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 1.16 by root, Sun Mar 25 02:37:00 2007 UTC vs.
Revision 1.48 by root, Sun Jul 1 22:20:00 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 27#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL // NYI
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
17 35
18// F_SKIPINVALID? 36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
19// F_EXECCODEREF? 37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
20// F_SELFCONVERT?
21 38
22#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
23#define F_DEFAULT 0 40#define F_DEFAULT (9UL << S_MAXDEPTH)
24 41
25#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
26#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
27 44
28#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
29#define SHORT_STRING_LEN 512 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
30 46
31#define SB do { 47#define SB do {
32#define SE } while (0) 48#define SE } while (0)
33 49
50#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value))
52# define inline inline
53#else
54# define expect(expr,value) (expr)
55# define inline static
56#endif
57
58#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1)
60
34static HV *json_stash; // JSON::XS:: 61static HV *json_stash, *json_boolean_stash; // JSON::XS::
62static SV *json_true, *json_false;
63
64typedef struct {
65 U32 flags;
66} JSON;
35 67
36///////////////////////////////////////////////////////////////////////////// 68/////////////////////////////////////////////////////////////////////////////
37// utility functions 69// utility functions
38 70
39static UV * 71inline void
40SvJSON (SV *sv)
41{
42 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
43 croak ("object is not of type JSON::XS");
44
45 return &SvUVX (SvRV (sv));
46}
47
48static void
49shrink (SV *sv) 72shrink (SV *sv)
50{ 73{
51 sv_utf8_downgrade (sv, 1); 74 sv_utf8_downgrade (sv, 1);
52 if (SvLEN (sv) > SvCUR (sv) + 1) 75 if (SvLEN (sv) > SvCUR (sv) + 1)
53 { 76 {
… …
62// decode an utf-8 character and return it, or (UV)-1 in 85// decode an utf-8 character and return it, or (UV)-1 in
63// case of an error. 86// case of an error.
64// we special-case "safe" characters from U+80 .. U+7FF, 87// we special-case "safe" characters from U+80 .. U+7FF,
65// but use the very good perl function to parse anything else. 88// but use the very good perl function to parse anything else.
66// note that we never call this function for a ascii codepoints 89// note that we never call this function for a ascii codepoints
67static UV 90inline UV
68decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 91decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
69{ 92{
70 if (s[0] > 0xdf || s[0] < 0xc2) 93 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
71 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 94 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
72 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 95 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
73 { 96 {
74 *clen = 2; 97 *clen = 2;
75 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 98 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
76 } 99 }
77 else 100 else
101 {
102 *clen = (STRLEN)-1;
78 return (UV)-1; 103 return (UV)-1;
104 }
79} 105}
80 106
81///////////////////////////////////////////////////////////////////////////// 107/////////////////////////////////////////////////////////////////////////////
82// encoder 108// encoder
83 109
… …
85typedef struct 111typedef struct
86{ 112{
87 char *cur; // SvPVX (sv) + current output position 113 char *cur; // SvPVX (sv) + current output position
88 char *end; // SvEND (sv) 114 char *end; // SvEND (sv)
89 SV *sv; // result scalar 115 SV *sv; // result scalar
90 UV flags; // F_* 116 JSON json;
91 int indent; // indentation level 117 U32 indent; // indentation level
92 int max_depth; // max. recursion level 118 U32 maxdepth; // max. indentation/recursion level
93} enc_t; 119} enc_t;
94 120
95static void 121inline void
96need (enc_t *enc, STRLEN len) 122need (enc_t *enc, STRLEN len)
97{ 123{
98 if (enc->cur + len >= enc->end) 124 if (expect_false (enc->cur + len >= enc->end))
99 { 125 {
100 STRLEN cur = enc->cur - SvPVX (enc->sv); 126 STRLEN cur = enc->cur - SvPVX (enc->sv);
101 SvGROW (enc->sv, cur + len + 1); 127 SvGROW (enc->sv, cur + len + 1);
102 enc->cur = SvPVX (enc->sv) + cur; 128 enc->cur = SvPVX (enc->sv) + cur;
103 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 129 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
104 } 130 }
105} 131}
106 132
107static void 133inline void
108encode_ch (enc_t *enc, char ch) 134encode_ch (enc_t *enc, char ch)
109{ 135{
110 need (enc, 1); 136 need (enc, 1);
111 *enc->cur++ = ch; 137 *enc->cur++ = ch;
112} 138}
… …
120 146
121 while (str < end) 147 while (str < end)
122 { 148 {
123 unsigned char ch = *(unsigned char *)str; 149 unsigned char ch = *(unsigned char *)str;
124 150
125 if (ch >= 0x20 && ch < 0x80) // most common case 151 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
126 { 152 {
127 if (ch == '"') // but with slow exceptions 153 if (expect_false (ch == '"')) // but with slow exceptions
128 { 154 {
129 need (enc, len += 1); 155 need (enc, len += 1);
130 *enc->cur++ = '\\'; 156 *enc->cur++ = '\\';
131 *enc->cur++ = '"'; 157 *enc->cur++ = '"';
132 } 158 }
133 else if (ch == '\\') 159 else if (expect_false (ch == '\\'))
134 { 160 {
135 need (enc, len += 1); 161 need (enc, len += 1);
