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 1.94 by root, Thu Nov 20 03:59:53 2008 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// guarentees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
23
9#define F_ASCII 0x00000001 24#define F_ASCII 0x00000001UL
25#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 26#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 27#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 28#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 29#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 30#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 31#define F_ALLOW_NONREF 0x00000100UL
32#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL
36#define F_ALLOW_UNKNOWN 0x00002000UL
37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
17 38
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0
20 40
21#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
42#define INDENT_STEP 3 // spaces per indentation level
43
44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 45
23#define SB do { 46#define SB do {
24#define SE } while (0) 47#define SE } while (0)
25 48
26static HV *json_stash; 49#if __GNUC__ >= 3
50# define expect(expr,value) __builtin_expect ((expr), (value))
51# define INLINE static inline
52#else
53# define expect(expr,value) (expr)
54# define INLINE static
55#endif
56
57#define expect_false(expr) expect ((expr) != 0, 0)
58#define expect_true(expr) expect ((expr) != 0, 1)
59
60#define IN_RANGE_INC(type,val,beg,end) \
61 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
62 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
63
64#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
65
66#ifdef USE_ITHREADS
67# define JSON_SLOW 1
68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
69#else
70# define JSON_SLOW 0
71# define JSON_STASH json_stash
72#endif
73
74static HV *json_stash, *json_boolean_stash; // JSON::XS::
75static SV *json_true, *json_false;
76
77enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash
81 INCR_M_JSON // outside anything, count nesting
82};
83
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85
86typedef struct {
87 U32 flags;
88 U32 max_depth;
89 STRLEN max_size;
90
91 SV *cb_object;
92 HV *cb_sk_object;
93
94 // for the incremental parser
95 SV *incr_text; // the source text so far
96 STRLEN incr_pos; // the current offset into the text
97 int incr_nest; // {[]}-nesting level
98 unsigned char incr_mode;
99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
107
108/////////////////////////////////////////////////////////////////////////////
109// utility functions
110
111INLINE SV *
112get_bool (const char *name)
113{
114 SV *sv = get_sv (name, 1);
115
116 SvREADONLY_on (sv);
117 SvREADONLY_on (SvRV (sv));
118
119 return sv;
120}
121
122INLINE void
123shrink (SV *sv)
124{
125 sv_utf8_downgrade (sv, 1);
126
127 if (SvLEN (sv) > SvCUR (sv) + 1)
128 {
129#ifdef SvPV_shrink_to_cur
130 SvPV_shrink_to_cur (sv);
131#elif defined (SvPV_renew)
132 SvPV_renew (sv, SvCUR (sv) + 1);
133#endif
134 }
135}
136
137// decode an utf-8 character and return it, or (UV)-1 in
138// case of an error.
139// we special-case "safe" characters from U+80 .. U+7FF,
140// but use the very good perl function to parse anything else.
141// note that we never call this function for a ascii codepoints
142INLINE UV
143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
144{
145 if (expect_true (len >= 2
146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
148 {
149 *clen = 2;
150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
151 }
152 else
153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
154}
155
156// likewise for encoding, also never called for ascii codepoints
157// this function takes advantage of this fact, although current gccs
158// seem to optimise the check for >= 0x80 away anyways
159INLINE unsigned char *
160encode_utf8 (unsigned char *s, UV ch)
161{
162 if (expect_false (ch < 0x000080))
163 *s++ = ch;
164 else if (expect_true (ch < 0x000800))
165 *s++ = 0xc0 | ( ch >> 6),
166 *s++ = 0x80 | ( ch & 0x3f);
167 else if ( ch < 0x010000)
168 *s++ = 0xe0 | ( ch >> 12),
169 *s++ = 0x80 | ((ch >> 6) & 0x3f),
170 *s++ = 0x80 | ( ch & 0x3f);
171 else if ( ch < 0x110000)
172 *s++ = 0xf0 | ( ch >> 18),
173 *s++ = 0x80 | ((ch >> 12) & 0x3f),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176
177 return s;
178}
179
180/////////////////////////////////////////////////////////////////////////////
181// encoder
27 182
28// structure used for encoding JSON 183// structure used for encoding JSON
29typedef struct 184typedef struct
30{ 185{
31 char *cur; 186 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
34 SV *sv; 188 SV *sv; // result scalar
35 UV flags; 189 JSON json;
36 int max_recurse; 190 U32 indent; // indentation level
37 int indent; 191 UV limit; // escape character values >= this value when encoding
38} enc_t; 192} enc_t;
39 193
40// structure used for decoding JSON 194INLINE 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) 195need (enc_t *enc, STRLEN len)
62{ 196{
63 if (enc->cur + len >= enc->end) 197 if (expect_false (enc->cur + len >= enc->end))
64 { 198 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 200 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 203 }
70} 204}
71 205
72static void 206INLINE void
73encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
74{ 208{
75 need (enc, 1); 209 need (enc, 1);
76 *enc->cur++ = ch; 210 *enc->cur++ = ch;
77} 211}
… …
85 219
86 while (str < end) 220 while (str < end)
87 { 221 {
88 unsigned char ch = *(unsigned char *)str; 222 unsigned char ch = *(unsigned char *)str;
89 223
90 if (ch == '"') 224 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 225 {
226 if (expect_false (ch == '"')) // but with slow exceptions
227 {
92 need (enc, len += 1); 228 need (enc, len += 1);
93 *enc->cur++ = '\\'; 229 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 230 *enc->cur++ = '"';
95 ++str;
96 } 231 }
97 else if (ch == '\\') 232 else if (expect_false (ch == '\\'))
98 { 233 {
99 need (enc, len += 1); 234 need (enc, len += 1);
100 *enc->cur++ = '\\'; 235 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 236 *enc->cur++ = '\\';
102 ++str;
103 } 237 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 238 else
105 {
106 *enc->cur++ = ch; 239 *enc->cur++ = ch;
107 ++str; 240
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; 241 ++str;
122 } 242 }
123 else 243 else
124 { 244 {
125 STRLEN clen; 245 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 { 246 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 247 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
131 if (clen < 0) 248 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
132 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 249 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
