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 1.51 by root, Mon Jul 2 00:48:18 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
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040 27#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
16 39
17#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
18#define F_DEFAULT 0 41#define F_DEFAULT (9UL << S_MAXDEPTH)
19 42
20#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level
45
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
21 47
22#define SB do { 48#define SB do {
23#define SE } while (0) 49#define SE } while (0)
24 50
25static HV *json_stash; 51#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value))
53# define inline inline
54#else
55# define expect(expr,value) (expr)
56# define inline static
57#endif
58
59#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1)
61
62static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false;
64
65typedef struct {
66 U32 flags;
67 SV *cb_object, *cb_sk_object;
68} JSON;
69
70/////////////////////////////////////////////////////////////////////////////
71// utility functions
72
73inline void
74shrink (SV *sv)
75{
76 sv_utf8_downgrade (sv, 1);
77 if (SvLEN (sv) > SvCUR (sv) + 1)
78 {
79#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew)
82 SvPV_renew (sv, SvCUR (sv) + 1);
83#endif
84 }
85}
86
87// decode an utf-8 character and return it, or (UV)-1 in
88// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints
92inline UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
98 {
99 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 }
102 else
103 {
104 *clen = (STRLEN)-1;
105 return (UV)-1;
106 }
107}
108
109/////////////////////////////////////////////////////////////////////////////
110// encoder
26 111
27// structure used for encoding JSON 112// structure used for encoding JSON
28typedef struct 113typedef struct
29{ 114{
30 char *cur; 115 char *cur; // SvPVX (sv) + current output position
31 STRLEN len; // SvLEN (sv)
32 char *end; // SvEND (sv) 116 char *end; // SvEND (sv)
33 SV *sv; 117 SV *sv; // result scalar
34 UV flags; 118 JSON json;
35 int max_recurse; 119 U32 indent; // indentation level
36 int indent; 120 U32 maxdepth; // max. indentation/recursion level
37} enc_t; 121} enc_t;
38 122
39// structure used for decoding JSON 123inline void
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
59static void
60need (enc_t *enc, STRLEN len) 124need (enc_t *enc, STRLEN len)
61{ 125{
62 if (enc->cur + len >= enc->end) 126 if (expect_false (enc->cur + len >= enc->end))
63 { 127 {
64 STRLEN cur = enc->cur - SvPVX (enc->sv); 128 STRLEN cur = enc->cur - SvPVX (enc->sv);
65 SvGROW (enc->sv, cur + len + 1); 129 SvGROW (enc->sv, cur + len + 1);
66 enc->cur = SvPVX (enc->sv) + cur; 130 enc->cur = SvPVX (enc->sv) + cur;
67 enc->end = SvEND (enc->sv); 131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
68 } 132 }
69} 133}
70 134
71static void 135inline void
72encode_ch (enc_t *enc, char ch) 136encode_ch (enc_t *enc, char ch)
73{ 137{
74 need (enc, 1); 138 need (enc, 1);
75 *enc->cur++ = ch; 139 *enc->cur++ = ch;
76} 140}
… …
78static void 142static void
79encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 143encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
80{ 144{
81 char *end = str + len; 145 char *end = str + len;
82 146
147 need (enc, len);
148
83 while (str < end) 149 while (str < end)
84 { 150 {
85 unsigned char ch = *(unsigned char *)str; 151 unsigned char ch = *(unsigned char *)str;
152
86 if (ch >= 0x20 && ch < 0x80) // most common case 153 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
87 { 154 {
155 if (expect_false (ch == '"')) // but with slow exceptions
156 {
157 need (enc, len += 1);
158 *enc->cur++ = '\\';
159 *enc->cur++ = '"';
160 }
161 else if (expect_false (ch == '\\'))
162 {
163 need (enc, len += 1);
164 *enc->cur++ = '\\';
165 *enc->cur++ = '\\';
166 }
167 else
88 *enc->cur++ = ch; 168 *enc->cur++ = ch;
169
89 str++; 170 ++str;
90 } 171 }
91 else 172 else
92 { 173 {
93 STRLEN clen; 174 switch (ch)
94 UV uch;
95
96 if (is_utf8)
97 {
98 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
99 if (clen < 0)
100 croak ("malformed UTF-8 character in string, cannot convert to JSON");
101 } 175 {
102 else 176 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
103 { 177 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
104 uch = ch; 178 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
105 clen = 1; 179 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
106 } 180 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
107 181
108 need (enc, len += 6); 182 default:
109
110 if (uch < 0xa0 || enc->flags & F_ASCII)
111 {
112 if (uch > 0xFFFFUL)
113 { 183 {
184 STRLEN clen;
185 UV uch;
186
187 if (is_utf8)
188 {
189 uch = decode_utf8 (str, end - str, &clen);
190 if (clen == (STRLEN)-1)
191 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
192 }
193 else
194 {
195 uch = ch;
114 len += 6; 196 clen = 1;
197 }
198
199 if (uch > 0x10FFFFUL)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
203 {
204 if (uch > 0xFFFFUL)
205 {
115 need (enc, len += 6); 206 need (enc, len += 11);
116 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
117 (uch - 0x10000) / 0x400 + 0xD800, 208 (int)((uch - 0x10000) / 0x400 + 0xD800),
118 (uch - 0x10000) % 0x400 + 0xDC00); 209 (int)((uch - 0x10000) % 0x400 + 0xDC00));
119 enc->cur += 12; 210 enc->cur += 12;
211 }
212 else
213 {
214 static char hexdigit [16] = "0123456789abcdef";
215 need (enc, len += 5);
216 *enc->cur++ = '\\';
217 *enc->cur++ = 'u';
218 *enc->cur++ = hexdigit [ uch >> 12 ];
219 *enc->cur++ = hexdigit [(uch >> 8) & 15];