136 *enc->cur++ = '\\'; 162 *enc->cur++ = '\\';
137 *enc->cur++ = '\\'; 163 *enc->cur++ = '\\';
138 } 164 }
… …
156 STRLEN clen; 182 STRLEN clen;
157 UV uch; 183 UV uch;
158 184
159 if (is_utf8) 185 if (is_utf8)
160 { 186 {
161 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
162 uch = decode_utf8 (str, end - str, &clen); 187 uch = decode_utf8 (str, end - str, &clen);
163 if (clen == (STRLEN)-1) 188 if (clen == (STRLEN)-1)
164 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 189 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
165 } 190 }
166 else 191 else
… …
170 } 195 }
171 196
172 if (uch > 0x10FFFFUL) 197 if (uch > 0x10FFFFUL)
173 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 198 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
174 199
175 if (uch < 0x80 || enc->flags & F_ASCII) 200 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
176 { 201 {
177 if (uch > 0xFFFFUL) 202 if (uch > 0xFFFFUL)
178 { 203 {
179 need (enc, len += 11); 204 need (enc, len += 11);
180 sprintf (enc->cur, "\\u%04x\\u%04x", 205 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
194 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 219 *enc->cur++ = hexdigit [(uch >> 0) & 15];
195 } 220 }
196 221
197 str += clen; 222 str += clen;
198 } 223 }
224 else if (enc->json.flags & F_LATIN1)
225 {
226 *enc->cur++ = uch;
227 str += clen;
228 }
199 else if (is_utf8) 229 else if (is_utf8)
200 { 230 {
201 need (enc, len += clen); 231 need (enc, len += clen);
202 do 232 do
203 { 233 {
… …
205 } 235 }
206 while (--clen); 236 while (--clen);
207 } 237 }
208 else 238 else
209 { 239 {
210 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 240 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
211 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 241 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
212 ++str; 242 ++str;
213 } 243 }
214 } 244 }
215 } 245 }
… …
217 247
218 --len; 248 --len;
219 } 249 }
220} 250}
221 251
222static void 252inline void
223encode_indent (enc_t *enc) 253encode_indent (enc_t *enc)
224{ 254{
225 if (enc->flags & F_INDENT) 255 if (enc->json.flags & F_INDENT)
226 { 256 {
227 int spaces = enc->indent * INDENT_STEP; 257 int spaces = enc->indent * INDENT_STEP;
228 258
229 need (enc, spaces); 259 need (enc, spaces);
230 memset (enc->cur, ' ', spaces); 260 memset (enc->cur, ' ', spaces);
231 enc->cur += spaces; 261 enc->cur += spaces;
232 } 262 }
233} 263}
234 264
235static void 265inline void
236encode_space (enc_t *enc) 266encode_space (enc_t *enc)
237{ 267{
238 need (enc, 1); 268 need (enc, 1);
239 encode_ch (enc, ' '); 269 encode_ch (enc, ' ');
240} 270}
241 271
242static void 272inline void
243encode_nl (enc_t *enc) 273encode_nl (enc_t *enc)
244{ 274{
245 if (enc->flags & F_INDENT) 275 if (enc->json.flags & F_INDENT)
246 { 276 {
247 need (enc, 1); 277 need (enc, 1);
248 encode_ch (enc, '\n'); 278 encode_ch (enc, '\n');
249 } 279 }
250} 280}
251 281
252static void 282inline void
253encode_comma (enc_t *enc) 283encode_comma (enc_t *enc)
254{ 284{
255 encode_ch (enc, ','); 285 encode_ch (enc, ',');
256 286
257 if (enc->flags & F_INDENT) 287 if (enc->json.flags & F_INDENT)
258 encode_nl (enc); 288 encode_nl (enc);
259 else if (enc->flags & F_SPACE_AFTER) 289 else if (enc->json.flags & F_SPACE_AFTER)
260 encode_space (enc); 290 encode_space (enc);
261} 291}
262 292
263static void encode_sv (enc_t *enc, SV *sv); 293static void encode_sv (enc_t *enc, SV *sv);
264 294
265static void 295static void
266encode_av (enc_t *enc, AV *av) 296encode_av (enc_t *enc, AV *av)
267{ 297{
268 int i, len = av_len (av); 298 int i, len = av_len (av);
299
300 if (enc->indent >= enc->maxdepth)
301 croak ("data structure too deep (hit recursion limit)");
269 302
270 encode_ch (enc, '['); encode_nl (enc); 303 encode_ch (enc, '['); encode_nl (enc);
271 ++enc->indent; 304 ++enc->indent;
272 305
273 for (i = 0; i <= len; ++i) 306 for (i = 0; i <= len; ++i)
… …
304 else 337 else
305 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 338 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
306 339
307 encode_ch (enc, '"'); 340 encode_ch (enc, '"');
308 341
309 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 342 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
310 encode_ch (enc, ':'); 343 encode_ch (enc, ':');
311 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 344 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
312 encode_sv (enc, HeVAL (he)); 345 encode_sv (enc, HeVAL (he));
313} 346}
314 347
315// compare hash entries, used when all keys are bytestrings 348// compare hash entries, used when all keys are bytestrings
316static int 349static int
… …
339 372
340static void 373static void
341encode_hv (enc_t *enc, HV *hv) 374encode_hv (enc_t *enc, HV *hv)
342{ 375{
343 int count, i; 376 int count, i;
377
378 if (enc->indent >= enc->maxdepth)
379 croak ("data structure too deep (hit recursion limit)");
344 380
345 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 381 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
346 382
347 if ((count = hv_iterinit (hv))) 383 if ((count = hv_iterinit (hv)))
348 { 384 {
349 // for canonical output we have to sort by keys first 385 // for canonical output we have to sort by keys first
350 // actually, this is mostly due to the stupid so-called 386 // actually, this is mostly due to the stupid so-called
351 // security workaround added somewhere in 5.8.x. 387 // security workaround added somewhere in 5.8.x.
