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 1.71 by root, Wed Mar 19 03:17:38 2008 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_RELAXED 0x00001000UL
33
34#define F_MAXDEPTH 0xf8000000UL
35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39
40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 42
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 44#define F_DEFAULT (9UL << S_MAXDEPTH)
20 45
21#define INIT_SIZE 32 // initial scalar size to be allocated 46#define INIT_SIZE 32 // initial scalar size to be allocated
47#define INDENT_STEP 3 // spaces per indentation level
48
49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 50
23#define SB do { 51#define SB do {
24#define SE } while (0) 52#define SE } while (0)
25 53
26static HV *json_stash; 54#if __GNUC__ >= 3
55# define expect(expr,value) __builtin_expect ((expr), (value))
56# define INLINE static inline
57#else
58# define expect(expr,value) (expr)
59# define INLINE static
60#endif
61
62#define expect_false(expr) expect ((expr) != 0, 0)
63#define expect_true(expr) expect ((expr) != 0, 1)
64
65#ifdef USE_ITHREADS
66# define JSON_SLOW 1
67# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
68#else
69# define JSON_SLOW 0
70# define JSON_STASH json_stash
71#endif
72
73static HV *json_stash, *json_boolean_stash; // JSON::XS::
74static SV *json_true, *json_false;
75
76typedef struct {
77 U32 flags;
78 SV *cb_object;
79 HV *cb_sk_object;
80} JSON;
81
82/////////////////////////////////////////////////////////////////////////////
83// utility functions
84
85INLINE void
86shrink (SV *sv)
87{
88 sv_utf8_downgrade (sv, 1);
89 if (SvLEN (sv) > SvCUR (sv) + 1)
90 {
91#ifdef SvPV_shrink_to_cur
92 SvPV_shrink_to_cur (sv);
93#elif defined (SvPV_renew)
94 SvPV_renew (sv, SvCUR (sv) + 1);
95#endif
96 }
97}
98
99// decode an utf-8 character and return it, or (UV)-1 in
100// case of an error.
101// we special-case "safe" characters from U+80 .. U+7FF,
102// but use the very good perl function to parse anything else.
103// note that we never call this function for a ascii codepoints
104INLINE UV
105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
106{
107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
110 {
111 *clen = 2;
112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
113 }
114 else
115 {
116 *clen = (STRLEN)-1;
117 return (UV)-1;
118 }
119}
120
121/////////////////////////////////////////////////////////////////////////////
122// encoder
27 123
28// structure used for encoding JSON 124// structure used for encoding JSON
29typedef struct 125typedef struct
30{ 126{
31 char *cur; 127 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 128 char *end; // SvEND (sv)
34 SV *sv; 129 SV *sv; // result scalar
35 UV flags; 130 JSON json;
36 int max_recurse; 131 U32 indent; // indentation level
37 int indent; 132 U32 maxdepth; // max. indentation/recursion level
133 UV limit; // escape character values >= this value when encoding
38} enc_t; 134} enc_t;
39 135
40// structure used for decoding JSON 136INLINE 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) 137need (enc_t *enc, STRLEN len)
62{ 138{
63 if (enc->cur + len >= enc->end) 139 if (expect_false (enc->cur + len >= enc->end))
64 { 140 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 141 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 142 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 143 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 144 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 145 }
70} 146}
71 147
72static void 148INLINE void
73encode_ch (enc_t *enc, char ch) 149encode_ch (enc_t *enc, char ch)
74{ 150{
75 need (enc, 1); 151 need (enc, 1);
76 *enc->cur++ = ch; 152 *enc->cur++ = ch;
77} 153}
85 161
86 while (str < end) 162 while (str < end)
87 { 163 {
88 unsigned char ch = *(unsigned char *)str; 164 unsigned char ch = *(unsigned char *)str;
89 165
90 if (ch == '"') 166 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 167 {
168 if (expect_false (ch == '"')) // but with slow exceptions
169 {
92 need (enc, len += 1); 170 need (enc, len += 1);
93 *enc->cur++ = '\\'; 171 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 172 *enc->cur++ = '"';
95 ++str;
96 } 173 }
97 else if (ch == '\\') 174 else if (expect_false (ch == '\\'))
98 { 175 {
99 need (enc, len += 1); 176 need (enc, len += 1);
100 *enc->cur++ = '\\'; 177 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 178 *enc->cur++ = '\\';
102 ++str;
103 } 179 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 180 else
105 {
106 *enc->cur++ = ch; 181 *enc->cur++ = ch;
107 ++str; 182
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; 183 ++str;
122 } 184 }
123 else 185 else
124 { 186 {
125 STRLEN clen; 187 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 { 188 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 189 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
131 if (clen == (STRLEN)-1) 190 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
132 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 191 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
133 } 192 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
134 else 193 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
135 {
136 uch = ch;
137 clen = 1;
138 }
139 194
140 if (uch < 0x80 || enc->flags & F_ASCII) 195 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 196 {
197 STRLEN clen;
198 UV uch;
199
200 if (is_utf8)
201 {
202 uch = decode_utf8 (str, end - str, &clen);
203 if (clen == (STRLEN)-1)
204 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
205 }
206 else
207 {
208 uch = ch;
209 clen = 1;
210 }
211
212 if (uch < 0x20 || uch >= enc->limit)
213 {
214 if (uch > 0xFFFFUL)
215 {
216 if (uch > 0x10FFFFUL)
217 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
218
144 need (enc, len += 11); 219 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 220 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 221 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 222 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 223 enc->cur += 12;
224 }
225 else
226 {
227 static char hexdigit [16] = "0123456789abcdef";
