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 1.48 by root, Sun Jul 1 22:20:00 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 // NYI
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35
36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
16 38
17#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
18#define F_DEFAULT 0 40#define F_DEFAULT (9UL << S_MAXDEPTH)
19 41
20#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level
44
45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
21 46
22#define SB do { 47#define SB do {
23#define SE } while (0) 48#define SE } while (0)
24 49
25static HV *json_stash; 50#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value))
52# define inline inline
53#else
54# define expect(expr,value) (expr)
55# define inline static
56#endif
57
58#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1)
60
61static HV *json_stash, *json_boolean_stash; // JSON::XS::
62static SV *json_true, *json_false;
63
64typedef struct {
65 U32 flags;
66} JSON;
67
68/////////////////////////////////////////////////////////////////////////////
69// utility functions
70
71inline void
72shrink (SV *sv)
73{
74 sv_utf8_downgrade (sv, 1);
75 if (SvLEN (sv) > SvCUR (sv) + 1)
76 {
77#ifdef SvPV_shrink_to_cur
78 SvPV_shrink_to_cur (sv);
79#elif defined (SvPV_renew)
80 SvPV_renew (sv, SvCUR (sv) + 1);
81#endif
82 }
83}
84
85// decode an utf-8 character and return it, or (UV)-1 in
86// case of an error.
87// we special-case "safe" characters from U+80 .. U+7FF,
88// but use the very good perl function to parse anything else.
89// note that we never call this function for a ascii codepoints
90inline UV
91decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
92{
93 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
94 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
95 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
96 {
97 *clen = 2;
98 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
99 }
100 else
101 {
102 *clen = (STRLEN)-1;
103 return (UV)-1;
104 }
105}
106
107/////////////////////////////////////////////////////////////////////////////
108// encoder
26 109
27// structure used for encoding JSON 110// structure used for encoding JSON
28typedef struct 111typedef struct
29{ 112{
30 char *cur; 113 char *cur; // SvPVX (sv) + current output position
31 STRLEN len; // SvLEN (sv)
32 char *end; // SvEND (sv) 114 char *end; // SvEND (sv)
33 SV *sv; 115 SV *sv; // result scalar
34 UV flags; 116 JSON json;
35 int max_recurse; 117 U32 indent; // indentation level
36 int indent; 118 U32 maxdepth; // max. indentation/recursion level
37} enc_t; 119} enc_t;
38 120
39// structure used for decoding JSON 121inline 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) 122need (enc_t *enc, STRLEN len)
61{ 123{
62 if (enc->cur + len >= enc->end) 124 if (expect_false (enc->cur + len >= enc->end))
63 { 125 {
64 STRLEN cur = enc->cur - SvPVX (enc->sv); 126 STRLEN cur = enc->cur - SvPVX (enc->sv);
65 SvGROW (enc->sv, cur + len + 1); 127 SvGROW (enc->sv, cur + len + 1);
66 enc->cur = SvPVX (enc->sv) + cur; 128 enc->cur = SvPVX (enc->sv) + cur;
67 enc->end = SvEND (enc->sv); 129 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
68 } 130 }
69} 131}
70 132
71static void 133inline void
72encode_ch (enc_t *enc, char ch) 134encode_ch (enc_t *enc, char ch)
73{ 135{
74 need (enc, 1); 136 need (enc, 1);
75 *enc->cur++ = ch; 137 *enc->cur++ = ch;
76} 138}
… …
78static void 140static void
79encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 141encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
80{ 142{
81 char *end = str + len; 143 char *end = str + len;
82 144
145 need (enc, len);
146
83 while (str < end) 147 while (str < end)
84 { 148 {
85 unsigned char ch = *(unsigned char *)str; 149 unsigned char ch = *(unsigned char *)str;
150
86 if (ch >= 0x20 && ch < 0x80) // most common case 151 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
87 { 152 {
153 if (expect_false (ch == '"')) // but with slow exceptions
154 {
155 need (enc, len += 1);
156 *enc->cur++ = '\\';
157 *enc->cur++ = '"';
158 }
159 else if (expect_false (ch == '\\'))
160 {
161 need (enc, len += 1);
162 *enc->cur++ = '\\';
163 *enc->cur++ = '\\';
164 }
165 else
88 *enc->cur++ = ch; 166 *enc->cur++ = ch;
167
89 str++; 168 ++str;
90 } 169 }
91 else 170 else
92 { 171 {
93 STRLEN clen; 172 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 } 173 {
102 else 174 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
103 { 175 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
104 uch = ch; 176 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
105 clen = 1; 177 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
106 } 178 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
107 179
108 need (enc, len += 6); 180 default:
109
110 if (uch < 0xa0 || enc->flags & F_ASCII)
111 {
112 if (uch > 0xFFFFUL)
113 { 181 {
182 STRLEN clen;
183 UV uch;
184
185 if (is_utf8)
186 {
187 uch = decode_utf8 (str, end - str, &clen);
188 if (clen == (STRLEN)-1)
189 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
190 }
191 else
192 {
193 uch = ch;
114 len += 6; 194 clen = 1;
195 }
196
197 if (uch > 0x10FFFFUL)
198 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
199
200 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
201 {
202 if (uch > 0xFFFFUL)
203 {
115 need (enc, len += 6); 204 need (enc, len += 11);
116 sprintf (enc->cur, "\\u%04x\\u%04x", 205 sprintf (enc->cur, "\\u%04x\\u%04x",
117 (uch - 0x10000) / 0x400 + 0xD800, 206 (int)((uch - 0x10000) / 0x400 + 0xD800),
118 (uch - 0x10000) % 0x400 + 0xDC00); 207 (int)((uch - 0x10000) % 0x400 + 0xDC00));
119 enc->cur += 12; 208 enc->cur += 12;
