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 1.18 by root, Sun Mar 25 21:19:13 2007 UTC vs.
Revision 1.62 by root, Sun Aug 26 22:27:32 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include <stdio.h>
9#include <float.h>
8 10
11#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h
13#endif
14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
9#define F_ASCII 0x00000001UL 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002UL 23#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004UL 24#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008UL 25#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010UL 26#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020UL 27#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080UL 28#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100UL 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
17#define F_MAXDEPTH 0xf8000000UL 32#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27 33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
19 37
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24 40
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (13UL << S_MAXDEPTH) 42#define F_DEFAULT (9UL << S_MAXDEPTH)
27 43
28#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 45#define INDENT_STEP 3 // spaces per indentation level
30 46
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size 47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
33 48
34#define SB do { 49#define SB do {
35#define SE } while (0) 50#define SE } while (0)
36 51
52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value))
54# define inline inline
55#else
56# define expect(expr,value) (expr)
57# define inline static
58#endif
59
60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#ifdef USE_ITHREADS
64# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else
67# define JSON_SLOW 0
68# define JSON_STASH json_stash
69#endif
70
37static HV *json_stash; // JSON::XS:: 71static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false;
73
74typedef struct {
75 U32 flags;
76 SV *cb_object;
77 HV *cb_sk_object;
78} JSON;
38 79
39///////////////////////////////////////////////////////////////////////////// 80/////////////////////////////////////////////////////////////////////////////
40// utility functions 81// utility functions
41 82
42static UV * 83inline void
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv) 84shrink (SV *sv)
53{ 85{
54 sv_utf8_downgrade (sv, 1); 86 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1) 87 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 88 {
… …
65// decode an utf-8 character and return it, or (UV)-1 in 97// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error. 98// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 99// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 100// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 101// note that we never call this function for a ascii codepoints
70static UV 102inline UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 104{
73 if (s[0] > 0xdf || s[0] < 0xc2) 105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 108 {
77 *clen = 2; 109 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 111 }
80 else 112 else
113 {
114 *clen = (STRLEN)-1;
81 return (UV)-1; 115 return (UV)-1;
116 }
82} 117}
83 118
84///////////////////////////////////////////////////////////////////////////// 119/////////////////////////////////////////////////////////////////////////////
85// encoder 120// encoder
86 121
… …
88typedef struct 123typedef struct
89{ 124{
90 char *cur; // SvPVX (sv) + current output position 125 char *cur; // SvPVX (sv) + current output position
91 char *end; // SvEND (sv) 126 char *end; // SvEND (sv)
92 SV *sv; // result scalar 127 SV *sv; // result scalar
93 U32 flags; // F_* 128 JSON json;
94 U32 indent; // indentation level 129 U32 indent; // indentation level
95 U32 maxdepth; // max. indentation/recursion level 130 U32 maxdepth; // max. indentation/recursion level
96} enc_t; 131} enc_t;
97 132
98static void 133inline void
99need (enc_t *enc, STRLEN len) 134need (enc_t *enc, STRLEN len)
100{ 135{
101 if (enc->cur + len >= enc->end) 136 if (expect_false (enc->cur + len >= enc->end))
102 { 137 {
103 STRLEN cur = enc->cur - SvPVX (enc->sv); 138 STRLEN cur = enc->cur - SvPVX (enc->sv);
104 SvGROW (enc->sv, cur + len + 1); 139 SvGROW (enc->sv, cur + len + 1);
105 enc->cur = SvPVX (enc->sv) + cur; 140 enc->cur = SvPVX (enc->sv) + cur;
106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107 } 142 }
108} 143}
109 144
110static void 145inline void
111encode_ch (enc_t *enc, char ch) 146encode_ch (enc_t *enc, char ch)
112{ 147{
113 need (enc, 1); 148 need (enc, 1);
114 *enc->cur++ = ch; 149 *enc->cur++ = ch;
115} 150}
… …
123 158
124 while (str < end) 159 while (str < end)
125 { 160 {
126 unsigned char ch = *(unsigned char *)str; 161 unsigned char ch = *(unsigned char *)str;
127 162
128 if (ch >= 0x20 && ch < 0x80) // most common case 163 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129 { 164 {
130 if (ch == '"') // but with slow exceptions 165 if (expect_false (ch == '"')) // but with slow exceptions
131 { 166 {
132 need (enc, len += 1); 167 need (enc, len += 1);
133 *enc->cur++ = '\\'; 168 *enc->cur++ = '\\';
134 *enc->cur++ = '"'; 169 *enc->cur++ = '"';
135 } 170 }
136 else if (ch == '\\') 171 else if (expect_false (ch == '\\'))
137 { 172 {
138 need (enc, len += 1); 173 need (enc, len += 1);
139 *enc->cur++ = '\\'; 174 *enc->cur++ = '\\';
140 *enc->cur++ = '\\'; 175 *enc->cur++ = '\\';
141 } 176 }
… …
159 STRLEN clen; 194 STRLEN clen;
160 UV uch; 195 UV uch;
161 196
162 if (is_utf8) 197 if (is_utf8)
163 { 198 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen); 199 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1) 200 if (clen == (STRLEN)-1)
167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 201 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
168 } 202 }
169 else 203 else
… …
173 } 207 }
174 208
175 if (uch > 0x10FFFFUL) 209 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177 211
