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 1.2 by root, Thu Mar 22 17:28:50 2007 UTC vs.
Revision 1.14 by root, Sat Mar 24 22:55:43 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_ALLOW_NONREF 0x00000080
15#define F_JSON_RPC 0x00000040 16#define F_SHRINK 0x00000100
16 17
17#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
18#define F_DEFAULT 0 19#define F_DEFAULT 0
19 20
20#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 512 // special-case strings of up to this size
21 26
22#define SB do { 27#define SB do {
23#define SE } while (0) 28#define SE } while (0)
24 29
25static 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
26 79
27// structure used for encoding JSON 80// structure used for encoding JSON
28typedef struct 81typedef struct
29{ 82{
30 char *cur; 83 char *cur; // SvPVX (sv) + current output position
31 STRLEN len; // SvLEN (sv)
32 char *end; // SvEND (sv) 84 char *end; // SvEND (sv)
33 SV *sv; 85 SV *sv; // result scalar
34 UV flags; 86 UV flags; // F_*
35 int max_recurse; 87 int indent; // indentation level
36 int indent; 88 int max_depth; // max. recursion level
37} enc_t; 89} enc_t;
38
39// structure used for decoding JSON
40typedef struct
41{
42 char *cur;
43 char *end;
44 char *err;
45 UV flags;
46} dec_t;
47
48static UV *
49SvJSON (SV *sv)
50{
51 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
52 croak ("object is not of type JSON::XS");
53
54 return &SvUVX (SvRV (sv));
55}
56
57/////////////////////////////////////////////////////////////////////////////
58 90
59static void 91static void
60need (enc_t *enc, STRLEN len) 92need (enc_t *enc, STRLEN len)
61{ 93{
62 if (enc->cur + len >= enc->end) 94 if (enc->cur + len >= enc->end)
63 { 95 {
64 STRLEN cur = enc->cur - SvPVX (enc->sv); 96 STRLEN cur = enc->cur - SvPVX (enc->sv);
65 SvGROW (enc->sv, cur + len + 1); 97 SvGROW (enc->sv, cur + len + 1);
66 enc->cur = SvPVX (enc->sv) + cur; 98 enc->cur = SvPVX (enc->sv) + cur;
67 enc->end = SvEND (enc->sv); 99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);
68 } 100 }
69} 101}
70 102
71static void 103static void
72encode_ch (enc_t *enc, char ch) 104encode_ch (enc_t *enc, char ch)
… …
78static void 110static void
79encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 111encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
80{ 112{
81 char *end = str + len; 113 char *end = str + len;
82 114
115 need (enc, len);
116
83 while (str < end) 117 while (str < end)
84 { 118 {
85 unsigned char ch = *(unsigned char *)str; 119 unsigned char ch = *(unsigned char *)str;
120
86 if (ch >= 0x20 && ch < 0x80) // most common case 121 if (ch >= 0x20 && ch < 0x80) // most common case
87 { 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
88 *enc->cur++ = ch; 136 *enc->cur++ = ch;
137
89 str++; 138 ++str;
90 } 139 }
91 else 140 else
92 { 141 {
93 STRLEN clen; 142 switch (ch)
94 UV uch;
95
96 if (is_utf8)
97 { 143 {
98 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;
99 if (clen < 0) 145 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
100 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;
101 } 147 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
102 else 148 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
103 {
104 uch = ch;
105 clen = 1;
106 }
107 149
108 need (enc, len += 6); 150 default:
109
110 if (uch < 0xa0 || enc->flags & F_ASCII)
111 {
112 if (uch > 0xFFFFUL)
113 { 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;
