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 1.3 by root, Thu Mar 22 18:10:29 2007 UTC vs.
Revision 1.17 by root, Sun Mar 25 02:46:41 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? <=> { &__class__ => }
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 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)
64 { 99 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 100 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 101 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 102 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvEND (enc->sv); 103 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);
69 } 104 }
70} 105}
71 106
72static void 107static void
73encode_ch (enc_t *enc, char ch) 108encode_ch (enc_t *enc, char ch)
… …
79static void 114static void
80encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 115encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
81{ 116{
82 char *end = str + len; 117 char *end = str + len;
83 118
119 need (enc, len);
120
84 while (str < end) 121 while (str < end)
85 { 122 {
86 unsigned char ch = *(unsigned char *)str; 123 unsigned char ch = *(unsigned char *)str;
124
87 if (ch >= 0x20 && ch < 0x80) // most common case 125 if (ch >= 0x20 && ch < 0x80) // most common case
88 { 126 {
127 if (ch == '"') // but with slow exceptions
128 {
129 need (enc, len += 1);
130 *enc->cur++ = '\\';
131 *enc->cur++ = '"';
132 }
133 else if (ch == '\\')
134 {
135 need (enc, len += 1);
136 *enc->cur++ = '\\';
137 *enc->cur++ = '\\';
138 }
139 else
89 *enc->cur++ = ch; 140 *enc->cur++ = ch;
141
90 str++; 142 ++str;
91 } 143 }
92 else 144 else
93 { 145 {
94 STRLEN clen; 146 switch (ch)
95 UV uch;
96
97 if (is_utf8)
98 { 147 {
99 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;
100 if (clen < 0) 149 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
101 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;
102 } 151 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
103 else 152 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
104 {
105 uch = ch;
106 clen = 1;
107 }
108 153
109 need (enc, len += 6); 154 default:
110
111 if (uch < 0xa0 || enc->flags & F_ASCII)
112 {
113 if (uch > 0xFFFFUL)
114 { 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;
115 len += 6; 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 {
116 need (enc, len += 6); 179 need (enc, len += 11);
117 sprintf (enc->cur, "\\u%04x\\u%04x", 180 sprintf (enc->cur, "\\u%04x\\u%04x",
118 (uch - 0x10000) / 0x400 + 0xD800, 181 (int)((uch - 0x10000) / 0x400 + 0xD800),
119 (uch - 0x10000) % 0x400 + 0xDC00); 182 (int)((uch - 0x10000) % 0x400 + 0xDC00));
120 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;
121 } 198 }
199 else if (is_utf8)
200 {
201 need (enc, len += clen);
202 do
203 {
204 *enc->cur++ = *str++;
205 }
206 while (--clen);
207 }
122 else 208 else
123 { 209 {
124 sprintf (enc->cur, "\\u%04x", uch); 210 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
125 enc->cur += 6; 211 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
212 ++str;
213 }
126 } 214 }
127 } 215 }
128 else if (is_utf8)
129 {
130 memcpy (enc->cur, str, clen);
131 enc->cur += clen;
132 }
133 else
134 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
135
136 str += clen;
137 } 216 }
138 217
139 --len; 218 --len;
140 } 219 }
141} 220}
142 221
143#define INDENT SB \ 222static void
223encode_indent (enc_t *enc)
224{
144 if (enc->flags & F_INDENT) \ 225 if (enc->flags & F_INDENT)
145 { \ 226 {
146 int i_; \ 227 int spaces = enc->indent * INDENT_STEP;
147 need (enc, enc->indent); \
148 for (i_ = enc->indent * 3; i_--; )\
149 encode_ch (enc, ' '); \
150 } \
151 SE
152 228
153#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 229 need (enc, spaces);