228 need (enc, len += 5);
229 *enc->cur++ = '\\';
230 *enc->cur++ = 'u';
231 *enc->cur++ = hexdigit [ uch >> 12 ];
232 *enc->cur++ = hexdigit [(uch >> 8) & 15];
233 *enc->cur++ = hexdigit [(uch >> 4) & 15];
234 *enc->cur++ = hexdigit [(uch >> 0) & 15];
235 }
236
237 str += clen;
238 }
239 else if (enc->json.flags & F_LATIN1)
240 {
241 *enc->cur++ = uch;
242 str += clen;
243 }
244 else if (is_utf8)
245 {
246 need (enc, len += clen);
247 do
248 {
249 *enc->cur++ = *str++;
250 }
251 while (--clen);
252 }
253 else
254 {
255 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
256 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
257 ++str;
258 }
149 } 259 }
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 } 260 }
179 } 261 }
180 262
181 --len; 263 --len;
182 } 264 }
183} 265}
184 266
185#define INDENT SB \ 267INLINE void
268encode_indent (enc_t *enc)
269{
186 if (enc->flags & F_INDENT) \ 270 if (enc->json.flags & F_INDENT)
187 { \ 271 {
188 int i_; \ 272 int spaces = enc->indent * INDENT_STEP;
189 need (enc, enc->indent); \ 273
190 for (i_ = enc->indent * 3; i_--; )\ 274 need (enc, spaces);
275 memset (enc->cur, ' ', spaces);
276 enc->cur += spaces;
277 }
278}
279
280INLINE void
281encode_space (enc_t *enc)
282{
283 need (enc, 1);
284 encode_ch (enc, ' ');
285}
286
287INLINE void
288encode_nl (enc_t *enc)
289{
290 if (enc->json.flags & F_INDENT)
291 {
292 need (enc, 1);
191 encode_ch (enc, ' '); \ 293 encode_ch (enc, '\n');
192 } \ 294 }
193 SE 295}
194 296
195#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 297INLINE void
196#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 298encode_comma (enc_t *enc)
197#define COMMA SB \ 299{
198 encode_ch (enc, ','); \ 300 encode_ch (enc, ',');
301
199 if (enc->flags & F_INDENT) \ 302 if (enc->json.flags & F_INDENT)
200 NL; \ 303 encode_nl (enc);
201 else if (enc->flags & F_SPACE_AFTER) \ 304 else if (enc->json.flags & F_SPACE_AFTER)
202 SPACE; \ 305 encode_space (enc);
203 SE 306}
204 307
205static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
206 309
207static void 310static void
208encode_av (enc_t *enc, AV *av) 311encode_av (enc_t *enc, AV *av)
209{ 312{
210 int i, len = av_len (av); 313 int i, len = av_len (av);
211 314
315 if (enc->indent >= enc->maxdepth)
316 croak ("data structure too deep (hit recursion limit)");
317
212 encode_ch (enc, '['); NL; 318 encode_ch (enc, '[');
213 ++enc->indent; 319
320 if (len >= 0)
321 {
322 encode_nl (enc); ++enc->indent;
214 323
215 for (i = 0; i <= len; ++i) 324 for (i = 0; i <= len; ++i)
216 { 325 {
217 INDENT; 326 SV **svp = av_fetch (av, i, 0);
218 encode_sv (enc, *av_fetch (av, i, 0));
219 327
328 encode_indent (enc);
329
330 if (svp)
331 encode_sv (enc, *svp);
332 else
333 encode_str (enc, "null", 4, 0);
334
220 if (i < len) 335 if (i < len)
221 COMMA; 336 encode_comma (enc);
222 } 337 }
223 338
224 NL; 339 encode_nl (enc); --enc->indent; encode_indent (enc);
225 340 }
226 --enc->indent; 341
227 INDENT; encode_ch (enc, ']'); 342 encode_ch (enc, ']');
228} 343}
229 344
230static void 345static void
231encode_he (enc_t *enc, HE *he) 346encode_hk (enc_t *enc, HE *he)
232{ 347{
233 encode_ch (enc, '"'); 348 encode_ch (enc, '"');
234 349
235 if (HeKLEN (he) == HEf_SVKEY) 350 if (HeKLEN (he) == HEf_SVKEY)
236 { 351 {
246 else 361 else
247 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 362 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
248 363
249 encode_ch (enc, '"'); 364 encode_ch (enc, '"');
250 365
251 if (enc->flags & F_SPACE_BEFORE) SPACE; 366 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
252 encode_ch (enc, ':'); 367 encode_ch (enc, ':');
253 if (enc->flags & F_SPACE_AFTER ) SPACE; 368 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
254 encode_sv (enc, HeVAL (he));
255} 369}
256 370
257// compare hash entries, used when all keys are bytestrings 371// compare hash entries, used when all keys are bytestrings
258static int 372static int
259he_cmp_fast (const void *a_, const void *b_) 373he_cmp_fast (const void *a_, const void *b_)
264 HE *b = *(HE **)b_; 378 HE *b = *(HE **)b_;
265 379
266 STRLEN la = HeKLEN (a); 380 STRLEN la = HeKLEN (a);
267 STRLEN lb = HeKLEN (b); 381 STRLEN lb = HeKLEN (b);
268 382
269 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 383 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
270 cmp = la < lb ? -1 : la == lb ? 0 : 1; 384 cmp = lb - la;
271 385
272 return cmp; 386 return cmp;
273} 387}
274 388
275// compare hash entries, used when some keys are sv's or utf-x 389// compare hash entries, used when some keys are sv's or utf-x
276static int 390static int
277he_cmp_slow (const void *a, const void *b) 391he_cmp_slow (const void *a, const void *b)
278{ 392{
279 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 393 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
280} 394}
281 395
282static void 396static void
283encode_hv (enc_t *enc, HV *hv) 397encode_hv (enc_t *enc, HV *hv)
284{ 398{
399 HE *he;
285 int count, i; 400 int count;
286 401
287 encode_ch (enc, '{'); NL; ++enc->indent; 402 if (enc->indent >= enc->maxdepth)
403 croak ("data structure too deep (hit recursion limit)");
288 404
289 if ((count = hv_iterinit (hv))) 405 encode_ch (enc, '{');
290 { 406
291 // for canonical output we have to sort by keys first 407 // for canonical output we have to sort by keys first
292 // actually, this is mostly due to the stupid so-called 408 // actually, this is mostly due to the stupid so-called
293 // security workaround added somewhere in 5.8.x. 409 // security workaround added somewhere in 5.8.x.
294 // that randomises hash orderings 410 // that randomises hash orderings
295 if (enc->flags & F_CANONICAL) 411 if (enc->json.flags & F_CANONICAL)
412 {
413 int count = hv_iterinit (hv);
414
415 if (SvMAGICAL (hv))
296 { 416 {
297 HE *he, *hes [count]; 417 // need to count by iterating. could improve by dynamically building the vector below
418 // but I don't care for the speed of this special case.
419 // note also that we will run into undefined behaviour when the two iterations
420 // do not result in the same count, something I might care for in some later release.
