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 1.4 by root, Thu Mar 22 21:13:58 2007 UTC vs.
Revision 1.38 by root, Mon Jun 11 03:18:07 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
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#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
9#define F_ASCII 0x00000001 14#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 16#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 17#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 18#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 19#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 20#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 21#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL
23#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27
25
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
17 30
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 32#define F_DEFAULT (9UL << S_MAXDEPTH)
20 33
21#define INIT_SIZE 32 // initial scalar size to be allocated 34#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level
36
37#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 38
23#define SB do { 39#define SB do {
24#define SE } while (0) 40#define SE } while (0)
25 41
42#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value))
44# define inline inline
45#else
46# define expect(expr,value) (expr)
47# define inline static
48#endif
49
50#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1)
52
26static HV *json_stash; 53static HV *json_stash; // JSON::XS::
54
55/////////////////////////////////////////////////////////////////////////////
56// utility functions
57
58static UV *
59SvJSON (SV *sv)
60{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62 croak ("object is not of type JSON::XS");
63
64 return &SvUVX (SvRV (sv));
65}
66
67static void
68shrink (SV *sv)
69{
70 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1)
72 {
73#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew)
76 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif
78 }
79}
80
81// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints
86inline UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
92 {
93 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 }
96 else
97 {
98 *clen = (STRLEN)-1;
99 return (UV)-1;
100 }
101}
102
103/////////////////////////////////////////////////////////////////////////////
104// encoder
27 105
28// structure used for encoding JSON 106// structure used for encoding JSON
29typedef struct 107typedef struct
30{ 108{
31 char *cur; 109 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 110 char *end; // SvEND (sv)
34 SV *sv; 111 SV *sv; // result scalar
35 UV flags; 112 U32 flags; // F_*
36 int max_recurse; 113 U32 indent; // indentation level
37 int indent; 114 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 115} enc_t;
39 116
40// structure used for decoding JSON 117inline void
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58/////////////////////////////////////////////////////////////////////////////
59
60static void
61need (enc_t *enc, STRLEN len) 118need (enc_t *enc, STRLEN len)
62{ 119{
63 if (enc->cur + len >= enc->end) 120 if (expect_false (enc->cur + len >= enc->end))
64 { 121 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 122 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 123 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 124 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 126 }
70} 127}
71 128
72static void 129inline void
73encode_ch (enc_t *enc, char ch) 130encode_ch (enc_t *enc, char ch)
74{ 131{
75 need (enc, 1); 132 need (enc, 1);
76 *enc->cur++ = ch; 133 *enc->cur++ = ch;
77} 134}
… …
85 142
86 while (str < end) 143 while (str < end)
87 { 144 {
88 unsigned char ch = *(unsigned char *)str; 145 unsigned char ch = *(unsigned char *)str;
89 146
90 if (ch == '"') 147 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 148 {
149 if (expect_false (ch == '"')) // but with slow exceptions
150 {
92 need (enc, len += 1); 151 need (enc, len += 1);
93 *enc->cur++ = '\\'; 152 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 153 *enc->cur++ = '"';
95 ++str;
96 } 154 }
97 else if (ch == '\\') 155 else if (expect_false (ch == '\\'))
98 { 156 {
99 need (enc, len += 1); 157 need (enc, len += 1);
100 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
102 ++str;
103 } 160 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 161 else
105 {
106 *enc->cur++ = ch; 162 *enc->cur++ = ch;
107 ++str; 163
108 }
109 else if (ch == '\015')
110 {
111 need (enc, len += 1);
112 *enc->cur++ = '\\';
113 *enc->cur++ = 'r';
114 ++str;
115 }
116 else if (ch == '\012')
117 {
118 need (enc, len += 1);
119 *enc->cur++ = '\\';
120 *enc->cur++ = 'n';
121 ++str; 164 ++str;
122 } 165 }
123 else 166 else
124 { 167 {
125 STRLEN clen; 168 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 { 169 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 170 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
131 if (clen < 0) 171 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
132 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 172 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
133 } 173 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
134 else 174 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
135 {
136 uch = ch;
137 clen = 1;
138 }
139 175
140 if (uch < 0x80 || enc->flags & F_ASCII) 176 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 177 {
178 STRLEN clen;
179 UV uch;
180
181 if (is_utf8)
182 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 }
188 else
189 {
190 uch = ch;
191 clen = 1;
192 }
193