114 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 {
115 need (enc, len += 6); 175 need (enc, len += 11);
116 sprintf (enc->cur, "\\u%04x\\u%04x", 176 sprintf (enc->cur, "\\u%04x\\u%04x",
117 (uch - 0x10000) / 0x400 + 0xD800, 177 (int)((uch - 0x10000) / 0x400 + 0xD800),
118 (uch - 0x10000) % 0x400 + 0xDC00); 178 (int)((uch - 0x10000) % 0x400 + 0xDC00));
119 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;
120 } 194 }
195 else if (is_utf8)
196 {
197 need (enc, len += clen);
198 do
199 {
200 *enc->cur++ = *str++;
201 }
202 while (--clen);
203 }
121 else 204 else
122 { 205 {
123 sprintf (enc->cur, "\\u%04x", uch); 206 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
124 enc->cur += 6; 207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str;
209 }
125 } 210 }
126 } 211 }
127 else if (is_utf8)
128 {
129 memcpy (enc->cur, str, clen);
130 enc->cur += clen;
131 }
132 else
133 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
134
135 str += clen;
136 } 212 }
137 213
138 --len; 214 --len;
139 } 215 }
140} 216}
141 217
142#define INDENT SB \ 218static void
219encode_indent (enc_t *enc)
220{
143 if (enc->flags & F_INDENT) \ 221 if (enc->flags & F_INDENT)
144 { \ 222 {
145 int i_; \ 223 int spaces = enc->indent * INDENT_STEP;
146 need (enc, enc->indent); \
147 for (i_ = enc->indent * 3; i_--; )\
148 encode_ch (enc, ' '); \
149 } \
150 SE
151 224
152#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 225 need (enc, spaces);
153#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 226 memset (enc->cur, ' ', spaces);
154#define COMMA SB \ 227 enc->cur += spaces;
228 }
229}
230
231static void
232encode_space (enc_t *enc)
233{
234 need (enc, 1);
155 encode_ch (enc, ','); \ 235 encode_ch (enc, ' ');
236}
237
238static void
239encode_nl (enc_t *enc)
240{
156 if (enc->flags & F_INDENT) \ 241 if (enc->flags & F_INDENT)
157 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);
158 else if (enc->flags & F_SPACE_AFTER) \ 255 else if (enc->flags & F_SPACE_AFTER)
159 SPACE; \ 256 encode_space (enc);
160 SE 257}
161 258
162static void encode_sv (enc_t *enc, SV *sv); 259static void encode_sv (enc_t *enc, SV *sv);
163 260
164static void 261static void
165encode_av (enc_t *enc, AV *av) 262encode_av (enc_t *enc, AV *av)
166{ 263{
167 int i, len = av_len (av); 264 int i, len = av_len (av);
168 265
169 encode_ch (enc, '['); NL; 266 encode_ch (enc, '['); encode_nl (enc);
170 ++enc->indent; 267 ++enc->indent;
171 268
172 for (i = 0; i <= len; ++i) 269 for (i = 0; i <= len; ++i)
173 { 270 {
174 INDENT; 271 encode_indent (enc);
175 encode_sv (enc, *av_fetch (av, i, 0)); 272 encode_sv (enc, *av_fetch (av, i, 0));
176 273
177 if (i < len) 274 if (i < len)
178 COMMA; 275 encode_comma (enc);
179 } 276 }
180 277
181 NL; 278 encode_nl (enc);
182 279
183 --enc->indent; 280 --enc->indent;
184 INDENT; encode_ch (enc, ']'); 281 encode_indent (enc); encode_ch (enc, ']');
185} 282}
186 283
187static void 284static void
188encode_he (enc_t *enc, HE *he) 285encode_he (enc_t *enc, HE *he)
189{ 286{
… …
191 288
192 if (HeKLEN (he) == HEf_SVKEY) 289 if (HeKLEN (he) == HEf_SVKEY)
193 { 290 {
194 SV *sv = HeSVKEY (he); 291 SV *sv = HeSVKEY (he);
195 STRLEN len; 292 STRLEN len;
293 char *str;
294
295 SvGETMAGIC (sv);
196 char *str = SvPV (sv, len); 296 str = SvPV (sv, len);
197 297
198 encode_str (enc, str, len, SvUTF8 (sv)); 298 encode_str (enc, str, len, SvUTF8 (sv));
199 } 299 }
200 else 300 else
201 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
202 302