209 }
210 else
211 {
212 static char hexdigit [16] = "0123456789abcdef";
213 need (enc, len += 5);
214 *enc->cur++ = '\\';
215 *enc->cur++ = 'u';
216 *enc->cur++ = hexdigit [ uch >> 12 ];
217 *enc->cur++ = hexdigit [(uch >> 8) & 15];
218 *enc->cur++ = hexdigit [(uch >> 4) & 15];
219 *enc->cur++ = hexdigit [(uch >> 0) & 15];
220 }
221
222 str += clen;
223 }
224 else if (enc->json.flags & F_LATIN1)
225 {
226 *enc->cur++ = uch;
227 str += clen;
228 }
229 else if (is_utf8)
230 {
231 need (enc, len += clen);
232 do
233 {
234 *enc->cur++ = *str++;
235 }
236 while (--clen);
237 }
238 else
239 {
240 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
241 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
242 ++str;
243 }
120 } 244 }
121 else
122 {
123 sprintf (enc->cur, "\\u%04x", uch);
124 enc->cur += 6;
125 }
126 }
127 else if (is_utf8)
128 { 245 }
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 } 246 }
137 247
138 --len; 248 --len;
139 } 249 }
140} 250}
141 251
142#define INDENT SB \ 252inline void
253encode_indent (enc_t *enc)
254{
143 if (enc->flags & F_INDENT) \ 255 if (enc->json.flags & F_INDENT)
144 { \ 256 {
145 int i_; \ 257 int spaces = enc->indent * INDENT_STEP;
146 need (enc, enc->indent); \ 258
147 for (i_ = enc->indent * 3; i_--; )\ 259 need (enc, spaces);
260 memset (enc->cur, ' ', spaces);
261 enc->cur += spaces;
262 }
263}
264
265inline void
266encode_space (enc_t *enc)
267{
268 need (enc, 1);
269 encode_ch (enc, ' ');
270}
271
272inline void
273encode_nl (enc_t *enc)
274{
275 if (enc->json.flags & F_INDENT)
276 {
277 need (enc, 1);
148 encode_ch (enc, ' '); \ 278 encode_ch (enc, '\n');
149 } \ 279 }
150 SE 280}
151 281
152#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 282inline void
153#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 283encode_comma (enc_t *enc)
154#define COMMA SB \ 284{
155 encode_ch (enc, ','); \ 285 encode_ch (enc, ',');
286
156 if (enc->flags & F_INDENT) \ 287 if (enc->json.flags & F_INDENT)
157 NL; \ 288 encode_nl (enc);
158 else if (enc->flags & F_SPACE_AFTER) \ 289 else if (enc->json.flags & F_SPACE_AFTER)
159 SPACE; \ 290 encode_space (enc);
160 SE 291}
161 292
162static void encode_sv (enc_t *enc, SV *sv); 293static void encode_sv (enc_t *enc, SV *sv);
163 294
164static void 295static void
165encode_av (enc_t *enc, AV *av) 296encode_av (enc_t *enc, AV *av)
166{ 297{
167 int i, len = av_len (av); 298 int i, len = av_len (av);
168 299
169 encode_ch (enc, '['); NL; 300 if (enc->indent >= enc->maxdepth)
301 croak ("data structure too deep (hit recursion limit)");
302
303 encode_ch (enc, '['); encode_nl (enc);
170 ++enc->indent; 304 ++enc->indent;
171 305
172 for (i = 0; i <= len; ++i) 306 for (i = 0; i <= len; ++i)
173 { 307 {
174 INDENT; 308 encode_indent (enc);
175 encode_sv (enc, *av_fetch (av, i, 0)); 309 encode_sv (enc, *av_fetch (av, i, 0));
176 310
177 if (i < len) 311 if (i < len)
178 COMMA; 312 encode_comma (enc);
179 } 313 }
180 314
181 NL; 315 encode_nl (enc);
182 316
183 --enc->indent; 317 --enc->indent;
184 INDENT; encode_ch (enc, ']'); 318 encode_indent (enc); encode_ch (enc, ']');
185} 319}
186 320
187static void 321static void
188encode_he (enc_t *enc, HE *he) 322encode_he (enc_t *enc, HE *he)
189{ 323{
… …
191 325
192 if (HeKLEN (he) == HEf_SVKEY) 326 if (HeKLEN (he) == HEf_SVKEY)
193 { 327 {
194 SV *sv = HeSVKEY (he); 328 SV *sv = HeSVKEY (he);
195 STRLEN len; 329 STRLEN len;
330 char *str;
331
332 SvGETMAGIC (sv);
196 char *str = SvPV (sv, len); 333 str = SvPV (sv, len);
197 334
198 encode_str (enc, str, len, SvUTF8 (sv)); 335 encode_str (enc, str, len, SvUTF8 (sv));
199 } 336 }
200 else 337 else
201 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 338 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
202 339
203 encode_ch (enc, '"'); 340 encode_ch (enc, '"');
204 341
205 if (enc->flags & F_SPACE_BEFORE) SPACE; 342 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
206 encode_ch (enc, ':'); 343 encode_ch (enc, ':');
207 if (enc->flags & F_SPACE_AFTER ) SPACE; 344 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
208 encode_sv (enc, HeVAL (he)); 345 encode_sv (enc, HeVAL (he));
209} 346}
210 347
211// compare hash entries, used when all keys are bytestrings 348// compare hash entries, used when all keys are bytestrings
212static int 349static int
… …
218 HE *b = *(HE **)b_; 355 HE *b = *(HE **)b_;
219 356
220 STRLEN la = HeKLEN (a); 357 STRLEN la = HeKLEN (a);
221 STRLEN lb = HeKLEN (b); 358 STRLEN lb = HeKLEN (b);
222 359
223 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 360 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
224 cmp = la < lb ? -1 : la == lb ? 0 : 1; 361 cmp = la - lb;
225 362
226 return cmp; 363 return cmp;
227} 364}
228 365
229// compare hash entries, used when some keys are sv's or utf-x 366// compare hash entries, used when some keys are sv's or utf-x
… …
236static void 373static void
237encode_hv (enc_t *enc, HV *hv) 374encode_hv (enc_t *enc, HV *hv)
238{ 375{
239 int count, i; 376 int count, i;
240 377
378 if (enc->indent >= enc->maxdepth)
379 croak ("data structure too deep (hit recursion limit)");
380
241 encode_ch (enc, '{'); NL; ++enc->indent; 381 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
242 382
243 if ((count = hv_iterinit (hv))) 383 if ((count = hv_iterinit (hv)))
244 { 384 {
245 // for canonical output we have to sort by keys first 385 // for canonical output we have to sort by keys first
246 // actually, this is mostly due to the stupid so-called 386 // actually, this is mostly due to the stupid so-called
247 // security workaround added somewhere in 5.8.x. 387 // security workaround added somewhere in 5.8.x.
