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 1.4 by root, Thu Mar 22 21:13:58 2007 UTC vs.
Revision 1.54 by root, Tue Jul 10 16:22:31 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 27#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 39
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 41#define F_DEFAULT (9UL << S_MAXDEPTH)
20 42
21#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level
45
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 47
23#define SB do { 48#define SB do {
24#define SE } while (0) 49#define SE } while (0)
25 50
26static HV *json_stash; 51#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value))
53# define inline inline
54#else
55# define expect(expr,value) (expr)
56# define inline static
57#endif
58
59#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1)
61
62static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false;
64
65typedef struct {
66 U32 flags;
67 SV *cb_object;
68 HV *cb_sk_object;
69} JSON;
70
71/////////////////////////////////////////////////////////////////////////////
72// utility functions
73
74inline void
75shrink (SV *sv)
76{
77 sv_utf8_downgrade (sv, 1);
78 if (SvLEN (sv) > SvCUR (sv) + 1)
79 {
80#ifdef SvPV_shrink_to_cur
81 SvPV_shrink_to_cur (sv);
82#elif defined (SvPV_renew)
83 SvPV_renew (sv, SvCUR (sv) + 1);
84#endif
85 }
86}
87
88// decode an utf-8 character and return it, or (UV)-1 in
89// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints
93inline UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
99 {
100 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 }
103 else
104 {
105 *clen = (STRLEN)-1;
106 return (UV)-1;
107 }
108}
109
110/////////////////////////////////////////////////////////////////////////////
111// encoder
27 112
28// structure used for encoding JSON 113// structure used for encoding JSON
29typedef struct 114typedef struct
30{ 115{
31 char *cur; 116 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
34 SV *sv; 118 SV *sv; // result scalar
35 UV flags; 119 JSON json;
36 int max_recurse; 120 U32 indent; // indentation level
37 int indent; 121 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 122} enc_t;
39 123
40// structure used for decoding JSON 124inline 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) 125need (enc_t *enc, STRLEN len)
62{ 126{
63 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
64 { 128 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 133 }
70} 134}
71 135
72static void 136inline void
73encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
74{ 138{
75 need (enc, 1); 139 need (enc, 1);
76 *enc->cur++ = ch; 140 *enc->cur++ = ch;
77} 141}
… …
85 149
86 while (str < end) 150 while (str < end)
87 { 151 {
88 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
89 153
90 if (ch == '"') 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 155 {
156 if (expect_false (ch == '"')) // but with slow exceptions
157 {
92 need (enc, len += 1); 158 need (enc, len += 1);
93 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 160 *enc->cur++ = '"';
95 ++str;
96 } 161 }
97 else if (ch == '\\') 162 else if (expect_false (ch == '\\'))
98 { 163 {
99 need (enc, len += 1); 164 need (enc, len += 1);
100 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
102 ++str;
103 } 167 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 168 else
105 {
106 *enc->cur++ = ch; 169 *enc->cur++ = ch;
107 ++str; 170
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; 171 ++str;
122 } 172 }
123 else 173 else
124 { 174 {
125 STRLEN clen; 175 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 < 0)
132 croak ("malformed UTF-8 character in string, cannot convert to JSON");
133 } 176 {
134 else 177 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
135 { 178 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
136 uch = ch; 179 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
137 clen = 1; 180 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
138 } 181 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
139 182
140 if (uch < 0x80 || enc->flags & F_ASCII) 183 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 184 {
185 STRLEN clen;
186 UV uch;
187
188 if (is_utf8)
189 {
190 uch = decode_utf8 (str, end - str, &clen);
191 if (clen == (STRLEN)-1)
192 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
193 }
194 else
195 {
196 uch = ch;
197 clen = 1;
198 }
199
200 if (uch > 0x10FFFFUL)
201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
202
203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
204 {
205 if (uch > 0xFFFFUL)
206 {
144 need (enc, len += 11); 207 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 209 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 210 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 211 enc->cur += 12;
212 }
213 else
214 {
215 static char hexdigit [16] = "0123456789abcdef";
216 need (enc, len += 5);
217 *enc->cur++ = '\\';
218 *enc->cur++ = 'u';
219 *enc->cur++ = hexdigit [ uch >> 12 ];
220 *enc->cur++ = hexdigit [(uch >> 8) & 15];
221 *enc->cur++ = hexdigit [(uch >> 4) & 15];
222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
223 }
224
225 str += clen;
226 }
227 else if (enc->json.flags & F_LATIN1)
228 {
229 *enc->cur++ = uch;
230 str += clen;
231 }
232 else if (is_utf8)
233 {
234 need (enc, len += clen);
235 do
236 {
237 *enc->cur++ = *str++;
238 }
239 while (--clen);
240 }
241 else
242 {
243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
245 ++str;
246 }
149 } 247 }
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 while (clen--)
168 *enc->cur++ = *str++;
169 }
170 else
171 {
172 need (enc, 10); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str;
175 } 248 }
176 } 249 }
177 250
178 --len; 251 --len;
179 } 252 }
180} 253}
181 254
182#define INDENT SB \ 255inline void
256encode_indent (enc_t *enc)
257{
183 if (enc->flags & F_INDENT) \ 258 if (enc->json.flags & F_INDENT)
184 { \ 259 {
185 int i_; \ 260 int spaces = enc->indent * INDENT_STEP;
186 need (enc, enc->indent); \ 261
187 for (i_ = enc->indent * 3; i_--; )\ 262 need (enc, spaces);
263 memset (enc->cur, ' ', spaces);
264 enc->cur += spaces;
