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 1.14 by root, Sat Mar 24 22:55:43 2007 UTC vs.
Revision 1.59 by root, Mon Aug 13 16:14:20 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include <stdio.h>
9#include <float.h>
8 10
11#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h
13#endif
14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
9#define F_ASCII 0x00000001 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 23#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 24#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 25#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 26#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 27#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 28#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
32#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37
38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 40
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 42#define F_DEFAULT (9UL << S_MAXDEPTH)
20 43
21#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 45#define INDENT_STEP 3 // spaces per indentation level
23 46
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#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
26 48
27#define SB do { 49#define SB do {
28#define SE } while (0) 50#define SE } while (0)
29 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
30static 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;
31 79
32///////////////////////////////////////////////////////////////////////////// 80/////////////////////////////////////////////////////////////////////////////
33// utility functions 81// utility functions
34 82
35static UV * 83inline void
36SvJSON (SV *sv)
37{
38 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
39 croak ("object is not of type JSON::XS");
40
41 return &SvUVX (SvRV (sv));
42}
43
44static void
45shrink (SV *sv) 84shrink (SV *sv)
46{ 85{
47 sv_utf8_downgrade (sv, 1); 86 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 87 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 88 {
… …
58// 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
59// case of an error. 98// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF, 99// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else. 100// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints 101// note that we never call this function for a ascii codepoints
63static UV 102inline UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{ 104{
66 if (s[0] > 0xdf || s[0] < 0xc2) 105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69 { 108 {
70 *clen = 2; 109 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 111 }
73 else 112 else
113 {
114 *clen = (STRLEN)-1;
74 return (UV)-1; 115 return (UV)-1;
116 }
75} 117}
76 118
77///////////////////////////////////////////////////////////////////////////// 119/////////////////////////////////////////////////////////////////////////////
78// encoder 120// encoder
79 121
… …
81typedef struct 123typedef struct
82{ 124{
83 char *cur; // SvPVX (sv) + current output position 125 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 126 char *end; // SvEND (sv)
85 SV *sv; // result scalar 127 SV *sv; // result scalar
86 UV flags; // F_* 128 JSON json;
87 int indent; // indentation level 129 U32 indent; // indentation level
88 int max_depth; // max. recursion level 130 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 131} enc_t;
90 132
91static void 133inline void
92need (enc_t *enc, STRLEN len) 134need (enc_t *enc, STRLEN len)
93{ 135{
94 if (enc->cur + len >= enc->end) 136 if (expect_false (enc->cur + len >= enc->end))
95 { 137 {
96 STRLEN cur = enc->cur - SvPVX (enc->sv); 138 STRLEN cur = enc->cur - SvPVX (enc->sv);
97 SvGROW (enc->sv, cur + len + 1); 139 SvGROW (enc->sv, cur + len + 1);
98 enc->cur = SvPVX (enc->sv) + cur; 140 enc->cur = SvPVX (enc->sv) + cur;
99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100 } 142 }
101} 143}
102 144
103static void 145inline void
104encode_ch (enc_t *enc, char ch) 146encode_ch (enc_t *enc, char ch)
105{ 147{
106 need (enc, 1); 148 need (enc, 1);
107 *enc->cur++ = ch; 149 *enc->cur++ = ch;
108} 150}
… …
116 158
117 while (str < end) 159 while (str < end)
118 { 160 {
119 unsigned char ch = *(unsigned char *)str; 161 unsigned char ch = *(unsigned char *)str;
120 162
121 if (ch >= 0x20 && ch < 0x80) // most common case 163 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122 { 164 {
123 if (ch == '"') // but with slow exceptions 165 if (expect_false (ch == '"')) // but with slow exceptions
124 { 166 {
125 need (enc, len += 1); 167 need (enc, len += 1);
126 *enc->cur++ = '\\'; 168 *enc->cur++ = '\\';
127 *enc->cur++ = '"'; 169 *enc->cur++ = '"';
128 } 170 }
129 else if (ch == '\\') 171 else if (expect_false (ch == '\\'))
130 { 172 {
131 need (enc, len += 1); 173 need (enc, len += 1);
132 *enc->cur++ = '\\'; 174 *enc->cur++ = '\\';
133 *enc->cur++ = '\\'; 175 *enc->cur++ = '\\';
134 } 176 }
… …
152 STRLEN clen; 194 STRLEN clen;
153 UV uch; 195 UV uch;
154 196
155 if (is_utf8) 197 if (is_utf8)