133 } 250 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
134 else 251 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
135 {
136 uch = ch;
137 clen = 1;
138 }
139 252
140 if (uch < 0x80 || enc->flags & F_ASCII) 253 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 254 {
255 STRLEN clen;
256 UV uch;
257
258 if (is_utf8)
259 {
260 uch = decode_utf8 (str, end - str, &clen);
261 if (clen == (STRLEN)-1)
262 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
263 }
264 else
265 {
266 uch = ch;
267 clen = 1;
268 }
269
270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
271 {
272 if (uch >= 0x10000UL)
273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
144 need (enc, len += 11); 277 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 281 enc->cur += 12;
282 }
283 else
284 {
285 need (enc, len += 5);
286 *enc->cur++ = '\\';
287 *enc->cur++ = 'u';
288 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
289 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
290 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
291 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
292 }
293
294 str += clen;
295 }
296 else if (enc->json.flags & F_LATIN1)
297 {
298 *enc->cur++ = uch;
299 str += clen;
300 }
301 else if (is_utf8)
302 {
303 need (enc, len += clen);
304 do
305 {
306 *enc->cur++ = *str++;
307 }
308 while (--clen);
309 }
310 else
311 {
312 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
313 enc->cur = encode_utf8 (enc->cur, uch);
314 ++str;
315 }
149 } 316 }
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 } 317 }
176 } 318 }
177 319
178 --len; 320 --len;
179 } 321 }
180} 322}
181 323
182#define INDENT SB \ 324INLINE void
325encode_indent (enc_t *enc)
326{
183 if (enc->flags & F_INDENT) \ 327 if (enc->json.flags & F_INDENT)
184 { \ 328 {
185 int i_; \ 329 int spaces = enc->indent * INDENT_STEP;
186 need (enc, enc->indent); \ 330
187 for (i_ = enc->indent * 3; i_--; )\ 331 need (enc, spaces);
332 memset (enc->cur, ' ', spaces);
333 enc->cur += spaces;
334 }
335}
336
337INLINE void
338encode_space (enc_t *enc)
339{
340 need (enc, 1);
341 encode_ch (enc, ' ');
342}
343
344INLINE void
345encode_nl (enc_t *enc)
346{
347 if (enc->json.flags & F_INDENT)
348 {
349 need (enc, 1);
188 encode_ch (enc, ' '); \ 350 encode_ch (enc, '\n');
189 } \ 351 }
190 SE 352}
191 353
192#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 354INLINE void
193#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 355encode_comma (enc_t *enc)
194#define COMMA SB \ 356{
195 encode_ch (enc, ','); \ 357 encode_ch (enc, ',');
358
196 if (enc->flags & F_INDENT) \ 359 if (enc->json.flags & F_INDENT)
197 NL; \ 360 encode_nl (enc);
198 else if (enc->flags & F_SPACE_AFTER) \ 361 else if (enc->json.flags & F_SPACE_AFTER)
199 SPACE; \ 362 encode_space (enc);
200 SE 363}
201 364
202static void encode_sv (enc_t *enc, SV *sv); 365static void encode_sv (enc_t *enc, SV *sv);
203 366
204static void 367static void
205encode_av (enc_t *enc, AV *av) 368encode_av (enc_t *enc, AV *av)
206{ 369{
207 int i, len = av_len (av); 370 int i, len = av_len (av);
208 371
372 if (enc->indent >= enc->json.max_depth)
373 croak (ERR_NESTING_EXCEEDED);
374
209 encode_ch (enc, '['); NL; 375 encode_ch (enc, '[');
210 ++enc->indent; 376
377 if (len >= 0)
378 {
379 encode_nl (enc); ++enc->indent;
211 380
212 for (i = 0; i <= len; ++i) 381 for (i = 0; i <= len; ++i)
213 { 382 {
214 INDENT; 383 SV **svp = av_fetch (av, i, 0);
215 encode_sv (enc, *av_fetch (av, i, 0));
216 384
385 encode_indent (enc);
386
387 if (svp)
388 encode_sv (enc, *svp);
389 else
390 encode_str (enc, "null", 4, 0);
391
217 if (i < len) 392 if (i < len)
218 COMMA; 393 encode_comma (enc);
219 } 394 }
220 395
221 NL; 396 encode_nl (enc); --enc->indent; encode_indent (enc);
222 397 }
223 --enc->indent; 398
224 INDENT; encode_ch (enc, ']'); 399 encode_ch (enc, ']');
225} 400}
226 401
227static void 402static void
228encode_he (enc_t *enc, HE *he) 403encode_hk (enc_t *enc, HE *he)
229{ 404{
230 encode_ch (enc, '"'); 405 encode_ch (enc, '"');
231 406
232 if (HeKLEN (he) == HEf_SVKEY) 407 if (HeKLEN (he) == HEf_SVKEY)
233 { 408 {
… …
243 else 418 else
244 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 419 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
245 420
246 encode_ch (enc, '"'); 421 encode_ch (enc, '"');
247 422
248 if (enc->flags & F_SPACE_BEFORE) SPACE; 423 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
249 encode_ch (enc, ':'); 424 encode_ch (enc, ':');
250 if (enc->flags & F_SPACE_AFTER ) SPACE; 425 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
251 encode_sv (enc, HeVAL (he));
252} 426}
253 427
254// compare hash entries, used when all keys are bytestrings 428// compare hash entries, used when all keys are bytestrings
255static int 429static int
256he_cmp_fast (const void *a_, const void *b_) 430he_cmp_fast (const void *a_, const void *b_)
… …
261 HE *b = *(HE **)b_; 435 HE *b = *(HE **)b_;
262 436
263 STRLEN la = HeKLEN (a); 437 STRLEN la = HeKLEN (a);
264 STRLEN lb = HeKLEN (b); 438 STRLEN lb = HeKLEN (b);
265 439
266 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 440 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
267 cmp = la < lb ? -1 : la == lb ? 0 : 1; 441 cmp = lb - la;
268 442
269 return cmp; 443 return cmp;
270} 444}
271 445
272// compare hash entries, used when some keys are sv's or utf-x 446// compare hash entries, used when some keys are sv's or utf-x
273static int 447static int
274he_cmp_slow (const void *a, const void *b) 448he_cmp_slow (const void *a, const void *b)
275{ 449{
276 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 450 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
277} 451}
278 452
279static void 453static void
280encode_hv (enc_t *enc, HV *hv) 454encode_hv (enc_t *enc, HV *hv)
281{ 455{
282 int count, i; 456 HE *he;
283 457
284 encode_ch (enc, '{'); NL; ++enc->indent; 458 if (enc->indent >= enc->json.max_depth)
459 croak (ERR_NESTING_EXCEEDED);
285 460
286 if ((count = hv_iterinit (hv))) 461 encode_ch (enc, '{');
287 { 462
288 // for canonical output we have to sort by keys first 463 // for canonical output we have to sort by keys first
289 // actually, this is mostly due to the stupid so-called 464 // actually, this is mostly due to the stupid so-called
290 // security workaround added somewhere in 5.8.x. 465 // security workaround added somewhere in 5.8.x
291 // that randomises hash orderings 466 // that randomises hash orderings
292 if (enc->flags & F_CANONICAL) 467 if (enc->json.flags & F_CANONICAL)
468 {
469 int count = hv_iterinit (hv);
470
471 if (SvMAGICAL (hv))
293 { 472 {
294 HE *he, *hes [count]; 473 // need to count by iterating. could improve by dynamically building the vector below
474 // but I don't care for the speed of this special case.
