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 1.13 by root, Sat Mar 24 22:55:16 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// decode an utf-8 character and return it, or (UV)-1 in
59// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints
63static UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{
66 if (s[0] > 0xdf || s[0] < 0xc2)
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69 {
70 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 }
73 else
74 return (UV)-1;
75}
76
77/////////////////////////////////////////////////////////////////////////////
78// encoder
27 79
28// structure used for encoding JSON 80// structure used for encoding JSON
29typedef struct 81typedef struct
30{ 82{
31 char *cur; 83 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 84 char *end; // SvEND (sv)
34 SV *sv; 85 SV *sv; // result scalar
35 UV flags; 86 UV flags; // F_*
36 int max_recurse; 87 int indent; // indentation level
37 int indent; 88 int max_depth; // max. recursion level
38} enc_t; 89} 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 90
60static void 91static void
61need (enc_t *enc, STRLEN len) 92need (enc_t *enc, STRLEN len)
62{ 93{
63 if (enc->cur + len >= enc->end) 94 if (enc->cur + len >= enc->end)
64 { 95 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 96 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 97 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 98 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvEND (enc->sv); 99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);
69 } 100 }
70} 101}
71 102
72static void 103static void
73encode_ch (enc_t *enc, char ch) 104encode_ch (enc_t *enc, char ch)
… …
79static void 110static void
80encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 111encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
81{ 112{
82 char *end = str + len; 113 char *end = str + len;
83 114
115 need (enc, len);
116
84 while (str < end) 117 while (str < end)
85 { 118 {
86 unsigned char ch = *(unsigned char *)str; 119 unsigned char ch = *(unsigned char *)str;
120
87 if (ch >= 0x20 && ch < 0x80) // most common case 121 if (ch >= 0x20 && ch < 0x80) // most common case
88 { 122 {
123 if (ch == '"') // but with slow exceptions
124 {
125 need (enc, len += 1);
126 *enc->cur++ = '\\';
127 *enc->cur++ = '"';
128 }
129 else if (ch == '\\')
130 {
131 need (enc, len += 1);
132 *enc->cur++ = '\\';
133 *enc->cur++ = '\\';
134 }
135 else
89 *enc->cur++ = ch; 136 *enc->cur++ = ch;
137
90 str++; 138 ++str;
91 } 139 }
92 else 140 else
93 { 141 {
94 STRLEN clen; 142 switch (ch)
95 UV uch;
96
97 if (is_utf8)
98 { 143 {
99 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 144 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
100 if (clen < 0) 145 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"); 146 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
102 } 147 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
103 else 148 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
104 {
105 uch = ch;
106 clen = 1;
107 }
108 149
109 need (enc, len += 6); 150 default:
110
111 if (uch < 0xa0 || enc->flags & F_ASCII)
112 {
113 if (uch > 0xFFFFUL)
114 { 151 {
152 STRLEN clen;
153 UV uch;
154
155 if (is_utf8)
156 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1)
160 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
161 }
162 else
163 {
164 uch = ch;
115 len += 6; 165 clen = 1;
166 }
167
168 if (uch > 0x10FFFFUL)
169 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170
171 if (uch < 0x80 || enc->flags & F_ASCII)
172 {
173 if (uch > 0xFFFFUL)
174 {
116 need (enc, len += 6); 175 need (enc, len += 11);