156 { 198 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen); 199 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1) 200 if (clen == (STRLEN)-1)
160 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);
161 } 202 }
162 else 203 else
… …
166 } 207 }
167 208
168 if (uch > 0x10FFFFUL) 209 if (uch > 0x10FFFFUL)
169 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);
170 211
171 if (uch < 0x80 || enc->flags & F_ASCII) 212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
172 { 213 {
173 if (uch > 0xFFFFUL) 214 if (uch > 0xFFFFUL)
174 { 215 {
175 need (enc, len += 11); 216 need (enc, len += 11);
176 sprintf (enc->cur, "\\u%04x\\u%04x", 217 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
190 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 231 *enc->cur++ = hexdigit [(uch >> 0) & 15];
191 } 232 }
192 233
193 str += clen; 234 str += clen;
194 } 235 }
236 else if (enc->json.flags & F_LATIN1)
237 {
238 *enc->cur++ = uch;
239 str += clen;
240 }
195 else if (is_utf8) 241 else if (is_utf8)
196 { 242 {
197 need (enc, len += clen); 243 need (enc, len += clen);
198 do 244 do
199 { 245 {
… …
201 } 247 }
202 while (--clen); 248 while (--clen);
203 } 249 }
204 else 250 else
205 { 251 {
206 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
207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str; 254 ++str;
209 } 255 }
210 } 256 }
211 } 257 }
… …
213 259
214 --len; 260 --len;
215 } 261 }
216} 262}
217 263
218static void 264inline void
219encode_indent (enc_t *enc) 265encode_indent (enc_t *enc)
220{ 266{
221 if (enc->flags & F_INDENT) 267 if (enc->json.flags & F_INDENT)
222 { 268 {
223 int spaces = enc->indent * INDENT_STEP; 269 int spaces = enc->indent * INDENT_STEP;
224 270
225 need (enc, spaces); 271 need (enc, spaces);
226 memset (enc->cur, ' ', spaces); 272 memset (enc->cur, ' ', spaces);
227 enc->cur += spaces; 273 enc->cur += spaces;
228 } 274 }
229} 275}
230 276
231static void 277inline void
232encode_space (enc_t *enc) 278encode_space (enc_t *enc)
233{ 279{
234 need (enc, 1); 280 need (enc, 1);
235 encode_ch (enc, ' '); 281 encode_ch (enc, ' ');
236} 282}
237 283
238static void 284inline void
239encode_nl (enc_t *enc) 285encode_nl (enc_t *enc)
240{ 286{
241 if (enc->flags & F_INDENT) 287 if (enc->json.flags & F_INDENT)
242 { 288 {
243 need (enc, 1); 289 need (enc, 1);
244 encode_ch (enc, '\n'); 290 encode_ch (enc, '\n');
245 } 291 }
246} 292}
247 293
248static void 294inline void
249encode_comma (enc_t *enc) 295encode_comma (enc_t *enc)
250{ 296{
251 encode_ch (enc, ','); 297 encode_ch (enc, ',');
252 298
253 if (enc->flags & F_INDENT) 299 if (enc->json.flags & F_INDENT)
254 encode_nl (enc); 300 encode_nl (enc);
255 else if (enc->flags & F_SPACE_AFTER) 301 else if (enc->json.flags & F_SPACE_AFTER)
256 encode_space (enc); 302 encode_space (enc);
257} 303}
258 304
259static void encode_sv (enc_t *enc, SV *sv); 305static void encode_sv (enc_t *enc, SV *sv);
260 306
261static void 307static void
262encode_av (enc_t *enc, AV *av) 308encode_av (enc_t *enc, AV *av)
263{ 309{
264 int i, len = av_len (av); 310 int i, len = av_len (av);
265 311
312 if (enc->indent >= enc->maxdepth)
313 croak ("data structure too deep (hit recursion limit)");
314
266 encode_ch (enc, '['); encode_nl (enc); 315 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 316 ++enc->indent;
268 317
269 for (i = 0; i <= len; ++i) 318 for (i = 0; i <= len; ++i)
270 { 319 {
320 SV **svp = av_fetch (av, i, 0);
321
271 encode_indent (enc); 322 encode_indent (enc);
272 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);
273 328
274 if (i < len) 329 if (i < len)
275 encode_comma (enc); 330 encode_comma (enc);
276 } 331 }
277 332
… …
300 else 355 else
301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 356 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
302 357
303 encode_ch (enc, '"'); 358 encode_ch (enc, '"');
304 359
305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 360 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
306 encode_ch (enc, ':'); 361 encode_ch (enc, ':');
307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 362 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
308 encode_sv (enc, HeVAL (he)); 363 encode_sv (enc, HeVAL (he));
309} 364}
310 365
311// compare hash entries, used when all keys are bytestrings 366// compare hash entries, used when all keys are bytestrings
312static int 367static int
… …
335 390
336static void 391static void
337encode_hv (enc_t *enc, HV *hv) 392encode_hv (enc_t *enc, HV *hv)
338{ 393{
339 int count, i; 394 int count, i;
395
396 if (enc->indent >= enc->maxdepth)
397 croak ("data structure too deep (hit recursion limit)");
340 398
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 400
343 if ((count = hv_iterinit (hv))) 401 if ((count = hv_iterinit (hv)))
344 { 402 {
345 // for canonical output we have to sort by keys first 403 // for canonical output we have to sort by keys first
346 // actually, this is mostly due to the stupid so-called 404 // actually, this is mostly due to the stupid so-called
347 // security workaround added somewhere in 5.8.x. 405 // security workaround added somewhere in 5.8.x.