220 *enc->cur++ = hexdigit [(uch >> 4) & 15];
221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
222 }
223
224 str += clen;
225 }
226 else if (enc->json.flags & F_LATIN1)
227 {
228 *enc->cur++ = uch;
229 str += clen;
230 }
231 else if (is_utf8)
232 {
233 need (enc, len += clen);
234 do
235 {
236 *enc->cur++ = *str++;
237 }
238 while (--clen);
239 }
240 else
241 {
242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
244 ++str;
245 }
120 } 246 }
121 else
122 {
123 sprintf (enc->cur, "\\u%04x", uch);
124 enc->cur += 6;
125 }
126 }
127 else if (is_utf8)
128 { 247 }
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 } 248 }
137 249
138 --len; 250 --len;
139 } 251 }
140} 252}
141 253
142#define INDENT SB \ 254inline void
255encode_indent (enc_t *enc)
256{
143 if (enc->flags & F_INDENT) \ 257 if (enc->json.flags & F_INDENT)
144 { \ 258 {
145 int i_; \ 259 int spaces = enc->indent * INDENT_STEP;
146 need (enc, enc->indent); \ 260
147 for (i_ = enc->indent * 3; i_--; )\ 261 need (enc, spaces);
262 memset (enc->cur, ' ', spaces);
263 enc->cur += spaces;
264 }
265}
266
267inline void
268encode_space (enc_t *enc)
269{
270 need (enc, 1);
271 encode_ch (enc, ' ');
272}
273
274inline void
275encode_nl (enc_t *enc)
276{
277 if (enc->json.flags & F_INDENT)
278 {
279 need (enc, 1);
148 encode_ch (enc, ' '); \ 280 encode_ch (enc, '\n');
149 } \ 281 }
150 SE 282}
151 283
152#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 284inline void
153#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 285encode_comma (enc_t *enc)
154#define COMMA SB \ 286{
155 encode_ch (enc, ','); \ 287 encode_ch (enc, ',');
288
156 if (enc->flags & F_INDENT) \ 289 if (enc->json.flags & F_INDENT)
157 NL; \ 290 encode_nl (enc);
158 else if (enc->flags & F_SPACE_AFTER) \ 291 else if (enc->json.flags & F_SPACE_AFTER)
159 SPACE; \ 292 encode_space (enc);
160 SE 293}
161 294
162static void encode_sv (enc_t *enc, SV *sv); 295static void encode_sv (enc_t *enc, SV *sv);
163 296
164static void 297static void
165encode_av (enc_t *enc, AV *av) 298encode_av (enc_t *enc, AV *av)
166{ 299{
167 int i, len = av_len (av); 300 int i, len = av_len (av);
168 301
169 encode_ch (enc, '['); NL; 302 if (enc->indent >= enc->maxdepth)
303 croak ("data structure too deep (hit recursion limit)");
304
305 encode_ch (enc, '['); encode_nl (enc);
170 ++enc->indent; 306 ++enc->indent;
171 307
172 for (i = 0; i <= len; ++i) 308 for (i = 0; i <= len; ++i)
173 { 309 {
174 INDENT; 310 encode_indent (enc);
175 encode_sv (enc, *av_fetch (av, i, 0)); 311 encode_sv (enc, *av_fetch (av, i, 0));
176 312
177 if (i < len) 313 if (i < len)
178 COMMA; 314 encode_comma (enc);
179 } 315 }
180 316
181 NL; 317 encode_nl (enc);
182 318
183 --enc->indent; 319 --enc->indent;
184 INDENT; encode_ch (enc, ']'); 320 encode_indent (enc); encode_ch (enc, ']');
185} 321}
186 322
187static void 323static void
188encode_he (enc_t *enc, HE *he) 324encode_he (enc_t *enc, HE *he)
189{ 325{
… …
191 327
192 if (HeKLEN (he) == HEf_SVKEY) 328 if (HeKLEN (he) == HEf_SVKEY)
193 { 329 {
194 SV *sv = HeSVKEY (he); 330 SV *sv = HeSVKEY (he);
195 STRLEN len; 331 STRLEN len;
332 char *str;
333
334 SvGETMAGIC (sv);
196 char *str = SvPV (sv, len); 335 str = SvPV (sv, len);
197 336
198 encode_str (enc, str, len, SvUTF8 (sv)); 337 encode_str (enc, str, len, SvUTF8 (sv));
199 } 338 }
200 else 339 else
201 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
202 341
203 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
204 343
205 if (enc->flags & F_SPACE_BEFORE) SPACE; 344 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
206 encode_ch (enc, ':'); 345 encode_ch (enc, ':');
207 if (enc->flags & F_SPACE_AFTER ) SPACE; 346 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
208 encode_sv (enc, HeVAL (he)); 347 encode_sv (enc, HeVAL (he));
209} 348}
210 349
211// compare hash entries, used when all keys are bytestrings 350// compare hash entries, used when all keys are bytestrings
212static int 351static int
… …
218 HE *b = *(HE **)b_; 357 HE *b = *(HE **)b_;
219 358
220 STRLEN la = HeKLEN (a); 359 STRLEN la = HeKLEN (a);
221 STRLEN lb = HeKLEN (b); 360 STRLEN lb = HeKLEN (b);
222 361
223 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
224 cmp = la < lb ? -1 : la == lb ? 0 : 1; 363 cmp = la - lb;
225 364
226 return cmp; 365 return cmp;
227} 366}
228 367
229// compare hash entries, used when some keys are sv's or utf-x 368// compare hash entries, used when some keys are sv's or utf-x
… …
236static void 375static void
237encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
238{ 377{
239 int count, i; 378 int count, i;
240 379
380 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)");
382
241 encode_ch (enc, '{'); NL; ++enc->indent; 383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
242 384
243 if ((count = hv_iterinit (hv))) 385 if ((count = hv_iterinit (hv)))
244 { 386 {
245 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
246 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
247 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
248 // that randomises hash orderings 390 // that randomises hash orderings
249 if (enc->flags & F_CANONICAL) 391 if (enc->json.flags & F_CANONICAL)
250 { 392 {
251 HE *he, *hes [count];
252 int fast = 1; 393 int fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE));
397#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif
253 400
254 i = 0; 401 i = 0;
255 while ((he = hv_iternext (hv))) 402 while ((he = hv_iternext (hv)))
256 { 403 {
257 hes [i++] = he; 404 hes [i++] = he;
… …
263 410
264 if (fast) 411 if (fast)
265 qsort (hes, count, sizeof (HE *), he_cmp_fast); 412 qsort (hes, count, sizeof (HE *), he_cmp_fast);
