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 1.2 by root, Thu Mar 22 17:28:50 2007 UTC vs.
Revision 1.105 by root, Tue Jan 19 01:07:27 2010 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 <limits.h>
10#include <float.h>
8 11
12#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h
14#endif
15
16// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24
9#define F_ASCII 0x00000001 25#define F_ASCII 0x00000001UL
26#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 27#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 28#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 29#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 30#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 31#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040 32#define F_ALLOW_NONREF 0x00000100UL
33#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
36#define F_RELAXED 0x00001000UL
37#define F_ALLOW_UNKNOWN 0x00002000UL
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
16 39
17#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
18#define F_DEFAULT 0
19 41
20#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level
44
45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
21 46
22#define SB do { 47#define SB do {
23#define SE } while (0) 48#define SE } while (0)
24 49
25static HV *json_stash; 50#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr), (value))
52# define INLINE static 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
61#define IN_RANGE_INC(type,val,beg,end) \
62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
67#ifdef USE_ITHREADS
68# define JSON_SLOW 1
69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
70#else
71# define JSON_SLOW 0
72# define JSON_STASH json_stash
73#endif
74
75static HV *json_stash, *json_boolean_stash; // JSON::XS::
76static SV *json_true, *json_false;
77
78enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash
82 INCR_M_C0, // inside comment in initial whitespace sequence
83 INCR_M_C1, // inside comment in other places
84 INCR_M_JSON // outside anything, count nesting
85};
86
87#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
88
89typedef struct {
90 U32 flags;
91 U32 max_depth;
92 STRLEN max_size;
93
94 SV *cb_object;
95 HV *cb_sk_object;
96
97 // for the incremental parser
98 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level
101 unsigned char incr_mode;
102} JSON;
103
104INLINE void
105json_init (JSON *json)
106{
107 Zero (json, 1, JSON);
108 json->max_depth = 512;
109}
110
111/////////////////////////////////////////////////////////////////////////////
112// utility functions
113
114INLINE SV *
115get_bool (const char *name)
116{
117 SV *sv = get_sv (name, 1);
118
119 SvREADONLY_on (sv);
120 SvREADONLY_on (SvRV (sv));
121
122 return sv;
123}
124
125INLINE void
126shrink (SV *sv)
127{
128 sv_utf8_downgrade (sv, 1);
129
130 if (SvLEN (sv) > SvCUR (sv) + 1)
131 {
132#ifdef SvPV_shrink_to_cur
133 SvPV_shrink_to_cur (sv);
134#elif defined (SvPV_renew)
135 SvPV_renew (sv, SvCUR (sv) + 1);
136#endif
137 }
138}
139
140// decode an utf-8 character and return it, or (UV)-1 in
141// case of an error.
142// we special-case "safe" characters from U+80 .. U+7FF,
143// but use the very good perl function to parse anything else.
144// note that we never call this function for a ascii codepoints
145INLINE UV
146decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
147{
148 if (expect_true (len >= 2
149 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
150 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
151 {
152 *clen = 2;
153 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
154 }
155 else
156 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
157}
158
159// likewise for encoding, also never called for ascii codepoints
160// this function takes advantage of this fact, although current gccs
161// seem to optimise the check for >= 0x80 away anyways
162INLINE unsigned char *
163encode_utf8 (unsigned char *s, UV ch)
164{
165 if (expect_false (ch < 0x000080))
166 *s++ = ch;
167 else if (expect_true (ch < 0x000800))
168 *s++ = 0xc0 | ( ch >> 6),
169 *s++ = 0x80 | ( ch & 0x3f);
170 else if ( ch < 0x010000)
171 *s++ = 0xe0 | ( ch >> 12),
172 *s++ = 0x80 | ((ch >> 6) & 0x3f),
173 *s++ = 0x80 | ( ch & 0x3f);
174 else if ( ch < 0x110000)
175 *s++ = 0xf0 | ( ch >> 18),
176 *s++ = 0x80 | ((ch >> 12) & 0x3f),
177 *s++ = 0x80 | ((ch >> 6) & 0x3f),
178 *s++ = 0x80 | ( ch & 0x3f);
179
180 return s;
181}
182
183// convert offset pointer to character index, sv must be string
184static STRLEN
185ptr_to_index (SV *sv, char *offset)
186{
187 return SvUTF8 (sv)
188 ? utf8_distance (offset, SvPVX (sv))
189 : offset - SvPVX (sv);
190}
191
192/////////////////////////////////////////////////////////////////////////////
193// fp hell
194
195// scan a group of digits, and a trailing exponent
196static void
197json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp)
198{
199 UV uaccum = 0;
200 int eaccum = 0;
201
202 for (;;)
203 {
204 U8 dig = (U8)*s - '0';
205
206 if (expect_false (dig >= 10))
207 {
208 if (dig == (U8)((U8)'.' - (U8)'0'))
209 {
210 ++s;
211 json_atof_scan1 (s, accum, expo, 1);
212 }
213 else if ((dig | ' ') == 'e' - '0')
214 {
215 int exp2 = 0;
216 int neg = 0;
217
218 ++s;
219
220 if (*s == '-')
221 {
222 ++s;
223 neg = 1;
224 }
225 else if (*s == '+')
226 ++s;
227
228 while ((dig = (U8)*s - '0') < 10)
229 exp2 = exp2 * 10 + *s++ - '0';
230
231 *expo += neg ? -exp2 : exp2;
232 }
233
234 break;
235 }
236
237 ++s;
238
239 uaccum = uaccum * 10 + dig;
240 ++eaccum;
241
242 // if we have too many digits, then recurse for more
243 // we actually do this for rather few digits
244 if (uaccum >= (UV_MAX - 9) / 10)
245 {
246 if (postdp) *expo -= eaccum;
247 json_atof_scan1 (s, accum, expo, postdp);
248 if (postdp) *expo += eaccum;
249
250 break;
251 }
252 }
253
254 // this relies greatly on the quality of the pow ()
255 // implementation of the platform, but a good
256 // implementation is hard to beat.
257 if (postdp) *expo -= eaccum;
258 *accum += uaccum * Perl_pow (10., *expo);
259 *expo += eaccum;
260}
261
262static NV
263json_atof (const char *s)
264{
265 NV accum = 0.;
266 int expo = 0;
267 int neg = 0;
268
269 if (*s == '-')
270 {
271 ++s;
272 neg = 1;
273 }
274
275 json_atof_scan1 (s, &accum, &expo, 0);
276
277 return neg ? -accum : accum;
278}
279/////////////////////////////////////////////////////////////////////////////
280// encoder
26 281
27// structure used for encoding JSON 282// structure used for encoding JSON
28typedef struct 283typedef struct
29{ 284{
30 char *cur; 285 char *cur; // SvPVX (sv) + current output position
31 STRLEN len; // SvLEN (sv)
32 char *end; // SvEND (sv) 286 char *end; // SvEND (sv)
33 SV *sv; 287 SV *sv; // result scalar
34 UV flags; 288 JSON json;
35 int max_recurse; 289 U32 indent; // indentation level
36 int indent; 290 UV limit; // escape character values >= this value when encoding
37} enc_t; 291} enc_t;
38 292
39// structure used for decoding JSON 293INLINE void
40typedef struct
41{
42 char *cur;
43 char *end;
44 char *err;
45 UV flags;
46} dec_t;
47
48static UV *
49SvJSON (SV *sv)
50{
51 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
52 croak ("object is not of type JSON::XS");
53
54 return &SvUVX (SvRV (sv));
55}
56
57/////////////////////////////////////////////////////////////////////////////
58
59static void
60need (enc_t *enc, STRLEN len) 294need (enc_t *enc, STRLEN len)
61{ 295{
62 if (enc->cur + len >= enc->end) 296 if (expect_false (enc->cur + len >= enc->end))
63 { 297 {
64 STRLEN cur = enc->cur - SvPVX (enc->sv); 298 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
65 SvGROW (enc->sv, cur + len + 1); 299 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
66 enc->cur = SvPVX (enc->sv) + cur; 300 enc->cur = SvPVX (enc->sv) + cur;
67 enc->end = SvEND (enc->sv); 301 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
68 } 302 }
69} 303}
70 304
71static void 305INLINE void
72encode_ch (enc_t *enc, char ch) 306encode_ch (enc_t *enc, char ch)
73{ 307{
74 need (enc, 1); 308 need (enc, 1);
75 *enc->cur++ = ch; 309 *enc->cur++ = ch;
76} 310}
… …
78static void 312static void
79encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 313encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
80{ 314{
81 char *end = str + len; 315 char *end = str + len;
82 316
317 need (enc, len);
318
83 while (str < end) 319 while (str < end)
84 { 320 {
85 unsigned char ch = *(unsigned char *)str; 321 unsigned char ch = *(unsigned char *)str;