178 if (uch < 0x80 || enc->flags & F_ASCII) 212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
179 { 213 {
180 if (uch > 0xFFFFUL) 214 if (uch > 0xFFFFUL)
181 { 215 {
182 need (enc, len += 11); 216 need (enc, len += 11);
183 sprintf (enc->cur, "\\u%04x\\u%04x", 217 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
197 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 231 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 } 232 }
199 233
200 str += clen; 234 str += clen;
201 } 235 }
236 else if (enc->json.flags & F_LATIN1)
237 {
238 *enc->cur++ = uch;
239 str += clen;
240 }
202 else if (is_utf8) 241 else if (is_utf8)
203 { 242 {
204 need (enc, len += clen); 243 need (enc, len += clen);
205 do 244 do
206 { 245 {
… …
208 } 247 }
209 while (--clen); 248 while (--clen);
210 } 249 }
211 else 250 else
212 { 251 {
213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 252 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str; 254 ++str;
216 } 255 }
217 } 256 }
218 } 257 }
… …
220 259
221 --len; 260 --len;
222 } 261 }
223} 262}
224 263
225static void 264inline void
226encode_indent (enc_t *enc) 265encode_indent (enc_t *enc)
227{ 266{
228 if (enc->flags & F_INDENT) 267 if (enc->json.flags & F_INDENT)
229 { 268 {
230 int spaces = enc->indent * INDENT_STEP; 269 int spaces = enc->indent * INDENT_STEP;
231 270
232 need (enc, spaces); 271 need (enc, spaces);
233 memset (enc->cur, ' ', spaces); 272 memset (enc->cur, ' ', spaces);
234 enc->cur += spaces; 273 enc->cur += spaces;
235 } 274 }
236} 275}
237 276
238static void 277inline void
239encode_space (enc_t *enc) 278encode_space (enc_t *enc)
240{ 279{
241 need (enc, 1); 280 need (enc, 1);
242 encode_ch (enc, ' '); 281 encode_ch (enc, ' ');
243} 282}
244 283
245static void 284inline void
246encode_nl (enc_t *enc) 285encode_nl (enc_t *enc)
247{ 286{
248 if (enc->flags & F_INDENT) 287 if (enc->json.flags & F_INDENT)
249 { 288 {
250 need (enc, 1); 289 need (enc, 1);
251 encode_ch (enc, '\n'); 290 encode_ch (enc, '\n');
252 } 291 }
253} 292}
254 293
255static void 294inline void
256encode_comma (enc_t *enc) 295encode_comma (enc_t *enc)
257{ 296{
258 encode_ch (enc, ','); 297 encode_ch (enc, ',');
259 298
260 if (enc->flags & F_INDENT) 299 if (enc->json.flags & F_INDENT)
261 encode_nl (enc); 300 encode_nl (enc);
262 else if (enc->flags & F_SPACE_AFTER) 301 else if (enc->json.flags & F_SPACE_AFTER)
263 encode_space (enc); 302 encode_space (enc);
264} 303}
265 304
266static void encode_sv (enc_t *enc, SV *sv); 305static void encode_sv (enc_t *enc, SV *sv);
267 306
268static void 307static void
269encode_av (enc_t *enc, AV *av) 308encode_av (enc_t *enc, AV *av)
270{ 309{
271 int i, len = av_len (av); 310 int i, len = av_len (av);
272 311
312 if (enc->indent >= enc->maxdepth)
313 croak ("data structure too deep (hit recursion limit)");
314
273 encode_ch (enc, '['); encode_nl (enc); 315 encode_ch (enc, '['); encode_nl (enc);
274 ++enc->indent; 316 ++enc->indent;
275 317
276 for (i = 0; i <= len; ++i) 318 for (i = 0; i <= len; ++i)
277 { 319 {
320 SV **svp = av_fetch (av, i, 0);
321
278 encode_indent (enc); 322 encode_indent (enc);
279 encode_sv (enc, *av_fetch (av, i, 0)); 323
324 if (svp)
325 encode_sv (enc, *svp);
326 else
327 encode_str (enc, "null", 4, 0);
280 328
281 if (i < len) 329 if (i < len)
282 encode_comma (enc); 330 encode_comma (enc);
283 } 331 }
284 332
… …
287 --enc->indent; 335 --enc->indent;
288 encode_indent (enc); encode_ch (enc, ']'); 336 encode_indent (enc); encode_ch (enc, ']');
289} 337}
290 338
291static void 339static void
292encode_he (enc_t *enc, HE *he) 340encode_hk (enc_t *enc, HE *he)
293{ 341{
294 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
295 343
296 if (HeKLEN (he) == HEf_SVKEY) 344 if (HeKLEN (he) == HEf_SVKEY)
297 { 345 {
… …
307 else 355 else
308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 356 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
309 357
310 encode_ch (enc, '"'); 358 encode_ch (enc, '"');
311 359
312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 360 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
313 encode_ch (enc, ':'); 361 encode_ch (enc, ':');
314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 362 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
315 encode_sv (enc, HeVAL (he));
316} 363}
317 364
318// compare hash entries, used when all keys are bytestrings 365// compare hash entries, used when all keys are bytestrings
319static int 366static int
320he_cmp_fast (const void *a_, const void *b_) 367he_cmp_fast (const void *a_, const void *b_)
… …
325 HE *b = *(HE **)b_; 372 HE *b = *(HE **)b_;
326 373
327 STRLEN la = HeKLEN (a); 374 STRLEN la = HeKLEN (a);
328 STRLEN lb = HeKLEN (b); 375 STRLEN lb = HeKLEN (b);
329 376
330 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 377 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
331 cmp = la - lb; 378 cmp = lb - la;
332 379
333 return cmp; 380 return cmp;
334} 381}
335 382
336// compare hash entries, used when some keys are sv's or utf-x 383// compare hash entries, used when some keys are sv's or utf-x
337static int 384static int
338he_cmp_slow (const void *a, const void *b) 385he_cmp_slow (const void *a, const void *b)
339{ 386{
340 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 387 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
341} 388}
342 389
343static void 390static void
344encode_hv (enc_t *enc, HV *hv) 391encode_hv (enc_t *enc, HV *hv)
345{ 392{
393 HE *he;
346 int count, i; 394 int count;
395
396 if (enc->indent >= enc->maxdepth)
397 croak ("data structure too deep (hit recursion limit)");
347 398
348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
349 400
350 if ((count = hv_iterinit (hv)))
351 {
352 // for canonical output we have to sort by keys first 401 // for canonical output we have to sort by keys first
353 // actually, this is mostly due to the stupid so-called 402 // actually, this is mostly due to the stupid so-called
354 // security workaround added somewhere in 5.8.x. 403 // security workaround added somewhere in 5.8.x.