266 else 413 else
267 { 414 {
268 // hack to disable "use bytes" 415 // hack to forcefully disable "use bytes"
269 COP *oldcop = PL_curcop, cop; 416 COP cop = *PL_curcop;
270 cop.op_private = 0; 417 cop.op_private = 0;
418
419 ENTER;
420 SAVETMPS;
421
422 SAVEVPTR (PL_curcop);
271 PL_curcop = &cop; 423 PL_curcop = &cop;
272 424
273 SAVETMPS;
274 qsort (hes, count, sizeof (HE *), he_cmp_slow); 425 qsort (hes, count, sizeof (HE *), he_cmp_slow);
426
275 FREETMPS; 427 FREETMPS;
276 428 LEAVE;
277 PL_curcop = oldcop;
278 } 429 }
279 430
280 for (i = 0; i < count; ++i) 431 for (i = 0; i < count; ++i)
281 { 432 {
282 INDENT; 433 encode_indent (enc);
283 encode_he (enc, hes [i]); 434 encode_he (enc, hes [i]);
284 435
285 if (i < count - 1) 436 if (i < count - 1)
286 COMMA; 437 encode_comma (enc);
287 }
288
289 NL; 438 }
439
440 encode_nl (enc);
290 } 441 }
291 else 442 else
292 { 443 {
293 SV *sv;
294 HE *he = hv_iternext (hv); 444 HE *he = hv_iternext (hv);
295 445
296 for (;;) 446 for (;;)
297 { 447 {
298 INDENT; 448 encode_indent (enc);
299 encode_he (enc, he); 449 encode_he (enc, he);
300 450
301 if (!(he = hv_iternext (hv))) 451 if (!(he = hv_iternext (hv)))
302 break; 452 break;
303 453
304 COMMA; 454 encode_comma (enc);
305 }
306
307 NL; 455 }
456
457 encode_nl (enc);
308 } 458 }
309 } 459 }
310 460
311 --enc->indent; INDENT; encode_ch (enc, '}'); 461 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462}
463
464// encode objects, arrays and special \0=false and \1=true values.
465static void
466encode_rv (enc_t *enc, SV *sv)
467{
468 svtype svt;
469
470 SvGETMAGIC (sv);
471 svt = SvTYPE (sv);
472
473 if (expect_false (SvOBJECT (sv)))
474 {
475 if (SvSTASH (sv) == json_boolean_stash)
476 {
477 if (SvIV (sv))
478 encode_str (enc, "true", 4, 0);
479 else
480 encode_str (enc, "false", 5, 0);
481 }
482 else
483 {
484#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 {
487 // not yet
488 }
489#endif
490 if (enc->json.flags & F_CONV_BLESSED)
491 {
492 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
494
495 if (to_json)
496 {
497 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
498 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
499
500 // calling with G_SCALAR ensures that we always get a 1 reutrn value
501 // check anyways.
502 PUTBACK;
503 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
504 SPAGAIN;
505
506 encode_sv (enc, POPs);
507
508 FREETMPS; LEAVE;
509 }
510 else if (enc->json.flags & F_ALLOW_BLESSED)
511 encode_str (enc, "null", 4, 0);
512 else
513 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
514 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
515 }
516 else if (enc->json.flags & F_ALLOW_BLESSED)
517 encode_str (enc, "null", 4, 0);
518 else
519 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
520 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
521 }
522 }
523 else if (svt == SVt_PVHV)
524 encode_hv (enc, (HV *)sv);
525 else if (svt == SVt_PVAV)
526 encode_av (enc, (AV *)sv);
527 else if (svt < SVt_PVAV)
528 {
529 STRLEN len = 0;
530 char *pv = svt ? SvPV (sv, len) : 0;
531
532 if (len == 1 && *pv == '1')
533 encode_str (enc, "true", 4, 0);
534 else if (len == 1 && *pv == '0')
535 encode_str (enc, "false", 5, 0);
536 else
537 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
538 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
539 }
540 else
541 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312} 543}
313 544
314static void 545static void
315encode_sv (enc_t *enc, SV *sv) 546encode_sv (enc_t *enc, SV *sv)
316{ 547{
548 SvGETMAGIC (sv);
549
317 if (SvPOKp (sv)) 550 if (SvPOKp (sv))
318 { 551 {
319 STRLEN len; 552 STRLEN len;
320 char *str = SvPV (sv, len); 553 char *str = SvPV (sv, len);
321 encode_ch (enc, '"'); 554 encode_ch (enc, '"');
322 encode_str (enc, str, len, SvUTF8 (sv)); 555 encode_str (enc, str, len, SvUTF8 (sv));
323 encode_ch (enc, '"'); 556 encode_ch (enc, '"');
324 } 557 }
325 else if (SvNOKp (sv)) 558 else if (SvNOKp (sv))
326 { 559 {
560 // trust that perl will do the right thing w.r.t. JSON syntax.
327 need (enc, NV_DIG + 32); 561 need (enc, NV_DIG + 32);
328 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 562 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
329 enc->cur += strlen (enc->cur); 563 enc->cur += strlen (enc->cur);
330 } 564 }
331 else if (SvIOKp (sv)) 565 else if (SvIOKp (sv))
332 { 566 {
333 need (enc, 64); 567 // we assume we can always read an IV as a UV
568 if (SvUV (sv) & ~(UV)0x7fff)
569 {
570 // large integer, use the (rather slow) snprintf way.
571 need (enc, sizeof (UV) * 3);
334 enc->cur += 572 enc->cur +=
335 SvIsUV(sv) 573 SvIsUV(sv)
336 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 574 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
337 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 575 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
576 }
577 else
578 {
579 // optimise the "small number case"
580 // code will likely be branchless and use only a single multiplication
581 I32 i = SvIV (sv);
582 U32 u;
583 char digit, nz = 0;
584
585 need (enc, 6);
586
587 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
588 u = i < 0 ? -i : i;
589
590 // convert to 4.28 fixed-point representation
591 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
592
593 // now output digit by digit, each time masking out the integer part
594 // and multiplying by 5 while moving the decimal point one to the right,
595 // resulting in a net multiplication by 10.
596 // we always write the digit to memory but conditionally increment
597 // the pointer, to ease the usage of conditional move instructions.
