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 1.4 by root, Thu Mar 22 21:13:58 2007 UTC vs.
Revision 1.110 by root, Wed Jun 1 13:01:09 2011 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
16#define F_ALLOW_NONREF 0x00000080 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
17 39
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0
20 41
21#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
46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
22 48
23#define SB do { 49#define SB do {
24#define SE } while (0) 50#define SE } while (0)
25 51
26static HV *json_stash; 52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr), (value))
54# define INLINE static inline
55#else
56# define expect(expr,value) (expr)
57# define INLINE static
58#endif
59
60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#define IN_RANGE_INC(type,val,beg,end) \
64 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
65 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
66
67#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
68
69#ifdef USE_ITHREADS
70# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
72#else
73# define JSON_SLOW 0
74# define JSON_STASH json_stash
75#endif
76
77static HV *json_stash, *json_boolean_stash; // JSON::XS::
78static SV *json_true, *json_false;
79
80enum {
81 INCR_M_WS = 0, // initial whitespace skipping, must be 0
82 INCR_M_STR, // inside string
83 INCR_M_BS, // inside backslash
84 INCR_M_C0, // inside comment in initial whitespace sequence
85 INCR_M_C1, // inside comment in other places
86 INCR_M_JSON // outside anything, count nesting
87};
88
89#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
90
91typedef struct {
92 U32 flags;
93 U32 max_depth;
94 STRLEN max_size;
95
96 SV *cb_object;
97 HV *cb_sk_object;
98
99 // for the incremental parser
100 SV *incr_text; // the source text so far
101 STRLEN incr_pos; // the current offset into the text
102 int incr_nest; // {[]}-nesting level
103 unsigned char incr_mode;
104} JSON;
105
106INLINE void
107json_init (JSON *json)
108{
109 Zero (json, 1, JSON);
110 json->max_depth = 512;
111}
112
113/////////////////////////////////////////////////////////////////////////////
114// utility functions
115
116INLINE SV *
117get_bool (const char *name)
118{
119 SV *sv = get_sv (name, 1);
120
121 SvREADONLY_on (sv);
122 SvREADONLY_on (SvRV (sv));
123
124 return sv;
125}
126
127INLINE void
128shrink (SV *sv)
129{
130 sv_utf8_downgrade (sv, 1);
131
132 if (SvLEN (sv) > SvCUR (sv) + 1)
133 {
134#ifdef SvPV_shrink_to_cur
135 SvPV_shrink_to_cur (sv);
136#elif defined (SvPV_renew)
137 SvPV_renew (sv, SvCUR (sv) + 1);
138#endif
139 }
140}
141
142// decode an utf-8 character and return it, or (UV)-1 in
143// case of an error.
144// we special-case "safe" characters from U+80 .. U+7FF,
145// but use the very good perl function to parse anything else.
146// note that we never call this function for a ascii codepoints
147INLINE UV
148decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
149{
150 if (expect_true (len >= 2
151 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
152 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
153 {
154 *clen = 2;
155 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
156 }
157 else
158 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
159}
160
161// likewise for encoding, also never called for ascii codepoints
162// this function takes advantage of this fact, although current gccs
163// seem to optimise the check for >= 0x80 away anyways
164INLINE unsigned char *
165encode_utf8 (unsigned char *s, UV ch)
166{
167 if (expect_false (ch < 0x000080))
168 *s++ = ch;
169 else if (expect_true (ch < 0x000800))
170 *s++ = 0xc0 | ( ch >> 6),
171 *s++ = 0x80 | ( ch & 0x3f);
172 else if ( ch < 0x010000)
173 *s++ = 0xe0 | ( ch >> 12),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176 else if ( ch < 0x110000)
177 *s++ = 0xf0 | ( ch >> 18),
178 *s++ = 0x80 | ((ch >> 12) & 0x3f),
179 *s++ = 0x80 | ((ch >> 6) & 0x3f),
180 *s++ = 0x80 | ( ch & 0x3f);
181
182 return s;
183}
184
185// convert offset pointer to character index, sv must be string
186static STRLEN
187ptr_to_index (SV *sv, char *offset)
188{
189 return SvUTF8 (sv)
190 ? utf8_distance (offset, SvPVX (sv))
191 : offset - SvPVX (sv);
192}
193
194/////////////////////////////////////////////////////////////////////////////
195// fp hell
196
197// scan a group of digits, and a trailing exponent
198static void
199json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
200{
201 UV uaccum = 0;
202 int eaccum = 0;
203
204 // if we recurse too deep, skip all remaining digits
205 // to avoid a stack overflow attack
206 if (expect_false (--maxdepth <= 0))
207 while (((U8)*s - '0') < 10)
208 ++s;
209
210 for (;;)
211 {
212 U8 dig = (U8)*s - '0';
213
214 if (expect_false (dig >= 10))
215 {
216 if (dig == (U8)((U8)'.' - (U8)'0'))
217 {
218 ++s;
219 json_atof_scan1 (s, accum, expo, 1, maxdepth);
220 }
221 else if ((dig | ' ') == 'e' - '0')
222 {
223 int exp2 = 0;
224 int neg = 0;
225
226 ++s;
227
228 if (*s == '-')
229 {
230 ++s;
231 neg = 1;
232 }
233 else if (*s == '+')
234 ++s;
235
236 while ((dig = (U8)*s - '0') < 10)
237 exp2 = exp2 * 10 + *s++ - '0';
238
239 *expo += neg ? -exp2 : exp2;
240 }
241
242 break;
243 }
244
245 ++s;
246
247 uaccum = uaccum * 10 + dig;
248 ++eaccum;
249
250 // if we have too many digits, then recurse for more
251 // we actually do this for rather few digits
252 if (uaccum >= (UV_MAX - 9) / 10)
253 {
254 if (postdp) *expo -= eaccum;
255 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
256 if (postdp) *expo += eaccum;
257
258 break;
259 }
260 }
261
262 // this relies greatly on the quality of the pow ()
263 // implementation of the platform, but a good
264 // implementation is hard to beat.
265 // (IEEE 754 conformant ones are required to be exact)
266 if (postdp) *expo -= eaccum;
267 *accum += uaccum * Perl_pow (10., *expo);
268 *expo += eaccum;
269}
270
271static NV
272json_atof (const char *s)
273{
274 NV accum = 0.;
275 int expo = 0;
276 int neg = 0;
277
278 if (*s == '-')
279 {
280 ++s;
281 neg = 1;
282 }
283
284 // a recursion depth of ten gives us >>500 bits
285 json_atof_scan1 (s, &accum, &expo, 0, 10);
286
287 return neg ? -accum : accum;
288}
289/////////////////////////////////////////////////////////////////////////////
290// encoder
27 291
28// structure used for encoding JSON 292// structure used for encoding JSON
29typedef struct 293typedef struct
30{ 294{
31 char *cur; 295 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 296 char *end; // SvEND (sv)
34 SV *sv; 297 SV *sv; // result scalar
35 UV flags; 298 JSON json;
36 int max_recurse; 299 U32 indent; // indentation level
37 int indent; 300 UV limit; // escape character values >= this value when encoding
38} enc_t; 301} enc_t;
39 302
40// structure used for decoding JSON 303INLINE void
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58/////////////////////////////////////////////////////////////////////////////
59
60static void
61need (enc_t *enc, STRLEN len) 304need (enc_t *enc, STRLEN len)
62{ 305{
63 if (enc->cur + len >= enc->end) 306 if (expect_false (enc->cur + len >= enc->end))
64 { 307 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 308 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 309 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 310 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 311 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 312 }
70} 313}
71 314
72static void 315INLINE void
73encode_ch (enc_t *enc, char ch) 316encode_ch (enc_t *enc, char ch)
74{ 317{
75 need (enc, 1); 318 need (enc, 1);
76 *enc->cur++ = ch; 319 *enc->cur++ = ch;
77} 320}
… …
85 328
86 while (str < end) 329 while (str < end)
87 { 330 {
88 unsigned char ch = *(unsigned char *)str; 331 unsigned char ch = *(unsigned char *)str;
89 332
90 if (ch == '"') 333 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 334 {
335 if (expect_false (ch == '"')) // but with slow exceptions
336 {
92 need (enc, len += 1); 337 need (enc, len += 1);
93 *enc->cur++ = '\\'; 338 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 339 *enc->cur++ = '"';
95 ++str;
96 } 340 }
97 else if (ch == '\\') 341 else if (expect_false (ch == '\\'))
98 { 342 {
99 need (enc, len += 1); 343 need (enc, len += 1);
100 *enc->cur++ = '\\'; 344 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 345 *enc->cur++ = '\\';