194 if (uch > 0x10FFFFUL)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 {
199 if (uch > 0xFFFFUL)
200 {
144 need (enc, len += 11); 201 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 202 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 203 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 204 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 205 enc->cur += 12;
206 }
207 else
208 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5);
211 *enc->cur++ = '\\';
212 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15];
217 }
218
219 str += clen;
220 }
221 else if (enc->flags & F_LATIN1)
222 {
223 *enc->cur++ = uch;
224 str += clen;
225 }
226 else if (is_utf8)
227 {
228 need (enc, len += clen);
229 do
230 {
231 *enc->cur++ = *str++;
232 }
233 while (--clen);
234 }
235 else
236 {
237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
239 ++str;
240 }
149 } 241 }
150 else
151 {
152 static char hexdigit [16] = "0123456789abcdef";
153 need (enc, len += 5);
154 *enc->cur++ = '\\';
155 *enc->cur++ = 'u';
156 *enc->cur++ = hexdigit [ uch >> 12 ];
157 *enc->cur++ = hexdigit [(uch >> 8) & 15];
158 *enc->cur++ = hexdigit [(uch >> 4) & 15];
159 *enc->cur++ = hexdigit [(uch >> 0) & 15];
160 }
161
162 str += clen;
163 }
164 else if (is_utf8)
165 {
166 need (enc, len += clen);
167 while (clen--)
168 *enc->cur++ = *str++;
169 }
170 else
171 {
172 need (enc, 10); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str;
175 } 242 }
176 } 243 }
177 244
178 --len; 245 --len;
179 } 246 }
180} 247}
181 248
182#define INDENT SB \ 249inline void
250encode_indent (enc_t *enc)
251{
183 if (enc->flags & F_INDENT) \ 252 if (enc->flags & F_INDENT)
184 { \ 253 {
185 int i_; \ 254 int spaces = enc->indent * INDENT_STEP;
186 need (enc, enc->indent); \
187 for (i_ = enc->indent * 3; i_--; )\
188 encode_ch (enc, ' '); \
189 } \
190 SE
191 255
192#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 256 need (enc, spaces);
193#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 257 memset (enc->cur, ' ', spaces);
194#define COMMA SB \ 258 enc->cur += spaces;
259 }
260}
261
262inline void
263encode_space (enc_t *enc)
264{
265 need (enc, 1);
195 encode_ch (enc, ','); \ 266 encode_ch (enc, ' ');
267}
268
269inline void
270encode_nl (enc_t *enc)
271{
196 if (enc->flags & F_INDENT) \ 272 if (enc->flags & F_INDENT)
197 NL; \ 273 {
274 need (enc, 1);
275 encode_ch (enc, '\n');
276 }
277}
278
279inline void
280encode_comma (enc_t *enc)
281{
282 encode_ch (enc, ',');
283
284 if (enc->flags & F_INDENT)
285 encode_nl (enc);
198 else if (enc->flags & F_SPACE_AFTER) \ 286 else if (enc->flags & F_SPACE_AFTER)
199 SPACE; \ 287 encode_space (enc);
200 SE 288}
201 289
202static void encode_sv (enc_t *enc, SV *sv); 290static void encode_sv (enc_t *enc, SV *sv);
203 291
204static void 292static void
205encode_av (enc_t *enc, AV *av) 293encode_av (enc_t *enc, AV *av)
206{ 294{
207 int i, len = av_len (av); 295 int i, len = av_len (av);
208 296
209 encode_ch (enc, '['); NL; 297 if (enc->indent >= enc->maxdepth)
298 croak ("data structure too deep (hit recursion limit)");
299
300 encode_ch (enc, '['); encode_nl (enc);
210 ++enc->indent; 301 ++enc->indent;
211 302
212 for (i = 0; i <= len; ++i) 303 for (i = 0; i <= len; ++i)
213 { 304 {
214 INDENT; 305 encode_indent (enc);
215 encode_sv (enc, *av_fetch (av, i, 0)); 306 encode_sv (enc, *av_fetch (av, i, 0));
216 307
217 if (i < len) 308 if (i < len)
218 COMMA; 309 encode_comma (enc);
219 } 310 }
220 311
221 NL; 312 encode_nl (enc);
222 313
223 --enc->indent; 314 --enc->indent;
224 INDENT; encode_ch (enc, ']'); 315 encode_indent (enc); encode_ch (enc, ']');
225} 316}
226 317
227static void 318static void
228encode_he (enc_t *enc, HE *he) 319encode_he (enc_t *enc, HE *he)
229{ 320{
… …
243 else 334 else
244 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
245 336
246 encode_ch (enc, '"'); 337 encode_ch (enc, '"');
247 338
248 if (enc->flags & F_SPACE_BEFORE) SPACE; 339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
249 encode_ch (enc, ':'); 340 encode_ch (enc, ':');
250 if (enc->flags & F_SPACE_AFTER ) SPACE; 341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
251 encode_sv (enc, HeVAL (he)); 342 encode_sv (enc, HeVAL (he));
252} 343}
253 344
254// compare hash entries, used when all keys are bytestrings 345// compare hash entries, used when all keys are bytestrings
255static int 346static int
… …
261 HE *b = *(HE **)b_; 352 HE *b = *(HE **)b_;
262 353
263 STRLEN la = HeKLEN (a); 354 STRLEN la = HeKLEN (a);
264 STRLEN lb = HeKLEN (b); 355 STRLEN lb = HeKLEN (b);
265 356
266 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
267 cmp = la < lb ? -1 : la == lb ? 0 : 1; 358 cmp = la - lb;
268 359
269 return cmp; 360 return cmp;
270} 361}
271 362
272// compare hash entries, used when some keys are sv's or utf-x 363// compare hash entries, used when some keys are sv's or utf-x
… …
279static void 370static void
280encode_hv (enc_t *enc, HV *hv) 371encode_hv (enc_t *enc, HV *hv)
281{ 372{
282 int count, i; 373 int count, i;
283 374
375 if (enc->indent >= enc->maxdepth)
376 croak ("data structure too deep (hit recursion limit)");
377
284 encode_ch (enc, '{'); NL; ++enc->indent; 378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
285 379
286 if ((count = hv_iterinit (hv))) 380 if ((count = hv_iterinit (hv)))
287 { 381 {
288 // for canonical output we have to sort by keys first 382 // for canonical output we have to sort by keys first
289 // actually, this is mostly due to the stupid so-called 383 // actually, this is mostly due to the stupid so-called
290 // security workaround added somewhere in 5.8.x. 384 // security workaround added somewhere in 5.8.x.