322
86 if (ch >= 0x20 && ch < 0x80) // most common case 323 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
87 { 324 {
325 if (expect_false (ch == '"')) // but with slow exceptions
326 {
327 need (enc, len += 1);
328 *enc->cur++ = '\\';
329 *enc->cur++ = '"';
330 }
331 else if (expect_false (ch == '\\'))
332 {
333 need (enc, len += 1);
334 *enc->cur++ = '\\';
335 *enc->cur++ = '\\';
336 }
337 else
88 *enc->cur++ = ch; 338 *enc->cur++ = ch;
339
89 str++; 340 ++str;
90 } 341 }
91 else 342 else
92 { 343 {
93 STRLEN clen; 344 switch (ch)
94 UV uch;
95
96 if (is_utf8)
97 { 345 {
98 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 346 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
99 if (clen < 0) 347 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
100 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 348 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
101 } 349 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
102 else 350 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
103 {
104 uch = ch;
105 clen = 1;
106 }
107 351
108 need (enc, len += 6); 352 default:
109
110 if (uch < 0xa0 || enc->flags & F_ASCII)
111 {
112 if (uch > 0xFFFFUL)
113 { 353 {
354 STRLEN clen;
355 UV uch;
356
357 if (is_utf8)
358 {
359 uch = decode_utf8 (str, end - str, &clen);
360 if (clen == (STRLEN)-1)
361 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
362 }
363 else
364 {
365 uch = ch;
114 len += 6; 366 clen = 1;
367 }
368
369 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
370 {
371 if (uch >= 0x10000UL)
372 {
373 if (uch >= 0x110000UL)
374 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
375
115 need (enc, len += 6); 376 need (enc, len += 11);
116 sprintf (enc->cur, "\\u%04x\\u%04x", 377 sprintf (enc->cur, "\\u%04x\\u%04x",
117 (uch - 0x10000) / 0x400 + 0xD800, 378 (int)((uch - 0x10000) / 0x400 + 0xD800),
118 (uch - 0x10000) % 0x400 + 0xDC00); 379 (int)((uch - 0x10000) % 0x400 + 0xDC00));
119 enc->cur += 12; 380 enc->cur += 12;
381 }
382 else
383 {
384 need (enc, len += 5);
385 *enc->cur++ = '\\';
386 *enc->cur++ = 'u';
387 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
388 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
389 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
390 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
391 }
392
393 str += clen;
120 } 394 }
395 else if (enc->json.flags & F_LATIN1)
396 {
397 *enc->cur++ = uch;
398 str += clen;
399 }
400 else if (is_utf8)
401 {
402 need (enc, len += clen);
403 do
404 {
405 *enc->cur++ = *str++;
406 }
407 while (--clen);
408 }
121 else 409 else
122 { 410 {
123 sprintf (enc->cur, "\\u%04x", uch); 411 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
124 enc->cur += 6; 412 enc->cur = encode_utf8 (enc->cur, uch);
413 ++str;
414 }
125 } 415 }
126 } 416 }
127 else if (is_utf8)
128 {
129 memcpy (enc->cur, str, clen);
130 enc->cur += clen;
131 }
132 else
133 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
134
135 str += clen;
136 } 417 }
137 418
138 --len; 419 --len;
139 } 420 }
140} 421}
141 422
142#define INDENT SB \ 423INLINE void
424encode_indent (enc_t *enc)
425{
143 if (enc->flags & F_INDENT) \ 426 if (enc->json.flags & F_INDENT)
144 { \ 427 {
145 int i_; \ 428 int spaces = enc->indent * INDENT_STEP;
146 need (enc, enc->indent); \ 429
147 for (i_ = enc->indent * 3; i_--; )\ 430 need (enc, spaces);
431 memset (enc->cur, ' ', spaces);
432 enc->cur += spaces;
433 }
434}
435
436INLINE void
437encode_space (enc_t *enc)
438{
439 need (enc, 1);
440 encode_ch (enc, ' ');
441}
442
443INLINE void
444encode_nl (enc_t *enc)
445{
446 if (enc->json.flags & F_INDENT)
447 {
448 need (enc, 1);
148 encode_ch (enc, ' '); \ 449 encode_ch (enc, '\n');
149 } \ 450 }
150 SE 451}
151 452
152#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 453INLINE void
153#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 454encode_comma (enc_t *enc)
154#define COMMA SB \ 455{
155 encode_ch (enc, ','); \ 456 encode_ch (enc, ',');
457
156 if (enc->flags & F_INDENT) \ 458 if (enc->json.flags & F_INDENT)
157 NL; \ 459 encode_nl (enc);
158 else if (enc->flags & F_SPACE_AFTER) \ 460 else if (enc->json.flags & F_SPACE_AFTER)
159 SPACE; \ 461 encode_space (enc);
160 SE 462}
161 463
162static void encode_sv (enc_t *enc, SV *sv); 464static void encode_sv (enc_t *enc, SV *sv);
163 465
164static void 466static void
165encode_av (enc_t *enc, AV *av) 467encode_av (enc_t *enc, AV *av)
166{ 468{
167 int i, len = av_len (av); 469 int i, len = av_len (av);
168 470
471 if (enc->indent >= enc->json.max_depth)
472 croak (ERR_NESTING_EXCEEDED);
473
169 encode_ch (enc, '['); NL; 474 encode_ch (enc, '[');
170 ++enc->indent; 475
476 if (len >= 0)
477 {
478 encode_nl (enc); ++enc->indent;
171 479
172 for (i = 0; i <= len; ++i) 480 for (i = 0; i <= len; ++i)
173 { 481 {
174 INDENT; 482 SV **svp = av_fetch (av, i, 0);
175 encode_sv (enc, *av_fetch (av, i, 0));
176 483
484 encode_indent (enc);
485
486 if (svp)
487 encode_sv (enc, *svp);
488 else
489 encode_str (enc, "null", 4, 0);
490
177 if (i < len) 491 if (i < len)
178 COMMA; 492 encode_comma (enc);
179 } 493 }
180 494
181 NL; 495 encode_nl (enc); --enc->indent; encode_indent (enc);
182 496 }
183 --enc->indent; 497
184 INDENT; encode_ch (enc, ']'); 498 encode_ch (enc, ']');
185} 499}
186 500
187static void 501static void
188encode_he (enc_t *enc, HE *he) 502encode_hk (enc_t *enc, HE *he)
189{ 503{
190 encode_ch (enc, '"'); 504 encode_ch (enc, '"');
191 505
192 if (HeKLEN (he) == HEf_SVKEY) 506 if (HeKLEN (he) == HEf_SVKEY)
193 { 507 {
194 SV *sv = HeSVKEY (he); 508 SV *sv = HeSVKEY (he);
195 STRLEN len; 509 STRLEN len;
510 char *str;
511
512 SvGETMAGIC (sv);
196 char *str = SvPV (sv, len); 513 str = SvPV (sv, len);
197 514
198 encode_str (enc, str, len, SvUTF8 (sv)); 515 encode_str (enc, str, len, SvUTF8 (sv));
199 } 516 }
200 else 517 else
201 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 518 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
202 519
203 encode_ch (enc, '"'); 520 encode_ch (enc, '"');
204 521
205 if (enc->flags & F_SPACE_BEFORE) SPACE; 522 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
206 encode_ch (enc, ':'); 523 encode_ch (enc, ':');
207 if (enc->flags & F_SPACE_AFTER ) SPACE; 524 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
208 encode_sv (enc, HeVAL (he));
209} 525}
210 526
211// compare hash entries, used when all keys are bytestrings 527// compare hash entries, used when all keys are bytestrings
212static int 528static int
213he_cmp_fast (const void *a_, const void *b_) 529he_cmp_fast (const void *a_, const void *b_)
… …
218 HE *b = *(HE **)b_; 534 HE *b = *(HE **)b_;
219 535
220 STRLEN la = HeKLEN (a); 536 STRLEN la = HeKLEN (a);
221 STRLEN lb = HeKLEN (b); 537 STRLEN lb = HeKLEN (b);
222 538
223 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 539 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
224 cmp = la < lb ? -1 : la == lb ? 0 : 1; 540 cmp = lb - la;
225 541
226 return cmp; 542 return cmp;
227} 543}
228 544
229// compare hash entries, used when some keys are sv's or utf-x 545// compare hash entries, used when some keys are sv's or utf-x
230static int 546static int
231he_cmp_slow (const void *a, const void *b) 547he_cmp_slow (const void *a, const void *b)
232{ 548{
233 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 549 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
234} 550}
235 551
236static void 552static void
237encode_hv (enc_t *enc, HV *hv) 553encode_hv (enc_t *enc, HV *hv)
238{ 554{
239 int count, i; 555 HE *he;
240 556
241 encode_ch (enc, '{'); NL; ++enc->indent; 557 if (enc->indent >= enc->json.max_depth)
558 croak (ERR_NESTING_EXCEEDED);
242 559
243 if ((count = hv_iterinit (hv))) 560 encode_ch (enc, '{');
244 { 561
245 // for canonical output we have to sort by keys first 562 // for canonical output we have to sort by keys first
246 // actually, this is mostly due to the stupid so-called 563 // actually, this is mostly due to the stupid so-called
247 // security workaround added somewhere in 5.8.x. 564 // security workaround added somewhere in 5.8.x
248 // that randomises hash orderings 565 // that randomises hash orderings
249 if (enc->flags & F_CANONICAL) 566 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
567 {
568 int count = hv_iterinit (hv);
569
570 if (SvMAGICAL (hv))
250 { 571 {
251 HE *he, *hes [count]; 572 // need to count by iterating. could improve by dynamically building the vector below
573 // but I don't care for the speed of this special case.