117 sprintf (enc->cur, "\\u%04x\\u%04x", 176 sprintf (enc->cur, "\\u%04x\\u%04x",
118 (uch - 0x10000) / 0x400 + 0xD800, 177 (int)((uch - 0x10000) / 0x400 + 0xD800),
119 (uch - 0x10000) % 0x400 + 0xDC00); 178 (int)((uch - 0x10000) % 0x400 + 0xDC00));
120 enc->cur += 12; 179 enc->cur += 12;
180 }
181 else
182 {
183 static char hexdigit [16] = "0123456789abcdef";
184 need (enc, len += 5);
185 *enc->cur++ = '\\';
186 *enc->cur++ = 'u';
187 *enc->cur++ = hexdigit [ uch >> 12 ];
188 *enc->cur++ = hexdigit [(uch >> 8) & 15];
189 *enc->cur++ = hexdigit [(uch >> 4) & 15];
190 *enc->cur++ = hexdigit [(uch >> 0) & 15];
191 }
192
193 str += clen;
121 } 194 }
195 else if (is_utf8)
196 {
197 need (enc, len += clen);
198 do
199 {
200 *enc->cur++ = *str++;
201 }
202 while (--clen);
203 }
122 else 204 else
123 { 205 {
124 sprintf (enc->cur, "\\u%04x", uch); 206 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
125 enc->cur += 6; 207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str;
209 }
126 } 210 }
127 } 211 }
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 } 212 }
138 213
139 --len; 214 --len;
140 } 215 }
141} 216}
142 217
143#define INDENT SB \ 218static void
219encode_indent (enc_t *enc)
220{
144 if (enc->flags & F_INDENT) \ 221 if (enc->flags & F_INDENT)
145 { \ 222 {
146 int i_; \ 223 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
152 224
153#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 225 need (enc, spaces);
154#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 226 memset (enc->cur, ' ', spaces);
155#define COMMA SB \ 227 enc->cur += spaces;
228 }
229}
230
231static void
232encode_space (enc_t *enc)
233{
234 need (enc, 1);
156 encode_ch (enc, ','); \ 235 encode_ch (enc, ' ');
236}
237
238static void
239encode_nl (enc_t *enc)
240{
157 if (enc->flags & F_INDENT) \ 241 if (enc->flags & F_INDENT)
158 NL; \ 242 {
243 need (enc, 1);
244 encode_ch (enc, '\n');
245 }
246}
247
248static void
249encode_comma (enc_t *enc)
250{
251 encode_ch (enc, ',');
252
253 if (enc->flags & F_INDENT)
254 encode_nl (enc);
159 else if (enc->flags & F_SPACE_AFTER) \ 255 else if (enc->flags & F_SPACE_AFTER)
160 SPACE; \ 256 encode_space (enc);
161 SE 257}
162 258
163static void encode_sv (enc_t *enc, SV *sv); 259static void encode_sv (enc_t *enc, SV *sv);
164 260
165static void 261static void
166encode_av (enc_t *enc, AV *av) 262encode_av (enc_t *enc, AV *av)
167{ 263{
168 int i, len = av_len (av); 264 int i, len = av_len (av);
169 265
170 encode_ch (enc, '['); NL; 266 encode_ch (enc, '['); encode_nl (enc);
171 ++enc->indent; 267 ++enc->indent;
172 268
173 for (i = 0; i <= len; ++i) 269 for (i = 0; i <= len; ++i)
174 { 270 {
175 INDENT; 271 encode_indent (enc);
176 encode_sv (enc, *av_fetch (av, i, 0)); 272 encode_sv (enc, *av_fetch (av, i, 0));
177 273
178 if (i < len) 274 if (i < len)
179 COMMA; 275 encode_comma (enc);
180 } 276 }
181 277
182 NL; 278 encode_nl (enc);
183 279
184 --enc->indent; 280 --enc->indent;
185 INDENT; encode_ch (enc, ']'); 281 encode_indent (enc); encode_ch (enc, ']');
186} 282}
187 283
188static void 284static void
189encode_he (enc_t *enc, HE *he) 285encode_he (enc_t *enc, HE *he)
190{ 286{
… …
192 288
193 if (HeKLEN (he) == HEf_SVKEY) 289 if (HeKLEN (he) == HEf_SVKEY)
194 { 290 {
195 SV *sv = HeSVKEY (he); 291 SV *sv = HeSVKEY (he);
196 STRLEN len; 292 STRLEN len;
293 char *str;
294
295 SvGETMAGIC (sv);
197 char *str = SvPV (sv, len); 296 str = SvPV (sv, len);
198 297
199 encode_str (enc, str, len, SvUTF8 (sv)); 298 encode_str (enc, str, len, SvUTF8 (sv));
200 } 299 }
201 else 300 else
202 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
203 302