248 // that randomises hash orderings 388 // that randomises hash orderings
249 if (enc->flags & F_CANONICAL) 389 if (enc->json.flags & F_CANONICAL)
250 { 390 {
251 HE *he, *hes [count];
252 int fast = 1; 391 int fast = 1;
392 HE *he;
393#if defined(__BORLANDC__) || defined(_MSC_VER)
394 HE **hes = _alloca (count * sizeof (HE));
395#else
396 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
397#endif
253 398
254 i = 0; 399 i = 0;
255 while ((he = hv_iternext (hv))) 400 while ((he = hv_iternext (hv)))
256 { 401 {
257 hes [i++] = he; 402 hes [i++] = he;
… …
263 408
264 if (fast) 409 if (fast)
265 qsort (hes, count, sizeof (HE *), he_cmp_fast); 410 qsort (hes, count, sizeof (HE *), he_cmp_fast);
266 else 411 else
267 { 412 {
268 // hack to disable "use bytes" 413 // hack to forcefully disable "use bytes"
269 COP *oldcop = PL_curcop, cop; 414 COP cop = *PL_curcop;
270 cop.op_private = 0; 415 cop.op_private = 0;
416
417 ENTER;
418 SAVETMPS;
419
420 SAVEVPTR (PL_curcop);
271 PL_curcop = &cop; 421 PL_curcop = &cop;
272 422
273 SAVETMPS;
274 qsort (hes, count, sizeof (HE *), he_cmp_slow); 423 qsort (hes, count, sizeof (HE *), he_cmp_slow);
424
275 FREETMPS; 425 FREETMPS;
276 426 LEAVE;
277 PL_curcop = oldcop;
278 } 427 }
279 428
280 for (i = 0; i < count; ++i) 429 for (i = 0; i < count; ++i)
281 { 430 {
282 INDENT; 431 encode_indent (enc);
283 encode_he (enc, hes [i]); 432 encode_he (enc, hes [i]);
284 433
285 if (i < count - 1) 434 if (i < count - 1)
286 COMMA; 435 encode_comma (enc);
287 }
288
289 NL; 436 }
437
438 encode_nl (enc);
290 } 439 }
291 else 440 else
292 { 441 {
293 SV *sv;
294 HE *he = hv_iternext (hv); 442 HE *he = hv_iternext (hv);
295 443
296 for (;;) 444 for (;;)
297 { 445 {
298 INDENT; 446 encode_indent (enc);
299 encode_he (enc, he); 447 encode_he (enc, he);
300 448
301 if (!(he = hv_iternext (hv))) 449 if (!(he = hv_iternext (hv)))
302 break; 450 break;
303 451
304 COMMA; 452 encode_comma (enc);
305 }
306
307 NL; 453 }
454
455 encode_nl (enc);
308 } 456 }
309 } 457 }
310 458
311 --enc->indent; INDENT; encode_ch (enc, '}'); 459 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
460}
461
462// encode objects, arrays and special \0=false and \1=true values.
463static void
464encode_rv (enc_t *enc, SV *sv)
465{
466 svtype svt;
467
468 SvGETMAGIC (sv);
469 svt = SvTYPE (sv);
470
471 if (expect_false (SvOBJECT (sv)))
472 {
473 if (SvSTASH (sv) == json_boolean_stash)
474 {
475 if (SvIV (sv) == 0)
476 encode_str (enc, "false", 5, 0);
477 else
478 encode_str (enc, "true", 4, 0);
479 }
480 else
481 {
482#if 0
483 if (0 && sv_derived_from (rv, "JSON::Literal"))
484 {
485 // not yet
486 }
487#endif
488 if (enc->json.flags & F_CONV_BLESSED)
489 {
490 // we re-bless the reference to get overload and other niceties right
491 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
492
493 if (to_json)
494 {
495 dSP;
496 ENTER;
497 SAVETMPS;
498 PUSHMARK (SP);
499 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
500
501 // calling with G_SCALAR ensures that we always get a 1 reutrn value
502 // check anyways.
503 PUTBACK;
504 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
505 SPAGAIN;
506
507 encode_sv (enc, POPs);
508
509 FREETMPS;
510 LEAVE;
511 }
512 else if (enc->json.flags & F_ALLOW_BLESSED)
513 encode_str (enc, "null", 4, 0);
514 else
515 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
516 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
517 }
518 else if (enc->json.flags & F_ALLOW_BLESSED)
519 encode_str (enc, "null", 4, 0);
520 else
521 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
522 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
523 }
524 }
525 else if (svt == SVt_PVHV)
526 encode_hv (enc, (HV *)sv);
527 else if (svt == SVt_PVAV)
528 encode_av (enc, (AV *)sv);
529 else if (svt < SVt_PVAV)
530 {
531 if (SvNIOK (sv) && SvIV (sv) == 0)
532 encode_str (enc, "false", 5, 0);
533 else if (SvNIOK (sv) && SvIV (sv) == 1)
534 encode_str (enc, "true", 4, 0);
535 else
536 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
537 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
538 }
539 else
540 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
541 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312} 542}
313 543
314static void 544static void
315encode_sv (enc_t *enc, SV *sv) 545encode_sv (enc_t *enc, SV *sv)
316{ 546{
547 SvGETMAGIC (sv);
548
317 if (SvPOKp (sv)) 549 if (SvPOKp (sv))
318 { 550 {
319 STRLEN len; 551 STRLEN len;
320 char *str = SvPV (sv, len); 552 char *str = SvPV (sv, len);
321 encode_ch (enc, '"'); 553 encode_ch (enc, '"');
322 encode_str (enc, str, len, SvUTF8 (sv)); 554 encode_str (enc, str, len, SvUTF8 (sv));
323 encode_ch (enc, '"'); 555 encode_ch (enc, '"');
324 } 556 }
325 else if (SvNOKp (sv)) 557 else if (SvNOKp (sv))
326 { 558 {
559 // trust that perl will do the right thing w.r.t. JSON syntax.