291 // that randomises hash orderings 385 // that randomises hash orderings
292 if (enc->flags & F_CANONICAL) 386 if (enc->flags & F_CANONICAL)
293 { 387 {
294 HE *he, *hes [count];
295 int fast = 1; 388 int fast = 1;
389 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE));
392#else
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
394#endif
296 395
297 i = 0; 396 i = 0;
298 while ((he = hv_iternext (hv))) 397 while ((he = hv_iternext (hv)))
299 { 398 {
300 hes [i++] = he; 399 hes [i++] = he;
… …
306 405
307 if (fast) 406 if (fast)
308 qsort (hes, count, sizeof (HE *), he_cmp_fast); 407 qsort (hes, count, sizeof (HE *), he_cmp_fast);
309 else 408 else
310 { 409 {
311 // hack to disable "use bytes" 410 // hack to forcefully disable "use bytes"
312 COP *oldcop = PL_curcop, cop; 411 COP cop = *PL_curcop;
313 cop.op_private = 0; 412 cop.op_private = 0;
413
414 ENTER;
415 SAVETMPS;
416
417 SAVEVPTR (PL_curcop);
314 PL_curcop = &cop; 418 PL_curcop = &cop;
315 419
316 SAVETMPS;
317 qsort (hes, count, sizeof (HE *), he_cmp_slow); 420 qsort (hes, count, sizeof (HE *), he_cmp_slow);
421
318 FREETMPS; 422 FREETMPS;
319 423 LEAVE;
320 PL_curcop = oldcop;
321 } 424 }
322 425
323 for (i = 0; i < count; ++i) 426 for (i = 0; i < count; ++i)
324 { 427 {
325 INDENT; 428 encode_indent (enc);
326 encode_he (enc, hes [i]); 429 encode_he (enc, hes [i]);
327 430
328 if (i < count - 1) 431 if (i < count - 1)
329 COMMA; 432 encode_comma (enc);
330 } 433 }
331 434
332 NL; 435 encode_nl (enc);
333 } 436 }
334 else 437 else
335 { 438 {
336 SV *sv;
337 HE *he = hv_iternext (hv); 439 HE *he = hv_iternext (hv);
338 440
339 for (;;) 441 for (;;)
340 { 442 {
341 INDENT; 443 encode_indent (enc);
342 encode_he (enc, he); 444 encode_he (enc, he);
343 445
344 if (!(he = hv_iternext (hv))) 446 if (!(he = hv_iternext (hv)))
345 break; 447 break;
346 448
347 COMMA; 449 encode_comma (enc);
348 } 450 }
349 451
350 NL; 452 encode_nl (enc);
351 } 453 }
352 } 454 }
353 455
354 --enc->indent; INDENT; encode_ch (enc, '}'); 456 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
457}
458
459// encode objects, arrays and special \0=false and \1=true values.
460static void
461encode_rv (enc_t *enc, SV *sv)
462{
463 svtype svt;
464
465 SvGETMAGIC (sv);
466 svt = SvTYPE (sv);
467
468 if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV)
473 {
474 if (SvNIOK (sv) && SvIV (sv) == 0)
475 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1)
477 encode_str (enc, "true", 4, 0);
478 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 }
482 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
355} 485}
356 486
357static void 487static void
358encode_sv (enc_t *enc, SV *sv) 488encode_sv (enc_t *enc, SV *sv)
359{ 489{
… …
373 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
374 enc->cur += strlen (enc->cur); 504 enc->cur += strlen (enc->cur);
375 } 505 }
376 else if (SvIOKp (sv)) 506 else if (SvIOKp (sv))
377 { 507 {
378 need (enc, 64); 508 // we assume we can always read an IV as a UV
509 if (SvUV (sv) & ~(UV)0x7fff)
510 {
511 need (enc, sizeof (UV) * 3);
379 enc->cur += 512 enc->cur +=
380 SvIsUV(sv) 513 SvIsUV(sv)
381 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 514 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
382 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 515 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
516 }
517 else
518 {
519 // optimise the "small number case"
520 // code will likely be branchless and use only a single multiplication
521 I32 i = SvIV (sv);
522 U32 u;
523
524 need (enc, 6);
525
526 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
527 u = i < 0 ? -i : i;
528
529 // convert to 4.28 fixed-point representation
530 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531
532 char digit, nz = 0;
533
534 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
535 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
536 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
537 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
538 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1;
539 }
383 } 540 }
384 else if (SvROK (sv)) 541 else if (SvROK (sv))
385 { 542 encode_rv (enc, SvRV (sv));
386 if (!--enc->max_recurse)
387 croak ("data structure too deep (hit recursion limit)");
388
389 sv = SvRV (sv);
390
391 switch (SvTYPE (sv))
392 {
393 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
394 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
395
396 default:
397 croak ("JSON can only represent references to arrays or hashes");
398 }
399 }
400 else if (!SvOK (sv)) 543 else if (!SvOK (sv))
401 encode_str (enc, "null", 4, 0); 544 encode_str (enc, "null", 4, 0);
402 else 545 else
403 croak ("encountered perl type that JSON cannot handle"); 546 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
547 SvPV_nolen (sv), SvFLAGS (sv));
404} 548}
405 549
406static SV * 550static SV *