352 // that randomises hash orderings 388 // that randomises hash orderings
353 if (enc->flags & F_CANONICAL) 389 if (enc->json.flags & F_CANONICAL)
354 { 390 {
355 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
356 int fast = 1; 391 int fast = 1;
392 HE *he;
393#if defined(__BORLANDC__) || defined(_MSC_VER)
394 HE **hes = _alloca (count * sizeof (HE));
395#else
396 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
397#endif
357 398
358 i = 0; 399 i = 0;
359 while ((he = hv_iternext (hv))) 400 while ((he = hv_iternext (hv)))
360 { 401 {
361 hes [i++] = he; 402 hes [i++] = he;
… …
396 437
397 encode_nl (enc); 438 encode_nl (enc);
398 } 439 }
399 else 440 else
400 { 441 {
401 SV *sv;
402 HE *he = hv_iternext (hv); 442 HE *he = hv_iternext (hv);
403 443
404 for (;;) 444 for (;;)
405 { 445 {
406 encode_indent (enc); 446 encode_indent (enc);
… …
415 encode_nl (enc); 455 encode_nl (enc);
416 } 456 }
417 } 457 }
418 458
419 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 459 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
460}
461
462// encode objects, arrays and special \0=false and \1=true values.
463static void
464encode_rv (enc_t *enc, SV *sv)
465{
466 svtype svt;
467
468 SvGETMAGIC (sv);
469 svt = SvTYPE (sv);
470
471 if (expect_false (SvOBJECT (sv)))
472 {
473 if (SvSTASH (sv) == json_boolean_stash)
474 {
475 if (SvIV (sv) == 0)
476 encode_str (enc, "false", 5, 0);
477 else
478 encode_str (enc, "true", 4, 0);
479 }
480 else
481 {
482#if 0
483 if (0 && sv_derived_from (rv, "JSON::Literal"))
484 {
485 // not yet
486 }
487#endif
488 if (enc->json.flags & F_CONV_BLESSED)
489 {
490 // we re-bless the reference to get overload and other niceties right
491 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
492
493 if (to_json)
494 {
495 dSP;
496 ENTER;
497 SAVETMPS;
498 PUSHMARK (SP);
499 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
500
501 // calling with G_SCALAR ensures that we always get a 1 reutrn value
502 // check anyways.
503 PUTBACK;
504 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
505 SPAGAIN;
506
507 encode_sv (enc, POPs);
508
509 FREETMPS;
510 LEAVE;
511 }
512 else if (enc->json.flags & F_ALLOW_BLESSED)
513 encode_str (enc, "null", 4, 0);
514 else
515 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
516 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
517 }
518 else if (enc->json.flags & F_ALLOW_BLESSED)
519 encode_str (enc, "null", 4, 0);
520 else
521 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
522 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
523 }
524 }
525 else if (svt == SVt_PVHV)
526 encode_hv (enc, (HV *)sv);
527 else if (svt == SVt_PVAV)
528 encode_av (enc, (AV *)sv);
529 else if (svt < SVt_PVAV)
530 {
531 if (SvNIOK (sv) && SvIV (sv) == 0)
532 encode_str (enc, "false", 5, 0);
533 else if (SvNIOK (sv) && SvIV (sv) == 1)
534 encode_str (enc, "true", 4, 0);
535 else
536 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
537 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
538 }
539 else
540 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
541 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
420} 542}
421 543
422static void 544static void
423encode_sv (enc_t *enc, SV *sv) 545encode_sv (enc_t *enc, SV *sv)
424{ 546{
… …
432 encode_str (enc, str, len, SvUTF8 (sv)); 554 encode_str (enc, str, len, SvUTF8 (sv));
433 encode_ch (enc, '"'); 555 encode_ch (enc, '"');
434 } 556 }
435 else if (SvNOKp (sv)) 557 else if (SvNOKp (sv))
436 { 558 {
559 // trust that perl will do the right thing w.r.t. JSON syntax.
437 need (enc, NV_DIG + 32); 560 need (enc, NV_DIG + 32);
438 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 561 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
439 enc->cur += strlen (enc->cur); 562 enc->cur += strlen (enc->cur);
440 } 563 }
441 else if (SvIOKp (sv)) 564 else if (SvIOKp (sv))
442 { 565 {
443 need (enc, 64); 566 // we assume we can always read an IV as a UV
567 if (SvUV (sv) & ~(UV)0x7fff)
568 {
569 // large integer, use the (rather slow) snprintf way.
570 need (enc, sizeof (UV) * 3);
444 enc->cur += 571 enc->cur +=
445 SvIsUV(sv) 572 SvIsUV(sv)
446 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 573 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
447 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 574 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
575 }
576 else
577 {
578 // optimise the "small number case"
579 // code will likely be branchless and use only a single multiplication
580 I32 i = SvIV (sv);
581 U32 u;
582 char digit, nz = 0;
583
584 need (enc, 6);
585
586 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
587 u = i < 0 ? -i : i;
588
589 // convert to 4.28 fixed-point representation
590 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
591
592 // now output digit by digit, each time masking out the integer part
593 // and multiplying by 5 while moving the decimal point one to the right,
594 // resulting in a net multiplication by 10.
595 // we always write the digit to memory but conditionally increment
596 // the pointer, to ease the usage of conditional move instructions.