265 }
266}
267
268inline void
269encode_space (enc_t *enc)
270{
271 need (enc, 1);
272 encode_ch (enc, ' ');
273}
274
275inline void
276encode_nl (enc_t *enc)
277{
278 if (enc->json.flags & F_INDENT)
279 {
280 need (enc, 1);
188 encode_ch (enc, ' '); \ 281 encode_ch (enc, '\n');
189 } \ 282 }
190 SE 283}
191 284
192#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 285inline void
193#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 286encode_comma (enc_t *enc)
194#define COMMA SB \ 287{
195 encode_ch (enc, ','); \ 288 encode_ch (enc, ',');
289
196 if (enc->flags & F_INDENT) \ 290 if (enc->json.flags & F_INDENT)
197 NL; \ 291 encode_nl (enc);
198 else if (enc->flags & F_SPACE_AFTER) \ 292 else if (enc->json.flags & F_SPACE_AFTER)
199 SPACE; \ 293 encode_space (enc);
200 SE 294}
201 295
202static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
203 297
204static void 298static void
205encode_av (enc_t *enc, AV *av) 299encode_av (enc_t *enc, AV *av)
206{ 300{
207 int i, len = av_len (av); 301 int i, len = av_len (av);
208 302
209 encode_ch (enc, '['); NL; 303 if (enc->indent >= enc->maxdepth)
304 croak ("data structure too deep (hit recursion limit)");
305
306 encode_ch (enc, '['); encode_nl (enc);
210 ++enc->indent; 307 ++enc->indent;
211 308
212 for (i = 0; i <= len; ++i) 309 for (i = 0; i <= len; ++i)
213 { 310 {
214 INDENT; 311 encode_indent (enc);
215 encode_sv (enc, *av_fetch (av, i, 0)); 312 encode_sv (enc, *av_fetch (av, i, 0));
216 313
217 if (i < len) 314 if (i < len)
218 COMMA; 315 encode_comma (enc);
219 } 316 }
220 317
221 NL; 318 encode_nl (enc);
222 319
223 --enc->indent; 320 --enc->indent;
224 INDENT; encode_ch (enc, ']'); 321 encode_indent (enc); encode_ch (enc, ']');
225} 322}
226 323
227static void 324static void
228encode_he (enc_t *enc, HE *he) 325encode_he (enc_t *enc, HE *he)
229{ 326{
… …
243 else 340 else
244 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 341 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
245 342
246 encode_ch (enc, '"'); 343 encode_ch (enc, '"');
247 344
248 if (enc->flags & F_SPACE_BEFORE) SPACE; 345 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
249 encode_ch (enc, ':'); 346 encode_ch (enc, ':');
250 if (enc->flags & F_SPACE_AFTER ) SPACE; 347 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
251 encode_sv (enc, HeVAL (he)); 348 encode_sv (enc, HeVAL (he));
252} 349}
253 350
254// compare hash entries, used when all keys are bytestrings 351// compare hash entries, used when all keys are bytestrings
255static int 352static int
… …
261 HE *b = *(HE **)b_; 358 HE *b = *(HE **)b_;
262 359
263 STRLEN la = HeKLEN (a); 360 STRLEN la = HeKLEN (a);
264 STRLEN lb = HeKLEN (b); 361 STRLEN lb = HeKLEN (b);
265 362
266 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 363 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
267 cmp = la < lb ? -1 : la == lb ? 0 : 1; 364 cmp = la - lb;
268 365
269 return cmp; 366 return cmp;
270} 367}
271 368
272// compare hash entries, used when some keys are sv's or utf-x 369// compare hash entries, used when some keys are sv's or utf-x
… …
279static void 376static void
280encode_hv (enc_t *enc, HV *hv) 377encode_hv (enc_t *enc, HV *hv)
281{ 378{
282 int count, i; 379 int count, i;
283 380
381 if (enc->indent >= enc->maxdepth)
382 croak ("data structure too deep (hit recursion limit)");
383
284 encode_ch (enc, '{'); NL; ++enc->indent; 384 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
285 385
286 if ((count = hv_iterinit (hv))) 386 if ((count = hv_iterinit (hv)))
287 { 387 {
288 // for canonical output we have to sort by keys first 388 // for canonical output we have to sort by keys first
289 // actually, this is mostly due to the stupid so-called 389 // actually, this is mostly due to the stupid so-called
290 // security workaround added somewhere in 5.8.x. 390 // security workaround added somewhere in 5.8.x.
291 // that randomises hash orderings 391 // that randomises hash orderings
292 if (enc->flags & F_CANONICAL) 392 if (enc->json.flags & F_CANONICAL)
293 { 393 {
294 HE *he, *hes [count];
295 int fast = 1; 394 int fast = 1;
395 HE *he;
396#if defined(__BORLANDC__) || defined(_MSC_VER)
397 HE **hes = _alloca (count * sizeof (HE));
398#else
399 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
400#endif
296 401
297 i = 0; 402 i = 0;
298 while ((he = hv_iternext (hv))) 403 while ((he = hv_iternext (hv)))
299 { 404 {
300 hes [i++] = he; 405 hes [i++] = he;
… …
306 411
307 if (fast) 412 if (fast)
308 qsort (hes, count, sizeof (HE *), he_cmp_fast); 413 qsort (hes, count, sizeof (HE *), he_cmp_fast);
309 else 414 else
310 { 415 {
311 // hack to disable "use bytes" 416 // hack to forcefully disable "use bytes"
312 COP *oldcop = PL_curcop, cop; 417 COP cop = *PL_curcop;
313 cop.op_private = 0; 418 cop.op_private = 0;
419
420 ENTER;
421 SAVETMPS;
422
423 SAVEVPTR (PL_curcop);
314 PL_curcop = &cop; 424 PL_curcop = &cop;
315 425
316 SAVETMPS;
317 qsort (hes, count, sizeof (HE *), he_cmp_slow); 426 qsort (hes, count, sizeof (HE *), he_cmp_slow);
427
318 FREETMPS; 428 FREETMPS;
319 429 LEAVE;
320 PL_curcop = oldcop;
321 } 430 }
322 431
323 for (i = 0; i < count; ++i) 432 for (i = 0; i < count; ++i)
324 { 433 {
325 INDENT; 434 encode_indent (enc);
326 encode_he (enc, hes [i]); 435 encode_he (enc, hes [i]);
327 436
328 if (i < count - 1) 437 if (i < count - 1)
329 COMMA; 438 encode_comma (enc);
330 }
331
332 NL; 439 }
440
441 encode_nl (enc);
333 } 442 }
334 else 443 else
335 { 444 {
336 SV *sv;
337 HE *he = hv_iternext (hv); 445 HE *he = hv_iternext (hv);
338 446
339 for (;;) 447 for (;;)
340 { 448 {
341 INDENT; 449 encode_indent (enc);
342 encode_he (enc, he); 450 encode_he (enc, he);
343 451
344 if (!(he = hv_iternext (hv))) 452 if (!(he = hv_iternext (hv)))
345 break; 453 break;
346 454
347 COMMA; 455 encode_comma (enc);
348 }
349
350 NL; 456 }
457
458 encode_nl (enc);
351 } 459 }
352 } 460 }
353 461
354 --enc->indent; INDENT; encode_ch (enc, '}'); 462 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
463}
464
465// encode objects, arrays and special \0=false and \1=true values.
