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 1.1 by root, Thu Mar 22 16:40:16 2007 UTC vs.
Revision 1.18 by root, Sun Mar 25 21:19:13 2007 UTC

… …
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8 8
9#define F_ASCII 0x00000001 9#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020UL
15#define F_ALLOW_NONREF 0x00000080UL
15#define F_JSON_RPC 0x00000040 16#define F_SHRINK 0x00000100UL
17#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27
16 19
17#define F_DEFAULT 0 20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (13UL << S_MAXDEPTH)
18 27
19#define INIT_SIZE 32 // initial scalar size to be allocated 28#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level
30
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size
20 33
21#define SB do { 34#define SB do {
22#define SE } while (0) 35#define SE } while (0)
23 36
24static HV *json_stash; 37static HV *json_stash; // JSON::XS::
38
39/////////////////////////////////////////////////////////////////////////////
40// utility functions
41
42static UV *
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv)
53{
54 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1)
56 {
57#ifdef SvPV_shrink_to_cur
58 SvPV_shrink_to_cur (sv);
59#elif defined (SvPV_renew)
60 SvPV_renew (sv, SvCUR (sv) + 1);
61#endif
62 }
63}
64
65// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints
70static UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{
73 if (s[0] > 0xdf || s[0] < 0xc2)
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 {
77 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 }
80 else
81 return (UV)-1;
82}
83
84/////////////////////////////////////////////////////////////////////////////
85// encoder
25 86
26// structure used for encoding JSON 87// structure used for encoding JSON
27typedef struct 88typedef struct
28{ 89{
29 char *cur; 90 char *cur; // SvPVX (sv) + current output position
30 STRLEN len; // SvLEN (sv)
31 char *end; // SvEND (sv) 91 char *end; // SvEND (sv)
32 SV *sv; 92 SV *sv; // result scalar
33 UV flags; 93 U32 flags; // F_*
34 int max_recurse; 94 U32 indent; // indentation level
35 int indent; 95 U32 maxdepth; // max. indentation/recursion level
36} enc_t; 96} enc_t;
37
38// structure used for decoding JSON
39typedef struct
40{
41 char *cur;
42 char *end;
43 char *err;
44 UV flags;
45} dec_t;
46
47static UV *
48SvJSON (SV *sv)
49{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
51 croak ("object is not of type JSON::XS");
52
53 return &SvUVX (SvRV (sv));
54}
55
56/////////////////////////////////////////////////////////////////////////////
57 97
58static void 98static void
59need (enc_t *enc, STRLEN len) 99need (enc_t *enc, STRLEN len)
60{ 100{
61 if (enc->cur + len >= enc->end) 101 if (enc->cur + len >= enc->end)
62 { 102 {
63 STRLEN cur = enc->cur - SvPVX (enc->sv); 103 STRLEN cur = enc->cur - SvPVX (enc->sv);
64 SvGROW (enc->sv, cur + len + 1); 104 SvGROW (enc->sv, cur + len + 1);
65 enc->cur = SvPVX (enc->sv) + cur; 105 enc->cur = SvPVX (enc->sv) + cur;
66 enc->end = SvEND (enc->sv); 106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);
67 } 107 }
68} 108}
69 109
70static void 110static void
71encode_ch (enc_t *enc, char ch) 111encode_ch (enc_t *enc, char ch)
… …
77static void 117static void
78encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 118encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
79{ 119{
80 char *end = str + len; 120 char *end = str + len;
81 121
122 need (enc, len);
123
82 while (str < end) 124 while (str < end)
83 { 125 {
84 unsigned char ch = *(unsigned char *)str; 126 unsigned char ch = *(unsigned char *)str;
127
85 if (ch >= 0x20 && ch < 0x80) // most common case 128 if (ch >= 0x20 && ch < 0x80) // most common case
86 { 129 {
130 if (ch == '"') // but with slow exceptions
131 {
132 need (enc, len += 1);
133 *enc->cur++ = '\\';
134 *enc->cur++ = '"';
135 }
136 else if (ch == '\\')
137 {
138 need (enc, len += 1);
139 *enc->cur++ = '\\';
140 *enc->cur++ = '\\';
141 }
142 else
87 *enc->cur++ = ch; 143 *enc->cur++ = ch;
144
88 str++; 145 ++str;
89 } 146 }
90 else 147 else
91 { 148 {
92 STRLEN clen; 149 switch (ch)
93 UV uch;