475 // note also that we will run into undefined behaviour when the two iterations
476 // do not result in the same count, something I might care for in some later release.
477
478 count = 0;
479 while (hv_iternext (hv))
480 ++count;
481
482 hv_iterinit (hv);
483 }
484
485 if (count)
486 {
295 int fast = 1; 487 int i, fast = 1;
488#if defined(__BORLANDC__) || defined(_MSC_VER)
489 HE **hes = _alloca (count * sizeof (HE));
490#else
491 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
492#endif
296 493
297 i = 0; 494 i = 0;
298 while ((he = hv_iternext (hv))) 495 while ((he = hv_iternext (hv)))
299 { 496 {
300 hes [i++] = he; 497 hes [i++] = he;
… …
306 503
307 if (fast) 504 if (fast)
308 qsort (hes, count, sizeof (HE *), he_cmp_fast); 505 qsort (hes, count, sizeof (HE *), he_cmp_fast);
309 else 506 else
310 { 507 {
311 // hack to disable "use bytes" 508 // hack to forcefully disable "use bytes"
312 COP *oldcop = PL_curcop, cop; 509 COP cop = *PL_curcop;
313 cop.op_private = 0; 510 cop.op_private = 0;
511
512 ENTER;
513 SAVETMPS;
514
515 SAVEVPTR (PL_curcop);
314 PL_curcop = &cop; 516 PL_curcop = &cop;
315 517
316 SAVETMPS;
317 qsort (hes, count, sizeof (HE *), he_cmp_slow); 518 qsort (hes, count, sizeof (HE *), he_cmp_slow);
519
318 FREETMPS; 520 FREETMPS;
319 521 LEAVE;
320 PL_curcop = oldcop;
321 } 522 }
322 523
323 for (i = 0; i < count; ++i) 524 encode_nl (enc); ++enc->indent;
525
526 while (count--)
324 { 527 {
325 INDENT; 528 encode_indent (enc);
529 he = hes [count];
326 encode_he (enc, hes [i]); 530 encode_hk (enc, he);
531 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
327 532
328 if (i < count - 1) 533 if (count)
329 COMMA; 534 encode_comma (enc);
330 } 535 }
331 536
537 encode_nl (enc); --enc->indent; encode_indent (enc);
538 }
539 }
540 else
541 {
542 if (hv_iterinit (hv) || SvMAGICAL (hv))
543 if ((he = hv_iternext (hv)))
332 NL; 544 {
545 encode_nl (enc); ++enc->indent;
546
547 for (;;)
548 {
549 encode_indent (enc);
550 encode_hk (enc, he);
551 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
552
553 if (!(he = hv_iternext (hv)))
554 break;
555
556 encode_comma (enc);
557 }
558
559 encode_nl (enc); --enc->indent; encode_indent (enc);
560 }
561 }
562
563 encode_ch (enc, '}');
564}
565
566// encode objects, arrays and special \0=false and \1=true values.