598 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
599 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
600 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
601 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
602 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
603 }
338 } 604 }
339 else if (SvROK (sv)) 605 else if (SvROK (sv))
340 { 606 encode_rv (enc, SvRV (sv));
341 if (!--enc->max_recurse)
342 croak ("data structure too deep (hit recursion limit)");
343
344 sv = SvRV (sv);
345
346 switch (SvTYPE (sv))
347 {
348 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
349 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
350
351 default:
352 croak ("JSON can only represent references to arrays or hashes");
353 }
354 }
355 else if (!SvOK (sv)) 607 else if (!SvOK (sv))
356 encode_str (enc, "null", 4, 0); 608 encode_str (enc, "null", 4, 0);
357 else 609 else
358 croak ("encountered perl type that JSON cannot handle"); 610 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
611 SvPV_nolen (sv), SvFLAGS (sv));
359} 612}
360 613
361static SV * 614static SV *
362encode_json (SV *scalar, UV flags) 615encode_json (SV *scalar, JSON *json)
363{ 616{
364 enc_t enc; 617 enc_t enc;
365 enc.flags = flags; 618
619 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
620 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
621
622 enc.json = *json;
366 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 623 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
367 enc.cur = SvPVX (enc.sv); 624 enc.cur = SvPVX (enc.sv);
368 enc.end = SvEND (enc.sv); 625 enc.end = SvEND (enc.sv);
369 enc.max_recurse = 0;
370 enc.indent = 0; 626 enc.indent = 0;
627 enc.maxdepth = DEC_DEPTH (enc.json.flags);
371 628
372 SvPOK_only (enc.sv); 629 SvPOK_only (enc.sv);
373 encode_sv (&enc, scalar); 630 encode_sv (&enc, scalar);
374 631
632 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
633 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
634
375 if (!(flags & (F_ASCII | F_UTF8))) 635 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
376 SvUTF8_on (enc.sv); 636 SvUTF8_on (enc.sv);
377 637
378 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 638 if (enc.json.flags & F_SHRINK)
639 shrink (enc.sv);
640
379 return enc.sv; 641 return enc.sv;
380} 642}
381 643
382///////////////////////////////////////////////////////////////////////////// 644/////////////////////////////////////////////////////////////////////////////
645// decoder
383 646
384#define WS \ 647// structure used for decoding JSON
648typedef struct
649{
650 char *cur; // current parser pointer
651 char *end; // end of input string
652 const char *err; // parse error, if != 0
653 JSON json;
654 U32 depth; // recursion depth
655 U32 maxdepth; // recursion depth limit
656} dec_t;
657
658inline void
659decode_ws (dec_t *dec)
660{
385 for (;;) \ 661 for (;;)
386 { \ 662 {
387 char ch = *dec->cur; \ 663 char ch = *dec->cur;
664
388 if (ch > 0x20 \ 665 if (ch > 0x20
389 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 666 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
390 break; \ 667 break;
668
391 ++dec->cur; \ 669 ++dec->cur;
392 } 670 }
671}
393 672
394#define ERR(reason) SB dec->err = reason; goto fail; SE 673#define ERR(reason) SB dec->err = reason; goto fail; SE
674
395#define EXPECT_CH(ch) SB \ 675#define EXPECT_CH(ch) SB \
396 if (*dec->cur != ch) \ 676 if (*dec->cur != ch) \
397 ERR (# ch " expected"); \ 677 ERR (# ch " expected"); \
398 ++dec->cur; \ 678 ++dec->cur; \
399 SE 679 SE
400 680
681#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
682#define DEC_DEC_DEPTH --dec->depth
683
401static SV *decode_sv (dec_t *dec); 684static SV *decode_sv (dec_t *dec);
402
403#define APPEND_CH(ch) SB \
404 SvGROW (sv, cur + 1 + 1); \
405 SvPVX (sv)[cur++] = (ch); \
406 SE
407 685
408static signed char decode_hexdigit[256]; 686static signed char decode_hexdigit[256];
409 687
410static UV 688static UV
411decode_4hex (dec_t *dec) 689decode_4hex (dec_t *dec)
412{ 690{
413 signed char d1, d2, d3, d4; 691 signed char d1, d2, d3, d4;
692 unsigned char *cur = (unsigned char *)dec->cur;
414 693
415 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 694 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
416 if (d1 < 0) ERR ("four hexadecimal digits expected"); 695 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
417 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 696 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
418 if (d2 < 0) ERR ("four hexadecimal digits expected"); 697 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
419 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
420 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");
423 698
424 dec->cur += 4; 699 dec->cur += 4;
425 700
426 return ((UV)d1) << 12 701 return ((UV)d1) << 12
427 | ((UV)d2) << 8 702 | ((UV)d2) << 8
… …
433} 708}
434 709
435static SV * 710static SV *
436decode_str (dec_t *dec) 711decode_str (dec_t *dec)
437{ 712{
438 SV *sv = NEWSV (0,2); 713 SV *sv = 0;
439 STRLEN cur = 0;
440 int utf8 = 0; 714 int utf8 = 0;
715 char *dec_cur = dec->cur;
441 716
442 for (;;) 717 do
443 { 718 {