355 // that randomises hash orderings 404 // that randomises hash orderings
356 if (enc->flags & F_CANONICAL) 405 if (enc->json.flags & F_CANONICAL)
406 {
407 int count = hv_iterinit (hv);
408
409 if (SvMAGICAL (hv))
357 { 410 {
411 // need to count by iterating. could improve by dynamically building the vector below
412 // but I don't care for the speed of this special case.
413 // note also that we will run into undefined behaviour when the two iterations
414 // do not result in the same count, something I might care for in some later release.
415
416 count = 0;
417 while (hv_iternext (hv))
418 ++count;
419
420 hv_iterinit (hv);
421 }
422
423 if (count)
424 {
425 int i, fast = 1;
426#if defined(__BORLANDC__) || defined(_MSC_VER)
427 HE **hes = _alloca (count * sizeof (HE));
428#else
358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 429 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
359 int fast = 1; 430#endif
360 431
361 i = 0; 432 i = 0;
362 while ((he = hv_iternext (hv))) 433 while ((he = hv_iternext (hv)))
363 { 434 {
364 hes [i++] = he; 435 hes [i++] = he;
… …
386 457
387 FREETMPS; 458 FREETMPS;
388 LEAVE; 459 LEAVE;
389 } 460 }
390 461
391 for (i = 0; i < count; ++i) 462 while (count--)
392 { 463 {
393 encode_indent (enc); 464 encode_indent (enc);
465 he = hes [count];
394 encode_he (enc, hes [i]); 466 encode_hk (enc, he);
467 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
395 468
396 if (i < count - 1) 469 if (count)
397 encode_comma (enc); 470 encode_comma (enc);
398 } 471 }
399 472 }
473 }
474 else
475 {
476 if (hv_iterinit (hv) || SvMAGICAL (hv))
477 if ((he = hv_iternext (hv)))
478 for (;;)
479 {
400 encode_nl (enc); 480 encode_indent (enc);
481 encode_hk (enc, he);
482 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
483
484 if (!(he = hv_iternext (hv)))
485 break;
486
487 encode_comma (enc);
488 }
489 }
490
491 encode_nl (enc);
492
493 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
494}
495
496// encode objects, arrays and special \0=false and \1=true values.
497static void
498encode_rv (enc_t *enc, SV *sv)
499{
500 svtype svt;
501
502 SvGETMAGIC (sv);
503 svt = SvTYPE (sv);
504
505 if (expect_false (SvOBJECT (sv)))
506 {
507 HV *stash = !JSON_SLOW || json_boolean_stash
508 ? json_boolean_stash
509 : gv_stashpv ("JSON::XS::Boolean", 1);
510
511 if (SvSTASH (sv) == stash)
512 {
513 if (SvIV (sv))
514 encode_str (enc, "true", 4, 0);
515 else
516 encode_str (enc, "false", 5, 0);
401 } 517 }
402 else 518 else
403 { 519 {
404 SV *sv; 520#if 0
405 HE *he = hv_iternext (hv); 521 if (0 && sv_derived_from (rv, "JSON::Literal"))
406
407 for (;;)
408 {
409 encode_indent (enc);
410 encode_he (enc, he);
411
412 if (!(he = hv_iternext (hv)))
413 break;
414
415 encode_comma (enc);
416 } 522 {
523 // not yet
524 }
525#endif
526 if (enc->json.flags & F_CONV_BLESSED)
527 {
528 // we re-bless the reference to get overload and other niceties right
529 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
417 530
531 if (to_json)
532 {
533 dSP;
534
535 ENTER; SAVETMPS; PUSHMARK (SP);
536 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
537
538 // calling with G_SCALAR ensures that we always get a 1 return value
539 PUTBACK;
540 call_sv ((SV *)GvCV (to_json), G_SCALAR);
541 SPAGAIN;
542
543 // catch this surprisingly common error
544 if (SvROK (TOPs) && SvRV (TOPs) == sv)
545 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
546
547 sv = POPs;
548 PUTBACK;
549
418 encode_nl (enc); 550 encode_sv (enc, sv);
551
552 FREETMPS; LEAVE;
553 }
554 else if (enc->json.flags & F_ALLOW_BLESSED)
555 encode_str (enc, "null", 4, 0);
556 else
557 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
558 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
559 }
560 else if (enc->json.flags & F_ALLOW_BLESSED)
561 encode_str (enc, "null", 4, 0);
562 else
563 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
564 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
419 } 565 }
420 } 566 }
567 else if (svt == SVt_PVHV)
568 encode_hv (enc, (HV *)sv);
569 else if (svt == SVt_PVAV)
570 encode_av (enc, (AV *)sv);
571 else if (svt < SVt_PVAV)
572 {
573 STRLEN len = 0;
574 char *pv = svt ? SvPV (sv, len) : 0;
421 575
422 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 576 if (len == 1 && *pv == '1')
577 encode_str (enc, "true", 4, 0);
578 else if (len == 1 && *pv == '0')
579 encode_str (enc, "false", 5, 0);
580 else
581 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
582 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
583 }
584 else
585 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
586 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
423} 587}
424 588
425static void 589static void
426encode_sv (enc_t *enc, SV *sv) 590encode_sv (enc_t *enc, SV *sv)
427{ 591{
… …
435 encode_str (enc, str, len, SvUTF8 (sv)); 599 encode_str (enc, str, len, SvUTF8 (sv));
436 encode_ch (enc, '"'); 600 encode_ch (enc, '"');
437 } 601 }
438 else if (SvNOKp (sv)) 602 else if (SvNOKp (sv))
439 { 603 {
604 // trust that perl will do the right thing w.r.t. JSON syntax.