102 ++str;
103 } 346 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 347 else
105 {
106 *enc->cur++ = ch; 348 *enc->cur++ = ch;
107 ++str; 349
108 }
109 else if (ch == '\015')
110 {
111 need (enc, len += 1);
112 *enc->cur++ = '\\';
113 *enc->cur++ = 'r';
114 ++str;
115 }
116 else if (ch == '\012')
117 {
118 need (enc, len += 1);
119 *enc->cur++ = '\\';
120 *enc->cur++ = 'n';
121 ++str; 350 ++str;
122 } 351 }
123 else 352 else
124 { 353 {
125 STRLEN clen; 354 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
131 if (clen < 0)
132 croak ("malformed UTF-8 character in string, cannot convert to JSON");
133 } 355 {
134 else 356 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
135 { 357 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
136 uch = ch; 358 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
137 clen = 1; 359 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
138 } 360 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
139 361
140 if (uch < 0x80 || enc->flags & F_ASCII) 362 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 363 {
364 STRLEN clen;
365 UV uch;
366
367 if (is_utf8)
368 {
369 uch = decode_utf8 (str, end - str, &clen);
370 if (clen == (STRLEN)-1)
371 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
372 }
373 else
374 {
375 uch = ch;
376 clen = 1;
377 }
378
379 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
380 {
381 if (uch >= 0x10000UL)
382 {
383 if (uch >= 0x110000UL)
384 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
385
144 need (enc, len += 11); 386 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 387 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 388 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 389 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 390 enc->cur += 12;
391 }
392 else
393 {
394 need (enc, len += 5);
395 *enc->cur++ = '\\';
396 *enc->cur++ = 'u';
397 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
398 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
399 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
400 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
401 }
402
403 str += clen;
404 }
405 else if (enc->json.flags & F_LATIN1)
406 {
407 *enc->cur++ = uch;
408 str += clen;
409 }
410 else if (is_utf8)
411 {
412 need (enc, len += clen);
413 do
414 {
415 *enc->cur++ = *str++;
416 }
417 while (--clen);
418 }
419 else
420 {
421 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
422 enc->cur = encode_utf8 (enc->cur, uch);
423 ++str;
424 }
149 } 425 }
150 else
151 {
152 static char hexdigit [16] = "0123456789abcdef";
153 need (enc, len += 5);
154 *enc->cur++ = '\\';
155 *enc->cur++ = 'u';
156 *enc->cur++ = hexdigit [ uch >> 12 ];
157 *enc->cur++ = hexdigit [(uch >> 8) & 15];
158 *enc->cur++ = hexdigit [(uch >> 4) & 15];
159 *enc->cur++ = hexdigit [(uch >> 0) & 15];
160 }
161
162 str += clen;
163 }
164 else if (is_utf8)
165 {
166 need (enc, len += clen);
167 while (clen--)
168 *enc->cur++ = *str++;
169 }
170 else
171 {
172 need (enc, 10); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str;
175 } 426 }
176 } 427 }
177 428
178 --len; 429 --len;
179 } 430 }
180} 431}
181 432
182#define INDENT SB \ 433INLINE void
434encode_indent (enc_t *enc)
435{
183 if (enc->flags & F_INDENT) \ 436 if (enc->json.flags & F_INDENT)
184 { \ 437 {
185 int i_; \ 438 int spaces = enc->indent * INDENT_STEP;
186 need (enc, enc->indent); \ 439
187 for (i_ = enc->indent * 3; i_--; )\ 440 need (enc, spaces);
441 memset (enc->cur, ' ', spaces);
442 enc->cur += spaces;
443 }
444}
445
446INLINE void
447encode_space (enc_t *enc)
448{
449 need (enc, 1);
450 encode_ch (enc, ' ');
451}
452
453INLINE void
454encode_nl (enc_t *enc)
455{
456 if (enc->json.flags & F_INDENT)
457 {
458 need (enc, 1);
188 encode_ch (enc, ' '); \ 459 encode_ch (enc, '\n');
189 } \ 460 }
190 SE 461}
191 462
192#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 463INLINE void
193#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 464encode_comma (enc_t *enc)
194#define COMMA SB \ 465{
195 encode_ch (enc, ','); \ 466 encode_ch (enc, ',');
467
196 if (enc->flags & F_INDENT) \ 468 if (enc->json.flags & F_INDENT)
197 NL; \ 469 encode_nl (enc);
198 else if (enc->flags & F_SPACE_AFTER) \ 470 else if (enc->json.flags & F_SPACE_AFTER)
199 SPACE; \ 471 encode_space (enc);
200 SE 472}
201 473
202static void encode_sv (enc_t *enc, SV *sv); 474static void encode_sv (enc_t *enc, SV *sv);
203 475
204static void 476static void
205encode_av (enc_t *enc, AV *av) 477encode_av (enc_t *enc, AV *av)
206{ 478{
207 int i, len = av_len (av); 479 int i, len = av_len (av);
208 480
481 if (enc->indent >= enc->json.max_depth)
482 croak (ERR_NESTING_EXCEEDED);
483
209 encode_ch (enc, '['); NL; 484 encode_ch (enc, '[');
210 ++enc->indent; 485
486 if (len >= 0)
487 {
488 encode_nl (enc); ++enc->indent;
211 489
212 for (i = 0; i <= len; ++i) 490 for (i = 0; i <= len; ++i)
213 { 491 {
214 INDENT; 492 SV **svp = av_fetch (av, i, 0);
215 encode_sv (enc, *av_fetch (av, i, 0));
216 493
494 encode_indent (enc);
495
496 if (svp)
497 encode_sv (enc, *svp);
498 else
499 encode_str (enc, "null", 4, 0);
500
217 if (i < len) 501 if (i < len)
218 COMMA; 502 encode_comma (enc);
219 } 503 }
220 504
221 NL; 505 encode_nl (enc); --enc->indent; encode_indent (enc);
222 506 }
223 --enc->indent; 507
224 INDENT; encode_ch (enc, ']'); 508 encode_ch (enc, ']');
225} 509}
226 510
227static void 511static void
228encode_he (enc_t *enc, HE *he) 512encode_hk (enc_t *enc, HE *he)
229{ 513{
230 encode_ch (enc, '"'); 514 encode_ch (enc, '"');
231 515
232 if (HeKLEN (he) == HEf_SVKEY) 516 if (HeKLEN (he) == HEf_SVKEY)
233 { 517 {
… …
243 else 527 else
244 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 528 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
245 529
246 encode_ch (enc, '"'); 530 encode_ch (enc, '"');
247 531
248 if (enc->flags & F_SPACE_BEFORE) SPACE; 532 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
249 encode_ch (enc, ':'); 533 encode_ch (enc, ':');
250 if (enc->flags & F_SPACE_AFTER ) SPACE; 534 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
251 encode_sv (enc, HeVAL (he));
252} 535}
253 536
254// compare hash entries, used when all keys are bytestrings 537// compare hash entries, used when all keys are bytestrings
255static int 538static int
256he_cmp_fast (const void *a_, const void *b_) 539he_cmp_fast (const void *a_, const void *b_)
… …
261 HE *b = *(HE **)b_; 544 HE *b = *(HE **)b_;
262 545
263 STRLEN la = HeKLEN (a); 546 STRLEN la = HeKLEN (a);
264 STRLEN lb = HeKLEN (b); 547 STRLEN lb = HeKLEN (b);
265 548
266 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 549 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
267 cmp = la < lb ? -1 : la == lb ? 0 : 1; 550 cmp = lb - la;
268 551
269 return cmp; 552 return cmp;
270} 553}
271 554
272// compare hash entries, used when some keys are sv's or utf-x 555// compare hash entries, used when some keys are sv's or utf-x
273static int 556static int
274he_cmp_slow (const void *a, const void *b) 557he_cmp_slow (const void *a, const void *b)
275{ 558{
276 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 559 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
277} 560}
278 561
279static void 562static void
280encode_hv (enc_t *enc, HV *hv) 563encode_hv (enc_t *enc, HV *hv)
281{ 564{
282 int count, i; 565 HE *he;
283 566
284 encode_ch (enc, '{'); NL; ++enc->indent; 567 if (enc->indent >= enc->json.max_depth)
568 croak (ERR_NESTING_EXCEEDED);
285 569
286 if ((count = hv_iterinit (hv))) 570 encode_ch (enc, '{');
287 { 571
288 // for canonical output we have to sort by keys first 572 // for canonical output we have to sort by keys first
289 // actually, this is mostly due to the stupid so-called 573 // actually, this is mostly due to the stupid so-called
290 // security workaround added somewhere in 5.8.x. 574 // security workaround added somewhere in 5.8.x
291 // that randomises hash orderings 575 // that randomises hash orderings
292 if (enc->flags & F_CANONICAL) 576 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
577 {
578 int count = hv_iterinit (hv);
579
580 if (SvMAGICAL (hv))
293 { 581 {
294 HE *he, *hes [count]; 582 // need to count by iterating. could improve by dynamically building the vector below
583 // but I don't care for the speed of this special case.
584 // note also that we will run into undefined behaviour when the two iterations
585 // do not result in the same count, something I might care for in some later release.
586
587 count = 0;
588 while (hv_iternext (hv))
589 ++count;
590
591 hv_iterinit (hv);
592 }
593
594 if (count)
595 {
295 int fast = 1; 596 int i, fast = 1;
597#if defined(__BORLANDC__) || defined(_MSC_VER)
598 HE **hes = _alloca (count * sizeof (HE));
599#else
600 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
601#endif
296 602
297 i = 0; 603 i = 0;
298 while ((he = hv_iternext (hv))) 604 while ((he = hv_iternext (hv)))
299 { 605 {
300 hes [i++] = he; 606 hes [i++] = he;
… …
306 612
307 if (fast) 613 if (fast)
308 qsort (hes, count, sizeof (HE *), he_cmp_fast); 614 qsort (hes, count, sizeof (HE *), he_cmp_fast);
309 else 615 else
310 { 616 {
311 // hack to disable "use bytes" 617 // hack to forcefully disable "use bytes"
312 COP *oldcop = PL_curcop, cop; 618 COP cop = *PL_curcop;
313 cop.op_private = 0; 619 cop.op_private = 0;
620
621 ENTER;
622 SAVETMPS;
623
624 SAVEVPTR (PL_curcop);
314 PL_curcop = &cop; 625 PL_curcop = &cop;
315 626
316 SAVETMPS;
317 qsort (hes, count, sizeof (HE *), he_cmp_slow); 627 qsort (hes, count, sizeof (HE *), he_cmp_slow);
628
318 FREETMPS; 629 FREETMPS;
630 LEAVE;
631 }
319 632
320 PL_curcop = oldcop; 633 encode_nl (enc); ++enc->indent;
634
635 while (count--)
321 } 636 {
637 encode_indent (enc);
638 he = hes [count];
639 encode_hk (enc, he);
640 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
322 641
323 for (i = 0; i < count; ++i) 642 if (count)
643 encode_comma (enc);
644 }
645
646 encode_nl (enc); --enc->indent; encode_indent (enc);
647 }
648 }
649 else
650 {
651 if (hv_iterinit (hv) || SvMAGICAL (hv))
652 if ((he = hv_iternext (hv)))
653 {
654 encode_nl (enc); ++enc->indent;
655
656 for (;;)
324 { 657 {
325 INDENT; 658 encode_indent (enc);
326 encode_he (enc, hes [i]); 659 encode_hk (enc, he);
660 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
327 661
328 if (i < count - 1) 662 if (!(he = hv_iternext (hv)))
329 COMMA; 663 break;
664
665 encode_comma (enc);
330 } 666 }
331 667
668 encode_nl (enc); --enc->indent; encode_indent (enc);
332 NL; 669 }
670 }
671
672 encode_ch (enc, '}');
673}
674
675// encode objects, arrays and special \0=false and \1=true values.
676static void
677encode_rv (enc_t *enc, SV *sv)
678{
679 svtype svt;
680
681 SvGETMAGIC (sv);
682 svt = SvTYPE (sv);
683
684 if (expect_false (SvOBJECT (sv)))
685 {
686 HV *stash = !JSON_SLOW || json_boolean_stash
687 ? json_boolean_stash
688 : gv_stashpv ("JSON::XS::Boolean", 1);
689
690 if (SvSTASH (sv) == stash)
691 {
692 if (SvIV (sv))
693 encode_str (enc, "true", 4, 0);
694 else
695 encode_str (enc, "false", 5, 0);
333 } 696 }
334 else 697 else
335 { 698 {
336 SV *sv; 699#if 0
337 HE *he = hv_iternext (hv); 700 if (0 && sv_derived_from (rv, "JSON::Literal"))
338
339 for (;;)
340 {
341 INDENT;
342 encode_he (enc, he);
343
344 if (!(he = hv_iternext (hv)))
345 break;
346
347 COMMA;
348 } 701 {
349 702 // not yet
350 NL; 703 }
704#endif
705 if (enc->json.flags & F_CONV_BLESSED)
706 {
707 // we re-bless the reference to get overload and other niceties right
708 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
709
710 if (to_json)
711 {
712 dSP;
713
714 ENTER; SAVETMPS; PUSHMARK (SP);
715 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
716
717 // calling with G_SCALAR ensures that we always get a 1 return value
718 PUTBACK;
719 call_sv ((SV *)GvCV (to_json), G_SCALAR);
720 SPAGAIN;
721
722 // catch this surprisingly common error
723 if (SvROK (TOPs) && SvRV (TOPs) == sv)
724 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
725
726 sv = POPs;
727 PUTBACK;
728
729 encode_sv (enc, sv);
730
731 FREETMPS; LEAVE;
732 }
733 else if (enc->json.flags & F_ALLOW_BLESSED)
734 encode_str (enc, "null", 4, 0);
735 else
736 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
737 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
738 }
739 else if (enc->json.flags & F_ALLOW_BLESSED)
740 encode_str (enc, "null", 4, 0);
741 else
742 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
743 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
351 } 744 }
352 } 745 }
746 else if (svt == SVt_PVHV)
747 encode_hv (enc, (HV *)sv);
748 else if (svt == SVt_PVAV)
749 encode_av (enc, (AV *)sv);
750 else if (svt < SVt_PVAV)
751 {
752 STRLEN len = 0;
753 char *pv = svt ? SvPV (sv, len) : 0;
353 754
354 --enc->indent; INDENT; encode_ch (enc, '}'); 755 if (len == 1 && *pv == '1')
756 encode_str (enc, "true", 4, 0);
757 else if (len == 1 && *pv == '0')
758 encode_str (enc, "false", 5, 0);
759 else if (enc->json.flags & F_ALLOW_UNKNOWN)
760 encode_str (enc, "null", 4, 0);
761 else
762 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
763 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
764 }
765 else if (enc->json.flags & F_ALLOW_UNKNOWN)
766 encode_str (enc, "null", 4, 0);
767 else
768 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
769 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
355} 770}
356 771
357static void 772static void
358encode_sv (enc_t *enc, SV *sv) 773encode_sv (enc_t *enc, SV *sv)
359{ 774{
… …
367 encode_str (enc, str, len, SvUTF8 (sv)); 782 encode_str (enc, str, len, SvUTF8 (sv));
368 encode_ch (enc, '"'); 783 encode_ch (enc, '"');
369 } 784 }
370 else if (SvNOKp (sv)) 785 else if (SvNOKp (sv))
371 { 786 {
787 // trust that perl will do the right thing w.r.t. JSON syntax.
372 need (enc, NV_DIG + 32); 788 need (enc, NV_DIG + 32);
373 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 789 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
374 enc->cur += strlen (enc->cur); 790 enc->cur += strlen (enc->cur);
375 } 791 }
376 else if (SvIOKp (sv)) 792 else if (SvIOKp (sv))
377 { 793 {
794 // we assume we can always read an IV as a UV and vice versa
795 // we assume two's complement
796 // we assume no aliasing issues in the union
797 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
798 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
799 {
800 // optimise the "small number case"
801 // code will likely be branchless and use only a single multiplication
802 // works for numbers up to 59074
803 I32 i = SvIVX (sv);
804 U32 u;
805 char digit, nz = 0;
806
378 need (enc, 64); 807 need (enc, 6);
808
809 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
810 u = i < 0 ? -i : i;
811
812 // convert to 4.28 fixed-point representation
813 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
814
815 // now output digit by digit, each time masking out the integer part
816 // and multiplying by 5 while moving the decimal point one to the right,
817 // resulting in a net multiplication by 10.
818 // we always write the digit to memory but conditionally increment
819 // the pointer, to enable the use of conditional move instructions.
820 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
821 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
822 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
823 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
824 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
825 }
826 else
827 {
828 // large integer, use the (rather slow) snprintf way.
829 need (enc, IVUV_MAXCHARS);
379 enc->cur += 830 enc->cur +=
380 SvIsUV(sv) 831 SvIsUV(sv)
381 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 832 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
382 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 833 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
834 }
383 } 835 }
384 else if (SvROK (sv)) 836 else if (SvROK (sv))
385 { 837 encode_rv (enc, SvRV (sv));
386 if (!--enc->max_recurse) 838 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
387 croak ("data structure too deep (hit recursion limit)");
388
389 sv = SvRV (sv);
390
391 switch (SvTYPE (sv))
392 {
393 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
394 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
395
396 default:
397 croak ("JSON can only represent references to arrays or hashes");
398 }
399 }
400 else if (!SvOK (sv))
401 encode_str (enc, "null", 4, 0); 839 encode_str (enc, "null", 4, 0);
402 else 840 else
403 croak ("encountered perl type that JSON cannot handle"); 841 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
842 SvPV_nolen (sv), SvFLAGS (sv));
404} 843}
405 844
406static SV * 845static SV *
407encode_json (SV *scalar, UV flags) 846encode_json (SV *scalar, JSON *json)
408{ 847{
409 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
410 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");
411
412 enc_t enc; 848 enc_t enc;
413 enc.flags = flags; 849
850 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
851 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
852
853 enc.json = *json;
414 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 854 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
415 enc.cur = SvPVX (enc.sv); 855 enc.cur = SvPVX (enc.sv);
416 enc.end = SvEND (enc.sv); 856 enc.end = SvEND (enc.sv);
417 enc.max_recurse = 0;
418 enc.indent = 0; 857 enc.indent = 0;
858 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
859 : enc.json.flags & F_LATIN1 ? 0x000100UL
860 : 0x110000UL;
419 861
420 SvPOK_only (enc.sv); 862 SvPOK_only (enc.sv);
421 encode_sv (&enc, scalar); 863 encode_sv (&enc, scalar);
864 encode_nl (&enc);
422 865
866 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
867 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
868
423 if (!(flags & (F_ASCII | F_UTF8))) 869 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
424 SvUTF8_on (enc.sv); 870 SvUTF8_on (enc.sv);
425 871
426 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 872 if (enc.json.flags & F_SHRINK)
873 shrink (enc.sv);
874
427 return enc.sv; 875 return enc.sv;
428} 876}
429 877
430///////////////////////////////////////////////////////////////////////////// 878/////////////////////////////////////////////////////////////////////////////
879// decoder
431 880
432#define WS \ 881// structure used for decoding JSON
882typedef struct
883{
884 char *cur; // current parser pointer
885 char *end; // end of input string
886 const char *err; // parse error, if != 0
887 JSON json;
888 U32 depth; // recursion depth
889 U32 maxdepth; // recursion depth limit
890} dec_t;
891
892INLINE void
893decode_comment (dec_t *dec)
894{
895 // only '#'-style comments allowed a.t.m.
896
897 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
898 ++dec->cur;
899}
900
901INLINE void
902decode_ws (dec_t *dec)
903{
433 for (;;) \ 904 for (;;)
434 { \ 905 {
435 char ch = *dec->cur; \ 906 char ch = *dec->cur;
907
436 if (ch > 0x20 \ 908 if (ch > 0x20)
909 {
910 if (expect_false (ch == '#'))
911 {
912 if (dec->json.flags & F_RELAXED)
913 decode_comment (dec);
914 else
915 break;
916 }
917 else
918 break;
919 }
437 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 920 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
438 break; \ 921 break; // parse error, but let higher level handle it, gives better error messages
922
439 ++dec->cur; \ 923 ++dec->cur;
440 } 924 }
925}
441 926
442#define ERR(reason) SB dec->err = reason; goto fail; SE 927#define ERR(reason) SB dec->err = reason; goto fail; SE
928
443#define EXPECT_CH(ch) SB \ 929#define EXPECT_CH(ch) SB \
444 if (*dec->cur != ch) \ 930 if (*dec->cur != ch) \
445 ERR (# ch " expected"); \ 931 ERR (# ch " expected"); \
446 ++dec->cur; \ 932 ++dec->cur; \
447 SE 933 SE
448 934
935#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
936#define DEC_DEC_DEPTH --dec->depth
937
449static SV *decode_sv (dec_t *dec); 938static SV *decode_sv (dec_t *dec);
450 939
451static signed char decode_hexdigit[256]; 940static signed char decode_hexdigit[256];
452 941
453static UV 942static UV
454decode_4hex (dec_t *dec) 943decode_4hex (dec_t *dec)
455{ 944{
456 signed char d1, d2, d3, d4; 945 signed char d1, d2, d3, d4;
946 unsigned char *cur = (unsigned char *)dec->cur;
457 947
458 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 948 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
459 if (d1 < 0) ERR ("four hexadecimal digits expected"); 949 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
460 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 950 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
461 if (d2 < 0) ERR ("four hexadecimal digits expected"); 951 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
462 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
463 if (d3 < 0) ERR ("four hexadecimal digits expected");
464 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
465 if (d4 < 0) ERR ("four hexadecimal digits expected");
466 952
467 dec->cur += 4; 953 dec->cur += 4;
468 954
469 return ((UV)d1) << 12 955 return ((UV)d1) << 12
470 | ((UV)d2) << 8 956 | ((UV)d2) << 8
… …
473 959
474fail: 960fail:
475 return (UV)-1; 961 return (UV)-1;
476} 962}
477 963
478#define APPEND_GROW(n) SB \
479 if (cur + (n) >= end) \
480 { \
481 STRLEN ofs = cur - SvPVX (sv); \
482 SvGROW (sv, ofs + (n) + 1); \
483 cur = SvPVX (sv) + ofs; \
484 end = SvEND (sv); \
485 } \
486 SE
487
488#define APPEND_CH(ch) SB \
489 APPEND_GROW (1); \
490 *cur++ = (ch); \
491 SE
492
493static SV * 964static SV *
494decode_str (dec_t *dec) 965decode_str (dec_t *dec)
495{ 966{
496 SV *sv = NEWSV (0,2); 967 SV *sv = 0;
497 int utf8 = 0; 968 int utf8 = 0;
498 char *cur = SvPVX (sv); 969 char *dec_cur = dec->cur;
499 char *end = SvEND (sv);
500 970
501 for (;;) 971 do
502 { 972 {
503 unsigned char ch = *(unsigned char *)dec->cur; 973 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
974 char *cur = buf;
504 975
505 if (ch == '"') 976 do
506 break;
507 else if (ch == '\\')
508 { 977 {
509 switch (*++dec->cur) 978 unsigned char ch = *(unsigned char *)dec_cur++;
979
980 if (expect_false (ch == '"'))
981 {
982 --dec_cur;
983 break;
510 { 984 }
511 case '\\': 985 else if (expect_false (ch == '\\'))
512 case '/': 986 {
513 case '"': APPEND_CH (*dec->cur++); break; 987 switch (*dec_cur)
514
515 case 'b': APPEND_CH ('\010'); ++dec->cur; break;
516 case 't': APPEND_CH ('\011'); ++dec->cur; break;
517 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
518 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
519 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
520
521 case 'u':
522 { 988 {
523 UV lo, hi; 989 case '\\':
524 ++dec->cur; 990 case '/':
991 case '"': *cur++ = *dec_cur++; break;
525 992
526 hi = decode_4hex (dec); 993 case 'b': ++dec_cur; *cur++ = '\010'; break;
527 if (hi == (UV)-1) 994 case 't': ++dec_cur; *cur++ = '\011'; break;
528 goto fail; 995 case 'n': ++dec_cur; *cur++ = '\012'; break;
996 case 'f': ++dec_cur; *cur++ = '\014'; break;
997 case 'r': ++dec_cur; *cur++ = '\015'; break;
529 998
530 // possibly a surrogate pair 999 case 'u':
531 if (hi >= 0xd800 && hi < 0xdc00)
532 { 1000 {
533 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 1001 UV lo, hi;
534 ERR ("illegal surrogate character"); 1002 ++dec_cur;
535 1003
536 dec->cur += 2; 1004 dec->cur = dec_cur;
537
538 lo = decode_4hex (dec); 1005 hi = decode_4hex (dec);
1006 dec_cur = dec->cur;
539 if (lo == (UV)-1) 1007 if (hi == (UV)-1)
540 goto fail; 1008 goto fail;
541 1009
1010 // possibly a surrogate pair
1011 if (hi >= 0xd800)
1012 if (hi < 0xdc00)
1013 {
1014 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
1015 ERR ("missing low surrogate character in surrogate pair");
1016
1017 dec_cur += 2;
1018
1019 dec->cur = dec_cur;
1020 lo = decode_4hex (dec);
1021 dec_cur = dec->cur;
1022 if (lo == (UV)-1)
1023 goto fail;
1024
542 if (lo < 0xdc00 || lo >= 0xe000) 1025 if (lo < 0xdc00 || lo >= 0xe000)
543 ERR ("surrogate pair expected"); 1026 ERR ("surrogate pair expected");
544 1027
545 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 1028 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
1029 }
1030 else if (hi < 0xe000)
1031 ERR ("missing high surrogate character in surrogate pair");
1032
1033 if (hi >= 0x80)
1034 {
1035 utf8 = 1;
1036
1037 cur = encode_utf8 (cur, hi);
1038 }
1039 else
1040 *cur++ = hi;
546 } 1041 }
547 else if (lo >= 0xdc00 && lo < 0xe000)
548 ERR ("illegal surrogate character");
549
550 if (hi >= 0x80)
551 { 1042 break;
552 utf8 = 1;
553 1043
554 APPEND_GROW (4); // at most 4 bytes for 21 bits
555 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
556 }
557 else 1044 default:
558 APPEND_CH (hi); 1045 --dec_cur;
1046 ERR ("illegal backslash escape sequence in string");
559 } 1047 }
1048 }
1049 else if (expect_true (ch >= 0x20 && ch < 0x80))
1050 *cur++ = ch;
1051 else if (ch >= 0x80)
1052 {
1053 STRLEN clen;
1054
1055 --dec_cur;
1056
1057 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
1058 if (clen == (STRLEN)-1)
1059 ERR ("malformed UTF-8 character in JSON string");
1060
1061 do
1062 *cur++ = *dec_cur++;
1063 while (--clen);
1064
1065 utf8 = 1;
1066 }
1067 else
1068 {
1069 --dec_cur;
1070
1071 if (!ch)
1072 ERR ("unexpected end of string while parsing JSON string");
560 break; 1073 else
1074 ERR ("invalid character encountered while parsing JSON string");
561 } 1075 }
562 } 1076 }
563 else if (ch >= 0x20 && ch <= 0x7f) 1077 while (cur < buf + SHORT_STRING_LEN);
564 APPEND_CH (*dec->cur++); 1078
565 else if (ch >= 0x80) 1079 {
1080 STRLEN len = cur - buf;
1081
1082 if (sv)
566 { 1083 {
567 STRLEN clen; 1084 STRLEN cur = SvCUR (sv);
568 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
569 if (clen < 0)
570 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
571 1085
572 APPEND_GROW (clen); 1086 if (SvLEN (sv) <= cur + len)
573 memcpy (cur, dec->cur, clen); 1087 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
574 cur += clen; 1088
575 dec->cur += clen; 1089 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1090 SvCUR_set (sv, SvCUR (sv) + len);
576 } 1091 }
577 else 1092 else
578 ERR ("invalid character encountered"); 1093 sv = newSVpvn (buf, len);
579 } 1094 }
1095 }
1096 while (*dec_cur != '"');
580 1097
581 ++dec->cur; 1098 ++dec_cur;
582 1099
583 SvCUR_set (sv, cur - SvPVX (sv)); 1100 if (sv)
584 1101 {
585 SvPOK_only (sv); 1102 SvPOK_only (sv);
586 *SvEND (sv) = 0; 1103 *SvEND (sv) = 0;
587 1104
588 if (utf8) 1105 if (utf8)
589 SvUTF8_on (sv); 1106 SvUTF8_on (sv);
1107 }
1108 else
1109 sv = newSVpvn ("", 0);
590 1110
1111 dec->cur = dec_cur;
591 return sv; 1112 return sv;
592 1113
593fail: 1114fail:
594 SvREFCNT_dec (sv); 1115 dec->cur = dec_cur;
595 return 0; 1116 return 0;
596} 1117}
597 1118
598static SV * 1119static SV *
599decode_num (dec_t *dec) 1120decode_num (dec_t *dec)
… …
609 { 1130 {
610 ++dec->cur; 1131 ++dec->cur;
611 if (*dec->cur >= '0' && *dec->cur <= '9') 1132 if (*dec->cur >= '0' && *dec->cur <= '9')
612 ERR ("malformed number (leading zero must not be followed by another digit)"); 1133 ERR ("malformed number (leading zero must not be followed by another digit)");
613 } 1134 }
614 1135 else if (*dec->cur < '0' || *dec->cur > '9')
615 // int 1136 ERR ("malformed number (no digits after initial minus)");
1137 else
1138 do
1139 {
1140 ++dec->cur;
1141 }
616 while (*dec->cur >= '0' && *dec->cur <= '9') 1142 while (*dec->cur >= '0' && *dec->cur <= '9');
617 ++dec->cur;
618 1143
619 // [frac] 1144 // [frac]
620 if (*dec->cur == '.') 1145 if (*dec->cur == '.')
621 { 1146 {
622 is_nv = 1; 1147 ++dec->cur;
1148
1149 if (*dec->cur < '0' || *dec->cur > '9')
1150 ERR ("malformed number (no digits after decimal point)");
623 1151
624 do 1152 do
625 { 1153 {
626 ++dec->cur; 1154 ++dec->cur;
627 } 1155 }
628 while (*dec->cur >= '0' && *dec->cur <= '9'); 1156 while (*dec->cur >= '0' && *dec->cur <= '9');
1157
1158 is_nv = 1;
629 } 1159 }
630 1160
631 // [exp] 1161 // [exp]
632 if (*dec->cur == 'e' || *dec->cur == 'E') 1162 if (*dec->cur == 'e' || *dec->cur == 'E')
633 { 1163 {
634 is_nv = 1;
635
636 ++dec->cur; 1164 ++dec->cur;
1165
637 if (*dec->cur == '-' || *dec->cur == '+') 1166 if (*dec->cur == '-' || *dec->cur == '+')
638 ++dec->cur; 1167 ++dec->cur;
639 1168
1169 if (*dec->cur < '0' || *dec->cur > '9')
1170 ERR ("malformed number (no digits after exp sign)");
1171
1172 do
1173 {
1174 ++dec->cur;
1175 }
640 while (*dec->cur >= '0' && *dec->cur <= '9') 1176 while (*dec->cur >= '0' && *dec->cur <= '9');
641 ++dec->cur; 1177
1178 is_nv = 1;
642 } 1179 }
643 1180
644 if (!is_nv) 1181 if (!is_nv)
645 { 1182 {
646 UV uv; 1183 int len = dec->cur - start;
647 int numtype = grok_number (start, dec->cur - start, &uv); 1184
648 if (numtype & IS_NUMBER_IN_UV) 1185 // special case the rather common 1..5-digit-int case
649 if (numtype & IS_NUMBER_NEG) 1186 if (*start == '-')
1187 switch (len)
650 { 1188 {
651 if (uv < (UV)IV_MIN) 1189 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
652 return newSViv (-(IV)uv); 1190 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1191 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1192 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1193 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
653 } 1194 }
1195 else
1196 switch (len)
1197 {
1198 case 1: return newSViv ( start [0] - '0' * 1);
1199 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1200 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1201 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1202 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1203 }
1204
1205 {
1206 UV uv;
1207 int numtype = grok_number (start, len, &uv);
1208 if (numtype & IS_NUMBER_IN_UV)
1209 if (numtype & IS_NUMBER_NEG)
1210 {
1211 if (uv < (UV)IV_MIN)
1212 return newSViv (-(IV)uv);
1213 }
654 else 1214 else
655 return newSVuv (uv); 1215 return newSVuv (uv);
656 } 1216 }
657 1217
1218 len -= *start == '-' ? 1 : 0;
1219
1220 // does not fit into IV or UV, try NV
1221 if (len <= NV_DIG)
1222 // fits into NV without loss of precision
1223 return newSVnv (json_atof (start));
1224
1225 // everything else fails, convert it to a string
1226 return newSVpvn (start, dec->cur - start);
1227 }
1228
1229 // loss of precision here
658 return newSVnv (Atof (start)); 1230 return newSVnv (json_atof (start));
659 1231
660fail: 1232fail:
661 return 0; 1233 return 0;
662} 1234}
663 1235
664static SV * 1236static SV *
665decode_av (dec_t *dec) 1237decode_av (dec_t *dec)
666{ 1238{
667 AV *av = newAV (); 1239 AV *av = newAV ();
668 1240
1241 DEC_INC_DEPTH;
1242 decode_ws (dec);
1243
1244 if (*dec->cur == ']')
1245 ++dec->cur;
1246 else
669 for (;;) 1247 for (;;)
670 { 1248 {
671 SV *value; 1249 SV *value;
672 1250
673 value = decode_sv (dec); 1251 value = decode_sv (dec);
674 if (!value) 1252 if (!value)
675 goto fail; 1253 goto fail;
676 1254
677 av_push (av, value); 1255 av_push (av, value);
678 1256
679 WS; 1257 decode_ws (dec);
680 1258
681 if (*dec->cur == ']') 1259 if (*dec->cur == ']')
682 { 1260 {
683 ++dec->cur; 1261 ++dec->cur;
684 break; 1262 break;
1263 }
685 } 1264
686
687 if (*dec->cur != ',') 1265 if (*dec->cur != ',')
688 ERR (", or ] expected while parsing array"); 1266 ERR (", or ] expected while parsing array");
689 1267
690 ++dec->cur; 1268 ++dec->cur;
1269
1270 decode_ws (dec);
1271
1272 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1273 {
1274 ++dec->cur;
1275 break;
1276 }
691 } 1277 }
692 1278
1279 DEC_DEC_DEPTH;
693 return newRV_noinc ((SV *)av); 1280 return newRV_noinc ((SV *)av);
694 1281
695fail: 1282fail:
696 SvREFCNT_dec (av); 1283 SvREFCNT_dec (av);
1284 DEC_DEC_DEPTH;
697 return 0; 1285 return 0;
698} 1286}
699 1287
700static SV * 1288static SV *
701decode_hv (dec_t *dec) 1289decode_hv (dec_t *dec)
702{ 1290{
1291 SV *sv;
703 HV *hv = newHV (); 1292 HV *hv = newHV ();
704 1293
1294 DEC_INC_DEPTH;
1295 decode_ws (dec);
1296
1297 if (*dec->cur == '}')
1298 ++dec->cur;
1299 else
705 for (;;) 1300 for (;;)
706 { 1301 {
707 SV *key, *value;
708
709 WS; EXPECT_CH ('"'); 1302 EXPECT_CH ('"');
710 1303
711 key = decode_str (dec); 1304 // heuristic: assume that
712 if (!key) 1305 // a) decode_str + hv_store_ent are abysmally slow.
713 goto fail; 1306 // b) most hash keys are short, simple ascii text.
714 1307 // => try to "fast-match" such strings to avoid
715 WS; EXPECT_CH (':'); 1308 // the overhead of decode_str + hv_store_ent.
716
717 value = decode_sv (dec);
718 if (!value)
719 { 1309 {
1310 SV *value;
1311 char *p = dec->cur;
1312 char *e = p + 24; // only try up to 24 bytes
1313
1314 for (;;)
1315 {
1316 // the >= 0x80 is false on most architectures
1317 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1318 {
1319 // slow path, back up and use decode_str
1320 SV *key = decode_str (dec);
1321 if (!key)
1322 goto fail;
1323
1324 decode_ws (dec); EXPECT_CH (':');
1325
1326 decode_ws (dec);
1327 value = decode_sv (dec);
1328 if (!value)
1329 {
1330 SvREFCNT_dec (key);
1331 goto fail;
1332 }
1333
1334 hv_store_ent (hv, key, value, 0);
720 SvREFCNT_dec (key); 1335 SvREFCNT_dec (key);
1336
1337 break;
1338 }
1339 else if (*p == '"')
1340 {
1341 // fast path, got a simple key
1342 char *key = dec->cur;
1343 int len = p - key;
1344 dec->cur = p + 1;
1345
1346 decode_ws (dec); EXPECT_CH (':');
1347
1348 decode_ws (dec);
1349 value = decode_sv (dec);
1350 if (!value)
721 goto fail; 1351 goto fail;
1352
1353 hv_store (hv, key, len, value, 0);
1354
1355 break;
1356 }
1357
1358 ++p;
1359 }
722 } 1360 }
723 1361
724 //TODO: optimise 1362 decode_ws (dec);
725 hv_store_ent (hv, key, value, 0);
726 1363
727 WS;
728
729 if (*dec->cur == '}') 1364 if (*dec->cur == '}')
1365 {
1366 ++dec->cur;
1367 break;
1368 }
1369
1370 if (*dec->cur != ',')
1371 ERR (", or } expected while parsing object/hash");
1372
1373 ++dec->cur;
1374
1375 decode_ws (dec);
1376
1377 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1378 {
1379 ++dec->cur;
1380 break;
1381 }
1382 }
1383
1384 DEC_DEC_DEPTH;
1385 sv = newRV_noinc ((SV *)hv);
1386
1387 // check filter callbacks
1388 if (dec->json.flags & F_HOOK)
1389 {
1390 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
730 { 1391 {
731 ++dec->cur; 1392 HE *cb, *he;
732 break; 1393
1394 hv_iterinit (hv);
1395 he = hv_iternext (hv);
1396 hv_iterinit (hv);
1397
1398 // the next line creates a mortal sv each time its called.
1399 // might want to optimise this for common cases.
1400 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1401
1402 if (cb)
1403 {
1404 dSP;
1405 int count;
1406
1407 ENTER; SAVETMPS; PUSHMARK (SP);
1408 XPUSHs (HeVAL (he));
1409 sv_2mortal (sv);
1410
1411 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1412
1413 if (count == 1)
1414 {
1415 sv = newSVsv (POPs);
1416 FREETMPS; LEAVE;
1417 return sv;
1418 }
1419
1420 SvREFCNT_inc (sv);
1421 FREETMPS; LEAVE;
1422 }
733 } 1423 }
734 1424
735 if (*dec->cur != ',') 1425 if (dec->json.cb_object)
736 ERR (", or } expected while parsing object/hash"); 1426 {
1427 dSP;
1428 int count;
737 1429
738 ++dec->cur; 1430 ENTER; SAVETMPS; PUSHMARK (SP);
739 } 1431 XPUSHs (sv_2mortal (sv));
740 1432
741 return newRV_noinc ((SV *)hv); 1433 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1434
1435 if (count == 1)
1436 {
1437 sv = newSVsv (POPs);
1438 FREETMPS; LEAVE;
1439 return sv;
1440 }
1441
1442 SvREFCNT_inc (sv);
1443 FREETMPS; LEAVE;
1444 }
1445 }
1446
1447 return sv;
742 1448
743fail: 1449fail:
744 SvREFCNT_dec (hv); 1450 SvREFCNT_dec (hv);
1451 DEC_DEC_DEPTH;
745 return 0; 1452 return 0;
746} 1453}
747 1454
748static SV * 1455static SV *
749decode_sv (dec_t *dec) 1456decode_sv (dec_t *dec)
750{ 1457{
751 WS; 1458 // the beauty of JSON: you need exactly one character lookahead
1459 // to parse everything.
752 switch (*dec->cur) 1460 switch (*dec->cur)
753 { 1461 {
754 case '"': ++dec->cur; return decode_str (dec); 1462 case '"': ++dec->cur; return decode_str (dec);
755 case '[': ++dec->cur; return decode_av (dec); 1463 case '[': ++dec->cur; return decode_av (dec);
756 case '{': ++dec->cur; return decode_hv (dec); 1464 case '{': ++dec->cur; return decode_hv (dec);
757 1465
758 case '-': 1466 case '-':
759 case '0': case '1': case '2': case '3': case '4': 1467 case '0': case '1': case '2': case '3': case '4':
760 case '5': case '6': case '7': case '8': case '9': 1468 case '5': case '6': case '7': case '8': case '9':
761 return decode_num (dec); 1469 return decode_num (dec);
762 1470
763 case 't': 1471 case 't':
764 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1472 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
765 { 1473 {
766 dec->cur += 4; 1474 dec->cur += 4;
1475#if JSON_SLOW
1476 json_true = get_bool ("JSON::XS::true");
1477#endif
767 return newSViv (1); 1478 return newSVsv (json_true);
768 } 1479 }
769 else 1480 else
770 ERR ("'true' expected"); 1481 ERR ("'true' expected");
771 1482
772 break; 1483 break;
773 1484
774 case 'f': 1485 case 'f':
775 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1486 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
776 { 1487 {
777 dec->cur += 5; 1488 dec->cur += 5;
1489#if JSON_SLOW
1490 json_false = get_bool ("JSON::XS::false");
1491#endif
778 return newSViv (0); 1492 return newSVsv (json_false);
779 } 1493 }
780 else 1494 else
781 ERR ("'false' expected"); 1495 ERR ("'false' expected");
782 1496
783 break; 1497 break;
784 1498
785 case 'n': 1499 case 'n':
786 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1500 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
787 { 1501 {
788 dec->cur += 4; 1502 dec->cur += 4;
789 return newSViv (1); 1503 return newSVsv (&PL_sv_undef);
790 } 1504 }
791 else 1505 else
792 ERR ("'null' expected"); 1506 ERR ("'null' expected");
793 1507
794 break; 1508 break;
795 1509
796 default: 1510 default:
797 ERR ("malformed json string"); 1511 ERR ("malformed JSON string, neither array, object, number, string or atom");
798 break; 1512 break;
799 } 1513 }
800 1514
801fail: 1515fail:
802 return 0; 1516 return 0;
803} 1517}
804 1518
805static SV * 1519static SV *
806decode_json (SV *string, UV flags) 1520decode_json (SV *string, JSON *json, char **offset_return)
807{ 1521{
1522 dec_t dec;
808 SV *sv; 1523 SV *sv;
809 1524
810 if (!(flags & F_UTF8)) 1525 /* work around bugs in 5.10 where manipulating magic values
1526 * will perl ignore the magic in subsequent accesses.
1527 * also make a copy of non-PV values, to get them into a clean
1528 * state (SvPV should do that, but it's buggy, see below).
1529 */
1530 /*SvGETMAGIC (string);*/
1531 if (SvMAGICAL (string) || !SvPOK (string))
1532 string = sv_2mortal (newSVsv (string));
1533
1534 SvUPGRADE (string, SVt_PV);
1535
1536 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1537 * assertion with -DDEBUGGING, although SvCUR is documented to
1538 * return the xpv_cur field which certainly exists after upgrading.
1539 * according to nicholas clark, calling SvPOK fixes this.
1540 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1541 * and hope for the best.
1542 * Damnit, SvPV_nolen still trips over yet another assertion. This
1543 * assertion business is seriously broken, try yet another workaround
1544 * for the broken -DDEBUGGING.
1545 */
1546 {
1547#ifdef DEBUGGING
1548 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1549#else
1550 STRLEN offset = SvCUR (string);
1551#endif
1552
1553 if (offset > json->max_size && json->max_size)
1554 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1555 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1556 }
1557
1558 if (DECODE_WANTS_OCTETS (json))
1559 sv_utf8_downgrade (string, 0);
1560 else
811 sv_utf8_upgrade (string); 1561 sv_utf8_upgrade (string);
812 1562
813 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1563 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
814 1564
815 dec_t dec; 1565 dec.json = *json;
816 dec.flags = flags;
817 dec.cur = SvPVX (string); 1566 dec.cur = SvPVX (string);
818 dec.end = SvEND (string); 1567 dec.end = SvEND (string);
819 dec.err = 0; 1568 dec.err = 0;
1569 dec.depth = 0;
820 1570
821 *dec.end = 1; // invalid anywhere 1571 if (dec.json.cb_object || dec.json.cb_sk_object)
1572 dec.json.flags |= F_HOOK;
1573
1574 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1575
1576 decode_ws (&dec);
822 sv = decode_sv (&dec); 1577 sv = decode_sv (&dec);
823 *dec.end = 0; 1578
1579 if (offset_return)
1580 *offset_return = dec.cur;
1581
1582 if (!(offset_return || !sv))
1583 {
1584 // check for trailing garbage
1585 decode_ws (&dec);
1586
1587 if (*dec.cur)
1588 {
1589 dec.err = "garbage after JSON object";
1590 SvREFCNT_dec (sv);
1591 sv = 0;
1592 }
1593 }
824 1594
825 if (!sv) 1595 if (!sv)
826 { 1596 {
827 IV offset = utf8_distance (dec.cur, SvPVX (string));
828 SV *uni = sv_newmortal (); 1597 SV *uni = sv_newmortal ();
829 1598
1599 // horrible hack to silence warning inside pv_uni_display
1600 COP cop = *PL_curcop;
1601 cop.cop_warnings = pWARN_NONE;
1602 ENTER;
1603 SAVEVPTR (PL_curcop);
1604 PL_curcop = &cop;
830 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1605 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1606 LEAVE;
1607
831 croak ("%s, at character %d (%s)", 1608 croak ("%s, at character offset %d (before \"%s\")",
832 dec.err, 1609 dec.err,
833 (int)offset, 1610 ptr_to_index (string, dec.cur),
834 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1611 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
835 } 1612 }
836 1613
837 sv = sv_2mortal (sv); 1614 sv = sv_2mortal (sv);
838 1615
839 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1616 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
840 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1617 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
841 1618
842 return sv; 1619 return sv;
843} 1620}
844 1621
1622/////////////////////////////////////////////////////////////////////////////
1623// incremental parser
1624
1625static void
1626incr_parse (JSON *self)
1627{
1628 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1629
1630 // the state machine here is a bit convoluted and could be simplified a lot
1631 // but this would make it slower, so...
1632
1633 for (;;)
1634 {
1635 //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
1636 switch (self->incr_mode)
1637 {
1638 // only used for initial whitespace skipping
1639 case INCR_M_WS:
1640 for (;;)
1641 {
1642 if (*p > 0x20)
1643 {
1644 if (*p == '#')
1645 {
1646 self->incr_mode = INCR_M_C0;
1647 goto incr_m_c;
1648 }
1649 else
1650 {
1651 self->incr_mode = INCR_M_JSON;
1652 goto incr_m_json;
1653 }
1654 }
1655 else if (!*p)
1656 goto interrupt;
1657
1658 ++p;
1659 }
1660
1661 // skip a single char inside a string (for \\-processing)
1662 case INCR_M_BS:
1663 if (!*p)
1664 goto interrupt;
1665
1666 ++p;
1667 self->incr_mode = INCR_M_STR;
1668 goto incr_m_str;
1669
1670 // inside #-style comments
1671 case INCR_M_C0:
1672 case INCR_M_C1:
1673 incr_m_c:
1674 for (;;)
1675 {
1676 if (*p == '\n')
1677 {
1678 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1679 break;
1680 }
1681 else if (!*p)
1682 goto interrupt;
1683
1684 ++p;
1685 }
1686
1687 break;
1688
1689 // inside a string
1690 case INCR_M_STR:
1691 incr_m_str:
1692 for (;;)
1693 {
1694 if (*p == '"')
1695 {
1696 ++p;
1697 self->incr_mode = INCR_M_JSON;
1698
1699 if (!self->incr_nest)
1700 goto interrupt;
1701
1702 goto incr_m_json;
1703 }
1704 else if (*p == '\\')
1705 {
1706 ++p; // "virtually" consumes character after \
1707
1708 if (!*p) // if at end of string we have to switch modes
1709 {
1710 self->incr_mode = INCR_M_BS;
1711 goto interrupt;
1712 }
1713 }
1714 else if (!*p)
1715 goto interrupt;
1716
1717 ++p;
1718 }
1719
1720 // after initial ws, outside string
1721 case INCR_M_JSON:
1722 incr_m_json:
1723 for (;;)
1724 {
1725 switch (*p++)
1726 {
1727 case 0:
1728 --p;
1729 goto interrupt;
1730
1731 case 0x09:
1732 case 0x0a:
1733 case 0x0d:
1734 case 0x20:
1735 if (!self->incr_nest)
1736 {
1737 --p; // do not eat the whitespace, let the next round do it
1738 goto interrupt;
1739 }
1740 break;
1741
1742 case '"':
1743 self->incr_mode = INCR_M_STR;
1744 goto incr_m_str;
1745
1746 case '[':
1747 case '{':
1748 if (++self->incr_nest > self->max_depth)
1749 croak (ERR_NESTING_EXCEEDED);
1750 break;
1751
1752 case ']':
1753 case '}':
1754 if (--self->incr_nest <= 0)
1755 goto interrupt;
1756 break;
1757
1758 case '#':
1759 self->incr_mode = INCR_M_C1;
1760 goto incr_m_c;
1761 }
1762 }
1763 }
1764
1765 modechange:
1766 ;
1767 }
1768
1769interrupt:
1770 self->incr_pos = p - SvPVX (self->incr_text);
1771 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1772 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1773}
1774
1775/////////////////////////////////////////////////////////////////////////////
1776// XS interface functions
1777
845MODULE = JSON::XS PACKAGE = JSON::XS 1778MODULE = JSON::XS PACKAGE = JSON::XS
846 1779
847BOOT: 1780BOOT:
848{ 1781{
849 int i; 1782 int i;
850 1783
851 memset (decode_hexdigit, 0xff, 256);
852 for (i = 10; i--; ) 1784 for (i = 0; i < 256; ++i)
853 decode_hexdigit ['0' + i] = i; 1785 decode_hexdigit [i] =
1786 i >= '0' && i <= '9' ? i - '0'
1787 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1788 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1789 : -1;
854 1790
855 for (i = 7; i--; )
856 {
857 decode_hexdigit ['a' + i] = 10 + i;
858 decode_hexdigit ['A' + i] = 10 + i;
859 }
860
861 json_stash = gv_stashpv ("JSON::XS", 1); 1791 json_stash = gv_stashpv ("JSON::XS" , 1);
1792 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1793
1794 json_true = get_bool ("JSON::XS::true");
1795 json_false = get_bool ("JSON::XS::false");
1796
1797 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
862} 1798}
863 1799
864PROTOTYPES: DISABLE 1800PROTOTYPES: DISABLE
865 1801
866SV *new (char *dummy) 1802void CLONE (...)
