv (enc_t *enc, SV *sv)
362{ 400{
… …
384 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
385 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
386 } 424 }
387 else if (SvROK (sv)) 425 else if (SvROK (sv))
388 { 426 {
389 if (!--enc->max_recurse) 427 SV *rv = SvRV (sv);
428
429 if (enc->indent >= enc->max_depth)
390 croak ("data structure too deep (hit recursion limit)"); 430 croak ("data structure too deep (hit recursion limit)");
391 431
392 sv = SvRV (sv);
393
394 switch (SvTYPE (sv)) 432 switch (SvTYPE (rv))
395 { 433 {
396 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 434 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
397 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 435 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
398 436
399 default: 437 default:
400 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));
401 } 440 }
402 } 441 }
403 else if (!SvOK (sv)) 442 else if (!SvOK (sv))
404 encode_str (enc, "null", 4, 0); 443 encode_str (enc, "null", 4, 0);
405 else 444 else
406 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));
407} 447}
408 448
409static SV * 449static SV *
410encode_json (SV *scalar, UV flags) 450encode_json (SV *scalar, UV flags)
411{ 451{
412 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
413 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)");
414 454
415 enc_t enc; 455 enc_t enc;
416 enc.flags = flags; 456 enc.flags = flags;
417 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
418 enc.cur = SvPVX (enc.sv); 458 enc.cur = SvPVX (enc.sv);
419 enc.end = SvEND (enc.sv); 459 enc.end = SvEND (enc.sv);
420 enc.max_recurse = 0;
421 enc.indent = 0; 460 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL;
422 462
423 SvPOK_only (enc.sv); 463 SvPOK_only (enc.sv);
424 encode_sv (&enc, scalar); 464 encode_sv (&enc, scalar);
425 465
426 if (!(flags & (F_ASCII | F_UTF8))) 466 if (!(flags & (F_ASCII | F_UTF8)))
427 SvUTF8_on (enc.sv); 467 SvUTF8_on (enc.sv);
428 468
429 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
430 return enc.sv; 474 return enc.sv;
431} 475}
432 476
433///////////////////////////////////////////////////////////////////////////// 477/////////////////////////////////////////////////////////////////////////////
478// decoder
434 479
435#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{
436 for (;;) \ 492 for (;;)
437 { \ 493 {
438 char ch = *dec->cur; \ 494 char ch = *dec->cur;
495
439 if (ch > 0x20 \ 496 if (ch > 0x20
440 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 497 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
441 break; \ 498 break;
499
442 ++dec->cur; \ 500 ++dec->cur;
443 } 501 }
502}
444 503
445#define ERR(reason) SB dec->err = reason; goto fail; SE 504#define ERR(reason) SB dec->err = reason; goto fail; SE
446#define EXPECT_CH(ch) SB \ 505#define EXPECT_CH(ch) SB \
447 if (*dec->cur != ch) \ 506 if (*dec->cur != ch) \
448 ERR (# ch " expected"); \ 507 ERR (# ch " expected"); \
… …
455 514
456static UV 515static UV
457decode_4hex (dec_t *dec) 516decode_4hex (dec_t *dec)
458{ 517{
459 signed char d1, d2, d3, d4; 518 signed char d1, d2, d3, d4;
519 unsigned char *cur = (unsigned char *)dec->cur;
460 520
461 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
462 if (d1 < 0) ERR ("four hexadecimal digits expected"); 521 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
463 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