597 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
598 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
599 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
600 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
601 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
602 }
448 } 603 }
449 else if (SvROK (sv)) 604 else if (SvROK (sv))
450 { 605 encode_rv (enc, SvRV (sv));
451 SV *rv = SvRV (sv);
452
453 if (enc->indent >= enc->max_depth)
454 croak ("data structure too deep (hit recursion limit)");
455
456 switch (SvTYPE (rv))
457 {
458 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
459 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
460
461 default:
462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv));
464 }
465 }
466 else if (!SvOK (sv)) 606 else if (!SvOK (sv))
467 encode_str (enc, "null", 4, 0); 607 encode_str (enc, "null", 4, 0);
468 else 608 else
469 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 609 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
470 SvPV_nolen (sv), SvFLAGS (sv)); 610 SvPV_nolen (sv), SvFLAGS (sv));
471} 611}
472 612
473static SV * 613static SV *
474encode_json (SV *scalar, UV flags) 614encode_json (SV *scalar, JSON *json)
475{ 615{
616 enc_t enc;
617
476 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 618 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
477 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 619 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
478 620
479 enc_t enc; 621 enc.json = *json;
480 enc.flags = flags;
481 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 622 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
482 enc.cur = SvPVX (enc.sv); 623 enc.cur = SvPVX (enc.sv);
483 enc.end = SvEND (enc.sv); 624 enc.end = SvEND (enc.sv);
484 enc.indent = 0; 625 enc.indent = 0;
485 enc.max_depth = 0x7fffffffUL; 626 enc.maxdepth = DEC_DEPTH (enc.json.flags);
486 627
487 SvPOK_only (enc.sv); 628 SvPOK_only (enc.sv);
488 encode_sv (&enc, scalar); 629 encode_sv (&enc, scalar);
489 630
631 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
632 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
633
490 if (!(flags & (F_ASCII | F_UTF8))) 634 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
491 SvUTF8_on (enc.sv); 635 SvUTF8_on (enc.sv);
492 636
493 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
494
495 if (enc.flags & F_SHRINK) 637 if (enc.json.flags & F_SHRINK)
496 shrink (enc.sv); 638 shrink (enc.sv);
497 639
498 return enc.sv; 640 return enc.sv;
499} 641}
500 642
… …
505typedef struct 647typedef struct
506{ 648{
507 char *cur; // current parser pointer 649 char *cur; // current parser pointer
508 char *end; // end of input string 650 char *end; // end of input string
509 const char *err; // parse error, if != 0 651 const char *err; // parse error, if != 0
510 UV flags; // F_* 652 JSON json;
653 U32 depth; // recursion depth
654 U32 maxdepth; // recursion depth limit
511} dec_t; 655} dec_t;
512 656
513static void 657inline void
514decode_ws (dec_t *dec) 658decode_ws (dec_t *dec)
515{ 659{
516 for (;;) 660 for (;;)
517 { 661 {
518 char ch = *dec->cur; 662 char ch = *dec->cur;
… …
524 ++dec->cur; 668 ++dec->cur;
525 } 669 }
526} 670}
527 671
528#define ERR(reason) SB dec->err = reason; goto fail; SE 672#define ERR(reason) SB dec->err = reason; goto fail; SE
673
529#define EXPECT_CH(ch) SB \ 674#define EXPECT_CH(ch) SB \
530 if (*dec->cur != ch) \ 675 if (*dec->cur != ch) \
531 ERR (# ch " expected"); \ 676 ERR (# ch " expected"); \
532 ++dec->cur; \ 677 ++dec->cur; \
533 SE 678 SE
534 679
680#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
681#define DEC_DEC_DEPTH --dec->depth
682
535static SV *decode_sv (dec_t *dec); 683static SV *decode_sv (dec_t *dec);
536 684
537static signed char decode_hexdigit[256]; 685static signed char decode_hexdigit[256];
538 686
539static UV 687static UV
540decode_4hex (dec_t *dec) 688decode_4hex (dec_t *dec)
541{ 689{
542 signed char d1, d2, d3, d4; 690 signed char d1, d2, d3, d4;
543 unsigned char *cur = (unsigned char *)dec->cur; 691 unsigned char *cur = (unsigned char *)dec->cur;
544 692
545 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 693 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
546 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 694 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
547 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 695 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
548 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 696 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
549 697
550 dec->cur += 4; 698 dec->cur += 4;
551 699
552 return ((UV)d1) << 12 700 return ((UV)d1) << 12
553 | ((UV)d2) << 8 701 | ((UV)d2) << 8
… …
561static SV * 709static SV *
562decode_str (dec_t *dec) 710decode_str (dec_t *dec)
563{ 711{
564 SV *sv = 0; 712 SV *sv = 0;
565 int utf8 = 0; 713 int utf8 = 0;
714 char *dec_cur = dec->cur;
