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 1.52 by root, Mon Jul 2 02:57:11 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;
68 HV *cb_sk_object;
69} JSON;
70
71/////////////////////////////////////////////////////////////////////////////
72// utility functions
73
74inline void
75shrink (SV *sv)
76{
77 sv_utf8_downgrade (sv, 1);
78 if (SvLEN (sv) > SvCUR (sv) + 1)
79 {
80#ifdef SvPV_shrink_to_cur
81 SvPV_shrink_to_cur (sv);
82#elif defined (SvPV_renew)
83 SvPV_renew (sv, SvCUR (sv) + 1);
84#endif
85 }
86}
87
88// decode an utf-8 character and return it, or (UV)-1 in
89// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints
93inline UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
99 {
100 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 }
103 else
104 {
105 *clen = (STRLEN)-1;
106 return (UV)-1;
107 }
108}
109
110/////////////////////////////////////////////////////////////////////////////
111// encoder
26 112
27// structure used for encoding JSON 113// structure used for encoding JSON
28typedef struct 114typedef struct
29{ 115{
30 char *cur; 116 char *cur; // SvPVX (sv) + current output position
31 STRLEN len; // SvLEN (sv)
32 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
33 SV *sv; 118 SV *sv; // result scalar
34 UV flags; 119 JSON json;
35 int max_recurse; 120 U32 indent; // indentation level
36 int indent; 121 U32 maxdepth; // max. indentation/recursion level
37} enc_t; 122} enc_t;
38 123
39// structure used for decoding JSON 124inline 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) 125need (enc_t *enc, STRLEN len)
61{ 126{
62 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
63 { 128 {
64 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
65 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
66 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
67 enc->end = SvEND (enc->sv); 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
68 } 133 }
69} 134}
70 135
71static void 136inline void
72encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
73{ 138{
74 need (enc, 1); 139 need (enc, 1);
75 *enc->cur++ = ch; 140 *enc->cur++ = ch;
76} 141}
… …
78static void 143static void
79encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 144encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
80{ 145{
81 char *end = str + len; 146 char *end = str + len;
82 147
148 need (enc, len);
149
83 while (str < end) 150 while (str < end)
84 { 151 {
85 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
153
86 if (ch >= 0x20 && ch < 0x80) // most common case 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
87 { 155 {
156 if (expect_false (ch == '"')) // but with slow exceptions
157 {
158 need (enc, len += 1);
159 *enc->cur++ = '\\';
160 *enc->cur++ = '"';
161 }
162 else if (expect_false (ch == '\\'))
163 {
164 need (enc, len += 1);
165 *enc->cur++ = '\\';
166 *enc->cur++ = '\\';
167 }
168 else
88 *enc->cur++ = ch; 169 *enc->cur++ = ch;
170
89 str++; 171 ++str;
90 } 172 }
91 else 173 else
92 { 174 {
93 STRLEN clen; 175 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 } 176 {
102 else 177 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
103 { 178 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
104 uch = ch; 179 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
105 clen = 1; 180 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
106 } 181 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
107 182
108 need (enc, len += 6); 183 default:
109
110 if (uch < 0xa0 || enc->flags & F_ASCII)
111 {
112 if (uch > 0xFFFFUL)
113 { 184 {
185 STRLEN clen;
186 UV uch;
187
188 if (is_utf8)
189 {
190 uch = decode_utf8 (str, end - str, &clen);
191 if (clen == (STRLEN)-1)
192 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
193 }
194 else
195 {
196 uch = ch;
114 len += 6; 197 clen = 1;
198 }
199
200 if (uch > 0x10FFFFUL)
201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
202
203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
204 {
205 if (uch > 0xFFFFUL)
206 {
115 need (enc, len += 6); 207 need (enc, len += 11);
116 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
117 (uch - 0x10000) / 0x400 + 0xD800, 209 (int)((uch - 0x10000) / 0x400 + 0xD800),
118 (uch - 0x10000) % 0x400 + 0xDC00); 210 (int)((uch - 0x10000) % 0x400 + 0xDC00));
119 enc->cur += 12; 211 enc->cur += 12;
212 }
213 else
214 {
215 static char hexdigit [16] = "0123456789abcdef";
216 need (enc, len += 5);
217 *enc->cur++ = '\\';
218 *enc->cur++ = 'u';
219 *enc->cur++ = hexdigit [ uch >> 12 ];
220 *enc->cur++ = hexdigit [(uch >> 8) & 15];
221 *enc->cur++ = hexdigit [(uch >> 4) & 15];
222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
223 }
224
225 str += clen;
226 }
227 else if (enc->json.flags & F_LATIN1)
228 {
229 *enc->cur++ = uch;
230 str += clen;
231 }
232 else if (is_utf8)
233 {
234 need (enc, len += clen);
235 do
236 {
237 *enc->cur++ = *str++;
238 }
239 while (--clen);
240 }
241 else
242 {