407encode_json (SV *scalar, UV flags) 551encode_json (SV *scalar, U32 flags)
408{ 552{
553 enc_t enc;
554
409 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
410 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
411 557
412 enc_t enc;
413 enc.flags = flags; 558 enc.flags = flags;
414 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
415 enc.cur = SvPVX (enc.sv); 560 enc.cur = SvPVX (enc.sv);
416 enc.end = SvEND (enc.sv); 561 enc.end = SvEND (enc.sv);
417 enc.max_recurse = 0;
418 enc.indent = 0; 562 enc.indent = 0;
563 enc.maxdepth = DEC_DEPTH (flags);
419 564
420 SvPOK_only (enc.sv); 565 SvPOK_only (enc.sv);
421 encode_sv (&enc, scalar); 566 encode_sv (&enc, scalar);
422 567
568 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570
423 if (!(flags & (F_ASCII | F_UTF8))) 571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
424 SvUTF8_on (enc.sv); 572 SvUTF8_on (enc.sv);
425 573
426 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 574 if (enc.flags & F_SHRINK)
575 shrink (enc.sv);
576
427 return enc.sv; 577 return enc.sv;
428} 578}
429 579
430///////////////////////////////////////////////////////////////////////////// 580/////////////////////////////////////////////////////////////////////////////
581// decoder
431 582
432#define WS \ 583// structure used for decoding JSON
584typedef struct
585{
586 char *cur; // current parser pointer
587 char *end; // end of input string
588 const char *err; // parse error, if != 0
589 U32 flags; // F_*
590 U32 depth; // recursion depth
591 U32 maxdepth; // recursion depth limit
592} dec_t;
593
594inline void
595decode_ws (dec_t *dec)
596{
433 for (;;) \ 597 for (;;)
434 { \ 598 {
435 char ch = *dec->cur; \ 599 char ch = *dec->cur;
600
436 if (ch > 0x20 \ 601 if (ch > 0x20
437 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 602 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
438 break; \ 603 break;
604
439 ++dec->cur; \ 605 ++dec->cur;
440 } 606 }
607}
441 608
442#define ERR(reason) SB dec->err = reason; goto fail; SE 609#define ERR(reason) SB dec->err = reason; goto fail; SE
610
443#define EXPECT_CH(ch) SB \ 611#define EXPECT_CH(ch) SB \
444 if (*dec->cur != ch) \ 612 if (*dec->cur != ch) \
445 ERR (# ch " expected"); \ 613 ERR (# ch " expected"); \
446 ++dec->cur; \ 614 ++dec->cur; \
447 SE 615 SE
448 616
617#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
618#define DEC_DEC_DEPTH --dec->depth
619
449static SV *decode_sv (dec_t *dec); 620static SV *decode_sv (dec_t *dec);
450 621
451static signed char decode_hexdigit[256]; 622static signed char decode_hexdigit[256];
452 623
453static UV 624static UV
454decode_4hex (dec_t *dec) 625decode_4hex (dec_t *dec)
455{ 626{
456 signed char d1, d2, d3, d4; 627 signed char d1, d2, d3, d4;
628 unsigned char *cur = (unsigned char *)dec->cur;
457 629
458 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");
459 if (d1 < 0) ERR ("four hexadecimal digits expected"); 631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected");
460 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected");
461 if (d2 < 0) ERR ("four hexadecimal digits expected"); 633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected");
462 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
463 if (d3 < 0) ERR ("four hexadecimal digits expected");
464 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
465 if (d4 < 0) ERR ("four hexadecimal digits expected");
466 634
467 dec->cur += 4; 635 dec->cur += 4;
468 636
469 return ((UV)d1) << 12 637 return ((UV)d1) << 12
470 | ((UV)d2) << 8 638 | ((UV)d2) << 8
… …
473 641
474fail: 642fail:
475 return (UV)-1; 643 return (UV)-1;
476} 644}
477 645
478#define APPEND_GROW(n) SB \
479 if (cur + (n) >= end) \
480 { \
481 STRLEN ofs = cur - SvPVX (sv); \
482 SvGROW (sv, ofs + (n) + 1); \
483 cur = SvPVX (sv) + ofs; \
484 end = SvEND (sv); \
485 } \
486 SE
487
488#define APPEND_CH(ch) SB \
489 APPEND_GROW (1); \
490 *cur++ = (ch); \
491 SE
492
493static SV * 646static SV *
494decode_str (dec_t *dec) 647decode_str (dec_t *dec)
495{ 648{
496 SV *sv = NEWSV (0,2); 649 SV *sv = 0;
497 int utf8 = 0; 650 int utf8 = 0;
498 char *cur = SvPVX (sv); 651 char *dec_cur = dec->cur;
499 char *end = SvEND (sv);
500 652
501 for (;;) 653 do
502 { 654 {
503 unsigned char ch = *(unsigned char *)dec->cur; 655 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
656 char *cur = buf;
504 657
505 if (ch == '"') 658 do
506 break;
507 else if (ch == '\\')
508 { 659 {
509 switch (*++dec->cur) 660 unsigned char ch = *(unsigned char *)dec_cur++;
661
662 if (expect_false (ch == '"'))
510 { 663 {
511 case '\\': 664 --dec_cur;
512 case '/': 665 break;
513 case '"': APPEND_CH (*dec->cur++); break; 666 }
514 667 else if (expect_false (ch == '\\'))
515 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 668 {
516 case 't': APPEND_CH ('\011'); ++dec->cur; break; 669 switch (*dec_cur)
517 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