464 if (d2 < 0) ERR ("four hexadecimal digits expected"); 522 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
465 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
466 if (d3 < 0) ERR ("four hexadecimal digits expected"); 523 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
467 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
468 if (d4 < 0) ERR ("four hexadecimal digits expected"); 524 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
469 525
470 dec->cur += 4; 526 dec->cur += 4;
471 527
472 return ((UV)d1) << 12 528 return ((UV)d1) << 12
473 | ((UV)d2) << 8 529 | ((UV)d2) << 8
… …
476 532
477fail: 533fail:
478 return (UV)-1; 534 return (UV)-1;
479} 535}
480 536
481#define APPEND_GROW(n) SB \
482 if (cur + (n) >= end) \
483 { \
484 STRLEN ofs = cur - SvPVX (sv); \
485 SvGROW (sv, ofs + (n) + 1); \
486 cur = SvPVX (sv) + ofs; \
487 end = SvEND (sv); \
488 } \
489 SE
490
491#define APPEND_CH(ch) SB \
492 APPEND_GROW (1); \
493 *cur++ = (ch); \
494 SE
495
496static SV * 537static SV *
497decode_str (dec_t *dec) 538decode_str (dec_t *dec)
498{ 539{
499 SV *sv = NEWSV (0,2); 540 SV *sv = 0;
500 int utf8 = 0; 541 int utf8 = 0;
501 char *cur = SvPVX (sv);
502 char *end = SvEND (sv);
503 542
504 for (;;) 543 do
505 { 544 {
506 unsigned char ch = *(unsigned char *)dec->cur; 545 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
546 char *cur = buf;
507 547
508 if (ch == '"') 548 do
509 break;
510 else if (ch == '\\')
511 { 549 {
512 switch (*++dec->cur) 550 unsigned char ch = *(unsigned char *)dec->cur++;
551
552 if (ch == '"')
513 { 553 {
514 case '\\': 554 --dec->cur;
515 case '/': 555 break;
516 case '"': APPEND_CH (*dec->cur++); break; 556 }
517 557 else if (ch == '\\')
518 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 558 {
519 case 't': APPEND_CH ('\011'); ++dec->cur; break; 559 switch (*dec->cur)
520 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
521 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
522 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
523
524 case 'u':
525 { 560 {
526 UV lo, hi; 561 case '\\':
527 ++dec->cur; 562 case '/':
563 case '"': *cur++ = *dec->cur++; break;
528 564
529 hi = decode_4hex (dec); 565 case 'b': ++dec->cur; *cur++ = '\010'; break;
530 if (hi == (UV)-1) 566 case 't': ++dec->cur; *cur++ = '\011'; break;
531 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;
532 570
533 // possibly a surrogate pair 571 case 'u':
534 if (hi >= 0xd800 && hi < 0xdc00)
535 { 572 {
536 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 573 UV lo, hi;
537 ERR ("missing low surrogate character in surrogate pair");
538
539 dec->cur += 2; 574 ++dec->cur;
540 575
541 lo = decode_4hex (dec); 576 hi = decode_4hex (dec);
542 if (lo == (UV)-1) 577 if (hi == (UV)-1)
543 goto fail; 578 goto fail;
544 579
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
545 if (lo < 0xdc00 || lo >= 0xe000) 593 if (lo < 0xdc00 || lo >= 0xe000)
546 ERR ("surrogate pair expected"); 594 ERR ("surrogate pair expected");
547 595
548 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;
549 } 609 }
550 else if (hi >= 0xdc00 && hi < 0xe000)
551 ERR ("missing high surrogate character in surrogate pair");
552
553 if (hi >= 0x80)
554 { 610 break;
555 utf8 = 1;
556 611
557 APPEND_GROW (4); // at most 4 bytes for 21 bits
558 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