243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
245 ++str;
246 }
120 } 247 }
121 else
122 {
123 sprintf (enc->cur, "\\u%04x", uch);
124 enc->cur += 6;
125 }
126 }
127 else if (is_utf8)
128 { 248 }
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 } 249 }
137 250
138 --len; 251 --len;
139 } 252 }
140} 253}
141 254
142#define INDENT SB \ 255inline void
256encode_indent (enc_t *enc)
257{
143 if (enc->flags & F_INDENT) \ 258 if (enc->json.flags & F_INDENT)
144 { \ 259 {
145 int i_; \ 260 int spaces = enc->indent * INDENT_STEP;
146 need (enc, enc->indent); \ 261
147 for (i_ = enc->indent * 3; i_--; )\ 262 need (enc, spaces);
263 memset (enc->cur, ' ', spaces);
264 enc->cur += spaces;
265 }
266}
267
268inline void
269encode_space (enc_t *enc)
270{
271 need (enc, 1);
272 encode_ch (enc, ' ');
273}
274
275inline void
276encode_nl (enc_t *enc)
277{
278 if (enc->json.flags & F_INDENT)
279 {
280 need (enc, 1);
148 encode_ch (enc, ' '); \ 281 encode_ch (enc, '\n');
149 } \ 282 }
150 SE 283}
151 284
152#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 285inline void
153#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 286encode_comma (enc_t *enc)
154#define COMMA SB \ 287{
155 encode_ch (enc, ','); \ 288 encode_ch (enc, ',');
289
156 if (enc->flags & F_INDENT) \ 290 if (enc->json.flags & F_INDENT)
157 NL; \ 291 encode_nl (enc);
158 else if (enc->flags & F_SPACE_AFTER) \ 292 else if (enc->json.flags & F_SPACE_AFTER)
159 SPACE; \ 293 encode_space (enc);
160 SE 294}
161 295
162static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
163 297
164static void 298static void
165encode_av (enc_t *enc, AV *av) 299encode_av (enc_t *enc, AV *av)
166{ 300{
167 int i, len = av_len (av); 301 int i, len = av_len (av);
168 302
169 encode_ch (enc, '['); NL; 303 if (enc->indent >= enc->maxdepth)
304 croak ("data structure too deep (hit recursion limit)");
305
306 encode_ch (enc, '['); encode_nl (enc);
170 ++enc->indent; 307 ++enc->indent;
171 308
172 for (i = 0; i <= len; ++i) 309 for (i = 0; i <= len; ++i)
173 { 310 {
174 INDENT; 311 encode_indent (enc);
175 encode_sv (enc, *av_fetch (av, i, 0)); 312 encode_sv (enc, *av_fetch (av, i, 0));
176 313
177 if (i < len) 314 if (i < len)
178 COMMA; 315 encode_comma (enc);
179 } 316 }
180 317
181 NL; 318 encode_nl (enc);
182 319
183 --enc->indent; 320 --enc->indent;
184 INDENT; encode_ch (enc, ']'); 321 encode_indent (enc); encode_ch (enc, ']');
185} 322}
186 323
187static void 324static void
188encode_he (enc_t *enc, HE *he) 325encode_he (enc_t *enc, HE *he)
189{ 326{
… …
191 328
192 if (HeKLEN (he) == HEf_SVKEY) 329 if (HeKLEN (he) == HEf_SVKEY)
193 { 330 {
194 SV *sv = HeSVKEY (he); 331 SV *sv = HeSVKEY (he);
195 STRLEN len; 332 STRLEN len;
333 char *str;
334
335 SvGETMAGIC (sv);
196 char *str = SvPV (sv, len); 336 str = SvPV (sv, len);
197 337
198 encode_str (enc, str, len, SvUTF8 (sv)); 338 encode_str (enc, str, len, SvUTF8 (sv));
199 } 339 }
200 else 340 else
201 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 341 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
202 342
203 encode_ch (enc, '"'); 343 encode_ch (enc, '"');
204 344
205 if (enc->flags & F_SPACE_BEFORE) SPACE; 345 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
206 encode_ch (enc, ':'); 346 encode_ch (enc, ':');
207 if (enc->flags & F_SPACE_AFTER ) SPACE; 347 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
208 encode_sv (enc, HeVAL (he)); 348 encode_sv (enc, HeVAL (he));
209} 349}
210 350
211// compare hash entries, used when all keys are bytestrings 351// compare hash entries, used when all keys are bytestrings
212static int 352static int
… …
218 HE *b = *(HE **)b_; 358 HE *b = *(HE **)b_;
219 359
220 STRLEN la = HeKLEN (a); 360 STRLEN la = HeKLEN (a);
221 STRLEN lb = HeKLEN (b); 361 STRLEN lb = HeKLEN (b);
222 362
223 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 363 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
224 cmp = la < lb ? -1 : la == lb ? 0 : 1; 364 cmp = la - lb;
225 365
226 return cmp; 366 return cmp;
227} 367}
228 368
229// compare hash entries, used when some keys are sv's or utf-x 369// compare hash entries, used when some keys are sv's or utf-x
… …
236static void 376static void
237encode_hv (enc_t *enc, HV *hv) 377encode_hv (enc_t *enc, HV *hv)
238{ 378{
239 int count, i; 379 int count, i;
240 380
381 if (enc->indent >= enc->maxdepth)
382 croak ("data structure too deep (hit recursion limit)");
383
241 encode_ch (enc, '{'); NL; ++enc->indent; 384 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
242 385
243 if ((count = hv_iterinit (hv))) 386 if ((count = hv_iterinit (hv)))
244 { 387 {
245 // for canonical output we have to sort by keys first 388 // for canonical output we have to sort by keys first
246 // actually, this is mostly due to the stupid so-called 389 // actually, this is mostly due to the stupid so-called
247 // security workaround added somewhere in 5.8.x. 390 // security workaround added somewhere in 5.8.x.