518 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
519 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
520
521 case 'u':
522 { 670 {
523 UV lo, hi; 671 case '\\':
524 ++dec->cur; 672 case '/':
673 case '"': *cur++ = *dec_cur++; break;
525 674
526 hi = decode_4hex (dec); 675 case 'b': ++dec_cur; *cur++ = '\010'; break;
527 if (hi == (UV)-1) 676 case 't': ++dec_cur; *cur++ = '\011'; break;
528 goto fail; 677 case 'n': ++dec_cur; *cur++ = '\012'; break;
678 case 'f': ++dec_cur; *cur++ = '\014'; break;
679 case 'r': ++dec_cur; *cur++ = '\015'; break;
529 680
530 // possibly a surrogate pair 681 case 'u':
531 if (hi >= 0xd800 && hi < 0xdc00)
532 { 682 {
533 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 683 UV lo, hi;
534 ERR ("illegal surrogate character"); 684 ++dec_cur;
535 685
536 dec->cur += 2; 686 dec->cur = dec_cur;
537
538 lo = decode_4hex (dec); 687 hi = decode_4hex (dec);
688 dec_cur = dec->cur;
539 if (lo == (UV)-1) 689 if (hi == (UV)-1)
540 goto fail; 690 goto fail;
541 691
692 // possibly a surrogate pair
693 if (hi >= 0xd800)
694 if (hi < 0xdc00)
695 {
696 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
697 ERR ("missing low surrogate character in surrogate pair");
698
699 dec_cur += 2;
700
701 dec->cur = dec_cur;
702 lo = decode_4hex (dec);
703 dec_cur = dec->cur;
704 if (lo == (UV)-1)
705 goto fail;
706
542 if (lo < 0xdc00 || lo >= 0xe000) 707 if (lo < 0xdc00 || lo >= 0xe000)
543 ERR ("surrogate pair expected"); 708 ERR ("surrogate pair expected");
544 709
545 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 710 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
711 }
712 else if (hi < 0xe000)
713 ERR ("missing high surrogate character in surrogate pair");
714
715 if (hi >= 0x80)
716 {
717 utf8 = 1;
718
719 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
720 }
721 else
722 *cur++ = hi;
546 } 723 }
547 else if (lo >= 0xdc00 && lo < 0xe000)
548 ERR ("illegal surrogate character");
549
550 if (hi >= 0x80)
551 { 724 break;
552 utf8 = 1;
553 725
554 APPEND_GROW (4); // at most 4 bytes for 21 bits
555 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
556 }
557 else 726 default:
558 APPEND_CH (hi); 727 --dec_cur;
728 ERR ("illegal backslash escape sequence in string");
559 } 729 }
730 }
731 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
732 *cur++ = ch;
733 else if (ch >= 0x80)
734 {
735 STRLEN clen;
736 UV uch;
737
738 --dec_cur;
739
740 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
741 if (clen == (STRLEN)-1)
742 ERR ("malformed UTF-8 character in JSON string");
743
744 do
745 *cur++ = *dec_cur++;
746 while (--clen);
747
748 utf8 = 1;
749 }
750 else
751 {
752 --dec_cur;
753
754 if (!ch)
755 ERR ("unexpected end of string while parsing JSON string");
560 break; 756 else
757 ERR ("invalid character encountered while parsing JSON string");
561 } 758 }
562 } 759 }
563 else if (ch >= 0x20 && ch <= 0x7f) 760 while (cur < buf + SHORT_STRING_LEN);
564 APPEND_CH (*dec->cur++); 761
565 else if (ch >= 0x80) 762 {
763 STRLEN len = cur - buf;
764
765 if (sv)
566 { 766 {
567 STRLEN clen; 767 SvGROW (sv, SvCUR (sv) + len + 1);
568 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 768 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
569 if (clen < 0) 769 SvCUR_set (sv, SvCUR (sv) + len);
570 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
571
572 APPEND_GROW (clen);
573 memcpy (cur, dec->cur, clen);
574 cur += clen;
575 dec->cur += clen;
576 } 770 }
577 else 771 else
578 ERR ("invalid character encountered"); 772 sv = newSVpvn (buf, len);
579 } 773 }
774 }
775 while (*dec_cur != '"');
580 776
581 ++dec->cur; 777 ++dec_cur;
582 778
583 SvCUR_set (sv, cur - SvPVX (sv)); 779 if (sv)
584 780 {
585 SvPOK_only (sv); 781 SvPOK_only (sv);
586 *SvEND (sv) = 0; 782 *SvEND (sv) = 0;
587 783
588 if (utf8) 784 if (utf8)
589 SvUTF8_on (sv); 785 SvUTF8_on (sv);
786 }
787 else
788 sv = newSVpvn ("", 0);
590 789
790 dec->cur = dec_cur;
591 return sv; 791 return sv;
592 792
593fail: 793fail:
594 SvREFCNT_dec (sv); 794 dec->cur = dec_cur;
595 return 0; 795 return 0;
596} 796}
597 797
598static SV * 798static SV *
599decode_num (dec_t *dec) 799decode_num (dec_t *dec)
… …
609 { 809 {
610 ++dec->cur; 810 ++dec->cur;
611 if (*dec->cur >= '0' && *dec->cur <= '9') 811 if (*dec->cur >= '0' && *dec->cur <= '9')
612 ERR ("malformed number (leading zero must not be followed by another digit)"); 812 ERR ("malformed number (leading zero must not be followed by another digit)");
613 } 813 }
614 814 else if (*dec->cur < '0' || *dec->cur > '9')
615 // int 815 ERR ("malformed number (no digits after initial minus)");
816 else
817 do
818 {
819 ++dec->cur;
820 }
616 while (*dec->cur >= '0' && *dec->cur <= '9') 821 while (*dec->cur >= '0' && *dec->cur <= '9');
617 ++dec->cur;
618 822
619 // [frac] 823 // [frac]
620 if (*dec->cur == '.') 824 if (*dec->cur == '.')