154#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 230 memset (enc->cur, ' ', spaces);
155#define COMMA SB \ 231 enc->cur += spaces;
232 }
233}
234
235static void
236encode_space (enc_t *enc)
237{
238 need (enc, 1);
156 encode_ch (enc, ','); \ 239 encode_ch (enc, ' ');
240}
241
242static void
243encode_nl (enc_t *enc)
244{
157 if (enc->flags & F_INDENT) \ 245 if (enc->flags & F_INDENT)
158 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);
159 else if (enc->flags & F_SPACE_AFTER) \ 259 else if (enc->flags & F_SPACE_AFTER)
160 SPACE; \ 260 encode_space (enc);
161 SE 261}
162 262
163static void encode_sv (enc_t *enc, SV *sv); 263static void encode_sv (enc_t *enc, SV *sv);
164 264
165static void 265static void
166encode_av (enc_t *enc, AV *av) 266encode_av (enc_t *enc, AV *av)
167{ 267{
168 int i, len = av_len (av); 268 int i, len = av_len (av);
169 269
170 encode_ch (enc, '['); NL; 270 encode_ch (enc, '['); encode_nl (enc);
171 ++enc->indent; 271 ++enc->indent;
172 272
173 for (i = 0; i <= len; ++i) 273 for (i = 0; i <= len; ++i)
174 { 274 {
175 INDENT; 275 encode_indent (enc);
176 encode_sv (enc, *av_fetch (av, i, 0)); 276 encode_sv (enc, *av_fetch (av, i, 0));
177 277
178 if (i < len) 278 if (i < len)
179 COMMA; 279 encode_comma (enc);
180 } 280 }
181 281
182 NL; 282 encode_nl (enc);
183 283
184 --enc->indent; 284 --enc->indent;
185 INDENT; encode_ch (enc, ']'); 285 encode_indent (enc); encode_ch (enc, ']');
186} 286}
187 287
188static void 288static void
189encode_he (enc_t *enc, HE *he) 289encode_he (enc_t *enc, HE *he)
190{ 290{
… …
192 292
193 if (HeKLEN (he) == HEf_SVKEY) 293 if (HeKLEN (he) == HEf_SVKEY)
194 { 294 {
195 SV *sv = HeSVKEY (he); 295 SV *sv = HeSVKEY (he);
196 STRLEN len; 296 STRLEN len;
297 char *str;
298
299 SvGETMAGIC (sv);
197 char *str = SvPV (sv, len); 300 str = SvPV (sv, len);
198 301
199 encode_str (enc, str, len, SvUTF8 (sv)); 302 encode_str (enc, str, len, SvUTF8 (sv));
200 } 303 }
201 else 304 else
202 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 305 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
203 306
204 encode_ch (enc, '"'); 307 encode_ch (enc, '"');
205 308
206 if (enc->flags & F_SPACE_BEFORE) SPACE; 309 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
207 encode_ch (enc, ':'); 310 encode_ch (enc, ':');
208 if (enc->flags & F_SPACE_AFTER ) SPACE; 311 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
209 encode_sv (enc, HeVAL (he)); 312 encode_sv (enc, HeVAL (he));
210} 313}
211 314
212// compare hash entries, used when all keys are bytestrings 315// compare hash entries, used when all keys are bytestrings
213static int 316static int
… …
219 HE *b = *(HE **)b_; 322 HE *b = *(HE **)b_;
220 323
221 STRLEN la = HeKLEN (a); 324 STRLEN la = HeKLEN (a);
222 STRLEN lb = HeKLEN (b); 325 STRLEN lb = HeKLEN (b);
223 326
224 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 327 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
225 cmp = la < lb ? -1 : la == lb ? 0 : 1; 328 cmp = la - lb;
226 329
227 return cmp; 330 return cmp;
228} 331}
229 332
230// 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
… …
237static void 340static void
238encode_hv (enc_t *enc, HV *hv) 341encode_hv (enc_t *enc, HV *hv)
239{ 342{
240 int count, i; 343 int count, i;
241 344
242 encode_ch (enc, '{'); NL; ++enc->indent; 345 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
243 346
244 if ((count = hv_iterinit (hv))) 347 if ((count = hv_iterinit (hv)))
245 { 348 {
246 // for canonical output we have to sort by keys first 349 // for canonical output we have to sort by keys first
247 // actually, this is mostly due to the stupid so-called 350 // actually, this is mostly due to the stupid so-called
248 // security workaround added somewhere in 5.8.x. 351 // security workaround added somewhere in 5.8.x.