444 unsigned char ch = *(unsigned char *)dec->cur; 719 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
720 char *cur = buf;
445 721
446 if (ch == '"') 722 do
447 break;
448 else if (ch == '\\')
449 { 723 {
450 switch (*++dec->cur) 724 unsigned char ch = *(unsigned char *)dec_cur++;
725
726 if (expect_false (ch == '"'))
727 {
728 --dec_cur;
729 break;
451 { 730 }
452 case '\\': 731 else if (expect_false (ch == '\\'))
453 case '/': 732 {
454 case '"': APPEND_CH (*dec->cur++); break; 733 switch (*dec_cur)
455
456 case 'b': APPEND_CH ('\010'); ++dec->cur; break;
457 case 't': APPEND_CH ('\011'); ++dec->cur; break;
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 { 734 {
464 UV lo, hi; 735 case '\\':
465 ++dec->cur; 736 case '/':
737 case '"': *cur++ = *dec_cur++; break;
466 738
467 hi = decode_4hex (dec); 739 case 'b': ++dec_cur; *cur++ = '\010'; break;
468 if (hi == (UV)-1) 740 case 't': ++dec_cur; *cur++ = '\011'; break;
469 goto fail; 741 case 'n': ++dec_cur; *cur++ = '\012'; break;
742 case 'f': ++dec_cur; *cur++ = '\014'; break;
743 case 'r': ++dec_cur; *cur++ = '\015'; break;
470 744
471 // possibly a surrogate pair 745 case 'u':
472 if (hi >= 0xd800 && hi < 0xdc00)
473 { 746 {
474 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 747 UV lo, hi;
475 ERR ("illegal surrogate character"); 748 ++dec_cur;
476 749
477 dec->cur += 2; 750 dec->cur = dec_cur;
478
479 lo = decode_4hex (dec); 751 hi = decode_4hex (dec);
752 dec_cur = dec->cur;
480 if (lo == (UV)-1) 753 if (hi == (UV)-1)
481 goto fail; 754 goto fail;
482 755
756 // possibly a surrogate pair
757 if (hi >= 0xd800)
758 if (hi < 0xdc00)
759 {
760 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
761 ERR ("missing low surrogate character in surrogate pair");
762
763 dec_cur += 2;
764
765 dec->cur = dec_cur;
766 lo = decode_4hex (dec);
767 dec_cur = dec->cur;
768 if (lo == (UV)-1)
769 goto fail;
770
483 if (lo < 0xdc00 || lo >= 0xe000) 771 if (lo < 0xdc00 || lo >= 0xe000)
484 ERR ("surrogate pair expected"); 772 ERR ("surrogate pair expected");
485 773
486 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 774 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
775 }
776 else if (hi < 0xe000)
777 ERR ("missing high surrogate character in surrogate pair");
778
779 if (hi >= 0x80)
780 {
781 utf8 = 1;
782
783 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
784 }
785 else
786 *cur++ = hi;
487 } 787 }
488 else if (lo >= 0xdc00 && lo < 0xe000)
489 ERR ("illegal surrogate character");
490
491 if (hi >= 0x80)
492 { 788 break;
493 utf8 = 1;
494 789
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 790 default:
499 APPEND_CH (hi); 791 --dec_cur;
792 ERR ("illegal backslash escape sequence in string");
500 } 793 }
794 }
795 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
796 *cur++ = ch;
797 else if (ch >= 0x80)
798 {
799 STRLEN clen;
800 UV uch;
801
802 --dec_cur;
803
804 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
805 if (clen == (STRLEN)-1)
806 ERR ("malformed UTF-8 character in JSON string");
807
808 do
809 *cur++ = *dec_cur++;
810 while (--clen);
811
812 utf8 = 1;
813 }
814 else
815 {
816 --dec_cur;
817
818 if (!ch)
819 ERR ("unexpected end of string while parsing JSON string");
501 break; 820 else
821 ERR ("invalid character encountered while parsing JSON string");
502 } 822 }
503 } 823 }
504 else if (ch >= 0x20 && ch <= 0x7f) 824 while (cur < buf + SHORT_STRING_LEN);
505 APPEND_CH (*dec->cur++); 825
506 else if (ch >= 0x80) 826 {
827 STRLEN len = cur - buf;
828
829 if (sv)
507 { 830 {
508 STRLEN clen; 831 SvGROW (sv, SvCUR (sv) + len + 1);
509 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 832 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
510 if (clen < 0) 833 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 } 834 }
517 else 835 else
518 ERR ("invalid character encountered"); 836 sv = newSVpvn (buf, len);
519 } 837 }
838 }
839 while (*dec_cur != '"');
520 840
521 ++dec->cur; 841 ++dec_cur;
522 842
843 if (sv)
844 {
523 SvPOK_only (sv); 845 SvPOK_only (sv);
524
525 SvCUR_set (sv, cur);
526 *SvEND (sv) = 0; 846 *SvEND (sv) = 0;
527 847
528 if (utf8) 848 if (utf8)
529 SvUTF8_on (sv); 849 SvUTF8_on (sv);
850 }
851 else
852 sv = newSVpvn ("", 0);
530 853
854 dec->cur = dec_cur;
531 return sv; 855 return sv;
532 856
533fail: 857fail:
534 SvREFCNT_dec (sv); 858 dec->cur = dec_cur;
535 return 0; 859 return 0;
536} 860}
537 861
538static SV * 862static SV *
539decode_num (dec_t *dec) 863decode_num (dec_t *dec)
… …
549 { 873 {
550 ++dec->cur; 874 ++dec->cur;
551 if (*dec->cur >= '0' && *dec->cur <= '9') 875 if (*dec->cur >= '0' && *dec->cur <= '9')
552 ERR ("malformed number (leading zero must not be followed by another digit)"); 876 ERR ("malformed number (leading zero must not be followed by another digit)");
553 } 877 }
554 878 else if (*dec->cur < '0' || *dec->cur > '9')
555 // int 879 ERR ("malformed number (no digits after initial minus)");
880 else
881 do
882 {
883 ++dec->cur;
884 }
556 while (*dec->cur >= '0' && *dec->cur <= '9') 885 while (*dec->cur >= '0' && *dec->cur <= '9');
557 ++dec->cur;
558 886
559 // [frac] 887 // [frac]
560 if (*dec->cur == '.') 888 if (*dec->cur == '.')