621 { 825 {
622 is_nv = 1; 826 ++dec->cur;
827
828 if (*dec->cur < '0' || *dec->cur > '9')
829 ERR ("malformed number (no digits after decimal point)");
623 830
624 do 831 do
625 { 832 {
626 ++dec->cur; 833 ++dec->cur;
627 } 834 }
628 while (*dec->cur >= '0' && *dec->cur <= '9'); 835 while (*dec->cur >= '0' && *dec->cur <= '9');
836
837 is_nv = 1;
629 } 838 }
630 839
631 // [exp] 840 // [exp]
632 if (*dec->cur == 'e' || *dec->cur == 'E') 841 if (*dec->cur == 'e' || *dec->cur == 'E')
633 { 842 {
634 is_nv = 1;
635
636 ++dec->cur; 843 ++dec->cur;
844
637 if (*dec->cur == '-' || *dec->cur == '+') 845 if (*dec->cur == '-' || *dec->cur == '+')
638 ++dec->cur; 846 ++dec->cur;
639 847
848 if (*dec->cur < '0' || *dec->cur > '9')
849 ERR ("malformed number (no digits after exp sign)");
850
851 do
852 {
853 ++dec->cur;
854 }
640 while (*dec->cur >= '0' && *dec->cur <= '9') 855 while (*dec->cur >= '0' && *dec->cur <= '9');
641 ++dec->cur; 856
857 is_nv = 1;
642 } 858 }
643 859
644 if (!is_nv) 860 if (!is_nv)
645 { 861 {
646 UV uv; 862 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
647 int numtype = grok_number (start, dec->cur - start, &uv); 863 if (*start == '-')
648 if (numtype & IS_NUMBER_IN_UV) 864 switch (dec->cur - start)
649 if (numtype & IS_NUMBER_NEG)
650 { 865 {
651 if (uv < (UV)IV_MIN) 866 case 2: return newSViv (-( start [1] - '0' ));
652 return newSViv (-(IV)uv); 867 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
868 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
869 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
653 } 870 }
871 else
872 switch (dec->cur - start)
873 {
874 case 1: return newSViv ( start [0] - '0' );
875 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
876 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
877 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
878 }
879
880 {
881 UV uv;
882 int numtype = grok_number (start, dec->cur - start, &uv);
883 if (numtype & IS_NUMBER_IN_UV)
884 if (numtype & IS_NUMBER_NEG)
885 {
886 if (uv < (UV)IV_MIN)
887 return newSViv (-(IV)uv);
888 }
654 else 889 else
655 return newSVuv (uv); 890 return newSVuv (uv);
891 }
656 } 892 }
657 893
658 return newSVnv (Atof (start)); 894 return newSVnv (Atof (start));
659 895
660fail: 896fail:
… …
664static SV * 900static SV *
665decode_av (dec_t *dec) 901decode_av (dec_t *dec)
666{ 902{
667 AV *av = newAV (); 903 AV *av = newAV ();
668 904
905 DEC_INC_DEPTH;
906 decode_ws (dec);
907
908 if (*dec->cur == ']')
909 ++dec->cur;
910 else
669 for (;;) 911 for (;;)
670 { 912 {
671 SV *value; 913 SV *value;
672 914
673 value = decode_sv (dec); 915 value = decode_sv (dec);
674 if (!value) 916 if (!value)
675 goto fail; 917 goto fail;
676 918
677 av_push (av, value); 919 av_push (av, value);
678 920
679 WS; 921 decode_ws (dec);
680 922
681 if (*dec->cur == ']') 923 if (*dec->cur == ']')
682 { 924 {
683 ++dec->cur; 925 ++dec->cur;
684 break; 926 break;
927 }
685 } 928
686
687 if (*dec->cur != ',') 929 if (*dec->cur != ',')
688 ERR (", or ] expected while parsing array"); 930 ERR (", or ] expected while parsing array");
689 931
690 ++dec->cur; 932 ++dec->cur;
691 } 933 }
692 934
935 DEC_DEC_DEPTH;
693 return newRV_noinc ((SV *)av); 936 return newRV_noinc ((SV *)av);
694 937
695fail: 938fail:
696 SvREFCNT_dec (av); 939 SvREFCNT_dec (av);
940 DEC_DEC_DEPTH;
697 return 0; 941 return 0;
698} 942}
699 943
700static SV * 944static SV *
701decode_hv (dec_t *dec) 945decode_hv (dec_t *dec)
702{ 946{
703 HV *hv = newHV (); 947 HV *hv = newHV ();
704 948
949 DEC_INC_DEPTH;
950 decode_ws (dec);
951
952 if (*dec->cur == '}')
953 ++dec->cur;
954 else
705 for (;;) 955 for (;;)
706 { 956 {
707 SV *key, *value; 957 SV *key, *value;
708 958
709 WS; EXPECT_CH ('"'); 959 decode_ws (dec); EXPECT_CH ('"');
710 960
711 key = decode_str (dec); 961 key = decode_str (dec);
712 if (!key) 962 if (!key)
713 goto fail;
714
715 WS; EXPECT_CH (':');
716
717 value = decode_sv (dec);
718 if (!value)
719 {
720 SvREFCNT_dec (key);
721 goto fail; 963 goto fail;
964
965 decode_ws (dec); EXPECT_CH (':');
966
967 value = decode_sv (dec);