466static void
467encode_rv (enc_t *enc, SV *sv)
468{
469 svtype svt;
470
471 SvGETMAGIC (sv);
472 svt = SvTYPE (sv);
473
474 if (expect_false (SvOBJECT (sv)))
475 {
476 if (SvSTASH (sv) == json_boolean_stash)
477 {
478 if (SvIV (sv))
479 encode_str (enc, "true", 4, 0);
480 else
481 encode_str (enc, "false", 5, 0);
482 }
483 else
484 {
485#if 0
486 if (0 && sv_derived_from (rv, "JSON::Literal"))
487 {
488 // not yet
489 }
490#endif
491 if (enc->json.flags & F_CONV_BLESSED)
492 {
493 // we re-bless the reference to get overload and other niceties right
494 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
495
496 if (to_json)
497 {
498 int count;
499 dSP;
500
501 ENTER; SAVETMPS; PUSHMARK (SP);
502 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
503
504 // calling with G_SCALAR ensures that we always get a 1 return value
505 PUTBACK;
506 call_sv ((SV *)GvCV (to_json), G_SCALAR);
507 SPAGAIN;
508
509 // catch this surprisingly common error
510 if (SvROK (TOPs) && SvRV (TOPs) == sv)
511 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
512
513 sv = POPs;
514 PUTBACK;
515
516 encode_sv (enc, sv);
517
518 FREETMPS; LEAVE;
519 }
520 else if (enc->json.flags & F_ALLOW_BLESSED)
521 encode_str (enc, "null", 4, 0);
522 else
523 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
524 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
525 }
526 else if (enc->json.flags & F_ALLOW_BLESSED)
527 encode_str (enc, "null", 4, 0);
528 else
529 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
530 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
531 }
532 }
533 else if (svt == SVt_PVHV)
534 encode_hv (enc, (HV *)sv);
535 else if (svt == SVt_PVAV)
536 encode_av (enc, (AV *)sv);
537 else if (svt < SVt_PVAV)
538 {
539 STRLEN len = 0;
540 char *pv = svt ? SvPV (sv, len) : 0;
541
542 if (len == 1 && *pv == '1')
543 encode_str (enc, "true", 4, 0);
544 else if (len == 1 && *pv == '0')
545 encode_str (enc, "false", 5, 0);
546 else
547 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
548 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
549 }
550 else
551 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
552 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
355} 553}
356 554
357static void 555static void
358encode_sv (enc_t *enc, SV *sv) 556encode_sv (enc_t *enc, SV *sv)
359{ 557{
… …
367 encode_str (enc, str, len, SvUTF8 (sv)); 565 encode_str (enc, str, len, SvUTF8 (sv));
368 encode_ch (enc, '"'); 566 encode_ch (enc, '"');
369 } 567 }
370 else if (SvNOKp (sv)) 568 else if (SvNOKp (sv))
371 { 569 {
570 // trust that perl will do the right thing w.r.t. JSON syntax.
372 need (enc, NV_DIG + 32); 571 need (enc, NV_DIG + 32);
373 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 572 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
374 enc->cur += strlen (enc->cur); 573 enc->cur += strlen (enc->cur);
375 } 574 }
376 else if (SvIOKp (sv)) 575 else if (SvIOKp (sv))
377 { 576 {
378 need (enc, 64); 577 // we assume we can always read an IV as a UV
578 if (SvUV (sv) & ~(UV)0x7fff)
579 {
580 // large integer, use the (rather slow) snprintf way.
581 need (enc, sizeof (UV) * 3);
379 enc->cur += 582 enc->cur +=
380 SvIsUV(sv) 583 SvIsUV(sv)
381 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 584 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
382 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 585 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
586 }
587 else
588 {
589 // optimise the "small number case"
590 // code will likely be branchless and use only a single multiplication
591 I32 i = SvIV (sv);
592 U32 u;
593 char digit, nz = 0;
594
595 need (enc, 6);
596
597 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
598 u = i < 0 ? -i : i;
599
600 // convert to 4.28 fixed-point representation
601 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
602
603 // now output digit by digit, each time masking out the integer part
604 // and multiplying by 5 while moving the decimal point one to the right,
605 // resulting in a net multiplication by 10.
606 // we always write the digit to memory but conditionally increment
607 // the pointer, to ease the usage of conditional move instructions.