574 // note also that we will run into undefined behaviour when the two iterations
575 // do not result in the same count, something I might care for in some later release.
576
577 count = 0;
578 while (hv_iternext (hv))
579 ++count;
580
581 hv_iterinit (hv);
582 }
583
584 if (count)
585 {
252 int fast = 1; 586 int i, fast = 1;
587#if defined(__BORLANDC__) || defined(_MSC_VER)
588 HE **hes = _alloca (count * sizeof (HE));
589#else
590 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
591#endif
253 592
254 i = 0; 593 i = 0;
255 while ((he = hv_iternext (hv))) 594 while ((he = hv_iternext (hv)))
256 { 595 {
257 hes [i++] = he; 596 hes [i++] = he;
… …
263 602
264 if (fast) 603 if (fast)
265 qsort (hes, count, sizeof (HE *), he_cmp_fast); 604 qsort (hes, count, sizeof (HE *), he_cmp_fast);
266 else 605 else
267 { 606 {
268 // hack to disable "use bytes" 607 // hack to forcefully disable "use bytes"
269 COP *oldcop = PL_curcop, cop; 608 COP cop = *PL_curcop;
270 cop.op_private = 0; 609 cop.op_private = 0;
610
611 ENTER;
612 SAVETMPS;
613
614 SAVEVPTR (PL_curcop);
271 PL_curcop = &cop; 615 PL_curcop = &cop;
272 616
273 SAVETMPS;
274 qsort (hes, count, sizeof (HE *), he_cmp_slow); 617 qsort (hes, count, sizeof (HE *), he_cmp_slow);
618
275 FREETMPS; 619 FREETMPS;
276 620 LEAVE;
277 PL_curcop = oldcop;
278 } 621 }
279 622
280 for (i = 0; i < count; ++i) 623 encode_nl (enc); ++enc->indent;
624
625 while (count--)
281 { 626 {
282 INDENT; 627 encode_indent (enc);
628 he = hes [count];
283 encode_he (enc, hes [i]); 629 encode_hk (enc, he);
630 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
284 631
285 if (i < count - 1) 632 if (count)
286 COMMA; 633 encode_comma (enc);
287 } 634 }
288 635
636 encode_nl (enc); --enc->indent; encode_indent (enc);
637 }
638 }
639 else
640 {
641 if (hv_iterinit (hv) || SvMAGICAL (hv))
642 if ((he = hv_iternext (hv)))
289 NL; 643 {
644 encode_nl (enc); ++enc->indent;
645
646 for (;;)
647 {
648 encode_indent (enc);
649 encode_hk (enc, he);
650 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
651
652 if (!(he = hv_iternext (hv)))
653 break;
654
655 encode_comma (enc);
656 }
657
658 encode_nl (enc); --enc->indent; encode_indent (enc);
659 }
660 }
661
662 encode_ch (enc, '}');
663}
664
665// encode objects, arrays and special \0=false and \1=true values.
666static void
667encode_rv (enc_t *enc, SV *sv)
668{
669 svtype svt;
670
671 SvGETMAGIC (sv);
672 svt = SvTYPE (sv);
673
674 if (expect_false (SvOBJECT (sv)))
675 {
676 HV *stash = !JSON_SLOW || json_boolean_stash
677 ? json_boolean_stash
678 : gv_stashpv ("JSON::XS::Boolean", 1);
679
680 if (SvSTASH (sv) == stash)
681 {
682 if (SvIV (sv))
683 encode_str (enc, "true", 4, 0);
684 else
685 encode_str (enc, "false", 5, 0);
290 } 686 }
291 else 687 else
292 { 688 {
293 SV *sv; 689#if 0
294 HE *he = hv_iternext (hv); 690 if (0 && sv_derived_from (rv, "JSON::Literal"))
295
296 for (;;)
297 { 691 {
298 INDENT; 692 // not yet
299 encode_he (enc, he);
300
301 if (!(he = hv_iternext (hv)))
302 break;
303
304 COMMA;
305 } 693 }
306 694#endif
695 if (enc->json.flags & F_CONV_BLESSED)
307 NL; 696 {
697 // we re-bless the reference to get overload and other niceties right
698 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
699
700 if (to_json)
701 {
702 dSP;
703
704 ENTER; SAVETMPS; PUSHMARK (SP);
705 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
706
707 // calling with G_SCALAR ensures that we always get a 1 return value
708 PUTBACK;
709 call_sv ((SV *)GvCV (to_json), G_SCALAR);
710 SPAGAIN;
711
712 // catch this surprisingly common error
713 if (SvROK (TOPs) && SvRV (TOPs) == sv)
714 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
715
716 sv = POPs;
717 PUTBACK;
718
719 encode_sv (enc, sv);
720
721 FREETMPS; LEAVE;
722 }
723 else if (enc->json.flags & F_ALLOW_BLESSED)
724 encode_str (enc, "null", 4, 0);
725 else
726 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
727 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
728 }
729 else if (enc->json.flags & F_ALLOW_BLESSED)
730 encode_str (enc, "null", 4, 0);
731 else
732 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
733 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
308 } 734 }
309 } 735 }
736 else if (svt == SVt_PVHV)
737 encode_hv (enc, (HV *)sv);
738 else if (svt == SVt_PVAV)
739 encode_av (enc, (AV *)sv);
740 else if (svt < SVt_PVAV)
741 {
742 STRLEN len = 0;
743 char *pv = svt ? SvPV (sv, len) : 0;
310 744
311 --enc->indent; INDENT; encode_ch (enc, '}'); 745 if (len == 1 && *pv == '1')
746 encode_str (enc, "true", 4, 0);
747 else if (len == 1 && *pv == '0')
748 encode_str (enc, "false", 5, 0);
749 else if (enc->json.flags & F_ALLOW_UNKNOWN)
750 encode_str (enc, "null", 4, 0);
751 else
752 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
753 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
754 }
755 else if (enc->json.flags & F_ALLOW_UNKNOWN)
756 encode_str (enc, "null", 4, 0);
757 else
758 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
759 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312} 760}
313 761
314static void 762static void
315encode_sv (enc_t *enc, SV *sv) 763encode_sv (enc_t *enc, SV *sv)
316{ 764{
765 SvGETMAGIC (sv);
766
317 if (SvPOKp (sv)) 767 if (SvPOKp (sv))
318 { 768 {
319 STRLEN len; 769 STRLEN len;
320 char *str = SvPV (sv, len); 770 char *str = SvPV (sv, len);
321 encode_ch (enc, '"'); 771 encode_ch (enc, '"');
322 encode_str (enc, str, len, SvUTF8 (sv)); 772 encode_str (enc, str, len, SvUTF8 (sv));
323 encode_ch (enc, '"'); 773 encode_ch (enc, '"');
324 } 774 }
325 else if (SvNOKp (sv)) 775 else if (SvNOKp (sv))
326 { 776 {
777 // trust that perl will do the right thing w.r.t. JSON syntax.
327 need (enc, NV_DIG + 32); 778 need (enc, NV_DIG + 32);
328 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 779 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
329 enc->cur += strlen (enc->cur); 780 enc->cur += strlen (enc->cur);
330 } 781 }
331 else if (SvIOKp (sv)) 782 else if (SvIOKp (sv))
332 { 783 {
784 // we assume we can always read an IV as a UV and vice versa
785 // we assume two's complement
786 // we assume no aliasing issues in the union
787 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
788 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
789 {
790 // optimise the "small number case"
791 // code will likely be branchless and use only a single multiplication
792 // works for numbers up to 59074
793 I32 i = SvIVX (sv);
794 U32 u;
795 char digit, nz = 0;
796
333 need (enc, 64); 797 need (enc, 6);
798
799 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
800 u = i < 0 ? -i : i;
801
802 // convert to 4.28 fixed-point representation
803 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
804
805 // now output digit by digit, each time masking out the integer part
806 // and multiplying by 5 while moving the decimal point one to the right,
807 // resulting in a net multiplication by 10.