566 715
567 do 716 do
568 { 717 {
569 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 718 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
570 char *cur = buf; 719 char *cur = buf;
571 720
572 do 721 do
573 { 722 {
574 unsigned char ch = *(unsigned char *)dec->cur++; 723 unsigned char ch = *(unsigned char *)dec_cur++;
575 724
576 if (ch == '"') 725 if (expect_false (ch == '"'))
577 { 726 {
578 --dec->cur; 727 --dec_cur;
579 break; 728 break;
580 } 729 }
581 else if (ch == '\\') 730 else if (expect_false (ch == '\\'))
582 { 731 {
583 switch (*dec->cur) 732 switch (*dec_cur)
584 { 733 {
585 case '\\': 734 case '\\':
586 case '/': 735 case '/':
587 case '"': *cur++ = *dec->cur++; break; 736 case '"': *cur++ = *dec_cur++; break;
588 737
589 case 'b': ++dec->cur; *cur++ = '\010'; break; 738 case 'b': ++dec_cur; *cur++ = '\010'; break;
590 case 't': ++dec->cur; *cur++ = '\011'; break; 739 case 't': ++dec_cur; *cur++ = '\011'; break;
591 case 'n': ++dec->cur; *cur++ = '\012'; break; 740 case 'n': ++dec_cur; *cur++ = '\012'; break;
592 case 'f': ++dec->cur; *cur++ = '\014'; break; 741 case 'f': ++dec_cur; *cur++ = '\014'; break;
593 case 'r': ++dec->cur; *cur++ = '\015'; break; 742 case 'r': ++dec_cur; *cur++ = '\015'; break;
594 743
595 case 'u': 744 case 'u':
596 { 745 {
597 UV lo, hi; 746 UV lo, hi;
598 ++dec->cur; 747 ++dec_cur;
599 748
749 dec->cur = dec_cur;
600 hi = decode_4hex (dec); 750 hi = decode_4hex (dec);
751 dec_cur = dec->cur;
601 if (hi == (UV)-1) 752 if (hi == (UV)-1)
602 goto fail; 753 goto fail;
603 754
604 // possibly a surrogate pair 755 // possibly a surrogate pair
605 if (hi >= 0xd800) 756 if (hi >= 0xd800)
606 if (hi < 0xdc00) 757 if (hi < 0xdc00)
607 { 758 {
608 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 759 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
609 ERR ("missing low surrogate character in surrogate pair"); 760 ERR ("missing low surrogate character in surrogate pair");
610 761
611 dec->cur += 2; 762 dec_cur += 2;
612 763
764 dec->cur = dec_cur;
613 lo = decode_4hex (dec); 765 lo = decode_4hex (dec);
766 dec_cur = dec->cur;
614 if (lo == (UV)-1) 767 if (lo == (UV)-1)
615 goto fail; 768 goto fail;
616 769
617 if (lo < 0xdc00 || lo >= 0xe000) 770 if (lo < 0xdc00 || lo >= 0xe000)
618 ERR ("surrogate pair expected"); 771 ERR ("surrogate pair expected");
… …
632 *cur++ = hi; 785 *cur++ = hi;
633 } 786 }
634 break; 787 break;
635 788
636 default: 789 default:
637 --dec->cur; 790 --dec_cur;
638 ERR ("illegal backslash escape sequence in string"); 791 ERR ("illegal backslash escape sequence in string");
639 } 792 }
640 } 793 }
641 else if (ch >= 0x20 && ch <= 0x7f) 794 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
642 *cur++ = ch; 795 *cur++ = ch;
643 else if (ch >= 0x80) 796 else if (ch >= 0x80)
644 { 797 {
645 --dec->cur;
646
647 STRLEN clen; 798 STRLEN clen;
799 UV uch;
800
801 --dec_cur;
802
648 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 803 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
649 if (clen == (STRLEN)-1) 804 if (clen == (STRLEN)-1)
650 ERR ("malformed UTF-8 character in JSON string"); 805 ERR ("malformed UTF-8 character in JSON string");
651 806
652 do 807 do
653 {
654 *cur++ = *dec->cur++; 808 *cur++ = *dec_cur++;
655 }
656 while (--clen); 809 while (--clen);
657 810
658 utf8 = 1; 811 utf8 = 1;
659 } 812 }
660 else if (!ch)
661 ERR ("unexpected end of string while parsing json string");
662 else 813 else
814 {
815 --dec_cur;
816
817 if (!ch)
818 ERR ("unexpected end of string while parsing JSON string");
819 else
663 ERR ("invalid character encountered"); 820 ERR ("invalid character encountered while parsing JSON string");
664 821 }
665 } 822 }
666 while (cur < buf + SHORT_STRING_LEN); 823 while (cur < buf + SHORT_STRING_LEN);
667 824
825 {
668 STRLEN len = cur - buf; 826 STRLEN len = cur - buf;
669 827
670 if (sv) 828 if (sv)
671 { 829 {
672 SvGROW (sv, SvCUR (sv) + len + 1); 830 SvGROW (sv, SvCUR (sv) + len + 1);
673 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 831 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
674 SvCUR_set (sv, SvCUR (sv) + len); 832 SvCUR_set (sv, SvCUR (sv) + len);
675 } 833 }
676 else 834 else
677 sv = newSVpvn (buf, len); 835 sv = newSVpvn (buf, len);
678 } 836 }
837 }
679 while (*dec->cur != '"'); 838 while (*dec_cur != '"');
680 839
681 ++dec->cur; 840 ++dec_cur;
682 841
683 if (sv) 842 if (sv)
684 { 843 {
685 SvPOK_only (sv); 844 SvPOK_only (sv);
686 *SvEND (sv) = 0; 845 *SvEND (sv) = 0;
… …
689 SvUTF8_on (sv); 848 SvUTF8_on (sv);
690 } 849 }
691 else 850 else
692 sv = newSVpvn ("", 0); 851 sv = newSVpvn ("", 0);
693 852
853 dec->cur = dec_cur;
694 return sv; 854 return sv;
695 855
696fail: 856fail:
857 dec->cur = dec_cur;
697 return 0; 858 return 0;
698} 859}
699 860
700static SV * 861static SV *
701decode_num (dec_t *dec) 862decode_num (dec_t *dec)
… …
759 is_nv = 1; 920 is_nv = 1;
760 } 921 }
761 922