608 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
609 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
610 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
611 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
612 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
613 }
383 } 614 }
384 else if (SvROK (sv)) 615 else if (SvROK (sv))
385 { 616 encode_rv (enc, SvRV (sv));
386 if (!--enc->max_recurse)
387 croak ("data structure too deep (hit recursion limit)");
388
389 sv = SvRV (sv);
390
391 switch (SvTYPE (sv))
392 {
393 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
394 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
395
396 default:
397 croak ("JSON can only represent references to arrays or hashes");
398 }
399 }
400 else if (!SvOK (sv)) 617 else if (!SvOK (sv))
401 encode_str (enc, "null", 4, 0); 618 encode_str (enc, "null", 4, 0);
402 else 619 else
403 croak ("encountered perl type that JSON cannot handle"); 620 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
621 SvPV_nolen (sv), SvFLAGS (sv));
404} 622}
405 623
406static SV * 624static SV *
407encode_json (SV *scalar, UV flags) 625encode_json (SV *scalar, JSON *json)
408{ 626{
409 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
410 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");
411
412 enc_t enc; 627 enc_t enc;
413 enc.flags = flags; 628
629 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
630 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
631
632 enc.json = *json;
414 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 633 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
415 enc.cur = SvPVX (enc.sv); 634 enc.cur = SvPVX (enc.sv);
416 enc.end = SvEND (enc.sv); 635 enc.end = SvEND (enc.sv);
417 enc.max_recurse = 0;
418 enc.indent = 0; 636 enc.indent = 0;
637 enc.maxdepth = DEC_DEPTH (enc.json.flags);
419 638
420 SvPOK_only (enc.sv); 639 SvPOK_only (enc.sv);
421 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
422 641
642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
643 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
644
423 if (!(flags & (F_ASCII | F_UTF8))) 645 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
424 SvUTF8_on (enc.sv); 646 SvUTF8_on (enc.sv);
425 647
426 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 648 if (enc.json.flags & F_SHRINK)
649 shrink (enc.sv);
650
427 return enc.sv; 651 return enc.sv;
428} 652}
429 653
430///////////////////////////////////////////////////////////////////////////// 654/////////////////////////////////////////////////////////////////////////////
655// decoder
431 656
432#define WS \ 657// structure used for decoding JSON
658typedef struct
659{
660 char *cur; // current parser pointer
661 char *end; // end of input string
662 const char *err; // parse error, if != 0
663 JSON json;
664 U32 depth; // recursion depth
665 U32 maxdepth; // recursion depth limit
666} dec_t;
667
668inline void
669decode_ws (dec_t *dec)
670{
433 for (;;) \ 671 for (;;)
434 { \ 672 {
435 char ch = *dec->cur; \ 673 char ch = *dec->cur;
674
436 if (ch > 0x20 \ 675 if (ch > 0x20
437 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 676 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
438 break; \ 677 break;
678
439 ++dec->cur; \ 679 ++dec->cur;
440 } 680 }
681}
441 682
442#define ERR(reason) SB dec->err = reason; goto fail; SE 683#define ERR(reason) SB dec->err = reason; goto fail; SE
684
443#define EXPECT_CH(ch) SB \ 685#define EXPECT_CH(ch) SB \
444 if (*dec->cur != ch) \ 686 if (*dec->cur != ch) \
445 ERR (# ch " expected"); \ 687 ERR (# ch " expected"); \
446 ++dec->cur; \ 688 ++dec->cur; \
447 SE 689 SE
448 690
691#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
692#define DEC_DEC_DEPTH --dec->depth
693
449static SV *decode_sv (dec_t *dec); 694static SV *decode_sv (dec_t *dec);
450 695
451static signed char decode_hexdigit[256]; 696static signed char decode_hexdigit[256];
452 697
453static UV 698static UV
454decode_4hex (dec_t *dec) 699decode_4hex (dec_t *dec)
455{ 700{
456 signed char d1, d2, d3, d4; 701 signed char d1, d2, d3, d4;
702 unsigned char *cur = (unsigned char *)dec->cur;
457 703
458 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 704 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
459 if (d1 < 0) ERR ("four hexadecimal digits expected"); 705 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
460 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 706 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
461 if (d2 < 0) ERR ("four hexadecimal digits expected"); 707 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
462 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
463 if (d3 < 0) ERR ("four hexadecimal digits expected");
464 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
465 if (d4 < 0) ERR ("four hexadecimal digits expected");
466 708
467 dec->cur += 4; 709 dec->cur += 4;
468 710
469 return ((UV)d1) << 12 711 return ((UV)d1) << 12
470 | ((UV)d2) << 8 712 | ((UV)d2) << 8
… …
473 715
474fail: 716fail:
475 return (UV)-1; 717 return (UV)-1;
476} 718}
477 719
478#define APPEND_GROW(n) SB \
479 if (cur + (n) >= end) \
480 { \
481 STRLEN ofs = cur - SvPVX (sv); \
482 SvGROW (sv, ofs + (n) + 1); \
483 cur = SvPVX (sv) + ofs; \
484 end = SvEND (sv); \
485 } \
486 SE
487
488#define APPEND_CH(ch) SB \
489 APPEND_GROW (1); \
490 *cur++ = (ch); \
491 SE
492
493static SV * 720static SV *
494decode_str (dec_t *dec) 721decode_str (dec_t *dec)
495{ 722{
496 SV *sv = NEWSV (0,2); 723 SV *sv = 0;
497 int utf8 = 0; 724 int utf8 = 0;
498 char *cur = SvPVX (sv); 725 char *dec_cur = dec->cur;
499 char *end = SvEND (sv);
500 726
501 for (;;) 727 do
502 { 728 {
503 unsigned char ch = *(unsigned char *)dec->cur; 729 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