808 // we always write the digit to memory but conditionally increment
809 // the pointer, to enable the use of conditional move instructions.
810 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
811 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
812 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
813 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
814 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
815 }
816 else
817 {
818 // large integer, use the (rather slow) snprintf way.
819 need (enc, IVUV_MAXCHARS);
334 enc->cur += 820 enc->cur +=
335 SvIsUV(sv) 821 SvIsUV(sv)
336 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 822 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
337 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 823 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
824 }
338 } 825 }
339 else if (SvROK (sv)) 826 else if (SvROK (sv))
340 { 827 encode_rv (enc, SvRV (sv));
341 if (!--enc->max_recurse) 828 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
342 croak ("data structure too deep (hit recursion limit)");
343
344 sv = SvRV (sv);
345
346 switch (SvTYPE (sv))
347 {
348 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
349 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
350
351 default:
352 croak ("JSON can only represent references to arrays or hashes");
353 }
354 }
355 else if (!SvOK (sv))
356 encode_str (enc, "null", 4, 0); 829 encode_str (enc, "null", 4, 0);
357 else 830 else
358 croak ("encountered perl type that JSON cannot handle"); 831 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
832 SvPV_nolen (sv), SvFLAGS (sv));
359} 833}
360 834
361static SV * 835static SV *
362encode_json (SV *scalar, UV flags) 836encode_json (SV *scalar, JSON *json)
363{ 837{
364 enc_t enc; 838 enc_t enc;
365 enc.flags = flags; 839
840 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
841 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
842
843 enc.json = *json;
366 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 844 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
367 enc.cur = SvPVX (enc.sv); 845 enc.cur = SvPVX (enc.sv);
368 enc.end = SvEND (enc.sv); 846 enc.end = SvEND (enc.sv);
369 enc.max_recurse = 0;
370 enc.indent = 0; 847 enc.indent = 0;
848 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
849 : enc.json.flags & F_LATIN1 ? 0x000100UL
850 : 0x110000UL;
371 851
372 SvPOK_only (enc.sv); 852 SvPOK_only (enc.sv);
373 encode_sv (&enc, scalar); 853 encode_sv (&enc, scalar);
854 encode_nl (&enc);
374 855
856 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
857 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
858
375 if (!(flags & (F_ASCII | F_UTF8))) 859 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
376 SvUTF8_on (enc.sv); 860 SvUTF8_on (enc.sv);
377 861
378 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 862 if (enc.json.flags & F_SHRINK)
863 shrink (enc.sv);
864
379 return enc.sv; 865 return enc.sv;
380} 866}
381 867
382///////////////////////////////////////////////////////////////////////////// 868/////////////////////////////////////////////////////////////////////////////
869// decoder
383 870
384#define WS \ 871// structure used for decoding JSON
872typedef struct
873{
874 char *cur; // current parser pointer
875 char *end; // end of input string
876 const char *err; // parse error, if != 0
877 JSON json;
878 U32 depth; // recursion depth
879 U32 maxdepth; // recursion depth limit
880} dec_t;
881
882INLINE void
883decode_comment (dec_t *dec)
884{
885 // only '#'-style comments allowed a.t.m.
886
887 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
888 ++dec->cur;
889}
890
891INLINE void
892decode_ws (dec_t *dec)
893{
385 for (;;) \ 894 for (;;)
386 { \ 895 {
387 char ch = *dec->cur; \ 896 char ch = *dec->cur;
897
388 if (ch > 0x20 \ 898 if (ch > 0x20)
899 {
900 if (expect_false (ch == '#'))
901 {
902 if (dec->json.flags & F_RELAXED)
903 decode_comment (dec);
904 else
905 break;
906 }
907 else
908 break;
909 }
389 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 910 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
390 break; \ 911 break; // parse error, but let higher level handle it, gives better error messages
912
391 ++dec->cur; \ 913 ++dec->cur;
392 } 914 }
915}
393 916
394#define ERR(reason) SB dec->err = reason; goto fail; SE 917#define ERR(reason) SB dec->err = reason; goto fail; SE
918
395#define EXPECT_CH(ch) SB \ 919#define EXPECT_CH(ch) SB \
396 if (*dec->cur != ch) \ 920 if (*dec->cur != ch) \
397 ERR (# ch " expected"); \ 921 ERR (# ch " expected"); \
398 ++dec->cur; \ 922 ++dec->cur; \
399 SE 923 SE
400 924
925#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
926#define DEC_DEC_DEPTH --dec->depth
927
401static SV *decode_sv (dec_t *dec); 928static SV *decode_sv (dec_t *dec);
402
403#define APPEND_CH(ch) SB \
404 SvGROW (sv, cur + 1 + 1); \
405 SvPVX (sv)[cur++] = (ch); \
406 SE
407 929
408static signed char decode_hexdigit[256]; 930static signed char decode_hexdigit[256];
409 931
410static UV 932static UV
411decode_4hex (dec_t *dec) 933decode_4hex (dec_t *dec)
412{ 934{
413 signed char d1, d2, d3, d4; 935 signed char d1, d2, d3, d4;
936 unsigned char *cur = (unsigned char *)dec->cur;
414 937
415 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 938 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
416 if (d1 < 0) ERR ("four hexadecimal digits expected"); 939 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
417 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 940 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
418 if (d2 < 0) ERR ("four hexadecimal digits expected"); 941 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
419 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
420 if (d3 < 0) ERR ("four hexadecimal digits expected");
421 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
422 if (d4 < 0) ERR ("four hexadecimal digits expected");
423 942
424 dec->cur += 4; 943 dec->cur += 4;
425 944
426 return ((UV)d1) << 12 945 return ((UV)d1) << 12
427 | ((UV)d2) << 8 946 | ((UV)d2) << 8
… …
433} 952}
434 953
435static SV * 954static SV *
436decode_str (dec_t *dec) 955decode_str (dec_t *dec)
437{ 956{
438 SV *sv = NEWSV (0,2); 957 SV *sv = 0;
439 STRLEN cur = 0;
440 int utf8 = 0; 958 int utf8 = 0;
959 char *dec_cur = dec->cur;
441 960
442 for (;;) 961 do
443 { 962 {
444 unsigned char ch = *(unsigned char *)dec->cur; 963 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
964 char *cur = buf;
445 965
446 if (ch == '"') 966 do
447 break;
448 else if (ch == '\\')
449 { 967 {
450 switch (*++dec->cur) 968 unsigned char ch = *(unsigned char *)dec_cur++;
969
970 if (expect_false (ch == '"'))
451 { 971 {
452 case '\\': 972 --dec_cur;
453 case '/': 973 break;
454 case '"': APPEND_CH (*dec->cur++); break; 974 }
455 975 else if (expect_false (ch == '\\'))
456 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 976 {
457 case 't': APPEND_CH ('\011'); ++dec->cur; break; 977 switch (*dec_cur)
458 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
459 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
460 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
461
462 case 'u':
463 { 978 {
464 UV lo, hi; 979 case '\\':
465 ++dec->cur; 980 case '/':
981 case '"': *cur++ = *dec_cur++; break;
466 982
467 hi = decode_4hex (dec); 983 case 'b': ++dec_cur; *cur++ = '\010'; break;
468 if (hi == (UV)-1) 984 case 't': ++dec_cur; *cur++ = '\011'; break;
469 goto fail; 985 case 'n': ++dec_cur; *cur++ = '\012'; break;
986 case 'f': ++dec_cur; *cur++ = '\014'; break;
987 case 'r': ++dec_cur; *cur++ = '\015'; break;
470 988
471 // possibly a surrogate pair 989 case 'u':
472 if (hi >= 0xd800 && hi < 0xdc00)
473 { 990 {
474 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 991 UV lo, hi;
475 ERR ("illegal surrogate character"); 992 ++dec_cur;
476 993
477 dec->cur += 2; 994 dec->cur = dec_cur;
478
479 lo = decode_4hex (dec); 995 hi = decode_4hex (dec);
996 dec_cur = dec->cur;
480 if (lo == (UV)-1) 997 if (hi == (UV)-1)
481 goto fail; 998 goto fail;
482 999
1000 // possibly a surrogate pair
1001 if (hi >= 0xd800)
1002 if (hi < 0xdc00)
1003 {
1004 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
1005 ERR ("missing low surrogate character in surrogate pair");
1006
1007 dec_cur += 2;
1008
1009 dec->cur = dec_cur;
1010 lo = decode_4hex (dec);
1011 dec_cur = dec->cur;