867 CODE: 1803 CODE:
868 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1804 json_stash = 0;
1805 json_boolean_stash = 0;
1806
1807void new (char *klass)
1808 PPCODE:
1809{
1810 SV *pv = NEWSV (0, sizeof (JSON));
1811 SvPOK_only (pv);
1812 json_init ((JSON *)SvPVX (pv));
1813 XPUSHs (sv_2mortal (sv_bless (
1814 newRV_noinc (pv),
1815 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1816 )));
1817}
1818
1819void ascii (JSON *self, int enable = 1)
1820 ALIAS:
1821 ascii = F_ASCII
1822 latin1 = F_LATIN1
1823 utf8 = F_UTF8
1824 indent = F_INDENT
1825 canonical = F_CANONICAL
1826 space_before = F_SPACE_BEFORE
1827 space_after = F_SPACE_AFTER
1828 pretty = F_PRETTY
1829 allow_nonref = F_ALLOW_NONREF
1830 shrink = F_SHRINK
1831 allow_blessed = F_ALLOW_BLESSED
1832 convert_blessed = F_CONV_BLESSED
1833 relaxed = F_RELAXED
1834 allow_unknown = F_ALLOW_UNKNOWN
1835 PPCODE:
1836{
1837 if (enable)
1838 self->flags |= ix;
1839 else
1840 self->flags &= ~ix;
1841
1842 XPUSHs (ST (0));
1843}
1844
1845void get_ascii (JSON *self)
1846 ALIAS:
1847 get_ascii = F_ASCII
1848 get_latin1 = F_LATIN1
1849 get_utf8 = F_UTF8
1850 get_indent = F_INDENT
1851 get_canonical = F_CANONICAL
1852 get_space_before = F_SPACE_BEFORE
1853 get_space_after = F_SPACE_AFTER
1854 get_allow_nonref = F_ALLOW_NONREF
1855 get_shrink = F_SHRINK
1856 get_allow_blessed = F_ALLOW_BLESSED
1857 get_convert_blessed = F_CONV_BLESSED
1858 get_relaxed = F_RELAXED
1859 get_allow_unknown = F_ALLOW_UNKNOWN
1860 PPCODE:
1861 XPUSHs (boolSV (self->flags & ix));
1862
1863void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1864 PPCODE:
1865 self->max_depth = max_depth;
1866 XPUSHs (ST (0));
1867
1868U32 get_max_depth (JSON *self)
1869 CODE:
1870 RETVAL = self->max_depth;
869 OUTPUT: 1871 OUTPUT:
870 RETVAL 1872 RETVAL
871 1873
872SV *ascii (SV *self, int enable) 1874void max_size (JSON *self, U32 max_size = 0)
873 ALIAS: 1875 PPCODE:
874 ascii = F_ASCII 1876 self->max_size = max_size;
875 utf8 = F_UTF8 1877 XPUSHs (ST (0));
876 indent = F_INDENT 1878
877 canonical = F_CANONICAL 1879int get_max_size (JSON *self)
878 space_before = F_SPACE_BEFORE
879 space_after = F_SPACE_AFTER
880 json_rpc = F_JSON_RPC
881 pretty = F_PRETTY
882 allow_nonref = F_ALLOW_NONREF
883 CODE: 1880 CODE:
884{ 1881 RETVAL = self->max_size;
885 UV *uv = SvJSON (self);
886 if (enable)
887 *uv |= ix;
888 else
889 *uv &= ~ix;
890
891 RETVAL = newSVsv (self);
892}
893 OUTPUT: 1882 OUTPUT:
894 RETVAL 1883 RETVAL
895 1884
896void encode (SV *self, SV *scalar) 1885void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
897 PPCODE: 1886 PPCODE:
898 XPUSHs (encode_json (scalar, *SvJSON (self))); 1887{
1888 SvREFCNT_dec (self->cb_object);
1889 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
899 1890
900void decode (SV *self, SV *jsonstr) 1891 XPUSHs (ST (0));
1892}
1893
1894void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
901 PPCODE: 1895 PPCODE:
1896{
1897 if (!self->cb_sk_object)
1898 self->cb_sk_object = newHV ();
1899
1900 if (SvOK (cb))
1901 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1902 else
1903 {
1904 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1905
1906 if (!HvKEYS (self->cb_sk_object))
1907 {
1908 SvREFCNT_dec (self->cb_sk_object);
1909 self->cb_sk_object = 0;
1910 }
1911 }
1912
1913 XPUSHs (ST (0));
1914}
1915
1916void encode (JSON *self, SV *scalar)
1917 PPCODE:
1918 XPUSHs (encode_json (scalar, self));
1919
1920void decode (JSON *self, SV *jsonstr)
1921 PPCODE:
902 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1922 XPUSHs (decode_json (jsonstr, self, 0));
1923
1924void decode_prefix (JSON *self, SV *jsonstr)
1925 PPCODE:
1926{
1927 char *offset;
1928 EXTEND (SP, 2);
1929 PUSHs (decode_json (jsonstr, self, &offset));
1930 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1931}
1932
1933void incr_parse (JSON *self, SV *jsonstr = 0)
1934 PPCODE:
1935{
1936 if (!self->incr_text)
1937 self->incr_text = newSVpvn ("", 0);
1938
1939 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1940 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1941 if (DECODE_WANTS_OCTETS (self))
1942 {
1943 if (self->incr_pos)
1944 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1945 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1946
1947 sv_utf8_downgrade (self->incr_text, 0);
1948 }
1949 else
1950 {
1951 sv_utf8_upgrade (self->incr_text);
1952
1953 if (self->incr_pos)
1954 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1955 - (U8 *)SvPVX (self->incr_text);
1956 }
1957
1958 // append data, if any
1959 if (jsonstr)
1960 {
1961 /* make sure both strings have same encoding */
1962 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1963 if (SvUTF8 (jsonstr))
1964 sv_utf8_downgrade (jsonstr, 0);
1965 else
1966 sv_utf8_upgrade (jsonstr);
1967
1968 /* and then just blindly append */
1969 {
1970 STRLEN len;
1971 const char *str = SvPV (jsonstr, len);
1972 STRLEN cur = SvCUR (self->incr_text);
1973
1974 if (SvLEN (self->incr_text) <= cur + len)
1975 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1976
1977 Move (str, SvEND (self->incr_text), len, char);
1978 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1979 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1980 }
1981 }
1982
1983 if (GIMME_V != G_VOID)
1984 do
1985 {
1986 char *offset;
1987
1988 if (!INCR_DONE (self))
1989 {
1990 incr_parse (self);
1991
1992 if (self->incr_pos > self->max_size && self->max_size)
1993 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1994 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1995
1996 if (!INCR_DONE (self))
1997 break;
1998 }
1999
2000 XPUSHs (decode_json (self->incr_text, self, &offset));
2001
2002 self->incr_pos -= offset - SvPVX (self->incr_text);
2003 self->incr_nest = 0;
2004 self->incr_mode = 0;
2005
2006 sv_chop (self->incr_text, offset);
2007 }
2008 while (GIMME_V == G_ARRAY);
2009}
2010
2011SV *incr_text (JSON *self)
2012 ATTRS: lvalue
2013 CODE:
2014{
2015 if (self->incr_pos)
2016 croak ("incr_text can not be called when the incremental parser already started parsing");
2017
2018 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2019}
2020 OUTPUT:
2021 RETVAL
2022
2023void incr_skip (JSON *self)
2024 CODE:
2025{
2026 if (self->incr_pos)
2027 {
2028 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2029 self->incr_pos = 0;
2030 self->incr_nest = 0;
2031 self->incr_mode = 0;
2032 }
2033}
2034
2035void incr_reset (JSON *self)
2036 CODE:
2037{
2038 SvREFCNT_dec (self->incr_text);
2039 self->incr_text = 0;
2040 self->incr_pos = 0;
2041 self->incr_nest = 0;
2042 self->incr_mode = 0;
2043}
2044
2045void DESTROY (JSON *self)
2046 CODE:
2047 SvREFCNT_dec (self->cb_sk_object);
2048 SvREFCNT_dec (self->cb_object);
2049 SvREFCNT_dec (self->incr_text);
903 2050
904PROTOTYPES: ENABLE 2051PROTOTYPES: ENABLE
905 2052
906void to_json (SV *scalar) 2053void encode_json (SV *scalar)
2054 ALIAS:
2055 to_json_ = 0
2056 encode_json = F_UTF8
907 PPCODE: 2057 PPCODE:
2058{
2059 JSON json;
2060 json_init (&json);
2061 json.flags |= ix;
908 XPUSHs (encode_json (scalar, F_UTF8)); 2062 XPUSHs (encode_json (scalar, &json));
2063}
909 2064
910void from_json (SV *jsonstr) 2065void decode_json (SV *jsonstr)
2066 ALIAS:
2067 from_json_ = 0
2068 decode_json = F_UTF8
911 PPCODE: 2069 PPCODE:
2070{
2071 JSON json;
2072 json_init (&json);
2073 json.flags |= ix;
912 XPUSHs (decode_json (jsonstr, F_UTF8)); 2074 XPUSHs (decode_json (jsonstr, &json, 0));
2075}
913 2076

Diff Legend

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