730 char *cur = buf;
504 731
505 if (ch == '"') 732 do
506 break;
507 else if (ch == '\\')
508 { 733 {
509 switch (*++dec->cur) 734 unsigned char ch = *(unsigned char *)dec_cur++;
735
736 if (expect_false (ch == '"'))
737 {
738 --dec_cur;
739 break;
510 { 740 }
511 case '\\': 741 else if (expect_false (ch == '\\'))
512 case '/': 742 {
513 case '"': APPEND_CH (*dec->cur++); break; 743 switch (*dec_cur)
514
515 case 'b': APPEND_CH ('\010'); ++dec->cur; break;
516 case 't': APPEND_CH ('\011'); ++dec->cur; break;
517 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
518 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
519 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
520
521 case 'u':
522 { 744 {
523 UV lo, hi; 745 case '\\':
524 ++dec->cur; 746 case '/':
747 case '"': *cur++ = *dec_cur++; break;
525 748
526 hi = decode_4hex (dec); 749 case 'b': ++dec_cur; *cur++ = '\010'; break;
527 if (hi == (UV)-1) 750 case 't': ++dec_cur; *cur++ = '\011'; break;
528 goto fail; 751 case 'n': ++dec_cur; *cur++ = '\012'; break;
752 case 'f': ++dec_cur; *cur++ = '\014'; break;
753 case 'r': ++dec_cur; *cur++ = '\015'; break;
529 754
530 // possibly a surrogate pair 755 case 'u':
531 if (hi >= 0xd800 && hi < 0xdc00)
532 { 756 {
533 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 757 UV lo, hi;
534 ERR ("illegal surrogate character"); 758 ++dec_cur;
535 759
536 dec->cur += 2; 760 dec->cur = dec_cur;
537
538 lo = decode_4hex (dec); 761 hi = decode_4hex (dec);
762 dec_cur = dec->cur;
539 if (lo == (UV)-1) 763 if (hi == (UV)-1)
540 goto fail; 764 goto fail;
541 765
766 // possibly a surrogate pair
767 if (hi >= 0xd800)
768 if (hi < 0xdc00)
769 {
770 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
771 ERR ("missing low surrogate character in surrogate pair");
772
773 dec_cur += 2;
774
775 dec->cur = dec_cur;
776 lo = decode_4hex (dec);
777 dec_cur = dec->cur;
778 if (lo == (UV)-1)
779 goto fail;
780
542 if (lo < 0xdc00 || lo >= 0xe000) 781 if (lo < 0xdc00 || lo >= 0xe000)
543 ERR ("surrogate pair expected"); 782 ERR ("surrogate pair expected");
544 783
545 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 784 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
785 }
786 else if (hi < 0xe000)
787 ERR ("missing high surrogate character in surrogate pair");
788
789 if (hi >= 0x80)
790 {
791 utf8 = 1;
792
793 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
794 }
795 else
796 *cur++ = hi;
546 } 797 }
547 else if (lo >= 0xdc00 && lo < 0xe000)
548 ERR ("illegal surrogate character");
549
550 if (hi >= 0x80)
551 { 798 break;
552 utf8 = 1;
553 799
554 APPEND_GROW (4); // at most 4 bytes for 21 bits
555 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
556 }
557 else 800 default:
558 APPEND_CH (hi); 801 --dec_cur;
802 ERR ("illegal backslash escape sequence in string");
559 } 803 }
804 }
805 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
806 *cur++ = ch;
807 else if (ch >= 0x80)
808 {
809 STRLEN clen;
810 UV uch;
811
812 --dec_cur;
813
814 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
815 if (clen == (STRLEN)-1)
816 ERR ("malformed UTF-8 character in JSON string");
817
818 do
819 *cur++ = *dec_cur++;
820 while (--clen);
821
822 utf8 = 1;
823 }
824 else
825 {
826 --dec_cur;
827
828 if (!ch)
829 ERR ("unexpected end of string while parsing JSON string");
560 break; 830 else
831 ERR ("invalid character encountered while parsing JSON string");
561 } 832 }
562 } 833 }
563 else if (ch >= 0x20 && ch <= 0x7f) 834 while (cur < buf + SHORT_STRING_LEN);
564 APPEND_CH (*dec->cur++); 835
565 else if (ch >= 0x80) 836 {
837 STRLEN len = cur - buf;
838
839 if (sv)
566 { 840 {
567 STRLEN clen; 841 SvGROW (sv, SvCUR (sv) + len + 1);
568 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 842 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
569 if (clen < 0) 843 SvCUR_set (sv, SvCUR (sv) + len);
570 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
571
572 APPEND_GROW (clen);
573 memcpy (cur, dec->cur, clen);
574 cur += clen;
575 dec->cur += clen;
576 } 844 }
577 else 845 else
578 ERR ("invalid character encountered"); 846 sv = newSVpvn (buf, len);
579 } 847 }
848 }
849 while (*dec_cur != '"');
580 850
581 ++dec->cur; 851 ++dec_cur;
582 852
583 SvCUR_set (sv, cur - SvPVX (sv)); 853 if (sv)
584 854 {
585 SvPOK_only (sv); 855 SvPOK_only (sv);
586 *SvEND (sv) = 0; 856 *SvEND (sv) = 0;
587 857
588 if (utf8) 858 if (utf8)
589 SvUTF8_on (sv); 859 SvUTF8_on (sv);
860 }
861 else
862 sv = newSVpvn ("", 0);
590 863
864 dec->cur = dec_cur;
591 return sv; 865 return sv;
592 866
593fail: 867fail:
594 SvREFCNT_dec (sv); 868 dec->cur = dec_cur;
595 return 0; 869 return 0;
596} 870}
597 871
598static SV * 872static SV *
599decode_num (dec_t *dec) 873decode_num (dec_t *dec)
… …
609 { 883 {
610 ++dec->cur; 884 ++dec->cur;
611 if (*dec->cur >= '0' && *dec->cur <= '9') 885 if (*dec->cur >= '0' && *dec->cur <= '9')
612 ERR ("malformed number (leading zero must not be followed by another digit)"); 886 ERR ("malformed number (leading zero must not be followed by another digit)");
613 } 887 }
614 888 else if (*dec->cur < '0' || *dec->cur > '9')
615 // int 889 ERR ("malformed number (no digits after initial minus)");
890 else
891 do
892 {
893 ++dec->cur;
894 }
616 while (*dec->cur >= '0' && *dec->cur <= '9') 895 while (*dec->cur >= '0' && *dec->cur <= '9');
617 ++dec->cur;
618 896
619 // [frac] 897 // [frac]
620 if (*dec->cur == '.') 898 if (*dec->cur == '.')
621 { 899 {
622 is_nv = 1; 900 ++dec->cur;
901
902 if (*dec->cur < '0' || *dec->cur > '9')
903 ERR ("malformed number (no digits after decimal point)");
623 904
624 do 905 do
625 { 906 {
626 ++dec->cur; 907 ++dec->cur;
627 } 908 }
628 while (*dec->cur >= '0' && *dec->cur <= '9'); 909 while (*dec->cur >= '0' && *dec->cur <= '9');
910
911 is_nv = 1;