249 // that randomises hash orderings 352 // that randomises hash orderings
250 if (enc->flags & F_CANONICAL) 353 if (enc->flags & F_CANONICAL)
251 { 354 {
252 HE *he, *hes [count]; 355 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
253 int fast = 1; 356 int fast = 1;
254 357
255 i = 0; 358 i = 0;
256 while ((he = hv_iternext (hv))) 359 while ((he = hv_iternext (hv)))
257 { 360 {
… …
264 367
265 if (fast) 368 if (fast)
266 qsort (hes, count, sizeof (HE *), he_cmp_fast); 369 qsort (hes, count, sizeof (HE *), he_cmp_fast);
267 else 370 else
268 { 371 {
269 // hack to disable "use bytes" 372 // hack to forcefully disable "use bytes"
270 COP *oldcop = PL_curcop, cop; 373 COP cop = *PL_curcop;
271 cop.op_private = 0; 374 cop.op_private = 0;
375
376 ENTER;
377 SAVETMPS;
378
379 SAVEVPTR (PL_curcop);
272 PL_curcop = &cop; 380 PL_curcop = &cop;
273 381
274 SAVETMPS;
275 qsort (hes, count, sizeof (HE *), he_cmp_slow); 382 qsort (hes, count, sizeof (HE *), he_cmp_slow);
383
276 FREETMPS; 384 FREETMPS;
277 385 LEAVE;
278 PL_curcop = oldcop;
279 } 386 }
280 387
281 for (i = 0; i < count; ++i) 388 for (i = 0; i < count; ++i)
282 { 389 {
283 INDENT; 390 encode_indent (enc);
284 encode_he (enc, hes [i]); 391 encode_he (enc, hes [i]);
285 392
286 if (i < count - 1) 393 if (i < count - 1)
287 COMMA; 394 encode_comma (enc);
288 } 395 }
289 396
290 NL; 397 encode_nl (enc);
291 } 398 }
292 else 399 else
293 { 400 {
294 SV *sv; 401 SV *sv;
295 HE *he = hv_iternext (hv); 402 HE *he = hv_iternext (hv);
296 403
297 for (;;) 404 for (;;)
298 { 405 {
299 INDENT; 406 encode_indent (enc);
300 encode_he (enc, he); 407 encode_he (enc, he);
301 408
302 if (!(he = hv_iternext (hv))) 409 if (!(he = hv_iternext (hv)))
303 break; 410 break;
304 411
305 COMMA; 412 encode_comma (enc);
306 } 413 }
307 414
308 NL; 415 encode_nl (enc);
309 } 416 }
310 } 417 }
311 418
312 --enc->indent; INDENT; encode_ch (enc, '}'); 419 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
313} 420}
314 421
315static void 422static void
316encode_sv (enc_t *enc, SV *sv) 423encode_sv (enc_t *enc, SV *sv)
317{ 424{
425 SvGETMAGIC (sv);
426
318 if (SvPOKp (sv)) 427 if (SvPOKp (sv))
319 { 428 {
320 STRLEN len; 429 STRLEN len;
321 char *str = SvPV (sv, len); 430 char *str = SvPV (sv, len);
322 encode_ch (enc, '"'); 431 encode_ch (enc, '"');
… …
337 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 446 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
338 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 447 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
339 } 448 }
340 else if (SvROK (sv)) 449 else if (SvROK (sv))
341 { 450 {
342 if (!--enc->max_recurse) 451 SV *rv = SvRV (sv);
452
453 if (enc->indent >= enc->max_depth)
343 croak ("data structure too deep (hit recursion limit)"); 454 croak ("data structure too deep (hit recursion limit)");
344 455
345 sv = SvRV (sv);
346
347 switch (SvTYPE (sv)) 456 switch (SvTYPE (rv))
348 { 457 {
349 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 458 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
350 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 459 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
351 460
352 default: 461 default:
353 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));
354 } 464 }
355 } 465 }
356 else if (!SvOK (sv)) 466 else if (!SvOK (sv))
357 encode_str (enc, "null", 4, 0); 467 encode_str (enc, "null", 4, 0);
358 else 468 else
359 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));
360} 471}
361 472
362static SV * 473static SV *
363encode_json (SV *scalar, UV flags) 474encode_json (SV *scalar, UV flags)
364{ 475{
365 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 476 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