762 if (!is_nv) 923 if (!is_nv)
763 { 924 {
764 UV uv; 925 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
765 int numtype = grok_number (start, dec->cur - start, &uv); 926 if (*start == '-')
766 if (numtype & IS_NUMBER_IN_UV) 927 switch (dec->cur - start)
767 if (numtype & IS_NUMBER_NEG)
768 { 928 {
769 if (uv < (UV)IV_MIN) 929 case 2: return newSViv (-( start [1] - '0' * 1));
770 return newSViv (-(IV)uv); 930 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
931 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
932 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
771 } 933 }
934 else
935 switch (dec->cur - start)
936 {
937 case 1: return newSViv ( start [0] - '0' * 1);
938 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
939 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
940 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
941 }
942
943 {
944 UV uv;
945 int numtype = grok_number (start, dec->cur - start, &uv);
946 if (numtype & IS_NUMBER_IN_UV)
947 if (numtype & IS_NUMBER_NEG)
948 {
949 if (uv < (UV)IV_MIN)
950 return newSViv (-(IV)uv);
951 }
772 else 952 else
773 return newSVuv (uv); 953 return newSVuv (uv);
954
955 // here would likely be the place for bigint support
774 } 956 }
957 }
775 958
959 // if we ever support bigint or bigfloat, this is the place for bigfloat
776 return newSVnv (Atof (start)); 960 return newSVnv (Atof (start));
777 961
778fail: 962fail:
779 return 0; 963 return 0;
780} 964}
… …
782static SV * 966static SV *
783decode_av (dec_t *dec) 967decode_av (dec_t *dec)
784{ 968{
785 AV *av = newAV (); 969 AV *av = newAV ();
786 970
971 DEC_INC_DEPTH;
787 decode_ws (dec); 972 decode_ws (dec);
973
788 if (*dec->cur == ']') 974 if (*dec->cur == ']')
789 ++dec->cur; 975 ++dec->cur;
790 else 976 else
791 for (;;) 977 for (;;)
792 { 978 {
… …
810 ERR (", or ] expected while parsing array"); 996 ERR (", or ] expected while parsing array");
811 997
812 ++dec->cur; 998 ++dec->cur;
813 } 999 }
814 1000
1001 DEC_DEC_DEPTH;
815 return newRV_noinc ((SV *)av); 1002 return newRV_noinc ((SV *)av);
816 1003
817fail: 1004fail:
818 SvREFCNT_dec (av); 1005 SvREFCNT_dec (av);
1006 DEC_DEC_DEPTH;
819 return 0; 1007 return 0;
820} 1008}
821 1009
822static SV * 1010static SV *
823decode_hv (dec_t *dec) 1011decode_hv (dec_t *dec)
824{ 1012{
825 HV *hv = newHV (); 1013 HV *hv = newHV ();
826 1014
1015 DEC_INC_DEPTH;
827 decode_ws (dec); 1016 decode_ws (dec);
1017
828 if (*dec->cur == '}') 1018 if (*dec->cur == '}')
829 ++dec->cur; 1019 ++dec->cur;
830 else 1020 else
831 for (;;) 1021 for (;;)
832 { 1022 {
833 SV *key, *value;
834
835 decode_ws (dec); EXPECT_CH ('"'); 1023 decode_ws (dec); EXPECT_CH ('"');
836 1024
837 key = decode_str (dec); 1025 // heuristic: assume that
838 if (!key) 1026 // a) decode_str + hv_store_ent are abysmally slow.
839 goto fail; 1027 // b) most hash keys are short, simple ascii text.
1028 // => try to "fast-match" such strings to avoid
1029 // the overhead of decode_str + hv_store_ent.
1030 {
1031 SV *value;
1032 char *p = dec->cur;
1033 char *e = p + 24; // only try up to 24 bytes
840 1034
841 decode_ws (dec); EXPECT_CH (':'); 1035 for (;;)
842
843 value = decode_sv (dec);
844 if (!value)
845 { 1036 {
1037 // the >= 0x80 is true on most architectures
1038 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1039 {
1040 // slow path, back up and use decode_str
1041 SV *key = decode_str (dec);
1042 if (!key)
1043 goto fail;
1044
1045 decode_ws (dec); EXPECT_CH (':');
1046
1047 value = decode_sv (dec);
1048 if (!value)
1049 {
1050 SvREFCNT_dec (key);
1051 goto fail;
1052 }
1053
1054 hv_store_ent (hv, key, value, 0);
846 SvREFCNT_dec (key); 1055 SvREFCNT_dec (key);
1056
1057 break;
1058 }
1059 else if (*p == '"')
1060 {
1061 // fast path, got a simple key
1062 char *key = dec->cur;
1063 int len = p - key;
1064 dec->cur = p + 1;
1065
1066 decode_ws (dec); EXPECT_CH (':');
1067
1068 value = decode_sv (dec);
1069 if (!value)
847 goto fail; 1070 goto fail;
1071
1072 hv_store (hv, key, len, value, 0);
1073
1074 break;
1075 }
1076
1077 ++p;
848 } 1078 }
849 1079 }
850 //TODO: optimise
851 hv_store_ent (hv, key, value, 0);
852 1080
853 decode_ws (dec); 1081 decode_ws (dec);
854 1082
855 if (*dec->cur == '}') 1083 if (*dec->cur == '}')
856 { 1084 {
… …
862 ERR (", or } expected while parsing object/hash"); 1090 ERR (", or } expected while parsing object/hash");
863 1091
864 ++dec->cur; 1092 ++dec->cur;
865 } 1093 }
866 1094
1095 DEC_DEC_DEPTH;
867 return newRV_noinc ((SV *)hv); 1096 return newRV_noinc ((SV *)hv);
868 1097
869fail: 1098fail:
870 SvREFCNT_dec (hv); 1099 SvREFCNT_dec (hv);
1100 DEC_DEC_DEPTH;
871 return 0; 1101 return 0;
872} 1102}
873 1103
874static SV * 1104static SV *
875decode_sv (dec_t *dec) 1105decode_sv (dec_t *dec)
876{ 1106{
877 decode_ws (dec); 1107 decode_ws (dec);
1108
1109 // the beauty of JSON: you need exactly one character lookahead
1110 // to parse anything.