440 need (enc, NV_DIG + 32); 605 need (enc, NV_DIG + 32);
441 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 606 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
442 enc->cur += strlen (enc->cur); 607 enc->cur += strlen (enc->cur);
443 } 608 }
444 else if (SvIOKp (sv)) 609 else if (SvIOKp (sv))
445 { 610 {
446 need (enc, 64); 611 // we assume we can always read an IV as a UV
612 if (SvUV (sv) & ~(UV)0x7fff)
613 {
614 // large integer, use the (rather slow) snprintf way.
615 need (enc, sizeof (UV) * 3);
447 enc->cur += 616 enc->cur +=
448 SvIsUV(sv) 617 SvIsUV(sv)
449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 618 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 619 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
620 }
621 else
622 {
623 // optimise the "small number case"
624 // code will likely be branchless and use only a single multiplication
625 I32 i = SvIV (sv);
626 U32 u;
627 char digit, nz = 0;
628
629 need (enc, 6);
630
631 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
632 u = i < 0 ? -i : i;
633
634 // convert to 4.28 fixed-point representation
635 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
636
637 // now output digit by digit, each time masking out the integer part
638 // and multiplying by 5 while moving the decimal point one to the right,
639 // resulting in a net multiplication by 10.
640 // we always write the digit to memory but conditionally increment
641 // the pointer, to ease the usage of conditional move instructions.
642 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
643 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
644 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
645 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
646 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
647 }
451 } 648 }
452 else if (SvROK (sv)) 649 else if (SvROK (sv))
453 { 650 encode_rv (enc, SvRV (sv));
454 SV *rv = SvRV (sv);
455
456 if (enc->indent >= enc->maxdepth)
457 croak ("data structure too deep (hit recursion limit)");
458
459 switch (SvTYPE (rv))
460 {
461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
462 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
463
464 default:
465 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
466 SvPV_nolen (sv));
467 }
468 }
469 else if (!SvOK (sv)) 651 else if (!SvOK (sv))
470 encode_str (enc, "null", 4, 0); 652 encode_str (enc, "null", 4, 0);
471 else 653 else
472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 654 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv)); 655 SvPV_nolen (sv), SvFLAGS (sv));
474} 656}
475 657
476static SV * 658static SV *
477encode_json (SV *scalar, U32 flags) 659encode_json (SV *scalar, JSON *json)
478{ 660{
661 enc_t enc;
662
479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 663 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 664 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
481 665
482 enc_t enc; 666 enc.json = *json;
483 enc.flags = flags;
484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 667 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
485 enc.cur = SvPVX (enc.sv); 668 enc.cur = SvPVX (enc.sv);
486 enc.end = SvEND (enc.sv); 669 enc.end = SvEND (enc.sv);
487 enc.indent = 0; 670 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags); 671 enc.maxdepth = DEC_DEPTH (enc.json.flags);
489 672
490 SvPOK_only (enc.sv); 673 SvPOK_only (enc.sv);
491 encode_sv (&enc, scalar); 674 encode_sv (&enc, scalar);
492 675
676 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
677 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
678
493 if (!(flags & (F_ASCII | F_UTF8))) 679 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
494 SvUTF8_on (enc.sv); 680 SvUTF8_on (enc.sv);
495 681
496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497
498 if (enc.flags & F_SHRINK) 682 if (enc.json.flags & F_SHRINK)
499 shrink (enc.sv); 683 shrink (enc.sv);
500 684
501 return enc.sv; 685 return enc.sv;
502} 686}
503 687
… …
508typedef struct 692typedef struct
509{ 693{
510 char *cur; // current parser pointer 694 char *cur; // current parser pointer
511 char *end; // end of input string 695 char *end; // end of input string
512 const char *err; // parse error, if != 0 696 const char *err; // parse error, if != 0
513 U32 flags; // F_* 697 JSON json;
514 U32 depth; // recursion depth 698 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit 699 U32 maxdepth; // recursion depth limit
516} dec_t; 700} dec_t;
517 701
518static void 702inline void
519decode_ws (dec_t *dec) 703decode_ws (dec_t *dec)
520{ 704{
521 for (;;) 705 for (;;)
522 { 706 {
523 char ch = *dec->cur; 707 char ch = *dec->cur;
… …
549decode_4hex (dec_t *dec) 733decode_4hex (dec_t *dec)
550{ 734{
551 signed char d1, d2, d3, d4; 735 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur; 736 unsigned char *cur = (unsigned char *)dec->cur;
553 737
554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 738 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 739 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 740 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 741 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
558 742
559 dec->cur += 4; 743 dec->cur += 4;
560 744
561 return ((UV)d1) << 12 745 return ((UV)d1) << 12
562 | ((UV)d2) << 8 746 | ((UV)d2) << 8