204 encode_ch (enc, '"'); 303 encode_ch (enc, '"');
205 304
206 if (enc->flags & F_SPACE_BEFORE) SPACE; 305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
207 encode_ch (enc, ':'); 306 encode_ch (enc, ':');
208 if (enc->flags & F_SPACE_AFTER ) SPACE; 307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
209 encode_sv (enc, HeVAL (he)); 308 encode_sv (enc, HeVAL (he));
210} 309}
211 310
212// compare hash entries, used when all keys are bytestrings 311// compare hash entries, used when all keys are bytestrings
213static int 312static int
… …
219 HE *b = *(HE **)b_; 318 HE *b = *(HE **)b_;
220 319
221 STRLEN la = HeKLEN (a); 320 STRLEN la = HeKLEN (a);
222 STRLEN lb = HeKLEN (b); 321 STRLEN lb = HeKLEN (b);
223 322
224 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 323 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
225 cmp = la < lb ? -1 : la == lb ? 0 : 1; 324 cmp = la - lb;
226 325
227 return cmp; 326 return cmp;
228} 327}
229 328
230// compare hash entries, used when some keys are sv's or utf-x 329// compare hash entries, used when some keys are sv's or utf-x
… …
237static void 336static void
238encode_hv (enc_t *enc, HV *hv) 337encode_hv (enc_t *enc, HV *hv)
239{ 338{
240 int count, i; 339 int count, i;
241 340
242 encode_ch (enc, '{'); NL; ++enc->indent; 341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
243 342
244 if ((count = hv_iterinit (hv))) 343 if ((count = hv_iterinit (hv)))
245 { 344 {
246 // for canonical output we have to sort by keys first 345 // for canonical output we have to sort by keys first
247 // actually, this is mostly due to the stupid so-called 346 // actually, this is mostly due to the stupid so-called
248 // security workaround added somewhere in 5.8.x. 347 // security workaround added somewhere in 5.8.x.
249 // that randomises hash orderings 348 // that randomises hash orderings
250 if (enc->flags & F_CANONICAL) 349 if (enc->flags & F_CANONICAL)
251 { 350 {
252 HE *he, *hes [count]; 351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
253 int fast = 1; 352 int fast = 1;
254 353
255 i = 0; 354 i = 0;
256 while ((he = hv_iternext (hv))) 355 while ((he = hv_iternext (hv)))
257 { 356 {
… …
264 363
265 if (fast) 364 if (fast)
266 qsort (hes, count, sizeof (HE *), he_cmp_fast); 365 qsort (hes, count, sizeof (HE *), he_cmp_fast);
267 else 366 else
268 { 367 {
269 // hack to disable "use bytes" 368 // hack to forcefully disable "use bytes"
270 COP *oldcop = PL_curcop, cop; 369 COP cop = *PL_curcop;
271 cop.op_private = 0; 370 cop.op_private = 0;
371
372 ENTER;
373 SAVETMPS;
374
375 SAVEVPTR (PL_curcop);
272 PL_curcop = &cop; 376 PL_curcop = &cop;
273 377
274 SAVETMPS;
275 qsort (hes, count, sizeof (HE *), he_cmp_slow); 378 qsort (hes, count, sizeof (HE *), he_cmp_slow);
379
276 FREETMPS; 380 FREETMPS;
277 381 LEAVE;
278 PL_curcop = oldcop;
279 } 382 }
280 383
281 for (i = 0; i < count; ++i) 384 for (i = 0; i < count; ++i)
282 { 385 {
283 INDENT; 386 encode_indent (enc);
284 encode_he (enc, hes [i]); 387 encode_he (enc, hes [i]);
285 388
286 if (i < count - 1) 389 if (i < count - 1)
287 COMMA; 390 encode_comma (enc);
288 } 391 }
289 392
290 NL; 393 encode_nl (enc);
291 } 394 }
292 else 395 else
293 { 396 {
294 SV *sv; 397 SV *sv;
295 HE *he = hv_iternext (hv); 398 HE *he = hv_iternext (hv);
296 399
297 for (;;) 400 for (;;)
298 { 401 {
299 INDENT; 402 encode_indent (enc);
300 encode_he (enc, he); 403 encode_he (enc, he);
301 404
302 if (!(he = hv_iternext (hv))) 405 if (!(he = hv_iternext (hv)))
303 break; 406 break;
304 407
305 COMMA; 408 encode_comma (enc);
306 } 409 }
307 410
308 NL; 411 encode_nl (enc);
309 } 412 }
310 } 413 }