421
422 count = 0;
423 while (hv_iternext (hv))
424 ++count;
425
426 hv_iterinit (hv);
427 }
428
429 if (count)
430 {
298 int fast = 1; 431 int i, fast = 1;
432#if defined(__BORLANDC__) || defined(_MSC_VER)
433 HE **hes = _alloca (count * sizeof (HE));
434#else
435 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
436#endif
299 437
300 i = 0; 438 i = 0;
301 while ((he = hv_iternext (hv))) 439 while ((he = hv_iternext (hv)))
302 { 440 {
303 hes [i++] = he; 441 hes [i++] = he;
309 447
310 if (fast) 448 if (fast)
311 qsort (hes, count, sizeof (HE *), he_cmp_fast); 449 qsort (hes, count, sizeof (HE *), he_cmp_fast);
312 else 450 else
313 { 451 {
314 // hack to disable "use bytes" 452 // hack to forcefully disable "use bytes"
315 COP *oldcop = PL_curcop, cop; 453 COP cop = *PL_curcop;
316 cop.op_private = 0; 454 cop.op_private = 0;
455
456 ENTER;
457 SAVETMPS;
458
459 SAVEVPTR (PL_curcop);
317 PL_curcop = &cop; 460 PL_curcop = &cop;
318 461
319 SAVETMPS;
320 qsort (hes, count, sizeof (HE *), he_cmp_slow); 462 qsort (hes, count, sizeof (HE *), he_cmp_slow);
463
321 FREETMPS; 464 FREETMPS;
322 465 LEAVE;
323 PL_curcop = oldcop;
324 } 466 }
325 467
326 for (i = 0; i < count; ++i) 468 encode_nl (enc); ++enc->indent;
469
470 while (count--)
327 { 471 {
328 INDENT; 472 encode_indent (enc);
473 he = hes [count];
329 encode_he (enc, hes [i]); 474 encode_hk (enc, he);
475 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
330 476
331 if (i < count - 1) 477 if (count)
332 COMMA; 478 encode_comma (enc);
333 } 479 }
334 480
481 encode_nl (enc); --enc->indent; encode_indent (enc);
482 }
483 }
484 else
485 {
486 if (hv_iterinit (hv) || SvMAGICAL (hv))
487 if ((he = hv_iternext (hv)))
335 NL; 488 {
489 encode_nl (enc); ++enc->indent;
490
491 for (;;)
492 {
493 encode_indent (enc);
494 encode_hk (enc, he);
495 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
496
497 if (!(he = hv_iternext (hv)))
498 break;
499
500 encode_comma (enc);
501 }
502
503 encode_nl (enc); --enc->indent; encode_indent (enc);
504 }
505 }
506
507 encode_ch (enc, '}');
508}
509
510// encode objects, arrays and special \0=false and \1=true values.
511static void
512encode_rv (enc_t *enc, SV *sv)
513{
514 svtype svt;
515
516 SvGETMAGIC (sv);
517 svt = SvTYPE (sv);
518
519 if (expect_false (SvOBJECT (sv)))
520 {
521 HV *stash = !JSON_SLOW || json_boolean_stash
522 ? json_boolean_stash
523 : gv_stashpv ("JSON::XS::Boolean", 1);
524
525 if (SvSTASH (sv) == stash)
526 {
527 if (SvIV (sv))
528 encode_str (enc, "true", 4, 0);
529 else
530 encode_str (enc, "false", 5, 0);
336 } 531 }
337 else 532 else
338 { 533 {
339 SV *sv; 534#if 0
340 HE *he = hv_iternext (hv); 535 if (0 && sv_derived_from (rv, "JSON::Literal"))
341
342 for (;;)
343 { 536 {
344 INDENT; 537 // not yet
345 encode_he (enc, he);
346
347 if (!(he = hv_iternext (hv)))
348 break;
349
350 COMMA;
351 } 538 }
352 539#endif
540 if (enc->json.flags & F_CONV_BLESSED)
353 NL; 541 {
542 // we re-bless the reference to get overload and other niceties right
543 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
544
545 if (to_json)
546 {
547 dSP;
548
549 ENTER; SAVETMPS; PUSHMARK (SP);
550 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
551
552 // calling with G_SCALAR ensures that we always get a 1 return value
553 PUTBACK;
554 call_sv ((SV *)GvCV (to_json), G_SCALAR);
555 SPAGAIN;
556
557 // catch this surprisingly common error
558 if (SvROK (TOPs) && SvRV (TOPs) == sv)
559 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
560
561 sv = POPs;
562 PUTBACK;
563
564 encode_sv (enc, sv);
565
566 FREETMPS; LEAVE;
567 }
568 else if (enc->json.flags & F_ALLOW_BLESSED)
569 encode_str (enc, "null", 4, 0);
570 else
571 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
572 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
573 }
574 else if (enc->json.flags & F_ALLOW_BLESSED)
575 encode_str (enc, "null", 4, 0);
576 else
577 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
578 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
354 } 579 }
355 } 580 }
581 else if (svt == SVt_PVHV)
582 encode_hv (enc, (HV *)sv);
583 else if (svt == SVt_PVAV)
584 encode_av (enc, (AV *)sv);
585 else if (svt < SVt_PVAV)
586 {
587 STRLEN len = 0;
588 char *pv = svt ? SvPV (sv, len) : 0;
356 589
357 --enc->indent; INDENT; encode_ch (enc, '}'); 590 if (len == 1 && *pv == '1')
591 encode_str (enc, "true", 4, 0);
592 else if (len == 1 && *pv == '0')
593 encode_str (enc, "false", 5, 0);
594 else
595 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
596 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
597 }
598 else
599 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
600 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
358} 601}
359 602
360static void 603static void
361encode_sv (enc_t *enc, SV *sv) 604encode_sv (enc_t *enc, SV *sv)
362{ 605{
370 encode_str (enc, str, len, SvUTF8 (sv)); 613 encode_str (enc, str, len, SvUTF8 (sv));
371 encode_ch (enc, '"'); 614 encode_ch (enc, '"');
372 } 615 }
373 else if (SvNOKp (sv)) 616 else if (SvNOKp (sv))
374 { 617 {
618 // trust that perl will do the right thing w.r.t. JSON syntax.
375 need (enc, NV_DIG + 32); 619 need (enc, NV_DIG + 32);
376 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 620 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
377 enc->cur += strlen (enc->cur); 621 enc->cur += strlen (enc->cur);
378 } 622 }
379 else if (SvIOKp (sv)) 623 else if (SvIOKp (sv))
380 { 624 {
381 need (enc, 64); 625 // we assume we can always read an IV as a UV
626 if (SvUV (sv) & ~(UV)0x7fff)
627 {
628 // large integer, use the (rather slow) snprintf way.
629 need (enc, sizeof (UV) * 3);
382 enc->cur += 630 enc->cur +=
383 SvIsUV(sv) 631 SvIsUV(sv)
384 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 632 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
385 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 633 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
634 }
635 else
636 {
637 // optimise the "small number case"
638 // code will likely be branchless and use only a single multiplication
639 I32 i = SvIV (sv);
640 U32 u;
641 char digit, nz = 0;
642
643 need (enc, 6);
644
645 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
646 u = i < 0 ? -i : i;
647
648 // convert to 4.28 fixed-point representation
649 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
650
651 // now output digit by digit, each time masking out the integer part
652 // and multiplying by 5 while moving the decimal point one to the right,
653 // resulting in a net multiplication by 10.