561 { 889 {
562 is_nv = 1; 890 ++dec->cur;
891
892 if (*dec->cur < '0' || *dec->cur > '9')
893 ERR ("malformed number (no digits after decimal point)");
563 894
564 do 895 do
565 { 896 {
566 ++dec->cur; 897 ++dec->cur;
567 } 898 }
568 while (*dec->cur >= '0' && *dec->cur <= '9'); 899 while (*dec->cur >= '0' && *dec->cur <= '9');
900
901 is_nv = 1;
569 } 902 }
570 903
571 // [exp] 904 // [exp]
572 if (*dec->cur == 'e' || *dec->cur == 'E') 905 if (*dec->cur == 'e' || *dec->cur == 'E')
573 { 906 {
574 is_nv = 1;
575
576 ++dec->cur; 907 ++dec->cur;
908
577 if (*dec->cur == '-' || *dec->cur == '+') 909 if (*dec->cur == '-' || *dec->cur == '+')
578 ++dec->cur; 910 ++dec->cur;
579 911
912 if (*dec->cur < '0' || *dec->cur > '9')
913 ERR ("malformed number (no digits after exp sign)");
914
915 do
916 {
917 ++dec->cur;
918 }
580 while (*dec->cur >= '0' && *dec->cur <= '9') 919 while (*dec->cur >= '0' && *dec->cur <= '9');
581 ++dec->cur; 920
921 is_nv = 1;
582 } 922 }
583 923
584 if (!is_nv) 924 if (!is_nv)
585 { 925 {
586 UV uv; 926 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
587 int numtype = grok_number (start, dec->cur - start, &uv); 927 if (*start == '-')
588 if (numtype & IS_NUMBER_IN_UV) 928 switch (dec->cur - start)
589 if (numtype & IS_NUMBER_NEG)
590 { 929 {
591 if (uv < (UV)IV_MIN) 930 case 2: return newSViv (-( start [1] - '0' * 1));
592 return newSViv (-(IV)uv); 931 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
932 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
933 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
593 } 934 }
935 else
936 switch (dec->cur - start)
937 {
938 case 1: return newSViv ( start [0] - '0' * 1);
939 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
940 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
941 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
942 }
943
944 {
945 UV uv;
946 int numtype = grok_number (start, dec->cur - start, &uv);
947 if (numtype & IS_NUMBER_IN_UV)
948 if (numtype & IS_NUMBER_NEG)
949 {
950 if (uv < (UV)IV_MIN)
951 return newSViv (-(IV)uv);
952 }
594 else 953 else
595 return newSVuv (uv); 954 return newSVuv (uv);
955
956 // here would likely be the place for bigint support
596 } 957 }
958 }
597 959
960 // if we ever support bigint or bigfloat, this is the place for bigfloat
598 return newSVnv (Atof (start)); 961 return newSVnv (Atof (start));
599 962
600fail: 963fail:
601 return 0; 964 return 0;
602} 965}
… …
604static SV * 967static SV *
605decode_av (dec_t *dec) 968decode_av (dec_t *dec)
606{ 969{
607 AV *av = newAV (); 970 AV *av = newAV ();
608 971
972 DEC_INC_DEPTH;
973 decode_ws (dec);
974
975 if (*dec->cur == ']')
976 ++dec->cur;
977 else
609 for (;;) 978 for (;;)
610 { 979 {
611 SV *value; 980 SV *value;
612 981
613 value = decode_sv (dec); 982 value = decode_sv (dec);
614 if (!value) 983 if (!value)
615 goto fail; 984 goto fail;
616 985
617 av_push (av, value); 986 av_push (av, value);
618 987
619 WS; 988 decode_ws (dec);
620 989
621 if (*dec->cur == ']') 990 if (*dec->cur == ']')
622 { 991 {
623 ++dec->cur; 992 ++dec->cur;
624 break; 993 break;
994 }
625 } 995
626
627 if (*dec->cur != ',') 996 if (*dec->cur != ',')
628 ERR (", or ] expected while parsing array"); 997 ERR (", or ] expected while parsing array");
629 998
630 ++dec->cur; 999 ++dec->cur;
631 } 1000 }
632 1001
1002 DEC_DEC_DEPTH;
633 return newRV_noinc ((SV *)av); 1003 return newRV_noinc ((SV *)av);
634 1004
635fail: 1005fail:
636 SvREFCNT_dec (av); 1006 SvREFCNT_dec (av);
1007 DEC_DEC_DEPTH;
637 return 0; 1008 return 0;
638} 1009}
639 1010
640static SV * 1011static SV *
641decode_hv (dec_t *dec) 1012decode_hv (dec_t *dec)
642{ 1013{
1014 SV *sv;
643 HV *hv = newHV (); 1015 HV *hv = newHV ();
644 1016
1017 DEC_INC_DEPTH;
1018 decode_ws (dec);
1019
1020 if (*dec->cur == '}')
1021 ++dec->cur;
1022 else
645 for (;;) 1023 for (;;)
646 { 1024 {
647 SV *key, *value; 1025 decode_ws (dec); EXPECT_CH ('"');
648 1026
649 WS; EXPECT_CH ('"'); 1027 // heuristic: assume that
650 1028 // a) decode_str + hv_store_ent are abysmally slow.
651 key = decode_str (dec); 1029 // b) most hash keys are short, simple ascii text.
652 if (!key) 1030 // => try to "fast-match" such strings to avoid
653 goto fail; 1031 // the overhead of decode_str + hv_store_ent.
654
655 WS; EXPECT_CH (':');
656
657 value = decode_sv (dec);
658 if (!value)
659 { 1032 {
1033 SV *value;
1034 char *p = dec->cur;
1035 char *e = p + 24; // only try up to 24 bytes
1036
1037 for (;;)
1038 {
1039 // the >= 0x80 is true on most architectures
1040 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1041 {
1042 // slow path, back up and use decode_str
1043 SV *key = decode_str (dec);
1044 if (!key)
1045 goto fail;
1046
1047 decode_ws (dec); EXPECT_CH (':');
1048
1049 value = decode_sv (dec);
1050 if (!value)
1051 {
1052 SvREFCNT_dec (key);
1053 goto fail;
1054 }
1055
1056 hv_store_ent (hv, key, value, 0);
660 SvREFCNT_dec (key); 1057 SvREFCNT_dec (key);
1058
1059 break;
1060 }
1061 else if (*p == '"')
1062 {
1063 // fast path, got a simple key
1064 char *key = dec->cur;
1065 int len = p - key;
1066 dec->cur = p + 1;
1067
1068 decode_ws (dec); EXPECT_CH (':');
1069
1070 value = decode_sv (dec);
1071 if (!value)
661 goto fail; 1072 goto fail;
1073
1074 hv_store (hv, key, len, value, 0);
1075
1076 break;
1077 }
1078
1079 ++p;
1080 }
662 } 1081 }
663 1082
664 //TODO: optimise 1083 decode_ws (dec);
665 hv_store_ent (hv, key, value, 0);
666 1084
667 WS;
668
669 if (*dec->cur == '}') 1085 if (*dec->cur == '}')
1086 {
1087 ++dec->cur;
1088 break;
1089 }
1090
1091 if (*dec->cur != ',')
1092 ERR (", or } expected while parsing object/hash");
1093
1094 ++dec->cur;
1095 }
1096
1097 DEC_DEC_DEPTH;
1098 sv = newRV_noinc ((SV *)hv);
1099
1100 // check filter callbacks
1101 if (dec->json.flags & F_HOOK)
1102 {
1103 ENTER; SAVETMPS;
1104
1105 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
670 { 1106 {
671 ++dec->cur; 1107 int count;
672 break; 1108
1109 dSP; PUSHMARK (SP);
1110 XPUSHs (sv_2mortal (sv));
1111
1112 PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN;
1113
1114 if (count == 1)
1115 {
1116 sv = newSVsv (POPs);
1117 goto filter_ok;
1118 }
1119
1120 SvREFCNT_inc (sv);
673 } 1121 }
674 1122
675 if (*dec->cur != ',') 1123 if (dec->json.cb_object)
676 ERR (", or } expected while parsing object/hash"); 1124 {
1125 int count;
677 1126
678 ++dec->cur; 1127 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
679 } 1128 XPUSHs (sv_2mortal (sv));
680 1129
681 return newRV_noinc ((SV *)hv); 1130 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1131
1132 if (count == 1)
1133 {
1134 sv = newSVsv (POPs);
1135 goto filter_ok;
1136 }
1137
1138 SvREFCNT_inc (sv);
1139 }
1140
1141filter_ok:
1142 FREETMPS; LEAVE;
1143 }
1144
1145 return sv;
682 1146
683fail: 1147fail:
684 SvREFCNT_dec (hv); 1148 SvREFCNT_dec (hv);
1149 DEC_DEC_DEPTH;
685 return 0; 1150 return 0;
686} 1151}
687 1152
688static SV * 1153static SV *
689decode_sv (dec_t *dec) 1154decode_sv (dec_t *dec)
690{ 1155{
691 WS; 1156 decode_ws (dec);
1157
1158 // the beauty of JSON: you need exactly one character lookahead
1159 // to parse anything.