327 need (enc, NV_DIG + 32); 560 need (enc, NV_DIG + 32);
328 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 561 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
329 enc->cur += strlen (enc->cur); 562 enc->cur += strlen (enc->cur);
330 } 563 }
331 else if (SvIOKp (sv)) 564 else if (SvIOKp (sv))
332 { 565 {
333 need (enc, 64); 566 // we assume we can always read an IV as a UV
567 if (SvUV (sv) & ~(UV)0x7fff)
568 {
569 // large integer, use the (rather slow) snprintf way.
570 need (enc, sizeof (UV) * 3);
334 enc->cur += 571 enc->cur +=
335 SvIsUV(sv) 572 SvIsUV(sv)
336 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 573 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
337 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 574 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
575 }
576 else
577 {
578 // optimise the "small number case"
579 // code will likely be branchless and use only a single multiplication
580 I32 i = SvIV (sv);
581 U32 u;
582 char digit, nz = 0;
583
584 need (enc, 6);
585
586 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
587 u = i < 0 ? -i : i;
588
589 // convert to 4.28 fixed-point representation
590 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
591
592 // now output digit by digit, each time masking out the integer part
593 // and multiplying by 5 while moving the decimal point one to the right,
594 // resulting in a net multiplication by 10.
595 // we always write the digit to memory but conditionally increment
596 // the pointer, to ease the usage of conditional move instructions.
597 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
598 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
599 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
600 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
601 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
602 }
338 } 603 }
339 else if (SvROK (sv)) 604 else if (SvROK (sv))
340 { 605 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)) 606 else if (!SvOK (sv))
356 encode_str (enc, "null", 4, 0); 607 encode_str (enc, "null", 4, 0);
357 else 608 else
358 croak ("encountered perl type that JSON cannot handle"); 609 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
610 SvPV_nolen (sv), SvFLAGS (sv));
359} 611}
360 612
361static SV * 613static SV *
362encode_json (SV *scalar, UV flags) 614encode_json (SV *scalar, JSON *json)
363{ 615{
364 enc_t enc; 616 enc_t enc;
365 enc.flags = flags; 617
618 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
619 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
620
621 enc.json = *json;
366 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 622 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
367 enc.cur = SvPVX (enc.sv); 623 enc.cur = SvPVX (enc.sv);
368 enc.end = SvEND (enc.sv); 624 enc.end = SvEND (enc.sv);
369 enc.max_recurse = 0;
370 enc.indent = 0; 625 enc.indent = 0;
626 enc.maxdepth = DEC_DEPTH (enc.json.flags);
371 627
372 SvPOK_only (enc.sv); 628 SvPOK_only (enc.sv);
373 encode_sv (&enc, scalar); 629 encode_sv (&enc, scalar);
374 630
631 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
632 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
633
375 if (!(flags & (F_ASCII | F_UTF8))) 634 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
376 SvUTF8_on (enc.sv); 635 SvUTF8_on (enc.sv);
377 636
378 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 637 if (enc.json.flags & F_SHRINK)
638 shrink (enc.sv);
639
379 return enc.sv; 640 return enc.sv;
380} 641}
381 642
382///////////////////////////////////////////////////////////////////////////// 643/////////////////////////////////////////////////////////////////////////////
644// decoder
383 645
384#define WS \ 646// structure used for decoding JSON
647typedef struct
648{
649 char *cur; // current parser pointer
650 char *end; // end of input string
651 const char *err; // parse error, if != 0
652 JSON json;
653 U32 depth; // recursion depth
654 U32 maxdepth; // recursion depth limit
655} dec_t;
656
657inline void
658decode_ws (dec_t *dec)
659{
385 for (;;) \ 660 for (;;)
386 { \ 661 {
387 char ch = *dec->cur; \ 662 char ch = *dec->cur;
663
388 if (ch > 0x20 \ 664 if (ch > 0x20
389 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 665 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
390 break; \ 666 break;
667
391 ++dec->cur; \ 668 ++dec->cur;
392 } 669 }
670}
393 671
394#define ERR(reason) SB dec->err = reason; goto fail; SE 672#define ERR(reason) SB dec->err = reason; goto fail; SE
673
395#define EXPECT_CH(ch) SB \ 674#define EXPECT_CH(ch) SB \
396 if (*dec->cur != ch) \ 675 if (*dec->cur != ch) \
397 ERR (# ch " expected"); \ 676 ERR (# ch " expected"); \
398 ++dec->cur; \ 677 ++dec->cur; \
399 SE 678 SE
400 679
680#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
681#define DEC_DEC_DEPTH --dec->depth
682
401static SV *decode_sv (dec_t *dec); 683static 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
407 684
408static signed char decode_hexdigit[256]; 685static signed char decode_hexdigit[256];
409 686
410static UV 687static UV
411decode_4hex (dec_t *dec) 688decode_4hex (dec_t *dec)
412{ 689{
413 signed char d1, d2, d3, d4; 690 signed char d1, d2, d3, d4;
691 unsigned char *cur = (unsigned char *)dec->cur;
414 692
415 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 693 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"); 694 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]]; 695 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"); 696 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");