654 // we always write the digit to memory but conditionally increment
655 // the pointer, to ease the usage of conditional move instructions.
656 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
657 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
658 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
659 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
660 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
661 }
386 } 662 }
387 else if (SvROK (sv)) 663 else if (SvROK (sv))
388 { 664 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)) 665 else if (!SvOK (sv))
404 encode_str (enc, "null", 4, 0); 666 encode_str (enc, "null", 4, 0);
405 else 667 else
406 croak ("encountered perl type that JSON cannot handle"); 668 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
669 SvPV_nolen (sv), SvFLAGS (sv));
407} 670}
408 671
409static SV * 672static SV *
410encode_json (SV *scalar, UV flags) 673encode_json (SV *scalar, JSON *json)
411{ 674{
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; 675 enc_t enc;
416 enc.flags = flags; 676
677 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
678 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
679
680 enc.json = *json;
417 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 681 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
418 enc.cur = SvPVX (enc.sv); 682 enc.cur = SvPVX (enc.sv);
419 enc.end = SvEND (enc.sv); 683 enc.end = SvEND (enc.sv);
420 enc.max_recurse = 0;
421 enc.indent = 0; 684 enc.indent = 0;
685 enc.maxdepth = DEC_DEPTH (enc.json.flags);
686 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
687 : enc.json.flags & F_LATIN1 ? 0x000100UL
688 : 0x10FFFFUL;
422 689
423 SvPOK_only (enc.sv); 690 SvPOK_only (enc.sv);
424 encode_sv (&enc, scalar); 691 encode_sv (&enc, scalar);
425 692
693 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
694 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
695
426 if (!(flags & (F_ASCII | F_UTF8))) 696 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
427 SvUTF8_on (enc.sv); 697 SvUTF8_on (enc.sv);
428 698
429 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 699 if (enc.json.flags & F_SHRINK)
700 shrink (enc.sv);
701
430 return enc.sv; 702 return enc.sv;
431} 703}
432 704
433///////////////////////////////////////////////////////////////////////////// 705/////////////////////////////////////////////////////////////////////////////
706// decoder
434 707
435#define WS \ 708// structure used for decoding JSON
709typedef struct
710{
711 char *cur; // current parser pointer
712 char *end; // end of input string
713 const char *err; // parse error, if != 0
714 JSON json;
715 U32 depth; // recursion depth
716 U32 maxdepth; // recursion depth limit
717} dec_t;
718
719INLINE void
720decode_comment (dec_t *dec)
721{
722 // only '#'-style comments allowed a.t.m.
723
724 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
725 ++dec->cur;
726}
727
728INLINE void
729decode_ws (dec_t *dec)
730{
436 for (;;) \ 731 for (;;)
437 { \ 732 {
438 char ch = *dec->cur; \ 733 char ch = *dec->cur;
734
439 if (ch > 0x20 \ 735 if (ch > 0x20)
736 {
737 if (expect_false (ch == '#'))
738 {
739 if (dec->json.flags & F_RELAXED)
740 decode_comment (dec);
741 else
742 break;
743 }
744 else
745 break;
746 }
440 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 747 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
441 break; \ 748 break; // parse error, but let higher level handle it, gives better error messages
749
442 ++dec->cur; \ 750 ++dec->cur;
443 } 751 }
752}
444 753
445#define ERR(reason) SB dec->err = reason; goto fail; SE 754#define ERR(reason) SB dec->err = reason; goto fail; SE
755
446#define EXPECT_CH(ch) SB \ 756#define EXPECT_CH(ch) SB \
447 if (*dec->cur != ch) \ 757 if (*dec->cur != ch) \
448 ERR (# ch " expected"); \ 758 ERR (# ch " expected"); \
449 ++dec->cur; \ 759 ++dec->cur; \
450 SE 760 SE
451 761
762#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
763#define DEC_DEC_DEPTH --dec->depth
764
452static SV *decode_sv (dec_t *dec); 765static SV *decode_sv (dec_t *dec);
453 766
454static signed char decode_hexdigit[256]; 767static signed char decode_hexdigit[256];
455 768
456static UV 769static UV
457decode_4hex (dec_t *dec) 770decode_4hex (dec_t *dec)
458{ 771{
459 signed char d1, d2, d3, d4; 772 signed char d1, d2, d3, d4;
773 unsigned char *cur = (unsigned char *)dec->cur;
460 774
461 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 775 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"); 776 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]]; 777 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"); 778 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");
469 779
470 dec->cur += 4; 780 dec->cur += 4;
471 781
472 return ((UV)d1) << 12 782 return ((UV)d1) << 12
473 | ((UV)d2) << 8 783 | ((UV)d2) << 8
476 786
477fail: 787fail:
478 return (UV)-1; 788 return (UV)-1;
479} 789}
480 790
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 * 791static SV *
497decode_str (dec_t *dec) 792decode_str (dec_t *dec)
498{ 793{
499 SV *sv = NEWSV (0,2); 794 SV *sv = 0;
500 int utf8 = 0; 795 int utf8 = 0;
501 char *cur = SvPVX (sv); 796 char *dec_cur = dec->cur;
502 char *end = SvEND (sv);
503 797
504 for (;;) 798 do
505 { 799 {
506 unsigned char ch = *(unsigned char *)dec->cur; 800 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
801 char *cur = buf;
507 802
508 if (ch == '"') 803 do
509 break;
510 else if (ch == '\\')
511 { 804 {
512 switch (*++dec->cur) 805 unsigned char ch = *(unsigned char *)dec_cur++;
806
807 if (expect_false (ch == '"'))
513 { 808 {
514 case '\\': 809 --dec_cur;
515 case '/': 810 break;
516 case '"': APPEND_CH (*dec->cur++); break; 811 }
517 812 else if (expect_false (ch == '\\'))