692 switch (*dec->cur) 1160 switch (*dec->cur)
693 { 1161 {
694 case '"': ++dec->cur; return decode_str (dec); 1162 case '"': ++dec->cur; return decode_str (dec);
695 case '[': ++dec->cur; return decode_av (dec); 1163 case '[': ++dec->cur; return decode_av (dec);
696 case '{': ++dec->cur; return decode_hv (dec); 1164 case '{': ++dec->cur; return decode_hv (dec);
… …
702 1170
703 case 't': 1171 case 't':
704 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1172 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
705 { 1173 {
706 dec->cur += 4; 1174 dec->cur += 4;
707 return newSViv (1); 1175 return SvREFCNT_inc (json_true);
708 } 1176 }
709 else 1177 else
710 ERR ("'true' expected"); 1178 ERR ("'true' expected");
711 1179
712 break; 1180 break;
713 1181
714 case 'f': 1182 case 'f':
715 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1183 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
716 { 1184 {
717 dec->cur += 5; 1185 dec->cur += 5;
718 return newSViv (0); 1186 return SvREFCNT_inc (json_false);
719 } 1187 }
720 else 1188 else
721 ERR ("'false' expected"); 1189 ERR ("'false' expected");
722 1190
723 break; 1191 break;
724 1192
725 case 'n': 1193 case 'n':
726 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1194 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
727 { 1195 {
728 dec->cur += 4; 1196 dec->cur += 4;
729 return newSViv (1); 1197 return newSVsv (&PL_sv_undef);
730 } 1198 }
731 else 1199 else
732 ERR ("'null' expected"); 1200 ERR ("'null' expected");
733 1201
734 break; 1202 break;
735 1203
736 default: 1204 default:
737 ERR ("malformed json string"); 1205 ERR ("malformed JSON string, neither array, object, number, string or atom");
738 break; 1206 break;
739 } 1207 }
740 1208
741fail: 1209fail:
742 return 0; 1210 return 0;
743} 1211}
744 1212
745static SV * 1213static SV *
746decode_json (SV *string, UV flags) 1214decode_json (SV *string, JSON *json, UV *offset_return)
747{ 1215{
1216 dec_t dec;
1217 UV offset;
748 SV *sv; 1218 SV *sv;
749 1219
1220 SvGETMAGIC (string);
1221 SvUPGRADE (string, SVt_PV);
1222
1223 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1224 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1225 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1226
750 if (!(flags & F_UTF8)) 1227 if (json->flags & F_UTF8)
1228 sv_utf8_downgrade (string, 0);
1229 else
751 sv_utf8_upgrade (string); 1230 sv_utf8_upgrade (string);
752 1231
753 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1232 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
754 1233
755 dec_t dec; 1234 dec.json = *json;
756 dec.flags = flags;
757 dec.cur = SvPVX (string); 1235 dec.cur = SvPVX (string);
758 dec.end = SvEND (string); 1236 dec.end = SvEND (string);
759 dec.err = 0; 1237 dec.err = 0;
1238 dec.depth = 0;
1239 dec.maxdepth = DEC_DEPTH (dec.json.flags);
760 1240
761 *dec.end = 1; // invalid anywhere 1241 if (dec.json.cb_object || dec.json.cb_sk_object)
1242 dec.json.flags |= F_HOOK;
1243
1244 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
762 sv = decode_sv (&dec); 1245 sv = decode_sv (&dec);
763 *dec.end = 0; 1246
1247 if (!(offset_return || !sv))
1248 {
1249 // check for trailing garbage
1250 decode_ws (&dec);
1251
1252 if (*dec.cur)
1253 {
1254 dec.err = "garbage after JSON object";
1255 SvREFCNT_dec (sv);
1256 sv = 0;
1257 }
1258 }
1259
1260 if (offset_return || !sv)
1261 {
1262 offset = dec.json.flags & F_UTF8
1263 ? dec.cur - SvPVX (string)
1264 : utf8_distance (dec.cur, SvPVX (string));
1265
1266 if (offset_return)
1267 *offset_return = offset;
1268 }
764 1269
765 if (!sv) 1270 if (!sv)
766 { 1271 {
767 IV offset = utf8_distance (dec.cur, SvPVX (string));
768 SV *uni = sv_newmortal (); 1272 SV *uni = sv_newmortal ();
769 1273
1274 // horrible hack to silence warning inside pv_uni_display
1275 COP cop = *PL_curcop;
1276 cop.cop_warnings = pWARN_NONE;
1277 ENTER;
1278 SAVEVPTR (PL_curcop);
1279 PL_curcop = &cop;
770 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1280 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1281 LEAVE;
1282
771 croak ("%s, at character %d (%s)", 1283 croak ("%s, at character offset %d [\"%s\"]",
772 dec.err, 1284 dec.err,
773 (int)offset, 1285 (int)offset,
774 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1286 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
775 } 1287 }
776 1288
777 sv_dump (sv);//D
778 return sv_2mortal (sv); 1289 sv = sv_2mortal (sv);
1290
1291 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1292 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1293
1294 return sv;
779} 1295}
1296
1297/////////////////////////////////////////////////////////////////////////////