968 if (!value)
969 {
970 SvREFCNT_dec (key);
971 goto fail;
722 } 972 }
723 973
724 //TODO: optimise
725 hv_store_ent (hv, key, value, 0); 974 hv_store_ent (hv, key, value, 0);
975 SvREFCNT_dec (key);
726 976
727 WS; 977 decode_ws (dec);
728 978
729 if (*dec->cur == '}') 979 if (*dec->cur == '}')
730 { 980 {
731 ++dec->cur; 981 ++dec->cur;
732 break; 982 break;
733 } 983 }
734 984
735 if (*dec->cur != ',') 985 if (*dec->cur != ',')
736 ERR (", or } expected while parsing object/hash"); 986 ERR (", or } expected while parsing object/hash");
737 987
738 ++dec->cur; 988 ++dec->cur;
739 } 989 }
740 990
991 DEC_DEC_DEPTH;
741 return newRV_noinc ((SV *)hv); 992 return newRV_noinc ((SV *)hv);
742 993
743fail: 994fail:
744 SvREFCNT_dec (hv); 995 SvREFCNT_dec (hv);
996 DEC_DEC_DEPTH;
745 return 0; 997 return 0;
746} 998}
747 999
748static SV * 1000static SV *
749decode_sv (dec_t *dec) 1001decode_sv (dec_t *dec)
750{ 1002{
751 WS; 1003 decode_ws (dec);
752 switch (*dec->cur) 1004 switch (*dec->cur)
753 { 1005 {
754 case '"': ++dec->cur; return decode_str (dec); 1006 case '"': ++dec->cur; return decode_str (dec);
755 case '[': ++dec->cur; return decode_av (dec); 1007 case '[': ++dec->cur; return decode_av (dec);
756 case '{': ++dec->cur; return decode_hv (dec); 1008 case '{': ++dec->cur; return decode_hv (dec);
… …
784 1036
785 case 'n': 1037 case 'n':
786 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1038 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
787 { 1039 {
788 dec->cur += 4; 1040 dec->cur += 4;
789 return newSViv (1); 1041 return newSVsv (&PL_sv_undef);
790 } 1042 }
791 else 1043 else
792 ERR ("'null' expected"); 1044 ERR ("'null' expected");
793 1045
794 break; 1046 break;
795 1047
796 default: 1048 default:
797 ERR ("malformed json string"); 1049 ERR ("malformed JSON string, neither array, object, number, string or atom");
798 break; 1050 break;
799 } 1051 }
800 1052
801fail: 1053fail:
802 return 0; 1054 return 0;
803} 1055}
804 1056
805static SV * 1057static SV *
806decode_json (SV *string, UV flags) 1058decode_json (SV *string, U32 flags, UV *offset_return)
807{ 1059{
1060 dec_t dec;
1061 UV offset;
808 SV *sv; 1062 SV *sv;
809 1063
1064 SvGETMAGIC (string);
1065 SvUPGRADE (string, SVt_PV);
1066
810 if (!(flags & F_UTF8)) 1067 if (flags & F_UTF8)
1068 sv_utf8_downgrade (string, 0);
1069 else
811 sv_utf8_upgrade (string); 1070 sv_utf8_upgrade (string);
812 1071
813 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1072 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
814 1073
815 dec_t dec;
816 dec.flags = flags; 1074 dec.flags = flags;
817 dec.cur = SvPVX (string); 1075 dec.cur = SvPVX (string);
818 dec.end = SvEND (string); 1076 dec.end = SvEND (string);
819 dec.err = 0; 1077 dec.err = 0;
1078 dec.depth = 0;
1079 dec.maxdepth = DEC_DEPTH (dec.flags);
820 1080
821 *dec.end = 1; // invalid anywhere 1081 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
822 sv = decode_sv (&dec); 1082 sv = decode_sv (&dec);
823 *dec.end = 0; 1083
1084 if (!(offset_return || !sv))
1085 {
1086 // check for trailing garbage
1087 decode_ws (&dec);
1088
1089 if (*dec.cur)
1090 {
1091 dec.err = "garbage after JSON object";
1092 SvREFCNT_dec (sv);
1093 sv = 0;
1094 }
1095 }
1096
1097 if (offset_return || !sv)
1098 {
1099 offset = dec.flags & F_UTF8
1100 ? dec.cur - SvPVX (string)
1101 : utf8_distance (dec.cur, SvPVX (string));
1102
1103 if (offset_return)
1104 *offset_return = offset;
1105 }
824 1106
825 if (!sv) 1107 if (!sv)
826 { 1108 {
827 IV offset = utf8_distance (dec.cur, SvPVX (string));
828 SV *uni = sv_newmortal (); 1109 SV *uni = sv_newmortal ();
829 1110
1111 // horrible hack to silence warning inside pv_uni_display
1112 COP cop = *PL_curcop;
1113 cop.cop_warnings = pWARN_NONE;
1114 ENTER;
1115 SAVEVPTR (PL_curcop);
1116 PL_curcop = &cop;
830 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1117 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1118 LEAVE;
1119
831 croak ("%s, at character %d (%s)", 1120 croak ("%s, at character offset %d [\"%s\"]",
832 dec.err, 1121 dec.err,
833 (int)offset, 1122 (int)offset,