518 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 813 {
519 case 't': APPEND_CH ('\011'); ++dec->cur; break; 814 switch (*dec_cur)
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 { 815 {
526 UV lo, hi; 816 case '\\':
527 ++dec->cur; 817 case '/':
818 case '"': *cur++ = *dec_cur++; break;
528 819
529 hi = decode_4hex (dec); 820 case 'b': ++dec_cur; *cur++ = '\010'; break;
530 if (hi == (UV)-1) 821 case 't': ++dec_cur; *cur++ = '\011'; break;
531 goto fail; 822 case 'n': ++dec_cur; *cur++ = '\012'; break;
823 case 'f': ++dec_cur; *cur++ = '\014'; break;
824 case 'r': ++dec_cur; *cur++ = '\015'; break;
532 825
533 // possibly a surrogate pair 826 case 'u':
534 if (hi >= 0xd800 && hi < 0xdc00)
535 { 827 {
536 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 828 UV lo, hi;
537 ERR ("missing low surrogate character in surrogate pair"); 829 ++dec_cur;
538 830
539 dec->cur += 2; 831 dec->cur = dec_cur;
540
541 lo = decode_4hex (dec); 832 hi = decode_4hex (dec);
833 dec_cur = dec->cur;
542 if (lo == (UV)-1) 834 if (hi == (UV)-1)
543 goto fail; 835 goto fail;
544 836
837 // possibly a surrogate pair
838 if (hi >= 0xd800)
839 if (hi < 0xdc00)
840 {
841 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
842 ERR ("missing low surrogate character in surrogate pair");
843
844 dec_cur += 2;
845
846 dec->cur = dec_cur;
847 lo = decode_4hex (dec);
848 dec_cur = dec->cur;
849 if (lo == (UV)-1)
850 goto fail;
851
545 if (lo < 0xdc00 || lo >= 0xe000) 852 if (lo < 0xdc00 || lo >= 0xe000)
546 ERR ("surrogate pair expected"); 853 ERR ("surrogate pair expected");
547 854
548 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 855 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
856 }
857 else if (hi < 0xe000)
858 ERR ("missing high surrogate character in surrogate pair");
859
860 if (hi >= 0x80)
861 {
862 utf8 = 1;
863
864 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
865 }
866 else
867 *cur++ = hi;
549 } 868 }
550 else if (hi >= 0xdc00 && hi < 0xe000)
551 ERR ("missing high surrogate character in surrogate pair");
552
553 if (hi >= 0x80)
554 { 869 break;
555 utf8 = 1;
556 870
557 APPEND_GROW (4); // at most 4 bytes for 21 bits
558 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
559 }
560 else 871 default:
561 APPEND_CH (hi); 872 --dec_cur;
873 ERR ("illegal backslash escape sequence in string");
562 } 874 }
563 break; 875 }
876 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
877 *cur++ = ch;
878 else if (ch >= 0x80)
879 {
880 STRLEN clen;
881 UV uch;
564 882
565 default:
566 --dec->cur; 883 --dec_cur;
567 ERR ("illegal backslash escape sequence in string"); 884
885 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
886 if (clen == (STRLEN)-1)
887 ERR ("malformed UTF-8 character in JSON string");
888
889 do
890 *cur++ = *dec_cur++;
891 while (--clen);
892
893 utf8 = 1;
894 }
895 else
896 {
897 --dec_cur;
898
899 if (!ch)
900 ERR ("unexpected end of string while parsing JSON string");
901 else
902 ERR ("invalid character encountered while parsing JSON string");
568 } 903 }
569 } 904 }
570 else if (ch >= 0x20 && ch <= 0x7f) 905 while (cur < buf + SHORT_STRING_LEN);
571 APPEND_CH (*dec->cur++); 906
572 else if (ch >= 0x80)
573 { 907 {
574 STRLEN clen; 908 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");
578 909
579 APPEND_GROW (clen); 910 if (sv)
580 do
581 { 911 {
582 *cur++ = *dec->cur++; 912 SvGROW (sv, SvCUR (sv) + len + 1);
913 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
914 SvCUR_set (sv, SvCUR (sv) + len);
583 } 915 }
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 916 else
591 ERR ("invalid character encountered"); 917 sv = newSVpvn (buf, len);
592 } 918 }
919 }
920 while (*dec_cur != '"');
593 921
594 ++dec->cur; 922 ++dec_cur;
595 923
596 SvCUR_set (sv, cur - SvPVX (sv)); 924 if (sv)
597 925 {
598 SvPOK_only (sv); 926 SvPOK_only (sv);
599 *SvEND (sv) = 0; 927 *SvEND (sv) = 0;
600 928
601 if (utf8) 929 if (utf8)
602 SvUTF8_on (sv); 930 SvUTF8_on (sv);
931 }
932 else
933 sv = newSVpvn ("", 0);
603 934
935 dec->cur = dec_cur;
604 return sv; 936 return sv;
605 937
606fail: 938fail:
607 SvREFCNT_dec (sv); 939 dec->cur = dec_cur;
608 return 0; 940 return 0;
609} 941}
610 942
611static SV * 943static SV *
612decode_num (dec_t *dec) 944decode_num (dec_t *dec)
670 is_nv = 1; 1002 is_nv = 1;
671 } 1003 }
672 1004
673 if (!is_nv) 1005 if (!is_nv)
674 { 1006 {
675 UV uv; 1007 int len = dec->cur - start;
676 int numtype = grok_number (start, dec->cur - start, &uv); 1008
677 if (numtype & IS_NUMBER_IN_UV) 1009 // special case the rather common 1..5-digit-int case
678 if (numtype & IS_NUMBER_NEG) 1010 if (*start == '-')
1011 switch (len)
679 { 1012 {
680 if (uv < (UV)IV_MIN) 1013 case 2: return newSViv (-( start [1] - '0' * 1));
681 return newSViv (-(IV)uv); 1014 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1015 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1016 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1017 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
682 } 1018 }
1019 else
1020 switch (len)
1021 {
1022 case 1: return newSViv ( start [0] - '0' * 1);
1023 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1024 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1025 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1026 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1027 }
1028
1029 {
1030 UV uv;
1031 int numtype = grok_number (start, len, &uv);
1032 if (numtype & IS_NUMBER_IN_UV)
1033 if (numtype & IS_NUMBER_NEG)
1034 {
1035 if (uv < (UV)IV_MIN)
1036 return newSViv (-(IV)uv);
1037 }
683 else 1038 else
684 return newSVuv (uv); 1039 return newSVuv (uv);
685 } 1040 }
686 1041
1042 len -= *start == '-' ? 1 : 0;
1043
1044 // does not fit into IV or UV, try NV