423 697
424 dec->cur += 4; 698 dec->cur += 4;
425 699
426 return ((UV)d1) << 12 700 return ((UV)d1) << 12
427 | ((UV)d2) << 8 701 | ((UV)d2) << 8
… …
433} 707}
434 708
435static SV * 709static SV *
436decode_str (dec_t *dec) 710decode_str (dec_t *dec)
437{ 711{
438 SV *sv = NEWSV (0,2); 712 SV *sv = 0;
439 STRLEN cur = 0;
440 int utf8 = 0; 713 int utf8 = 0;
714 char *dec_cur = dec->cur;
441 715
442 for (;;) 716 do
443 { 717 {
444 unsigned char ch = *(unsigned char *)dec->cur; 718 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
719 char *cur = buf;
445 720
446 if (ch == '"') 721 do
447 break;
448 else if (ch == '\\')
449 { 722 {
450 switch (*++dec->cur) 723 unsigned char ch = *(unsigned char *)dec_cur++;
724
725 if (expect_false (ch == '"'))
726 {
727 --dec_cur;
728 break;
451 { 729 }
452 case '\\': 730 else if (expect_false (ch == '\\'))
453 case '/': 731 {
454 case '"': APPEND_CH (*dec->cur++); break; 732 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 { 733 {
464 UV lo, hi; 734 case '\\':
465 ++dec->cur; 735 case '/':
736 case '"': *cur++ = *dec_cur++; break;
466 737
467 hi = decode_4hex (dec); 738 case 'b': ++dec_cur; *cur++ = '\010'; break;
468 if (hi == (UV)-1) 739 case 't': ++dec_cur; *cur++ = '\011'; break;
469 goto fail; 740 case 'n': ++dec_cur; *cur++ = '\012'; break;
741 case 'f': ++dec_cur; *cur++ = '\014'; break;
742 case 'r': ++dec_cur; *cur++ = '\015'; break;
470 743
471 // possibly a surrogate pair 744 case 'u':
472 if (hi >= 0xd800 && hi < 0xdc00)
473 { 745 {
474 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 746 UV lo, hi;
475 ERR ("illegal surrogate character"); 747 ++dec_cur;
476 748
477 dec->cur += 2; 749 dec->cur = dec_cur;
478
479 lo = decode_4hex (dec); 750 hi = decode_4hex (dec);
751 dec_cur = dec->cur;
480 if (lo == (UV)-1) 752 if (hi == (UV)-1)
481 goto fail; 753 goto fail;
482 754
755 // possibly a surrogate pair
756 if (hi >= 0xd800)
757 if (hi < 0xdc00)
758 {
759 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
760 ERR ("missing low surrogate character in surrogate pair");
761
762 dec_cur += 2;
763
764 dec->cur = dec_cur;
765 lo = decode_4hex (dec);
766 dec_cur = dec->cur;
767 if (lo == (UV)-1)
768 goto fail;
769
483 if (lo < 0xdc00 || lo >= 0xe000) 770 if (lo < 0xdc00 || lo >= 0xe000)
484 ERR ("surrogate pair expected"); 771 ERR ("surrogate pair expected");
485 772
486 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 773 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
774 }
775 else if (hi < 0xe000)
776 ERR ("missing high surrogate character in surrogate pair");
777
778 if (hi >= 0x80)
779 {
780 utf8 = 1;
781
782 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
783 }
784 else
785 *cur++ = hi;
487 } 786 }
488 else if (lo >= 0xdc00 && lo < 0xe000)
489 ERR ("illegal surrogate character");
490
491 if (hi >= 0x80)
492 { 787 break;
493 utf8 = 1;
494 788
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 789 default:
499 APPEND_CH (hi); 790 --dec_cur;
791 ERR ("illegal backslash escape sequence in string");
500 } 792 }
793 }
794 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
795 *cur++ = ch;
796 else if (ch >= 0x80)
797 {
798 STRLEN clen;
799 UV uch;
800
801 --dec_cur;
802
803 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
804 if (clen == (STRLEN)-1)
805 ERR ("malformed UTF-8 character in JSON string");
806
807 do
808 *cur++ = *dec_cur++;
809 while (--clen);
810
811 utf8 = 1;
812 }
813 else
814 {
815 --dec_cur;
816
817 if (!ch)
818 ERR ("unexpected end of string while parsing JSON string");
501 break; 819 else
820 ERR ("invalid character encountered while parsing JSON string");
502 } 821 }
503 } 822 }
504 else if (ch >= 0x20 && ch <= 0x7f) 823 while (cur < buf + SHORT_STRING_LEN);
505 APPEND_CH (*dec->cur++); 824
506 else if (ch >= 0x80) 825 {
826 STRLEN len = cur - buf;
827
828 if (sv)
507 { 829 {
508 STRLEN clen; 830 SvGROW (sv, SvCUR (sv) + len + 1);
509 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 831 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
510 if (clen < 0) 832 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 } 833 }
517 else 834 else
518 ERR ("invalid character encountered"); 835 sv = newSVpvn (buf, len);
519 } 836 }
837 }
838 while (*dec_cur != '"');
520 839
521 ++dec->cur; 840 ++dec_cur;
522 841
842 if (sv)
843 {
523 SvPOK_only (sv); 844 SvPOK_only (sv);
524
525 SvCUR_set (sv, cur);
526 *SvEND (sv) = 0; 845 *SvEND (sv) = 0;
527 846
528 if (utf8) 847 if (utf8)
529 SvUTF8_on (sv); 848 SvUTF8_on (sv);
849 }
850 else
851 sv = newSVpvn ("", 0);
530 852
853 dec->cur = dec_cur;
531 return sv; 854 return sv;
532 855
533fail: 856fail:
534 SvREFCNT_dec (sv); 857 dec->cur = dec_cur;
535 return 0; 858 return 0;
536} 859}
537 860
538static SV * 861static SV *
539decode_num (dec_t *dec) 862decode_num (dec_t *dec)
… …
549 { 872 {
550 ++dec->cur; 873 ++dec->cur;
551 if (*dec->cur >= '0' && *dec->cur <= '9') 874 if (*dec->cur >= '0' && *dec->cur <= '9')
552 ERR ("malformed number (leading zero must not be followed by another digit)"); 875 ERR ("malformed number (leading zero must not be followed by another digit)");
553 } 876 }
554 877 else if (*dec->cur < '0' || *dec->cur > '9')
555 // int 878 ERR ("malformed number (no digits after initial minus)");
879 else
880 do
881 {
882 ++dec->cur;
883 }
556 while (*dec->cur >= '0' && *dec->cur <= '9') 884 while (*dec->cur >= '0' && *dec->cur <= '9');
557 ++dec->cur;
558 885
559 // [frac] 886 // [frac]
560 if (*dec->cur == '.') 887 if (*dec->cur == '.')