248 // that randomises hash orderings 391 // that randomises hash orderings
249 if (enc->flags & F_CANONICAL) 392 if (enc->json.flags & F_CANONICAL)
250 { 393 {
251 HE *he, *hes [count];
252 int fast = 1; 394 int fast = 1;
395 HE *he;
396#if defined(__BORLANDC__) || defined(_MSC_VER)
397 HE **hes = _alloca (count * sizeof (HE));
398#else
399 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
400#endif
253 401
254 i = 0; 402 i = 0;
255 while ((he = hv_iternext (hv))) 403 while ((he = hv_iternext (hv)))
256 { 404 {
257 hes [i++] = he; 405 hes [i++] = he;
… …
263 411
264 if (fast) 412 if (fast)
265 qsort (hes, count, sizeof (HE *), he_cmp_fast); 413 qsort (hes, count, sizeof (HE *), he_cmp_fast);
266 else 414 else
267 { 415 {
268 // hack to disable "use bytes" 416 // hack to forcefully disable "use bytes"
269 COP *oldcop = PL_curcop, cop; 417 COP cop = *PL_curcop;
270 cop.op_private = 0; 418 cop.op_private = 0;
419
420 ENTER;
421 SAVETMPS;
422
423 SAVEVPTR (PL_curcop);
271 PL_curcop = &cop; 424 PL_curcop = &cop;
272 425
273 SAVETMPS;
274 qsort (hes, count, sizeof (HE *), he_cmp_slow); 426 qsort (hes, count, sizeof (HE *), he_cmp_slow);
427
275 FREETMPS; 428 FREETMPS;
276 429 LEAVE;
277 PL_curcop = oldcop;
278 } 430 }
279 431
280 for (i = 0; i < count; ++i) 432 for (i = 0; i < count; ++i)
281 { 433 {
282 INDENT; 434 encode_indent (enc);
283 encode_he (enc, hes [i]); 435 encode_he (enc, hes [i]);
284 436
285 if (i < count - 1) 437 if (i < count - 1)
286 COMMA; 438 encode_comma (enc);
287 }
288
289 NL; 439 }
440
441 encode_nl (enc);
290 } 442 }
291 else 443 else
292 { 444 {
293 SV *sv;
294 HE *he = hv_iternext (hv); 445 HE *he = hv_iternext (hv);
295 446
296 for (;;) 447 for (;;)
297 { 448 {
298 INDENT; 449 encode_indent (enc);
299 encode_he (enc, he); 450 encode_he (enc, he);
300 451
301 if (!(he = hv_iternext (hv))) 452 if (!(he = hv_iternext (hv)))
302 break; 453 break;
303 454
304 COMMA; 455 encode_comma (enc);
305 }
306
307 NL; 456 }
457
458 encode_nl (enc);
308 } 459 }
309 } 460 }
310 461
311 --enc->indent; INDENT; encode_ch (enc, '}'); 462 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
463}
464
465// encode objects, arrays and special \0=false and \1=true values.
466static void
467encode_rv (enc_t *enc, SV *sv)
468{
469 svtype svt;
470
471 SvGETMAGIC (sv);
472 svt = SvTYPE (sv);
473
474 if (expect_false (SvOBJECT (sv)))
475 {
476 if (SvSTASH (sv) == json_boolean_stash)
477 {
478 if (SvIV (sv))
479 encode_str (enc, "true", 4, 0);
480 else
481 encode_str (enc, "false", 5, 0);
482 }
483 else
484 {
485#if 0
486 if (0 && sv_derived_from (rv, "JSON::Literal"))
487 {
488 // not yet
489 }
490#endif
491 if (enc->json.flags & F_CONV_BLESSED)
492 {
493 // we re-bless the reference to get overload and other niceties right
494 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
495
496 if (to_json)
497 {
498 dSP; ENTER; SAVETMPS; 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; LEAVE;
510 }
511 else if (enc->json.flags & F_ALLOW_BLESSED)
512 encode_str (enc, "null", 4, 0);
513 else
514 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
515 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
516 }
517 else if (enc->json.flags & F_ALLOW_BLESSED)
518 encode_str (enc, "null", 4, 0);
519 else
520 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
522 }
523 }
524 else if (svt == SVt_PVHV)
525 encode_hv (enc, (HV *)sv);
526 else if (svt == SVt_PVAV)
527 encode_av (enc, (AV *)sv);
528 else if (svt < SVt_PVAV)
529 {
530 STRLEN len = 0;
531 char *pv = svt ? SvPV (sv, len) : 0;
532
533 if (len == 1 && *pv == '1')
534 encode_str (enc, "true", 4, 0);
535 else if (len == 1 && *pv == '0')
536 encode_str (enc, "false", 5, 0);
537 else
538 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540 }
541 else
542 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
543 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312} 544}
313 545
314static void 546static void
315encode_sv (enc_t *enc, SV *sv) 547encode_sv (enc_t *enc, SV *sv)
316{ 548{
549 SvGETMAGIC (sv);
550
317 if (SvPOKp (sv)) 551 if (SvPOKp (sv))
318 { 552 {
319 STRLEN len; 553 STRLEN len;
320 char *str = SvPV (sv, len); 554 char *str = SvPV (sv, len);
321 encode_ch (enc, '"'); 555 encode_ch (enc, '"');
322 encode_str (enc, str, len, SvUTF8 (sv)); 556 encode_str (enc, str, len, SvUTF8 (sv));
323 encode_ch (enc, '"'); 557 encode_ch (enc, '"');
324 } 558 }
325 else if (SvNOKp (sv)) 559 else if (SvNOKp (sv))
326 { 560 {
561 // trust that perl will do the right thing w.r.t. JSON syntax.
327 need (enc, NV_DIG + 32); 562 need (enc, NV_DIG + 32);
328 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 563 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
329 enc->cur += strlen (enc->cur); 564 enc->cur += strlen (enc->cur);
330 } 565 }
331 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
332 { 567 {
333 need (enc, 64); 568 // we assume we can always read an IV as a UV
569 if (SvUV (sv) & ~(UV)0x7fff)
570 {
571 // large integer, use the (rather slow) snprintf way.
572 need (enc, sizeof (UV) * 3);
334 enc->cur += 573 enc->cur +=
335 SvIsUV(sv) 574 SvIsUV(sv)
336 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 575 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
337 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 576 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
577 }
578 else
579 {
580 // optimise the "small number case"
581 // code will likely be branchless and use only a single multiplication
582 I32 i = SvIV (sv);
583 U32 u;
584 char digit, nz = 0;
585
586 need (enc, 6);
587
588 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
589 u = i < 0 ? -i : i;
590
591 // convert to 4.28 fixed-point representation
592 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
593
594 // now output digit by digit, each time masking out the integer part
595 // and multiplying by 5 while moving the decimal point one to the right,
596 // resulting in a net multiplication by 10.
597 // we always write the digit to memory but conditionally increment
598 // the pointer, to ease the usage of conditional move instructions.