203 encode_ch (enc, '"'); 303 encode_ch (enc, '"');
204 304
205 if (enc->flags & F_SPACE_BEFORE) SPACE; 305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
206 encode_ch (enc, ':'); 306 encode_ch (enc, ':');
207 if (enc->flags & F_SPACE_AFTER ) SPACE; 307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
208 encode_sv (enc, HeVAL (he)); 308 encode_sv (enc, HeVAL (he));
209} 309}
210 310
211// compare hash entries, used when all keys are bytestrings 311// compare hash entries, used when all keys are bytestrings
212static int 312static int
… …
218 HE *b = *(HE **)b_; 318 HE *b = *(HE **)b_;
219 319
220 STRLEN la = HeKLEN (a); 320 STRLEN la = HeKLEN (a);
221 STRLEN lb = HeKLEN (b); 321 STRLEN lb = HeKLEN (b);
222 322
223 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 323 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
224 cmp = la < lb ? -1 : la == lb ? 0 : 1; 324 cmp = la - lb;
225 325
226 return cmp; 326 return cmp;
227} 327}
228 328
229// 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
… …
236static void 336static void
237encode_hv (enc_t *enc, HV *hv) 337encode_hv (enc_t *enc, HV *hv)
238{ 338{
239 int count, i; 339 int count, i;
240 340
241 encode_ch (enc, '{'); NL; ++enc->indent; 341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
242 342
243 if ((count = hv_iterinit (hv))) 343 if ((count = hv_iterinit (hv)))
244 { 344 {
245 // for canonical output we have to sort by keys first 345 // for canonical output we have to sort by keys first
246 // actually, this is mostly due to the stupid so-called 346 // actually, this is mostly due to the stupid so-called
247 // security workaround added somewhere in 5.8.x. 347 // security workaround added somewhere in 5.8.x.
248 // that randomises hash orderings 348 // that randomises hash orderings
249 if (enc->flags & F_CANONICAL) 349 if (enc->flags & F_CANONICAL)
250 { 350 {
251 HE *he, *hes [count]; 351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
252 int fast = 1; 352 int fast = 1;
253 353
254 i = 0; 354 i = 0;
255 while ((he = hv_iternext (hv))) 355 while ((he = hv_iternext (hv)))
256 { 356 {
… …
263 363
264 if (fast) 364 if (fast)
265 qsort (hes, count, sizeof (HE *), he_cmp_fast); 365 qsort (hes, count, sizeof (HE *), he_cmp_fast);
266 else 366 else
267 { 367 {
268 // hack to disable "use bytes" 368 // hack to forcefully disable "use bytes"
269 COP *oldcop = PL_curcop, cop; 369 COP cop = *PL_curcop;
270 cop.op_private = 0; 370 cop.op_private = 0;
371
372 ENTER;
373 SAVETMPS;
374
375 SAVEVPTR (PL_curcop);
271 PL_curcop = &cop; 376 PL_curcop = &cop;
272 377
273 SAVETMPS;
274 qsort (hes, count, sizeof (HE *), he_cmp_slow); 378 qsort (hes, count, sizeof (HE *), he_cmp_slow);
379
275 FREETMPS; 380 FREETMPS;
276 381 LEAVE;
277 PL_curcop = oldcop;
278 } 382 }
279 383
280 for (i = 0; i < count; ++i) 384 for (i = 0; i < count; ++i)
281 { 385 {
282 INDENT; 386 encode_indent (enc);
283 encode_he (enc, hes [i]); 387 encode_he (enc, hes [i]);
284 388
285 if (i < count - 1) 389 if (i < count - 1)
286 COMMA; 390 encode_comma (enc);
287 } 391 }
288 392
289 NL; 393 encode_nl (enc);
290 } 394 }
291 else 395 else
292 { 396 {
293 SV *sv; 397 SV *sv;
294 HE *he = hv_iternext (hv); 398 HE *he = hv_iternext (hv);
295 399
296 for (;;) 400 for (;;)
297 { 401 {
298 INDENT; 402 encode_indent (enc);
299 encode_he (enc, he); 403 encode_he (enc, he);
300 404
301 if (!(he = hv_iternext (hv))) 405 if (!(he = hv_iternext (hv)))
302 break; 406 break;
303 407
304 COMMA; 408 encode_comma (enc);