561 { 888 {
562 is_nv = 1; 889 ++dec->cur;
890
891 if (*dec->cur < '0' || *dec->cur > '9')
892 ERR ("malformed number (no digits after decimal point)");
563 893
564 do 894 do
565 { 895 {
566 ++dec->cur; 896 ++dec->cur;
567 } 897 }
568 while (*dec->cur >= '0' && *dec->cur <= '9'); 898 while (*dec->cur >= '0' && *dec->cur <= '9');
899
900 is_nv = 1;
569 } 901 }
570 902
571 // [exp] 903 // [exp]
572 if (*dec->cur == 'e' || *dec->cur == 'E') 904 if (*dec->cur == 'e' || *dec->cur == 'E')
573 { 905 {
574 is_nv = 1;
575
576 ++dec->cur; 906 ++dec->cur;
907
577 if (*dec->cur == '-' || *dec->cur == '+') 908 if (*dec->cur == '-' || *dec->cur == '+')
578 ++dec->cur; 909 ++dec->cur;
579 910
911 if (*dec->cur < '0' || *dec->cur > '9')
912 ERR ("malformed number (no digits after exp sign)");
913
914 do
915 {
916 ++dec->cur;
917 }
580 while (*dec->cur >= '0' && *dec->cur <= '9') 918 while (*dec->cur >= '0' && *dec->cur <= '9');
581 ++dec->cur; 919
920 is_nv = 1;
582 } 921 }
583 922
584 if (!is_nv) 923 if (!is_nv)
585 { 924 {
586 UV uv; 925 // 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); 926 if (*start == '-')
588 if (numtype & IS_NUMBER_IN_UV) 927 switch (dec->cur - start)
589 if (numtype & IS_NUMBER_NEG)
590 { 928 {
591 if (uv < (UV)IV_MIN) 929 case 2: return newSViv (-( start [1] - '0' * 1));
592 return newSViv (-(IV)uv); 930 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
931 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
932 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
593 } 933 }
934 else
935 switch (dec->cur - start)
936 {
937 case 1: return newSViv ( start [0] - '0' * 1);
938 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
939 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
940 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
941 }
942
943 {
944 UV uv;
945 int numtype = grok_number (start, dec->cur - start, &uv);
946 if (numtype & IS_NUMBER_IN_UV)
947 if (numtype & IS_NUMBER_NEG)
948 {
949 if (uv < (UV)IV_MIN)
950 return newSViv (-(IV)uv);
951 }
594 else 952 else
595 return newSVuv (uv); 953 return newSVuv (uv);
954
955 // here would likely be the place for bigint support
596 } 956 }
957 }
597 958
959 // if we ever support bigint or bigfloat, this is the place for bigfloat
598 return newSVnv (Atof (start)); 960 return newSVnv (Atof (start));
599 961
600fail: 962fail:
601 return 0; 963 return 0;
602} 964}
… …
604static SV * 966static SV *
605decode_av (dec_t *dec) 967decode_av (dec_t *dec)
606{ 968{
607 AV *av = newAV (); 969 AV *av = newAV ();
608 970
971 DEC_INC_DEPTH;
972 decode_ws (dec);
973
974 if (*dec->cur == ']')
975 ++dec->cur;
976 else
609 for (;;) 977 for (;;)
610 { 978 {
611 SV *value; 979 SV *value;
612 980
613 value = decode_sv (dec); 981 value = decode_sv (dec);
614 if (!value) 982 if (!value)
615 goto fail; 983 goto fail;
616 984
617 av_push (av, value); 985 av_push (av, value);
618 986
619 WS; 987 decode_ws (dec);
620 988
621 if (*dec->cur == ']') 989 if (*dec->cur == ']')
622 { 990 {
623 ++dec->cur; 991 ++dec->cur;
624 break; 992 break;
993 }
625 } 994
626
627 if (*dec->cur != ',') 995 if (*dec->cur != ',')
628 ERR (", or ] expected while parsing array"); 996 ERR (", or ] expected while parsing array");
629 997
630 ++dec->cur; 998 ++dec->cur;
631 } 999 }
632 1000
1001 DEC_DEC_DEPTH;
633 return newRV_noinc ((SV *)av); 1002 return newRV_noinc ((SV *)av);
634 1003
635fail: 1004fail:
636 SvREFCNT_dec (av); 1005 SvREFCNT_dec (av);
1006 DEC_DEC_DEPTH;
637 return 0; 1007 return 0;
638} 1008}
639 1009
640static SV * 1010static SV *
641decode_hv (dec_t *dec) 1011decode_hv (dec_t *dec)
642{ 1012{
643 HV *hv = newHV (); 1013 HV *hv = newHV ();
644 1014
1015 DEC_INC_DEPTH;
1016 decode_ws (dec);
1017
1018 if (*dec->cur == '}')
1019 ++dec->cur;
1020 else
645 for (;;) 1021 for (;;)
646 { 1022 {
647 SV *key, *value; 1023 decode_ws (dec); EXPECT_CH ('"');
648 1024
649 WS; EXPECT_CH ('"'); 1025 // heuristic: assume that
650 1026 // a) decode_str + hv_store_ent are abysmally slow.
651 key = decode_str (dec); 1027 // b) most hash keys are short, simple ascii text.
652 if (!key) 1028 // => try to "fast-match" such strings to avoid
653 goto fail; 1029 // the overhead of decode_str + hv_store_ent.
654
655 WS; EXPECT_CH (':');
656
657 value = decode_sv (dec);
658 if (!value)
659 { 1030 {
1031 SV *value;
1032 char *p = dec->cur;
1033 char *e = p + 24; // only try up to 24 bytes
1034
1035 for (;;)
1036 {
1037 // the >= 0x80 is true on most architectures
1038 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1039 {
1040 // slow path, back up and use decode_str
1041 SV *key = decode_str (dec);
1042 if (!key)
1043 goto fail;
1044
1045 decode_ws (dec); EXPECT_CH (':');
1046
1047 value = decode_sv (dec);
1048 if (!value)
1049 {
1050 SvREFCNT_dec (key);
1051 goto fail;
1052 }
1053
1054 hv_store_ent (hv, key, value, 0);
660 SvREFCNT_dec (key); 1055 SvREFCNT_dec (key);
1056
1057 break;
1058 }
1059 else if (*p == '"')
1060 {
1061 // fast path, got a simple key
1062 char *key = dec->cur;
1063 int len = p - key;
1064 dec->cur = p + 1;
1065
1066 decode_ws (dec); EXPECT_CH (':');
1067
1068 value = decode_sv (dec);
1069 if (!value)
661 goto fail; 1070 goto fail;
1071
1072 hv_store (hv, key, len, value, 0);
1073
1074 break;
1075 }
1076
1077 ++p;
1078 }
662 } 1079 }
663 1080
664 //TODO: optimise 1081 decode_ws (dec);
665 hv_store_ent (hv, key, value, 0);
666 1082
667 WS;
668
669 if (*dec->cur == '}') 1083 if (*dec->cur == '}')
670 { 1084 {
671 ++dec->cur; 1085 ++dec->cur;
672 break; 1086 break;
673 } 1087 }
674 1088
675 if (*dec->cur != ',') 1089 if (*dec->cur != ',')
676 ERR (", or } expected while parsing object/hash"); 1090 ERR (", or } expected while parsing object/hash");
677 1091
678 ++dec->cur; 1092 ++dec->cur;
679 } 1093 }
680 1094
1095 DEC_DEC_DEPTH;
681 return newRV_noinc ((SV *)hv); 1096 return newRV_noinc ((SV *)hv);
682 1097
683fail: 1098fail:
684 SvREFCNT_dec (hv); 1099 SvREFCNT_dec (hv);
1100 DEC_DEC_DEPTH;
685 return 0; 1101 return 0;
686} 1102}
687 1103
688static SV * 1104static SV *
689decode_sv (dec_t *dec) 1105decode_sv (dec_t *dec)
690{ 1106{
691 WS; 1107 decode_ws (dec);
1108
1109 // the beauty of JSON: you need exactly one character lookahead
1110 // to parse anything.
692 switch (*dec->cur) 1111 switch (*dec->cur)
693 { 1112 {
694 case '"': ++dec->cur; return decode_str (dec); 1113 case '"': ++dec->cur; return decode_str (dec);
695 case '[': ++dec->cur; return decode_av (dec); 1114 case '[': ++dec->cur; return decode_av (dec);
696 case '{': ++dec->cur; return decode_hv (dec); 1115 case '{': ++dec->cur; return decode_hv (dec);
… …
702 1121
703 case 't': 1122 case 't':
704 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1123 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
705 { 1124 {
706 dec->cur += 4; 1125 dec->cur += 4;
707 return newSViv (1); 1126 return SvREFCNT_inc (json_true);
708 } 1127 }
709 else 1128 else
710 ERR ("'true' expected"); 1129 ERR ("'true' expected");
711 1130
712 break; 1131 break;
713 1132
714 case 'f': 1133 case 'f':
715 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1134 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
716 { 1135 {
717 dec->cur += 5; 1136 dec->cur += 5;
718 return newSViv (0); 1137 return SvREFCNT_inc (json_false);
719 } 1138 }
720 else 1139 else
721 ERR ("'false' expected"); 1140 ERR ("'false' expected");
722 1141
723 break; 1142 break;
724 1143
725 case 'n': 1144 case 'n':
726 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1145 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
727 { 1146 {
728 dec->cur += 4; 1147 dec->cur += 4;
729 return newSViv (1); 1148 return newSVsv (&PL_sv_undef);
730 } 1149 }
731 else 1150 else
732 ERR ("'null' expected"); 1151 ERR ("'null' expected");
733 1152
734 break; 1153 break;
735 1154
736 default: 1155 default:
737 ERR ("malformed json string"); 1156 ERR ("malformed JSON string, neither array, object, number, string or atom");
738 break; 1157 break;
739 } 1158 }
740 1159
741fail: 1160fail:
742 return 0; 1161 return 0;
743} 1162}
744 1163
745static SV * 1164static SV *
746decode_json (SV *string, UV flags) 1165decode_json (SV *string, JSON *json, UV *offset_return)
747{ 1166{
1167 dec_t dec;
1168 UV offset;
748 SV *sv; 1169 SV *sv;
749 1170
1171 SvGETMAGIC (string);
1172 SvUPGRADE (string, SVt_PV);
1173
1174 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1175 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1176 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1177
750 if (!(flags & F_UTF8)) 1178 if (json->flags & F_UTF8)
1179 sv_utf8_downgrade (string, 0);
1180 else
751 sv_utf8_upgrade (string); 1181 sv_utf8_upgrade (string);
752 1182
753 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1183 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
754 1184
755 dec_t dec; 1185 dec.json = *json;
756 dec.flags = flags;
757 dec.cur = SvPVX (string); 1186 dec.cur = SvPVX (string);
758 dec.end = SvEND (string); 1187 dec.end = SvEND (string);
759 dec.err = 0; 1188 dec.err = 0;
1189 dec.depth = 0;
1190 dec.maxdepth = DEC_DEPTH (dec.json.flags);
760 1191
761 *dec.end = 1; // invalid anywhere 1192 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
762 sv = decode_sv (&dec); 1193 sv = decode_sv (&dec);
763 *dec.end = 0; 1194
1195 if (!(offset_return || !sv))
1196 {
1197 // check for trailing garbage
1198 decode_ws (&dec);
1199
1200 if (*dec.cur)
1201 {
1202 dec.err = "garbage after JSON object";
1203 SvREFCNT_dec (sv);
1204 sv = 0;
1205 }
1206 }
1207
1208 if (offset_return || !sv)
1209 {
1210 offset = dec.json.flags & F_UTF8
1211 ? dec.cur - SvPVX (string)