629 } 912 }
630 913
631 // [exp] 914 // [exp]
632 if (*dec->cur == 'e' || *dec->cur == 'E') 915 if (*dec->cur == 'e' || *dec->cur == 'E')
633 { 916 {
634 is_nv = 1;
635
636 ++dec->cur; 917 ++dec->cur;
918
637 if (*dec->cur == '-' || *dec->cur == '+') 919 if (*dec->cur == '-' || *dec->cur == '+')
638 ++dec->cur; 920 ++dec->cur;
639 921
922 if (*dec->cur < '0' || *dec->cur > '9')
923 ERR ("malformed number (no digits after exp sign)");
924
925 do
926 {
927 ++dec->cur;
928 }
640 while (*dec->cur >= '0' && *dec->cur <= '9') 929 while (*dec->cur >= '0' && *dec->cur <= '9');
641 ++dec->cur; 930
931 is_nv = 1;
642 } 932 }
643 933
644 if (!is_nv) 934 if (!is_nv)
645 { 935 {
646 UV uv; 936 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
647 int numtype = grok_number (start, dec->cur - start, &uv); 937 if (*start == '-')
648 if (numtype & IS_NUMBER_IN_UV) 938 switch (dec->cur - start)
649 if (numtype & IS_NUMBER_NEG)
650 { 939 {
651 if (uv < (UV)IV_MIN) 940 case 2: return newSViv (-( start [1] - '0' * 1));
652 return newSViv (-(IV)uv); 941 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
942 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
943 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
653 } 944 }
945 else
946 switch (dec->cur - start)
947 {
948 case 1: return newSViv ( start [0] - '0' * 1);
949 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
950 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
951 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
952 }
953
954 {
955 UV uv;
956 int numtype = grok_number (start, dec->cur - start, &uv);
957 if (numtype & IS_NUMBER_IN_UV)
958 if (numtype & IS_NUMBER_NEG)
959 {
960 if (uv < (UV)IV_MIN)
961 return newSViv (-(IV)uv);
962 }
654 else 963 else
655 return newSVuv (uv); 964 return newSVuv (uv);
965
966 // here would likely be the place for bigint support
656 } 967 }
968 }
657 969
970 // if we ever support bigint or bigfloat, this is the place for bigfloat
658 return newSVnv (Atof (start)); 971 return newSVnv (Atof (start));
659 972
660fail: 973fail:
661 return 0; 974 return 0;
662} 975}
… …
664static SV * 977static SV *
665decode_av (dec_t *dec) 978decode_av (dec_t *dec)
666{ 979{
667 AV *av = newAV (); 980 AV *av = newAV ();
668 981
982 DEC_INC_DEPTH;
983 decode_ws (dec);
984
985 if (*dec->cur == ']')
986 ++dec->cur;
987 else
669 for (;;) 988 for (;;)
670 { 989 {
671 SV *value; 990 SV *value;
672 991
673 value = decode_sv (dec); 992 value = decode_sv (dec);
674 if (!value) 993 if (!value)
675 goto fail; 994 goto fail;
676 995
677 av_push (av, value); 996 av_push (av, value);
678 997
679 WS; 998 decode_ws (dec);
680 999
681 if (*dec->cur == ']') 1000 if (*dec->cur == ']')
682 { 1001 {
683 ++dec->cur; 1002 ++dec->cur;
684 break; 1003 break;
1004 }
685 } 1005
686
687 if (*dec->cur != ',') 1006 if (*dec->cur != ',')
688 ERR (", or ] expected while parsing array"); 1007 ERR (", or ] expected while parsing array");
689 1008
690 ++dec->cur; 1009 ++dec->cur;
691 } 1010 }
692 1011
1012 DEC_DEC_DEPTH;
693 return newRV_noinc ((SV *)av); 1013 return newRV_noinc ((SV *)av);
694 1014
695fail: 1015fail:
696 SvREFCNT_dec (av); 1016 SvREFCNT_dec (av);
1017 DEC_DEC_DEPTH;
697 return 0; 1018 return 0;
698} 1019}
699 1020
700static SV * 1021static SV *
701decode_hv (dec_t *dec) 1022decode_hv (dec_t *dec)
702{ 1023{
1024 SV *sv;
703 HV *hv = newHV (); 1025 HV *hv = newHV ();
704 1026
1027 DEC_INC_DEPTH;
1028 decode_ws (dec);
1029
1030 if (*dec->cur == '}')
1031 ++dec->cur;
1032 else
705 for (;;) 1033 for (;;)
706 { 1034 {
707 SV *key, *value; 1035 decode_ws (dec); EXPECT_CH ('"');
708 1036
709 WS; EXPECT_CH ('"'); 1037 // heuristic: assume that
710 1038 // a) decode_str + hv_store_ent are abysmally slow.
711 key = decode_str (dec); 1039 // b) most hash keys are short, simple ascii text.
712 if (!key) 1040 // => try to "fast-match" such strings to avoid
713 goto fail; 1041 // the overhead of decode_str + hv_store_ent.
714
715 WS; EXPECT_CH (':');
716
717 value = decode_sv (dec);
718 if (!value)
719 { 1042 {
1043 SV *value;
1044 char *p = dec->cur;
1045 char *e = p + 24; // only try up to 24 bytes
1046
1047 for (;;)
1048 {
1049 // the >= 0x80 is true on most architectures
1050 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1051 {
1052 // slow path, back up and use decode_str
1053 SV *key = decode_str (dec);
1054 if (!key)
1055 goto fail;
1056
1057 decode_ws (dec); EXPECT_CH (':');
1058
1059 value = decode_sv (dec);
1060 if (!value)
1061 {
1062 SvREFCNT_dec (key);
1063 goto fail;
1064 }
1065
1066 hv_store_ent (hv, key, value, 0);
720 SvREFCNT_dec (key); 1067 SvREFCNT_dec (key);
1068
1069 break;
1070 }
1071 else if (*p == '"')
1072 {
1073 // fast path, got a simple key
1074 char *key = dec->cur;
1075 int len = p - key;
1076 dec->cur = p + 1;
1077
1078 decode_ws (dec); EXPECT_CH (':');
1079
1080 value = decode_sv (dec);
1081 if (!value)
721 goto fail; 1082 goto fail;
1083
1084 hv_store (hv, key, len, value, 0);
1085
1086 break;
1087 }
1088
1089 ++p;
1090 }
722 } 1091 }
723 1092
724 //TODO: optimise 1093 decode_ws (dec);
725 hv_store_ent (hv, key, value, 0);
726 1094
727 WS;
728
729 if (*dec->cur == '}') 1095 if (*dec->cur == '}')
1096 {
1097 ++dec->cur;
1098 break;
1099 }
1100
1101 if (*dec->cur != ',')
1102 ERR (", or } expected while parsing object/hash");
1103
1104 ++dec->cur;
1105 }
1106
1107 DEC_DEC_DEPTH;
1108 sv = newRV_noinc ((SV *)hv);
1109
1110 // check filter callbacks
1111 if (dec->json.flags & F_HOOK)
1112 {
1113 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
730 { 1114 {
731 ++dec->cur; 1115 HE *cb, *he;
732 break; 1116
1117 hv_iterinit (hv);
1118 he = hv_iternext (hv);
1119 hv_iterinit (hv);
1120
1121 // the next line creates a mortal sv each time its called.
1122 // might want to optimise this for common cases.
