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 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs.
Revision 1.61 by root, Sun Aug 26 21:56:47 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
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>
9#include <float.h>
8 10
11#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h
13#endif
14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
9#define F_ASCII 0x00000001 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 23#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 24#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 25#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 26#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 27#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 28#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
32#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37
38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 40
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 42#define F_DEFAULT (9UL << S_MAXDEPTH)
20 43
21#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
45#define INDENT_STEP 3 // spaces per indentation level
46
47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 48
23#define SB do { 49#define SB do {
24#define SE } while (0) 50#define SE } while (0)
25 51
26static HV *json_stash; 52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value))
54# define inline inline
55#else
56# define expect(expr,value) (expr)
57# define inline static
58#endif
59
60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#ifdef USE_ITHREADS
64# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else
67# define JSON_SLOW 0
68# define JSON_STASH json_stash
69#endif
70
71static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false;
73
74typedef struct {
75 U32 flags;
76 SV *cb_object;
77 HV *cb_sk_object;
78} JSON;
79
80/////////////////////////////////////////////////////////////////////////////
81// utility functions
82
83inline void
84shrink (SV *sv)
85{
86 sv_utf8_downgrade (sv, 1);
87 if (SvLEN (sv) > SvCUR (sv) + 1)
88 {
89#ifdef SvPV_shrink_to_cur
90 SvPV_shrink_to_cur (sv);
91#elif defined (SvPV_renew)
92 SvPV_renew (sv, SvCUR (sv) + 1);
93#endif
94 }
95}
96
97// decode an utf-8 character and return it, or (UV)-1 in
98// case of an error.
99// we special-case "safe" characters from U+80 .. U+7FF,
100// but use the very good perl function to parse anything else.
101// note that we never call this function for a ascii codepoints
102inline UV
103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
104{
105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
108 {
109 *clen = 2;
110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
111 }
112 else
113 {
114 *clen = (STRLEN)-1;
115 return (UV)-1;
116 }
117}
118
119/////////////////////////////////////////////////////////////////////////////
120// encoder
27 121
28// structure used for encoding JSON 122// structure used for encoding JSON
29typedef struct 123typedef struct
30{ 124{
31 char *cur; 125 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 126 char *end; // SvEND (sv)
34 SV *sv; 127 SV *sv; // result scalar
35 UV flags; 128 JSON json;
36 int max_recurse; 129 U32 indent; // indentation level
37 int indent; 130 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 131} enc_t;
39 132
40// structure used for decoding JSON 133inline void
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58/////////////////////////////////////////////////////////////////////////////
59
60static void
61need (enc_t *enc, STRLEN len) 134need (enc_t *enc, STRLEN len)
62{ 135{
63 if (enc->cur + len >= enc->end) 136 if (expect_false (enc->cur + len >= enc->end))
64 { 137 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 138 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 139 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 140 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 142 }
70} 143}
71 144
72static void 145inline void
73encode_ch (enc_t *enc, char ch) 146encode_ch (enc_t *enc, char ch)
74{ 147{
75 need (enc, 1); 148 need (enc, 1);
76 *enc->cur++ = ch; 149 *enc->cur++ = ch;
77} 150}
… …
85 158
86 while (str < end) 159 while (str < end)
87 { 160 {
88 unsigned char ch = *(unsigned char *)str; 161 unsigned char ch = *(unsigned char *)str;
89 162
90 if (ch == '"') 163 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 164 {
165 if (expect_false (ch == '"')) // but with slow exceptions
166 {
92 need (enc, len += 1); 167 need (enc, len += 1);
93 *enc->cur++ = '\\'; 168 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 169 *enc->cur++ = '"';
95 ++str;
96 } 170 }
97 else if (ch == '\\') 171 else if (expect_false (ch == '\\'))
98 { 172 {
99 need (enc, len += 1); 173 need (enc, len += 1);
100 *enc->cur++ = '\\'; 174 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 175 *enc->cur++ = '\\';
102 ++str;
103 } 176 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 177 else
105 {
106 *enc->cur++ = ch; 178 *enc->cur++ = ch;
107 ++str; 179
108 }
109 else if (ch == '\015')
110 {
111 need (enc, len += 1);
112 *enc->cur++ = '\\';
113 *enc->cur++ = 'r';
114 ++str;
115 }
116 else if (ch == '\012')
117 {
118 need (enc, len += 1);
119 *enc->cur++ = '\\';
120 *enc->cur++ = 'n';
121 ++str; 180 ++str;
122 } 181 }
123 else 182 else
124 { 183 {
125 STRLEN clen; 184 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
131 if (clen == (STRLEN)-1)
132 croak ("malformed UTF-8 character in string, cannot convert to JSON");
133 } 185 {
134 else 186 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
135 { 187 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
136 uch = ch; 188 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
137 clen = 1; 189 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
138 } 190 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
139 191
140 if (uch < 0x80 || enc->flags & F_ASCII) 192 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 193 {
194 STRLEN clen;
195 UV uch;
196
197 if (is_utf8)
198 {
199 uch = decode_utf8 (str, end - str, &clen);
200 if (clen == (STRLEN)-1)
201 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
202 }
203 else
204 {
205 uch = ch;
206 clen = 1;
207 }
208
209 if (uch > 0x10FFFFUL)
210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
211
212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
213 {
214 if (uch > 0xFFFFUL)
215 {
144 need (enc, len += 11); 216 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 217 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 218 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 219 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 220 enc->cur += 12;