1012 if (lo == (UV)-1)
1013 goto fail;
1014
483 if (lo < 0xdc00 || lo >= 0xe000) 1015 if (lo < 0xdc00 || lo >= 0xe000)
484 ERR ("surrogate pair expected"); 1016 ERR ("surrogate pair expected");
485 1017
486 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 1018 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
1019 }
1020 else if (hi < 0xe000)
1021 ERR ("missing high surrogate character in surrogate pair");
1022
1023 if (hi >= 0x80)
1024 {
1025 utf8 = 1;
1026
1027 cur = encode_utf8 (cur, hi);
1028 }
1029 else
1030 *cur++ = hi;
487 } 1031 }
488 else if (lo >= 0xdc00 && lo < 0xe000)
489 ERR ("illegal surrogate character");
490
491 if (hi >= 0x80)
492 { 1032 break;
493 utf8 = 1;
494 1033
495 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
496 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
497 }
498 else 1034 default:
499 APPEND_CH (hi); 1035 --dec_cur;
1036 ERR ("illegal backslash escape sequence in string");
500 } 1037 }
1038 }
1039 else if (expect_true (ch >= 0x20 && ch < 0x80))
1040 *cur++ = ch;
1041 else if (ch >= 0x80)
1042 {
1043 STRLEN clen;
1044
1045 --dec_cur;
1046
1047 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
1048 if (clen == (STRLEN)-1)
1049 ERR ("malformed UTF-8 character in JSON string");
1050
1051 do
1052 *cur++ = *dec_cur++;
1053 while (--clen);
1054
1055 utf8 = 1;
1056 }
1057 else
1058 {
1059 --dec_cur;
1060
1061 if (!ch)
1062 ERR ("unexpected end of string while parsing JSON string");
501 break; 1063 else
1064 ERR ("invalid character encountered while parsing JSON string");
502 } 1065 }
503 } 1066 }
504 else if (ch >= 0x20 && ch <= 0x7f) 1067 while (cur < buf + SHORT_STRING_LEN);
505 APPEND_CH (*dec->cur++); 1068
506 else if (ch >= 0x80) 1069 {
1070 STRLEN len = cur - buf;
1071
1072 if (sv)
507 { 1073 {
508 STRLEN clen; 1074 STRLEN cur = SvCUR (sv);
509 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
510 if (clen < 0)
511 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
512 1075
513 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits 1076 if (SvLEN (sv) <= cur + len)
514 memcpy (SvPVX (sv) + cur, dec->cur, clen); 1077 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
515 dec->cur += clen; 1078
1079 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1080 SvCUR_set (sv, SvCUR (sv) + len);
516 } 1081 }
517 else 1082 else
518 ERR ("invalid character encountered"); 1083 sv = newSVpvn (buf, len);
519 } 1084 }
1085 }
1086 while (*dec_cur != '"');
520 1087
521 ++dec->cur; 1088 ++dec_cur;
522 1089
1090 if (sv)
1091 {
523 SvPOK_only (sv); 1092 SvPOK_only (sv);
524
525 SvCUR_set (sv, cur);
526 *SvEND (sv) = 0; 1093 *SvEND (sv) = 0;
527 1094
528 if (utf8) 1095 if (utf8)
529 SvUTF8_on (sv); 1096 SvUTF8_on (sv);
1097 }
1098 else
1099 sv = newSVpvn ("", 0);
530 1100
1101 dec->cur = dec_cur;
531 return sv; 1102 return sv;
532 1103
533fail: 1104fail:
534 SvREFCNT_dec (sv); 1105 dec->cur = dec_cur;
535 return 0; 1106 return 0;
536} 1107}
537 1108
538static SV * 1109static SV *
539decode_num (dec_t *dec) 1110decode_num (dec_t *dec)
… …
549 { 1120 {
550 ++dec->cur; 1121 ++dec->cur;
551 if (*dec->cur >= '0' && *dec->cur <= '9') 1122 if (*dec->cur >= '0' && *dec->cur <= '9')
552 ERR ("malformed number (leading zero must not be followed by another digit)"); 1123 ERR ("malformed number (leading zero must not be followed by another digit)");
553 } 1124 }
554 1125 else if (*dec->cur < '0' || *dec->cur > '9')
555 // int 1126 ERR ("malformed number (no digits after initial minus)");
1127 else
1128 do
1129 {
1130 ++dec->cur;
1131 }
556 while (*dec->cur >= '0' && *dec->cur <= '9') 1132 while (*dec->cur >= '0' && *dec->cur <= '9');
557 ++dec->cur;
558 1133
559 // [frac] 1134 // [frac]
560 if (*dec->cur == '.') 1135 if (*dec->cur == '.')
561 { 1136 {
562 is_nv = 1; 1137 ++dec->cur;
1138
1139 if (*dec->cur < '0' || *dec->cur > '9')
1140 ERR ("malformed number (no digits after decimal point)");
563 1141
564 do 1142 do
565 { 1143 {
566 ++dec->cur; 1144 ++dec->cur;
567 } 1145 }
568 while (*dec->cur >= '0' && *dec->cur <= '9'); 1146 while (*dec->cur >= '0' && *dec->cur <= '9');
1147
1148 is_nv = 1;
569 } 1149 }
570 1150
571 // [exp] 1151 // [exp]
572 if (*dec->cur == 'e' || *dec->cur == 'E') 1152 if (*dec->cur == 'e' || *dec->cur == 'E')
573 { 1153 {
574 is_nv = 1;
575
576 ++dec->cur; 1154 ++dec->cur;
1155
577 if (*dec->cur == '-' || *dec->cur == '+') 1156 if (*dec->cur == '-' || *dec->cur == '+')
578 ++dec->cur; 1157 ++dec->cur;
579 1158
1159 if (*dec->cur < '0' || *dec->cur > '9')
1160 ERR ("malformed number (no digits after exp sign)");
1161
1162 do
1163 {
1164 ++dec->cur;
1165 }
580 while (*dec->cur >= '0' && *dec->cur <= '9') 1166 while (*dec->cur >= '0' && *dec->cur <= '9');
581 ++dec->cur; 1167
1168 is_nv = 1;
582 } 1169 }
583 1170
584 if (!is_nv) 1171 if (!is_nv)
585 { 1172 {
586 UV uv; 1173 int len = dec->cur - start;
587 int numtype = grok_number (start, dec->cur - start, &uv); 1174
588 if (numtype & IS_NUMBER_IN_UV) 1175 // special case the rather common 1..5-digit-int case
589 if (numtype & IS_NUMBER_NEG) 1176 if (*start == '-')
1177 switch (len)
590 { 1178 {
591 if (uv < (UV)IV_MIN) 1179 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
592 return newSViv (-(IV)uv); 1180 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1181 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1182 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1183 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
593 } 1184 }
1185 else
1186 switch (len)
1187 {
1188 case 1: return newSViv ( start [0] - '0' * 1);
1189 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1190 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1191 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1192 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1193 }
1194
1195 {
1196 UV uv;
1197 int numtype = grok_number (start, len, &uv);
1198 if (numtype & IS_NUMBER_IN_UV)
1199 if (numtype & IS_NUMBER_NEG)
1200 {
1201 if (uv < (UV)IV_MIN)
1202 return newSViv (-(IV)uv);
1203 }
594 else 1204 else
595 return newSVuv (uv); 1205 return newSVuv (uv);
596 } 1206 }
597 1207
1208 len -= *start == '-' ? 1 : 0;
1209
1210 // does not fit into IV or UV, try NV
1211 if (len <= NV_DIG)
1212 // fits into NV without loss of precision
1213 return newSVnv (json_atof (start));
1214
1215 // everything else fails, convert it to a string
1216 return newSVpvn (start, dec->cur - start);
1217 }
1218
1219 // loss of precision here
598 return newSVnv (Atof (start)); 1220 return newSVnv (json_atof (start));
599 1221
600fail: 1222fail:
601 return 0; 1223 return 0;
602} 1224}
603 1225
604static SV * 1226static SV *
605decode_av (dec_t *dec) 1227decode_av (dec_t *dec)
606{ 1228{
607 AV *av = newAV (); 1229 AV *av = newAV ();
608 1230
1231 DEC_INC_DEPTH;
1232 decode_ws (dec);
1233
1234 if (*dec->cur == ']')
1235 ++dec->cur;
1236 else
609 for (;;) 1237 for (;;)
610 { 1238 {
611 SV *value; 1239 SV *value;
612 1240
613 value = decode_sv (dec); 1241 value = decode_sv (dec);
614 if (!value) 1242 if (!value)
615 goto fail; 1243 goto fail;
616 1244
617 av_push (av, value); 1245 av_push (av, value);
618 1246
619 WS; 1247 decode_ws (dec);
620 1248
621 if (*dec->cur == ']') 1249 if (*dec->cur == ']')
622 { 1250 {
623 ++dec->cur; 1251 ++dec->cur;
624 break; 1252 break;
1253 }
625 } 1254
626
627 if (*dec->cur != ',') 1255 if (*dec->cur != ',')
628 ERR (", or ] expected while parsing array"); 1256 ERR (", or ] expected while parsing array");
629 1257
630 ++dec->cur; 1258 ++dec->cur;
1259
1260 decode_ws (dec);
1261
1262 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1263 {
1264 ++dec->cur;
1265 break;
1266 }
631 } 1267 }
632 1268
1269 DEC_DEC_DEPTH;
633 return newRV_noinc ((SV *)av); 1270 return newRV_noinc ((SV *)av);
634 1271
635fail: 1272fail:
636 SvREFCNT_dec (av); 1273 SvREFCNT_dec (av);
1274 DEC_DEC_DEPTH;
637 return 0; 1275 return 0;
638} 1276}
639 1277
640static SV * 1278static SV *
641decode_hv (dec_t *dec) 1279decode_hv (dec_t *dec)
642{ 1280{
1281 SV *sv;
643 HV *hv = newHV (); 1282 HV *hv = newHV ();
644 1283
1284 DEC_INC_DEPTH;
1285 decode_ws (dec);
1286
1287 if (*dec->cur == '}')
1288 ++dec->cur;
1289 else
645 for (;;) 1290 for (;;)
646 { 1291 {
647 SV *key, *value;
648
649 WS; EXPECT_CH ('"'); 1292 EXPECT_CH ('"');
650 1293
651 key = decode_str (dec); 1294 // heuristic: assume that
652 if (!key) 1295 // a) decode_str + hv_store_ent are abysmally slow.