1045 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1046 #if defined (LDBL_DIG)
1047 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1048 #endif
1049 )
1050 // fits into NV without loss of precision
1051 return newSVnv (Atof (start));
1052
1053 // everything else fails, convert it to a string
1054 return newSVpvn (start, dec->cur - start);
1055 }
1056
1057 // loss of precision here
687 return newSVnv (Atof (start)); 1058 return newSVnv (Atof (start));
688 1059
689fail: 1060fail:
690 return 0; 1061 return 0;
691} 1062}
693static SV * 1064static SV *
694decode_av (dec_t *dec) 1065decode_av (dec_t *dec)
695{ 1066{
696 AV *av = newAV (); 1067 AV *av = newAV ();
697 1068
698 WS; 1069 DEC_INC_DEPTH;
1070 decode_ws (dec);
1071
699 if (*dec->cur == ']') 1072 if (*dec->cur == ']')
700 ++dec->cur; 1073 ++dec->cur;
701 else 1074 else
702 for (;;) 1075 for (;;)
703 { 1076 {
707 if (!value) 1080 if (!value)
708 goto fail; 1081 goto fail;
709 1082
710 av_push (av, value); 1083 av_push (av, value);
711 1084
712 WS; 1085 decode_ws (dec);
713 1086
714 if (*dec->cur == ']') 1087 if (*dec->cur == ']')
715 { 1088 {
716 ++dec->cur; 1089 ++dec->cur;
717 break; 1090 break;
719 1092
720 if (*dec->cur != ',') 1093 if (*dec->cur != ',')
721 ERR (", or ] expected while parsing array"); 1094 ERR (", or ] expected while parsing array");
722 1095
723 ++dec->cur; 1096 ++dec->cur;
1097
1098 decode_ws (dec);
1099
1100 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1101 {
1102 ++dec->cur;
1103 break;
1104 }
724 } 1105 }
725 1106
1107 DEC_DEC_DEPTH;
726 return newRV_noinc ((SV *)av); 1108 return newRV_noinc ((SV *)av);
727 1109
728fail: 1110fail:
729 SvREFCNT_dec (av); 1111 SvREFCNT_dec (av);
1112 DEC_DEC_DEPTH;
730 return 0; 1113 return 0;
731} 1114}
732 1115
733static SV * 1116static SV *
734decode_hv (dec_t *dec) 1117decode_hv (dec_t *dec)
735{ 1118{
1119 SV *sv;
736 HV *hv = newHV (); 1120 HV *hv = newHV ();
737 1121
738 WS; 1122 DEC_INC_DEPTH;
1123 decode_ws (dec);
1124
739 if (*dec->cur == '}') 1125 if (*dec->cur == '}')
740 ++dec->cur; 1126 ++dec->cur;
741 else 1127 else
742 for (;;) 1128 for (;;)
743 { 1129 {
744 SV *key, *value;
745
746 WS; EXPECT_CH ('"'); 1130 EXPECT_CH ('"');
747 1131
748 key = decode_str (dec); 1132 // heuristic: assume that
749 if (!key) 1133 // a) decode_str + hv_store_ent are abysmally slow.
750 goto fail; 1134 // b) most hash keys are short, simple ascii text.
1135 // => try to "fast-match" such strings to avoid
1136 // the overhead of decode_str + hv_store_ent.
1137 {
1138 SV *value;
1139 char *p = dec->cur;
1140 char *e = p + 24; // only try up to 24 bytes
751 1141
752 WS; EXPECT_CH (':'); 1142 for (;;)
753
754 value = decode_sv (dec);
755 if (!value)
756 { 1143 {
1144 // the >= 0x80 is true on most architectures
1145 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1146 {
1147 // slow path, back up and use decode_str
1148 SV *key = decode_str (dec);
1149 if (!key)
1150 goto fail;
1151
1152 decode_ws (dec); EXPECT_CH (':');
1153
1154 decode_ws (dec);
1155 value = decode_sv (dec);
1156 if (!value)
1157 {
1158 SvREFCNT_dec (key);
1159 goto fail;
1160 }
1161
1162 hv_store_ent (hv, key, value, 0);
757 SvREFCNT_dec (key); 1163 SvREFCNT_dec (key);
1164
1165 break;
1166 }
1167 else if (*p == '"')
1168 {
1169 // fast path, got a simple key
1170 char *key = dec->cur;
1171 int len = p - key;
1172 dec->cur = p + 1;
1173
1174 decode_ws (dec); EXPECT_CH (':');
1175
1176 decode_ws (dec);
1177 value = decode_sv (dec);
1178 if (!value)
758 goto fail; 1179 goto fail;
1180
1181 hv_store (hv, key, len, value, 0);
1182
1183 break;
1184 }
1185
1186 ++p;
759 } 1187 }
760
761 //TODO: optimise
762 hv_store_ent (hv, key, value, 0);
763
764 WS; 1188 }
1189
1190 decode_ws (dec);
765 1191
766 if (*dec->cur == '}') 1192 if (*dec->cur == '}')
767 { 1193 {
768 ++dec->cur; 1194 ++dec->cur;
769 break; 1195 break;
771 1197
772 if (*dec->cur != ',') 1198 if (*dec->cur != ',')
773 ERR (", or } expected while parsing object/hash"); 1199 ERR (", or } expected while parsing object/hash");
774 1200
775 ++dec->cur; 1201 ++dec->cur;
1202
1203 decode_ws (dec);
1204
1205 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1206 {
1207 ++dec->cur;
1208 break;
1209 }
776 } 1210 }
777 1211
1212 DEC_DEC_DEPTH;
778 return newRV_noinc ((SV *)hv); 1213 sv = newRV_noinc ((SV *)hv);
1214
1215 // check filter callbacks
1216 if (dec->json.flags & F_HOOK)
1217 {
1218 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1219 {
1220 HE *cb, *he;
1221
1222 hv_iterinit (hv);
1223 he = hv_iternext (hv);
1224 hv_iterinit (hv);
1225
1226 // the next line creates a mortal sv each time its called.
1227 // might want to optimise this for common cases.
1228 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1229
1230 if (cb)
1231 {
1232 dSP;
1233 int count;
1234
1235 ENTER; SAVETMPS; PUSHMARK (SP);
1236 XPUSHs (HeVAL (he));
1237
1238 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1239
1240 if (count == 1)
1241 {
1242 sv = newSVsv (POPs);
1243 FREETMPS; LEAVE;
1244 return sv;
1245 }
1246
1247 FREETMPS; LEAVE;
1248 }
1249 }
1250
1251 if (dec->json.cb_object)
1252 {
1253 dSP;
1254 int count;
1255
1256 ENTER; SAVETMPS; PUSHMARK (SP);
1257 XPUSHs (sv_2mortal (sv));
1258
1259 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1260
1261 if (count == 1)
1262 {
1263 sv = newSVsv (POPs);
1264 FREETMPS; LEAVE;
1265 return sv;
1266 }
1267
1268 SvREFCNT_inc (sv);
1269 FREETMPS; LEAVE;
1270 }
1271 }
1272
1273 return sv;
779 1274
780fail: 1275fail:
781 SvREFCNT_dec (hv); 1276 SvREFCNT_dec (hv);
1277 DEC_DEC_DEPTH;
782 return 0; 1278 return 0;
783} 1279}
784 1280
785static SV * 1281static SV *
786decode_sv (dec_t *dec) 1282decode_sv (dec_t *dec)
787{ 1283{
788 WS; 1284 // the beauty of JSON: you need exactly one character lookahead
1285 // to parse anything.