878 switch (*dec->cur) 1111 switch (*dec->cur)
879 { 1112 {
880 case '"': ++dec->cur; return decode_str (dec); 1113 case '"': ++dec->cur; return decode_str (dec);
881 case '[': ++dec->cur; return decode_av (dec); 1114 case '[': ++dec->cur; return decode_av (dec);
882 case '{': ++dec->cur; return decode_hv (dec); 1115 case '{': ++dec->cur; return decode_hv (dec);
… …
888 1121
889 case 't': 1122 case 't':
890 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1123 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
891 { 1124 {
892 dec->cur += 4; 1125 dec->cur += 4;
893 return newSViv (1); 1126 return SvREFCNT_inc (json_true);
894 } 1127 }
895 else 1128 else
896 ERR ("'true' expected"); 1129 ERR ("'true' expected");
897 1130
898 break; 1131 break;
899 1132
900 case 'f': 1133 case 'f':
901 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1134 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
902 { 1135 {
903 dec->cur += 5; 1136 dec->cur += 5;
904 return newSViv (0); 1137 return SvREFCNT_inc (json_false);
905 } 1138 }
906 else 1139 else
907 ERR ("'false' expected"); 1140 ERR ("'false' expected");
908 1141
909 break; 1142 break;
… …
918 ERR ("'null' expected"); 1151 ERR ("'null' expected");
919 1152
920 break; 1153 break;
921 1154
922 default: 1155 default:
923 ERR ("malformed json string, neither array, object, number, string or atom"); 1156 ERR ("malformed JSON string, neither array, object, number, string or atom");
924 break; 1157 break;
925 } 1158 }
926 1159
927fail: 1160fail:
928 return 0; 1161 return 0;
929} 1162}
930 1163
931static SV * 1164static SV *
932decode_json (SV *string, UV flags) 1165decode_json (SV *string, JSON *json, UV *offset_return)
933{ 1166{
1167 dec_t dec;
1168 UV offset;
934 SV *sv; 1169 SV *sv;
935 1170
1171 SvGETMAGIC (string);
1172 SvUPGRADE (string, SVt_PV);
1173
1174 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1175 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1176 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1177
936 if (flags & F_UTF8) 1178 if (json->flags & F_UTF8)
937 sv_utf8_downgrade (string, 0); 1179 sv_utf8_downgrade (string, 0);
938 else 1180 else
939 sv_utf8_upgrade (string); 1181 sv_utf8_upgrade (string);
940 1182
941 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1183 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
942 1184
943 dec_t dec; 1185 dec.json = *json;
944 dec.flags = flags;
945 dec.cur = SvPVX (string); 1186 dec.cur = SvPVX (string);
946 dec.end = SvEND (string); 1187 dec.end = SvEND (string);
947 dec.err = 0; 1188 dec.err = 0;
1189 dec.depth = 0;
1190 dec.maxdepth = DEC_DEPTH (dec.json.flags);
948 1191
1192 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
949 sv = decode_sv (&dec); 1193 sv = decode_sv (&dec);
950 1194
1195 if (!(offset_return || !sv))
1196 {
1197 // check for trailing garbage
1198 decode_ws (&dec);
1199
1200 if (*dec.cur)
1201 {
1202 dec.err = "garbage after JSON object";
1203 SvREFCNT_dec (sv);
1204 sv = 0;
1205 }
1206 }
1207
1208 if (offset_return || !sv)
1209 {
1210 offset = dec.json.flags & F_UTF8
1211 ? dec.cur - SvPVX (string)
1212 : utf8_distance (dec.cur, SvPVX (string));
1213
1214 if (offset_return)
1215 *offset_return = offset;
1216 }
1217
951 if (!sv) 1218 if (!sv)
952 { 1219 {
953 IV offset = dec.flags & F_UTF8
954 ? dec.cur - SvPVX (string)
955 : utf8_distance (dec.cur, SvPVX (string));
956 SV *uni = sv_newmortal (); 1220 SV *uni = sv_newmortal ();
957 1221
958 // horrible hack to silence warning inside pv_uni_display 1222 // horrible hack to silence warning inside pv_uni_display
959 COP cop = *PL_curcop; 1223 COP cop = *PL_curcop;
960 cop.cop_warnings = pWARN_NONE; 1224 cop.cop_warnings = pWARN_NONE;
… …
962 SAVEVPTR (PL_curcop); 1226 SAVEVPTR (PL_curcop);
963 PL_curcop = &cop; 1227 PL_curcop = &cop;
964 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1228 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
965 LEAVE; 1229 LEAVE;
966 1230
967 croak ("%s, at character offset %d (%s)", 1231 croak ("%s, at character offset %d [\"%s\"]",
968 dec.err, 1232 dec.err,
969 (int)offset, 1233 (int)offset,
970 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1234 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
971 } 1235 }
972 1236
973 sv = sv_2mortal (sv); 1237 sv = sv_2mortal (sv);
974 1238
975 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1239 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