599 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
600 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
601 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
602 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
603 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
604 }
338 } 605 }
339 else if (SvROK (sv)) 606 else if (SvROK (sv))
340 { 607 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)) 608 else if (!SvOK (sv))
356 encode_str (enc, "null", 4, 0); 609 encode_str (enc, "null", 4, 0);
357 else 610 else
358 croak ("encountered perl type that JSON cannot handle"); 611 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
612 SvPV_nolen (sv), SvFLAGS (sv));
359} 613}
360 614
361static SV * 615static SV *
362encode_json (SV *scalar, UV flags) 616encode_json (SV *scalar, JSON *json)
363{ 617{
364 enc_t enc; 618 enc_t enc;
365 enc.flags = flags; 619
620 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
621 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
622
623 enc.json = *json;
366 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 624 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
367 enc.cur = SvPVX (enc.sv); 625 enc.cur = SvPVX (enc.sv);
368 enc.end = SvEND (enc.sv); 626 enc.end = SvEND (enc.sv);
369 enc.max_recurse = 0;
370 enc.indent = 0; 627 enc.indent = 0;
628 enc.maxdepth = DEC_DEPTH (enc.json.flags);
371 629
372 SvPOK_only (enc.sv); 630 SvPOK_only (enc.sv);
373 encode_sv (&enc, scalar); 631 encode_sv (&enc, scalar);
374 632
633 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
634 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
635
375 if (!(flags & (F_ASCII | F_UTF8))) 636 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
376 SvUTF8_on (enc.sv); 637 SvUTF8_on (enc.sv);
377 638
378 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 639 if (enc.json.flags & F_SHRINK)
640 shrink (enc.sv);
641
379 return enc.sv; 642 return enc.sv;
380} 643}
381 644
382///////////////////////////////////////////////////////////////////////////// 645/////////////////////////////////////////////////////////////////////////////
646// decoder
383 647
384#define WS \ 648// structure used for decoding JSON
649typedef struct
650{
651 char *cur; // current parser pointer
652 char *end; // end of input string
653 const char *err; // parse error, if != 0
654 JSON json;
655 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit
657} dec_t;
658
659inline void
660decode_ws (dec_t *dec)
661{
385 for (;;) \ 662 for (;;)
386 { \ 663 {
387 char ch = *dec->cur; \ 664 char ch = *dec->cur;
665
388 if (ch > 0x20 \ 666 if (ch > 0x20
389 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 667 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
390 break; \ 668 break;
669
391 ++dec->cur; \ 670 ++dec->cur;
392 } 671 }
672}
393 673
394#define ERR(reason) SB dec->err = reason; goto fail; SE 674#define ERR(reason) SB dec->err = reason; goto fail; SE
675
395#define EXPECT_CH(ch) SB \ 676#define EXPECT_CH(ch) SB \
396 if (*dec->cur != ch) \ 677 if (*dec->cur != ch) \
397 ERR (# ch " expected"); \ 678 ERR (# ch " expected"); \
398 ++dec->cur; \ 679 ++dec->cur; \
399 SE 680 SE
400 681
682#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
683#define DEC_DEC_DEPTH --dec->depth
684
401static SV *decode_sv (dec_t *dec); 685static 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
407 686
408static signed char decode_hexdigit[256]; 687static signed char decode_hexdigit[256];
409 688
410static UV 689static UV
411decode_4hex (dec_t *dec) 690decode_4hex (dec_t *dec)
412{ 691{
413 signed char d1, d2, d3, d4; 692 signed char d1, d2, d3, d4;
693 unsigned char *cur = (unsigned char *)dec->cur;
414 694
415 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 695 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"); 696 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]]; 697 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"); 698 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");
423 699
424 dec->cur += 4; 700 dec->cur += 4;
425 701
426 return ((UV)d1) << 12 702 return ((UV)d1) << 12
427 | ((UV)d2) << 8 703 | ((UV)d2) << 8
… …
433} 709}
434 710
435static SV * 711static SV *
436decode_str (dec_t *dec) 712decode_str (dec_t *dec)
437{ 713{
438 SV *sv = NEWSV (0,2); 714 SV *sv = 0;
439 STRLEN cur = 0;
440 int utf8 = 0; 715 int utf8 = 0;
716 char *dec_cur = dec->cur;
441 717
442 for (;;) 718 do
443 { 719 {
444 unsigned char ch = *(unsigned char *)dec->cur; 720 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
721 char *cur = buf;
445 722
446 if (ch == '"') 723 do
447 break;
448 else if (ch == '\\')
449 { 724 {
450 switch (*++dec->cur) 725 unsigned char ch = *(unsigned char *)dec_cur++;
726
727 if (expect_false (ch == '"'))
728 {
729 --dec_cur;
730 break;
451 { 731 }
452 case '\\': 732 else if (expect_false (ch == '\\'))
453 case '/': 733 {
454 case '"': APPEND_CH (*dec->cur++); break; 734 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 { 735 {
464 UV lo, hi; 736 case '\\':
465 ++dec->cur; 737 case '/':
738 case '"': *cur++ = *dec_cur++; break;
466 739
467 hi = decode_4hex (dec); 740 case 'b': ++dec_cur; *cur++ = '\010'; break;