653 goto fail; 1296 // b) most hash keys are short, simple ascii text.
654 1297 // => try to "fast-match" such strings to avoid
655 WS; EXPECT_CH (':'); 1298 // the overhead of decode_str + hv_store_ent.
656
657 value = decode_sv (dec);
658 if (!value)
659 { 1299 {
1300 SV *value;
1301 char *p = dec->cur;
1302 char *e = p + 24; // only try up to 24 bytes
1303
1304 for (;;)
1305 {
1306 // the >= 0x80 is false on most architectures
1307 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1308 {
1309 // slow path, back up and use decode_str
1310 SV *key = decode_str (dec);
1311 if (!key)
1312 goto fail;
1313
1314 decode_ws (dec); EXPECT_CH (':');
1315
1316 decode_ws (dec);
1317 value = decode_sv (dec);
1318 if (!value)
1319 {
1320 SvREFCNT_dec (key);
1321 goto fail;
1322 }
1323
1324 hv_store_ent (hv, key, value, 0);
660 SvREFCNT_dec (key); 1325 SvREFCNT_dec (key);
1326
1327 break;
1328 }
1329 else if (*p == '"')
1330 {
1331 // fast path, got a simple key
1332 char *key = dec->cur;
1333 int len = p - key;
1334 dec->cur = p + 1;
1335
1336 decode_ws (dec); EXPECT_CH (':');
1337
1338 decode_ws (dec);
1339 value = decode_sv (dec);
1340 if (!value)
661 goto fail; 1341 goto fail;
1342
1343 hv_store (hv, key, len, value, 0);
1344
1345 break;
1346 }
1347
1348 ++p;
1349 }
662 } 1350 }
663 1351
664 //TODO: optimise 1352 decode_ws (dec);
665 hv_store_ent (hv, key, value, 0);
666 1353
667 WS;
668
669 if (*dec->cur == '}') 1354 if (*dec->cur == '}')
1355 {
1356 ++dec->cur;
1357 break;
1358 }
1359
1360 if (*dec->cur != ',')
1361 ERR (", or } expected while parsing object/hash");
1362
1363 ++dec->cur;
1364
1365 decode_ws (dec);
1366
1367 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1368 {
1369 ++dec->cur;
1370 break;
1371 }
1372 }
1373
1374 DEC_DEC_DEPTH;
1375 sv = newRV_noinc ((SV *)hv);
1376
1377 // check filter callbacks
1378 if (dec->json.flags & F_HOOK)
1379 {
1380 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
670 { 1381 {
671 ++dec->cur; 1382 HE *cb, *he;
672 break; 1383
1384 hv_iterinit (hv);
1385 he = hv_iternext (hv);
1386 hv_iterinit (hv);
1387
1388 // the next line creates a mortal sv each time its called.
1389 // might want to optimise this for common cases.
1390 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1391
1392 if (cb)
1393 {
1394 dSP;
1395 int count;
1396
1397 ENTER; SAVETMPS; PUSHMARK (SP);
1398 XPUSHs (HeVAL (he));
1399
1400 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1401
1402 if (count == 1)
1403 {
1404 sv = newSVsv (POPs);
1405 FREETMPS; LEAVE;
1406 return sv;
1407 }
1408
1409 FREETMPS; LEAVE;
1410 }
673 } 1411 }
674 1412
675 if (*dec->cur != ',') 1413 if (dec->json.cb_object)
676 ERR (", or } expected while parsing object/hash"); 1414 {
1415 dSP;
1416 int count;
677 1417
678 ++dec->cur; 1418 ENTER; SAVETMPS; PUSHMARK (SP);
679 } 1419 XPUSHs (sv_2mortal (sv));
680 1420
681 return newRV_noinc ((SV *)hv); 1421 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1422
1423 if (count == 1)
1424 {
1425 sv = newSVsv (POPs);
1426 FREETMPS; LEAVE;
1427 return sv;
1428 }
1429
1430 SvREFCNT_inc (sv);
1431 FREETMPS; LEAVE;
1432 }
1433 }
1434
1435 return sv;
682 1436
683fail: 1437fail:
684 SvREFCNT_dec (hv); 1438 SvREFCNT_dec (hv);
1439 DEC_DEC_DEPTH;
685 return 0; 1440 return 0;
686} 1441}
687 1442
688static SV * 1443static SV *
689decode_sv (dec_t *dec) 1444decode_sv (dec_t *dec)
690{ 1445{
691 WS; 1446 // the beauty of JSON: you need exactly one character lookahead
1447 // to parse everything.
692 switch (*dec->cur) 1448 switch (*dec->cur)
693 { 1449 {
694 case '"': ++dec->cur; return decode_str (dec); 1450 case '"': ++dec->cur; return decode_str (dec);
695 case '[': ++dec->cur; return decode_av (dec); 1451 case '[': ++dec->cur; return decode_av (dec);
696 case '{': ++dec->cur; return decode_hv (dec); 1452 case '{': ++dec->cur; return decode_hv (dec);
697 1453
698 case '-': 1454 case '-':
699 case '0': case '1': case '2': case '3': case '4': 1455 case '0': case '1': case '2': case '3': case '4':
700 case '5': case '6': case '7': case '8': case '9': 1456 case '5': case '6': case '7': case '8': case '9':
701 return decode_num (dec); 1457 return decode_num (dec);
702 1458
703 case 't': 1459 case 't':
704 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1460 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
705 { 1461 {
706 dec->cur += 4; 1462 dec->cur += 4;
1463#if JSON_SLOW
1464 json_true = get_bool ("JSON::XS::true");
1465#endif
707 return newSViv (1); 1466 return newSVsv (json_true);
708 } 1467 }
709 else 1468 else
710 ERR ("'true' expected"); 1469 ERR ("'true' expected");
711 1470
712 break; 1471 break;
713 1472
714 case 'f': 1473 case 'f':
715 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1474 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
716 { 1475 {
717 dec->cur += 5; 1476 dec->cur += 5;
1477#if JSON_SLOW
1478 json_false = get_bool ("JSON::XS::false");
1479#endif
718 return newSViv (0); 1480 return newSVsv (json_false);
719 } 1481 }
720 else 1482 else
721 ERR ("'false' expected"); 1483 ERR ("'false' expected");
722 1484
723 break; 1485 break;
724 1486
725 case 'n': 1487 case 'n':
726 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1488 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
727 { 1489 {
728 dec->cur += 4; 1490 dec->cur += 4;
729 return newSViv (1); 1491 return newSVsv (&PL_sv_undef);
730 } 1492 }
731 else 1493 else
732 ERR ("'null' expected"); 1494 ERR ("'null' expected");
733 1495
734 break; 1496 break;
735 1497
736 default: 1498 default:
737 ERR ("malformed json string"); 1499 ERR ("malformed JSON string, neither array, object, number, string or atom");
738 break; 1500 break;
739 } 1501 }
740 1502
741fail: 1503fail:
742 return 0; 1504 return 0;
743} 1505}
744 1506
745static SV * 1507static SV *
746decode_json (SV *string, UV flags) 1508decode_json (SV *string, JSON *json, char **offset_return)
747{ 1509{
1510 dec_t dec;
748 SV *sv; 1511 SV *sv;
749 1512
1513 /* work around bugs in 5.10 where manipulating magic values
1514 * will perl ignore the magic in subsequent accesses
1515 */
1516 /*SvGETMAGIC (string);*/
1517 if (SvMAGICAL (string))
1518 string = sv_2mortal (newSVsv (string));
1519
1520 SvUPGRADE (string, SVt_PV);
1521
1522 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1523 * assertion with -DDEBUGGING, although SvCUR is documented to
1524 * return the xpv_cur field which certainly exists after upgrading.