834 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1123 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
835 } 1124 }
836 1125
837 sv = sv_2mortal (sv); 1126 sv = sv_2mortal (sv);
838 1127
839 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1128 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
840 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1129 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
841 1130
842 return sv; 1131 return sv;
843} 1132}
844 1133
1134/////////////////////////////////////////////////////////////////////////////
1135// XS interface functions
1136
845MODULE = JSON::XS PACKAGE = JSON::XS 1137MODULE = JSON::XS PACKAGE = JSON::XS
846 1138
847BOOT: 1139BOOT:
848{ 1140{
849 int i; 1141 int i;
850 1142
851 memset (decode_hexdigit, 0xff, 256); 1143 memset (decode_hexdigit, 0xff, 256);
1144
852 for (i = 10; i--; ) 1145 for (i = 0; i < 256; ++i)
853 decode_hexdigit ['0' + i] = i; 1146 decode_hexdigit [i] =
854 1147 i >= '0' && i <= '9' ? i - '0'
855 for (i = 7; i--; ) 1148 : i >= 'a' && i <= 'f' ? i - 'a' + 10
856 { 1149 : i >= 'A' && i <= 'F' ? i - 'A' + 10
857 decode_hexdigit ['a' + i] = 10 + i; 1150 : -1;
858 decode_hexdigit ['A' + i] = 10 + i;
859 }
860 1151
861 json_stash = gv_stashpv ("JSON::XS", 1); 1152 json_stash = gv_stashpv ("JSON::XS", 1);
862} 1153}
863 1154
864PROTOTYPES: DISABLE 1155PROTOTYPES: DISABLE
… …
867 CODE: 1158 CODE:
868 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1159 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
869 OUTPUT: 1160 OUTPUT:
870 RETVAL 1161 RETVAL
871 1162
872SV *ascii (SV *self, int enable) 1163SV *ascii (SV *self, int enable = 1)
873 ALIAS: 1164 ALIAS:
874 ascii = F_ASCII 1165 ascii = F_ASCII
1166 latin1 = F_LATIN1
875 utf8 = F_UTF8 1167 utf8 = F_UTF8
876 indent = F_INDENT 1168 indent = F_INDENT
877 canonical = F_CANONICAL 1169 canonical = F_CANONICAL
878 space_before = F_SPACE_BEFORE 1170 space_before = F_SPACE_BEFORE
879 space_after = F_SPACE_AFTER 1171 space_after = F_SPACE_AFTER
880 json_rpc = F_JSON_RPC
881 pretty = F_PRETTY 1172 pretty = F_PRETTY
882 allow_nonref = F_ALLOW_NONREF 1173 allow_nonref = F_ALLOW_NONREF
1174 shrink = F_SHRINK
883 CODE: 1175 CODE:
884{ 1176{
885 UV *uv = SvJSON (self); 1177 UV *uv = SvJSON (self);
886 if (enable) 1178 if (enable)
887 *uv |= ix; 1179 *uv |= ix;
… …
891 RETVAL = newSVsv (self); 1183 RETVAL = newSVsv (self);
892} 1184}
893 OUTPUT: 1185 OUTPUT:
894 RETVAL 1186 RETVAL
895 1187
1188SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1189 CODE:
1190{
1191 UV *uv = SvJSON (self);
1192 UV log2 = 0;
1193
1194 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1195
1196 while ((1UL << log2) < max_depth)
1197 ++log2;
1198
1199 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1200
1201 RETVAL = newSVsv (self);
1202}
1203 OUTPUT:
1204 RETVAL
1205
896void encode (SV *self, SV *scalar) 1206void encode (SV *self, SV *scalar)
897 PPCODE: 1207 PPCODE:
898 XPUSHs (encode_json (scalar, *SvJSON (self))); 1208 XPUSHs (encode_json (scalar, *SvJSON (self)));
899 1209
900void decode (SV *self, SV *jsonstr) 1210void decode (SV *self, SV *jsonstr)
901 PPCODE: 1211 PPCODE:
902 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1212 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1213
1214void decode_prefix (SV *self, SV *jsonstr)
1215 PPCODE:
1216{
1217 UV offset;
1218 EXTEND (SP, 2);
1219 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1220 PUSHs (sv_2mortal (newSVuv (offset)));
1221}
903 1222
904PROTOTYPES: ENABLE 1223PROTOTYPES: ENABLE
905 1224
906void to_json (SV *scalar) 1225void to_json (SV *scalar)
1226 ALIAS:
1227 objToJson = 0
907 PPCODE: 1228 PPCODE:
908 XPUSHs (encode_json (scalar, F_UTF8)); 1229 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
909 1230
910void from_json (SV *jsonstr) 1231void from_json (SV *jsonstr)
1232 ALIAS:
1233 jsonToObj = 0
911 PPCODE: 1234 PPCODE:
912 XPUSHs (decode_json (jsonstr, F_UTF8)); 1235 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
913 1236

Diff Legend

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