559 }
560 else 612 default:
561 APPEND_CH (hi); 613 --dec->cur;
614 ERR ("illegal backslash escape sequence in string");
562 } 615 }
563 break;
564
565 default:
566 --dec->cur;
567 ERR ("illegal backslash escape sequence in string");
568 } 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
569 } 641 }
570 else if (ch >= 0x20 && ch <= 0x7f) 642 while (cur < buf + SHORT_STRING_LEN);
571 APPEND_CH (*dec->cur++); 643
572 else if (ch >= 0x80) 644 STRLEN len = cur - buf;
645
646 if (sv)
573 { 647 {
574 STRLEN clen; 648 SvGROW (sv, SvCUR (sv) + len + 1);
575 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 649 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
576 if (clen == (STRLEN)-1) 650 SvCUR_set (sv, SvCUR (sv) + len);
577 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
578
579 APPEND_GROW (clen);
580 do
581 {
582 *cur++ = *dec->cur++;
583 }
584 while (--clen);
585
586 utf8 = 1;
587 } 651 }
588 else if (dec->cur == dec->end)
589 ERR ("unexpected end of string while parsing json string");
590 else 652 else
591 ERR ("invalid character encountered"); 653 sv = newSVpvn (buf, len);
592 } 654 }
655 while (*dec->cur != '"');
593 656
594 ++dec->cur; 657 ++dec->cur;
595 658
596 SvCUR_set (sv, cur - SvPVX (sv)); 659 if (sv)
597 660 {
598 SvPOK_only (sv); 661 SvPOK_only (sv);
599 *SvEND (sv) = 0; 662 *SvEND (sv) = 0;
600 663
601 if (utf8) 664 if (utf8)
602 SvUTF8_on (sv); 665 SvUTF8_on (sv);
666 }
667 else
668 sv = newSVpvn ("", 0);
603 669
604 return sv; 670 return sv;
605 671
606fail: 672fail:
607 SvREFCNT_dec (sv);
608 return 0; 673 return 0;
609} 674}
610 675
611static SV * 676static SV *
612decode_num (dec_t *dec) 677decode_num (dec_t *dec)
… …
693static SV * 758static SV *
694decode_av (dec_t *dec) 759decode_av (dec_t *dec)
695{ 760{
696 AV *av = newAV (); 761 AV *av = newAV ();
697 762
698 WS; 763 decode_ws (dec);
699 if (*dec->cur == ']') 764 if (*dec->cur == ']')
700 ++dec->cur; 765 ++dec->cur;
701 else 766 else
702 for (;;) 767 for (;;)
703 { 768 {
… …
707 if (!value) 772 if (!value)
708 goto fail; 773 goto fail;
709 774
710 av_push (av, value); 775 av_push (av, value);
711 776
712 WS; 777 decode_ws (dec);
713 778
714 if (*dec->cur == ']') 779 if (*dec->cur == ']')
715 { 780 {
716 ++dec->cur; 781 ++dec->cur;
717 break; 782 break;
… …
733static SV * 798static SV *
734decode_hv (dec_t *dec) 799decode_hv (dec_t *dec)
735{ 800{
736 HV *hv = newHV (); 801 HV *hv = newHV ();
737 802
738 WS; 803 decode_ws (dec);
739 if (*dec->cur == '}') 804 if (*dec->cur == '}')
740 ++dec->cur; 805 ++dec->cur;
741 else 806 else
742 for (;;) 807 for (;;)
743 { 808 {
744 SV *key, *value; 809 SV *key, *value;
745 810
746 WS; EXPECT_CH ('"'); 811 decode_ws (dec); EXPECT_CH ('"');
747 812
748 key = decode_str (dec); 813 key = decode_str (dec);
749 if (!key) 814 if (!key)
750 goto fail; 815 goto fail;
751 816
752 WS; EXPECT_CH (':'); 817 decode_ws (dec); EXPECT_CH (':');
753 818
754 value = decode_sv (dec); 819 value = decode_sv (dec);
755 if (!value) 820 if (!value)
756 { 821 {
757 SvREFCNT_dec (key); 822 SvREFCNT_dec (key);
… …
759 } 824 }
760 825
761 //TODO: optimise 826 //TODO: optimise
762 hv_store_ent (hv, key, value, 0); 827 hv_store_ent (hv, key, value, 0);
763 828
764 WS; 829 decode_ws (dec);
765 830
766 if (*dec->cur == '}') 831 if (*dec->cur == '}')
767 { 832 {
768 ++dec->cur; 833 ++dec->cur;
769 break; 834 break;
… …
783} 848}