1525 * according to nicholas clark, calling SvPOK fixes this.
1526 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1527 * and hope for the best.
1528 * Damnit, SvPV_nolen still trips over yet another assertion. This
1529 * assertion business is seriously broken, try yet another workaround
1530 * for the broken -DDEBUGGING.
1531 */
1532 {
1533#ifdef DEBUGGING
1534 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1535#else
1536 STRLEN offset = SvCUR (string);
1537#endif
1538
1539 if (offset > json->max_size && json->max_size)
1540 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1541 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1542 }
1543
750 if (!(flags & F_UTF8)) 1544 if (json->flags & F_UTF8)
1545 sv_utf8_downgrade (string, 0);
1546 else
751 sv_utf8_upgrade (string); 1547 sv_utf8_upgrade (string);
752 1548
753 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1549 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
754 1550
755 dec_t dec; 1551 dec.json = *json;
756 dec.flags = flags;
757 dec.cur = SvPVX (string); 1552 dec.cur = SvPVX (string);
758 dec.end = SvEND (string); 1553 dec.end = SvEND (string);
759 dec.err = 0; 1554 dec.err = 0;
1555 dec.depth = 0;
760 1556
761 *dec.end = 1; // invalid anywhere 1557 if (dec.json.cb_object || dec.json.cb_sk_object)
1558 dec.json.flags |= F_HOOK;
1559
1560 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1561
1562 decode_ws (&dec);
762 sv = decode_sv (&dec); 1563 sv = decode_sv (&dec);
763 *dec.end = 0; 1564
1565 if (offset_return)
1566 *offset_return = dec.cur;
1567
1568 if (!(offset_return || !sv))
1569 {
1570 // check for trailing garbage
1571 decode_ws (&dec);
1572
1573 if (*dec.cur)
1574 {
1575 dec.err = "garbage after JSON object";
1576 SvREFCNT_dec (sv);
1577 sv = 0;
1578 }
1579 }
764 1580
765 if (!sv) 1581 if (!sv)
766 { 1582 {
767 IV offset = utf8_distance (dec.cur, SvPVX (string));
768 SV *uni = sv_newmortal (); 1583 SV *uni = sv_newmortal ();
769 1584
1585 // horrible hack to silence warning inside pv_uni_display
1586 COP cop = *PL_curcop;
1587 cop.cop_warnings = pWARN_NONE;
1588 ENTER;
1589 SAVEVPTR (PL_curcop);
1590 PL_curcop = &cop;
770 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1591 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1592 LEAVE;
1593
771 croak ("%s, at character %d (%s)", 1594 croak ("%s, at character offset %d (before \"%s\")",
772 dec.err, 1595 dec.err,
773 (int)offset, 1596 ptr_to_index (string, dec.cur),
774 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1597 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
775 } 1598 }
776 1599
777 sv_dump (sv);//D
778 return sv_2mortal (sv); 1600 sv = sv_2mortal (sv);
1601
1602 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1603 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1604
1605 return sv;
779} 1606}
1607
1608/////////////////////////////////////////////////////////////////////////////
1609// incremental parser
1610
1611static void
1612incr_parse (JSON *self)
1613{
1614 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1615
1616 // the state machine here is a bit convoluted and could be simplified a lot
1617 // but this would make it slower, so...
1618
1619 for (;;)
1620 {
1621 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1622 switch (self->incr_mode)
1623 {
1624 // only used for initial whitespace skipping
1625 case INCR_M_WS:
1626 for (;;)
1627 {
1628 if (*p > 0x20)
1629 {
1630 if (*p == '#')
1631 {
1632 self->incr_mode = INCR_M_C0;
1633 goto incr_m_c;
1634 }
1635 else
1636 {
1637 self->incr_mode = INCR_M_JSON;
1638 goto incr_m_json;
1639 }
1640 }
1641 else if (!*p)
1642 goto interrupt;
1643
1644 ++p;
1645 }
1646
1647 // skip a single char inside a string (for \\-processing)
1648 case INCR_M_BS:
1649 if (!*p)
1650 goto interrupt;
1651
1652 ++p;
1653 self->incr_mode = INCR_M_STR;
1654 goto incr_m_str;
1655
1656 // inside #-style comments
1657 case INCR_M_C0:
1658 case INCR_M_C1:
1659 incr_m_c:
1660 for (;;)
1661 {
1662 if (*p == '\n')
1663 {
1664 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1665 break;
1666 }
1667 else if (!*p)
1668 goto interrupt;
1669
1670 ++p;
1671 }
1672
1673 break;
1674
1675 // inside a string
1676 case INCR_M_STR:
1677 incr_m_str:
1678 for (;;)
1679 {
1680 if (*p == '"')
1681 {
1682 ++p;
1683 self->incr_mode = INCR_M_JSON;
1684
1685 if (!self->incr_nest)
1686 goto interrupt;
1687
1688 goto incr_m_json;
1689 }
1690 else if (*p == '\\')
1691 {
1692 ++p; // "virtually" consumes character after \
1693
1694 if (!*p) // if at end of string we have to switch modes
1695 {
1696 self->incr_mode = INCR_M_BS;
1697 goto interrupt;
1698 }
1699 }
1700 else if (!*p)
1701 goto interrupt;
1702
1703 ++p;
1704 }
1705
1706 // after initial ws, outside string
1707 case INCR_M_JSON:
1708 incr_m_json:
1709 for (;;)
1710 {
1711 switch (*p++)
1712 {
1713 case 0:
1714 --p;
1715 goto interrupt;
1716
1717 case 0x09:
1718 case 0x0a:
1719 case 0x0d:
1720 case 0x20:
1721 if (!self->incr_nest)
1722 {
1723 --p; // do not eat the whitespace, let the next round do it
1724 goto interrupt;
1725 }
1726 break;
1727
1728 case '"':
1729 self->incr_mode = INCR_M_STR;
1730 goto incr_m_str;
1731
1732 case '[':
1733 case '{':
1734 if (++self->incr_nest > self->max_depth)
1735 croak (ERR_NESTING_EXCEEDED);
1736 break;
1737
1738 case ']':
1739 case '}':
1740 if (--self->incr_nest <= 0)
1741 goto interrupt;
1742 break;
1743
1744 case '#':
1745 self->incr_mode = INCR_M_C1;
1746 goto incr_m_c;
1747 }
1748 }
1749 }
1750
1751 modechange:
1752 ;
1753 }
1754
1755interrupt:
1756 self->incr_pos = p - SvPVX (self->incr_text);
1757 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1758 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1759}
1760
1761/////////////////////////////////////////////////////////////////////////////
1762// XS interface functions
780 1763
781MODULE = JSON::XS PACKAGE = JSON::XS 1764MODULE = JSON::XS PACKAGE = JSON::XS
782 1765
783BOOT: 1766BOOT:
784{ 1767{
785 int i; 1768 int i;
786 1769
787 memset (decode_hexdigit, 0xff, 256);
788 for (i = 10; i--; ) 1770 for (i = 0; i < 256; ++i)
789 decode_hexdigit ['0' + i] = i; 1771 decode_hexdigit [i] =
1772 i >= '0' && i <= '9' ? i - '0'
1773 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1774 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1775 : -1;
790 1776
791 for (i = 6; --i; )
792 {
793 decode_hexdigit ['a' + i] = 10 + i;
794 decode_hexdigit ['A' + i] = 10 + i;
795 }
796
797 json_stash = gv_stashpv ("JSON::XS", 1); 1777 json_stash = gv_stashpv ("JSON::XS" , 1);
798} 1778 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
799 1779
800SV *new (char *dummy) 1780 json_true = get_bool ("JSON::XS::true");
1781 json_false = get_bool ("JSON::XS::false");
1782
1783 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1784}
1785
1786PROTOTYPES: DISABLE
1787
1788void CLONE (...)