1212 : utf8_distance (dec.cur, SvPVX (string));
1213
1214 if (offset_return)
1215 *offset_return = offset;
1216 }
764 1217
765 if (!sv) 1218 if (!sv)
766 { 1219 {
767 IV offset = utf8_distance (dec.cur, SvPVX (string));
768 SV *uni = sv_newmortal (); 1220 SV *uni = sv_newmortal ();
769 1221
1222 // horrible hack to silence warning inside pv_uni_display
1223 COP cop = *PL_curcop;
1224 cop.cop_warnings = pWARN_NONE;
1225 ENTER;
1226 SAVEVPTR (PL_curcop);
1227 PL_curcop = &cop;
770 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1228 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1229 LEAVE;
1230
771 croak ("%s, at character %d (%s)", 1231 croak ("%s, at character offset %d [\"%s\"]",
772 dec.err, 1232 dec.err,
773 (int)offset, 1233 (int)offset,
774 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1234 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
775 } 1235 }
776 1236
777 sv_dump (sv);//D
778 return sv_2mortal (sv); 1237 sv = sv_2mortal (sv);
1238
1239 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1240 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1241
1242 return sv;
779} 1243}
1244
1245/////////////////////////////////////////////////////////////////////////////
1246// XS interface functions
780 1247
781MODULE = JSON::XS PACKAGE = JSON::XS 1248MODULE = JSON::XS PACKAGE = JSON::XS
782 1249
783BOOT: 1250BOOT:
784{ 1251{
785 int i; 1252 int i;
786 1253
787 memset (decode_hexdigit, 0xff, 256);
788 for (i = 10; i--; ) 1254 for (i = 0; i < 256; ++i)
789 decode_hexdigit ['0' + i] = i; 1255 decode_hexdigit [i] =
1256 i >= '0' && i <= '9' ? i - '0'
1257 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1258 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1259 : -1;
790 1260
791 for (i = 6; --i; )
792 {
793 decode_hexdigit ['a' + i] = 10 + i;
794 decode_hexdigit ['A' + i] = 10 + i;
795 }
796
797 json_stash = gv_stashpv ("JSON::XS", 1); 1261 json_stash = gv_stashpv ("JSON::XS" , 1);
798} 1262 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
799 1263
800SV *new (char *dummy) 1264 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1265 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1266}
1267
1268PROTOTYPES: DISABLE
1269
1270void new (char *klass)
801 CODE: 1271 PPCODE:
802 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1272{
803 OUTPUT: 1273 SV *pv = NEWSV (0, sizeof (JSON));
804 RETVAL 1274 SvPOK_only (pv);
1275 Zero (SvPVX (pv), 1, sizeof (JSON));
1276 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1277 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1278}
805 1279
806SV *ascii (SV *self, int enable) 1280void ascii (JSON *self, int enable = 1)
807 ALIAS: 1281 ALIAS:
808 ascii = F_ASCII 1282 ascii = F_ASCII
1283 latin1 = F_LATIN1
809 utf8 = F_UTF8 1284 utf8 = F_UTF8
810 indent = F_INDENT 1285 indent = F_INDENT
811 canonical = F_CANONICAL 1286 canonical = F_CANONICAL
812 space_before = F_SPACE_BEFORE 1287 space_before = F_SPACE_BEFORE
813 space_after = F_SPACE_AFTER 1288 space_after = F_SPACE_AFTER
814 json_rpc = F_JSON_RPC
815 pretty = F_PRETTY 1289 pretty = F_PRETTY
1290 allow_nonref = F_ALLOW_NONREF
1291 shrink = F_SHRINK
1292 allow_blessed = F_ALLOW_BLESSED
1293 convert_blessed = F_CONV_BLESSED
816 CODE: 1294 PPCODE:
817{ 1295{
818 UV *uv = SvJSON (self);
819 if (enable) 1296 if (enable)
820 *uv |= ix; 1297 self->flags |= ix;
821 else 1298 else
822 *uv &= ~ix; 1299 self->flags &= ~ix;
823 1300
824 RETVAL = newSVsv (self); 1301 XPUSHs (ST (0));
825} 1302}
826 OUTPUT:
827 RETVAL
828 1303
829void encode (SV *self, SV *scalar) 1304void max_depth (JSON *self, UV max_depth = 0x80000000UL)
830 PPCODE: 1305 PPCODE:
831 XPUSHs (encode_json (scalar, *SvJSON (self))); 1306{
1307 UV log2 = 0;
832 1308
833void decode (SV *self, SV *jsonstr) 1309 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1310
1311 while ((1UL << log2) < max_depth)
1312 ++log2;
1313
1314 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1315
1316 XPUSHs (ST (0));
1317}
1318
1319void max_size (JSON *self, UV max_size = 0)
834 PPCODE: 1320 PPCODE:
1321{
1322 UV log2 = 0;
1323
1324 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1325 if (max_size == 1) max_size = 2;
1326
1327 while ((1UL << log2) < max_size)
1328 ++log2;
1329
1330 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1331
1332 XPUSHs (ST (0));
1333}
1334
1335void encode (JSON *self, SV *scalar)
1336 PPCODE:
1337 XPUSHs (encode_json (scalar, self));
1338
1339void decode (JSON *self, SV *jsonstr)
1340 PPCODE:
835 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1341 XPUSHs (decode_json (jsonstr, self, 0));
1342
1343void decode_prefix (JSON *self, SV *jsonstr)
1344 PPCODE:
1345{
1346 UV offset;
1347 EXTEND (SP, 2);
1348 PUSHs (decode_json (jsonstr, self, &offset));
1349 PUSHs (sv_2mortal (newSVuv (offset)));
1350}
1351
1352PROTOTYPES: ENABLE
836 1353
837void to_json (SV *scalar) 1354void to_json (SV *scalar)
838 PPCODE: 1355 PPCODE:
1356{
1357 JSON json = { F_DEFAULT | F_UTF8 };
839 XPUSHs (encode_json (scalar, F_UTF8)); 1358 XPUSHs (encode_json (scalar, &json));
1359}
840 1360
841void from_json (SV *jsonstr) 1361void from_json (SV *jsonstr)
842 PPCODE: 1362 PPCODE:
1363{
1364 JSON json = { F_DEFAULT | F_UTF8 };
843 XPUSHs (decode_json (jsonstr, F_UTF8)); 1365 XPUSHs (decode_json (jsonstr, &json, 0));
1366}
844 1367

Diff Legend

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