976 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1240 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
977 1241
978 return sv; 1242 return sv;
979} 1243}
980 1244
… …
985 1249
986BOOT: 1250BOOT:
987{ 1251{
988 int i; 1252 int i;
989 1253
990 memset (decode_hexdigit, 0xff, 256);
991 for (i = 10; i--; ) 1254 for (i = 0; i < 256; ++i)
992 decode_hexdigit ['0' + i] = i; 1255 decode_hexdigit [i] =
1256 i >= '0' && i <= '9' ? i - '0'
1257 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1258 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1259 : -1;
993 1260
994 for (i = 7; i--; )
995 {
996 decode_hexdigit ['a' + i] = 10 + i;
997 decode_hexdigit ['A' + i] = 10 + i;
998 }
999
1000 json_stash = gv_stashpv ("JSON::XS", 1); 1261 json_stash = gv_stashpv ("JSON::XS" , 1);
1262 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1263
1264 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1265 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1001} 1266}
1002 1267
1003PROTOTYPES: DISABLE 1268PROTOTYPES: DISABLE
1004 1269
1005SV *new (char *dummy) 1270void new (char *klass)
1006 CODE: 1271 PPCODE:
1007 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1272{
1008 OUTPUT: 1273 SV *pv = NEWSV (0, sizeof (JSON));
1009 RETVAL 1274 SvPOK_only (pv);
1275 Zero (SvPVX (pv), 1, sizeof (JSON));
1276 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1277 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1278}
1010 1279
1011SV *ascii (SV *self, int enable = 1) 1280void ascii (JSON *self, int enable = 1)
1012 ALIAS: 1281 ALIAS:
1013 ascii = F_ASCII 1282 ascii = F_ASCII
1283 latin1 = F_LATIN1
1014 utf8 = F_UTF8 1284 utf8 = F_UTF8
1015 indent = F_INDENT 1285 indent = F_INDENT
1016 canonical = F_CANONICAL 1286 canonical = F_CANONICAL
1017 space_before = F_SPACE_BEFORE 1287 space_before = F_SPACE_BEFORE
1018 space_after = F_SPACE_AFTER 1288 space_after = F_SPACE_AFTER
1019 pretty = F_PRETTY 1289 pretty = F_PRETTY
1020 allow_nonref = F_ALLOW_NONREF 1290 allow_nonref = F_ALLOW_NONREF
1021 shrink = F_SHRINK 1291 shrink = F_SHRINK
1292 allow_blessed = F_ALLOW_BLESSED
1293 convert_blessed = F_CONV_BLESSED
1022 CODE: 1294 PPCODE:
1023{ 1295{
1024 UV *uv = SvJSON (self);
1025 if (enable) 1296 if (enable)
1026 *uv |= ix; 1297 self->flags |= ix;
1027 else 1298 else
1028 *uv &= ~ix; 1299 self->flags &= ~ix;
1029 1300
1030 RETVAL = newSVsv (self); 1301 XPUSHs (ST (0));
1031} 1302}
1032 OUTPUT:
1033 RETVAL
1034 1303
1035void encode (SV *self, SV *scalar) 1304void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1036 PPCODE: 1305 PPCODE:
1037 XPUSHs (encode_json (scalar, *SvJSON (self))); 1306{
1307 UV log2 = 0;
1038 1308
1039void decode (SV *self, SV *jsonstr) 1309 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1310
1311 while ((1UL << log2) < max_depth)
1312 ++log2;
1313
1314 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1315
1316 XPUSHs (ST (0));
1317}
1318
1319void max_size (JSON *self, UV max_size = 0)
1040 PPCODE: 1320 PPCODE:
1321{
1322 UV log2 = 0;
1323
1324 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1325 if (max_size == 1) max_size = 2;
1326
1327 while ((1UL << log2) < max_size)
1328 ++log2;
1329
1330 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1331
1332 XPUSHs (ST (0));
1333}
1334
1335void encode (JSON *self, SV *scalar)
1336 PPCODE:
1337 XPUSHs (encode_json (scalar, self));
1338
1339void decode (JSON *self, SV *jsonstr)
1340 PPCODE:
1041 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1341 XPUSHs (decode_json (jsonstr, self, 0));
1342
1343void decode_prefix (JSON *self, SV *jsonstr)
1344 PPCODE:
1345{
1346 UV offset;
1347 EXTEND (SP, 2);
1348 PUSHs (decode_json (jsonstr, self, &offset));
1349 PUSHs (sv_2mortal (newSVuv (offset)));
1350}
1042 1351
1043PROTOTYPES: ENABLE 1352PROTOTYPES: ENABLE
1044 1353
1045void to_json (SV *scalar) 1354void to_json (SV *scalar)
1046 ALIAS:
1047 objToJson = 0
1048 PPCODE: 1355 PPCODE:
1356{
1357 JSON json = { F_DEFAULT | F_UTF8 };
1049 XPUSHs (encode_json (scalar, F_UTF8)); 1358 XPUSHs (encode_json (scalar, &json));
1359}
1050 1360
1051void from_json (SV *jsonstr) 1361void from_json (SV *jsonstr)
1052 ALIAS:
1053 jsonToObj = 0
1054 PPCODE: 1362 PPCODE:
1363{
1364 JSON json = { F_DEFAULT | F_UTF8 };
1055 XPUSHs (decode_json (jsonstr, F_UTF8)); 1365 XPUSHs (decode_json (jsonstr, &json, 0));
1366}
1056 1367

Diff Legend

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