468 if (hi == (UV)-1) 741 case 't': ++dec_cur; *cur++ = '\011'; break;
469 goto fail; 742 case 'n': ++dec_cur; *cur++ = '\012'; break;
743 case 'f': ++dec_cur; *cur++ = '\014'; break;
744 case 'r': ++dec_cur; *cur++ = '\015'; break;
470 745
471 // possibly a surrogate pair 746 case 'u':
472 if (hi >= 0xd800 && hi < 0xdc00)
473 { 747 {
474 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 748 UV lo, hi;
475 ERR ("illegal surrogate character"); 749 ++dec_cur;
476 750
477 dec->cur += 2; 751 dec->cur = dec_cur;
478
479 lo = decode_4hex (dec); 752 hi = decode_4hex (dec);
753 dec_cur = dec->cur;
480 if (lo == (UV)-1) 754 if (hi == (UV)-1)
481 goto fail; 755 goto fail;
482 756
757 // possibly a surrogate pair
758 if (hi >= 0xd800)
759 if (hi < 0xdc00)
760 {
761 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
762 ERR ("missing low surrogate character in surrogate pair");
763
764 dec_cur += 2;
765
766 dec->cur = dec_cur;
767 lo = decode_4hex (dec);
768 dec_cur = dec->cur;
769 if (lo == (UV)-1)
770 goto fail;
771
483 if (lo < 0xdc00 || lo >= 0xe000) 772 if (lo < 0xdc00 || lo >= 0xe000)
484 ERR ("surrogate pair expected"); 773 ERR ("surrogate pair expected");
485 774
486 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 775 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
776 }
777 else if (hi < 0xe000)
778 ERR ("missing high surrogate character in surrogate pair");
779
780 if (hi >= 0x80)
781 {
782 utf8 = 1;
783
784 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
785 }
786 else
787 *cur++ = hi;
487 } 788 }
488 else if (lo >= 0xdc00 && lo < 0xe000)
489 ERR ("illegal surrogate character");
490
491 if (hi >= 0x80)
492 { 789 break;
493 utf8 = 1;
494 790
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 791 default:
499 APPEND_CH (hi); 792 --dec_cur;
793 ERR ("illegal backslash escape sequence in string");
500 } 794 }
795 }
796 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
797 *cur++ = ch;
798 else if (ch >= 0x80)
799 {
800 STRLEN clen;
801 UV uch;
802
803 --dec_cur;
804
805 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
806 if (clen == (STRLEN)-1)
807 ERR ("malformed UTF-8 character in JSON string");
808
809 do
810 *cur++ = *dec_cur++;
811 while (--clen);
812
813 utf8 = 1;
814 }
815 else
816 {
817 --dec_cur;
818
819 if (!ch)
820 ERR ("unexpected end of string while parsing JSON string");
501 break; 821 else
822 ERR ("invalid character encountered while parsing JSON string");
502 } 823 }
503 } 824 }
504 else if (ch >= 0x20 && ch <= 0x7f) 825 while (cur < buf + SHORT_STRING_LEN);
505 APPEND_CH (*dec->cur++); 826
506 else if (ch >= 0x80) 827 {
828 STRLEN len = cur - buf;
829
830 if (sv)
507 { 831 {
508 STRLEN clen; 832 SvGROW (sv, SvCUR (sv) + len + 1);
509 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 833 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
510 if (clen < 0) 834 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 } 835 }
517 else 836 else
518 ERR ("invalid character encountered"); 837 sv = newSVpvn (buf, len);
519 } 838 }
839 }
840 while (*dec_cur != '"');
520 841
521 ++dec->cur; 842 ++dec_cur;
522 843
844 if (sv)
845 {
523 SvPOK_only (sv); 846 SvPOK_only (sv);
524
525 SvCUR_set (sv, cur);
526 *SvEND (sv) = 0; 847 *SvEND (sv) = 0;
527 848
528 if (utf8) 849 if (utf8)
529 SvUTF8_on (sv); 850 SvUTF8_on (sv);
851 }
852 else
853 sv = newSVpvn ("", 0);
530 854
855 dec->cur = dec_cur;
531 return sv; 856 return sv;
532 857
533fail: 858fail:
534 SvREFCNT_dec (sv); 859 dec->cur = dec_cur;
535 return 0; 860 return 0;
536} 861}
537 862
538static SV * 863static SV *
539decode_num (dec_t *dec) 864decode_num (dec_t *dec)
… …
549 { 874 {
550 ++dec->cur; 875 ++dec->cur;
551 if (*dec->cur >= '0' && *dec->cur <= '9') 876 if (*dec->cur >= '0' && *dec->cur <= '9')
552 ERR ("malformed number (leading zero must not be followed by another digit)"); 877 ERR ("malformed number (leading zero must not be followed by another digit)");
553 } 878 }
554 879 else if (*dec->cur < '0' || *dec->cur > '9')
555 // int 880 ERR ("malformed number (no digits after initial minus)");
881 else
882 do
883 {
884 ++dec->cur;
885 }
556 while (*dec->cur >= '0' && *dec->cur <= '9') 886 while (*dec->cur >= '0' && *dec->cur <= '9');
557 ++dec->cur;
558 887
559 // [frac] 888 // [frac]
560 if (*dec->cur == '.') 889 if (*dec->cur == '.')
561 { 890 {
562 is_nv = 1; 891 ++dec->cur;
892
893 if (*dec->cur < '0' || *dec->cur > '9')
894 ERR ("malformed number (no digits after decimal point)");
563 895
564 do 896 do
565 { 897 {
566 ++dec->cur; 898 ++dec->cur;
567 } 899 }
568 while (*dec->cur >= '0' && *dec->cur <= '9'); 900 while (*dec->cur >= '0' && *dec->cur <= '9');
901
902 is_nv = 1;
569 } 903 }
570 904
571 // [exp] 905 // [exp]
572 if (*dec->cur == 'e' || *dec->cur == 'E') 906 if (*dec->cur == 'e' || *dec->cur == 'E')
573 { 907 {
574 is_nv = 1;
575
576 ++dec->cur; 908 ++dec->cur;
909
577 if (*dec->cur == '-' || *dec->cur == '+') 910 if (*dec->cur == '-' || *dec->cur == '+')
578 ++dec->cur; 911 ++dec->cur;
579 912
913 if (*dec->cur < '0' || *dec->cur > '9')
914 ERR ("malformed number (no digits after exp sign)");
915
916 do
917 {
918 ++dec->cur;
919 }
580 while (*dec->cur >= '0' && *dec->cur <= '9') 920 while (*dec->cur >= '0' && *dec->cur <= '9');