311 414
312 --enc->indent; INDENT; encode_ch (enc, '}'); 415 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
313} 416}
314 417
315static void 418static void
316encode_sv (enc_t *enc, SV *sv) 419encode_sv (enc_t *enc, SV *sv)
317{ 420{
421 SvGETMAGIC (sv);
422
318 if (SvPOKp (sv)) 423 if (SvPOKp (sv))
319 { 424 {
320 STRLEN len; 425 STRLEN len;
321 char *str = SvPV (sv, len); 426 char *str = SvPV (sv, len);
322 encode_ch (enc, '"'); 427 encode_ch (enc, '"');
… …
337 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
338 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
339 } 444 }
340 else if (SvROK (sv)) 445 else if (SvROK (sv))
341 { 446 {
342 if (!--enc->max_recurse) 447 SV *rv = SvRV (sv);
448
449 if (enc->indent >= enc->max_depth)
343 croak ("data structure too deep (hit recursion limit)"); 450 croak ("data structure too deep (hit recursion limit)");
344 451
345 sv = SvRV (sv);
346
347 switch (SvTYPE (sv)) 452 switch (SvTYPE (rv))
348 { 453 {
349 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 454 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
350 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 455 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
351 456
352 default: 457 default:
353 croak ("JSON can only represent references to arrays or hashes"); 458 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
459 SvPV_nolen (sv));
354 } 460 }
355 } 461 }
356 else if (!SvOK (sv)) 462 else if (!SvOK (sv))
357 encode_str (enc, "null", 4, 0); 463 encode_str (enc, "null", 4, 0);
358 else 464 else
359 croak ("encountered perl type that JSON cannot handle"); 465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv));
360} 467}
361 468
362static SV * 469static SV *
363encode_json (SV *scalar, UV flags) 470encode_json (SV *scalar, UV flags)
364{ 471{
365 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
366 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
367 474
368 enc_t enc; 475 enc_t enc;
369 enc.flags = flags; 476 enc.flags = flags;
370 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
371 enc.cur = SvPVX (enc.sv); 478 enc.cur = SvPVX (enc.sv);
372 enc.end = SvEND (enc.sv); 479 enc.end = SvEND (enc.sv);
373 enc.max_recurse = 0;
374 enc.indent = 0; 480 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL;
375 482
376 SvPOK_only (enc.sv); 483 SvPOK_only (enc.sv);
377 encode_sv (&enc, scalar); 484 encode_sv (&enc, scalar);
378 485
379 if (!(flags & (F_ASCII | F_UTF8))) 486 if (!(flags & (F_ASCII | F_UTF8)))
380 SvUTF8_on (enc.sv); 487 SvUTF8_on (enc.sv);
381 488
382 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK)
492 shrink (enc.sv);
493
383 return enc.sv; 494 return enc.sv;
384} 495}
385 496
386///////////////////////////////////////////////////////////////////////////// 497/////////////////////////////////////////////////////////////////////////////
498// decoder
387 499
388#define WS \ 500// structure used for decoding JSON
501typedef struct
502{
503 char *cur; // current parser pointer
504 char *end; // end of input string
505 const char *err; // parse error, if != 0
506 UV flags; // F_*
507} dec_t;
508
509static void
510decode_ws (dec_t *dec)
511{
389 for (;;) \ 512 for (;;)
390 { \ 513 {
391 char ch = *dec->cur; \ 514 char ch = *dec->cur;
515
392 if (ch > 0x20 \ 516 if (ch > 0x20
393 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 517 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
394 break; \ 518 break;
519
395 ++dec->cur; \ 520 ++dec->cur;
396 } 521 }
522}
397 523
398#define ERR(reason) SB dec->err = reason; goto fail; SE 524#define ERR(reason) SB dec->err = reason; goto fail; SE
399#define EXPECT_CH(ch) SB \ 525#define EXPECT_CH(ch) SB \