305 } 409 }
306 410
307 NL; 411 encode_nl (enc);
308 } 412 }
309 } 413 }
310 414
311 --enc->indent; INDENT; encode_ch (enc, '}'); 415 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
312} 416}
313 417
314static void 418static void
315encode_sv (enc_t *enc, SV *sv) 419encode_sv (enc_t *enc, SV *sv)
316{ 420{
421 SvGETMAGIC (sv);
422
317 if (SvPOKp (sv)) 423 if (SvPOKp (sv))
318 { 424 {
319 STRLEN len; 425 STRLEN len;
320 char *str = SvPV (sv, len); 426 char *str = SvPV (sv, len);
321 encode_ch (enc, '"'); 427 encode_ch (enc, '"');
… …
336 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
337 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
338 } 444 }
339 else if (SvROK (sv)) 445 else if (SvROK (sv))
340 { 446 {
341 if (!--enc->max_recurse) 447 SV *rv = SvRV (sv);
448
449 if (enc->indent >= enc->max_depth)
342 croak ("data structure too deep (hit recursion limit)"); 450 croak ("data structure too deep (hit recursion limit)");
343 451
344 sv = SvRV (sv);
345
346 switch (SvTYPE (sv)) 452 switch (SvTYPE (rv))
347 { 453 {
348 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 454 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
349 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 455 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
350 456
351 default: 457 default:
352 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));
353 } 460 }
354 } 461 }
355 else if (!SvOK (sv)) 462 else if (!SvOK (sv))
356 encode_str (enc, "null", 4, 0); 463 encode_str (enc, "null", 4, 0);
357 else 464 else
358 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));
359} 467}
360 468
361static SV * 469static SV *
362encode_json (SV *scalar, UV flags) 470encode_json (SV *scalar, UV flags)
363{ 471{
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474
364 enc_t enc; 475 enc_t enc;
365 enc.flags = flags; 476 enc.flags = flags;
366 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
367 enc.cur = SvPVX (enc.sv); 478 enc.cur = SvPVX (enc.sv);
368 enc.end = SvEND (enc.sv); 479 enc.end = SvEND (enc.sv);
369 enc.max_recurse = 0;
370 enc.indent = 0; 480 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL;
371 482
372 SvPOK_only (enc.sv); 483 SvPOK_only (enc.sv);
373 encode_sv (&enc, scalar); 484 encode_sv (&enc, scalar);
374 485
375 if (!(flags & (F_ASCII | F_UTF8))) 486 if (!(flags & (F_ASCII | F_UTF8)))
376 SvUTF8_on (enc.sv); 487 SvUTF8_on (enc.sv);
377 488
378 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
379 return enc.sv; 494 return enc.sv;
380} 495}
381 496
382///////////////////////////////////////////////////////////////////////////// 497/////////////////////////////////////////////////////////////////////////////
498// decoder
383 499
384#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{
385 for (;;) \ 512 for (;;)
386 { \ 513 {
387 char ch = *dec->cur; \ 514 char ch = *dec->cur;
515
388 if (ch > 0x20 \ 516 if (ch > 0x20
389 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 517 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
390 break; \ 518 break;
519
391 ++dec->cur; \ 520 ++dec->cur;
392 } 521 }
522}
393 523
394#define ERR(reason) SB dec->err = reason; goto fail; SE 524#define ERR(reason) SB dec->err = reason; goto fail; SE
395#define EXPECT_CH(ch) SB \ 525#define EXPECT_CH(ch) SB \
396 if (*dec->cur != ch) \ 526 if (*dec->cur != ch) \
397 ERR (# ch " expected"); \ 527 ERR (# ch " expected"); \
398 ++dec->cur; \ 528 ++dec->cur; \
399 SE 529 SE
400 530
401static SV *decode_sv (dec_t *dec); 531static SV *decode_sv (dec_t *dec);
402 532
403#define APPEND_CH(ch) SB \
404 SvGROW (sv, cur + 1 + 1); \
405 SvPVX (sv)[cur++] = (ch); \
406 SE
407
408static signed char decode_hexdigit[256]; 533static signed char decode_hexdigit[256];
409 534
410static UV 535static UV
411decode_4hex (dec_t *dec) 536decode_4hex (dec_t *dec)
412{ 537{
413 signed char d1, d2, d3, d4; 538 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur;
414 540
415 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
416 if (d1 < 0) ERR ("four hexadecimal digits expected"); 541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
417 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
418 if (d2 < 0) ERR ("four hexadecimal digits expected"); 542 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
419 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
420 if (d3 < 0) ERR ("four hexadecimal digits expected"); 543 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
421 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
422 if (d4 < 0) ERR ("four hexadecimal digits expected"); 544 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
423 545
424 dec->cur += 4; 546 dec->cur += 4;
425 547
426 return ((UV)d1) << 12 548 return ((UV)d1) << 12
427 | ((UV)d2) << 8 549 | ((UV)d2) << 8
… …
433} 555}
434 556
435static SV * 557static SV *
436decode_str (dec_t *dec) 558decode_str (dec_t *dec)
437{ 559{
438 SV *sv = NEWSV (0,2); 560 SV *sv = 0;
439 STRLEN cur = 0;
440 int utf8 = 0; 561 int utf8 = 0;
441 562
442 for (;;) 563 do
443 { 564 {
444 unsigned char ch = *(unsigned char *)dec->cur; 565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
566 char *cur = buf;
445 567
446 if (ch == '"') 568 do
447 break;
448 else if (ch == '\\')
449 { 569 {
450 switch (*++dec->cur) 570 unsigned char ch = *(unsigned char *)dec->cur++;
571
572 if (ch == '"')
451 { 573 {
452 case '\\': 574 --dec->cur;
453 case '/': 575 break;
454 case '"': APPEND_CH (*dec->cur++); break; 576 }
455 577 else if (ch == '\\')
456 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 578 {
457 case 't': APPEND_CH ('\011'); ++dec->cur; break; 579 switch (*dec->cur)
458 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
459 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
460 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
461
462 case 'u':
463 { 580 {
464 UV lo, hi; 581 case '\\':
465 ++dec->cur; 582 case '/':
583 case '"': *cur++ = *dec->cur++; break;
466 584
467 hi = decode_4hex (dec); 585 case 'b': ++dec->cur; *cur++ = '\010'; break;
468 if (hi == (UV)-1) 586 case 't': ++dec->cur; *cur++ = '\011'; break;
469 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;
470 590
471 // possibly a surrogate pair 591 case 'u':
472 if (hi >= 0xd800 && hi < 0xdc00)
473 { 592 {
474 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 593 UV lo, hi;
475 ERR ("illegal surrogate character");
476
477 dec->cur += 2; 594 ++dec->cur;
478 595
479 lo = decode_4hex (dec); 596 hi = decode_4hex (dec);
480 if (lo == (UV)-1) 597 if (hi == (UV)-1)
481 goto fail; 598 goto fail;
482 599
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
483 if (lo < 0xdc00 || lo >= 0xe000) 613 if (lo < 0xdc00 || lo >= 0xe000)
484 ERR ("surrogate pair expected"); 614 ERR ("surrogate pair expected");
485 615
486 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;
487 } 629 }
488 else if (lo >= 0xdc00 && lo < 0xe000)
489 ERR ("illegal surrogate character");
490
491 if (hi >= 0x80)
492 { 630 break;
493 utf8 = 1;
494 631
495 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
496 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
497 }
498 else 632 default:
499 APPEND_CH (hi); 633 --dec->cur;
634 ERR ("illegal backslash escape sequence in string");
500 } 635 }
501 break;
502 } 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
503 } 661 }
504 else if (ch >= 0x20 && ch <= 0x7f) 662 while (cur < buf + SHORT_STRING_LEN);
505 APPEND_CH (*dec->cur++); 663
506 else if (ch >= 0x80) 664 STRLEN len = cur - buf;
665
666 if (sv)
507 { 667 {
508 STRLEN clen; 668 SvGROW (sv, SvCUR (sv) + len + 1);
509 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
510 if (clen < 0) 670 SvCUR_set (sv, SvCUR (sv) + len);
511 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
512
513 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits
514 memcpy (SvPVX (sv) + cur, dec->cur, clen);
515 dec->cur += clen;
516 } 671 }
517 else 672 else
518 ERR ("invalid character encountered"); 673 sv = newSVpvn (buf, len);
519 } 674 }
675 while (*dec->cur != '"');
520 676
521 ++dec->cur; 677 ++dec->cur;
522 678
679 if (sv)
680 {
523 SvPOK_only (sv); 681 SvPOK_only (sv);
524
525 SvCUR_set (sv, cur);
526 *SvEND (sv) = 0; 682 *SvEND (sv) = 0;
527 683
528 if (utf8) 684 if (utf8)
529 SvUTF8_on (sv); 685 SvUTF8_on (sv);
686 }
687 else
688 sv = newSVpvn ("", 0);
530 689
531 return sv; 690 return sv;
532 691
533fail: 692fail:
534 SvREFCNT_dec (sv);
535 return 0; 693 return 0;
536} 694}
537 695
538static SV * 696static SV *
539decode_num (dec_t *dec) 697decode_num (dec_t *dec)
… …
549 { 707 {
550 ++dec->cur; 708 ++dec->cur;
551 if (*dec->cur >= '0' && *dec->cur <= '9') 709 if (*dec->cur >= '0' && *dec->cur <= '9')
552 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)");
553 } 711 }
554 712 else if (*dec->cur < '0' || *dec->cur > '9')
555 // int 713 ERR ("malformed number (no digits after initial minus)");
714 else
715 do
716 {
717 ++dec->cur;
718 }
556 while (*dec->cur >= '0' && *dec->cur <= '9') 719 while (*dec->cur >= '0' && *dec->cur <= '9');
557 ++dec->cur;
558 720
559 // [frac] 721 // [frac]
560 if (*dec->cur == '.') 722 if (*dec->cur == '.')
561 { 723 {
562 is_nv = 1; 724 ++dec->cur;
725
726 if (*dec->cur < '0' || *dec->cur > '9')
727 ERR ("malformed number (no digits after decimal point)");
563 728
564 do 729 do
565 { 730 {
566 ++dec->cur; 731 ++dec->cur;
567 } 732 }
568 while (*dec->cur >= '0' && *dec->cur <= '9'); 733 while (*dec->cur >= '0' && *dec->cur <= '9');
734
735 is_nv = 1;
569 } 736 }
570 737
571 // [exp] 738 // [exp]
572 if (*dec->cur == 'e' || *dec->cur == 'E') 739 if (*dec->cur == 'e' || *dec->cur == 'E')
573 { 740 {
574 is_nv = 1;
575
576 ++dec->cur; 741 ++dec->cur;
742
577 if (*dec->cur == '-' || *dec->cur == '+') 743 if (*dec->cur == '-' || *dec->cur == '+')
578 ++dec->cur; 744 ++dec->cur;
579 745
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 }
580 while (*dec->cur >= '0' && *dec->cur <= '9') 753 while (*dec->cur >= '0' && *dec->cur <= '9');
581 ++dec->cur; 754
755 is_nv = 1;
582 } 756 }
583 757
584 if (!is_nv) 758 if (!is_nv)
585 { 759 {
586 UV uv; 760 UV uv;
… …
604static SV * 778static SV *
605decode_av (dec_t *dec) 779decode_av (dec_t *dec)
606{ 780{
607 AV *av = newAV (); 781 AV *av = newAV ();
608 782
783 decode_ws (dec);
784 if (*dec->cur == ']')
785 ++dec->cur;
786 else
609 for (;;) 787 for (;;)
610 { 788 {
611 SV *value; 789 SV *value;
612 790
613 value = decode_sv (dec); 791 value = decode_sv (dec);
614 if (!value) 792 if (!value)
615 goto fail; 793 goto fail;
616 794
617 av_push (av, value); 795 av_push (av, value);
618 796
619 WS; 797 decode_ws (dec);
620 798
621 if (*dec->cur == ']') 799 if (*dec->cur == ']')
622 { 800 {
623 ++dec->cur; 801 ++dec->cur;
624 break; 802 break;
803 }
625 } 804
626
627 if (*dec->cur != ',') 805 if (*dec->cur != ',')
628 ERR (", or ] expected while parsing array"); 806 ERR (", or ] expected while parsing array");
629 807
630 ++dec->cur; 808 ++dec->cur;
631 } 809 }
632 810
633 return newRV_noinc ((SV *)av); 811 return newRV_noinc ((SV *)av);
634 812
635fail: 813fail:
636 SvREFCNT_dec (av); 814 SvREFCNT_dec (av);
… …
640static SV * 818static SV *
641decode_hv (dec_t *dec) 819decode_hv (dec_t *dec)
642{ 820{
643 HV *hv = newHV (); 821 HV *hv = newHV ();
644 822
823 decode_ws (dec);
824 if (*dec->cur == '}')
825 ++dec->cur;
826 else
645 for (;;) 827 for (;;)
646 { 828 {
647 SV *key, *value; 829 SV *key, *value;
648 830
649 WS; EXPECT_CH ('"'); 831 decode_ws (dec); EXPECT_CH ('"');
650 832
651 key = decode_str (dec); 833 key = decode_str (dec);
652 if (!key) 834 if (!key)
653 goto fail;
654
655 WS; EXPECT_CH (':');
656
657 value = decode_sv (dec);
658 if (!value)
659 {
660 SvREFCNT_dec (key);
661 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;
662 } 844 }
663 845
664 //TODO: optimise 846 //TODO: optimise
665 hv_store_ent (hv, key, value, 0); 847 hv_store_ent (hv, key, value, 0);
666 848
667 WS; 849 decode_ws (dec);
668 850
669 if (*dec->cur == '}') 851 if (*dec->cur == '}')
670 { 852 {
671 ++dec->cur; 853 ++dec->cur;
672 break; 854 break;
673 } 855 }
674 856
675 if (*dec->cur != ',') 857 if (*dec->cur != ',')
676 ERR (", or } expected while parsing object/hash"); 858 ERR (", or } expected while parsing object/hash");
677 859
678 ++dec->cur; 860 ++dec->cur;
679 } 861 }
680 862
681 return newRV_noinc ((SV *)hv); 863 return newRV_noinc ((SV *)hv);
682 864
683fail: 865fail:
684 SvREFCNT_dec (hv); 866 SvREFCNT_dec (hv);
… …
686} 868}
687 869
688static SV * 870static SV *
689decode_sv (dec_t *dec) 871decode_sv (dec_t *dec)
690{ 872{
691 WS; 873 decode_ws (dec);
692 switch (*dec->cur) 874 switch (*dec->cur)
693 { 875 {
694 case '"': ++dec->cur; return decode_str (dec); 876 case '"': ++dec->cur; return decode_str (dec);
695 case '[': ++dec->cur; return decode_av (dec); 877 case '[': ++dec->cur; return decode_av (dec);
696 case '{': ++dec->cur; return decode_hv (dec); 878 case '{': ++dec->cur; return decode_hv (dec);
… …
724 906
725 case 'n': 907 case 'n':
726 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 908 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
727 { 909 {
728 dec->cur += 4; 910 dec->cur += 4;
729 return newSViv (1); 911 return newSVsv (&PL_sv_undef);
730 } 912 }
731 else 913 else
732 ERR ("'null' expected"); 914 ERR ("'null' expected");
733 915
734 break; 916 break;
735 917
736 default: 918 default:
737 ERR ("malformed json string"); 919 ERR ("malformed json string, neither array, object, number, string or atom");
738 break; 920 break;
739 } 921 }
740 922
741fail: 923fail:
742 return 0; 924 return 0;
… …
745static SV * 927static SV *
746decode_json (SV *string, UV flags) 928decode_json (SV *string, UV flags)
747{ 929{
748 SV *sv; 930 SV *sv;
749 931
750 if (!(flags & F_UTF8)) 932 if (flags & F_UTF8)
933 sv_utf8_downgrade (string, 0);
934 else
751 sv_utf8_upgrade (string); 935 sv_utf8_upgrade (string);
752 936
753 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
754 938
755 dec_t dec; 939 dec_t dec;
756 dec.flags = flags; 940 dec.flags = flags;
757 dec.cur = SvPVX (string); 941 dec.cur = SvPVX (string);
758 dec.end = SvEND (string); 942 dec.end = SvEND (string);
759 dec.err = 0; 943 dec.err = 0;
760 944
761 *dec.end = 1; // invalid anywhere
762 sv = decode_sv (&dec); 945 sv = decode_sv (&dec);
763 *dec.end = 0;
764 946
765 if (!sv) 947 if (!sv)
766 { 948 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
767 IV offset = utf8_distance (dec.cur, SvPVX (string)); 951 : utf8_distance (dec.cur, SvPVX (string));
768 SV *uni = sv_newmortal (); 952 SV *uni = sv_newmortal ();
769 953
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;
770 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
771 croak ("%s, at character %d (%s)", 963 croak ("%s, at character offset %d (%s)",
772 dec.err, 964 dec.err,
773 (int)offset, 965 (int)offset,
774 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
775 } 967 }
776 968
777 sv_dump (sv);//D
778 return sv_2mortal (sv); 969 sv = sv_2mortal (sv);
970
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
972 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973
974 return sv;
779} 975}
976
977/////////////////////////////////////////////////////////////////////////////
978// XS interface functions
780 979
781MODULE = JSON::XS PACKAGE = JSON::XS 980MODULE = JSON::XS PACKAGE = JSON::XS
782 981
783BOOT: 982BOOT:
784{ 983{
… …
786 985
787 memset (decode_hexdigit, 0xff, 256); 986 memset (decode_hexdigit, 0xff, 256);
788 for (i = 10; i--; ) 987 for (i = 10; i--; )
789 decode_hexdigit ['0' + i] = i; 988 decode_hexdigit ['0' + i] = i;
790 989
791 for (i = 6; --i; ) 990 for (i = 7; i--; )
792 { 991 {
793 decode_hexdigit ['a' + i] = 10 + i; 992 decode_hexdigit ['a' + i] = 10 + i;
794 decode_hexdigit ['A' + i] = 10 + i; 993 decode_hexdigit ['A' + i] = 10 + i;
795 } 994 }
796 995
797 json_stash = gv_stashpv ("JSON::XS", 1); 996 json_stash = gv_stashpv ("JSON::XS", 1);
798} 997}
998
999PROTOTYPES: DISABLE
799 1000
800SV *new (char *dummy) 1001SV *new (char *dummy)
801 CODE: 1002 CODE:
802 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
803 OUTPUT: 1004 OUTPUT:
804 RETVAL 1005 RETVAL
805 1006
806SV *ascii (SV *self, int enable) 1007SV *ascii (SV *self, int enable = 1)
807 ALIAS: 1008 ALIAS:
808 ascii = F_ASCII 1009 ascii = F_ASCII
809 utf8 = F_UTF8 1010 utf8 = F_UTF8
810 indent = F_INDENT 1011 indent = F_INDENT
811 canonical = F_CANONICAL 1012 canonical = F_CANONICAL
812 space_before = F_SPACE_BEFORE 1013 space_before = F_SPACE_BEFORE
813 space_after = F_SPACE_AFTER 1014 space_after = F_SPACE_AFTER
814 json_rpc = F_JSON_RPC
815 pretty = F_PRETTY 1015 pretty = F_PRETTY
1016 allow_nonref = F_ALLOW_NONREF
1017 shrink = F_SHRINK
816 CODE: 1018 CODE:
817{ 1019{
818 UV *uv = SvJSON (self); 1020 UV *uv = SvJSON (self);
819 if (enable) 1021 if (enable)
820 *uv |= ix; 1022 *uv |= ix;
… …
832 1034
833void decode (SV *self, SV *jsonstr) 1035void decode (SV *self, SV *jsonstr)
834 PPCODE: 1036 PPCODE:
835 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1037 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
836 1038
1039PROTOTYPES: ENABLE
1040
837void to_json (SV *scalar) 1041void to_json (SV *scalar)
838 PPCODE: 1042 PPCODE:
839 XPUSHs (encode_json (scalar, F_UTF8)); 1043 XPUSHs (encode_json (scalar, F_UTF8));
840 1044
841void from_json (SV *jsonstr) 1045void from_json (SV *jsonstr)

Diff Legend

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