784 849
785static SV * 850static SV *
786decode_sv (dec_t *dec) 851decode_sv (dec_t *dec)
787{ 852{
788 WS; 853 decode_ws (dec);
789 switch (*dec->cur) 854 switch (*dec->cur)
790 { 855 {
791 case '"': ++dec->cur; return decode_str (dec); 856 case '"': ++dec->cur; return decode_str (dec);
792 case '[': ++dec->cur; return decode_av (dec); 857 case '[': ++dec->cur; return decode_av (dec);
793 case '{': ++dec->cur; return decode_hv (dec); 858 case '{': ++dec->cur; return decode_hv (dec);
… …
829 ERR ("'null' expected"); 894 ERR ("'null' expected");
830 895
831 break; 896 break;
832 897
833 default: 898 default:
834 ERR ("malformed json string"); 899 ERR ("malformed json string, neither array, object, number, string or atom");
835 break; 900 break;
836 } 901 }
837 902
838fail: 903fail:
839 return 0; 904 return 0;
… …
859 924
860 sv = decode_sv (&dec); 925 sv = decode_sv (&dec);
861 926
862 if (!sv) 927 if (!sv)
863 { 928 {
929 IV offset = dec.flags & F_UTF8
930 ? dec.cur - SvPVX (string)
864 IV offset = utf8_distance (dec.cur, SvPVX (string)); 931 : utf8_distance (dec.cur, SvPVX (string));
865 SV *uni = sv_newmortal (); 932 SV *uni = sv_newmortal ();
933
866 // horrible hack to silence warning inside pv_uni_display 934 // horrible hack to silence warning inside pv_uni_display
867 COP cop; 935 COP cop = *PL_curcop;
868 memset (&cop, 0, sizeof (cop));
869 cop.cop_warnings = pWARN_NONE; 936 cop.cop_warnings = pWARN_NONE;
937 ENTER;
870 SAVEVPTR (PL_curcop); 938 SAVEVPTR (PL_curcop);
871 PL_curcop = &cop; 939 PL_curcop = &cop;
872
873 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
874 croak ("%s, at character offset %d (%s)", 943 croak ("%s, at character offset %d (%s)",
875 dec.err, 944 dec.err,
876 (int)offset, 945 (int)offset,
877 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 946 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
878 } 947 }
879 948
880 sv = sv_2mortal (sv); 949 sv = sv_2mortal (sv);
881 950
882 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 951 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
883 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)");
884 953
885 return sv; 954 return sv;
886} 955}
956
957/////////////////////////////////////////////////////////////////////////////
958// XS interface functions
887 959
888MODULE = JSON::XS PACKAGE = JSON::XS 960MODULE = JSON::XS PACKAGE = JSON::XS
889 961
890BOOT: 962BOOT:
891{ 963{
… …
910 CODE: 982 CODE:
911 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 983 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
912 OUTPUT: 984 OUTPUT:
913 RETVAL 985 RETVAL
914 986
915SV *ascii (SV *self, int enable) 987SV *ascii (SV *self, int enable = 1)
916 ALIAS: 988 ALIAS:
917 ascii = F_ASCII 989 ascii = F_ASCII
918 utf8 = F_UTF8 990 utf8 = F_UTF8
919 indent = F_INDENT 991 indent = F_INDENT
920 canonical = F_CANONICAL 992 canonical = F_CANONICAL
921 space_before = F_SPACE_BEFORE 993 space_before = F_SPACE_BEFORE
922 space_after = F_SPACE_AFTER 994 space_after = F_SPACE_AFTER
923 json_rpc = F_JSON_RPC
924 pretty = F_PRETTY 995 pretty = F_PRETTY
925 allow_nonref = F_ALLOW_NONREF 996 allow_nonref = F_ALLOW_NONREF
997 shrink = F_SHRINK
926 CODE: 998 CODE:
927{ 999{
928 UV *uv = SvJSON (self); 1000 UV *uv = SvJSON (self);
929 if (enable) 1001 if (enable)
930 *uv |= ix; 1002 *uv |= ix;

Diff Legend

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