… …
570static SV * 754static SV *
571decode_str (dec_t *dec) 755decode_str (dec_t *dec)
572{ 756{
573 SV *sv = 0; 757 SV *sv = 0;
574 int utf8 = 0; 758 int utf8 = 0;
759 char *dec_cur = dec->cur;
575 760
576 do 761 do
577 { 762 {
578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 763 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
579 char *cur = buf; 764 char *cur = buf;
580 765
581 do 766 do
582 { 767 {
583 unsigned char ch = *(unsigned char *)dec->cur++; 768 unsigned char ch = *(unsigned char *)dec_cur++;
584 769
585 if (ch == '"') 770 if (expect_false (ch == '"'))
586 { 771 {
587 --dec->cur; 772 --dec_cur;
588 break; 773 break;
589 } 774 }
590 else if (ch == '\\') 775 else if (expect_false (ch == '\\'))
591 { 776 {
592 switch (*dec->cur) 777 switch (*dec_cur)
593 { 778 {
594 case '\\': 779 case '\\':
595 case '/': 780 case '/':
596 case '"': *cur++ = *dec->cur++; break; 781 case '"': *cur++ = *dec_cur++; break;
597 782
598 case 'b': ++dec->cur; *cur++ = '\010'; break; 783 case 'b': ++dec_cur; *cur++ = '\010'; break;
599 case 't': ++dec->cur; *cur++ = '\011'; break; 784 case 't': ++dec_cur; *cur++ = '\011'; break;
600 case 'n': ++dec->cur; *cur++ = '\012'; break; 785 case 'n': ++dec_cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break; 786 case 'f': ++dec_cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break; 787 case 'r': ++dec_cur; *cur++ = '\015'; break;
603 788
604 case 'u': 789 case 'u':
605 { 790 {
606 UV lo, hi; 791 UV lo, hi;
607 ++dec->cur; 792 ++dec_cur;
608 793
794 dec->cur = dec_cur;
609 hi = decode_4hex (dec); 795 hi = decode_4hex (dec);
796 dec_cur = dec->cur;
610 if (hi == (UV)-1) 797 if (hi == (UV)-1)
611 goto fail; 798 goto fail;
612 799
613 // possibly a surrogate pair 800 // possibly a surrogate pair
614 if (hi >= 0xd800) 801 if (hi >= 0xd800)
615 if (hi < 0xdc00) 802 if (hi < 0xdc00)
616 { 803 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 804 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair"); 805 ERR ("missing low surrogate character in surrogate pair");
619 806
620 dec->cur += 2; 807 dec_cur += 2;
621 808
809 dec->cur = dec_cur;
622 lo = decode_4hex (dec); 810 lo = decode_4hex (dec);
811 dec_cur = dec->cur;
623 if (lo == (UV)-1) 812 if (lo == (UV)-1)
624 goto fail; 813 goto fail;
625 814
626 if (lo < 0xdc00 || lo >= 0xe000) 815 if (lo < 0xdc00 || lo >= 0xe000)
627 ERR ("surrogate pair expected"); 816 ERR ("surrogate pair expected");
… …
641 *cur++ = hi; 830 *cur++ = hi;
642 } 831 }
643 break; 832 break;
644 833
645 default: 834 default:
646 --dec->cur; 835 --dec_cur;
647 ERR ("illegal backslash escape sequence in string"); 836 ERR ("illegal backslash escape sequence in string");
648 } 837 }
649 } 838 }
650 else if (ch >= 0x20 && ch <= 0x7f) 839 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
651 *cur++ = ch; 840 *cur++ = ch;
652 else if (ch >= 0x80) 841 else if (ch >= 0x80)
653 { 842 {
654 --dec->cur;
655
656 STRLEN clen; 843 STRLEN clen;
844 UV uch;
845
846 --dec_cur;
847
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 848 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
658 if (clen == (STRLEN)-1) 849 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string"); 850 ERR ("malformed UTF-8 character in JSON string");
660 851
661 do 852 do
662 {
663 *cur++ = *dec->cur++; 853 *cur++ = *dec_cur++;
664 }
665 while (--clen); 854 while (--clen);
666 855
667 utf8 = 1; 856 utf8 = 1;
668 } 857 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else 858 else
859 {
860 --dec_cur;
861
862 if (!ch)
863 ERR ("unexpected end of string while parsing JSON string");
864 else
672 ERR ("invalid character encountered"); 865 ERR ("invalid character encountered while parsing JSON string");
673 866 }
674 } 867 }
675 while (cur < buf + SHORT_STRING_LEN); 868 while (cur < buf + SHORT_STRING_LEN);
676 869
870 {
677 STRLEN len = cur - buf; 871 STRLEN len = cur - buf;
678 872
679 if (sv) 873 if (sv)
680 { 874 {
681 SvGROW (sv, SvCUR (sv) + len + 1); 875 SvGROW (sv, SvCUR (sv) + len + 1);
682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 876 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
683 SvCUR_set (sv, SvCUR (sv) + len); 877 SvCUR_set (sv, SvCUR (sv) + len);
684 } 878 }
685 else 879 else
686 sv = newSVpvn (buf, len); 880 sv = newSVpvn (buf, len);
687 } 881 }
882 }
688 while (*dec->cur != '"'); 883 while (*dec_cur != '"');
689 884
690 ++dec->cur; 885 ++dec_cur;
691 886
692 if (sv) 887 if (sv)
693 { 888 {
694 SvPOK_only (sv); 889 SvPOK_only (sv);
695 *SvEND (sv) = 0; 890 *SvEND (sv) = 0;
… …
698 SvUTF8_on (sv); 893 SvUTF8_on (sv);
699 } 894 }
700 else 895 else
701 sv = newSVpvn ("", 0); 896 sv = newSVpvn ("", 0);
702 897
898 dec->cur = dec_cur;
703 return sv; 899 return sv;
704 900
705fail: 901fail:
902 dec->cur = dec_cur;
706 return 0; 903 return 0;
707} 904}
708 905
709static SV * 906static SV *
710decode_num (dec_t *dec) 907decode_num (dec_t *dec)
… …
768 is_nv = 1; 965 is_nv = 1;
769 } 966 }
770 967
771 if (!is_nv) 968 if (!is_nv)
772 { 969 {
773 UV uv; 970 int len = dec->cur - start;
774 int numtype = grok_number (start, dec->cur - start, &uv); 971
775 if (numtype & IS_NUMBER_IN_UV) 972 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
776 if (numtype & IS_NUMBER_NEG) 973 if (*start == '-')
974 switch (len)
777 { 975 {