348 // that randomises hash orderings 406 // that randomises hash orderings
349 if (enc->flags & F_CANONICAL) 407 if (enc->json.flags & F_CANONICAL)
350 { 408 {
351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
352 int fast = 1; 409 int fast = 1;
410 HE *he;
411#if defined(__BORLANDC__) || defined(_MSC_VER)
412 HE **hes = _alloca (count * sizeof (HE));
413#else
414 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
415#endif
353 416
354 i = 0; 417 i = 0;
355 while ((he = hv_iternext (hv))) 418 while ((he = hv_iternext (hv)))
356 { 419 {
357 hes [i++] = he; 420 hes [i++] = he;
… …
392 455
393 encode_nl (enc); 456 encode_nl (enc);
394 } 457 }
395 else 458 else
396 { 459 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 460 HE *he = hv_iternext (hv);
399 461
400 for (;;) 462 for (;;)
401 { 463 {
402 encode_indent (enc); 464 encode_indent (enc);
… …
411 encode_nl (enc); 473 encode_nl (enc);
412 } 474 }
413 } 475 }
414 476
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 477 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
478}
479
480// encode objects, arrays and special \0=false and \1=true values.
481static void
482encode_rv (enc_t *enc, SV *sv)
483{
484 svtype svt;
485
486 SvGETMAGIC (sv);
487 svt = SvTYPE (sv);
488
489 if (expect_false (SvOBJECT (sv)))
490 {
491 HV *stash = !JSON_SLOW || json_boolean_stash
492 ? json_boolean_stash
493 : gv_stashpv ("JSON::XS::Boolean", 1);
494
495 if (SvSTASH (sv) == stash)
496 {
497 if (SvIV (sv))
498 encode_str (enc, "true", 4, 0);
499 else
500 encode_str (enc, "false", 5, 0);
501 }
502 else
503 {
504#if 0
505 if (0 && sv_derived_from (rv, "JSON::Literal"))
506 {
507 // not yet
508 }
509#endif
510 if (enc->json.flags & F_CONV_BLESSED)
511 {
512 // we re-bless the reference to get overload and other niceties right
513 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
514
515 if (to_json)
516 {
517 dSP;
518
519 ENTER; SAVETMPS; PUSHMARK (SP);
520 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
521
522 // calling with G_SCALAR ensures that we always get a 1 return value
523 PUTBACK;
524 call_sv ((SV *)GvCV (to_json), G_SCALAR);
525 SPAGAIN;
526
527 // catch this surprisingly common error
528 if (SvROK (TOPs) && SvRV (TOPs) == sv)
529 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
530
531 sv = POPs;
532 PUTBACK;
533
534 encode_sv (enc, sv);
535
536 FREETMPS; LEAVE;
537 }
538 else if (enc->json.flags & F_ALLOW_BLESSED)
539 encode_str (enc, "null", 4, 0);
540 else
541 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
543 }
544 else if (enc->json.flags & F_ALLOW_BLESSED)
545 encode_str (enc, "null", 4, 0);
546 else
547 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
548 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
549 }
550 }
551 else if (svt == SVt_PVHV)
552 encode_hv (enc, (HV *)sv);
553 else if (svt == SVt_PVAV)
554 encode_av (enc, (AV *)sv);
555 else if (svt < SVt_PVAV)
556 {
557 STRLEN len = 0;
558 char *pv = svt ? SvPV (sv, len) : 0;
559
560 if (len == 1 && *pv == '1')
561 encode_str (enc, "true", 4, 0);
562 else if (len == 1 && *pv == '0')
563 encode_str (enc, "false", 5, 0);
564 else
565 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
566 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
567 }
568 else
569 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
570 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 571}
417 572
418static void 573static void
419encode_sv (enc_t *enc, SV *sv) 574encode_sv (enc_t *enc, SV *sv)
420{ 575{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 583 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 584 encode_ch (enc, '"');
430 } 585 }
431 else if (SvNOKp (sv)) 586 else if (SvNOKp (sv))
432 { 587 {
588 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 589 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 590 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 591 enc->cur += strlen (enc->cur);
436 } 592 }
437 else if (SvIOKp (sv)) 593 else if (SvIOKp (sv))
438 { 594 {
439 need (enc, 64); 595 // we assume we can always read an IV as a UV
596 if (SvUV (sv) & ~(UV)0x7fff)
597 {
598 // large integer, use the (rather slow) snprintf way.
599 need (enc, sizeof (UV) * 3);
440 enc->cur += 600 enc->cur +=
441 SvIsUV(sv) 601 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 602 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 603 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
604 }
605 else
606 {
607 // optimise the "small number case"
608 // code will likely be branchless and use only a single multiplication
609 I32 i = SvIV (sv);
610 U32 u;
611 char digit, nz = 0;
612
613 need (enc, 6);
614
615 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
616 u = i < 0 ? -i : i;
617
618 // convert to 4.28 fixed-point representation
619 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
620
621 // now output digit by digit, each time masking out the integer part
622 // and multiplying by 5 while moving the decimal point one to the right,
623 // resulting in a net multiplication by 10.
624 // we always write the digit to memory but conditionally increment
625 // the pointer, to ease the usage of conditional move instructions.
626 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
627 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
628 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
629 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
630 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
631 }
444 } 632 }
445 else if (SvROK (sv)) 633 else if (SvROK (sv))
446 { 634 encode_rv (enc, SvRV (sv));
447 SV *rv = SvRV (sv);
448
449 if (enc->indent >= enc->max_depth)
450 croak ("data structure too deep (hit recursion limit)");
451
452 switch (SvTYPE (rv))
453 {
454 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
455 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
456
457 default:
458 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
459 SvPV_nolen (sv));
460 }
461 }
462 else if (!SvOK (sv)) 635 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 636 encode_str (enc, "null", 4, 0);
464 else 637 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 638 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 639 SvPV_nolen (sv), SvFLAGS (sv));
467} 640}
468 641
469static SV * 642static SV *
470encode_json (SV *scalar, UV flags) 643encode_json (SV *scalar, JSON *json)
471{ 644{
645 enc_t enc;
646
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 647 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 648 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 649
475 enc_t enc; 650 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 651 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 652 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 653 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 654 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 655 enc.maxdepth = DEC_DEPTH (enc.json.flags);
482 656
483 SvPOK_only (enc.sv); 657 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 658 encode_sv (&enc, scalar);
485 659
660 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
661 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
662
486 if (!(flags & (F_ASCII | F_UTF8))) 663 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 664 SvUTF8_on (enc.sv);
488 665
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 666 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 667 shrink (enc.sv);
493 668
494 return enc.sv; 669 return enc.sv;
495} 670}
496 671
… …
501typedef struct 676typedef struct
502{ 677{
503 char *cur; // current parser pointer 678 char *cur; // current parser pointer
504 char *end; // end of input string 679 char *end; // end of input string
505 const char *err; // parse error, if != 0 680 const char *err; // parse error, if != 0
506 UV flags; // F_* 681 JSON json;
682 U32 depth; // recursion depth
683 U32 maxdepth; // recursion depth limit
507} dec_t; 684} dec_t;
508 685
509static void 686inline void
510decode_ws (dec_t *dec) 687decode_ws (dec_t *dec)
511{ 688{
512 for (;;) 689 for (;;)
513 { 690 {
514 char ch = *dec->cur; 691 char ch = *dec->cur;
… …
520 ++dec->cur; 697 ++dec->cur;
521 } 698 }
522} 699}
523 700
524#define ERR(reason) SB dec->err = reason; goto fail; SE 701#define ERR(reason) SB dec->err = reason; goto fail; SE
702
525#define EXPECT_CH(ch) SB \ 703#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 704 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 705 ERR (# ch " expected"); \
528 ++dec->cur; \ 706 ++dec->cur; \
529 SE 707 SE
530 708
709#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
710#define DEC_DEC_DEPTH --dec->depth
711
531static SV *decode_sv (dec_t *dec); 712static SV *decode_sv (dec_t *dec);
532 713
533static signed char decode_hexdigit[256]; 714static signed char decode_hexdigit[256];
534 715
535static UV 716static UV
536decode_4hex (dec_t *dec) 717decode_4hex (dec_t *dec)
537{ 718{
538 signed char d1, d2, d3, d4; 719 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 720 unsigned char *cur = (unsigned char *)dec->cur;
540 721
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 722 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
542 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 723 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
543 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 724 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
544 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 725 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 726
546 dec->cur += 4; 727 dec->cur += 4;
547 728
548 return ((UV)d1) << 12 729 return ((UV)d1) << 12
549 | ((UV)d2) << 8 730 | ((UV)d2) << 8
… …
557static SV * 738static SV *
558decode_str (dec_t *dec) 739decode_str (dec_t *dec)
559{ 740{
560 SV *sv = 0; 741 SV *sv = 0;
561 int utf8 = 0; 742 int utf8 = 0;
743 char *dec_cur = dec->cur;
562 744
563 do 745 do
564 { 746 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 747 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 748 char *cur = buf;
567 749
568 do 750 do
569 { 751 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 752 unsigned char ch = *(unsigned char *)dec_cur++;
571 753
572 if (ch == '"') 754 if (expect_false (ch == '"'))
573 { 755 {
574 --dec->cur; 756 --dec_cur;
575 break; 757 break;
576 } 758 }
577 else if (ch == '\\') 759 else if (expect_false (ch == '\\'))
578 { 760 {
579 switch (*dec->cur) 761 switch (*dec_cur)
580 { 762 {
581 case '\\': 763 case '\\':
582 case '/': 764 case '/':
583 case '"': *cur++ = *dec->cur++; break; 765 case '"': *cur++ = *dec_cur++; break;
584 766
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 767 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 768 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 769 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 770 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 771 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 772
591 case 'u': 773 case 'u':
592 { 774 {
593 UV lo, hi; 775 UV lo, hi;
594 ++dec->cur; 776 ++dec_cur;
595 777
778 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 779 hi = decode_4hex (dec);
780 dec_cur = dec->cur;
597 if (hi == (UV)-1) 781 if (hi == (UV)-1)
598 goto fail; 782 goto fail;
599 783
600 // possibly a surrogate pair 784 // possibly a surrogate pair
601 if (hi >= 0xd800) 785 if (hi >= 0xd800)
602 if (hi < 0xdc00) 786 if (hi < 0xdc00)
603 { 787 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 788 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 789 ERR ("missing low surrogate character in surrogate pair");
606 790
607 dec->cur += 2; 791 dec_cur += 2;
608 792
793 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 794 lo = decode_4hex (dec);
795 dec_cur = dec->cur;
610 if (lo == (UV)-1) 796 if (lo == (UV)-1)
611 goto fail; 797 goto fail;
612 798
613 if (lo < 0xdc00 || lo >= 0xe000) 799 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 800 ERR ("surrogate pair expected");
… …
628 *cur++ = hi; 814 *cur++ = hi;
629 } 815 }
630 break; 816 break;
631 817
632 default: 818 default:
633 --dec->cur; 819 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 820 ERR ("illegal backslash escape sequence in string");
635 } 821 }
636 } 822 }
637 else if (ch >= 0x20 && ch <= 0x7f) 823 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638 *cur++ = ch; 824 *cur++ = ch;
639 else if (ch >= 0x80) 825 else if (ch >= 0x80)
640 { 826 {
641 --dec->cur;
642
643 STRLEN clen; 827 STRLEN clen;
828 UV uch;
829
830 --dec_cur;
831
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 832 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 833 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 834 ERR ("malformed UTF-8 character in JSON string");
647 835
648 do 836 do
649 {
650 *cur++ = *dec->cur++; 837 *cur++ = *dec_cur++;
651 }
652 while (--clen); 838 while (--clen);
653 839
654 utf8 = 1; 840 utf8 = 1;
655 } 841 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 842 else
843 {
844 --dec_cur;
845
846 if (!ch)
847 ERR ("unexpected end of string while parsing JSON string");
848 else
659 ERR ("invalid character encountered"); 849 ERR ("invalid character encountered while parsing JSON string");
660 850 }
661 } 851 }
662 while (cur < buf + SHORT_STRING_LEN); 852 while (cur < buf + SHORT_STRING_LEN);
663 853
854 {
664 STRLEN len = cur - buf; 855 STRLEN len = cur - buf;
665 856
666 if (sv) 857 if (sv)
667 { 858 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 859 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 860 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 861 SvCUR_set (sv, SvCUR (sv) + len);
671 } 862 }
672 else 863 else
673 sv = newSVpvn (buf, len); 864 sv = newSVpvn (buf, len);
674 } 865 }
866 }
675 while (*dec->cur != '"'); 867 while (*dec_cur != '"');
676 868
677 ++dec->cur; 869 ++dec_cur;
678 870
679 if (sv) 871 if (sv)
680 { 872 {
681 SvPOK_only (sv); 873 SvPOK_only (sv);
682 *SvEND (sv) = 0; 874 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 877 SvUTF8_on (sv);
686 } 878 }
687 else 879 else
688 sv = newSVpvn ("", 0); 880 sv = newSVpvn ("", 0);
689 881
882 dec->cur = dec_cur;
690 return sv; 883 return sv;
691 884
692fail: 885fail:
886 dec->cur = dec_cur;
693 return 0; 887 return 0;
694} 888}
695 889
696static SV * 890static SV *
697decode_num (dec_t *dec) 891decode_num (dec_t *dec)
… …
755 is_nv = 1; 949 is_nv = 1;
756 } 950 }
757 951
758 if (!is_nv) 952 if (!is_nv)
759 { 953 {
760 UV uv; 954 int len = dec->cur - start;
761 int numtype = grok_number (start, dec->cur - start, &uv); 955
762 if (numtype & IS_NUMBER_IN_UV) 956 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
763 if (numtype & IS_NUMBER_NEG) 957 if (*start == '-')
958 switch (len)
764 { 959 {
765 if (uv < (UV)IV_MIN) 960 case 2: return newSViv (-( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 961 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
962 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
963 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767 } 964 }
965 else
966 switch (len)
967 {
968 case 1: return newSViv ( start [0] - '0' * 1);
969 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
970 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
971 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
972 }
973
974 {
975 UV uv;
976 int numtype = grok_number (start, len, &uv);
977 if (numtype & IS_NUMBER_IN_UV)
978 if (numtype & IS_NUMBER_NEG)
979 {
980 if (uv < (UV)IV_MIN)
981 return newSViv (-(IV)uv);
982 }
768 else 983 else
769 return newSVuv (uv); 984 return newSVuv (uv);
770 } 985 }
771 986
987 len -= *start == '-' ? 1 : 0;
988
989 // does not fit into IV or UV, try NV
990 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
991 #if defined (LDBL_DIG)
992 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
993 #endif
994 )
995 // fits into NV without loss of precision
996 return newSVnv (Atof (start));
997
998 // everything else fails, convert it to a string
999 return newSVpvn (start, dec->cur - start);
1000 }
1001
1002 // loss of precision here
772 return newSVnv (Atof (start)); 1003 return newSVnv (Atof (start));
773 1004
774fail: 1005fail:
775 return 0; 1006 return 0;
776} 1007}
… …
778static SV * 1009static SV *
779decode_av (dec_t *dec) 1010decode_av (dec_t *dec)
780{ 1011{
781 AV *av = newAV (); 1012 AV *av = newAV ();
782 1013
1014 DEC_INC_DEPTH;
783 decode_ws (dec); 1015 decode_ws (dec);
1016
784 if (*dec->cur == ']') 1017 if (*dec->cur == ']')
785 ++dec->cur; 1018 ++dec->cur;
786 else 1019 else
787 for (;;) 1020 for (;;)
788 { 1021 {
… …
806 ERR (", or ] expected while parsing array"); 1039 ERR (", or ] expected while parsing array");
807 1040
808 ++dec->cur; 1041 ++dec->cur;
809 } 1042 }
810 1043
1044 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1045 return newRV_noinc ((SV *)av);
812 1046
813fail: 1047fail:
814 SvREFCNT_dec (av); 1048 SvREFCNT_dec (av);
1049 DEC_DEC_DEPTH;
815 return 0; 1050 return 0;
816} 1051}
817 1052
818static SV * 1053static SV *
819decode_hv (dec_t *dec) 1054decode_hv (dec_t *dec)
820{ 1055{
1056 SV *sv;
821 HV *hv = newHV (); 1057 HV *hv = newHV ();
822 1058
1059 DEC_INC_DEPTH;
823 decode_ws (dec); 1060 decode_ws (dec);
1061
824 if (*dec->cur == '}') 1062 if (*dec->cur == '}')
825 ++dec->cur; 1063 ++dec->cur;
826 else 1064 else
827 for (;;) 1065 for (;;)
828 { 1066 {
829 SV *key, *value;
830
831 decode_ws (dec); EXPECT_CH ('"'); 1067 decode_ws (dec); EXPECT_CH ('"');
832 1068
833 key = decode_str (dec); 1069 // heuristic: assume that
834 if (!key) 1070 // a) decode_str + hv_store_ent are abysmally slow.
835 goto fail; 1071 // b) most hash keys are short, simple ascii text.
1072 // => try to "fast-match" such strings to avoid
1073 // the overhead of decode_str + hv_store_ent.
1074 {
1075 SV *value;
1076 char *p = dec->cur;
1077 char *e = p + 24; // only try up to 24 bytes
836 1078
837 decode_ws (dec); EXPECT_CH (':'); 1079 for (;;)
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1080 {
1081 // the >= 0x80 is true on most architectures
1082 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1083 {
1084 // slow path, back up and use decode_str
1085 SV *key = decode_str (dec);
1086 if (!key)
1087 goto fail;
1088
1089 decode_ws (dec); EXPECT_CH (':');
1090
1091 value = decode_sv (dec);
1092 if (!value)
1093 {
1094 SvREFCNT_dec (key);
1095 goto fail;
1096 }
1097
1098 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1099 SvREFCNT_dec (key);
1100
1101 break;
1102 }
1103 else if (*p == '"')
1104 {
1105 // fast path, got a simple key
1106 char *key = dec->cur;
1107 int len = p - key;
1108 dec->cur = p + 1;
1109
1110 decode_ws (dec); EXPECT_CH (':');
1111
1112 value = decode_sv (dec);
1113 if (!value)
843 goto fail; 1114 goto fail;
1115
1116 hv_store (hv, key, len, value, 0);
1117
1118 break;
1119 }
1120
1121 ++p;
844 } 1122 }
845 1123 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1124
849 decode_ws (dec); 1125 decode_ws (dec);
850 1126
851 if (*dec->cur == '}') 1127 if (*dec->cur == '}')
852 { 1128 {
… …
858 ERR (", or } expected while parsing object/hash"); 1134 ERR (", or } expected while parsing object/hash");
859 1135
860 ++dec->cur; 1136 ++dec->cur;
861 } 1137 }
862 1138
1139 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 1140 sv = newRV_noinc ((SV *)hv);
1141
1142 // check filter callbacks
1143 if (dec->json.flags & F_HOOK)
1144 {
1145 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1146 {
1147 HE *cb, *he;
1148
1149 hv_iterinit (hv);
1150 he = hv_iternext (hv);
1151 hv_iterinit (hv);
1152
1153 // the next line creates a mortal sv each time its called.
1154 // might want to optimise this for common cases.
1155 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1156
1157 if (cb)
1158 {
1159 dSP;
1160 int count;
1161
1162 ENTER; SAVETMPS; PUSHMARK (SP);
1163 XPUSHs (HeVAL (he));
1164
1165 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1166
1167 if (count == 1)
1168 {
1169 sv = newSVsv (POPs);
1170 FREETMPS; LEAVE;
1171 return sv;
1172 }
1173
1174 FREETMPS; LEAVE;
1175 }
1176 }
1177
1178 if (dec->json.cb_object)
1179 {
1180 dSP;
1181 int count;
1182
1183 ENTER; SAVETMPS; PUSHMARK (SP);
1184 XPUSHs (sv_2mortal (sv));
1185
1186 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1187
1188 if (count == 1)
1189 {
1190 sv = newSVsv (POPs);
1191 FREETMPS; LEAVE;
1192 return sv;
1193 }
1194
1195 SvREFCNT_inc (sv);
1196 FREETMPS; LEAVE;
1197 }
1198 }
1199
1200 return sv;
864 1201
865fail: 1202fail:
866 SvREFCNT_dec (hv); 1203 SvREFCNT_dec (hv);
1204 DEC_DEC_DEPTH;
867 return 0; 1205 return 0;
868} 1206}
869 1207
870static SV * 1208static SV *
871decode_sv (dec_t *dec) 1209decode_sv (dec_t *dec)
872{ 1210{
873 decode_ws (dec); 1211 decode_ws (dec);
1212
1213 // the beauty of JSON: you need exactly one character lookahead
1214 // to parse anything.
874 switch (*dec->cur) 1215 switch (*dec->cur)
875 { 1216 {
876 case '"': ++dec->cur; return decode_str (dec); 1217 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1218 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1219 case '{': ++dec->cur; return decode_hv (dec);
… …
884 1225
885 case 't': 1226 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1227 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1228 {
888 dec->cur += 4; 1229 dec->cur += 4;
889 return newSViv (1); 1230 return SvREFCNT_inc (json_true);
890 } 1231 }
891 else 1232 else
892 ERR ("'true' expected"); 1233 ERR ("'true' expected");
893 1234
894 break; 1235 break;
895 1236
896 case 'f': 1237 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1238 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1239 {
899 dec->cur += 5; 1240 dec->cur += 5;
900 return newSViv (0); 1241 return SvREFCNT_inc (json_false);
901 } 1242 }
902 else 1243 else
903 ERR ("'false' expected"); 1244 ERR ("'false' expected");
904 1245
905 break; 1246 break;
… …
914 ERR ("'null' expected"); 1255 ERR ("'null' expected");
915 1256
916 break; 1257 break;
917 1258
918 default: 1259 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1260 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1261 break;
921 } 1262 }
922 1263
923fail: 1264fail:
924 return 0; 1265 return 0;
925} 1266}
926 1267
927static SV * 1268static SV *
928decode_json (SV *string, UV flags) 1269decode_json (SV *string, JSON *json, UV *offset_return)
929{ 1270{
1271 dec_t dec;
1272 UV offset;
930 SV *sv; 1273 SV *sv;
931 1274
1275 SvGETMAGIC (string);
1276 SvUPGRADE (string, SVt_PV);
1277
1278 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1279 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1280 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1281
932 if (flags & F_UTF8) 1282 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1283 sv_utf8_downgrade (string, 0);
934 else 1284 else
935 sv_utf8_upgrade (string); 1285 sv_utf8_upgrade (string);
936 1286
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1287 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1288
939 dec_t dec; 1289 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1290 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1291 dec.end = SvEND (string);
943 dec.err = 0; 1292 dec.err = 0;
1293 dec.depth = 0;
1294 dec.maxdepth = DEC_DEPTH (dec.json.flags);
944 1295
1296 if (dec.json.cb_object || dec.json.cb_sk_object)
1297 dec.json.flags |= F_HOOK;
1298
1299 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
945 sv = decode_sv (&dec); 1300 sv = decode_sv (&dec);
946 1301
1302 if (!(offset_return || !sv))
1303 {
1304 // check for trailing garbage
1305 decode_ws (&dec);
1306
1307 if (*dec.cur)
1308 {
1309 dec.err = "garbage after JSON object";
1310 SvREFCNT_dec (sv);
1311 sv = 0;
1312 }
1313 }
1314
1315 if (offset_return || !sv)
1316 {
1317 offset = dec.json.flags & F_UTF8
1318 ? dec.cur - SvPVX (string)
1319 : utf8_distance (dec.cur, SvPVX (string));
1320
1321 if (offset_return)
1322 *offset_return = offset;
1323 }
1324
947 if (!sv) 1325 if (!sv)
948 { 1326 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1327 SV *uni = sv_newmortal ();
953 1328
954 // horrible hack to silence warning inside pv_uni_display 1329 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1330 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1331 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1333 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1334 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1335 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1336 LEAVE;
962 1337
963 croak ("%s, at character offset %d (%s)", 1338 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1339 dec.err,
965 (int)offset, 1340 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1341 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1342 }
968 1343
969 sv = sv_2mortal (sv); 1344 sv = sv_2mortal (sv);
970 1345
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1346 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
972 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1347 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1348
974 return sv; 1349 return sv;
975} 1350}
976 1351
… …
981 1356
982BOOT: 1357BOOT:
983{ 1358{
984 int i; 1359 int i;
985 1360
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1361 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1362 decode_hexdigit [i] =
1363 i >= '0' && i <= '9' ? i - '0'
1364 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1365 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1366 : -1;
989 1367
990 for (i = 7; i--; ) 1368 json_stash = gv_stashpv ("JSON::XS" , 1);
1369 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1370
1371 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1372 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1373}
1374
1375PROTOTYPES: DISABLE
1376
1377void CLONE (...)
1378 CODE:
1379 json_stash = 0;
1380 json_boolean_stash = 0;
1381
1382void new (char *klass)
1383 PPCODE:
1384{
1385 SV *pv = NEWSV (0, sizeof (JSON));
1386 SvPOK_only (pv);
1387 Zero (SvPVX (pv), 1, JSON);
1388 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1389 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1390}
1391
1392void ascii (JSON *self, int enable = 1)
1393 ALIAS:
1394 ascii = F_ASCII
1395 latin1 = F_LATIN1
1396 utf8 = F_UTF8
1397 indent = F_INDENT
1398 canonical = F_CANONICAL
1399 space_before = F_SPACE_BEFORE
1400 space_after = F_SPACE_AFTER
1401 pretty = F_PRETTY
1402 allow_nonref = F_ALLOW_NONREF
1403 shrink = F_SHRINK
1404 allow_blessed = F_ALLOW_BLESSED
1405 convert_blessed = F_CONV_BLESSED
1406 PPCODE:
1407{
1408 if (enable)
1409 self->flags |= ix;
1410 else
1411 self->flags &= ~ix;
1412
1413 XPUSHs (ST (0));
1414}
1415
1416void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1417 PPCODE:
1418{
1419 UV log2 = 0;
1420
1421 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1422
1423 while ((1UL << log2) < max_depth)
1424 ++log2;
1425
1426 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1427
1428 XPUSHs (ST (0));
1429}
1430
1431void max_size (JSON *self, UV max_size = 0)
1432 PPCODE:
1433{
1434 UV log2 = 0;
1435
1436 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1437 if (max_size == 1) max_size = 2;
1438
1439 while ((1UL << log2) < max_size)
1440 ++log2;
1441
1442 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1443
1444 XPUSHs (ST (0));
1445}
1446
1447void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1448 PPCODE:
1449{
1450 SvREFCNT_dec (self->cb_object);
1451 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1452
1453 XPUSHs (ST (0));
1454}
1455
1456void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1457 PPCODE:
1458{
1459 if (!self->cb_sk_object)
1460 self->cb_sk_object = newHV ();
1461
1462 if (SvOK (cb))
1463 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1464 else
991 { 1465 {
992 decode_hexdigit ['a' + i] = 10 + i; 1466 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
993 decode_hexdigit ['A' + i] = 10 + i; 1467
1468 if (!HvKEYS (self->cb_sk_object))
1469 {
1470 SvREFCNT_dec (self->cb_sk_object);
1471 self->cb_sk_object = 0;
1472 }
994 } 1473 }
995 1474
996 json_stash = gv_stashpv ("JSON::XS", 1); 1475 XPUSHs (ST (0));
997} 1476}
998 1477
999PROTOTYPES: DISABLE 1478void encode (JSON *self, SV *scalar)
1479 PPCODE:
1480 XPUSHs (encode_json (scalar, self));
1000 1481
1001SV *new (char *dummy) 1482void decode (JSON *self, SV *jsonstr)
1483 PPCODE:
1484 XPUSHs (decode_json (jsonstr, self, 0));
1485
1486void decode_prefix (JSON *self, SV *jsonstr)
1487 PPCODE:
1488{
1489 UV offset;
1490 EXTEND (SP, 2);
1491 PUSHs (decode_json (jsonstr, self, &offset));
1492 PUSHs (sv_2mortal (newSVuv (offset)));
1493}
1494
1495void DESTROY (JSON *self)
1002 CODE: 1496 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1497 SvREFCNT_dec (self->cb_sk_object);
1004 OUTPUT: 1498 SvREFCNT_dec (self->cb_object);
1005 RETVAL
1006
1007SV *ascii (SV *self, int enable = 1)
1008 ALIAS:
1009 ascii = F_ASCII
1010 utf8 = F_UTF8
1011 indent = F_INDENT
1012 canonical = F_CANONICAL
1013 space_before = F_SPACE_BEFORE
1014 space_after = F_SPACE_AFTER
1015 pretty = F_PRETTY
1016 allow_nonref = F_ALLOW_NONREF
1017 shrink = F_SHRINK
1018 CODE:
1019{
1020 UV *uv = SvJSON (self);
1021 if (enable)
1022 *uv |= ix;
1023 else
1024 *uv &= ~ix;
1025
1026 RETVAL = newSVsv (self);
1027}
1028 OUTPUT:
1029 RETVAL
1030
1031void encode (SV *self, SV *scalar)
1032 PPCODE:
1033 XPUSHs (encode_json (scalar, *SvJSON (self)));
1034
1035void decode (SV *self, SV *jsonstr)
1036 PPCODE:
1037 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
1038 1499
1039PROTOTYPES: ENABLE 1500PROTOTYPES: ENABLE
1040 1501
1041void to_json (SV *scalar) 1502void to_json (SV *scalar)
1042 PPCODE: 1503 PPCODE:
1504{
1505 JSON json = { F_DEFAULT | F_UTF8 };
1043 XPUSHs (encode_json (scalar, F_UTF8)); 1506 XPUSHs (encode_json (scalar, &json));
1507}
1044 1508
1045void from_json (SV *jsonstr) 1509void from_json (SV *jsonstr)
1046 PPCODE: 1510 PPCODE:
1511{
1512 JSON json = { F_DEFAULT | F_UTF8 };
1047 XPUSHs (decode_json (jsonstr, F_UTF8)); 1513 XPUSHs (decode_json (jsonstr, &json, 0));
1514}
1048 1515

Diff Legend

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