366 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)");
367 478
368 enc_t enc; 479 enc_t enc;
369 enc.flags = flags; 480 enc.flags = flags;
370 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 481 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
371 enc.cur = SvPVX (enc.sv); 482 enc.cur = SvPVX (enc.sv);
372 enc.end = SvEND (enc.sv); 483 enc.end = SvEND (enc.sv);
373 enc.max_recurse = 0;
374 enc.indent = 0; 484 enc.indent = 0;
485 enc.max_depth = 0x7fffffffUL;
375 486
376 SvPOK_only (enc.sv); 487 SvPOK_only (enc.sv);
377 encode_sv (&enc, scalar); 488 encode_sv (&enc, scalar);
378 489
379 if (!(flags & (F_ASCII | F_UTF8))) 490 if (!(flags & (F_ASCII | F_UTF8)))
380 SvUTF8_on (enc.sv); 491 SvUTF8_on (enc.sv);
381 492
382 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
383 return enc.sv; 498 return enc.sv;
384} 499}
385 500
386///////////////////////////////////////////////////////////////////////////// 501/////////////////////////////////////////////////////////////////////////////
502// decoder
387 503
388#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{
389 for (;;) \ 516 for (;;)
390 { \ 517 {
391 char ch = *dec->cur; \ 518 char ch = *dec->cur;
519
392 if (ch > 0x20 \ 520 if (ch > 0x20
393 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 521 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
394 break; \ 522 break;
523
395 ++dec->cur; \ 524 ++dec->cur;
396 } 525 }
526}
397 527
398#define ERR(reason) SB dec->err = reason; goto fail; SE 528#define ERR(reason) SB dec->err = reason; goto fail; SE
399#define EXPECT_CH(ch) SB \ 529#define EXPECT_CH(ch) SB \
400 if (*dec->cur != ch) \ 530 if (*dec->cur != ch) \
401 ERR (# ch " expected"); \ 531 ERR (# ch " expected"); \
402 ++dec->cur; \ 532 ++dec->cur; \
403 SE 533 SE
404 534
405static SV *decode_sv (dec_t *dec); 535static SV *decode_sv (dec_t *dec);
406 536
407#define APPEND_CH(ch) SB \
408 SvGROW (sv, cur + 1 + 1); \
409 SvPVX (sv)[cur++] = (ch); \
410 SE
411
412static signed char decode_hexdigit[256]; 537static signed char decode_hexdigit[256];
413 538
414static UV 539static UV
415decode_4hex (dec_t *dec) 540decode_4hex (dec_t *dec)
416{ 541{
417 signed char d1, d2, d3, d4; 542 signed char d1, d2, d3, d4;
543 unsigned char *cur = (unsigned char *)dec->cur;
418 544
419 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
420 if (d1 < 0) ERR ("four hexadecimal digits expected"); 545 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
421 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
422 if (d2 < 0) ERR ("four hexadecimal digits expected"); 546 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
423 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
424 if (d3 < 0) ERR ("four hexadecimal digits expected"); 547 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
425 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
426 if (d4 < 0) ERR ("four hexadecimal digits expected"); 548 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
427 549
428 dec->cur += 4; 550 dec->cur += 4;
429 551
430 return ((UV)d1) << 12 552 return ((UV)d1) << 12
431 | ((UV)d2) << 8 553 | ((UV)d2) << 8
… …
437} 559}
438 560
439static SV * 561static SV *
440decode_str (dec_t *dec) 562decode_str (dec_t *dec)
441{ 563{
442 SV *sv = NEWSV (0,2); 564 SV *sv = 0;
443 STRLEN cur = 0;
444 int utf8 = 0; 565 int utf8 = 0;
445 566
446 for (;;) 567 do
447 { 568 {
448 unsigned char ch = *(unsigned char *)dec->cur; 569 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
570 char *cur = buf;
449 571
450 if (ch == '"') 572 do
451 break;
452 else if (ch == '\\')
453 { 573 {
454 switch (*++dec->cur) 574 unsigned char ch = *(unsigned char *)dec->cur++;
575
576 if (ch == '"')
455 { 577 {
456 case '\\': 578 --dec->cur;
457 case '/': 579 break;