778 if (uv < (UV)IV_MIN) 976 case 2: return newSViv (-( start [1] - '0' * 1));
779 return newSViv (-(IV)uv); 977 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
978 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
979 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
780 } 980 }
981 else
982 switch (len)
983 {
984 case 1: return newSViv ( start [0] - '0' * 1);
985 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
986 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
987 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
988 }
989
990 {
991 UV uv;
992 int numtype = grok_number (start, len, &uv);
993 if (numtype & IS_NUMBER_IN_UV)
994 if (numtype & IS_NUMBER_NEG)
995 {
996 if (uv < (UV)IV_MIN)
997 return newSViv (-(IV)uv);
998 }
781 else 999 else
782 return newSVuv (uv); 1000 return newSVuv (uv);
783 } 1001 }
784 1002
1003 len -= *start == '-' ? 1 : 0;
1004
1005 // does not fit into IV or UV, try NV
1006 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1007 #if defined (LDBL_DIG)
1008 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1009 #endif
1010 )
1011 // fits into NV without loss of precision
1012 return newSVnv (Atof (start));
1013
1014 // everything else fails, convert it to a string
1015 return newSVpvn (start, dec->cur - start);
1016 }
1017
1018 // loss of precision here
785 return newSVnv (Atof (start)); 1019 return newSVnv (Atof (start));
786 1020
787fail: 1021fail:
788 return 0; 1022 return 0;
789} 1023}
… …
833} 1067}
834 1068
835static SV * 1069static SV *
836decode_hv (dec_t *dec) 1070decode_hv (dec_t *dec)
837{ 1071{
1072 SV *sv;
838 HV *hv = newHV (); 1073 HV *hv = newHV ();
839 1074
840 DEC_INC_DEPTH; 1075 DEC_INC_DEPTH;
841 decode_ws (dec); 1076 decode_ws (dec);
842 1077
843 if (*dec->cur == '}') 1078 if (*dec->cur == '}')
844 ++dec->cur; 1079 ++dec->cur;
845 else 1080 else
846 for (;;) 1081 for (;;)
847 { 1082 {
848 SV *key, *value;
849
850 decode_ws (dec); EXPECT_CH ('"'); 1083 decode_ws (dec); EXPECT_CH ('"');
851 1084
852 key = decode_str (dec); 1085 // heuristic: assume that
853 if (!key) 1086 // a) decode_str + hv_store_ent are abysmally slow.
854 goto fail; 1087 // b) most hash keys are short, simple ascii text.
1088 // => try to "fast-match" such strings to avoid
1089 // the overhead of decode_str + hv_store_ent.
1090 {
1091 SV *value;
1092 char *p = dec->cur;
1093 char *e = p + 24; // only try up to 24 bytes
855 1094
856 decode_ws (dec); EXPECT_CH (':'); 1095 for (;;)
857
858 value = decode_sv (dec);
859 if (!value)
860 { 1096 {
1097 // the >= 0x80 is true on most architectures
1098 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1099 {
1100 // slow path, back up and use decode_str
1101 SV *key = decode_str (dec);
1102 if (!key)
1103 goto fail;
1104
1105 decode_ws (dec); EXPECT_CH (':');
1106
1107 value = decode_sv (dec);
1108 if (!value)
1109 {
1110 SvREFCNT_dec (key);
1111 goto fail;
1112 }
1113
1114 hv_store_ent (hv, key, value, 0);
861 SvREFCNT_dec (key); 1115 SvREFCNT_dec (key);
1116
1117 break;
1118 }
1119 else if (*p == '"')
1120 {
1121 // fast path, got a simple key
1122 char *key = dec->cur;
1123 int len = p - key;
1124 dec->cur = p + 1;
1125
1126 decode_ws (dec); EXPECT_CH (':');
1127
1128 value = decode_sv (dec);
1129 if (!value)
862 goto fail; 1130 goto fail;
1131
1132 hv_store (hv, key, len, value, 0);
1133
1134 break;
1135 }
1136
1137 ++p;
863 } 1138 }
864 1139 }
865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
867 1140
868 decode_ws (dec); 1141 decode_ws (dec);
869 1142
870 if (*dec->cur == '}') 1143 if (*dec->cur == '}')
871 { 1144 {
… …
878 1151
879 ++dec->cur; 1152 ++dec->cur;
880 } 1153 }
881 1154
882 DEC_DEC_DEPTH; 1155 DEC_DEC_DEPTH;
883 return newRV_noinc ((SV *)hv); 1156 sv = newRV_noinc ((SV *)hv);
1157
1158 // check filter callbacks
1159 if (dec->json.flags & F_HOOK)
1160 {
1161 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1162 {
1163 HE *cb, *he;
1164
1165 hv_iterinit (hv);
1166 he = hv_iternext (hv);
1167 hv_iterinit (hv);
1168
1169 // the next line creates a mortal sv each time its called.
1170 // might want to optimise this for common cases.
1171 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1172
1173 if (cb)
1174 {
1175 dSP;
1176 int count;
1177
1178 ENTER; SAVETMPS; PUSHMARK (SP);
1179 XPUSHs (HeVAL (he));
1180
1181 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1182
1183 if (count == 1)
1184 {
1185 sv = newSVsv (POPs);
1186 FREETMPS; LEAVE;
1187 return sv;
1188 }
1189
1190 FREETMPS; LEAVE;
1191 }
1192 }
1193
1194 if (dec->json.cb_object)
1195 {
1196 dSP;
1197 int count;
1198
1199 ENTER; SAVETMPS; PUSHMARK (SP);
1200 XPUSHs (sv_2mortal (sv));
1201
1202 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1203
1204 if (count == 1)
1205 {
1206 sv = newSVsv (POPs);
1207 FREETMPS; LEAVE;
1208 return sv;
1209 }
1210
1211 SvREFCNT_inc (sv);
1212 FREETMPS; LEAVE;
1213 }
1214 }
1215
1216 return sv;
884 1217
885fail: 1218fail:
886 SvREFCNT_dec (hv); 1219 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH; 1220 DEC_DEC_DEPTH;
888 return 0; 1221 return 0;
… …
890 1223
891static SV * 1224static SV *
892decode_sv (dec_t *dec) 1225decode_sv (dec_t *dec)
893{ 1226{
894 decode_ws (dec); 1227 decode_ws (dec);
1228
1229 // the beauty of JSON: you need exactly one character lookahead
1230 // to parse anything.