94
95 if (is_utf8)
96 { 150 {
97 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 151 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
98 if (clen < 0) 152 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
99 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 153 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
100 } 154 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
101 else 155 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
102 {
103 uch = ch;
104 clen = 1;
105 }
106 156
107 need (enc, len += 6); 157 default:
108
109 if (uch < 0xa0 || enc->flags & F_ASCII)
110 {
111 if (uch > 0xFFFFUL)
112 { 158 {
159 STRLEN clen;
160 UV uch;
161
162 if (is_utf8)
163 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1)
167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
168 }
169 else
170 {
171 uch = ch;
113 len += 6; 172 clen = 1;
173 }
174
175 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177
178 if (uch < 0x80 || enc->flags & F_ASCII)
179 {
180 if (uch > 0xFFFFUL)
181 {
114 need (enc, len += 6); 182 need (enc, len += 11);
115 sprintf (enc->cur, "\\u%04x\\u%04x", 183 sprintf (enc->cur, "\\u%04x\\u%04x",
116 (uch - 0x10000) / 0x400 + 0xD800, 184 (int)((uch - 0x10000) / 0x400 + 0xD800),
117 (uch - 0x10000) % 0x400 + 0xDC00); 185 (int)((uch - 0x10000) % 0x400 + 0xDC00));
118 enc->cur += 12; 186 enc->cur += 12;
187 }
188 else
189 {
190 static char hexdigit [16] = "0123456789abcdef";
191 need (enc, len += 5);
192 *enc->cur++ = '\\';
193 *enc->cur++ = 'u';
194 *enc->cur++ = hexdigit [ uch >> 12 ];
195 *enc->cur++ = hexdigit [(uch >> 8) & 15];
196 *enc->cur++ = hexdigit [(uch >> 4) & 15];
197 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 }
199
200 str += clen;
119 } 201 }
202 else if (is_utf8)
203 {
204 need (enc, len += clen);
205 do
206 {
207 *enc->cur++ = *str++;
208 }
209 while (--clen);
210 }
120 else 211 else
121 { 212 {
122 sprintf (enc->cur, "\\u%04x", uch); 213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
123 enc->cur += 6; 214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str;
216 }
124 } 217 }
125 } 218 }
126 else if (is_utf8)
127 {
128 memcpy (enc->cur, str, clen);
129 enc->cur += clen;
130 }
131 else
132 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
133
134 str += clen;
135 } 219 }
136 220
137 --len; 221 --len;
138 } 222 }
139} 223}
140 224
141#define INDENT SB \ 225static void
226encode_indent (enc_t *enc)
227{
142 if (enc->flags & F_INDENT) \ 228 if (enc->flags & F_INDENT)
143 { \ 229 {
144 int i_; \ 230 int spaces = enc->indent * INDENT_STEP;
145 need (enc, enc->indent); \
146 for (i_ = enc->indent * 3; i_--; )\
147 encode_ch (enc, ' '); \
148 } \
149 SE
150 231
151#define SPACE SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, ' '); } SE 232 need (enc, spaces);
152#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 233 memset (enc->cur, ' ', spaces);
153#define COMMA SB \ 234 enc->cur += spaces;
235 }
236}
237
238static void
239encode_space (enc_t *enc)
240{
241 need (enc, 1);
154 encode_ch (enc, ','); \ 242 encode_ch (enc, ' ');
243}
244
245static void
246encode_nl (enc_t *enc)
247{
155 if (enc->flags & F_INDENT) \ 248 if (enc->flags & F_INDENT)
156 NL; \ 249 {
250 need (enc, 1);
251 encode_ch (enc, '\n');
252 }
253}
254
255static void
256encode_comma (enc_t *enc)
257{
258 encode_ch (enc, ',');
259
260 if (enc->flags & F_INDENT)
261 encode_nl (enc);
157 else if (enc->flags & F_SPACE_AFTER) \ 262 else if (enc->flags & F_SPACE_AFTER)
158 SPACE; \ 263 encode_space (enc);
159 SE 264}
160 265
161static void encode_sv (enc_t *enc, SV *sv); 266static void encode_sv (enc_t *enc, SV *sv);
162 267
163static void 268static void
164encode_av (enc_t *enc, AV *av) 269encode_av (enc_t *enc, AV *av)
165{ 270{
166 int i, len = av_len (av); 271 int i, len = av_len (av);
167 272
168 encode_ch (enc, '['); NL; 273 encode_ch (enc, '['); encode_nl (enc);
169 ++enc->indent; 274 ++enc->indent;
170 275
171 for (i = 0; i <= len; ++i) 276 for (i = 0; i <= len; ++i)
172 { 277 {
173 INDENT; 278 encode_indent (enc);
174 encode_sv (enc, *av_fetch (av, i, 0)); 279 encode_sv (enc, *av_fetch (av, i, 0));
175 280