581 ++dec->cur; 921
922 is_nv = 1;
582 } 923 }
583 924
584 if (!is_nv) 925 if (!is_nv)
585 { 926 {
586 UV uv; 927 // 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); 928 if (*start == '-')
588 if (numtype & IS_NUMBER_IN_UV) 929 switch (dec->cur - start)
589 if (numtype & IS_NUMBER_NEG)
590 { 930 {
591 if (uv < (UV)IV_MIN) 931 case 2: return newSViv (-( start [1] - '0' * 1));
592 return newSViv (-(IV)uv); 932 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
933 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
934 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
593 } 935 }
936 else
937 switch (dec->cur - start)
938 {
939 case 1: return newSViv ( start [0] - '0' * 1);
940 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
941 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
942 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
943 }
944
945 {
946 UV uv;
947 int numtype = grok_number (start, dec->cur - start, &uv);
948 if (numtype & IS_NUMBER_IN_UV)
949 if (numtype & IS_NUMBER_NEG)
950 {
951 if (uv < (UV)IV_MIN)
952 return newSViv (-(IV)uv);
953 }
594 else 954 else
595 return newSVuv (uv); 955 return newSVuv (uv);
956
957 // here would likely be the place for bigint support
596 } 958 }
959 }
597 960
961 // if we ever support bigint or bigfloat, this is the place for bigfloat
598 return newSVnv (Atof (start)); 962 return newSVnv (Atof (start));
599 963
600fail: 964fail:
601 return 0; 965 return 0;
602} 966}
… …
604static SV * 968static SV *
605decode_av (dec_t *dec) 969decode_av (dec_t *dec)
606{ 970{
607 AV *av = newAV (); 971 AV *av = newAV ();
608 972
973 DEC_INC_DEPTH;
974 decode_ws (dec);
975
976 if (*dec->cur == ']')
977 ++dec->cur;
978 else
609 for (;;) 979 for (;;)
610 { 980 {
611 SV *value; 981 SV *value;
612 982
613 value = decode_sv (dec); 983 value = decode_sv (dec);
614 if (!value) 984 if (!value)
615 goto fail; 985 goto fail;
616 986
617 av_push (av, value); 987 av_push (av, value);
618 988
619 WS; 989 decode_ws (dec);
620 990
621 if (*dec->cur == ']') 991 if (*dec->cur == ']')
622 { 992 {
623 ++dec->cur; 993 ++dec->cur;
624 break; 994 break;
995 }
625 } 996
626
627 if (*dec->cur != ',') 997 if (*dec->cur != ',')
628 ERR (", or ] expected while parsing array"); 998 ERR (", or ] expected while parsing array");
629 999
630 ++dec->cur; 1000 ++dec->cur;
631 } 1001 }
632 1002
1003 DEC_DEC_DEPTH;
633 return newRV_noinc ((SV *)av); 1004 return newRV_noinc ((SV *)av);
634 1005
635fail: 1006fail:
636 SvREFCNT_dec (av); 1007 SvREFCNT_dec (av);
1008 DEC_DEC_DEPTH;
637 return 0; 1009 return 0;
638} 1010}
639 1011
640static SV * 1012static SV *
641decode_hv (dec_t *dec) 1013decode_hv (dec_t *dec)
642{ 1014{
1015 SV *sv;
643 HV *hv = newHV (); 1016 HV *hv = newHV ();
644 1017
1018 DEC_INC_DEPTH;
1019 decode_ws (dec);
1020
1021 if (*dec->cur == '}')
1022 ++dec->cur;
1023 else
645 for (;;) 1024 for (;;)
646 { 1025 {
647 SV *key, *value; 1026 decode_ws (dec); EXPECT_CH ('"');
648 1027
649 WS; EXPECT_CH ('"'); 1028 // heuristic: assume that
650 1029 // a) decode_str + hv_store_ent are abysmally slow.
651 key = decode_str (dec); 1030 // b) most hash keys are short, simple ascii text.
652 if (!key) 1031 // => try to "fast-match" such strings to avoid
653 goto fail; 1032 // the overhead of decode_str + hv_store_ent.
654
655 WS; EXPECT_CH (':');
656
657 value = decode_sv (dec);
658 if (!value)
659 { 1033 {
1034 SV *value;
1035 char *p = dec->cur;
1036 char *e = p + 24; // only try up to 24 bytes
1037
1038 for (;;)
1039 {
1040 // the >= 0x80 is true on most architectures
1041 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1042 {
1043 // slow path, back up and use decode_str
1044 SV *key = decode_str (dec);
1045 if (!key)
1046 goto fail;
1047
1048 decode_ws (dec); EXPECT_CH (':');
1049
1050 value = decode_sv (dec);
1051 if (!value)
1052 {
1053 SvREFCNT_dec (key);
1054 goto fail;
1055 }
1056
1057 hv_store_ent (hv, key, value, 0);
660 SvREFCNT_dec (key); 1058 SvREFCNT_dec (key);
1059
1060 break;
1061 }
1062 else if (*p == '"')
1063 {
1064 // fast path, got a simple key
1065 char *key = dec->cur;
1066 int len = p - key;
1067 dec->cur = p + 1;
1068
1069 decode_ws (dec); EXPECT_CH (':');
1070
1071 value = decode_sv (dec);
1072 if (!value)
661 goto fail; 1073 goto fail;
1074
1075 hv_store (hv, key, len, value, 0);
1076
1077 break;
1078 }
1079
1080 ++p;
1081 }
662 } 1082 }
663 1083
664 //TODO: optimise 1084 decode_ws (dec);
665 hv_store_ent (hv, key, value, 0);
666 1085
667 WS;
668
669 if (*dec->cur == '}') 1086 if (*dec->cur == '}')
1087 {
1088 ++dec->cur;
1089 break;
1090 }
1091
1092 if (*dec->cur != ',')
1093 ERR (", or } expected while parsing object/hash");
1094
1095 ++dec->cur;
1096 }
1097
1098 DEC_DEC_DEPTH;
1099 sv = newRV_noinc ((SV *)hv);
1100
1101 // check filter callbacks
1102 if (dec->json.flags & F_HOOK)
1103 {
1104 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
670 { 1105 {
671 ++dec->cur; 1106 HE *cb, *he;
672 break; 1107
1108 hv_iterinit (hv);
1109 he = hv_iternext (hv);
1110 hv_iterinit (hv);
1111
1112 // the next line creates a mortal sv each time its called.
1113 // might want to optimise this for common cases.