1123 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1124
1125 if (cb)
1126 {
1127 dSP;
1128 int count;
1129
1130 ENTER; SAVETMPS; PUSHMARK (SP);
1131 XPUSHs (HeVAL (he));
1132
1133 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1134
1135 if (count == 1)
1136 {
1137 sv = newSVsv (POPs);
1138 FREETMPS; LEAVE;
1139 return sv;
1140 }
1141
1142 FREETMPS; LEAVE;
1143 }
733 } 1144 }
734 1145
735 if (*dec->cur != ',') 1146 if (dec->json.cb_object)
736 ERR (", or } expected while parsing object/hash"); 1147 {
1148 dSP;
1149 int count;
737 1150
738 ++dec->cur; 1151 ENTER; SAVETMPS; PUSHMARK (SP);
739 } 1152 XPUSHs (sv_2mortal (sv));
740 1153
741 return newRV_noinc ((SV *)hv); 1154 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1155
1156 if (count == 1)
1157 {
1158 sv = newSVsv (POPs);
1159 FREETMPS; LEAVE;
1160 return sv;
1161 }
1162
1163 SvREFCNT_inc (sv);
1164 FREETMPS; LEAVE;
1165 }
1166 }
1167
1168 return sv;
742 1169
743fail: 1170fail:
744 SvREFCNT_dec (hv); 1171 SvREFCNT_dec (hv);
1172 DEC_DEC_DEPTH;
745 return 0; 1173 return 0;
746} 1174}
747 1175
748static SV * 1176static SV *
749decode_sv (dec_t *dec) 1177decode_sv (dec_t *dec)
750{ 1178{
751 WS; 1179 decode_ws (dec);
1180
1181 // the beauty of JSON: you need exactly one character lookahead
1182 // to parse anything.
752 switch (*dec->cur) 1183 switch (*dec->cur)
753 { 1184 {
754 case '"': ++dec->cur; return decode_str (dec); 1185 case '"': ++dec->cur; return decode_str (dec);
755 case '[': ++dec->cur; return decode_av (dec); 1186 case '[': ++dec->cur; return decode_av (dec);
756 case '{': ++dec->cur; return decode_hv (dec); 1187 case '{': ++dec->cur; return decode_hv (dec);
… …
762 1193
763 case 't': 1194 case 't':
764 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1195 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
765 { 1196 {
766 dec->cur += 4; 1197 dec->cur += 4;
767 return newSViv (1); 1198 return SvREFCNT_inc (json_true);
768 } 1199 }
769 else 1200 else
770 ERR ("'true' expected"); 1201 ERR ("'true' expected");
771 1202
772 break; 1203 break;
773 1204
774 case 'f': 1205 case 'f':
775 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1206 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
776 { 1207 {
777 dec->cur += 5; 1208 dec->cur += 5;
778 return newSViv (0); 1209 return SvREFCNT_inc (json_false);
779 } 1210 }
780 else 1211 else
781 ERR ("'false' expected"); 1212 ERR ("'false' expected");
782 1213
783 break; 1214 break;
784 1215
785 case 'n': 1216 case 'n':
786 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1217 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
787 { 1218 {
788 dec->cur += 4; 1219 dec->cur += 4;
789 return newSViv (1); 1220 return newSVsv (&PL_sv_undef);
790 } 1221 }
791 else 1222 else
792 ERR ("'null' expected"); 1223 ERR ("'null' expected");
793 1224
794 break; 1225 break;
795 1226
796 default: 1227 default:
797 ERR ("malformed json string"); 1228 ERR ("malformed JSON string, neither array, object, number, string or atom");
798 break; 1229 break;
799 } 1230 }
800 1231
801fail: 1232fail:
802 return 0; 1233 return 0;
803} 1234}
804 1235
805static SV * 1236static SV *
806decode_json (SV *string, UV flags) 1237decode_json (SV *string, JSON *json, UV *offset_return)
807{ 1238{
1239 dec_t dec;
1240 UV offset;
808 SV *sv; 1241 SV *sv;
809 1242
1243 SvGETMAGIC (string);
1244 SvUPGRADE (string, SVt_PV);
1245
1246 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1247 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1248 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1249
810 if (!(flags & F_UTF8)) 1250 if (json->flags & F_UTF8)
1251 sv_utf8_downgrade (string, 0);
1252 else
811 sv_utf8_upgrade (string); 1253 sv_utf8_upgrade (string);
812 1254
813 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1255 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
814 1256
815 dec_t dec; 1257 dec.json = *json;
816 dec.flags = flags;
817 dec.cur = SvPVX (string); 1258 dec.cur = SvPVX (string);
818 dec.end = SvEND (string); 1259 dec.end = SvEND (string);
819 dec.err = 0; 1260 dec.err = 0;
1261 dec.depth = 0;
1262 dec.maxdepth = DEC_DEPTH (dec.json.flags);
820 1263
821 *dec.end = 1; // invalid anywhere 1264 if (dec.json.cb_object || dec.json.cb_sk_object)
1265 dec.json.flags |= F_HOOK;
1266
1267 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
822 sv = decode_sv (&dec); 1268 sv = decode_sv (&dec);
823 *dec.end = 0; 1269
1270 if (!(offset_return || !sv))
1271 {
1272 // check for trailing garbage
1273 decode_ws (&dec);
1274
1275 if (*dec.cur)
1276 {
1277 dec.err = "garbage after JSON object";
1278 SvREFCNT_dec (sv);
1279 sv = 0;
1280 }
1281 }
1282
1283 if (offset_return || !sv)
1284 {
1285 offset = dec.json.flags & F_UTF8
1286 ? dec.cur - SvPVX (string)
1287 : utf8_distance (dec.cur, SvPVX (string));
1288
1289 if (offset_return)
1290 *offset_return = offset;
1291 }
824 1292
825 if (!sv) 1293 if (!sv)
826 { 1294 {
827 IV offset = utf8_distance (dec.cur, SvPVX (string));
828 SV *uni = sv_newmortal (); 1295 SV *uni = sv_newmortal ();
829 1296
1297 // horrible hack to silence warning inside pv_uni_display
1298 COP cop = *PL_curcop;
1299 cop.cop_warnings = pWARN_NONE;
1300 ENTER;
1301 SAVEVPTR (PL_curcop);
1302 PL_curcop = &cop;
830 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1303 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1304 LEAVE;
1305
831 croak ("%s, at character %d (%s)", 1306 croak ("%s, at character offset %d [\"%s\"]",
832 dec.err, 1307 dec.err,
833 (int)offset, 1308 (int)offset,