400 if (*dec->cur != ch) \ 526 if (*dec->cur != ch) \
401 ERR (# ch " expected"); \ 527 ERR (# ch " expected"); \
402 ++dec->cur; \ 528 ++dec->cur; \
403 SE 529 SE
404 530
405static SV *decode_sv (dec_t *dec); 531static SV *decode_sv (dec_t *dec);
406 532
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]; 533static signed char decode_hexdigit[256];
413 534
414static UV 535static UV
415decode_4hex (dec_t *dec) 536decode_4hex (dec_t *dec)
416{ 537{
417 signed char d1, d2, d3, d4; 538 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur;
418 540
419 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
420 if (d1 < 0) ERR ("four hexadecimal digits expected"); 541 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"); 542 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"); 543 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"); 544 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
427 545
428 dec->cur += 4; 546 dec->cur += 4;
429 547
430 return ((UV)d1) << 12 548 return ((UV)d1) << 12
431 | ((UV)d2) << 8 549 | ((UV)d2) << 8
… …
437} 555}
438 556
439static SV * 557static SV *
440decode_str (dec_t *dec) 558decode_str (dec_t *dec)
441{ 559{
442 SV *sv = NEWSV (0,2); 560 SV *sv = 0;
443 STRLEN cur = 0;
444 int utf8 = 0; 561 int utf8 = 0;
445 562
446 for (;;) 563 do
447 { 564 {
448 unsigned char ch = *(unsigned char *)dec->cur; 565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
566 char *cur = buf;
449 567
450 if (ch == '"') 568 do
451 break;
452 else if (ch == '\\')
453 { 569 {
454 switch (*++dec->cur) 570 unsigned char ch = *(unsigned char *)dec->cur++;
571
572 if (ch == '"')
455 { 573 {
456 case '\\': 574 --dec->cur;
457 case '/': 575 break;
458 case '"': APPEND_CH (*dec->cur++); break; 576 }
459 577 else if (ch == '\\')
460 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 578 {
461 case 't': APPEND_CH ('\011'); ++dec->cur; break; 579 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 { 580 {
468 UV lo, hi; 581 case '\\':
469 ++dec->cur; 582 case '/':
583 case '"': *cur++ = *dec->cur++; break;
470 584
471 hi = decode_4hex (dec); 585 case 'b': ++dec->cur; *cur++ = '\010'; break;
472 if (hi == (UV)-1) 586 case 't': ++dec->cur; *cur++ = '\011'; break;
473 goto fail; 587 case 'n': ++dec->cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break;
474 590
475 // possibly a surrogate pair 591 case 'u':
476 if (hi >= 0xd800 && hi < 0xdc00)
477 { 592 {
478 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 593 UV lo, hi;
479 ERR ("illegal surrogate character");
480
481 dec->cur += 2; 594 ++dec->cur;
482 595
483 lo = decode_4hex (dec); 596 hi = decode_4hex (dec);
484 if (lo == (UV)-1) 597 if (hi == (UV)-1)
485 goto fail; 598 goto fail;
486 599
600 // possibly a surrogate pair
601 if (hi >= 0xd800)
602 if (hi < 0xdc00)
603 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair");
606
607 dec->cur += 2;
608
609 lo = decode_4hex (dec);
610 if (lo == (UV)-1)
611 goto fail;
612
487 if (lo < 0xdc00 || lo >= 0xe000) 613 if (lo < 0xdc00 || lo >= 0xe000)
488 ERR ("surrogate pair expected"); 614 ERR ("surrogate pair expected");
489 615
490 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 616 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
617 }
618 else if (hi < 0xe000)
619 ERR ("missing high surrogate character in surrogate pair");
620
621 if (hi >= 0x80)
622 {
623 utf8 = 1;
624
625 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
626 }
627 else
628 *cur++ = hi;
491 } 629 }
492 else if (lo >= 0xdc00 && lo < 0xe000)
493 ERR ("illegal surrogate character");
494
495 if (hi >= 0x80)
496 { 630 break;
497 utf8 = 1;
498 631
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 632 default:
503 APPEND_CH (hi); 633 --dec->cur;
634 ERR ("illegal backslash escape sequence in string");
504 } 635 }
505 break;
506 } 636 }
637 else if (ch >= 0x20 && ch <= 0x7f)
638 *cur++ = ch;
639 else if (ch >= 0x80)
640 {
641 --dec->cur;
642
643 STRLEN clen;
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
645 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string");
647
648 do
649 {
650 *cur++ = *dec->cur++;
651 }
652 while (--clen);
653
654 utf8 = 1;
655 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else
659 ERR ("invalid character encountered");
660
507 } 661 }
508 else if (ch >= 0x20 && ch <= 0x7f) 662 while (cur < buf + SHORT_STRING_LEN);
509 APPEND_CH (*dec->cur++); 663
510 else if (ch >= 0x80) 664 STRLEN len = cur - buf;
665
666 if (sv)
511 { 667 {
512 STRLEN clen; 668 SvGROW (sv, SvCUR (sv) + len + 1);
513 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
514 if (clen < 0) 670 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 } 671 }
521 else 672 else
522 ERR ("invalid character encountered"); 673 sv = newSVpvn (buf, len);
523 } 674 }
675 while (*dec->cur != '"');
524 676
525 ++dec->cur; 677 ++dec->cur;
526 678
679 if (sv)
680 {
527 SvPOK_only (sv); 681 SvPOK_only (sv);
528
529 SvCUR_set (sv, cur);
530 *SvEND (sv) = 0; 682 *SvEND (sv) = 0;
531 683
532 if (utf8) 684 if (utf8)
533 SvUTF8_on (sv); 685 SvUTF8_on (sv);
686 }
687 else
688 sv = newSVpvn ("", 0);
534 689
535 return sv; 690 return sv;
536 691
537fail: 692fail:
538 SvREFCNT_dec (sv);
539 return 0; 693 return 0;
540} 694}
541 695
542static SV * 696static SV *
543decode_num (dec_t *dec) 697decode_num (dec_t *dec)
… …
553 { 707 {
554 ++dec->cur; 708 ++dec->cur;
555 if (*dec->cur >= '0' && *dec->cur <= '9') 709 if (*dec->cur >= '0' && *dec->cur <= '9')
556 ERR ("malformed number (leading zero must not be followed by another digit)"); 710 ERR ("malformed number (leading zero must not be followed by another digit)");
557 } 711 }
558 712 else if (*dec->cur < '0' || *dec->cur > '9')
559 // int 713 ERR ("malformed number (no digits after initial minus)");
714 else
715 do
716 {
717 ++dec->cur;
718 }
560 while (*dec->cur >= '0' && *dec->cur <= '9') 719 while (*dec->cur >= '0' && *dec->cur <= '9');
561 ++dec->cur;
562 720
563 // [frac] 721 // [frac]
564 if (*dec->cur == '.') 722 if (*dec->cur == '.')
565 { 723 {
566 is_nv = 1; 724 ++dec->cur;
725
726 if (*dec->cur < '0' || *dec->cur > '9')
727 ERR ("malformed number (no digits after decimal point)");
567 728
568 do 729 do
569 { 730 {
570 ++dec->cur; 731 ++dec->cur;
571 } 732 }
572 while (*dec->cur >= '0' && *dec->cur <= '9'); 733 while (*dec->cur >= '0' && *dec->cur <= '9');
734
735 is_nv = 1;
573 } 736 }
574 737
575 // [exp] 738 // [exp]
576 if (*dec->cur == 'e' || *dec->cur == 'E') 739 if (*dec->cur == 'e' || *dec->cur == 'E')
577 { 740 {
578 is_nv = 1;
579
580 ++dec->cur; 741 ++dec->cur;
742
581 if (*dec->cur == '-' || *dec->cur == '+') 743 if (*dec->cur == '-' || *dec->cur == '+')
582 ++dec->cur; 744 ++dec->cur;
583 745
746 if (*dec->cur < '0' || *dec->cur > '9')
747 ERR ("malformed number (no digits after exp sign)");
748
749 do
750 {
751 ++dec->cur;
752 }
584 while (*dec->cur >= '0' && *dec->cur <= '9') 753 while (*dec->cur >= '0' && *dec->cur <= '9');
585 ++dec->cur; 754
755 is_nv = 1;
586 } 756 }
587 757
588 if (!is_nv) 758 if (!is_nv)
589 { 759 {
590 UV uv; 760 UV uv;
… …
608static SV * 778static SV *
609decode_av (dec_t *dec) 779decode_av (dec_t *dec)
610{ 780{
611 AV *av = newAV (); 781 AV *av = newAV ();
612 782
783 decode_ws (dec);
784 if (*dec->cur == ']')
785 ++dec->cur;
786 else
613 for (;;) 787 for (;;)
614 { 788 {
615 SV *value; 789 SV *value;
616 790
617 value = decode_sv (dec); 791 value = decode_sv (dec);
618 if (!value) 792 if (!value)
619 goto fail; 793 goto fail;
620 794
621 av_push (av, value); 795 av_push (av, value);
622 796
623 WS; 797 decode_ws (dec);
624 798
625 if (*dec->cur == ']') 799 if (*dec->cur == ']')
626 { 800 {
627 ++dec->cur; 801 ++dec->cur;
628 break; 802 break;
803 }
629 } 804
630
631 if (*dec->cur != ',') 805 if (*dec->cur != ',')
632 ERR (", or ] expected while parsing array"); 806 ERR (", or ] expected while parsing array");
633 807
634 ++dec->cur; 808 ++dec->cur;
635 } 809 }
636 810
637 return newRV_noinc ((SV *)av); 811 return newRV_noinc ((SV *)av);
638 812
639fail: 813fail:
640 SvREFCNT_dec (av); 814 SvREFCNT_dec (av);
… …
644static SV * 818static SV *
645decode_hv (dec_t *dec) 819decode_hv (dec_t *dec)
646{ 820{
647 HV *hv = newHV (); 821 HV *hv = newHV ();
648 822
823 decode_ws (dec);
824 if (*dec->cur == '}')
825 ++dec->cur;
826 else
649 for (;;) 827 for (;;)
650 { 828 {
651 SV *key, *value; 829 SV *key, *value;
652 830
653 WS; EXPECT_CH ('"'); 831 decode_ws (dec); EXPECT_CH ('"');
654 832
655 key = decode_str (dec); 833 key = decode_str (dec);
656 if (!key) 834 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; 835 goto fail;
836
837 decode_ws (dec); EXPECT_CH (':');
838
839 value = decode_sv (dec);
840 if (!value)
841 {
842 SvREFCNT_dec (key);
843 goto fail;
666 } 844 }
667 845
668 //TODO: optimise 846 //TODO: optimise
669 hv_store_ent (hv, key, value, 0); 847 hv_store_ent (hv, key, value, 0);
670 848
671 WS; 849 decode_ws (dec);
672 850
673 if (*dec->cur == '}') 851 if (*dec->cur == '}')
674 { 852 {
675 ++dec->cur; 853 ++dec->cur;
676 break; 854 break;
677 } 855 }
678 856
679 if (*dec->cur != ',') 857 if (*dec->cur != ',')
680 ERR (", or } expected while parsing object/hash"); 858 ERR (", or } expected while parsing object/hash");
681 859
682 ++dec->cur; 860 ++dec->cur;
683 } 861 }
684 862
685 return newRV_noinc ((SV *)hv); 863 return newRV_noinc ((SV *)hv);
686 864
687fail: 865fail:
688 SvREFCNT_dec (hv); 866 SvREFCNT_dec (hv);
… …
690} 868}
691 869
692static SV * 870static SV *
693decode_sv (dec_t *dec) 871decode_sv (dec_t *dec)
694{ 872{
695 WS; 873 decode_ws (dec);
696 switch (*dec->cur) 874 switch (*dec->cur)
697 { 875 {
698 case '"': ++dec->cur; return decode_str (dec); 876 case '"': ++dec->cur; return decode_str (dec);
699 case '[': ++dec->cur; return decode_av (dec); 877 case '[': ++dec->cur; return decode_av (dec);
700 case '{': ++dec->cur; return decode_hv (dec); 878 case '{': ++dec->cur; return decode_hv (dec);
… …
728 906
729 case 'n': 907 case 'n':
730 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 908 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
731 { 909 {
732 dec->cur += 4; 910 dec->cur += 4;
733 return newSViv (1); 911 return newSVsv (&PL_sv_undef);
734 } 912 }
735 else 913 else
736 ERR ("'null' expected"); 914 ERR ("'null' expected");
737 915
738 break; 916 break;
739 917
740 default: 918 default:
741 ERR ("malformed json string"); 919 ERR ("malformed json string, neither array, object, number, string or atom");
742 break; 920 break;
743 } 921 }
744 922
745fail: 923fail:
746 return 0; 924 return 0;
… …
749static SV * 927static SV *
750decode_json (SV *string, UV flags) 928decode_json (SV *string, UV flags)
751{ 929{
752 SV *sv; 930 SV *sv;
753 931
754 if (!(flags & F_UTF8)) 932 if (flags & F_UTF8)
933 sv_utf8_downgrade (string, 0);
934 else
755 sv_utf8_upgrade (string); 935 sv_utf8_upgrade (string);
756 936
757 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
758 938
759 dec_t dec; 939 dec_t dec;
760 dec.flags = flags; 940 dec.flags = flags;
761 dec.cur = SvPVX (string); 941 dec.cur = SvPVX (string);
762 dec.end = SvEND (string); 942 dec.end = SvEND (string);
763 dec.err = 0; 943 dec.err = 0;
764 944
765 *dec.end = 1; // invalid anywhere
766 sv = decode_sv (&dec); 945 sv = decode_sv (&dec);
767 *dec.end = 0;
768 946
769 if (!sv) 947 if (!sv)
770 { 948 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
771 IV offset = utf8_distance (dec.cur, SvPVX (string)); 951 : utf8_distance (dec.cur, SvPVX (string));
772 SV *uni = sv_newmortal (); 952 SV *uni = sv_newmortal ();
773 953
954 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE;
957 ENTER;
958 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop;
774 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE;
962
775 croak ("%s, at character %d (%s)", 963 croak ("%s, at character offset %d (%s)",
776 dec.err, 964 dec.err,
777 (int)offset, 965 (int)offset,
778 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
779 } 967 }
780 968
781 sv = sv_2mortal (sv); 969 sv = sv_2mortal (sv);
782 970
783 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 971 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)"); 972 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
785 973
786 return sv; 974 return sv;
787} 975}
976
977/////////////////////////////////////////////////////////////////////////////
978// XS interface functions
788 979
789MODULE = JSON::XS PACKAGE = JSON::XS 980MODULE = JSON::XS PACKAGE = JSON::XS
790 981
791BOOT: 982BOOT:
792{ 983{
… …
794 985
795 memset (decode_hexdigit, 0xff, 256); 986 memset (decode_hexdigit, 0xff, 256);
796 for (i = 10; i--; ) 987 for (i = 10; i--; )
797 decode_hexdigit ['0' + i] = i; 988 decode_hexdigit ['0' + i] = i;
798 989
799 for (i = 6; --i; ) 990 for (i = 7; i--; )
800 { 991 {
801 decode_hexdigit ['a' + i] = 10 + i; 992 decode_hexdigit ['a' + i] = 10 + i;
802 decode_hexdigit ['A' + i] = 10 + i; 993 decode_hexdigit ['A' + i] = 10 + i;
803 } 994 }
804 995
805 json_stash = gv_stashpv ("JSON::XS", 1); 996 json_stash = gv_stashpv ("JSON::XS", 1);
806} 997}
998
999PROTOTYPES: DISABLE
807 1000
808SV *new (char *dummy) 1001SV *new (char *dummy)
809 CODE: 1002 CODE:
810 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
811 OUTPUT: 1004 OUTPUT:
812 RETVAL 1005 RETVAL
813 1006
814SV *ascii (SV *self, int enable) 1007SV *ascii (SV *self, int enable = 1)
815 ALIAS: 1008 ALIAS:
816 ascii = F_ASCII 1009 ascii = F_ASCII
817 utf8 = F_UTF8 1010 utf8 = F_UTF8
818 indent = F_INDENT 1011 indent = F_INDENT
819 canonical = F_CANONICAL 1012 canonical = F_CANONICAL
820 space_before = F_SPACE_BEFORE 1013 space_before = F_SPACE_BEFORE
821 space_after = F_SPACE_AFTER 1014 space_after = F_SPACE_AFTER
822 json_rpc = F_JSON_RPC
823 pretty = F_PRETTY 1015 pretty = F_PRETTY
824 allow_nonref = F_ALLOW_NONREF 1016 allow_nonref = F_ALLOW_NONREF
1017 shrink = F_SHRINK
825 CODE: 1018 CODE:
826{ 1019{
827 UV *uv = SvJSON (self); 1020 UV *uv = SvJSON (self);
828 if (enable) 1021 if (enable)
829 *uv |= ix; 1022 *uv |= ix;
… …
841 1034
842void decode (SV *self, SV *jsonstr) 1035void decode (SV *self, SV *jsonstr)
843 PPCODE: 1036 PPCODE:
844 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1037 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
845 1038
1039PROTOTYPES: ENABLE
1040
846void to_json (SV *scalar) 1041void to_json (SV *scalar)
847 PPCODE: 1042 PPCODE:
848 XPUSHs (encode_json (scalar, F_UTF8)); 1043 XPUSHs (encode_json (scalar, F_UTF8));
849 1044
850void from_json (SV *jsonstr) 1045void from_json (SV *jsonstr)

Diff Legend

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