567static void
568encode_rv (enc_t *enc, SV *sv)
569{
570 svtype svt;
571
572 SvGETMAGIC (sv);
573 svt = SvTYPE (sv);
574
575 if (expect_false (SvOBJECT (sv)))
576 {
577 HV *stash = !JSON_SLOW || json_boolean_stash
578 ? json_boolean_stash
579 : gv_stashpv ("JSON::XS::Boolean", 1);
580
581 if (SvSTASH (sv) == stash)
582 {
583 if (SvIV (sv))
584 encode_str (enc, "true", 4, 0);
585 else
586 encode_str (enc, "false", 5, 0);
333 } 587 }
334 else 588 else
335 { 589 {
336 SV *sv; 590#if 0
337 HE *he = hv_iternext (hv); 591 if (0 && sv_derived_from (rv, "JSON::Literal"))
338
339 for (;;)
340 { 592 {
341 INDENT; 593 // not yet
342 encode_he (enc, he);
343
344 if (!(he = hv_iternext (hv)))
345 break;
346
347 COMMA;
348 } 594 }
349 595#endif
596 if (enc->json.flags & F_CONV_BLESSED)
350 NL; 597 {
598 // we re-bless the reference to get overload and other niceties right
599 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
600
601 if (to_json)
602 {
603 dSP;
604
605 ENTER; SAVETMPS; PUSHMARK (SP);
606 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
607
608 // calling with G_SCALAR ensures that we always get a 1 return value
609 PUTBACK;
610 call_sv ((SV *)GvCV (to_json), G_SCALAR);
611 SPAGAIN;
612
613 // catch this surprisingly common error
614 if (SvROK (TOPs) && SvRV (TOPs) == sv)
615 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
616
617 sv = POPs;
618 PUTBACK;
619
620 encode_sv (enc, sv);
621
622 FREETMPS; LEAVE;
623 }
624 else if (enc->json.flags & F_ALLOW_BLESSED)
625 encode_str (enc, "null", 4, 0);
626 else
627 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
628 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
629 }
630 else if (enc->json.flags & F_ALLOW_BLESSED)
631 encode_str (enc, "null", 4, 0);
632 else
633 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
634 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
351 } 635 }
352 } 636 }
637 else if (svt == SVt_PVHV)
638 encode_hv (enc, (HV *)sv);
639 else if (svt == SVt_PVAV)
640 encode_av (enc, (AV *)sv);
641 else if (svt < SVt_PVAV)
642 {
643 STRLEN len = 0;
644 char *pv = svt ? SvPV (sv, len) : 0;
353 645
354 --enc->indent; INDENT; encode_ch (enc, '}'); 646 if (len == 1 && *pv == '1')
647 encode_str (enc, "true", 4, 0);
648 else if (len == 1 && *pv == '0')
649 encode_str (enc, "false", 5, 0);
650 else if (enc->json.flags & F_ALLOW_UNKNOWN)
651 encode_str (enc, "null", 4, 0);
652 else
653 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
654 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
655 }
656 else if (enc->json.flags & F_ALLOW_UNKNOWN)
657 encode_str (enc, "null", 4, 0);
658 else
659 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
355} 661}
356 662
357static void 663static void
358encode_sv (enc_t *enc, SV *sv) 664encode_sv (enc_t *enc, SV *sv)
359{ 665{
… …
367 encode_str (enc, str, len, SvUTF8 (sv)); 673 encode_str (enc, str, len, SvUTF8 (sv));
368 encode_ch (enc, '"'); 674 encode_ch (enc, '"');
369 } 675 }
370 else if (SvNOKp (sv)) 676 else if (SvNOKp (sv))
371 { 677 {
678 // trust that perl will do the right thing w.r.t. JSON syntax.
372 need (enc, NV_DIG + 32); 679 need (enc, NV_DIG + 32);
373 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 680 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
374 enc->cur += strlen (enc->cur); 681 enc->cur += strlen (enc->cur);
375 } 682 }
376 else if (SvIOKp (sv)) 683 else if (SvIOKp (sv))
377 { 684 {
685 // we assume we can always read an IV as a UV and vice versa
686 // we assume two's complement
687 // we assume no aliasing issues in the union
688 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
689 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
690 {
691 // optimise the "small number case"
692 // code will likely be branchless and use only a single multiplication
693 // works for numbers up to 59074
694 I32 i = SvIVX (sv);
695 U32 u;
696 char digit, nz = 0;
697
378 need (enc, 64); 698 need (enc, 6);
699
700 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
701 u = i < 0 ? -i : i;
702
703 // convert to 4.28 fixed-point representation
704 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
705
706 // now output digit by digit, each time masking out the integer part
707 // and multiplying by 5 while moving the decimal point one to the right,
708 // resulting in a net multiplication by 10.
709 // we always write the digit to memory but conditionally increment
710 // the pointer, to enable the use of conditional move instructions.
711 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
712 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
713 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
714 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
715 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
716 }
717 else
718 {
719 // large integer, use the (rather slow) snprintf way.
720 need (enc, IVUV_MAXCHARS);
379 enc->cur += 721 enc->cur +=
380 SvIsUV(sv) 722 SvIsUV(sv)
381 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 723 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
382 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 724 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
725 }
383 } 726 }
384 else if (SvROK (sv)) 727 else if (SvROK (sv))
385 { 728 encode_rv (enc, SvRV (sv));
386 if (!--enc->max_recurse) 729 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); 730 encode_str (enc, "null", 4, 0);
402 else 731 else
403 croak ("encountered perl type that JSON cannot handle"); 732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
733 SvPV_nolen (sv), SvFLAGS (sv));
404} 734}
405 735
406static SV * 736static SV *
407encode_json (SV *scalar, UV flags) 737encode_json (SV *scalar, JSON *json)
408{ 738{
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; 739 enc_t enc;
413 enc.flags = flags; 740
741 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
742 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
743
744 enc.json = *json;
414 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 745 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
415 enc.cur = SvPVX (enc.sv); 746 enc.cur = SvPVX (enc.sv);
416 enc.end = SvEND (enc.sv); 747 enc.end = SvEND (enc.sv);
417 enc.max_recurse = 0;
418 enc.indent = 0; 748 enc.indent = 0;
749 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
750 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL;
419 752
420 SvPOK_only (enc.sv); 753 SvPOK_only (enc.sv);
421 encode_sv (&enc, scalar); 754 encode_sv (&enc, scalar);
422 755
756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
758
423 if (!(flags & (F_ASCII | F_UTF8))) 759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
424 SvUTF8_on (enc.sv); 760 SvUTF8_on (enc.sv);
425 761
426 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 762 if (enc.json.flags & F_SHRINK)
763 shrink (enc.sv);
764
427 return enc.sv; 765 return enc.sv;
428} 766}
429 767
430///////////////////////////////////////////////////////////////////////////// 768/////////////////////////////////////////////////////////////////////////////
769// decoder
431 770
432#define WS \ 771// structure used for decoding JSON
772typedef struct
773{
774 char *cur; // current parser pointer
775 char *end; // end of input string
776 const char *err; // parse error, if != 0
777 JSON json;
778 U32 depth; // recursion depth
779 U32 maxdepth; // recursion depth limit
780} dec_t;
781
782INLINE void
783decode_comment (dec_t *dec)
784{
785 // only '#'-style comments allowed a.t.m.
786
787 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
788 ++dec->cur;
789}
790
791INLINE void
792decode_ws (dec_t *dec)
793{
433 for (;;) \ 794 for (;;)
434 { \ 795 {
435 char ch = *dec->cur; \ 796 char ch = *dec->cur;
797
436 if (ch > 0x20 \ 798 if (ch > 0x20)
799 {
800 if (expect_false (ch == '#'))
801 {
802 if (dec->json.flags & F_RELAXED)
803 decode_comment (dec);
804 else
805 break;
806 }
807 else
808 break;
809 }
437 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 810 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
438 break; \ 811 break; // parse error, but let higher level handle it, gives better error messages
812
439 ++dec->cur; \ 813 ++dec->cur;
440 } 814 }
815}
441 816
442#define ERR(reason) SB dec->err = reason; goto fail; SE 817#define ERR(reason) SB dec->err = reason; goto fail; SE
818
443#define EXPECT_CH(ch) SB \ 819#define EXPECT_CH(ch) SB \
444 if (*dec->cur != ch) \ 820 if (*dec->cur != ch) \
445 ERR (# ch " expected"); \ 821 ERR (# ch " expected"); \
446 ++dec->cur; \ 822 ++dec->cur; \
447 SE 823 SE
448 824
825#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
826#define DEC_DEC_DEPTH --dec->depth
827
449static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
450 829
451static signed char decode_hexdigit[256]; 830static signed char decode_hexdigit[256];
452 831
453static UV 832static UV
454decode_4hex (dec_t *dec) 833decode_4hex (dec_t *dec)
455{ 834{
456 signed char d1, d2, d3, d4; 835 signed char d1, d2, d3, d4;
836 unsigned char *cur = (unsigned char *)dec->cur;
457 837
458 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 838 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"); 839 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]]; 840 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"); 841 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");
466 842
467 dec->cur += 4; 843 dec->cur += 4;
468 844
469 return ((UV)d1) << 12 845 return ((UV)d1) << 12
470 | ((UV)d2) << 8 846 | ((UV)d2) << 8
… …
473 849
474fail: 850fail:
475 return (UV)-1; 851 return (UV)-1;
476} 852}
477 853
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 * 854static SV *
494decode_str (dec_t *dec) 855decode_str (dec_t *dec)
495{ 856{
496 SV *sv = NEWSV (0,2); 857 SV *sv = 0;
497 int utf8 = 0; 858 int utf8 = 0;
498 char *cur = SvPVX (sv); 859 char *dec_cur = dec->cur;
499 char *end = SvEND (sv);
500 860
501 for (;;) 861 do
502 { 862 {
503 unsigned char ch = *(unsigned char *)dec->cur; 863 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
864 char *cur = buf;
504 865
505 if (ch == '"') 866 do
506 break;
507 else if (ch == '\\')
508 { 867 {
509 switch (*++dec->cur) 868 unsigned char ch = *(unsigned char *)dec_cur++;
869
870 if (expect_false (ch == '"'))
510 { 871 {
511 case '\\': 872 --dec_cur;
512 case '/': 873 break;
513 case '"': APPEND_CH (*dec->cur++); break; 874 }
514 875 else if (expect_false (ch == '\\'))
515 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 876 {
516 case 't': APPEND_CH ('\011'); ++dec->cur; break; 877 switch (*dec_cur)
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 { 878 {
523 UV lo, hi; 879 case '\\':
524 ++dec->cur; 880 case '/':
881 case '"': *cur++ = *dec_cur++; break;
525 882
526 hi = decode_4hex (dec); 883 case 'b': ++dec_cur; *cur++ = '\010'; break;
527 if (hi == (UV)-1) 884 case 't': ++dec_cur; *cur++ = '\011'; break;
528 goto fail; 885 case 'n': ++dec_cur; *cur++ = '\012'; break;
886 case 'f': ++dec_cur; *cur++ = '\014'; break;
887 case 'r': ++dec_cur; *cur++ = '\015'; break;
529 888
530 // possibly a surrogate pair 889 case 'u':
531 if (hi >= 0xd800 && hi < 0xdc00)
532 { 890 {
533 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 891 UV lo, hi;
534 ERR ("illegal surrogate character"); 892 ++dec_cur;
535 893
536 dec->cur += 2; 894 dec->cur = dec_cur;
537
538 lo = decode_4hex (dec); 895 hi = decode_4hex (dec);
896 dec_cur = dec->cur;
539 if (lo == (UV)-1) 897 if (hi == (UV)-1)
540 goto fail; 898 goto fail;
541 899
900 // possibly a surrogate pair
901 if (hi >= 0xd800)
902 if (hi < 0xdc00)
903 {
904 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
905 ERR ("missing low surrogate character in surrogate pair");
906
907 dec_cur += 2;
908
909 dec->cur = dec_cur;
910 lo = decode_4hex (dec);
911 dec_cur = dec->cur;
912 if (lo == (UV)-1)
913 goto fail;
914
542 if (lo < 0xdc00 || lo >= 0xe000) 915 if (lo < 0xdc00 || lo >= 0xe000)
543 ERR ("surrogate pair expected"); 916 ERR ("surrogate pair expected");
544 917
545 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 918 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
919 }
920 else if (hi < 0xe000)
921 ERR ("missing high surrogate character in surrogate pair");
922
923 if (hi >= 0x80)
924 {
925 utf8 = 1;
926
927 cur = encode_utf8 (cur, hi);
928 }
929 else
930 *cur++ = hi;
546 } 931 }
547 else if (lo >= 0xdc00 && lo < 0xe000)
548 ERR ("illegal surrogate character");
549
550 if (hi >= 0x80)
551 { 932 break;
552 utf8 = 1;
553 933
554 APPEND_GROW (4); // at most 4 bytes for 21 bits
555 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
556 }
557 else 934 default:
558 APPEND_CH (hi); 935 --dec_cur;
936 ERR ("illegal backslash escape sequence in string");
559 } 937 }
938 }
939 else if (expect_true (ch >= 0x20 && ch < 0x80))
940 *cur++ = ch;
941 else if (ch >= 0x80)
942 {
943 STRLEN clen;
944 UV uch;
945
946 --dec_cur;
947
948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949 if (clen == (STRLEN)-1)
950 ERR ("malformed UTF-8 character in JSON string");
951
952 do
953 *cur++ = *dec_cur++;
954 while (--clen);
955
956 utf8 = 1;
957 }
958 else
959 {
960 --dec_cur;
961
962 if (!ch)
963 ERR ("unexpected end of string while parsing JSON string");
560 break; 964 else
965 ERR ("invalid character encountered while parsing JSON string");
561 } 966 }
562 } 967 }
563 else if (ch >= 0x20 && ch <= 0x7f) 968 while (cur < buf + SHORT_STRING_LEN);
564 APPEND_CH (*dec->cur++); 969
565 else if (ch >= 0x80) 970 {
971 STRLEN len = cur - buf;
972
973 if (sv)
566 { 974 {
567 STRLEN clen; 975 SvGROW (sv, SvCUR (sv) + len + 1);
568 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 976 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
569 if (clen < 0) 977 SvCUR_set (sv, SvCUR (sv) + len);
570 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
571
572 APPEND_GROW (clen);
573 memcpy (cur, dec->cur, clen);
574 cur += clen;
575 dec->cur += clen;
576 } 978 }
577 else 979 else
578 ERR ("invalid character encountered"); 980 sv = newSVpvn (buf, len);
579 } 981 }
982 }
983 while (*dec_cur != '"');
580 984
581 ++dec->cur; 985 ++dec_cur;
582 986
583 SvCUR_set (sv, cur - SvPVX (sv)); 987 if (sv)
584 988 {
585 SvPOK_only (sv); 989 SvPOK_only (sv);
586 *SvEND (sv) = 0; 990 *SvEND (sv) = 0;
587 991
588 if (utf8) 992 if (utf8)
589 SvUTF8_on (sv); 993 SvUTF8_on (sv);
994 }
995 else
996 sv = newSVpvn ("", 0);
590 997
998 dec->cur = dec_cur;
591 return sv; 999 return sv;
592 1000
593fail: 1001fail:
594 SvREFCNT_dec (sv); 1002 dec->cur = dec_cur;
595 return 0; 1003 return 0;
596} 1004}
597 1005
598static SV * 1006static SV *
599decode_num (dec_t *dec) 1007decode_num (dec_t *dec)
… …
609 { 1017 {
610 ++dec->cur; 1018 ++dec->cur;
611 if (*dec->cur >= '0' && *dec->cur <= '9') 1019 if (*dec->cur >= '0' && *dec->cur <= '9')
612 ERR ("malformed number (leading zero must not be followed by another digit)"); 1020 ERR ("malformed number (leading zero must not be followed by another digit)");
613 } 1021 }
614 1022 else if (*dec->cur < '0' || *dec->cur > '9')
615 // int 1023 ERR ("malformed number (no digits after initial minus)");
1024 else
1025 do
1026 {
1027 ++dec->cur;
1028 }
616 while (*dec->cur >= '0' && *dec->cur <= '9') 1029 while (*dec->cur >= '0' && *dec->cur <= '9');
617 ++dec->cur;
618 1030
619 // [frac] 1031 // [frac]
620 if (*dec->cur == '.') 1032 if (*dec->cur == '.')
621 { 1033 {
622 is_nv = 1; 1034 ++dec->cur;
1035
1036 if (*dec->cur < '0' || *dec->cur > '9')
1037 ERR ("malformed number (no digits after decimal point)");
623 1038
624 do 1039 do
625 { 1040 {
626 ++dec->cur; 1041 ++dec->cur;
627 } 1042 }
628 while (*dec->cur >= '0' && *dec->cur <= '9'); 1043 while (*dec->cur >= '0' && *dec->cur <= '9');
1044
1045 is_nv = 1;
629 } 1046 }
630 1047
631 // [exp] 1048 // [exp]
632 if (*dec->cur == 'e' || *dec->cur == 'E') 1049 if (*dec->cur == 'e' || *dec->cur == 'E')
633 { 1050 {
634 is_nv = 1;
635
636 ++dec->cur; 1051 ++dec->cur;
1052
637 if (*dec->cur == '-' || *dec->cur == '+') 1053 if (*dec->cur == '-' || *dec->cur == '+')
638 ++dec->cur; 1054 ++dec->cur;
639 1055
1056 if (*dec->cur < '0' || *dec->cur > '9')
1057 ERR ("malformed number (no digits after exp sign)");
1058
1059 do
1060 {
1061 ++dec->cur;
1062 }
640 while (*dec->cur >= '0' && *dec->cur <= '9') 1063 while (*dec->cur >= '0' && *dec->cur <= '9');
641 ++dec->cur; 1064
1065 is_nv = 1;
642 } 1066 }
643 1067
644 if (!is_nv) 1068 if (!is_nv)
645 { 1069 {
646 UV uv; 1070 int len = dec->cur - start;
647 int numtype = grok_number (start, dec->cur - start, &uv); 1071
648 if (numtype & IS_NUMBER_IN_UV) 1072 // special case the rather common 1..5-digit-int case
649 if (numtype & IS_NUMBER_NEG) 1073 if (*start == '-')
1074 switch (len)
650 { 1075 {
651 if (uv < (UV)IV_MIN) 1076 case 2: return newSViv (-( start [1] - '0' * 1));
652 return newSViv (-(IV)uv); 1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1078 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1079 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1080 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
653 } 1081 }
1082 else
1083 switch (len)
1084 {
1085 case 1: return newSViv ( start [0] - '0' * 1);
1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1087 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1088 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1089 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1090 }
1091
1092 {
1093 UV uv;
1094 int numtype = grok_number (start, len, &uv);
1095 if (numtype & IS_NUMBER_IN_UV)
1096 if (numtype & IS_NUMBER_NEG)
1097 {
1098 if (uv < (UV)IV_MIN)
1099 return newSViv (-(IV)uv);
1100 }
654 else 1101 else
655 return newSVuv (uv); 1102 return newSVuv (uv);
656 } 1103 }
657 1104
1105 len -= *start == '-' ? 1 : 0;
1106
1107 // does not fit into IV or UV, try NV
1108 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1109 #if defined (LDBL_DIG)
1110 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1111 #endif
1112 )
1113 // fits into NV without loss of precision
1114 return newSVnv (Atof (start));
1115
1116 // everything else fails, convert it to a string
1117 return newSVpvn (start, dec->cur - start);
1118 }
1119
1120 // loss of precision here
658 return newSVnv (Atof (start)); 1121 return newSVnv (Atof (start));
659 1122
660fail: 1123fail:
661 return 0; 1124 return 0;
662} 1125}
… …
664static SV * 1127static SV *
665decode_av (dec_t *dec) 1128decode_av (dec_t *dec)
666{ 1129{
667 AV *av = newAV (); 1130 AV *av = newAV ();
668 1131
1132 DEC_INC_DEPTH;
1133 decode_ws (dec);
1134
1135 if (*dec->cur == ']')
1136 ++dec->cur;
1137 else
669 for (;;) 1138 for (;;)
670 { 1139 {
671 SV *value; 1140 SV *value;
672 1141
673 value = decode_sv (dec); 1142 value = decode_sv (dec);
674 if (!value) 1143 if (!value)
675 goto fail; 1144 goto fail;
676 1145
677 av_push (av, value); 1146 av_push (av, value);
678 1147
679 WS; 1148 decode_ws (dec);
680 1149
681 if (*dec->cur == ']') 1150 if (*dec->cur == ']')
682 { 1151 {
683 ++dec->cur; 1152 ++dec->cur;
684 break; 1153 break;
1154 }
685 } 1155
686
687 if (*dec->cur != ',') 1156 if (*dec->cur != ',')
688 ERR (", or ] expected while parsing array"); 1157 ERR (", or ] expected while parsing array");
689 1158
690 ++dec->cur; 1159 ++dec->cur;
1160
1161 decode_ws (dec);
1162
1163 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1164 {
1165 ++dec->cur;
1166 break;
1167 }
691 } 1168 }
692 1169
1170 DEC_DEC_DEPTH;
693 return newRV_noinc ((SV *)av); 1171 return newRV_noinc ((SV *)av);
694 1172
695fail: 1173fail:
696 SvREFCNT_dec (av); 1174 SvREFCNT_dec (av);
1175 DEC_DEC_DEPTH;
697 return 0; 1176 return 0;
698} 1177}
699 1178
700static SV * 1179static SV *
701decode_hv (dec_t *dec) 1180decode_hv (dec_t *dec)
702{ 1181{
1182 SV *sv;
703 HV *hv = newHV (); 1183 HV *hv = newHV ();
704 1184
1185 DEC_INC_DEPTH;
1186 decode_ws (dec);
1187
1188 if (*dec->cur == '}')
1189 ++dec->cur;
1190 else
705 for (;;) 1191 for (;;)
706 { 1192 {
707 SV *key, *value;
708
709 WS; EXPECT_CH ('"'); 1193 EXPECT_CH ('"');
710 1194
711 key = decode_str (dec); 1195 // heuristic: assume that
712 if (!key) 1196 // a) decode_str + hv_store_ent are abysmally slow.
713 goto fail; 1197 // b) most hash keys are short, simple ascii text.
714 1198 // => try to "fast-match" such strings to avoid
715 WS; EXPECT_CH (':'); 1199 // the overhead of decode_str + hv_store_ent.
716
717 value = decode_sv (dec);
718 if (!value)
719 { 1200 {
1201 SV *value;
1202 char *p = dec->cur;
1203 char *e = p + 24; // only try up to 24 bytes
1204
1205 for (;;)
1206 {
1207 // the >= 0x80 is false on most architectures
1208 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1209 {
1210 // slow path, back up and use decode_str
1211 SV *key = decode_str (dec);
1212 if (!key)
1213 goto fail;
1214
1215 decode_ws (dec); EXPECT_CH (':');
1216
1217 decode_ws (dec);
1218 value = decode_sv (dec);
1219 if (!value)
1220 {
1221 SvREFCNT_dec (key);
1222 goto fail;
1223 }
1224
1225 hv_store_ent (hv, key, value, 0);
720 SvREFCNT_dec (key); 1226 SvREFCNT_dec (key);
1227
1228 break;
1229 }
1230 else if (*p == '"')
1231 {
1232 // fast path, got a simple key
1233 char *key = dec->cur;
1234 int len = p - key;
1235 dec->cur = p + 1;
1236
1237 decode_ws (dec); EXPECT_CH (':');
1238
1239 decode_ws (dec);
1240 value = decode_sv (dec);
1241 if (!value)
721 goto fail; 1242 goto fail;
1243
1244 hv_store (hv, key, len, value, 0);
1245
1246 break;
1247 }
1248
1249 ++p;
1250 }
722 } 1251 }
723 1252
724 //TODO: optimise 1253 decode_ws (dec);
725 hv_store_ent (hv, key, value, 0);
726 1254
727 WS;
728
729 if (*dec->cur == '}') 1255 if (*dec->cur == '}')
1256 {
1257 ++dec->cur;
1258 break;
1259 }
1260
1261 if (*dec->cur != ',')
1262 ERR (", or } expected while parsing object/hash");
1263
1264 ++dec->cur;
1265
1266 decode_ws (dec);
1267
1268 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1269 {
1270 ++dec->cur;
1271 break;
1272 }
1273 }
1274
1275 DEC_DEC_DEPTH;
1276 sv = newRV_noinc ((SV *)hv);
1277
1278 // check filter callbacks
1279 if (dec->json.flags & F_HOOK)
1280 {
1281 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
730 { 1282 {
731 ++dec->cur; 1283 HE *cb, *he;
732 break; 1284
1285 hv_iterinit (hv);
1286 he = hv_iternext (hv);
1287 hv_iterinit (hv);
1288
1289 // the next line creates a mortal sv each time its called.