895 switch (*dec->cur) 1231 switch (*dec->cur)
896 { 1232 {
897 case '"': ++dec->cur; return decode_str (dec); 1233 case '"': ++dec->cur; return decode_str (dec);
898 case '[': ++dec->cur; return decode_av (dec); 1234 case '[': ++dec->cur; return decode_av (dec);
899 case '{': ++dec->cur; return decode_hv (dec); 1235 case '{': ++dec->cur; return decode_hv (dec);
… …
905 1241
906 case 't': 1242 case 't':
907 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1243 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
908 { 1244 {
909 dec->cur += 4; 1245 dec->cur += 4;
910 return newSViv (1); 1246#if JSON_SLOW
1247 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1248#endif
1249 return SvREFCNT_inc (json_true);
911 } 1250 }
912 else 1251 else
913 ERR ("'true' expected"); 1252 ERR ("'true' expected");
914 1253
915 break; 1254 break;
916 1255
917 case 'f': 1256 case 'f':
918 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1257 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
919 { 1258 {
920 dec->cur += 5; 1259 dec->cur += 5;
921 return newSViv (0); 1260#if JSON_SLOW
1261 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1262#endif
1263 return SvREFCNT_inc (json_false);
922 } 1264 }
923 else 1265 else
924 ERR ("'false' expected"); 1266 ERR ("'false' expected");
925 1267
926 break; 1268 break;
… …
935 ERR ("'null' expected"); 1277 ERR ("'null' expected");
936 1278
937 break; 1279 break;
938 1280
939 default: 1281 default:
940 ERR ("malformed json string, neither array, object, number, string or atom"); 1282 ERR ("malformed JSON string, neither array, object, number, string or atom");
941 break; 1283 break;
942 } 1284 }
943 1285
944fail: 1286fail:
945 return 0; 1287 return 0;
946} 1288}
947 1289
948static SV * 1290static SV *
949decode_json (SV *string, U32 flags) 1291decode_json (SV *string, JSON *json, UV *offset_return)
950{ 1292{
1293 dec_t dec;
1294 UV offset;
951 SV *sv; 1295 SV *sv;
952 1296
1297 SvGETMAGIC (string);
1298 SvUPGRADE (string, SVt_PV);
1299
1300 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1301 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1302 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1303
953 if (flags & F_UTF8) 1304 if (json->flags & F_UTF8)
954 sv_utf8_downgrade (string, 0); 1305 sv_utf8_downgrade (string, 0);
955 else 1306 else
956 sv_utf8_upgrade (string); 1307 sv_utf8_upgrade (string);
957 1308
958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1309 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
959 1310
960 dec_t dec; 1311 dec.json = *json;
961 dec.flags = flags;
962 dec.cur = SvPVX (string); 1312 dec.cur = SvPVX (string);
963 dec.end = SvEND (string); 1313 dec.end = SvEND (string);
964 dec.err = 0; 1314 dec.err = 0;
965 dec.depth = 0; 1315 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags); 1316 dec.maxdepth = DEC_DEPTH (dec.json.flags);
967 1317
968 *SvEND (sv) = 0; // this shou[ld basically be a nop, too 1318 if (dec.json.cb_object || dec.json.cb_sk_object)
1319 dec.json.flags |= F_HOOK;
1320
1321 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
969 sv = decode_sv (&dec); 1322 sv = decode_sv (&dec);
970 1323
1324 if (!(offset_return || !sv))
1325 {
1326 // check for trailing garbage
1327 decode_ws (&dec);
1328
1329 if (*dec.cur)
1330 {
1331 dec.err = "garbage after JSON object";
1332 SvREFCNT_dec (sv);
1333 sv = 0;
1334 }
1335 }
1336
1337 if (offset_return || !sv)
1338 {
1339 offset = dec.json.flags & F_UTF8
1340 ? dec.cur - SvPVX (string)
1341 : utf8_distance (dec.cur, SvPVX (string));
1342
1343 if (offset_return)
1344 *offset_return = offset;
1345 }
1346
971 if (!sv) 1347 if (!sv)
972 { 1348 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
975 : utf8_distance (dec.cur, SvPVX (string));
976 SV *uni = sv_newmortal (); 1349 SV *uni = sv_newmortal ();
977 1350
978 // horrible hack to silence warning inside pv_uni_display 1351 // horrible hack to silence warning inside pv_uni_display
979 COP cop = *PL_curcop; 1352 COP cop = *PL_curcop;
980 cop.cop_warnings = pWARN_NONE; 1353 cop.cop_warnings = pWARN_NONE;
… …
982 SAVEVPTR (PL_curcop); 1355 SAVEVPTR (PL_curcop);
983 PL_curcop = &cop; 1356 PL_curcop = &cop;
984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1357 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE; 1358 LEAVE;
986 1359
987 croak ("%s, at character offset %d (%s)", 1360 croak ("%s, at character offset %d [\"%s\"]",
988 dec.err, 1361 dec.err,
989 (int)offset, 1362 (int)offset,
990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1363 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
991 } 1364 }
992 1365
993 sv = sv_2mortal (sv); 1366 sv = sv_2mortal (sv);
994 1367
995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1368 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
996 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1369 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
997 1370
998 return sv; 1371 return sv;
999} 1372}
1000 1373
… …
1004MODULE = JSON::XS PACKAGE = JSON::XS 1377MODULE = JSON::XS PACKAGE = JSON::XS
1005 1378
1006BOOT: 1379BOOT:
1007{ 1380{
1008 int i; 1381 int i;
1009
1010 memset (decode_hexdigit, 0xff, 256);
1011 1382
1012 for (i = 0; i < 256; ++i) 1383 for (i = 0; i < 256; ++i)
1013 decode_hexdigit [i] = 1384 decode_hexdigit [i] =
1014 i >= '0' && i <= '9' ? i - '0' 1385 i >= '0' && i <= '9' ? i - '0'
1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1386 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1387 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1017 : -1; 1388 : -1;
1018 1389
1019 json_stash = gv_stashpv ("JSON::XS", 1); 1390 json_stash = gv_stashpv ("JSON::XS" , 1);
1391 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1392
1393 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1394 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1020} 1395}
1021 1396
1022PROTOTYPES: DISABLE 1397PROTOTYPES: DISABLE
1023 1398
1024SV *new (char *dummy) 1399void CLONE (...)