1114 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1115
1116 if (cb)
1117 {
1118 int count;
1119 ENTER; SAVETMPS;
1120
1121 dSP; PUSHMARK (SP);
1122 XPUSHs (HeVAL (he));
1123
1124 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1125
1126 if (count == 1)
1127 {
1128 sv = newSVsv (POPs);
1129 FREETMPS; LEAVE;
1130 return sv;
1131 }
1132
1133 FREETMPS; LEAVE;
1134 }
673 } 1135 }
674 1136
675 if (*dec->cur != ',') 1137 if (dec->json.cb_object)
676 ERR (", or } expected while parsing object/hash"); 1138 {
1139 int count;
1140 ENTER; SAVETMPS;
677 1141
678 ++dec->cur; 1142 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
679 } 1143 XPUSHs (sv_2mortal (sv));
680 1144
681 return newRV_noinc ((SV *)hv); 1145 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1146
1147 if (count == 1)
1148 {
1149 sv = newSVsv (POPs);
1150 FREETMPS; LEAVE;
1151 return sv;
1152 }
1153
1154 SvREFCNT_inc (sv);
1155 FREETMPS; LEAVE;
1156 }
1157 }
1158
1159 return sv;
682 1160
683fail: 1161fail:
684 SvREFCNT_dec (hv); 1162 SvREFCNT_dec (hv);
1163 DEC_DEC_DEPTH;
685 return 0; 1164 return 0;
686} 1165}
687 1166
688static SV * 1167static SV *
689decode_sv (dec_t *dec) 1168decode_sv (dec_t *dec)
690{ 1169{
691 WS; 1170 decode_ws (dec);
1171
1172 // the beauty of JSON: you need exactly one character lookahead
1173 // to parse anything.
692 switch (*dec->cur) 1174 switch (*dec->cur)
693 { 1175 {
694 case '"': ++dec->cur; return decode_str (dec); 1176 case '"': ++dec->cur; return decode_str (dec);
695 case '[': ++dec->cur; return decode_av (dec); 1177 case '[': ++dec->cur; return decode_av (dec);
696 case '{': ++dec->cur; return decode_hv (dec); 1178 case '{': ++dec->cur; return decode_hv (dec);
… …
702 1184
703 case 't': 1185 case 't':
704 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1186 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
705 { 1187 {
706 dec->cur += 4; 1188 dec->cur += 4;
707 return newSViv (1); 1189 return SvREFCNT_inc (json_true);
708 } 1190 }
709 else 1191 else
710 ERR ("'true' expected"); 1192 ERR ("'true' expected");
711 1193
712 break; 1194 break;
713 1195
714 case 'f': 1196 case 'f':
715 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1197 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
716 { 1198 {
717 dec->cur += 5; 1199 dec->cur += 5;
718 return newSViv (0); 1200 return SvREFCNT_inc (json_false);
719 } 1201 }
720 else 1202 else
721 ERR ("'false' expected"); 1203 ERR ("'false' expected");
722 1204
723 break; 1205 break;
724 1206
725 case 'n': 1207 case 'n':
726 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1208 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
727 { 1209 {
728 dec->cur += 4; 1210 dec->cur += 4;
729 return newSViv (1); 1211 return newSVsv (&PL_sv_undef);
730 } 1212 }
731 else 1213 else
732 ERR ("'null' expected"); 1214 ERR ("'null' expected");
733 1215
734 break; 1216 break;
735 1217
736 default: 1218 default:
737 ERR ("malformed json string"); 1219 ERR ("malformed JSON string, neither array, object, number, string or atom");
738 break; 1220 break;
739 } 1221 }
740 1222
741fail: 1223fail:
742 return 0; 1224 return 0;
743} 1225}
744 1226
745static SV * 1227static SV *
746decode_json (SV *string, UV flags) 1228decode_json (SV *string, JSON *json, UV *offset_return)
747{ 1229{
1230 dec_t dec;
1231 UV offset;
748 SV *sv; 1232 SV *sv;
749 1233
1234 SvGETMAGIC (string);
1235 SvUPGRADE (string, SVt_PV);
1236
1237 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1238 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1239 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1240
750 if (!(flags & F_UTF8)) 1241 if (json->flags & F_UTF8)
1242 sv_utf8_downgrade (string, 0);
1243 else
751 sv_utf8_upgrade (string); 1244 sv_utf8_upgrade (string);
752 1245
753 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1246 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
754 1247
755 dec_t dec; 1248 dec.json = *json;
756 dec.flags = flags;
757 dec.cur = SvPVX (string); 1249 dec.cur = SvPVX (string);
758 dec.end = SvEND (string); 1250 dec.end = SvEND (string);
759 dec.err = 0; 1251 dec.err = 0;
1252 dec.depth = 0;
1253 dec.maxdepth = DEC_DEPTH (dec.json.flags);
760 1254
761 *dec.end = 1; // invalid anywhere 1255 if (dec.json.cb_object || dec.json.cb_sk_object)
1256 dec.json.flags |= F_HOOK;
1257
1258 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
762 sv = decode_sv (&dec); 1259 sv = decode_sv (&dec);
763 *dec.end = 0; 1260
1261 if (!(offset_return || !sv))
1262 {
1263 // check for trailing garbage
1264 decode_ws (&dec);
1265
1266 if (*dec.cur)
1267 {
1268 dec.err = "garbage after JSON object";
1269 SvREFCNT_dec (sv);
1270 sv = 0;
1271 }
1272 }
1273
1274 if (offset_return || !sv)
1275 {
1276 offset = dec.json.flags & F_UTF8
1277 ? dec.cur - SvPVX (string)
1278 : utf8_distance (dec.cur, SvPVX (string));
1279
1280 if (offset_return)
1281 *offset_return = offset;
1282 }
764 1283
765 if (!sv) 1284 if (!sv)
766 { 1285 {
767 IV offset = utf8_distance (dec.cur, SvPVX (string));
768 SV *uni = sv_newmortal (); 1286 SV *uni = sv_newmortal ();
769 1287
1288 // horrible hack to silence warning inside pv_uni_display
1289 COP cop = *PL_curcop;
1290 cop.cop_warnings = pWARN_NONE;
1291 ENTER;
1292 SAVEVPTR (PL_curcop);
1293 PL_curcop = &cop;
770 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1294 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1295 LEAVE;
1296
771 croak ("%s, at character %d (%s)", 1297 croak ("%s, at character offset %d [\"%s\"]",
772 dec.err, 1298 dec.err,
773 (int)offset, 1299 (int)offset,
774 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1300 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
775 } 1301 }
776 1302
777 sv_dump (sv);//D
778 return sv_2mortal (sv); 1303 sv = sv_2mortal (sv);
1304
1305 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1306 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1307
1308 return sv;
779} 1309}
1310
1311/////////////////////////////////////////////////////////////////////////////
1312// XS interface functions
780 1313
781MODULE = JSON::XS PACKAGE = JSON::XS 1314MODULE = JSON::XS PACKAGE = JSON::XS
782 1315
783BOOT: 1316BOOT:
784{ 1317{
785 int i; 1318 int i;
786 1319
787 memset (decode_hexdigit, 0xff, 256);
788 for (i = 10; i--; ) 1320 for (i = 0; i < 256; ++i)
789 decode_hexdigit ['0' + i] = i; 1321 decode_hexdigit [i] =
1322 i >= '0' && i <= '9' ? i - '0'
1323 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1324 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1325 : -1;
790 1326
791 for (i = 6; --i; ) 1327 json_stash = gv_stashpv ("JSON::XS" , 1);
1328 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1329
1330 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1331 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1332}
1333
1334PROTOTYPES: DISABLE
1335
1336void new (char *klass)
1337 PPCODE:
1338{
1339 SV *pv = NEWSV (0, sizeof (JSON));
1340 SvPOK_only (pv);
1341 Zero (SvPVX (pv), 1, JSON);
1342 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1343 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1344}
1345
1346void ascii (JSON *self, int enable = 1)
1347 ALIAS:
1348 ascii = F_ASCII
1349 latin1 = F_LATIN1
1350 utf8 = F_UTF8
1351 indent = F_INDENT
1352 canonical = F_CANONICAL
1353 space_before = F_SPACE_BEFORE
1354 space_after = F_SPACE_AFTER
1355 pretty = F_PRETTY
1356 allow_nonref = F_ALLOW_NONREF
1357 shrink = F_SHRINK
1358 allow_blessed = F_ALLOW_BLESSED
1359 convert_blessed = F_CONV_BLESSED
1360 PPCODE:
1361{
1362 if (enable)
1363 self->flags |= ix;
1364 else
1365 self->flags &= ~ix;
1366
1367 XPUSHs (ST (0));
1368}
1369
1370void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1371 PPCODE:
1372{
1373 UV log2 = 0;
1374
1375 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1376
1377 while ((1UL << log2) < max_depth)
1378 ++log2;
1379
1380 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1381
1382 XPUSHs (ST (0));
1383}
1384
1385void max_size (JSON *self, UV max_size = 0)
1386 PPCODE:
1387{
1388 UV log2 = 0;
1389
1390 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1391 if (max_size == 1) max_size = 2;
1392
1393 while ((1UL << log2) < max_size)
1394 ++log2;
1395
1396 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1397
1398 XPUSHs (ST (0));
1399}
1400
1401void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1402 PPCODE:
1403{
1404 SvREFCNT_dec (self->cb_object);
1405 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1406
1407 XPUSHs (ST (0));
1408}
1409
1410void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1411 PPCODE:
1412{
1413 if (!self->cb_sk_object)
1414 self->cb_sk_object = newHV ();
1415
1416 if (SvOK (cb))
1417 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1418 else
792 { 1419 {
793 decode_hexdigit ['a' + i] = 10 + i; 1420 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
794 decode_hexdigit ['A' + i] = 10 + i; 1421
1422 if (!HvKEYS (self->cb_sk_object))
1423 {
1424 SvREFCNT_dec (self->cb_sk_object);
1425 self->cb_sk_object = 0;
1426 }
795 } 1427 }
796 1428
797 json_stash = gv_stashpv ("JSON::XS", 1); 1429 XPUSHs (ST (0));
798} 1430}
799 1431
800SV *new (char *dummy) 1432void encode (JSON *self, SV *scalar)
1433 PPCODE:
1434 XPUSHs (encode_json (scalar, self));
1435
1436void decode (JSON *self, SV *jsonstr)
1437 PPCODE:
1438 XPUSHs (decode_json (jsonstr, self, 0));
1439
1440void decode_prefix (JSON *self, SV *jsonstr)
1441 PPCODE:
1442{
1443 UV offset;
1444 EXTEND (SP, 2);
1445 PUSHs (decode_json (jsonstr, self, &offset));
1446 PUSHs (sv_2mortal (newSVuv (offset)));
1447}
1448
1449void DESTROY (JSON *self)
801 CODE: 1450 CODE:
802 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1451 SvREFCNT_dec (self->cb_sk_object);
803 OUTPUT: 1452 SvREFCNT_dec (self->cb_object);
804 RETVAL
805 1453
806SV *ascii (SV *self, int enable) 1454PROTOTYPES: ENABLE
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}
826 OUTPUT:
827 RETVAL
828
829void encode (SV *self, SV *scalar)
830 PPCODE:
831 XPUSHs (encode_json (scalar, *SvJSON (self)));
832
833void decode (SV *self, SV *jsonstr)
834 PPCODE:
835 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
836 1455
837void to_json (SV *scalar) 1456void to_json (SV *scalar)
838 PPCODE: 1457 PPCODE:
1458{
1459 JSON json = { F_DEFAULT | F_UTF8 };
839 XPUSHs (encode_json (scalar, F_UTF8)); 1460 XPUSHs (encode_json (scalar, &json));
1461}
840 1462
841void from_json (SV *jsonstr) 1463void from_json (SV *jsonstr)
842 PPCODE: 1464 PPCODE:
1465{
1466 JSON json = { F_DEFAULT | F_UTF8 };
843 XPUSHs (decode_json (jsonstr, F_UTF8)); 1467 XPUSHs (decode_json (jsonstr, &json, 0));
1468}
844 1469

Diff Legend

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