221 }
222 else
223 {
224 static char hexdigit [16] = "0123456789abcdef";
225 need (enc, len += 5);
226 *enc->cur++ = '\\';
227 *enc->cur++ = 'u';
228 *enc->cur++ = hexdigit [ uch >> 12 ];
229 *enc->cur++ = hexdigit [(uch >> 8) & 15];
230 *enc->cur++ = hexdigit [(uch >> 4) & 15];
231 *enc->cur++ = hexdigit [(uch >> 0) & 15];
232 }
233
234 str += clen;
235 }
236 else if (enc->json.flags & F_LATIN1)
237 {
238 *enc->cur++ = uch;
239 str += clen;
240 }
241 else if (is_utf8)
242 {
243 need (enc, len += clen);
244 do
245 {
246 *enc->cur++ = *str++;
247 }
248 while (--clen);
249 }
250 else
251 {
252 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
254 ++str;
255 }
149 } 256 }
150 else
151 {
152 static char hexdigit [16] = "0123456789abcdef";
153 need (enc, len += 5);
154 *enc->cur++ = '\\';
155 *enc->cur++ = 'u';
156 *enc->cur++ = hexdigit [ uch >> 12 ];
157 *enc->cur++ = hexdigit [(uch >> 8) & 15];
158 *enc->cur++ = hexdigit [(uch >> 4) & 15];
159 *enc->cur++ = hexdigit [(uch >> 0) & 15];
160 }
161
162 str += clen;
163 }
164 else if (is_utf8)
165 {
166 need (enc, len += clen);
167 do
168 {
169 *enc->cur++ = *str++;
170 }
171 while (--clen);
172 }
173 else
174 {
175 need (enc, 10); // never more than 11 bytes needed
176 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
177 ++str;
178 } 257 }
179 } 258 }
180 259
181 --len; 260 --len;
182 } 261 }
183} 262}
184 263
185#define INDENT SB \ 264inline void
265encode_indent (enc_t *enc)
266{
186 if (enc->flags & F_INDENT) \ 267 if (enc->json.flags & F_INDENT)
187 { \ 268 {
188 int i_; \ 269 int spaces = enc->indent * INDENT_STEP;
189 need (enc, enc->indent); \ 270
190 for (i_ = enc->indent * 3; i_--; )\ 271 need (enc, spaces);
272 memset (enc->cur, ' ', spaces);
273 enc->cur += spaces;
274 }
275}
276
277inline void
278encode_space (enc_t *enc)
279{
280 need (enc, 1);
281 encode_ch (enc, ' ');
282}
283
284inline void
285encode_nl (enc_t *enc)
286{
287 if (enc->json.flags & F_INDENT)
288 {
289 need (enc, 1);
191 encode_ch (enc, ' '); \ 290 encode_ch (enc, '\n');
192 } \ 291 }
193 SE 292}
194 293
195#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 294inline void
196#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 295encode_comma (enc_t *enc)
197#define COMMA SB \ 296{
198 encode_ch (enc, ','); \ 297 encode_ch (enc, ',');
298
199 if (enc->flags & F_INDENT) \ 299 if (enc->json.flags & F_INDENT)
200 NL; \ 300 encode_nl (enc);
201 else if (enc->flags & F_SPACE_AFTER) \ 301 else if (enc->json.flags & F_SPACE_AFTER)
202 SPACE; \ 302 encode_space (enc);
203 SE 303}
204 304
205static void encode_sv (enc_t *enc, SV *sv); 305static void encode_sv (enc_t *enc, SV *sv);
206 306
207static void 307static void
208encode_av (enc_t *enc, AV *av) 308encode_av (enc_t *enc, AV *av)
209{ 309{
210 int i, len = av_len (av); 310 int i, len = av_len (av);
211 311
212 encode_ch (enc, '['); NL; 312 if (enc->indent >= enc->maxdepth)
313 croak ("data structure too deep (hit recursion limit)");
314
315 encode_ch (enc, '['); encode_nl (enc);
213 ++enc->indent; 316 ++enc->indent;
214 317
215 for (i = 0; i <= len; ++i) 318 for (i = 0; i <= len; ++i)
216 { 319 {
217 INDENT; 320 SV **svp = av_fetch (av, i, 0);
218 encode_sv (enc, *av_fetch (av, i, 0)); 321
322 encode_indent (enc);
323
324 if (svp)
325 encode_sv (enc, *svp);
326 else
327 encode_str (enc, "null", 4, 0);
219 328
220 if (i < len) 329 if (i < len)
221 COMMA; 330 encode_comma (enc);
222 } 331 }
223 332
224 NL; 333 encode_nl (enc);
225 334
226 --enc->indent; 335 --enc->indent;
227 INDENT; encode_ch (enc, ']'); 336 encode_indent (enc); encode_ch (enc, ']');
228} 337}
229 338
230static void 339static void
231encode_he (enc_t *enc, HE *he) 340encode_hk (enc_t *enc, HE *he)
232{ 341{
233 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
234 343
235 if (HeKLEN (he) == HEf_SVKEY) 344 if (HeKLEN (he) == HEf_SVKEY)
236 { 345 {
… …
246 else 355 else
247 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 356 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
248 357
249 encode_ch (enc, '"'); 358 encode_ch (enc, '"');
250 359
251 if (enc->flags & F_SPACE_BEFORE) SPACE; 360 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
252 encode_ch (enc, ':'); 361 encode_ch (enc, ':');
253 if (enc->flags & F_SPACE_AFTER ) SPACE; 362 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
254 encode_sv (enc, HeVAL (he));
255} 363}
256 364
257// compare hash entries, used when all keys are bytestrings 365// compare hash entries, used when all keys are bytestrings
258static int 366static int
259he_cmp_fast (const void *a_, const void *b_) 367he_cmp_fast (const void *a_, const void *b_)
… …
264 HE *b = *(HE **)b_; 372 HE *b = *(HE **)b_;
265 373
266 STRLEN la = HeKLEN (a); 374 STRLEN la = HeKLEN (a);
267 STRLEN lb = HeKLEN (b); 375 STRLEN lb = HeKLEN (b);
268 376
269 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 377 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
270 cmp = la < lb ? -1 : la == lb ? 0 : 1; 378 cmp = lb - la;
271 379
272 return cmp; 380 return cmp;
273} 381}
274 382
275// compare hash entries, used when some keys are sv's or utf-x 383// compare hash entries, used when some keys are sv's or utf-x
276static int 384static int
277he_cmp_slow (const void *a, const void *b) 385he_cmp_slow (const void *a, const void *b)
278{ 386{
279 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 387 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
280} 388}
281 389
282static void 390static void
283encode_hv (enc_t *enc, HV *hv) 391encode_hv (enc_t *enc, HV *hv)
284{ 392{
393 HE *he;
285 int count, i; 394 int count;
286 395
396 if (enc->indent >= enc->maxdepth)
397 croak ("data structure too deep (hit recursion limit)");
398
287 encode_ch (enc, '{'); NL; ++enc->indent; 399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
288 400
289 if ((count = hv_iterinit (hv)))
290 {
291 // for canonical output we have to sort by keys first 401 // for canonical output we have to sort by keys first
292 // actually, this is mostly due to the stupid so-called 402 // actually, this is mostly due to the stupid so-called
293 // security workaround added somewhere in 5.8.x. 403 // security workaround added somewhere in 5.8.x.
294 // that randomises hash orderings 404 // that randomises hash orderings
295 if (enc->flags & F_CANONICAL) 405 if (enc->json.flags & F_CANONICAL)
406 {
407 int count = hv_iterinit (hv);
408
409 if (SvMAGICAL (hv))
296 { 410 {
297 HE *he, *hes [count]; 411 // need to count by iterating. could improve by dynamically building the vector below
412 // but I don't care for the speed of this special case.
413 // note also that we will run into undefined behaviour when the two iterations
414 // do not result in the same count, something I might care for in some later release.
415
416 count = 0;
417 while (hv_iternext (hv))
418 ++count;
419
420 hv_iterinit (hv);
421 }
422
423 if (count)
424 {
298 int fast = 1; 425 int i, fast = 1;
426#if defined(__BORLANDC__) || defined(_MSC_VER)
427 HE **hes = _alloca (count * sizeof (HE));
428#else
429 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
430#endif
299 431
300 i = 0; 432 i = 0;
301 while ((he = hv_iternext (hv))) 433 while ((he = hv_iternext (hv)))
302 { 434 {
303 hes [i++] = he; 435 hes [i++] = he;
… …
309 441
310 if (fast) 442 if (fast)
311 qsort (hes, count, sizeof (HE *), he_cmp_fast); 443 qsort (hes, count, sizeof (HE *), he_cmp_fast);
312 else 444 else
313 { 445 {
314 // hack to disable "use bytes" 446 // hack to forcefully disable "use bytes"
315 COP *oldcop = PL_curcop, cop; 447 COP cop = *PL_curcop;
316 cop.op_private = 0; 448 cop.op_private = 0;
449
450 ENTER;
451 SAVETMPS;
452
453 SAVEVPTR (PL_curcop);
317 PL_curcop = &cop; 454 PL_curcop = &cop;
318 455
319 SAVETMPS;
320 qsort (hes, count, sizeof (HE *), he_cmp_slow); 456 qsort (hes, count, sizeof (HE *), he_cmp_slow);
457
321 FREETMPS; 458 FREETMPS;
322 459 LEAVE;
323 PL_curcop = oldcop;
324 }
325
326 for (i = 0; i < count; ++i)
327 { 460 }
328 INDENT;
329 encode_he (enc, hes [i]);
330 461
331 if (i < count - 1) 462 while (count--)
332 COMMA;
333 } 463 {
464 encode_indent (enc);
465 he = hes [count];
466 encode_hk (enc, he);
467 encode_sv (enc, SvMAGICAL (hv) ? hv_iterval (hv, he) : HeVAL (he));
334 468
469 if (count)
470 encode_comma (enc);
335 NL; 471 }
472 }
473 }
474 else
475 {
476 if (hv_iterinit (hv) || SvMAGICAL (hv))
477 if ((he = hv_iternext (hv)))
478 for (;;)
479 {
480 encode_indent (enc);
481 encode_hk (enc, he);
482 encode_sv (enc, SvMAGICAL (hv) ? hv_iterval (hv, he) : HeVAL (he));
483
484 if (!(he = hv_iternext (hv)))
485 break;
486
487 encode_comma (enc);
488 }
489 }
490
491 encode_nl (enc);
492
493 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
494}
495
496// encode objects, arrays and special \0=false and \1=true values.
497static void
498encode_rv (enc_t *enc, SV *sv)
499{
500 svtype svt;
501
502 SvGETMAGIC (sv);
503 svt = SvTYPE (sv);
504
505 if (expect_false (SvOBJECT (sv)))
506 {
507 HV *stash = !JSON_SLOW || json_boolean_stash
508 ? json_boolean_stash
509 : gv_stashpv ("JSON::XS::Boolean", 1);
510
511 if (SvSTASH (sv) == stash)
512 {
513 if (SvIV (sv))
514 encode_str (enc, "true", 4, 0);
515 else
516 encode_str (enc, "false", 5, 0);
336 } 517 }
337 else 518 else
338 { 519 {
339 SV *sv; 520#if 0
340 HE *he = hv_iternext (hv); 521 if (0 && sv_derived_from (rv, "JSON::Literal"))
341
342 for (;;)
343 {
344 INDENT;
345 encode_he (enc, he);
346
347 if (!(he = hv_iternext (hv)))
348 break;
349
350 COMMA;
351 } 522 {
352 523 // not yet
353 NL; 524 }
525#endif
526 if (enc->json.flags & F_CONV_BLESSED)
527 {
528 // we re-bless the reference to get overload and other niceties right
529 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
530
531 if (to_json)
532 {
533 dSP;
534
535 ENTER; SAVETMPS; PUSHMARK (SP);
536 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
537
538 // calling with G_SCALAR ensures that we always get a 1 return value
539 PUTBACK;
540 call_sv ((SV *)GvCV (to_json), G_SCALAR);
541 SPAGAIN;
542
543 // catch this surprisingly common error
544 if (SvROK (TOPs) && SvRV (TOPs) == sv)
545 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
546
547 sv = POPs;
548 PUTBACK;
549
550 encode_sv (enc, sv);
551
552 FREETMPS; LEAVE;
553 }
554 else if (enc->json.flags & F_ALLOW_BLESSED)
555 encode_str (enc, "null", 4, 0);
556 else
557 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
558 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
559 }
560 else if (enc->json.flags & F_ALLOW_BLESSED)
561 encode_str (enc, "null", 4, 0);
562 else
563 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
564 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
354 } 565 }
355 } 566 }
567 else if (svt == SVt_PVHV)
568 encode_hv (enc, (HV *)sv);
569 else if (svt == SVt_PVAV)
570 encode_av (enc, (AV *)sv);
571 else if (svt < SVt_PVAV)
572 {
573 STRLEN len = 0;
574 char *pv = svt ? SvPV (sv, len) : 0;
356 575
357 --enc->indent; INDENT; encode_ch (enc, '}'); 576 if (len == 1 && *pv == '1')
577 encode_str (enc, "true", 4, 0);
578 else if (len == 1 && *pv == '0')
579 encode_str (enc, "false", 5, 0);
580 else
581 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
582 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
583 }
584 else
585 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
586 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
358} 587}
359 588
360static void 589static void
361encode_sv (enc_t *enc, SV *sv) 590encode_sv (enc_t *enc, SV *sv)
362{ 591{
… …
370 encode_str (enc, str, len, SvUTF8 (sv)); 599 encode_str (enc, str, len, SvUTF8 (sv));
371 encode_ch (enc, '"'); 600 encode_ch (enc, '"');
372 } 601 }
373 else if (SvNOKp (sv)) 602 else if (SvNOKp (sv))
374 { 603 {
604 // trust that perl will do the right thing w.r.t. JSON syntax.
375 need (enc, NV_DIG + 32); 605 need (enc, NV_DIG + 32);
376 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 606 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
377 enc->cur += strlen (enc->cur); 607 enc->cur += strlen (enc->cur);
378 } 608 }
379 else if (SvIOKp (sv)) 609 else if (SvIOKp (sv))
380 { 610 {
381 need (enc, 64); 611 // we assume we can always read an IV as a UV
612 if (SvUV (sv) & ~(UV)0x7fff)
613 {
614 // large integer, use the (rather slow) snprintf way.
615 need (enc, sizeof (UV) * 3);
382 enc->cur += 616 enc->cur +=
383 SvIsUV(sv) 617 SvIsUV(sv)
384 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 618 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
385 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 619 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
620 }
621 else
622 {
623 // optimise the "small number case"
624 // code will likely be branchless and use only a single multiplication
625 I32 i = SvIV (sv);
626 U32 u;
627 char digit, nz = 0;
628
629 need (enc, 6);
630
631 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
632 u = i < 0 ? -i : i;
633
634 // convert to 4.28 fixed-point representation
635 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
636
637 // now output digit by digit, each time masking out the integer part
638 // and multiplying by 5 while moving the decimal point one to the right,
639 // resulting in a net multiplication by 10.
640 // we always write the digit to memory but conditionally increment
641 // the pointer, to ease the usage of conditional move instructions.
642 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
643 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
644 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
645 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
646 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
647 }
386 } 648 }
387 else if (SvROK (sv)) 649 else if (SvROK (sv))
388 { 650 encode_rv (enc, SvRV (sv));
389 if (!--enc->max_recurse)
390 croak ("data structure too deep (hit recursion limit)");
391
392 sv = SvRV (sv);
393
394 switch (SvTYPE (sv))
395 {
396 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
397 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
398
399 default:
400 croak ("JSON can only represent references to arrays or hashes");
401 }
402 }
403 else if (!SvOK (sv)) 651 else if (!SvOK (sv))
404 encode_str (enc, "null", 4, 0); 652 encode_str (enc, "null", 4, 0);
405 else 653 else
406 croak ("encountered perl type that JSON cannot handle"); 654 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
655 SvPV_nolen (sv), SvFLAGS (sv));
407} 656}
408 657
409static SV * 658static SV *
410encode_json (SV *scalar, UV flags) 659encode_json (SV *scalar, JSON *json)
411{ 660{
412 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
413 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");
414
415 enc_t enc; 661 enc_t enc;
416 enc.flags = flags; 662
663 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
664 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
665
666 enc.json = *json;
417 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 667 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
418 enc.cur = SvPVX (enc.sv); 668 enc.cur = SvPVX (enc.sv);
419 enc.end = SvEND (enc.sv); 669 enc.end = SvEND (enc.sv);
420 enc.max_recurse = 0;
421 enc.indent = 0; 670 enc.indent = 0;
671 enc.maxdepth = DEC_DEPTH (enc.json.flags);
422 672
423 SvPOK_only (enc.sv); 673 SvPOK_only (enc.sv);
424 encode_sv (&enc, scalar); 674 encode_sv (&enc, scalar);
425 675
676 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
677 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
678
426 if (!(flags & (F_ASCII | F_UTF8))) 679 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
427 SvUTF8_on (enc.sv); 680 SvUTF8_on (enc.sv);
428 681
429 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 682 if (enc.json.flags & F_SHRINK)
683 shrink (enc.sv);
684
430 return enc.sv; 685 return enc.sv;
431} 686}
432 687
433///////////////////////////////////////////////////////////////////////////// 688/////////////////////////////////////////////////////////////////////////////
689// decoder
434 690
435#define WS \ 691// structure used for decoding JSON
692typedef struct
693{
694 char *cur; // current parser pointer
695 char *end; // end of input string
696 const char *err; // parse error, if != 0
697 JSON json;
698 U32 depth; // recursion depth
699 U32 maxdepth; // recursion depth limit
700} dec_t;
701
702inline void
703decode_ws (dec_t *dec)
704{
436 for (;;) \ 705 for (;;)
437 { \ 706 {
438 char ch = *dec->cur; \ 707 char ch = *dec->cur;
708
439 if (ch > 0x20 \ 709 if (ch > 0x20
440 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 710 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
441 break; \ 711 break;
712
442 ++dec->cur; \ 713 ++dec->cur;
443 } 714 }
715}
444 716
445#define ERR(reason) SB dec->err = reason; goto fail; SE 717#define ERR(reason) SB dec->err = reason; goto fail; SE
718
446#define EXPECT_CH(ch) SB \ 719#define EXPECT_CH(ch) SB \
447 if (*dec->cur != ch) \ 720 if (*dec->cur != ch) \
448 ERR (# ch " expected"); \ 721 ERR (# ch " expected"); \
449 ++dec->cur; \ 722 ++dec->cur; \
450 SE 723 SE
451 724
725#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
726#define DEC_DEC_DEPTH --dec->depth
727
452static SV *decode_sv (dec_t *dec); 728static SV *decode_sv (dec_t *dec);
453 729
454static signed char decode_hexdigit[256]; 730static signed char decode_hexdigit[256];
455 731
456static UV 732static UV
457decode_4hex (dec_t *dec) 733decode_4hex (dec_t *dec)
458{ 734{
459 signed char d1, d2, d3, d4; 735 signed char d1, d2, d3, d4;
736 unsigned char *cur = (unsigned char *)dec->cur;
460 737
461 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 738 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
462 if (d1 < 0) ERR ("four hexadecimal digits expected"); 739 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
463 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 740 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
464 if (d2 < 0) ERR ("four hexadecimal digits expected"); 741 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
465 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
466 if (d3 < 0) ERR ("four hexadecimal digits expected");
467 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
468 if (d4 < 0) ERR ("four hexadecimal digits expected");
469 742
470 dec->cur += 4; 743 dec->cur += 4;
471 744
472 return ((UV)d1) << 12 745 return ((UV)d1) << 12
473 | ((UV)d2) << 8 746 | ((UV)d2) << 8
… …
476 749
477fail: 750fail:
478 return (UV)-1; 751 return (UV)-1;
479} 752}
480 753
481#define APPEND_GROW(n) SB \
482 if (cur + (n) >= end) \
483 { \
484 STRLEN ofs = cur - SvPVX (sv); \
485 SvGROW (sv, ofs + (n) + 1); \
486 cur = SvPVX (sv) + ofs; \
487 end = SvEND (sv); \
488 } \
489 SE
490
491#define APPEND_CH(ch) SB \
492 APPEND_GROW (1); \
493 *cur++ = (ch); \
494 SE
495
496static SV * 754static SV *
497decode_str (dec_t *dec) 755decode_str (dec_t *dec)
498{ 756{
499 SV *sv = NEWSV (0,2); 757 SV *sv = 0;
500 int utf8 = 0; 758 int utf8 = 0;
501 char *cur = SvPVX (sv); 759 char *dec_cur = dec->cur;
502 char *end = SvEND (sv);
503 760
504 for (;;) 761 do
505 { 762 {
506 unsigned char ch = *(unsigned char *)dec->cur; 763 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
764 char *cur = buf;
507 765
508 if (ch == '"') 766 do
509 break;
510 else if (ch == '\\')
511 { 767 {
512 switch (*++dec->cur) 768 unsigned char ch = *(unsigned char *)dec_cur++;
769
770 if (expect_false (ch == '"'))
771 {
772 --dec_cur;
773 break;
513 { 774 }
514 case '\\': 775 else if (expect_false (ch == '\\'))
515 case '/': 776 {
516 case '"': APPEND_CH (*dec->cur++); break; 777 switch (*dec_cur)
517
518 case 'b': APPEND_CH ('\010'); ++dec->cur; break;
519 case 't': APPEND_CH ('\011'); ++dec->cur; break;
520 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
521 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
522 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
523
524 case 'u':
525 { 778 {
526 UV lo, hi; 779 case '\\':
527 ++dec->cur; 780 case '/':
781 case '"': *cur++ = *dec_cur++; break;
528 782
529 hi = decode_4hex (dec); 783 case 'b': ++dec_cur; *cur++ = '\010'; break;
530 if (hi == (UV)-1) 784 case 't': ++dec_cur; *cur++ = '\011'; break;
531 goto fail; 785 case 'n': ++dec_cur; *cur++ = '\012'; break;
786 case 'f': ++dec_cur; *cur++ = '\014'; break;
787 case 'r': ++dec_cur; *cur++ = '\015'; break;
532 788
533 // possibly a surrogate pair 789 case 'u':
534 if (hi >= 0xd800 && hi < 0xdc00)
535 { 790 {
536 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 791 UV lo, hi;
537 ERR ("missing low surrogate character in surrogate pair"); 792 ++dec_cur;
538 793
539 dec->cur += 2; 794 dec->cur = dec_cur;
540
541 lo = decode_4hex (dec); 795 hi = decode_4hex (dec);
796 dec_cur = dec->cur;
542 if (lo == (UV)-1) 797 if (hi == (UV)-1)
543 goto fail; 798 goto fail;
544 799
800 // possibly a surrogate pair
801 if (hi >= 0xd800)
802 if (hi < 0xdc00)
803 {
804 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
805 ERR ("missing low surrogate character in surrogate pair");
806
807 dec_cur += 2;
808
809 dec->cur = dec_cur;
810 lo = decode_4hex (dec);
811 dec_cur = dec->cur;
812 if (lo == (UV)-1)
813 goto fail;
814
545 if (lo < 0xdc00 || lo >= 0xe000) 815 if (lo < 0xdc00 || lo >= 0xe000)
546 ERR ("surrogate pair expected"); 816 ERR ("surrogate pair expected");
547 817
548 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 818 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
819 }
820 else if (hi < 0xe000)
821 ERR ("missing high surrogate character in surrogate pair");
822
823 if (hi >= 0x80)
824 {
825 utf8 = 1;
826
827 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
828 }
829 else
830 *cur++ = hi;
549 } 831 }
550 else if (hi >= 0xdc00 && hi < 0xe000)
551 ERR ("missing high surrogate character in surrogate pair");
552
553 if (hi >= 0x80)
554 { 832 break;
555 utf8 = 1;
556 833
557 APPEND_GROW (4); // at most 4 bytes for 21 bits
558 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
559 }
560 else 834 default:
561 APPEND_CH (hi); 835 --dec_cur;
836 ERR ("illegal backslash escape sequence in string");
562 } 837 }
563 break; 838 }
839 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
840 *cur++ = ch;
841 else if (ch >= 0x80)
842 {
843 STRLEN clen;
844 UV uch;
564 845
565 default:
566 --dec->cur; 846 --dec_cur;
567 ERR ("illegal backslash escape sequence in string"); 847
848 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
849 if (clen == (STRLEN)-1)
850 ERR ("malformed UTF-8 character in JSON string");
851
852 do
853 *cur++ = *dec_cur++;
854 while (--clen);
855
856 utf8 = 1;
857 }
858 else
859 {
860 --dec_cur;
861
862 if (!ch)
863 ERR ("unexpected end of string while parsing JSON string");
864 else
865 ERR ("invalid character encountered while parsing JSON string");
568 } 866 }
569 } 867 }
570 else if (ch >= 0x20 && ch <= 0x7f) 868 while (cur < buf + SHORT_STRING_LEN);
571 APPEND_CH (*dec->cur++); 869
572 else if (ch >= 0x80)
573 { 870 {
574 STRLEN clen; 871 STRLEN len = cur - buf;
575 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
576 if (clen == (STRLEN)-1)
577 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
578 872
579 APPEND_GROW (clen); 873 if (sv)
580 do
581 { 874 {
582 *cur++ = *dec->cur++; 875 SvGROW (sv, SvCUR (sv) + len + 1);
876 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
877 SvCUR_set (sv, SvCUR (sv) + len);
583 } 878 }
584 while (--clen);
585
586 utf8 = 1;
587 }
588 else if (dec->cur == dec->end)
589 ERR ("unexpected end of string while parsing json string");
590 else 879 else
591 ERR ("invalid character encountered"); 880 sv = newSVpvn (buf, len);
592 } 881 }
882 }
883 while (*dec_cur != '"');
593 884
594 ++dec->cur; 885 ++dec_cur;
595 886
596 SvCUR_set (sv, cur - SvPVX (sv)); 887 if (sv)
597 888 {
598 SvPOK_only (sv); 889 SvPOK_only (sv);
599 *SvEND (sv) = 0; 890 *SvEND (sv) = 0;
600 891
601 if (utf8) 892 if (utf8)
602 SvUTF8_on (sv); 893 SvUTF8_on (sv);
894 }
895 else
896 sv = newSVpvn ("", 0);
603 897
898 dec->cur = dec_cur;
604 return sv; 899 return sv;
605 900
606fail: 901fail:
607 SvREFCNT_dec (sv); 902 dec->cur = dec_cur;
608 return 0; 903 return 0;
609} 904}
610 905
611static SV * 906static SV *
612decode_num (dec_t *dec) 907decode_num (dec_t *dec)
… …
670 is_nv = 1; 965 is_nv = 1;
671 } 966 }
672 967
673 if (!is_nv) 968 if (!is_nv)
674 { 969 {
675 UV uv; 970 int len = dec->cur - start;
676 int numtype = grok_number (start, dec->cur - start, &uv); 971
677 if (numtype & IS_NUMBER_IN_UV) 972 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
678 if (numtype & IS_NUMBER_NEG) 973 if (*start == '-')
974 switch (len)
679 { 975 {
680 if (uv < (UV)IV_MIN) 976 case 2: return newSViv (-( start [1] - '0' * 1));
681 return newSViv (-(IV)uv); 977 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
978 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
979 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
682 } 980 }
981 else
982 switch (len)
983 {
984 case 1: return newSViv ( start [0] - '0' * 1);
985 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
986 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
987 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
988 }
989
990 {
991 UV uv;
992 int numtype = grok_number (start, len, &uv);
993 if (numtype & IS_NUMBER_IN_UV)
994 if (numtype & IS_NUMBER_NEG)
995 {
996 if (uv < (UV)IV_MIN)
997 return newSViv (-(IV)uv);
998 }
683 else 999 else
684 return newSVuv (uv); 1000 return newSVuv (uv);
685 } 1001 }
686 1002
1003 len -= *start == '-' ? 1 : 0;
1004
1005 // does not fit into IV or UV, try NV
1006 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1007 #if defined (LDBL_DIG)
1008 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1009 #endif
1010 )
1011 // fits into NV without loss of precision
1012 return newSVnv (Atof (start));
1013
1014 // everything else fails, convert it to a string
1015 return newSVpvn (start, dec->cur - start);
1016 }
1017
1018 // loss of precision here
687 return newSVnv (Atof (start)); 1019 return newSVnv (Atof (start));
688 1020
689fail: 1021fail:
690 return 0; 1022 return 0;
691} 1023}
… …
693static SV * 1025static SV *
694decode_av (dec_t *dec) 1026decode_av (dec_t *dec)
695{ 1027{
696 AV *av = newAV (); 1028 AV *av = newAV ();
697 1029
698 WS; 1030 DEC_INC_DEPTH;
1031 decode_ws (dec);
1032
699 if (*dec->cur == ']') 1033 if (*dec->cur == ']')
700 ++dec->cur; 1034 ++dec->cur;
701 else 1035 else
702 for (;;) 1036 for (;;)
703 { 1037 {
… …
707 if (!value) 1041 if (!value)
708 goto fail; 1042 goto fail;
709 1043
710 av_push (av, value); 1044 av_push (av, value);
711 1045
712 WS; 1046 decode_ws (dec);
713 1047
714 if (*dec->cur == ']') 1048 if (*dec->cur == ']')
715 { 1049 {
716 ++dec->cur; 1050 ++dec->cur;
717 break; 1051 break;
… …
721 ERR (", or ] expected while parsing array"); 1055 ERR (", or ] expected while parsing array");
722 1056
723 ++dec->cur; 1057 ++dec->cur;
724 } 1058 }
725 1059
1060 DEC_DEC_DEPTH;
726 return newRV_noinc ((SV *)av); 1061 return newRV_noinc ((SV *)av);
727 1062
728fail: 1063fail:
729 SvREFCNT_dec (av); 1064 SvREFCNT_dec (av);
1065 DEC_DEC_DEPTH;
730 return 0; 1066 return 0;
731} 1067}
732 1068
733static SV * 1069static SV *
734decode_hv (dec_t *dec) 1070decode_hv (dec_t *dec)
735{ 1071{
1072 SV *sv;
736 HV *hv = newHV (); 1073 HV *hv = newHV ();
737 1074
738 WS; 1075 DEC_INC_DEPTH;
1076 decode_ws (dec);
1077
739 if (*dec->cur == '}') 1078 if (*dec->cur == '}')
740 ++dec->cur; 1079 ++dec->cur;
741 else 1080 else
742 for (;;) 1081 for (;;)
743 { 1082 {
1083 decode_ws (dec); EXPECT_CH ('"');
1084
1085 // heuristic: assume that
1086 // a) decode_str + hv_store_ent are abysmally slow.
1087 // b) most hash keys are short, simple ascii text.
1088 // => try to "fast-match" such strings to avoid
1089 // the overhead of decode_str + hv_store_ent.
1090 {
744 SV *key, *value; 1091 SV *value;
1092 char *p = dec->cur;
1093 char *e = p + 24; // only try up to 24 bytes
745 1094
746 WS; EXPECT_CH ('"'); 1095 for (;;)
747
748 key = decode_str (dec);
749 if (!key)
750 goto fail;
751
752 WS; EXPECT_CH (':');
753
754 value = decode_sv (dec);
755 if (!value)
756 { 1096 {
1097 // the >= 0x80 is true on most architectures
1098 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1099 {
1100 // slow path, back up and use decode_str
1101 SV *key = decode_str (dec);
1102 if (!key)
1103 goto fail;
1104
1105 decode_ws (dec); EXPECT_CH (':');
1106
1107 value = decode_sv (dec);
1108 if (!value)
1109 {
1110 SvREFCNT_dec (key);
1111 goto fail;
1112 }
1113
1114 hv_store_ent (hv, key, value, 0);
757 SvREFCNT_dec (key); 1115 SvREFCNT_dec (key);
1116
1117 break;
1118 }
1119 else if (*p == '"')
1120 {
1121 // fast path, got a simple key
1122 char *key = dec->cur;
1123 int len = p - key;
1124 dec->cur = p + 1;
1125
1126 decode_ws (dec); EXPECT_CH (':');
1127
1128 value = decode_sv (dec);
1129 if (!value)
758 goto fail; 1130 goto fail;
1131
1132 hv_store (hv, key, len, value, 0);
1133
1134 break;
1135 }
1136
1137 ++p;
759 } 1138 }
760
761 //TODO: optimise
762 hv_store_ent (hv, key, value, 0);
763
764 WS; 1139 }
1140
1141 decode_ws (dec);
765 1142
766 if (*dec->cur == '}') 1143 if (*dec->cur == '}')
767 { 1144 {
768 ++dec->cur; 1145 ++dec->cur;
769 break; 1146 break;
… …
773 ERR (", or } expected while parsing object/hash"); 1150 ERR (", or } expected while parsing object/hash");
774 1151
775 ++dec->cur; 1152 ++dec->cur;
776 } 1153 }
777 1154
1155 DEC_DEC_DEPTH;
778 return newRV_noinc ((SV *)hv); 1156 sv = newRV_noinc ((SV *)hv);
1157
1158 // check filter callbacks
1159 if (dec->json.flags & F_HOOK)
1160 {
1161 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1162 {
1163 HE *cb, *he;
1164
1165 hv_iterinit (hv);
1166 he = hv_iternext (hv);
1167 hv_iterinit (hv);
1168
1169 // the next line creates a mortal sv each time its called.