1298// XS interface functions
780 1299
781MODULE = JSON::XS PACKAGE = JSON::XS 1300MODULE = JSON::XS PACKAGE = JSON::XS
782 1301
783BOOT: 1302BOOT:
784{ 1303{
785 int i; 1304 int i;
786 1305
787 memset (decode_hexdigit, 0xff, 256);
788 for (i = 10; i--; ) 1306 for (i = 0; i < 256; ++i)
789 decode_hexdigit ['0' + i] = i; 1307 decode_hexdigit [i] =
1308 i >= '0' && i <= '9' ? i - '0'
1309 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1310 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1311 : -1;
790 1312
791 for (i = 6; --i; ) 1313 json_stash = gv_stashpv ("JSON::XS" , 1);
1314 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1315
1316 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1317 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1318}
1319
1320PROTOTYPES: DISABLE
1321
1322void new (char *klass)
1323 PPCODE:
1324{
1325 SV *pv = NEWSV (0, sizeof (JSON));
1326 SvPOK_only (pv);
1327 Zero (SvPVX (pv), 1, JSON);
1328 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1329 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1330}
1331
1332void ascii (JSON *self, int enable = 1)
1333 ALIAS:
1334 ascii = F_ASCII
1335 latin1 = F_LATIN1
1336 utf8 = F_UTF8
1337 indent = F_INDENT
1338 canonical = F_CANONICAL
1339 space_before = F_SPACE_BEFORE
1340 space_after = F_SPACE_AFTER
1341 pretty = F_PRETTY
1342 allow_nonref = F_ALLOW_NONREF
1343 shrink = F_SHRINK
1344 allow_blessed = F_ALLOW_BLESSED
1345 convert_blessed = F_CONV_BLESSED
1346 PPCODE:
1347{
1348 if (enable)
1349 self->flags |= ix;
1350 else
1351 self->flags &= ~ix;
1352
1353 XPUSHs (ST (0));
1354}
1355
1356void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1357 PPCODE:
1358{
1359 UV log2 = 0;
1360
1361 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1362
1363 while ((1UL << log2) < max_depth)
1364 ++log2;
1365
1366 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1367
1368 XPUSHs (ST (0));
1369}
1370
1371void max_size (JSON *self, UV max_size = 0)
1372 PPCODE:
1373{
1374 UV log2 = 0;
1375
1376 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1377 if (max_size == 1) max_size = 2;
1378
1379 while ((1UL << log2) < max_size)
1380 ++log2;
1381
1382 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1383
1384 XPUSHs (ST (0));
1385}
1386
1387void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1388 ALIAS:
1389 filter_json_single_key_object = 1
1390 PPCODE:
1391{
1392 SV **svp;
1393
1394 if (!SvOK (cb))
1395 cb = 0;
1396 else
1397 cb = newSVsv (cb);
1398
1399 switch (ix)
792 { 1400 {
793 decode_hexdigit ['a' + i] = 10 + i; 1401 case 0: svp = &self->cb_object ; break;
794 decode_hexdigit ['A' + i] = 10 + i; 1402 case 1: svp = &self->cb_sk_object; break;
795 } 1403 }
796 1404
797 json_stash = gv_stashpv ("JSON::XS", 1); 1405 if (*svp)
798} 1406 SvREFCNT_dec (*svp);
799 1407
800SV *new (char *dummy) 1408 *svp = cb;
801 CODE:
802 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
803 OUTPUT:
804 RETVAL
805 1409
806SV *ascii (SV *self, int enable) 1410 XPUSHs (ST (0));
807 ALIAS:
808 ascii = F_ASCII
809 utf8 = F_UTF8
810 indent = F_INDENT
811 canonical = F_CANONICAL
812 space_before = F_SPACE_BEFORE
813 space_after = F_SPACE_AFTER
814 json_rpc = F_JSON_RPC
815 pretty = F_PRETTY
816 CODE:
817{
818 UV *uv = SvJSON (self);
819 if (enable)
820 *uv |= ix;
821 else
822 *uv &= ~ix;
823
824 RETVAL = newSVsv (self);
825} 1411}
826 OUTPUT:
827 RETVAL
828 1412
829void encode (SV *self, SV *scalar) 1413void encode (JSON *self, SV *scalar)
830 PPCODE: 1414 PPCODE:
831 XPUSHs (encode_json (scalar, *SvJSON (self))); 1415 XPUSHs (encode_json (scalar, self));
832 1416
833void decode (SV *self, SV *jsonstr) 1417void decode (JSON *self, SV *jsonstr)
834 PPCODE: 1418 PPCODE:
835 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1419 XPUSHs (decode_json (jsonstr, self, 0));
1420
1421void decode_prefix (JSON *self, SV *jsonstr)
1422 PPCODE:
1423{
1424 UV offset;
1425 EXTEND (SP, 2);
1426 PUSHs (decode_json (jsonstr, self, &offset));
1427 PUSHs (sv_2mortal (newSVuv (offset)));
1428}
1429
1430PROTOTYPES: ENABLE
836 1431
837void to_json (SV *scalar) 1432void to_json (SV *scalar)
838 PPCODE: 1433 PPCODE:
1434{
1435 JSON json = { F_DEFAULT | F_UTF8 };
839 XPUSHs (encode_json (scalar, F_UTF8)); 1436 XPUSHs (encode_json (scalar, &json));
1437}
840 1438
841void from_json (SV *jsonstr) 1439void from_json (SV *jsonstr)
842 PPCODE: 1440 PPCODE:
1441{
1442 JSON json = { F_DEFAULT | F_UTF8 };
843 XPUSHs (decode_json (jsonstr, F_UTF8)); 1443 XPUSHs (decode_json (jsonstr, &json, 0));
1444}
844 1445

Diff Legend

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