789 switch (*dec->cur) 1286 switch (*dec->cur)
790 { 1287 {
791 case '"': ++dec->cur; return decode_str (dec); 1288 case '"': ++dec->cur; return decode_str (dec);
792 case '[': ++dec->cur; return decode_av (dec); 1289 case '[': ++dec->cur; return decode_av (dec);
793 case '{': ++dec->cur; return decode_hv (dec); 1290 case '{': ++dec->cur; return decode_hv (dec);
799 1296
800 case 't': 1297 case 't':
801 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1298 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
802 { 1299 {
803 dec->cur += 4; 1300 dec->cur += 4;
804 return newSViv (1); 1301#if JSON_SLOW
1302 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1303#endif
1304 return SvREFCNT_inc (json_true);
805 } 1305 }
806 else 1306 else
807 ERR ("'true' expected"); 1307 ERR ("'true' expected");
808 1308
809 break; 1309 break;
810 1310
811 case 'f': 1311 case 'f':
812 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1312 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
813 { 1313 {
814 dec->cur += 5; 1314 dec->cur += 5;
815 return newSViv (0); 1315#if JSON_SLOW
1316 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1317#endif
1318 return SvREFCNT_inc (json_false);
816 } 1319 }
817 else 1320 else
818 ERR ("'false' expected"); 1321 ERR ("'false' expected");
819 1322
820 break; 1323 break;
829 ERR ("'null' expected"); 1332 ERR ("'null' expected");
830 1333
831 break; 1334 break;
832 1335
833 default: 1336 default:
834 ERR ("malformed json string"); 1337 ERR ("malformed JSON string, neither array, object, number, string or atom");
835 break; 1338 break;
836 } 1339 }
837 1340
838fail: 1341fail:
839 return 0; 1342 return 0;
840} 1343}
841 1344
842static SV * 1345static SV *
843decode_json (SV *string, UV flags) 1346decode_json (SV *string, JSON *json, UV *offset_return)
844{ 1347{
1348 dec_t dec;
1349 UV offset;
845 SV *sv; 1350 SV *sv;
846 1351
1352 SvGETMAGIC (string);
1353 SvUPGRADE (string, SVt_PV);
1354
1355 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1356 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1357 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1358
847 if (flags & F_UTF8) 1359 if (json->flags & F_UTF8)
848 sv_utf8_downgrade (string, 0); 1360 sv_utf8_downgrade (string, 0);
849 else 1361 else
850 sv_utf8_upgrade (string); 1362 sv_utf8_upgrade (string);
851 1363
852 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1364 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
853 1365
854 dec_t dec; 1366 dec.json = *json;
855 dec.flags = flags;
856 dec.cur = SvPVX (string); 1367 dec.cur = SvPVX (string);
857 dec.end = SvEND (string); 1368 dec.end = SvEND (string);
858 dec.err = 0; 1369 dec.err = 0;
1370 dec.depth = 0;
1371 dec.maxdepth = DEC_DEPTH (dec.json.flags);
859 1372
1373 if (dec.json.cb_object || dec.json.cb_sk_object)
1374 dec.json.flags |= F_HOOK;
1375
1376 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1377
1378 decode_ws (&dec);
860 sv = decode_sv (&dec); 1379 sv = decode_sv (&dec);
861 1380
1381 if (!(offset_return || !sv))
1382 {
1383 // check for trailing garbage
1384 decode_ws (&dec);
1385
1386 if (*dec.cur)
1387 {
1388 dec.err = "garbage after JSON object";
1389 SvREFCNT_dec (sv);
1390 sv = 0;
1391 }
1392 }
1393
1394 if (offset_return || !sv)
1395 {
1396 offset = dec.json.flags & F_UTF8
1397 ? dec.cur - SvPVX (string)
1398 : utf8_distance (dec.cur, SvPVX (string));
1399
1400 if (offset_return)
1401 *offset_return = offset;
1402 }
1403
862 if (!sv) 1404 if (!sv)
863 { 1405 {
864 IV offset = utf8_distance (dec.cur, SvPVX (string));
865 SV *uni = sv_newmortal (); 1406 SV *uni = sv_newmortal ();
1407
866 // horrible hack to silence warning inside pv_uni_display 1408 // horrible hack to silence warning inside pv_uni_display
867 COP cop; 1409 COP cop = *PL_curcop;
868 memset (&cop, 0, sizeof (cop));
869 cop.cop_warnings = pWARN_NONE; 1410 cop.cop_warnings = pWARN_NONE;
1411 ENTER;
870 SAVEVPTR (PL_curcop); 1412 SAVEVPTR (PL_curcop);
871 PL_curcop = &cop; 1413 PL_curcop = &cop;
872
873 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1414 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1415 LEAVE;
1416
874 croak ("%s, at character offset %d (%s)", 1417 croak ("%s, at character offset %d [\"%s\"]",
875 dec.err, 1418 dec.err,
876 (int)offset, 1419 (int)offset,
877 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1420 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
878 } 1421 }
879 1422
880 sv = sv_2mortal (sv); 1423 sv = sv_2mortal (sv);
881 1424
882 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1425 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)"); 1426 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
884 1427
885 return sv; 1428 return sv;
886} 1429}
887 1430
1431/////////////////////////////////////////////////////////////////////////////
1432// XS interface functions
1433
888MODULE = JSON::XS PACKAGE = JSON::XS 1434MODULE = JSON::XS PACKAGE = JSON::XS
889 1435
890BOOT: 1436BOOT:
891{ 1437{
892 int i; 1438 int i;
893 1439
894 memset (decode_hexdigit, 0xff, 256);
895 for (i = 10; i--; ) 1440 for (i = 0; i < 256; ++i)
896 decode_hexdigit ['0' + i] = i; 1441 decode_hexdigit [i] =
1442 i >= '0' && i <= '9' ? i - '0'
1443 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1444 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1445 : -1;
897 1446
898 for (i = 7; i--; )
899 {
900 decode_hexdigit ['a' + i] = 10 + i;
901 decode_hexdigit ['A' + i] = 10 + i;
902 }
903
904 json_stash = gv_stashpv ("JSON::XS", 1); 1447 json_stash = gv_stashpv ("JSON::XS" , 1);
1448 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1449
1450 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1451 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
905} 1452}
906 1453
907PROTOTYPES: DISABLE 1454PROTOTYPES: DISABLE
908 1455
909SV *new (char *dummy) 1456void CLONE (...)