1025 CODE: 1400 CODE:
1026 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1401 json_stash = 0;
1027 OUTPUT: 1402 json_boolean_stash = 0;
1028 RETVAL
1029 1403
1404void new (char *klass)
1405 PPCODE:
1406{
1407 SV *pv = NEWSV (0, sizeof (JSON));
1408 SvPOK_only (pv);
1409 Zero (SvPVX (pv), 1, JSON);
1410 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1411 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1412}
1413
1030SV *ascii (SV *self, int enable = 1) 1414void ascii (JSON *self, int enable = 1)
1031 ALIAS: 1415 ALIAS:
1032 ascii = F_ASCII 1416 ascii = F_ASCII
1417 latin1 = F_LATIN1
1033 utf8 = F_UTF8 1418 utf8 = F_UTF8
1034 indent = F_INDENT 1419 indent = F_INDENT
1035 canonical = F_CANONICAL 1420 canonical = F_CANONICAL
1036 space_before = F_SPACE_BEFORE 1421 space_before = F_SPACE_BEFORE
1037 space_after = F_SPACE_AFTER 1422 space_after = F_SPACE_AFTER
1038 pretty = F_PRETTY 1423 pretty = F_PRETTY
1039 allow_nonref = F_ALLOW_NONREF 1424 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK 1425 shrink = F_SHRINK
1426 allow_blessed = F_ALLOW_BLESSED
1427 convert_blessed = F_CONV_BLESSED
1041 CODE: 1428 PPCODE:
1042{ 1429{
1043 UV *uv = SvJSON (self);
1044 if (enable) 1430 if (enable)
1045 *uv |= ix; 1431 self->flags |= ix;
1046 else 1432 else
1047 *uv &= ~ix; 1433 self->flags &= ~ix;
1048 1434
1049 RETVAL = newSVsv (self); 1435 XPUSHs (ST (0));
1050} 1436}
1051 OUTPUT:
1052 RETVAL
1053 1437
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1438void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1055 CODE: 1439 PPCODE:
1056{ 1440{
1057 UV *uv = SvJSON (self);
1058 UV log2 = 0; 1441 UV log2 = 0;
1059 1442
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1443 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1061 1444
1062 while ((1UL << log2) < max_depth) 1445 while ((1UL << log2) < max_depth)
1063 ++log2; 1446 ++log2;
1064 1447
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1448 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066 1449
1067 RETVAL = newSVsv (self); 1450 XPUSHs (ST (0));
1068} 1451}
1069 OUTPUT:
1070 RETVAL
1071 1452
1072void encode (SV *self, SV *scalar) 1453void max_size (JSON *self, UV max_size = 0)
1073 PPCODE: 1454 PPCODE:
1074 XPUSHs (encode_json (scalar, *SvJSON (self))); 1455{
1456 UV log2 = 0;
1075 1457
1076void decode (SV *self, SV *jsonstr) 1458 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1459 if (max_size == 1) max_size = 2;
1460
1461 while ((1UL << log2) < max_size)
1462 ++log2;
1463
1464 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1465
1466 XPUSHs (ST (0));
1467}
1468
1469void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1470 PPCODE:
1471{
1472 SvREFCNT_dec (self->cb_object);
1473 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1474
1475 XPUSHs (ST (0));
1476}
1477
1478void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1479 PPCODE:
1480{
1481 if (!self->cb_sk_object)
1482 self->cb_sk_object = newHV ();
1483
1484 if (SvOK (cb))
1485 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1486 else
1487 {
1488 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1489
1490 if (!HvKEYS (self->cb_sk_object))
1491 {
1492 SvREFCNT_dec (self->cb_sk_object);
1493 self->cb_sk_object = 0;
1494 }
1495 }
1496
1497 XPUSHs (ST (0));
1498}
1499
1500void encode (JSON *self, SV *scalar)
1501 PPCODE:
1502 XPUSHs (encode_json (scalar, self));
1503
1504void decode (JSON *self, SV *jsonstr)
1505 PPCODE:
1078 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1506 XPUSHs (decode_json (jsonstr, self, 0));
1507
1508void decode_prefix (JSON *self, SV *jsonstr)
1509 PPCODE:
1510{
1511 UV offset;
1512 EXTEND (SP, 2);
1513 PUSHs (decode_json (jsonstr, self, &offset));
1514 PUSHs (sv_2mortal (newSVuv (offset)));
1515}
1516
1517void DESTROY (JSON *self)
1518 CODE:
1519 SvREFCNT_dec (self->cb_sk_object);
1520 SvREFCNT_dec (self->cb_object);
1079 1521
1080PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
1081 1523
1082void to_json (SV *scalar) 1524void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
1085 PPCODE: 1525 PPCODE:
1526{
1527 JSON json = { F_DEFAULT | F_UTF8 };
1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1528 XPUSHs (encode_json (scalar, &json));
1529}
1087 1530
1088void from_json (SV *jsonstr) 1531void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
1091 PPCODE: 1532 PPCODE:
1533{
1534 JSON json = { F_DEFAULT | F_UTF8 };
1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1535 XPUSHs (decode_json (jsonstr, &json, 0));
1536}
1093 1537

Diff Legend

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