458 case '"': APPEND_CH (*dec->cur++); break; 580 }
459 581 else if (ch == '\\')
460 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 582 {
461 case 't': APPEND_CH ('\011'); ++dec->cur; break; 583 switch (*dec->cur)
462 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
463 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
464 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
465
466 case 'u':
467 { 584 {
468 UV lo, hi; 585 case '\\':
469 ++dec->cur; 586 case '/':
587 case '"': *cur++ = *dec->cur++; break;
470 588
471 hi = decode_4hex (dec); 589 case 'b': ++dec->cur; *cur++ = '\010'; break;
472 if (hi == (UV)-1) 590 case 't': ++dec->cur; *cur++ = '\011'; break;
473 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;
474 594
475 // possibly a surrogate pair 595 case 'u':
476 if (hi >= 0xd800 && hi < 0xdc00)
477 { 596 {
478 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 597 UV lo, hi;
479 ERR ("illegal surrogate character");
480
481 dec->cur += 2; 598 ++dec->cur;
482 599
483 lo = decode_4hex (dec); 600 hi = decode_4hex (dec);
484 if (lo == (UV)-1) 601 if (hi == (UV)-1)
485 goto fail; 602 goto fail;
486 603
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
487 if (lo < 0xdc00 || lo >= 0xe000) 617 if (lo < 0xdc00 || lo >= 0xe000)
488 ERR ("surrogate pair expected"); 618 ERR ("surrogate pair expected");
489 619
490 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;
491 } 633 }
492 else if (lo >= 0xdc00 && lo < 0xe000)
493 ERR ("illegal surrogate character");
494
495 if (hi >= 0x80)
496 { 634 break;
497 utf8 = 1;
498 635
499 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
500 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
501 }
502 else 636 default:
503 APPEND_CH (hi); 637 --dec->cur;
638 ERR ("illegal backslash escape sequence in string");
504 } 639 }
505 break;
506 } 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
507 } 665 }
508 else if (ch >= 0x20 && ch <= 0x7f) 666 while (cur < buf + SHORT_STRING_LEN);
509 APPEND_CH (*dec->cur++); 667
510 else if (ch >= 0x80) 668 STRLEN len = cur - buf;
669
670 if (sv)
511 { 671 {
512 STRLEN clen; 672 SvGROW (sv, SvCUR (sv) + len + 1);
513 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 673 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
514 if (clen < 0) 674 SvCUR_set (sv, SvCUR (sv) + len);
515 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
516
517 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits
518 memcpy (SvPVX (sv) + cur, dec->cur, clen);
519 dec->cur += clen;
520 } 675 }
521 else 676 else
522 ERR ("invalid character encountered"); 677 sv = newSVpvn (buf, len);
523 } 678 }
679 while (*dec->cur != '"');
524 680
525 ++dec->cur; 681 ++dec->cur;
526 682
683 if (sv)
684 {
527 SvPOK_only (sv); 685 SvPOK_only (sv);
528
529 SvCUR_set (sv, cur);
530 *SvEND (sv) = 0; 686 *SvEND (sv) = 0;
531 687
532 if (utf8) 688 if (utf8)
533 SvUTF8_on (sv); 689 SvUTF8_on (sv);
690 }
691 else
692 sv = newSVpvn ("", 0);
534 693
535 return sv; 694 return sv;
536 695
537fail: 696fail:
538 SvREFCNT_dec (sv);
539 return 0; 697 return 0;
540} 698}
541 699
542static SV * 700static SV *
543decode_num (dec_t *dec) 701decode_num (dec_t *dec)
… …
553 { 711 {
554 ++dec->cur; 712 ++dec->cur;
555 if (*dec->cur >= '0' && *dec->cur <= '9') 713 if (*dec->cur >= '0' && *dec->cur <= '9')
556 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)");
557 } 715 }
558 716 else if (*dec->cur < '0' || *dec->cur > '9')
559 // int 717 ERR ("malformed number (no digits after initial minus)");
718 else
719 do
720 {
721 ++dec->cur;
722 }
560 while (*dec->cur >= '0' && *dec->cur <= '9') 723 while (*dec->cur >= '0' && *dec->cur <= '9');
561 ++dec->cur;
562 724
563 // [frac] 725 // [frac]
564 if (*dec->cur == '.') 726 if (*dec->cur == '.')
565 { 727 {
566 is_nv = 1; 728 ++dec->cur;
729
730 if (*dec->cur < '0' || *dec->cur > '9')
731 ERR ("malformed number (no digits after decimal point)");
567 732
568 do 733 do
569 { 734 {
570 ++dec->cur; 735 ++dec->cur;
571 } 736 }
572 while (*dec->cur >= '0' && *dec->cur <= '9'); 737 while (*dec->cur >= '0' && *dec->cur <= '9');
738
739 is_nv = 1;
573 } 740 }
574 741
575 // [exp] 742 // [exp]
576 if (*dec->cur == 'e' || *dec->cur == 'E') 743 if (*dec->cur == 'e' || *dec->cur == 'E')
577 { 744 {
578 is_nv = 1;
579
580 ++dec->cur; 745 ++dec->cur;
746
581 if (*dec->cur == '-' || *dec->cur == '+') 747 if (*dec->cur == '-' || *dec->cur == '+')
582 ++dec->cur; 748 ++dec->cur;
583 749
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 }
584 while (*dec->cur >= '0' && *dec->cur <= '9') 757 while (*dec->cur >= '0' && *dec->cur <= '9');
585 ++dec->cur; 758
759 is_nv = 1;
586 } 760 }
587 761
588 if (!is_nv) 762 if (!is_nv)
589 { 763 {
590 UV uv; 764 UV uv;
… …
608static SV * 782static SV *
609decode_av (dec_t *dec) 783decode_av (dec_t *dec)
610{ 784{
611 AV *av = newAV (); 785 AV *av = newAV ();
612 786
787 decode_ws (dec);
788 if (*dec->cur == ']')
789 ++dec->cur;
790 else
613 for (;;) 791 for (;;)
614 { 792 {
615 SV *value; 793 SV *value;
616 794
617 value = decode_sv (dec); 795 value = decode_sv (dec);
618 if (!value) 796 if (!value)
619 goto fail; 797 goto fail;
620 798
621 av_push (av, value); 799 av_push (av, value);
622 800
623 WS; 801 decode_ws (dec);
624 802
625 if (*dec->cur == ']') 803 if (*dec->cur == ']')
626 { 804 {
627 ++dec->cur; 805 ++dec->cur;
628 break; 806 break;
807 }
629 } 808
630
631 if (*dec->cur != ',') 809 if (*dec->cur != ',')
632 ERR (", or ] expected while parsing array"); 810 ERR (", or ] expected while parsing array");
633 811
634 ++dec->cur; 812 ++dec->cur;
635 } 813 }
636 814
637 return newRV_noinc ((SV *)av); 815 return newRV_noinc ((SV *)av);
638 816
639fail: 817fail:
640 SvREFCNT_dec (av); 818 SvREFCNT_dec (av);
… …
644static SV * 822static SV *
645decode_hv (dec_t *dec) 823decode_hv (dec_t *dec)
646{ 824{
647 HV *hv = newHV (); 825 HV *hv = newHV ();
648 826
827 decode_ws (dec);
828 if (*dec->cur == '}')
829 ++dec->cur;
830 else
649 for (;;) 831 for (;;)
650 { 832 {
651 SV *key, *value; 833 SV *key, *value;
652 834
653 WS; EXPECT_CH ('"'); 835 decode_ws (dec); EXPECT_CH ('"');
654 836
655 key = decode_str (dec); 837 key = decode_str (dec);
656 if (!key) 838 if (!key)
657 goto fail;
658
659 WS; EXPECT_CH (':');
660
661 value = decode_sv (dec);
662 if (!value)
663 {
664 SvREFCNT_dec (key);
665 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;
666 } 848 }
667 849
668 //TODO: optimise 850 //TODO: optimise
669 hv_store_ent (hv, key, value, 0); 851 hv_store_ent (hv, key, value, 0);
670 852
671 WS; 853 decode_ws (dec);
672 854
673 if (*dec->cur == '}') 855 if (*dec->cur == '}')
674 { 856 {
675 ++dec->cur; 857 ++dec->cur;
676 break; 858 break;
677 } 859 }
678 860
679 if (*dec->cur != ',') 861 if (*dec->cur != ',')
680 ERR (", or } expected while parsing object/hash"); 862 ERR (", or } expected while parsing object/hash");
681 863
682 ++dec->cur; 864 ++dec->cur;
683 } 865 }
684 866
685 return newRV_noinc ((SV *)hv); 867 return newRV_noinc ((SV *)hv);
686 868
687fail: 869fail:
688 SvREFCNT_dec (hv); 870 SvREFCNT_dec (hv);
… …
690} 872}
691 873
692static SV * 874static SV *
693decode_sv (dec_t *dec) 875decode_sv (dec_t *dec)
694{ 876{
695 WS; 877 decode_ws (dec);
696 switch (*dec->cur) 878 switch (*dec->cur)
697 { 879 {
698 case '"': ++dec->cur; return decode_str (dec); 880 case '"': ++dec->cur; return decode_str (dec);
699 case '[': ++dec->cur; return decode_av (dec); 881 case '[': ++dec->cur; return decode_av (dec);
700 case '{': ++dec->cur; return decode_hv (dec); 882 case '{': ++dec->cur; return decode_hv (dec);
… …
728 910
729 case 'n': 911 case 'n':
730 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 912 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
731 { 913 {
732 dec->cur += 4; 914 dec->cur += 4;
733 return newSViv (1); 915 return newSVsv (&PL_sv_undef);
734 } 916 }
735 else 917 else
736 ERR ("'null' expected"); 918 ERR ("'null' expected");
737 919
738 break; 920 break;
739 921
740 default: 922 default:
741 ERR ("malformed json string"); 923 ERR ("malformed json string, neither array, object, number, string or atom");
742 break; 924 break;
743 } 925 }
744 926
745fail: 927fail:
746 return 0; 928 return 0;
… …
749static SV * 931static SV *
750decode_json (SV *string, UV flags) 932decode_json (SV *string, UV flags)
751{ 933{
752 SV *sv; 934 SV *sv;
753 935
754 if (!(flags & F_UTF8)) 936 if (flags & F_UTF8)
937 sv_utf8_downgrade (string, 0);
938 else
755 sv_utf8_upgrade (string); 939 sv_utf8_upgrade (string);
756 940
757 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 941 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
758 942
759 dec_t dec; 943 dec_t dec;
760 dec.flags = flags; 944 dec.flags = flags;
761 dec.cur = SvPVX (string); 945 dec.cur = SvPVX (string);
762 dec.end = SvEND (string); 946 dec.end = SvEND (string);
763 dec.err = 0; 947 dec.err = 0;
764 948
765 *dec.end = 1; // invalid anywhere
766 sv = decode_sv (&dec); 949 sv = decode_sv (&dec);
767 *dec.end = 0;
768 950
769 if (!sv) 951 if (!sv)
770 { 952 {
953 IV offset = dec.flags & F_UTF8
954 ? dec.cur - SvPVX (string)
771 IV offset = utf8_distance (dec.cur, SvPVX (string)); 955 : utf8_distance (dec.cur, SvPVX (string));
772 SV *uni = sv_newmortal (); 956 SV *uni = sv_newmortal ();
773 957
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;
774 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
775 croak ("%s, at character %d (%s)", 967 croak ("%s, at character offset %d (%s)",
776 dec.err, 968 dec.err,
777 (int)offset, 969 (int)offset,
778 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 970 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
779 } 971 }
780 972
781 sv = sv_2mortal (sv); 973 sv = sv_2mortal (sv);
782 974
783 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 975 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
784 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)");
785 977
786 return sv; 978 return sv;
787} 979}
980
981/////////////////////////////////////////////////////////////////////////////
982// XS interface functions
788 983
789MODULE = JSON::XS PACKAGE = JSON::XS 984MODULE = JSON::XS PACKAGE = JSON::XS
790 985
791BOOT: 986BOOT:
792{ 987{
… …
794 989
795 memset (decode_hexdigit, 0xff, 256); 990 memset (decode_hexdigit, 0xff, 256);
796 for (i = 10; i--; ) 991 for (i = 10; i--; )
797 decode_hexdigit ['0' + i] = i; 992 decode_hexdigit ['0' + i] = i;
798 993
799 for (i = 6; --i; ) 994 for (i = 7; i--; )
800 { 995 {
801 decode_hexdigit ['a' + i] = 10 + i; 996 decode_hexdigit ['a' + i] = 10 + i;
802 decode_hexdigit ['A' + i] = 10 + i; 997 decode_hexdigit ['A' + i] = 10 + i;
803 } 998 }
804 999
805 json_stash = gv_stashpv ("JSON::XS", 1); 1000 json_stash = gv_stashpv ("JSON::XS", 1);
806} 1001}
1002
1003PROTOTYPES: DISABLE
807 1004
808SV *new (char *dummy) 1005SV *new (char *dummy)
809 CODE: 1006 CODE:
810 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1007 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
811 OUTPUT: 1008 OUTPUT:
812 RETVAL 1009 RETVAL
813 1010
814SV *ascii (SV *self, int enable) 1011SV *ascii (SV *self, int enable = 1)
815 ALIAS: 1012 ALIAS:
816 ascii = F_ASCII 1013 ascii = F_ASCII
817 utf8 = F_UTF8 1014 utf8 = F_UTF8
818 indent = F_INDENT 1015 indent = F_INDENT
819 canonical = F_CANONICAL 1016 canonical = F_CANONICAL
820 space_before = F_SPACE_BEFORE 1017 space_before = F_SPACE_BEFORE
821 space_after = F_SPACE_AFTER 1018 space_after = F_SPACE_AFTER
822 json_rpc = F_JSON_RPC
823 pretty = F_PRETTY 1019 pretty = F_PRETTY
824 allow_nonref = F_ALLOW_NONREF 1020 allow_nonref = F_ALLOW_NONREF
1021 shrink = F_SHRINK
825 CODE: 1022 CODE:
826{ 1023{
827 UV *uv = SvJSON (self); 1024 UV *uv = SvJSON (self);
828 if (enable) 1025 if (enable)
829 *uv |= ix; 1026 *uv |= ix;
… …
841 1038
842void decode (SV *self, SV *jsonstr) 1039void decode (SV *self, SV *jsonstr)
843 PPCODE: 1040 PPCODE:
844 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1041 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
845 1042
1043PROTOTYPES: ENABLE
1044
846void to_json (SV *scalar) 1045void to_json (SV *scalar)
1046 ALIAS:
1047 objToJson = 0
847 PPCODE: 1048 PPCODE:
848 XPUSHs (encode_json (scalar, F_UTF8)); 1049 XPUSHs (encode_json (scalar, F_UTF8));
849 1050
850void from_json (SV *jsonstr) 1051void from_json (SV *jsonstr)
1052 ALIAS:
1053 jsonToObj = 0
851 PPCODE: 1054 PPCODE:
852 XPUSHs (decode_json (jsonstr, F_UTF8)); 1055 XPUSHs (decode_json (jsonstr, F_UTF8));
853 1056

Diff Legend

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