910 CODE: 1457 CODE:
911 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1458 json_stash = 0;
1459 json_boolean_stash = 0;
1460
1461void new (char *klass)
1462 PPCODE:
1463{
1464 SV *pv = NEWSV (0, sizeof (JSON));
1465 SvPOK_only (pv);
1466 Zero (SvPVX (pv), 1, JSON);
1467 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1468 XPUSHs (sv_2mortal (sv_bless (
1469 newRV_noinc (pv),
1470 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1471 )));
1472}
1473
1474void ascii (JSON *self, int enable = 1)
1475 ALIAS:
1476 ascii = F_ASCII
1477 latin1 = F_LATIN1
1478 utf8 = F_UTF8
1479 indent = F_INDENT
1480 canonical = F_CANONICAL
1481 space_before = F_SPACE_BEFORE
1482 space_after = F_SPACE_AFTER
1483 pretty = F_PRETTY
1484 allow_nonref = F_ALLOW_NONREF
1485 shrink = F_SHRINK
1486 allow_blessed = F_ALLOW_BLESSED
1487 convert_blessed = F_CONV_BLESSED
1488 relaxed = F_RELAXED
1489 PPCODE:
1490{
1491 if (enable)
1492 self->flags |= ix;
1493 else
1494 self->flags &= ~ix;
1495
1496 XPUSHs (ST (0));
1497}
1498
1499void get_ascii (JSON *self)
1500 ALIAS:
1501 get_ascii = F_ASCII
1502 get_latin1 = F_LATIN1
1503 get_utf8 = F_UTF8
1504 get_indent = F_INDENT
1505 get_canonical = F_CANONICAL
1506 get_space_before = F_SPACE_BEFORE
1507 get_space_after = F_SPACE_AFTER
1508 get_allow_nonref = F_ALLOW_NONREF
1509 get_shrink = F_SHRINK
1510 get_allow_blessed = F_ALLOW_BLESSED
1511 get_convert_blessed = F_CONV_BLESSED
1512 get_relaxed = F_RELAXED
1513 PPCODE:
1514 XPUSHs (boolSV (self->flags & ix));
1515
1516void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1517 PPCODE:
1518{
1519 UV log2 = 0;
1520
1521 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1522
1523 while ((1UL << log2) < max_depth)
1524 ++log2;
1525
1526 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1527
1528 XPUSHs (ST (0));
1529}
1530
1531U32 get_max_depth (JSON *self)
1532 CODE:
1533 RETVAL = DEC_DEPTH (self->flags);
912 OUTPUT: 1534 OUTPUT:
913 RETVAL 1535 RETVAL
914 1536
915SV *ascii (SV *self, int enable) 1537void max_size (JSON *self, UV max_size = 0)
916 ALIAS: 1538 PPCODE:
917 ascii = F_ASCII 1539{
918 utf8 = F_UTF8 1540 UV log2 = 0;
919 indent = F_INDENT 1541
920 canonical = F_CANONICAL 1542 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
921 space_before = F_SPACE_BEFORE 1543 if (max_size == 1) max_size = 2;
922 space_after = F_SPACE_AFTER 1544
923 json_rpc = F_JSON_RPC 1545 while ((1UL << log2) < max_size)
924 pretty = F_PRETTY 1546 ++log2;
925 allow_nonref = F_ALLOW_NONREF 1547
1548 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1549
1550 XPUSHs (ST (0));
1551}
1552
1553int get_max_size (JSON *self)
926 CODE: 1554 CODE:
927{ 1555 RETVAL = DEC_SIZE (self->flags);
928 UV *uv = SvJSON (self);
929 if (enable)
930 *uv |= ix;
931 else
932 *uv &= ~ix;
933
934 RETVAL = newSVsv (self);
935}
936 OUTPUT: 1556 OUTPUT:
937 RETVAL 1557 RETVAL
938 1558
939void encode (SV *self, SV *scalar) 1559void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
940 PPCODE: 1560 PPCODE:
941 XPUSHs (encode_json (scalar, *SvJSON (self))); 1561{
1562 SvREFCNT_dec (self->cb_object);
1563 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
942 1564
943void decode (SV *self, SV *jsonstr) 1565 XPUSHs (ST (0));
1566}
1567
1568void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
944 PPCODE: 1569 PPCODE:
1570{
1571 if (!self->cb_sk_object)
1572 self->cb_sk_object = newHV ();
1573
1574 if (SvOK (cb))
1575 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1576 else
1577 {
1578 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1579
1580 if (!HvKEYS (self->cb_sk_object))
1581 {
1582 SvREFCNT_dec (self->cb_sk_object);
1583 self->cb_sk_object = 0;
1584 }
1585 }
1586
1587 XPUSHs (ST (0));
1588}
1589
1590void encode (JSON *self, SV *scalar)
1591 PPCODE:
1592 XPUSHs (encode_json (scalar, self));
1593
1594void decode (JSON *self, SV *jsonstr)
1595 PPCODE:
945 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1596 XPUSHs (decode_json (jsonstr, self, 0));
1597
1598void decode_prefix (JSON *self, SV *jsonstr)
1599 PPCODE:
1600{
1601 UV offset;
1602 EXTEND (SP, 2);
1603 PUSHs (decode_json (jsonstr, self, &offset));
1604 PUSHs (sv_2mortal (newSVuv (offset)));
1605}
1606
1607void DESTROY (JSON *self)
1608 CODE:
1609 SvREFCNT_dec (self->cb_sk_object);
1610 SvREFCNT_dec (self->cb_object);
946 1611
947PROTOTYPES: ENABLE 1612PROTOTYPES: ENABLE
948 1613
949void to_json (SV *scalar) 1614void encode_json (SV *scalar)
950 PPCODE: 1615 PPCODE:
1616{
1617 JSON json = { F_DEFAULT | F_UTF8 };
951 XPUSHs (encode_json (scalar, F_UTF8)); 1618 XPUSHs (encode_json (scalar, &json));
1619}
952 1620
953void from_json (SV *jsonstr) 1621void decode_json (SV *jsonstr)
954 PPCODE: 1622 PPCODE:
1623{
1624 JSON json = { F_DEFAULT | F_UTF8 };
955 XPUSHs (decode_json (jsonstr, F_UTF8)); 1625 XPUSHs (decode_json (jsonstr, &json, 0));
1626}
956 1627

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines