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 1.15 by root, Sun Mar 25 02:32:40 2007 UTC vs.
Revision 1.57 by root, Mon Aug 13 16:05:42 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#else
66# define JSON_SLOW 1
67#endif
68
30static HV *json_stash; // JSON::XS:: 69static HV *json_stash, *json_boolean_stash; // JSON::XS::
70static SV *json_true, *json_false;
71
72typedef struct {
73 U32 flags;
74 SV *cb_object;
75 HV *cb_sk_object;
76} JSON;
31 77
32///////////////////////////////////////////////////////////////////////////// 78/////////////////////////////////////////////////////////////////////////////
33// utility functions 79// utility functions
34 80
35static UV * 81inline 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) 82shrink (SV *sv)
46{ 83{
47 sv_utf8_downgrade (sv, 1); 84 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 85 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 86 {
… …
58// decode an utf-8 character and return it, or (UV)-1 in 95// decode an utf-8 character and return it, or (UV)-1 in
59// case of an error. 96// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF, 97// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else. 98// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints 99// note that we never call this function for a ascii codepoints
63static UV 100inline UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 101decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{ 102{
66 if (s[0] > 0xdf || s[0] < 0xc2) 103 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 104 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 105 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69 { 106 {
70 *clen = 2; 107 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 108 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 109 }
73 else 110 else
111 {
112 *clen = (STRLEN)-1;
74 return (UV)-1; 113 return (UV)-1;
114 }
75} 115}
76 116
77///////////////////////////////////////////////////////////////////////////// 117/////////////////////////////////////////////////////////////////////////////
78// encoder 118// encoder
79 119
… …
81typedef struct 121typedef struct
82{ 122{
83 char *cur; // SvPVX (sv) + current output position 123 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 124 char *end; // SvEND (sv)
85 SV *sv; // result scalar 125 SV *sv; // result scalar
86 UV flags; // F_* 126 JSON json;
87 int indent; // indentation level 127 U32 indent; // indentation level
88 int max_depth; // max. recursion level 128 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 129} enc_t;
90 130
91static void 131inline void
92need (enc_t *enc, STRLEN len) 132need (enc_t *enc, STRLEN len)
93{ 133{
94 if (enc->cur + len >= enc->end) 134 if (expect_false (enc->cur + len >= enc->end))
95 { 135 {
96 STRLEN cur = enc->cur - SvPVX (enc->sv); 136 STRLEN cur = enc->cur - SvPVX (enc->sv);
97 SvGROW (enc->sv, cur + len + 1); 137 SvGROW (enc->sv, cur + len + 1);
98 enc->cur = SvPVX (enc->sv) + cur; 138 enc->cur = SvPVX (enc->sv) + cur;
99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 139 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100 } 140 }
101} 141}
102 142
103static void 143inline void
104encode_ch (enc_t *enc, char ch) 144encode_ch (enc_t *enc, char ch)
105{ 145{
106 need (enc, 1); 146 need (enc, 1);
107 *enc->cur++ = ch; 147 *enc->cur++ = ch;
108} 148}
… …
116 156
117 while (str < end) 157 while (str < end)
118 { 158 {
119 unsigned char ch = *(unsigned char *)str; 159 unsigned char ch = *(unsigned char *)str;
120 160
121 if (ch >= 0x20 && ch < 0x80) // most common case 161 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122 { 162 {
123 if (ch == '"') // but with slow exceptions 163 if (expect_false (ch == '"')) // but with slow exceptions
124 { 164 {
125 need (enc, len += 1); 165 need (enc, len += 1);
126 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
127 *enc->cur++ = '"'; 167 *enc->cur++ = '"';
128 } 168 }
129 else if (ch == '\\') 169 else if (expect_false (ch == '\\'))
130 { 170 {
131 need (enc, len += 1); 171 need (enc, len += 1);
132 *enc->cur++ = '\\'; 172 *enc->cur++ = '\\';
133 *enc->cur++ = '\\'; 173 *enc->cur++ = '\\';
134 } 174 }
… …
152 STRLEN clen; 192 STRLEN clen;
153 UV uch; 193 UV uch;
154 194
155 if (is_utf8) 195 if (is_utf8)
156 { 196 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen); 197 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1) 198 if (clen == (STRLEN)-1)
160 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 199 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
161 } 200 }
162 else 201 else
… …
166 } 205 }
167 206
168 if (uch > 0x10FFFFUL) 207 if (uch > 0x10FFFFUL)
169 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 208 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170 209
171 if (uch < 0x80 || enc->flags & F_ASCII) 210 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
172 { 211 {
173 if (uch > 0xFFFFUL) 212 if (uch > 0xFFFFUL)
174 { 213 {
175 need (enc, len += 11); 214 need (enc, len += 11);
176 sprintf (enc->cur, "\\u%04x\\u%04x", 215 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
190 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 229 *enc->cur++ = hexdigit [(uch >> 0) & 15];
191 } 230 }
192 231
193 str += clen; 232 str += clen;
194 } 233 }
234 else if (enc->json.flags & F_LATIN1)
235 {
236 *enc->cur++ = uch;
237 str += clen;
238 }
195 else if (is_utf8) 239 else if (is_utf8)
196 { 240 {
197 need (enc, len += clen); 241 need (enc, len += clen);
198 do 242 do
199 { 243 {
… …
201 } 245 }
202 while (--clen); 246 while (--clen);
203 } 247 }
204 else 248 else
205 { 249 {
206 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 250 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 251 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str; 252 ++str;
209 } 253 }
210 } 254 }
211 } 255 }
… …
213 257
214 --len; 258 --len;
215 } 259 }
216} 260}
217 261
218static void 262inline void
219encode_indent (enc_t *enc) 263encode_indent (enc_t *enc)
220{ 264{
221 if (enc->flags & F_INDENT) 265 if (enc->json.flags & F_INDENT)
222 { 266 {
223 int spaces = enc->indent * INDENT_STEP; 267 int spaces = enc->indent * INDENT_STEP;
224 268
225 need (enc, spaces); 269 need (enc, spaces);
226 memset (enc->cur, ' ', spaces); 270 memset (enc->cur, ' ', spaces);
227 enc->cur += spaces; 271 enc->cur += spaces;
228 } 272 }
229} 273}
230 274
231static void 275inline void
232encode_space (enc_t *enc) 276encode_space (enc_t *enc)
233{ 277{
234 need (enc, 1); 278 need (enc, 1);
235 encode_ch (enc, ' '); 279 encode_ch (enc, ' ');
236} 280}
237 281
238static void 282inline void
239encode_nl (enc_t *enc) 283encode_nl (enc_t *enc)
240{ 284{
241 if (enc->flags & F_INDENT) 285 if (enc->json.flags & F_INDENT)
242 { 286 {
243 need (enc, 1); 287 need (enc, 1);
244 encode_ch (enc, '\n'); 288 encode_ch (enc, '\n');
245 } 289 }
246} 290}
247 291
248static void 292inline void
249encode_comma (enc_t *enc) 293encode_comma (enc_t *enc)
250{ 294{
251 encode_ch (enc, ','); 295 encode_ch (enc, ',');
252 296
253 if (enc->flags & F_INDENT) 297 if (enc->json.flags & F_INDENT)
254 encode_nl (enc); 298 encode_nl (enc);
255 else if (enc->flags & F_SPACE_AFTER) 299 else if (enc->json.flags & F_SPACE_AFTER)
256 encode_space (enc); 300 encode_space (enc);
257} 301}
258 302
259static void encode_sv (enc_t *enc, SV *sv); 303static void encode_sv (enc_t *enc, SV *sv);
260 304
261static void 305static void
262encode_av (enc_t *enc, AV *av) 306encode_av (enc_t *enc, AV *av)
263{ 307{
264 int i, len = av_len (av); 308 int i, len = av_len (av);
265 309
310 if (enc->indent >= enc->maxdepth)
311 croak ("data structure too deep (hit recursion limit)");
312
266 encode_ch (enc, '['); encode_nl (enc); 313 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 314 ++enc->indent;
268 315
269 for (i = 0; i <= len; ++i) 316 for (i = 0; i <= len; ++i)
270 { 317 {
318 SV **svp = av_fetch (av, i, 0);
319
271 encode_indent (enc); 320 encode_indent (enc);
272 encode_sv (enc, *av_fetch (av, i, 0)); 321
322 if (svp)
323 encode_sv (enc, *svp);
324 else
325 encode_str (enc, "null", 4, 0);
273 326
274 if (i < len) 327 if (i < len)
275 encode_comma (enc); 328 encode_comma (enc);
276 } 329 }
277 330
… …
300 else 353 else
301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 354 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
302 355
303 encode_ch (enc, '"'); 356 encode_ch (enc, '"');
304 357
305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 358 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
306 encode_ch (enc, ':'); 359 encode_ch (enc, ':');
307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 360 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
308 encode_sv (enc, HeVAL (he)); 361 encode_sv (enc, HeVAL (he));
309} 362}
310 363
311// compare hash entries, used when all keys are bytestrings 364// compare hash entries, used when all keys are bytestrings
312static int 365static int
… …
335 388
336static void 389static void
337encode_hv (enc_t *enc, HV *hv) 390encode_hv (enc_t *enc, HV *hv)
338{ 391{
339 int count, i; 392 int count, i;
393
394 if (enc->indent >= enc->maxdepth)
395 croak ("data structure too deep (hit recursion limit)");
340 396
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 397 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 398
343 if ((count = hv_iterinit (hv))) 399 if ((count = hv_iterinit (hv)))
344 { 400 {
345 // for canonical output we have to sort by keys first 401 // for canonical output we have to sort by keys first
346 // actually, this is mostly due to the stupid so-called 402 // actually, this is mostly due to the stupid so-called
347 // security workaround added somewhere in 5.8.x. 403 // security workaround added somewhere in 5.8.x.
348 // that randomises hash orderings 404 // that randomises hash orderings
349 if (enc->flags & F_CANONICAL) 405 if (enc->json.flags & F_CANONICAL)
350 { 406 {
351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
352 int fast = 1; 407 int fast = 1;
408 HE *he;
409#if defined(__BORLANDC__) || defined(_MSC_VER)
410 HE **hes = _alloca (count * sizeof (HE));
411#else
412 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
413#endif
353 414
354 i = 0; 415 i = 0;
355 while ((he = hv_iternext (hv))) 416 while ((he = hv_iternext (hv)))
356 { 417 {
357 hes [i++] = he; 418 hes [i++] = he;
… …
392 453
393 encode_nl (enc); 454 encode_nl (enc);
394 } 455 }
395 else 456 else
396 { 457 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 458 HE *he = hv_iternext (hv);
399 459
400 for (;;) 460 for (;;)
401 { 461 {
402 encode_indent (enc); 462 encode_indent (enc);
… …
411 encode_nl (enc); 471 encode_nl (enc);
412 } 472 }
413 } 473 }
414 474
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 475 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
476}
477
478// encode objects, arrays and special \0=false and \1=true values.
479static void
480encode_rv (enc_t *enc, SV *sv)
481{
482 svtype svt;
483
484 SvGETMAGIC (sv);
485 svt = SvTYPE (sv);
486
487 if (expect_false (SvOBJECT (sv)))
488 {
489 HV *stash = !JSON_SLOW || json_boolean_stash
490 ? json_boolean_stash
491 : gv_stashpv ("JSON::XS::Boolean", 1);
492
493 if (SvSTASH (sv) == stash)
494 {
495 if (SvIV (sv))
496 encode_str (enc, "true", 4, 0);
497 else
498 encode_str (enc, "false", 5, 0);
499 }
500 else
501 {
502#if 0
503 if (0 && sv_derived_from (rv, "JSON::Literal"))
504 {
505 // not yet
506 }
507#endif
508 if (enc->json.flags & F_CONV_BLESSED)
509 {
510 // we re-bless the reference to get overload and other niceties right
511 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
512
513 if (to_json)
514 {
515 int count;
516 dSP;
517
518 ENTER; SAVETMPS; PUSHMARK (SP);
519 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
520
521 // calling with G_SCALAR ensures that we always get a 1 return value
522 PUTBACK;
523 call_sv ((SV *)GvCV (to_json), G_SCALAR);
524 SPAGAIN;
525
526 // catch this surprisingly common error
527 if (SvROK (TOPs) && SvRV (TOPs) == sv)
528 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
529
530 sv = POPs;
531 PUTBACK;
532
533 encode_sv (enc, sv);
534
535 FREETMPS; LEAVE;
536 }
537 else if (enc->json.flags & F_ALLOW_BLESSED)
538 encode_str (enc, "null", 4, 0);
539 else
540 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
541 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
542 }
543 else if (enc->json.flags & F_ALLOW_BLESSED)
544 encode_str (enc, "null", 4, 0);
545 else
546 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
547 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
548 }
549 }
550 else if (svt == SVt_PVHV)
551 encode_hv (enc, (HV *)sv);
552 else if (svt == SVt_PVAV)
553 encode_av (enc, (AV *)sv);
554 else if (svt < SVt_PVAV)
555 {
556 STRLEN len = 0;
557 char *pv = svt ? SvPV (sv, len) : 0;
558
559 if (len == 1 && *pv == '1')
560 encode_str (enc, "true", 4, 0);
561 else if (len == 1 && *pv == '0')
562 encode_str (enc, "false", 5, 0);
563 else
564 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
565 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
566 }
567 else
568 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
569 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 570}
417 571
418static void 572static void
419encode_sv (enc_t *enc, SV *sv) 573encode_sv (enc_t *enc, SV *sv)
420{ 574{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 582 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 583 encode_ch (enc, '"');
430 } 584 }
431 else if (SvNOKp (sv)) 585 else if (SvNOKp (sv))
432 { 586 {
587 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 588 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 589 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 590 enc->cur += strlen (enc->cur);
436 } 591 }
437 else if (SvIOKp (sv)) 592 else if (SvIOKp (sv))
438 { 593 {
439 need (enc, 64); 594 // we assume we can always read an IV as a UV
595 if (SvUV (sv) & ~(UV)0x7fff)
596 {
597 // large integer, use the (rather slow) snprintf way.
598 need (enc, sizeof (UV) * 3);
440 enc->cur += 599 enc->cur +=
441 SvIsUV(sv) 600 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 601 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 602 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
603 }
604 else
605 {
606 // optimise the "small number case"
607 // code will likely be branchless and use only a single multiplication
608 I32 i = SvIV (sv);
609 U32 u;
610 char digit, nz = 0;
611
612 need (enc, 6);
613
614 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
615 u = i < 0 ? -i : i;
616
617 // convert to 4.28 fixed-point representation
618 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
619
620 // now output digit by digit, each time masking out the integer part
621 // and multiplying by 5 while moving the decimal point one to the right,
622 // resulting in a net multiplication by 10.
623 // we always write the digit to memory but conditionally increment
624 // the pointer, to ease the usage of conditional move instructions.
625 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
626 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
627 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
628 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
629 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
630 }
444 } 631 }
445 else if (SvROK (sv)) 632 else if (SvROK (sv))
446 { 633 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)) 634 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 635 encode_str (enc, "null", 4, 0);
464 else 636 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 637 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 638 SvPV_nolen (sv), SvFLAGS (sv));
467} 639}
468 640
469static SV * 641static SV *
470encode_json (SV *scalar, UV flags) 642encode_json (SV *scalar, JSON *json)
471{ 643{
644 enc_t enc;
645
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 646 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 647 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 648
475 enc_t enc; 649 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 650 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 651 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 652 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 653 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 654 enc.maxdepth = DEC_DEPTH (enc.json.flags);
482 655
483 SvPOK_only (enc.sv); 656 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 657 encode_sv (&enc, scalar);
485 658
659 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
660 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
661
486 if (!(flags & (F_ASCII | F_UTF8))) 662 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 663 SvUTF8_on (enc.sv);
488 664
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 665 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 666 shrink (enc.sv);
493 667
494 return enc.sv; 668 return enc.sv;
495} 669}
496 670
… …
501typedef struct 675typedef struct
502{ 676{
503 char *cur; // current parser pointer 677 char *cur; // current parser pointer
504 char *end; // end of input string 678 char *end; // end of input string
505 const char *err; // parse error, if != 0 679 const char *err; // parse error, if != 0
506 UV flags; // F_* 680 JSON json;
681 U32 depth; // recursion depth
682 U32 maxdepth; // recursion depth limit
507} dec_t; 683} dec_t;
508 684
509static void 685inline void
510decode_ws (dec_t *dec) 686decode_ws (dec_t *dec)
511{ 687{
512 for (;;) 688 for (;;)
513 { 689 {
514 char ch = *dec->cur; 690 char ch = *dec->cur;
… …
520 ++dec->cur; 696 ++dec->cur;
521 } 697 }
522} 698}
523 699
524#define ERR(reason) SB dec->err = reason; goto fail; SE 700#define ERR(reason) SB dec->err = reason; goto fail; SE
701
525#define EXPECT_CH(ch) SB \ 702#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 703 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 704 ERR (# ch " expected"); \
528 ++dec->cur; \ 705 ++dec->cur; \
529 SE 706 SE
530 707
708#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
709#define DEC_DEC_DEPTH --dec->depth
710
531static SV *decode_sv (dec_t *dec); 711static SV *decode_sv (dec_t *dec);
532 712
533static signed char decode_hexdigit[256]; 713static signed char decode_hexdigit[256];
534 714
535static UV 715static UV
536decode_4hex (dec_t *dec) 716decode_4hex (dec_t *dec)
537{ 717{
538 signed char d1, d2, d3, d4; 718 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 719 unsigned char *cur = (unsigned char *)dec->cur;
540 720
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 721 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"); 722 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"); 723 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"); 724 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 725
546 dec->cur += 4; 726 dec->cur += 4;
547 727
548 return ((UV)d1) << 12 728 return ((UV)d1) << 12
549 | ((UV)d2) << 8 729 | ((UV)d2) << 8
… …
557static SV * 737static SV *
558decode_str (dec_t *dec) 738decode_str (dec_t *dec)
559{ 739{
560 SV *sv = 0; 740 SV *sv = 0;
561 int utf8 = 0; 741 int utf8 = 0;
742 char *dec_cur = dec->cur;
562 743
563 do 744 do
564 { 745 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 746 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 747 char *cur = buf;
567 748
568 do 749 do
569 { 750 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 751 unsigned char ch = *(unsigned char *)dec_cur++;
571 752
572 if (ch == '"') 753 if (expect_false (ch == '"'))
573 { 754 {
574 --dec->cur; 755 --dec_cur;
575 break; 756 break;
576 } 757 }
577 else if (ch == '\\') 758 else if (expect_false (ch == '\\'))
578 { 759 {
579 switch (*dec->cur) 760 switch (*dec_cur)
580 { 761 {
581 case '\\': 762 case '\\':
582 case '/': 763 case '/':
583 case '"': *cur++ = *dec->cur++; break; 764 case '"': *cur++ = *dec_cur++; break;
584 765
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 766 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 767 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 768 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 769 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 770 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 771
591 case 'u': 772 case 'u':
592 { 773 {
593 UV lo, hi; 774 UV lo, hi;
594 ++dec->cur; 775 ++dec_cur;
595 776
777 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 778 hi = decode_4hex (dec);
779 dec_cur = dec->cur;
597 if (hi == (UV)-1) 780 if (hi == (UV)-1)
598 goto fail; 781 goto fail;
599 782
600 // possibly a surrogate pair 783 // possibly a surrogate pair
601 if (hi >= 0xd800) 784 if (hi >= 0xd800)
602 if (hi < 0xdc00) 785 if (hi < 0xdc00)
603 { 786 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 787 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 788 ERR ("missing low surrogate character in surrogate pair");
606 789
607 dec->cur += 2; 790 dec_cur += 2;
608 791
792 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 793 lo = decode_4hex (dec);
794 dec_cur = dec->cur;
610 if (lo == (UV)-1) 795 if (lo == (UV)-1)
611 goto fail; 796 goto fail;
612 797
613 if (lo < 0xdc00 || lo >= 0xe000) 798 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 799 ERR ("surrogate pair expected");
… …
628 *cur++ = hi; 813 *cur++ = hi;
629 } 814 }
630 break; 815 break;
631 816
632 default: 817 default:
633 --dec->cur; 818 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 819 ERR ("illegal backslash escape sequence in string");
635 } 820 }
636 } 821 }
637 else if (ch >= 0x20 && ch <= 0x7f) 822 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638 *cur++ = ch; 823 *cur++ = ch;
639 else if (ch >= 0x80) 824 else if (ch >= 0x80)
640 { 825 {
641 --dec->cur;
642
643 STRLEN clen; 826 STRLEN clen;
827 UV uch;
828
829 --dec_cur;
830
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 831 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 832 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 833 ERR ("malformed UTF-8 character in JSON string");
647 834
648 do 835 do
649 {
650 *cur++ = *dec->cur++; 836 *cur++ = *dec_cur++;
651 }
652 while (--clen); 837 while (--clen);
653 838
654 utf8 = 1; 839 utf8 = 1;
655 } 840 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 841 else
842 {
843 --dec_cur;
844
845 if (!ch)
846 ERR ("unexpected end of string while parsing JSON string");
847 else
659 ERR ("invalid character encountered"); 848 ERR ("invalid character encountered while parsing JSON string");
660 849 }
661 } 850 }
662 while (cur < buf + SHORT_STRING_LEN); 851 while (cur < buf + SHORT_STRING_LEN);
663 852
853 {
664 STRLEN len = cur - buf; 854 STRLEN len = cur - buf;
665 855
666 if (sv) 856 if (sv)
667 { 857 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 858 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 859 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 860 SvCUR_set (sv, SvCUR (sv) + len);
671 } 861 }
672 else 862 else
673 sv = newSVpvn (buf, len); 863 sv = newSVpvn (buf, len);
674 } 864 }
865 }
675 while (*dec->cur != '"'); 866 while (*dec_cur != '"');
676 867
677 ++dec->cur; 868 ++dec_cur;
678 869
679 if (sv) 870 if (sv)
680 { 871 {
681 SvPOK_only (sv); 872 SvPOK_only (sv);
682 *SvEND (sv) = 0; 873 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 876 SvUTF8_on (sv);
686 } 877 }
687 else 878 else
688 sv = newSVpvn ("", 0); 879 sv = newSVpvn ("", 0);
689 880
881 dec->cur = dec_cur;
690 return sv; 882 return sv;
691 883
692fail: 884fail:
885 dec->cur = dec_cur;
693 return 0; 886 return 0;
694} 887}
695 888
696static SV * 889static SV *
697decode_num (dec_t *dec) 890decode_num (dec_t *dec)
… …
755 is_nv = 1; 948 is_nv = 1;
756 } 949 }
757 950
758 if (!is_nv) 951 if (!is_nv)
759 { 952 {
760 UV uv; 953 int len = dec->cur - start;
761 int numtype = grok_number (start, dec->cur - start, &uv); 954
762 if (numtype & IS_NUMBER_IN_UV) 955 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
763 if (numtype & IS_NUMBER_NEG) 956 if (*start == '-')
957 switch (len)
764 { 958 {
765 if (uv < (UV)IV_MIN) 959 case 2: return newSViv (-( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 960 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
961 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
962 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767 } 963 }
964 else
965 switch (len)
966 {
967 case 1: return newSViv ( start [0] - '0' * 1);
968 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
969 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
970 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
971 }
972
973 {
974 UV uv;
975 int numtype = grok_number (start, len, &uv);
976 if (numtype & IS_NUMBER_IN_UV)
977 if (numtype & IS_NUMBER_NEG)
978 {
979 if (uv < (UV)IV_MIN)
980 return newSViv (-(IV)uv);
981 }
768 else 982 else
769 return newSVuv (uv); 983 return newSVuv (uv);
770 } 984 }
771 985
986 len -= *start == '-' ? 1 : 0;
987
988 // does not fit into IV or UV, try NV
989 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
990 #if defined (LDBL_DIG)
991 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
992 #endif
993 )
994 // fits into NV without loss of precision
995 return newSVnv (Atof (start));
996
997 // everything else fails, convert it to a string
998 return newSVpvn (start, dec->cur - start);
999 }
1000
1001 // loss of precision here
772 return newSVnv (Atof (start)); 1002 return newSVnv (Atof (start));
773 1003
774fail: 1004fail:
775 return 0; 1005 return 0;
776} 1006}
… …
778static SV * 1008static SV *
779decode_av (dec_t *dec) 1009decode_av (dec_t *dec)
780{ 1010{
781 AV *av = newAV (); 1011 AV *av = newAV ();
782 1012
1013 DEC_INC_DEPTH;
783 decode_ws (dec); 1014 decode_ws (dec);
1015
784 if (*dec->cur == ']') 1016 if (*dec->cur == ']')
785 ++dec->cur; 1017 ++dec->cur;
786 else 1018 else
787 for (;;) 1019 for (;;)
788 { 1020 {
… …
806 ERR (", or ] expected while parsing array"); 1038 ERR (", or ] expected while parsing array");
807 1039
808 ++dec->cur; 1040 ++dec->cur;
809 } 1041 }
810 1042
1043 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1044 return newRV_noinc ((SV *)av);
812 1045
813fail: 1046fail:
814 SvREFCNT_dec (av); 1047 SvREFCNT_dec (av);
1048 DEC_DEC_DEPTH;
815 return 0; 1049 return 0;
816} 1050}
817 1051
818static SV * 1052static SV *
819decode_hv (dec_t *dec) 1053decode_hv (dec_t *dec)
820{ 1054{
1055 SV *sv;
821 HV *hv = newHV (); 1056 HV *hv = newHV ();
822 1057
1058 DEC_INC_DEPTH;
823 decode_ws (dec); 1059 decode_ws (dec);
1060
824 if (*dec->cur == '}') 1061 if (*dec->cur == '}')
825 ++dec->cur; 1062 ++dec->cur;
826 else 1063 else
827 for (;;) 1064 for (;;)
828 { 1065 {
829 SV *key, *value;
830
831 decode_ws (dec); EXPECT_CH ('"'); 1066 decode_ws (dec); EXPECT_CH ('"');
832 1067
833 key = decode_str (dec); 1068 // heuristic: assume that
834 if (!key) 1069 // a) decode_str + hv_store_ent are abysmally slow.
835 goto fail; 1070 // b) most hash keys are short, simple ascii text.
1071 // => try to "fast-match" such strings to avoid
1072 // the overhead of decode_str + hv_store_ent.
1073 {
1074 SV *value;
1075 char *p = dec->cur;
1076 char *e = p + 24; // only try up to 24 bytes
836 1077
837 decode_ws (dec); EXPECT_CH (':'); 1078 for (;;)
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1079 {
1080 // the >= 0x80 is true on most architectures
1081 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1082 {
1083 // slow path, back up and use decode_str
1084 SV *key = decode_str (dec);
1085 if (!key)
1086 goto fail;
1087
1088 decode_ws (dec); EXPECT_CH (':');
1089
1090 value = decode_sv (dec);
1091 if (!value)
1092 {
1093 SvREFCNT_dec (key);
1094 goto fail;
1095 }
1096
1097 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1098 SvREFCNT_dec (key);
1099
1100 break;
1101 }
1102 else if (*p == '"')
1103 {
1104 // fast path, got a simple key
1105 char *key = dec->cur;
1106 int len = p - key;
1107 dec->cur = p + 1;
1108
1109 decode_ws (dec); EXPECT_CH (':');
1110
1111 value = decode_sv (dec);
1112 if (!value)
843 goto fail; 1113 goto fail;
1114
1115 hv_store (hv, key, len, value, 0);
1116
1117 break;
1118 }
1119
1120 ++p;
844 } 1121 }
845 1122 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1123
849 decode_ws (dec); 1124 decode_ws (dec);
850 1125
851 if (*dec->cur == '}') 1126 if (*dec->cur == '}')
852 { 1127 {
… …
858 ERR (", or } expected while parsing object/hash"); 1133 ERR (", or } expected while parsing object/hash");
859 1134
860 ++dec->cur; 1135 ++dec->cur;
861 } 1136 }
862 1137
1138 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 1139 sv = newRV_noinc ((SV *)hv);
1140
1141 // check filter callbacks
1142 if (dec->json.flags & F_HOOK)
1143 {
1144 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1145 {
1146 HE *cb, *he;
1147
1148 hv_iterinit (hv);
1149 he = hv_iternext (hv);
1150 hv_iterinit (hv);
1151
1152 // the next line creates a mortal sv each time its called.
1153 // might want to optimise this for common cases.
1154 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1155
1156 if (cb)
1157 {
1158 dSP;
1159 int count;
1160
1161 ENTER; SAVETMPS; PUSHMARK (SP);
1162 XPUSHs (HeVAL (he));
1163
1164 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1165
1166 if (count == 1)
1167 {
1168 sv = newSVsv (POPs);
1169 FREETMPS; LEAVE;
1170 return sv;
1171 }
1172
1173 FREETMPS; LEAVE;
1174 }
1175 }
1176
1177 if (dec->json.cb_object)
1178 {
1179 dSP;
1180 int count;
1181
1182 ENTER; SAVETMPS; PUSHMARK (SP);
1183 XPUSHs (sv_2mortal (sv));
1184
1185 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1186
1187 if (count == 1)
1188 {
1189 sv = newSVsv (POPs);
1190 FREETMPS; LEAVE;
1191 return sv;
1192 }
1193
1194 SvREFCNT_inc (sv);
1195 FREETMPS; LEAVE;
1196 }
1197 }
1198
1199 return sv;
864 1200
865fail: 1201fail:
866 SvREFCNT_dec (hv); 1202 SvREFCNT_dec (hv);
1203 DEC_DEC_DEPTH;
867 return 0; 1204 return 0;
868} 1205}
869 1206
870static SV * 1207static SV *
871decode_sv (dec_t *dec) 1208decode_sv (dec_t *dec)
872{ 1209{
873 decode_ws (dec); 1210 decode_ws (dec);
1211
1212 // the beauty of JSON: you need exactly one character lookahead
1213 // to parse anything.
874 switch (*dec->cur) 1214 switch (*dec->cur)
875 { 1215 {
876 case '"': ++dec->cur; return decode_str (dec); 1216 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1217 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1218 case '{': ++dec->cur; return decode_hv (dec);
… …
884 1224
885 case 't': 1225 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1226 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1227 {
888 dec->cur += 4; 1228 dec->cur += 4;
889 return newSViv (1); 1229 return SvREFCNT_inc (json_true);
890 } 1230 }
891 else 1231 else
892 ERR ("'true' expected"); 1232 ERR ("'true' expected");
893 1233
894 break; 1234 break;
895 1235
896 case 'f': 1236 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1237 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1238 {
899 dec->cur += 5; 1239 dec->cur += 5;
900 return newSViv (0); 1240 return SvREFCNT_inc (json_false);
901 } 1241 }
902 else 1242 else
903 ERR ("'false' expected"); 1243 ERR ("'false' expected");
904 1244
905 break; 1245 break;
… …
914 ERR ("'null' expected"); 1254 ERR ("'null' expected");
915 1255
916 break; 1256 break;
917 1257
918 default: 1258 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1259 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1260 break;
921 } 1261 }
922 1262
923fail: 1263fail:
924 return 0; 1264 return 0;
925} 1265}
926 1266
927static SV * 1267static SV *
928decode_json (SV *string, UV flags) 1268decode_json (SV *string, JSON *json, UV *offset_return)
929{ 1269{
1270 dec_t dec;
1271 UV offset;
930 SV *sv; 1272 SV *sv;
931 1273
1274 SvGETMAGIC (string);
1275 SvUPGRADE (string, SVt_PV);
1276
1277 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1278 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1279 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1280
932 if (flags & F_UTF8) 1281 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1282 sv_utf8_downgrade (string, 0);
934 else 1283 else
935 sv_utf8_upgrade (string); 1284 sv_utf8_upgrade (string);
936 1285
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1286 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1287
939 dec_t dec; 1288 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1289 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1290 dec.end = SvEND (string);
943 dec.err = 0; 1291 dec.err = 0;
1292 dec.depth = 0;
1293 dec.maxdepth = DEC_DEPTH (dec.json.flags);
944 1294
1295 if (dec.json.cb_object || dec.json.cb_sk_object)
1296 dec.json.flags |= F_HOOK;
1297
1298 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
945 sv = decode_sv (&dec); 1299 sv = decode_sv (&dec);
946 1300
1301 if (!(offset_return || !sv))
1302 {
1303 // check for trailing garbage
1304 decode_ws (&dec);
1305
1306 if (*dec.cur)
1307 {
1308 dec.err = "garbage after JSON object";
1309 SvREFCNT_dec (sv);
1310 sv = 0;
1311 }
1312 }
1313
1314 if (offset_return || !sv)
1315 {
1316 offset = dec.json.flags & F_UTF8
1317 ? dec.cur - SvPVX (string)
1318 : utf8_distance (dec.cur, SvPVX (string));
1319
1320 if (offset_return)
1321 *offset_return = offset;
1322 }
1323
947 if (!sv) 1324 if (!sv)
948 { 1325 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1326 SV *uni = sv_newmortal ();
953 1327
954 // horrible hack to silence warning inside pv_uni_display 1328 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1329 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1330 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1332 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1333 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1334 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1335 LEAVE;
962 1336
963 croak ("%s, at character offset %d (%s)", 1337 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1338 dec.err,
965 (int)offset, 1339 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1340 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1341 }
968 1342
969 sv = sv_2mortal (sv); 1343 sv = sv_2mortal (sv);
970 1344
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1345 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)"); 1346 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1347
974 return sv; 1348 return sv;
975} 1349}
976 1350
… …
981 1355
982BOOT: 1356BOOT:
983{ 1357{
984 int i; 1358 int i;
985 1359
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1360 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1361 decode_hexdigit [i] =
1362 i >= '0' && i <= '9' ? i - '0'
1363 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1364 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1365 : -1;
989 1366
990 for (i = 7; i--; ) 1367 json_stash = gv_stashpv ("JSON::XS" , 1);
1368 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1369
1370 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1371 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1372}
1373
1374PROTOTYPES: DISABLE
1375
1376void CLONE (...)
1377 CODE:
1378 json_stash = 0;
1379 json_boolean_stash = 0;
1380
1381void new (char *klass)
1382 PPCODE:
1383{
1384 HV *stash = !JSON_SLOW || json_stash
1385 ? json_stash
1386 : gv_stashpv ("JSON::XS", 1);
1387 SV *pv = NEWSV (0, sizeof (JSON));
1388 SvPOK_only (pv);
1389 Zero (SvPVX (pv), 1, JSON);
1390 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1391 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), stash)));
1392}
1393
1394void ascii (JSON *self, int enable = 1)
1395 ALIAS:
1396 ascii = F_ASCII
1397 latin1 = F_LATIN1
1398 utf8 = F_UTF8
1399 indent = F_INDENT
1400 canonical = F_CANONICAL
1401 space_before = F_SPACE_BEFORE
1402 space_after = F_SPACE_AFTER
1403 pretty = F_PRETTY
1404 allow_nonref = F_ALLOW_NONREF
1405 shrink = F_SHRINK
1406 allow_blessed = F_ALLOW_BLESSED
1407 convert_blessed = F_CONV_BLESSED
1408 PPCODE:
1409{
1410 if (enable)
1411 self->flags |= ix;
1412 else
1413 self->flags &= ~ix;
1414
1415 XPUSHs (ST (0));
1416}
1417
1418void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1419 PPCODE:
1420{
1421 UV log2 = 0;
1422
1423 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1424
1425 while ((1UL << log2) < max_depth)
1426 ++log2;
1427
1428 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1429
1430 XPUSHs (ST (0));
1431}
1432
1433void max_size (JSON *self, UV max_size = 0)
1434 PPCODE:
1435{
1436 UV log2 = 0;
1437
1438 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1439 if (max_size == 1) max_size = 2;
1440
1441 while ((1UL << log2) < max_size)
1442 ++log2;
1443
1444 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1445
1446 XPUSHs (ST (0));
1447}
1448
1449void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1450 PPCODE:
1451{
1452 SvREFCNT_dec (self->cb_object);
1453 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1454
1455 XPUSHs (ST (0));
1456}
1457
1458void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1459 PPCODE:
1460{
1461 if (!self->cb_sk_object)
1462 self->cb_sk_object = newHV ();
1463
1464 if (SvOK (cb))
1465 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1466 else
991 { 1467 {
992 decode_hexdigit ['a' + i] = 10 + i; 1468 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
993 decode_hexdigit ['A' + i] = 10 + i; 1469
1470 if (!HvKEYS (self->cb_sk_object))
1471 {
1472 SvREFCNT_dec (self->cb_sk_object);
1473 self->cb_sk_object = 0;
1474 }
994 } 1475 }
995 1476
996 json_stash = gv_stashpv ("JSON::XS", 1); 1477 XPUSHs (ST (0));
997} 1478}
998 1479
999PROTOTYPES: DISABLE 1480void encode (JSON *self, SV *scalar)
1481 PPCODE:
1482 XPUSHs (encode_json (scalar, self));
1000 1483
1001SV *new (char *dummy) 1484void decode (JSON *self, SV *jsonstr)
1485 PPCODE:
1486 XPUSHs (decode_json (jsonstr, self, 0));
1487
1488void decode_prefix (JSON *self, SV *jsonstr)
1489 PPCODE:
1490{
1491 UV offset;
1492 EXTEND (SP, 2);
1493 PUSHs (decode_json (jsonstr, self, &offset));
1494 PUSHs (sv_2mortal (newSVuv (offset)));
1495}
1496
1497void DESTROY (JSON *self)
1002 CODE: 1498 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1499 SvREFCNT_dec (self->cb_sk_object);
1004 OUTPUT: 1500 SvREFCNT_dec (self->cb_object);
1005 RETVAL
1006 1501
1007SV *ascii (SV *self, int enable = 1) 1502PROTOTYPES: ENABLE
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 1503
1026 RETVAL = newSVsv (self); 1504void to_json (SV *scalar)
1027}
1028 OUTPUT:
1029 RETVAL
1030
1031void encode (SV *self, SV *scalar)
1032 PPCODE: 1505 PPCODE:
1506{
1507 JSON json = { F_DEFAULT | F_UTF8 };
1033 XPUSHs (encode_json (scalar, *SvJSON (self))); 1508 XPUSHs (encode_json (scalar, &json));
1509}
1034 1510
1035void decode (SV *self, SV *jsonstr) 1511void from_json (SV *jsonstr)
1036 PPCODE: 1512 PPCODE:
1037 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1513{
1038 1514 JSON json = { F_DEFAULT | F_UTF8 };
1039PROTOTYPES: ENABLE
1040
1041void to_json (SV *scalar)
1042 ALIAS:
1043 objToJson = 0
1044 PPCODE:
1045 XPUSHs (encode_json (scalar, F_UTF8));
1046
1047void from_json (SV *jsonstr)
1048 ALIAS:
1049 jsonToObj = 0
1050 PPCODE:
1051 XPUSHs (decode_json (jsonstr, F_UTF8)); 1515 XPUSHs (decode_json (jsonstr, &json, 0));
1516}
1052 1517

Diff Legend

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