834 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1309 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
835 } 1310 }
836 1311
837 sv = sv_2mortal (sv); 1312 sv = sv_2mortal (sv);
838 1313
839 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1314 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
840 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1315 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
841 1316
842 return sv; 1317 return sv;
843} 1318}
844 1319
1320/////////////////////////////////////////////////////////////////////////////
1321// XS interface functions
1322
845MODULE = JSON::XS PACKAGE = JSON::XS 1323MODULE = JSON::XS PACKAGE = JSON::XS
846 1324
847BOOT: 1325BOOT:
848{ 1326{
849 int i; 1327 int i;
850 1328
851 memset (decode_hexdigit, 0xff, 256);
852 for (i = 10; i--; ) 1329 for (i = 0; i < 256; ++i)
853 decode_hexdigit ['0' + i] = i; 1330 decode_hexdigit [i] =
1331 i >= '0' && i <= '9' ? i - '0'
1332 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1333 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1334 : -1;
854 1335
855 for (i = 7; i--; ) 1336 json_stash = gv_stashpv ("JSON::XS" , 1);
1337 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1338
1339 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1340 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1341}
1342
1343PROTOTYPES: DISABLE
1344
1345void new (char *klass)
1346 PPCODE:
1347{
1348 SV *pv = NEWSV (0, sizeof (JSON));
1349 SvPOK_only (pv);
1350 Zero (SvPVX (pv), 1, JSON);
1351 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1352 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1353}
1354
1355void ascii (JSON *self, int enable = 1)
1356 ALIAS:
1357 ascii = F_ASCII
1358 latin1 = F_LATIN1
1359 utf8 = F_UTF8
1360 indent = F_INDENT
1361 canonical = F_CANONICAL
1362 space_before = F_SPACE_BEFORE
1363 space_after = F_SPACE_AFTER
1364 pretty = F_PRETTY
1365 allow_nonref = F_ALLOW_NONREF
1366 shrink = F_SHRINK
1367 allow_blessed = F_ALLOW_BLESSED
1368 convert_blessed = F_CONV_BLESSED
1369 PPCODE:
1370{
1371 if (enable)
1372 self->flags |= ix;
1373 else
1374 self->flags &= ~ix;
1375
1376 XPUSHs (ST (0));
1377}
1378
1379void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1380 PPCODE:
1381{
1382 UV log2 = 0;
1383
1384 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1385
1386 while ((1UL << log2) < max_depth)
1387 ++log2;
1388
1389 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1390
1391 XPUSHs (ST (0));
1392}
1393
1394void max_size (JSON *self, UV max_size = 0)
1395 PPCODE:
1396{
1397 UV log2 = 0;
1398
1399 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1400 if (max_size == 1) max_size = 2;
1401
1402 while ((1UL << log2) < max_size)
1403 ++log2;
1404
1405 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1406
1407 XPUSHs (ST (0));
1408}
1409
1410void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1411 PPCODE:
1412{
1413 SvREFCNT_dec (self->cb_object);
1414 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1415
1416 XPUSHs (ST (0));
1417}
1418
1419void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1420 PPCODE:
1421{
1422 if (!self->cb_sk_object)
1423 self->cb_sk_object = newHV ();
1424
1425 if (SvOK (cb))
1426 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1427 else
856 { 1428 {
857 decode_hexdigit ['a' + i] = 10 + i; 1429 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
858 decode_hexdigit ['A' + i] = 10 + i; 1430
1431 if (!HvKEYS (self->cb_sk_object))
1432 {
1433 SvREFCNT_dec (self->cb_sk_object);
1434 self->cb_sk_object = 0;
1435 }
859 } 1436 }
860 1437
861 json_stash = gv_stashpv ("JSON::XS", 1); 1438 XPUSHs (ST (0));
862} 1439}
863 1440
864PROTOTYPES: DISABLE 1441void encode (JSON *self, SV *scalar)
1442 PPCODE:
1443 XPUSHs (encode_json (scalar, self));
865 1444
866SV *new (char *dummy) 1445void decode (JSON *self, SV *jsonstr)
1446 PPCODE:
1447 XPUSHs (decode_json (jsonstr, self, 0));
1448
1449void decode_prefix (JSON *self, SV *jsonstr)
1450 PPCODE:
1451{
1452 UV offset;
1453 EXTEND (SP, 2);
1454 PUSHs (decode_json (jsonstr, self, &offset));
1455 PUSHs (sv_2mortal (newSVuv (offset)));
1456}
1457
1458void DESTROY (JSON *self)
867 CODE: 1459 CODE:
868 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1460 SvREFCNT_dec (self->cb_sk_object);
869 OUTPUT: 1461 SvREFCNT_dec (self->cb_object);
870 RETVAL
871
872SV *ascii (SV *self, int enable)
873 ALIAS:
874 ascii = F_ASCII
875 utf8 = F_UTF8
876 indent = F_INDENT
877 canonical = F_CANONICAL
878 space_before = F_SPACE_BEFORE
879 space_after = F_SPACE_AFTER
880 json_rpc = F_JSON_RPC
881 pretty = F_PRETTY
882 allow_nonref = F_ALLOW_NONREF
883 CODE:
884{
885 UV *uv = SvJSON (self);
886 if (enable)
887 *uv |= ix;
888 else
889 *uv &= ~ix;
890
891 RETVAL = newSVsv (self);
892}
893 OUTPUT:
894 RETVAL
895
896void encode (SV *self, SV *scalar)
897 PPCODE:
898 XPUSHs (encode_json (scalar, *SvJSON (self)));
899
900void decode (SV *self, SV *jsonstr)
901 PPCODE:
902 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
903 1462
904PROTOTYPES: ENABLE 1463PROTOTYPES: ENABLE
905 1464
906void to_json (SV *scalar) 1465void to_json (SV *scalar)
907 PPCODE: 1466 PPCODE:
1467{
1468 JSON json = { F_DEFAULT | F_UTF8 };
908 XPUSHs (encode_json (scalar, F_UTF8)); 1469 XPUSHs (encode_json (scalar, &json));
1470}
909 1471
910void from_json (SV *jsonstr) 1472void from_json (SV *jsonstr)
911 PPCODE: 1473 PPCODE:
1474{
1475 JSON json = { F_DEFAULT | F_UTF8 };
912 XPUSHs (decode_json (jsonstr, F_UTF8)); 1476 XPUSHs (decode_json (jsonstr, &json, 0));
1477}
913 1478

Diff Legend

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