176 if (i < len) 281 if (i < len)
177 COMMA; 282 encode_comma (enc);
178 } 283 }
179 284
180 NL; 285 encode_nl (enc);
181 286
182 --enc->indent; 287 --enc->indent;
183 INDENT; encode_ch (enc, ']'); 288 encode_indent (enc); encode_ch (enc, ']');
184} 289}
185 290
186static void 291static void
187encode_he (enc_t *enc, HE *he) 292encode_he (enc_t *enc, HE *he)
188{ 293{
… …
190 295
191 if (HeKLEN (he) == HEf_SVKEY) 296 if (HeKLEN (he) == HEf_SVKEY)
192 { 297 {
193 SV *sv = HeSVKEY (he); 298 SV *sv = HeSVKEY (he);
194 STRLEN len; 299 STRLEN len;
300 char *str;
301
302 SvGETMAGIC (sv);
195 char *str = SvPV (sv, len); 303 str = SvPV (sv, len);
196 304
197 encode_str (enc, str, len, SvUTF8 (sv)); 305 encode_str (enc, str, len, SvUTF8 (sv));
198 } 306 }
199 else 307 else
200 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
201 309
202 encode_ch (enc, '"'); 310 encode_ch (enc, '"');
203 311
204 if (enc->flags & F_SPACE_BEFORE) SPACE; 312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
205 encode_ch (enc, ':'); 313 encode_ch (enc, ':');
206 if (enc->flags & F_SPACE_AFTER ) SPACE; 314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
207 encode_sv (enc, HeVAL (he)); 315 encode_sv (enc, HeVAL (he));
208} 316}
209 317
210// compare hash entries, used when all keys are bytestrings 318// compare hash entries, used when all keys are bytestrings
211static int 319static int
… …
217 HE *b = *(HE **)b_; 325 HE *b = *(HE **)b_;
218 326
219 STRLEN la = HeKLEN (a); 327 STRLEN la = HeKLEN (a);
220 STRLEN lb = HeKLEN (b); 328 STRLEN lb = HeKLEN (b);
221 329
222 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 330 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
223 cmp = la < lb ? -1 : la == lb ? 0 : 1; 331 cmp = la - lb;
224 332
225 return cmp; 333 return cmp;
226} 334}
227 335
228// compare hash entries, used when some keys are sv's or utf-x 336// compare hash entries, used when some keys are sv's or utf-x
… …
235static void 343static void
236encode_hv (enc_t *enc, HV *hv) 344encode_hv (enc_t *enc, HV *hv)
237{ 345{
238 int count, i; 346 int count, i;
239 347
240 encode_ch (enc, '{'); NL; ++enc->indent; 348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
241 349
242 if ((count = hv_iterinit (hv))) 350 if ((count = hv_iterinit (hv)))
243 { 351 {
244 // for canonical output we have to sort by keys first 352 // for canonical output we have to sort by keys first
245 // actually, this is mostly due to the stupid so-called 353 // actually, this is mostly due to the stupid so-called
246 // security workaround added somewhere in 5.8.x. 354 // security workaround added somewhere in 5.8.x.
247 // that randomises hash orderings 355 // that randomises hash orderings
248 if (enc->flags & F_CANONICAL) 356 if (enc->flags & F_CANONICAL)
249 { 357 {
250 HE *he, *hes [count]; 358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
251 int fast = 1; 359 int fast = 1;
252 360
253 i = 0; 361 i = 0;
254 while ((he = hv_iternext (hv))) 362 while ((he = hv_iternext (hv)))
255 { 363 {
… …
262 370
263 if (fast) 371 if (fast)
264 qsort (hes, count, sizeof (HE *), he_cmp_fast); 372 qsort (hes, count, sizeof (HE *), he_cmp_fast);
265 else 373 else
266 { 374 {
267 // hack to disable "use bytes" 375 // hack to forcefully disable "use bytes"
268 COP *oldcop = PL_curcop, cop; 376 COP cop = *PL_curcop;
269 cop.op_private = 0; 377 cop.op_private = 0;
378
379 ENTER;
380 SAVETMPS;
381
382 SAVEVPTR (PL_curcop);
270 PL_curcop = &cop; 383 PL_curcop = &cop;
271 384
272 SAVETMPS;
273 qsort (hes, count, sizeof (HE *), he_cmp_slow); 385 qsort (hes, count, sizeof (HE *), he_cmp_slow);
386
274 FREETMPS; 387 FREETMPS;
275 388 LEAVE;
276 PL_curcop = oldcop;
277 } 389 }
278 390
279 for (i = 0; i < count; ++i) 391 for (i = 0; i < count; ++i)
280 { 392 {
281 INDENT; 393 encode_indent (enc);
282 encode_he (enc, hes [i]); 394 encode_he (enc, hes [i]);
283 395
284 if (i < count - 1) 396 if (i < count - 1)
285 COMMA; 397 encode_comma (enc);
286 } 398 }
287 399
288 NL; 400 encode_nl (enc);
289 } 401 }
290 else 402 else
291 { 403 {
292 SV *sv; 404 SV *sv;
293 HE *he = hv_iternext (hv); 405 HE *he = hv_iternext (hv);
294 406
295 for (;;) 407 for (;;)
296 { 408 {
297 INDENT; 409 encode_indent (enc);
298 encode_he (enc, he); 410 encode_he (enc, he);
299 411
300 if (!(he = hv_iternext (hv))) 412 if (!(he = hv_iternext (hv)))
301 break; 413 break;
302 414
303 COMMA; 415 encode_comma (enc);
304 } 416 }
305 417
306 NL; 418 encode_nl (enc);
307 } 419 }
308 } 420 }
309 421
310 --enc->indent; INDENT; encode_ch (enc, '}'); 422 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
311} 423}
312 424
313static void 425static void
314encode_sv (enc_t *enc, SV *sv) 426encode_sv (enc_t *enc, SV *sv)
315{ 427{
428 SvGETMAGIC (sv);
429
316 if (SvPOKp (sv)) 430 if (SvPOKp (sv))
317 { 431 {
318 STRLEN len; 432 STRLEN len;
319 char *str = SvPV (sv, len); 433 char *str = SvPV (sv, len);
320 encode_ch (enc, '"'); 434 encode_ch (enc, '"');
… …
335 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
336 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
337 } 451 }
338 else if (SvROK (sv)) 452 else if (SvROK (sv))
339 { 453 {
340 if (!--enc->max_recurse) 454 SV *rv = SvRV (sv);
455
456 if (enc->indent >= enc->maxdepth)
341 croak ("data structure too deep (hit recursion limit)"); 457 croak ("data structure too deep (hit recursion limit)");
342 458
343 sv = SvRV (sv);
344
345 switch (SvTYPE (sv)) 459 switch (SvTYPE (rv))
346 { 460 {
347 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
348 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 462 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
349 463
350 default: 464 default:
351 croak ("JSON can only represent references to arrays or hashes"); 465 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
466 SvPV_nolen (sv));
352 } 467 }
353 } 468 }
354 else if (!SvOK (sv)) 469 else if (!SvOK (sv))
355 encode_str (enc, "null", 4, 0); 470 encode_str (enc, "null", 4, 0);
356 else 471 else
357 croak ("encountered perl type that JSON cannot handle"); 472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv));
358} 474}
359 475
360static SV * 476static SV *
361encode_json (SV *scalar, UV flags) 477encode_json (SV *scalar, U32 flags)
362{ 478{
479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
481
363 enc_t enc; 482 enc_t enc;
364 enc.flags = flags; 483 enc.flags = flags;
365 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
366 enc.cur = SvPVX (enc.sv); 485 enc.cur = SvPVX (enc.sv);
367 enc.end = SvEND (enc.sv); 486 enc.end = SvEND (enc.sv);
368 enc.max_recurse = 0;
369 enc.indent = 0; 487 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags);
370 489
371 SvPOK_only (enc.sv); 490 SvPOK_only (enc.sv);
372 encode_sv (&enc, scalar); 491 encode_sv (&enc, scalar);
373 492
374 if (!(flags & (F_ASCII | F_UTF8))) 493 if (!(flags & (F_ASCII | F_UTF8)))
375 SvUTF8_on (enc.sv); 494 SvUTF8_on (enc.sv);
376 495
377 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497
498 if (enc.flags & F_SHRINK)
499 shrink (enc.sv);
500
378 return enc.sv; 501 return enc.sv;
379} 502}
380 503
381///////////////////////////////////////////////////////////////////////////// 504/////////////////////////////////////////////////////////////////////////////
505// decoder
382 506
383#define WS \ 507// structure used for decoding JSON
508typedef struct
509{
510 char *cur; // current parser pointer
511 char *end; // end of input string
512 const char *err; // parse error, if != 0
513 U32 flags; // F_*
514 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit
516} dec_t;
517
518static void
519decode_ws (dec_t *dec)
520{
384 for (;;) \ 521 for (;;)
385 { \ 522 {
386 char ch = *dec->cur; \ 523 char ch = *dec->cur;
524
387 if (ch > 0x20 \ 525 if (ch > 0x20
388 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 526 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
389 break; \ 527 break;
528
390 ++dec->cur; \ 529 ++dec->cur;
391 } 530 }
531}
392 532
393#define ERR(reason) SB dec->err = reason; goto fail; SE 533#define ERR(reason) SB dec->err = reason; goto fail; SE
534
394#define EXPECT_CH(ch) SB \ 535#define EXPECT_CH(ch) SB \
395 if (*dec->cur != ch) \ 536 if (*dec->cur != ch) \
396 ERR (# ch " expected"); \ 537 ERR (# ch " expected"); \
397 ++dec->cur; \ 538 ++dec->cur; \
398 SE 539 SE
399 540
541#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
542#define DEC_DEC_DEPTH --dec->depth
543
400static SV *decode_sv (dec_t *dec); 544static SV *decode_sv (dec_t *dec);
401
402#define APPEND_CH(ch) SB \
403 SvGROW (sv, cur + 1 + 1); \
404 SvPVX (sv)[cur++] = (ch); \
405 SE
406 545
407static signed char decode_hexdigit[256]; 546static signed char decode_hexdigit[256];
408 547
409static UV 548static UV
410decode_4hex (dec_t *dec) 549decode_4hex (dec_t *dec)
411{ 550{
412 signed char d1, d2, d3, d4; 551 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur;
413 553
414 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
415 if (d1 < 0) ERR ("four hexadecimal digits expected"); 554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
416 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
417 if (d2 < 0) ERR ("four hexadecimal digits expected"); 555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
418 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
419 if (d3 < 0) ERR ("four hexadecimal digits expected"); 556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
420 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
421 if (d4 < 0) ERR ("four hexadecimal digits expected"); 557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
422 558
423 dec->cur += 4; 559 dec->cur += 4;
424 560
425 return ((UV)d1) << 12 561 return ((UV)d1) << 12
426 | ((UV)d2) << 8 562 | ((UV)d2) << 8
… …
432} 568}
433 569
434static SV * 570static SV *
435decode_str (dec_t *dec) 571decode_str (dec_t *dec)
436{ 572{
437 SV *sv = NEWSV (0,2); 573 SV *sv = 0;
438 STRLEN cur = 0;
439 int utf8 = 0; 574 int utf8 = 0;
440 575
441 for (;;) 576 do
442 { 577 {
443 unsigned char ch = *(unsigned char *)dec->cur; 578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
579 char *cur = buf;
444 580
445 if (ch == '"') 581 do
446 break;
447 else if (ch == '\\')
448 { 582 {
449 switch (*++dec->cur) 583 unsigned char ch = *(unsigned char *)dec->cur++;
584
585 if (ch == '"')
450 { 586 {
451 case '\\': 587 --dec->cur;
452 case '/': 588 break;
453 case '"': APPEND_CH (*dec->cur++); break; 589 }
454 590 else if (ch == '\\')
455 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 591 {
456 case 't': APPEND_CH ('\011'); ++dec->cur; break; 592 switch (*dec->cur)
457 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
458 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
459 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
460
461 case 'u':
462 { 593 {
463 UV lo, hi; 594 case '\\':
464 ++dec->cur; 595 case '/':
596 case '"': *cur++ = *dec->cur++; break;
465 597
466 hi = decode_4hex (dec); 598 case 'b': ++dec->cur; *cur++ = '\010'; break;
467 if (hi == (UV)-1) 599 case 't': ++dec->cur; *cur++ = '\011'; break;
468 goto fail; 600 case 'n': ++dec->cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break;
469 603
470 // possibly a surrogate pair 604 case 'u':
471 if (hi >= 0xd800 && hi < 0xdc00)
472 { 605 {
473 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 606 UV lo, hi;
474 ERR ("illegal surrogate character");
475
476 dec->cur += 2; 607 ++dec->cur;
477 608
478 lo = decode_4hex (dec); 609 hi = decode_4hex (dec);
479 if (lo == (UV)-1) 610 if (hi == (UV)-1)
480 goto fail; 611 goto fail;
481 612
613 // possibly a surrogate pair
614 if (hi >= 0xd800)
615 if (hi < 0xdc00)
616 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair");
619
620 dec->cur += 2;
621
622 lo = decode_4hex (dec);
623 if (lo == (UV)-1)
624 goto fail;
625
482 if (lo < 0xdc00 || lo >= 0xe000) 626 if (lo < 0xdc00 || lo >= 0xe000)
483 ERR ("surrogate pair expected"); 627 ERR ("surrogate pair expected");
484 628
485 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 629 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
630 }
631 else if (hi < 0xe000)
632 ERR ("missing high surrogate character in surrogate pair");
633
634 if (hi >= 0x80)
635 {
636 utf8 = 1;
637
638 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
639 }
640 else
641 *cur++ = hi;
486 } 642 }
487 else if (lo >= 0xdc00 && lo < 0xe000)
488 ERR ("illegal surrogate character");
489
490 if (hi >= 0x80)
491 { 643 break;
492 utf8 = 1;
493 644
494 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
495 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
496 }
497 else 645 default:
498 APPEND_CH (hi); 646 --dec->cur;
647 ERR ("illegal backslash escape sequence in string");
499 } 648 }
500 break;
501 } 649 }
650 else if (ch >= 0x20 && ch <= 0x7f)
651 *cur++ = ch;
652 else if (ch >= 0x80)
653 {
654 --dec->cur;
655
656 STRLEN clen;
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
658 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string");
660
661 do
662 {
663 *cur++ = *dec->cur++;
664 }
665 while (--clen);
666
667 utf8 = 1;
668 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else
672 ERR ("invalid character encountered");
673
502 } 674 }
503 else if (ch >= 0x20 && ch <= 0x7f) 675 while (cur < buf + SHORT_STRING_LEN);
504 APPEND_CH (*dec->cur++); 676
505 else if (ch >= 0x80) 677 STRLEN len = cur - buf;
678
679 if (sv)
506 { 680 {
507 STRLEN clen; 681 SvGROW (sv, SvCUR (sv) + len + 1);
508 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
509 if (clen < 0) 683 SvCUR_set (sv, SvCUR (sv) + len);
510 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
511
512 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits
513 memcpy (SvPVX (sv) + cur, dec->cur, clen);
514 dec->cur += clen;
515 } 684 }
516 else 685 else
517 ERR ("invalid character encountered"); 686 sv = newSVpvn (buf, len);
518 } 687 }
688 while (*dec->cur != '"');
519 689
520 ++dec->cur; 690 ++dec->cur;
521 691
692 if (sv)
693 {
522 SvPOK_only (sv); 694 SvPOK_only (sv);
523
524 SvCUR_set (sv, cur);
525 *SvEND (sv) = 0; 695 *SvEND (sv) = 0;
526 696
527 if (utf8) 697 if (utf8)
528 SvUTF8_on (sv); 698 SvUTF8_on (sv);
699 }
700 else
701 sv = newSVpvn ("", 0);
529 702
530 return sv; 703 return sv;
531 704
532fail: 705fail:
533 SvREFCNT_dec (sv);
534 return 0; 706 return 0;
535} 707}
536 708
537static SV * 709static SV *
538decode_num (dec_t *dec) 710decode_num (dec_t *dec)
… …
548 { 720 {
549 ++dec->cur; 721 ++dec->cur;
550 if (*dec->cur >= '0' && *dec->cur <= '9') 722 if (*dec->cur >= '0' && *dec->cur <= '9')
551 ERR ("malformed number (leading zero must not be followed by another digit)"); 723 ERR ("malformed number (leading zero must not be followed by another digit)");
552 } 724 }
553 725 else if (*dec->cur < '0' || *dec->cur > '9')
554 // int 726 ERR ("malformed number (no digits after initial minus)");
727 else
728 do
729 {
730 ++dec->cur;
731 }
555 while (*dec->cur >= '0' && *dec->cur <= '9') 732 while (*dec->cur >= '0' && *dec->cur <= '9');
556 ++dec->cur;
557 733
558 // [frac] 734 // [frac]
559 if (*dec->cur == '.') 735 if (*dec->cur == '.')
560 { 736 {
561 is_nv = 1; 737 ++dec->cur;
738
739 if (*dec->cur < '0' || *dec->cur > '9')
740 ERR ("malformed number (no digits after decimal point)");
562 741
563 do 742 do
564 { 743 {
565 ++dec->cur; 744 ++dec->cur;
566 } 745 }
567 while (*dec->cur >= '0' && *dec->cur <= '9'); 746 while (*dec->cur >= '0' && *dec->cur <= '9');
747
748 is_nv = 1;
568 } 749 }
569 750
570 // [exp] 751 // [exp]
571 if (*dec->cur == 'e' || *dec->cur == 'E') 752 if (*dec->cur == 'e' || *dec->cur == 'E')
572 { 753 {
573 is_nv = 1;
574
575 ++dec->cur; 754 ++dec->cur;
755
576 if (*dec->cur == '-' || *dec->cur == '+') 756 if (*dec->cur == '-' || *dec->cur == '+')
577 ++dec->cur; 757 ++dec->cur;
578 758
759 if (*dec->cur < '0' || *dec->cur > '9')
760 ERR ("malformed number (no digits after exp sign)");
761
762 do
763 {
764 ++dec->cur;
765 }
579 while (*dec->cur >= '0' && *dec->cur <= '9') 766 while (*dec->cur >= '0' && *dec->cur <= '9');
580 ++dec->cur; 767
768 is_nv = 1;
581 } 769 }
582 770
583 if (!is_nv) 771 if (!is_nv)
584 { 772 {
585 UV uv; 773 UV uv;
… …
603static SV * 791static SV *
604decode_av (dec_t *dec) 792decode_av (dec_t *dec)
605{ 793{
606 AV *av = newAV (); 794 AV *av = newAV ();
607 795
796 DEC_INC_DEPTH;
797 decode_ws (dec);
798
799 if (*dec->cur == ']')
800 ++dec->cur;
801 else
608 for (;;) 802 for (;;)
609 { 803 {
610 SV *value; 804 SV *value;
611 805
612 value = decode_sv (dec); 806 value = decode_sv (dec);
613 if (!value) 807 if (!value)
614 goto fail; 808 goto fail;
615 809
616 av_push (av, value); 810 av_push (av, value);
617 811
618 WS; 812 decode_ws (dec);
619 813
620 if (*dec->cur == ']') 814 if (*dec->cur == ']')
621 { 815 {
622 ++dec->cur; 816 ++dec->cur;
623 break; 817 break;
818 }
624 } 819
625
626 if (*dec->cur != ',') 820 if (*dec->cur != ',')
627 ERR (", or ] expected while parsing array"); 821 ERR (", or ] expected while parsing array");
628 822
629 ++dec->cur; 823 ++dec->cur;
630 } 824 }
631 825
826 DEC_DEC_DEPTH;
632 return newRV_noinc ((SV *)av); 827 return newRV_noinc ((SV *)av);
633 828
634fail: 829fail:
635 SvREFCNT_dec (av); 830 SvREFCNT_dec (av);
831 DEC_DEC_DEPTH;
636 return 0; 832 return 0;
637} 833}
638 834
639static SV * 835static SV *
640decode_hv (dec_t *dec) 836decode_hv (dec_t *dec)
641{ 837{
642 HV *hv = newHV (); 838 HV *hv = newHV ();
643 839
840 DEC_INC_DEPTH;
841 decode_ws (dec);
842
843 if (*dec->cur == '}')
844 ++dec->cur;
845 else
644 for (;;) 846 for (;;)
645 { 847 {
646 SV *key, *value; 848 SV *key, *value;
647 849
648 WS; EXPECT_CH ('"'); 850 decode_ws (dec); EXPECT_CH ('"');
649 851
650 key = decode_str (dec); 852 key = decode_str (dec);
651 if (!key) 853 if (!key)
652 goto fail;
653
654 WS; EXPECT_CH (':');
655
656 value = decode_sv (dec);
657 if (!value)
658 {
659 SvREFCNT_dec (key);
660 goto fail; 854 goto fail;
855
856 decode_ws (dec); EXPECT_CH (':');
857
858 value = decode_sv (dec);
859 if (!value)
860 {
861 SvREFCNT_dec (key);
862 goto fail;
661 } 863 }
662 864
663 //TODO: optimise
664 hv_store_ent (hv, key, value, 0); 865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
665 867
666 WS; 868 decode_ws (dec);
667 869
668 if (*dec->cur == '}') 870 if (*dec->cur == '}')
669 { 871 {
670 ++dec->cur; 872 ++dec->cur;
671 break; 873 break;
672 } 874 }
673 875
674 if (*dec->cur != ',') 876 if (*dec->cur != ',')
675 ERR (", or } expected while parsing object/hash"); 877 ERR (", or } expected while parsing object/hash");
676 878
677 ++dec->cur; 879 ++dec->cur;
678 } 880 }
679 881
882 DEC_DEC_DEPTH;
680 return newRV_noinc ((SV *)hv); 883 return newRV_noinc ((SV *)hv);
681 884
682fail: 885fail:
683 SvREFCNT_dec (hv); 886 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH;
684 return 0; 888 return 0;
685} 889}
686 890
687static SV * 891static SV *
688decode_sv (dec_t *dec) 892decode_sv (dec_t *dec)
689{ 893{
690 WS; 894 decode_ws (dec);
691 switch (*dec->cur) 895 switch (*dec->cur)
692 { 896 {
693 case '"': ++dec->cur; return decode_str (dec); 897 case '"': ++dec->cur; return decode_str (dec);
694 case '[': ++dec->cur; return decode_av (dec); 898 case '[': ++dec->cur; return decode_av (dec);
695 case '{': ++dec->cur; return decode_hv (dec); 899 case '{': ++dec->cur; return decode_hv (dec);
… …
723 927
724 case 'n': 928 case 'n':
725 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 929 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
726 { 930 {
727 dec->cur += 4; 931 dec->cur += 4;
728 return newSViv (1); 932 return newSVsv (&PL_sv_undef);
729 } 933 }
730 else 934 else
731 ERR ("'null' expected"); 935 ERR ("'null' expected");
732 936
733 break; 937 break;
734 938
735 default: 939 default:
736 ERR ("malformed json string"); 940 ERR ("malformed json string, neither array, object, number, string or atom");
737 break; 941 break;
738 } 942 }
739 943
740fail: 944fail:
741 return 0; 945 return 0;
742} 946}
743 947
744static SV * 948static SV *
745decode_json (SV *string, UV flags) 949decode_json (SV *string, U32 flags)
746{ 950{
747 SV *sv; 951 SV *sv;
748 952
749 if (!(flags & F_UTF8)) 953 if (flags & F_UTF8)
954 sv_utf8_downgrade (string, 0);
955 else
750 sv_utf8_upgrade (string); 956 sv_utf8_upgrade (string);
751 957
752 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
753 959
754 dec_t dec; 960 dec_t dec;
755 dec.flags = flags; 961 dec.flags = flags;
756 dec.cur = SvPVX (string); 962 dec.cur = SvPVX (string);
757 dec.end = SvEND (string); 963 dec.end = SvEND (string);
758 dec.err = 0; 964 dec.err = 0;
965 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags);
759 967
760 *dec.end = 1; // invalid anywhere 968 *SvEND (sv) = 0; // this shou[ld basically be a nop, too
761 sv = decode_sv (&dec); 969 sv = decode_sv (&dec);
762 *dec.end = 0;
763 970
764 if (!sv) 971 if (!sv)
765 { 972 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
766 IV offset = utf8_distance (dec.cur, SvPVX (string)); 975 : utf8_distance (dec.cur, SvPVX (string));
767 SV *uni = sv_newmortal (); 976 SV *uni = sv_newmortal ();
768 977
978 // horrible hack to silence warning inside pv_uni_display
979 COP cop = *PL_curcop;
980 cop.cop_warnings = pWARN_NONE;
981 ENTER;
982 SAVEVPTR (PL_curcop);
983 PL_curcop = &cop;
769 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE;
986
770 croak ("%s, at character %d (%s)", 987 croak ("%s, at character offset %d (%s)",
771 dec.err, 988 dec.err,
772 (int)offset, 989 (int)offset,
773 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
774 } 991 }
775 992
776 sv_dump (sv);//D
777 return sv_2mortal (sv); 993 sv = sv_2mortal (sv);
994
995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
996 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
997
998 return sv;
778} 999}
1000
1001/////////////////////////////////////////////////////////////////////////////
1002// XS interface functions
779 1003
780MODULE = JSON::XS PACKAGE = JSON::XS 1004MODULE = JSON::XS PACKAGE = JSON::XS
781 1005
782BOOT: 1006BOOT:
783{ 1007{
784 int i; 1008 int i;
785 1009
786 memset (decode_hexdigit, 0xff, 256); 1010 memset (decode_hexdigit, 0xff, 256);
1011
787 for (i = 10; i--; ) 1012 for (i = 0; i < 256; ++i)
788 decode_hexdigit ['0' + i] = i; 1013 decode_hexdigit [i] =
789 1014 i >= '0' && i <= '9' ? i - '0'
790 for (i = 6; --i; ) 1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10
791 { 1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10
792 decode_hexdigit ['a' + i] = 10 + i; 1017 : -1;
793 decode_hexdigit ['A' + i] = 10 + i;
794 }
795 1018
796 json_stash = gv_stashpv ("JSON::XS", 1); 1019 json_stash = gv_stashpv ("JSON::XS", 1);
797} 1020}
1021
1022PROTOTYPES: DISABLE
798 1023
799SV *new (char *dummy) 1024SV *new (char *dummy)
800 CODE: 1025 CODE:
801 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1026 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
802 OUTPUT: 1027 OUTPUT:
803 RETVAL 1028 RETVAL
804 1029
805SV *ascii (SV *self, int enable) 1030SV *ascii (SV *self, int enable = 1)
806 ALIAS: 1031 ALIAS:
807 ascii = F_ASCII 1032 ascii = F_ASCII
808 utf8 = F_UTF8 1033 utf8 = F_UTF8
809 indent = F_INDENT 1034 indent = F_INDENT
810 canonical = F_CANONICAL 1035 canonical = F_CANONICAL
811 space_before = F_SPACE_BEFORE 1036 space_before = F_SPACE_BEFORE
812 space_after = F_SPACE_AFTER 1037 space_after = F_SPACE_AFTER
813 json_rpc = F_JSON_RPC 1038 pretty = F_PRETTY
1039 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK
814 CODE: 1041 CODE:
815{ 1042{
816 UV *uv = SvJSON (self); 1043 UV *uv = SvJSON (self);
817 if (enable) 1044 if (enable)
818 *uv |= ix; 1045 *uv |= ix;
… …
822 RETVAL = newSVsv (self); 1049 RETVAL = newSVsv (self);
823} 1050}
824 OUTPUT: 1051 OUTPUT:
825 RETVAL 1052 RETVAL
826 1053
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL)
1055 CODE:
1056{
1057 UV *uv = SvJSON (self);
1058 UV log2 = 0;
1059
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1061
1062 while ((1UL << log2) < max_depth)
1063 ++log2;
1064
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066
1067 RETVAL = newSVsv (self);
1068}
1069 OUTPUT:
1070 RETVAL
1071
827void encode (SV *self, SV *scalar) 1072void encode (SV *self, SV *scalar)
828 PPCODE: 1073 PPCODE:
829 XPUSHs (encode_json (scalar, *SvJSON (self))); 1074 XPUSHs (encode_json (scalar, *SvJSON (self)));
830 1075
831void decode (SV *self, SV *jsondata) 1076void decode (SV *self, SV *jsonstr)
832 PPCODE: 1077 PPCODE:
833 XPUSHs (decode_json (jsondata, *SvJSON (self))); 1078 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
834 1079
1080PROTOTYPES: ENABLE
1081
1082void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
1085 PPCODE:
1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1087
1088void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
1091 PPCODE:
1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1093

Diff Legend

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