1170 // might want to optimise this for common cases.
1171 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1172
1173 if (cb)
1174 {
1175 dSP;
1176 int count;
1177
1178 ENTER; SAVETMPS; PUSHMARK (SP);
1179 XPUSHs (HeVAL (he));
1180
1181 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1182
1183 if (count == 1)
1184 {
1185 sv = newSVsv (POPs);
1186 FREETMPS; LEAVE;
1187 return sv;
1188 }
1189
1190 FREETMPS; LEAVE;
1191 }
1192 }
1193
1194 if (dec->json.cb_object)
1195 {
1196 dSP;
1197 int count;
1198
1199 ENTER; SAVETMPS; PUSHMARK (SP);
1200 XPUSHs (sv_2mortal (sv));
1201
1202 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1203
1204 if (count == 1)
1205 {
1206 sv = newSVsv (POPs);
1207 FREETMPS; LEAVE;
1208 return sv;
1209 }
1210
1211 SvREFCNT_inc (sv);
1212 FREETMPS; LEAVE;
1213 }
1214 }
1215
1216 return sv;
779 1217
780fail: 1218fail:
781 SvREFCNT_dec (hv); 1219 SvREFCNT_dec (hv);
1220 DEC_DEC_DEPTH;
782 return 0; 1221 return 0;
783} 1222}
784 1223
785static SV * 1224static SV *
786decode_sv (dec_t *dec) 1225decode_sv (dec_t *dec)
787{ 1226{
788 WS; 1227 decode_ws (dec);
1228
1229 // the beauty of JSON: you need exactly one character lookahead
1230 // to parse anything.
789 switch (*dec->cur) 1231 switch (*dec->cur)
790 { 1232 {
791 case '"': ++dec->cur; return decode_str (dec); 1233 case '"': ++dec->cur; return decode_str (dec);
792 case '[': ++dec->cur; return decode_av (dec); 1234 case '[': ++dec->cur; return decode_av (dec);
793 case '{': ++dec->cur; return decode_hv (dec); 1235 case '{': ++dec->cur; return decode_hv (dec);
… …
799 1241
800 case 't': 1242 case 't':
801 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1243 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
802 { 1244 {
803 dec->cur += 4; 1245 dec->cur += 4;
804 return newSViv (1); 1246#if JSON_SLOW
1247 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1248#endif
1249 return SvREFCNT_inc (json_true);
805 } 1250 }
806 else 1251 else
807 ERR ("'true' expected"); 1252 ERR ("'true' expected");
808 1253
809 break; 1254 break;
810 1255
811 case 'f': 1256 case 'f':
812 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1257 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
813 { 1258 {
814 dec->cur += 5; 1259 dec->cur += 5;
815 return newSViv (0); 1260#if JSON_SLOW
1261 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1262#endif
1263 return SvREFCNT_inc (json_false);
816 } 1264 }
817 else 1265 else
818 ERR ("'false' expected"); 1266 ERR ("'false' expected");
819 1267
820 break; 1268 break;
… …
829 ERR ("'null' expected"); 1277 ERR ("'null' expected");
830 1278
831 break; 1279 break;
832 1280
833 default: 1281 default:
834 ERR ("malformed json string"); 1282 ERR ("malformed JSON string, neither array, object, number, string or atom");
835 break; 1283 break;
836 } 1284 }
837 1285
838fail: 1286fail:
839 return 0; 1287 return 0;
840} 1288}
841 1289
842static SV * 1290static SV *
843decode_json (SV *string, UV flags) 1291decode_json (SV *string, JSON *json, UV *offset_return)
844{ 1292{
1293 dec_t dec;
1294 UV offset;
845 SV *sv; 1295 SV *sv;
846 1296
1297 SvGETMAGIC (string);
1298 SvUPGRADE (string, SVt_PV);
1299
1300 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1301 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1302 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1303
847 if (flags & F_UTF8) 1304 if (json->flags & F_UTF8)
848 sv_utf8_downgrade (string, 0); 1305 sv_utf8_downgrade (string, 0);
849 else 1306 else
850 sv_utf8_upgrade (string); 1307 sv_utf8_upgrade (string);
851 1308
852 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1309 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
853 1310
854 dec_t dec; 1311 dec.json = *json;
855 dec.flags = flags;
856 dec.cur = SvPVX (string); 1312 dec.cur = SvPVX (string);
857 dec.end = SvEND (string); 1313 dec.end = SvEND (string);
858 dec.err = 0; 1314 dec.err = 0;
1315 dec.depth = 0;
1316 dec.maxdepth = DEC_DEPTH (dec.json.flags);
859 1317
1318 if (dec.json.cb_object || dec.json.cb_sk_object)
1319 dec.json.flags |= F_HOOK;
1320
1321 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
860 sv = decode_sv (&dec); 1322 sv = decode_sv (&dec);
861 1323
1324 if (!(offset_return || !sv))
1325 {
1326 // check for trailing garbage
1327 decode_ws (&dec);
1328
1329 if (*dec.cur)
1330 {
1331 dec.err = "garbage after JSON object";
1332 SvREFCNT_dec (sv);
1333 sv = 0;
1334 }
1335 }
1336
1337 if (offset_return || !sv)
1338 {
1339 offset = dec.json.flags & F_UTF8
1340 ? dec.cur - SvPVX (string)
1341 : utf8_distance (dec.cur, SvPVX (string));
1342
1343 if (offset_return)
1344 *offset_return = offset;
1345 }
1346
862 if (!sv) 1347 if (!sv)
863 { 1348 {
864 IV offset = utf8_distance (dec.cur, SvPVX (string));
865 SV *uni = sv_newmortal (); 1349 SV *uni = sv_newmortal ();
1350
866 // horrible hack to silence warning inside pv_uni_display 1351 // horrible hack to silence warning inside pv_uni_display
867 COP cop; 1352 COP cop = *PL_curcop;
868 memset (&cop, 0, sizeof (cop));
869 cop.cop_warnings = pWARN_NONE; 1353 cop.cop_warnings = pWARN_NONE;
1354 ENTER;
870 SAVEVPTR (PL_curcop); 1355 SAVEVPTR (PL_curcop);
871 PL_curcop = &cop; 1356 PL_curcop = &cop;
872
873 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1357 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1358 LEAVE;
1359
874 croak ("%s, at character offset %d (%s)", 1360 croak ("%s, at character offset %d [\"%s\"]",
875 dec.err, 1361 dec.err,
876 (int)offset, 1362 (int)offset,
877 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1363 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
878 } 1364 }
879 1365
880 sv = sv_2mortal (sv); 1366 sv = sv_2mortal (sv);
881 1367
882 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1368 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
883 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1369 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
884 1370
885 return sv; 1371 return sv;
886} 1372}
887 1373
1374/////////////////////////////////////////////////////////////////////////////
1375// XS interface functions
1376
888MODULE = JSON::XS PACKAGE = JSON::XS 1377MODULE = JSON::XS PACKAGE = JSON::XS
889 1378
890BOOT: 1379BOOT:
891{ 1380{
892 int i; 1381 int i;
893 1382
894 memset (decode_hexdigit, 0xff, 256);
895 for (i = 10; i--; ) 1383 for (i = 0; i < 256; ++i)
896 decode_hexdigit ['0' + i] = i; 1384 decode_hexdigit [i] =
1385 i >= '0' && i <= '9' ? i - '0'
1386 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1387 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1388 : -1;
897 1389
898 for (i = 7; i--; ) 1390 json_stash = gv_stashpv ("JSON::XS" , 1);
1391 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1392
1393 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1394 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1395}
1396
1397PROTOTYPES: DISABLE
1398
1399void CLONE (...)
1400 CODE:
1401 json_stash = 0;
1402 json_boolean_stash = 0;
1403
1404void new (char *klass)
1405 PPCODE:
1406{
1407 SV *pv = NEWSV (0, sizeof (JSON));
1408 SvPOK_only (pv);
1409 Zero (SvPVX (pv), 1, JSON);
1410 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1411 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1412}
1413
1414void ascii (JSON *self, int enable = 1)
1415 ALIAS:
1416 ascii = F_ASCII
1417 latin1 = F_LATIN1
1418 utf8 = F_UTF8
1419 indent = F_INDENT
1420 canonical = F_CANONICAL
1421 space_before = F_SPACE_BEFORE
1422 space_after = F_SPACE_AFTER
1423 pretty = F_PRETTY
1424 allow_nonref = F_ALLOW_NONREF
1425 shrink = F_SHRINK
1426 allow_blessed = F_ALLOW_BLESSED
1427 convert_blessed = F_CONV_BLESSED
1428 PPCODE:
1429{
1430 if (enable)
1431 self->flags |= ix;
1432 else
1433 self->flags &= ~ix;
1434
1435 XPUSHs (ST (0));
1436}
1437
1438void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1439 PPCODE:
1440{
1441 UV log2 = 0;
1442
1443 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1444
1445 while ((1UL << log2) < max_depth)
1446 ++log2;
1447
1448 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1449
1450 XPUSHs (ST (0));
1451}
1452
1453void max_size (JSON *self, UV max_size = 0)
1454 PPCODE:
1455{
1456 UV log2 = 0;
1457
1458 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1459 if (max_size == 1) max_size = 2;
1460
1461 while ((1UL << log2) < max_size)
1462 ++log2;
1463
1464 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1465
1466 XPUSHs (ST (0));
1467}
1468
1469void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1470 PPCODE:
1471{
1472 SvREFCNT_dec (self->cb_object);
1473 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1474
1475 XPUSHs (ST (0));
1476}
1477
1478void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1479 PPCODE:
1480{
1481 if (!self->cb_sk_object)
1482 self->cb_sk_object = newHV ();
1483
1484 if (SvOK (cb))
1485 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1486 else
899 { 1487 {
900 decode_hexdigit ['a' + i] = 10 + i; 1488 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
901 decode_hexdigit ['A' + i] = 10 + i; 1489
1490 if (!HvKEYS (self->cb_sk_object))
1491 {
1492 SvREFCNT_dec (self->cb_sk_object);
1493 self->cb_sk_object = 0;
1494 }
902 } 1495 }
903 1496
904 json_stash = gv_stashpv ("JSON::XS", 1); 1497 XPUSHs (ST (0));
905} 1498}
906 1499
907PROTOTYPES: DISABLE 1500void encode (JSON *self, SV *scalar)
1501 PPCODE:
1502 XPUSHs (encode_json (scalar, self));
908 1503
909SV *new (char *dummy) 1504void decode (JSON *self, SV *jsonstr)
1505 PPCODE:
1506 XPUSHs (decode_json (jsonstr, self, 0));
1507
1508void decode_prefix (JSON *self, SV *jsonstr)
1509 PPCODE:
1510{
1511 UV offset;
1512 EXTEND (SP, 2);
1513 PUSHs (decode_json (jsonstr, self, &offset));
1514 PUSHs (sv_2mortal (newSVuv (offset)));
1515}
1516
1517void DESTROY (JSON *self)
910 CODE: 1518 CODE:
911 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1519 SvREFCNT_dec (self->cb_sk_object);
912 OUTPUT: 1520 SvREFCNT_dec (self->cb_object);
913 RETVAL
914
915SV *ascii (SV *self, int enable)
916 ALIAS:
917 ascii = F_ASCII
918 utf8 = F_UTF8
919 indent = F_INDENT
920 canonical = F_CANONICAL
921 space_before = F_SPACE_BEFORE
922 space_after = F_SPACE_AFTER
923 json_rpc = F_JSON_RPC
924 pretty = F_PRETTY
925 allow_nonref = F_ALLOW_NONREF
926 CODE:
927{
928 UV *uv = SvJSON (self);
929 if (enable)
930 *uv |= ix;
931 else
932 *uv &= ~ix;
933
934 RETVAL = newSVsv (self);
935}
936 OUTPUT:
937 RETVAL
938
939void encode (SV *self, SV *scalar)
940 PPCODE:
941 XPUSHs (encode_json (scalar, *SvJSON (self)));
942
943void decode (SV *self, SV *jsonstr)
944 PPCODE:
945 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
946 1521
947PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
948 1523
949void to_json (SV *scalar) 1524void to_json (SV *scalar)
950 PPCODE: 1525 PPCODE:
1526{
1527 JSON json = { F_DEFAULT | F_UTF8 };
951 XPUSHs (encode_json (scalar, F_UTF8)); 1528 XPUSHs (encode_json (scalar, &json));
1529}
952 1530
953void from_json (SV *jsonstr) 1531void from_json (SV *jsonstr)
954 PPCODE: 1532 PPCODE:
1533{
1534 JSON json = { F_DEFAULT | F_UTF8 };
955 XPUSHs (decode_json (jsonstr, F_UTF8)); 1535 XPUSHs (decode_json (jsonstr, &json, 0));
1536}
956 1537

Diff Legend

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