801 CODE: 1789 CODE:
802 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1790 json_stash = 0;
1791 json_boolean_stash = 0;
1792
1793void new (char *klass)
1794 PPCODE:
1795{
1796 SV *pv = NEWSV (0, sizeof (JSON));
1797 SvPOK_only (pv);
1798 json_init ((JSON *)SvPVX (pv));
1799 XPUSHs (sv_2mortal (sv_bless (
1800 newRV_noinc (pv),
1801 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1802 )));
1803}
1804
1805void ascii (JSON *self, int enable = 1)
1806 ALIAS:
1807 ascii = F_ASCII
1808 latin1 = F_LATIN1
1809 utf8 = F_UTF8
1810 indent = F_INDENT
1811 canonical = F_CANONICAL
1812 space_before = F_SPACE_BEFORE
1813 space_after = F_SPACE_AFTER
1814 pretty = F_PRETTY
1815 allow_nonref = F_ALLOW_NONREF
1816 shrink = F_SHRINK
1817 allow_blessed = F_ALLOW_BLESSED
1818 convert_blessed = F_CONV_BLESSED
1819 relaxed = F_RELAXED
1820 allow_unknown = F_ALLOW_UNKNOWN
1821 PPCODE:
1822{
1823 if (enable)
1824 self->flags |= ix;
1825 else
1826 self->flags &= ~ix;
1827
1828 XPUSHs (ST (0));
1829}
1830
1831void get_ascii (JSON *self)
1832 ALIAS:
1833 get_ascii = F_ASCII
1834 get_latin1 = F_LATIN1
1835 get_utf8 = F_UTF8
1836 get_indent = F_INDENT
1837 get_canonical = F_CANONICAL
1838 get_space_before = F_SPACE_BEFORE
1839 get_space_after = F_SPACE_AFTER
1840 get_allow_nonref = F_ALLOW_NONREF
1841 get_shrink = F_SHRINK
1842 get_allow_blessed = F_ALLOW_BLESSED
1843 get_convert_blessed = F_CONV_BLESSED
1844 get_relaxed = F_RELAXED
1845 get_allow_unknown = F_ALLOW_UNKNOWN
1846 PPCODE:
1847 XPUSHs (boolSV (self->flags & ix));
1848
1849void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1850 PPCODE:
1851 self->max_depth = max_depth;
1852 XPUSHs (ST (0));
1853
1854U32 get_max_depth (JSON *self)
1855 CODE:
1856 RETVAL = self->max_depth;
803 OUTPUT: 1857 OUTPUT:
804 RETVAL 1858 RETVAL
805 1859
806SV *ascii (SV *self, int enable) 1860void max_size (JSON *self, U32 max_size = 0)
807 ALIAS: 1861 PPCODE:
808 ascii = F_ASCII 1862 self->max_size = max_size;
809 utf8 = F_UTF8 1863 XPUSHs (ST (0));
810 indent = F_INDENT 1864
811 canonical = F_CANONICAL 1865int get_max_size (JSON *self)
812 space_before = F_SPACE_BEFORE
813 space_after = F_SPACE_AFTER
814 json_rpc = F_JSON_RPC
815 pretty = F_PRETTY
816 CODE: 1866 CODE:
817{ 1867 RETVAL = self->max_size;
818 UV *uv = SvJSON (self);
819 if (enable)
820 *uv |= ix;
821 else
822 *uv &= ~ix;
823
824 RETVAL = newSVsv (self);
825}
826 OUTPUT: 1868 OUTPUT:
827 RETVAL 1869 RETVAL
828 1870
829void encode (SV *self, SV *scalar) 1871void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
830 PPCODE: 1872 PPCODE:
831 XPUSHs (encode_json (scalar, *SvJSON (self))); 1873{
1874 SvREFCNT_dec (self->cb_object);
1875 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
832 1876
833void decode (SV *self, SV *jsonstr) 1877 XPUSHs (ST (0));
1878}
1879
1880void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
834 PPCODE: 1881 PPCODE:
835 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1882{
1883 if (!self->cb_sk_object)
1884 self->cb_sk_object = newHV ();
836 1885
837void to_json (SV *scalar) 1886 if (SvOK (cb))
1887 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1888 else
1889 {
1890 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1891
1892 if (!HvKEYS (self->cb_sk_object))
1893 {
1894 SvREFCNT_dec (self->cb_sk_object);
1895 self->cb_sk_object = 0;
1896 }
1897 }
1898
1899 XPUSHs (ST (0));
1900}
1901
1902void encode (JSON *self, SV *scalar)
838 PPCODE: 1903 PPCODE:
839 XPUSHs (encode_json (scalar, F_UTF8)); 1904 XPUSHs (encode_json (scalar, self));
840 1905
841void from_json (SV *jsonstr) 1906void decode (JSON *self, SV *jsonstr)
842 PPCODE: 1907 PPCODE:
843 XPUSHs (decode_json (jsonstr, F_UTF8)); 1908 XPUSHs (decode_json (jsonstr, self, 0));
844 1909
1910void decode_prefix (JSON *self, SV *jsonstr)
1911 PPCODE:
1912{
1913 char *offset;
1914 EXTEND (SP, 2);
1915 PUSHs (decode_json (jsonstr, self, &offset));
1916 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1917}
1918
1919void incr_parse (JSON *self, SV *jsonstr = 0)
1920 PPCODE:
1921{
1922 if (!self->incr_text)
1923 self->incr_text = newSVpvn ("", 0);
1924
1925 // append data, if any
1926 if (jsonstr)
1927 {
1928 if (SvUTF8 (jsonstr))
1929 {
1930 if (!SvUTF8 (self->incr_text))
1931 {
1932 /* utf-8-ness differs, need to upgrade */
1933 sv_utf8_upgrade (self->incr_text);
1934
1935 if (self->incr_pos)
1936 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1937 - (U8 *)SvPVX (self->incr_text);
1938 }
1939 }
1940 else if (SvUTF8 (self->incr_text))
1941 sv_utf8_upgrade (jsonstr);
1942
1943 {
1944 STRLEN len;
1945 const char *str = SvPV (jsonstr, len);
1946 STRLEN cur = SvCUR (self->incr_text);
1947
1948 if (SvLEN (self->incr_text) <= cur + len)
1949 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1950
1951 Move (str, SvEND (self->incr_text), len, char);
1952 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1953 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1954 }
1955 }
1956
1957 if (GIMME_V != G_VOID)
1958 do
1959 {
1960 char *offset;
1961
1962 if (!INCR_DONE (self))
1963 {
1964 incr_parse (self);
1965
1966 if (self->incr_pos > self->max_size && self->max_size)
1967 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1968 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1969
1970 if (!INCR_DONE (self))
1971 break;
1972 }
1973
1974 XPUSHs (decode_json (self->incr_text, self, &offset));
1975
1976 self->incr_pos -= offset - SvPVX (self->incr_text);
1977 self->incr_nest = 0;
1978 self->incr_mode = 0;
1979
1980 sv_chop (self->incr_text, offset);
1981 }
1982 while (GIMME_V == G_ARRAY);
1983}
1984
1985SV *incr_text (JSON *self)
1986 ATTRS: lvalue
1987 CODE:
1988{
1989 if (self->incr_pos)
1990 croak ("incr_text can not be called when the incremental parser already started parsing");
1991
1992 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1993}
1994 OUTPUT:
1995 RETVAL
1996
1997void incr_skip (JSON *self)
1998 CODE:
1999{
2000 if (self->incr_pos)
2001 {
2002 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2003 self->incr_pos = 0;
2004 self->incr_nest = 0;
2005 self->incr_mode = 0;
2006 }
2007}
2008
2009void incr_reset (JSON *self)
2010 CODE:
2011{
2012 SvREFCNT_dec (self->incr_text);
2013 self->incr_text = 0;
2014 self->incr_pos = 0;
2015 self->incr_nest = 0;
2016 self->incr_mode = 0;
2017}
2018
2019void DESTROY (JSON *self)
2020 CODE:
2021 SvREFCNT_dec (self->cb_sk_object);
2022 SvREFCNT_dec (self->cb_object);
2023 SvREFCNT_dec (self->incr_text);
2024
2025PROTOTYPES: ENABLE
2026
2027void encode_json (SV *scalar)
2028 ALIAS:
2029 to_json_ = 0
2030 encode_json = F_UTF8
2031 PPCODE:
2032{
2033 JSON json;
2034 json_init (&json);
2035 json.flags |= ix;
2036 XPUSHs (encode_json (scalar, &json));
2037}
2038
2039void decode_json (SV *jsonstr)
2040 ALIAS:
2041 from_json_ = 0
2042 decode_json = F_UTF8
2043 PPCODE:
2044{
2045 JSON json;
2046 json_init (&json);
2047 json.flags |= ix;
2048 XPUSHs (decode_json (jsonstr, &json, 0));
2049}
2050

Diff Legend

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