1290 // might want to optimise this for common cases.
1291 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1292
1293 if (cb)
1294 {
1295 dSP;
1296 int count;
1297
1298 ENTER; SAVETMPS; PUSHMARK (SP);
1299 XPUSHs (HeVAL (he));
1300
1301 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1302
1303 if (count == 1)
1304 {
1305 sv = newSVsv (POPs);
1306 FREETMPS; LEAVE;
1307 return sv;
1308 }
1309
1310 FREETMPS; LEAVE;
1311 }
733 } 1312 }
734 1313
735 if (*dec->cur != ',') 1314 if (dec->json.cb_object)
736 ERR (", or } expected while parsing object/hash"); 1315 {
1316 dSP;
1317 int count;
737 1318
738 ++dec->cur; 1319 ENTER; SAVETMPS; PUSHMARK (SP);
739 } 1320 XPUSHs (sv_2mortal (sv));
740 1321
741 return newRV_noinc ((SV *)hv); 1322 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1323
1324 if (count == 1)
1325 {
1326 sv = newSVsv (POPs);
1327 FREETMPS; LEAVE;
1328 return sv;
1329 }
1330
1331 SvREFCNT_inc (sv);
1332 FREETMPS; LEAVE;
1333 }
1334 }
1335
1336 return sv;
742 1337
743fail: 1338fail:
744 SvREFCNT_dec (hv); 1339 SvREFCNT_dec (hv);
1340 DEC_DEC_DEPTH;
745 return 0; 1341 return 0;
746} 1342}
747 1343
748static SV * 1344static SV *
749decode_sv (dec_t *dec) 1345decode_sv (dec_t *dec)
750{ 1346{
751 WS; 1347 // the beauty of JSON: you need exactly one character lookahead
1348 // to parse everything.
752 switch (*dec->cur) 1349 switch (*dec->cur)
753 { 1350 {
754 case '"': ++dec->cur; return decode_str (dec); 1351 case '"': ++dec->cur; return decode_str (dec);
755 case '[': ++dec->cur; return decode_av (dec); 1352 case '[': ++dec->cur; return decode_av (dec);
756 case '{': ++dec->cur; return decode_hv (dec); 1353 case '{': ++dec->cur; return decode_hv (dec);
757 1354
758 case '-': 1355 case '-':
759 case '0': case '1': case '2': case '3': case '4': 1356 case '0': case '1': case '2': case '3': case '4':
760 case '5': case '6': case '7': case '8': case '9': 1357 case '5': case '6': case '7': case '8': case '9':
761 return decode_num (dec); 1358 return decode_num (dec);
762 1359
763 case 't': 1360 case 't':
764 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1361 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
765 { 1362 {
766 dec->cur += 4; 1363 dec->cur += 4;
1364#if JSON_SLOW
1365 json_true = get_bool ("JSON::XS::true");
1366#endif
767 return newSViv (1); 1367 return newSVsv (json_true);
768 } 1368 }
769 else 1369 else
770 ERR ("'true' expected"); 1370 ERR ("'true' expected");
771 1371
772 break; 1372 break;
773 1373
774 case 'f': 1374 case 'f':
775 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1375 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
776 { 1376 {
777 dec->cur += 5; 1377 dec->cur += 5;
1378#if JSON_SLOW
1379 json_false = get_bool ("JSON::XS::false");
1380#endif
778 return newSViv (0); 1381 return newSVsv (json_false);
779 } 1382 }
780 else 1383 else
781 ERR ("'false' expected"); 1384 ERR ("'false' expected");
782 1385
783 break; 1386 break;
784 1387
785 case 'n': 1388 case 'n':
786 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1389 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
787 { 1390 {
788 dec->cur += 4; 1391 dec->cur += 4;
789 return newSViv (1); 1392 return newSVsv (&PL_sv_undef);
790 } 1393 }
791 else 1394 else
792 ERR ("'null' expected"); 1395 ERR ("'null' expected");
793 1396
794 break; 1397 break;
795 1398
796 default: 1399 default:
797 ERR ("malformed json string"); 1400 ERR ("malformed JSON string, neither array, object, number, string or atom");
798 break; 1401 break;
799 } 1402 }
800 1403
801fail: 1404fail:
802 return 0; 1405 return 0;
803} 1406}
804 1407
805static SV * 1408static SV *
806decode_json (SV *string, UV flags) 1409decode_json (SV *string, JSON *json, STRLEN *offset_return)
807{ 1410{
1411 dec_t dec;
1412 STRLEN offset;
808 SV *sv; 1413 SV *sv;
809 1414
1415 /* work around bugs in 5.10 where manipulating magic values
1416 * will perl ignore the magic in subsequent accesses
1417 */
1418 /*SvGETMAGIC (string);*/
1419 if (SvMAGICAL (string))
1420 string = sv_2mortal (newSVsv (string));
1421
1422 SvUPGRADE (string, SVt_PV);
1423
1424 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1425 * assertion with -DDEBUGGING, although SvCUR is documented to
1426 * return the xpv_cur field which certainly exists after upgrading.
1427 * according to nicholas clark, calling SvPOK fixes this.
1428 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1429 * and hope for the best.
1430 * Damnit, SvPV_nolen still trips over yet another assertion. This
1431 * assertion business is seriously broken, try yet another workaround
1432 * for the broken -DDEBUGGING.
1433 */
1434#ifdef DEBUGGING
1435 offset = SvOK (string) ? sv_len (string) : 0;
1436#else
1437 offset = SvCUR (string);
1438#endif
1439
1440 if (offset > json->max_size && json->max_size)
1441 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1442 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1443
810 if (!(flags & F_UTF8)) 1444 if (json->flags & F_UTF8)
1445 sv_utf8_downgrade (string, 0);
1446 else
811 sv_utf8_upgrade (string); 1447 sv_utf8_upgrade (string);
812 1448
813 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1449 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
814 1450
815 dec_t dec; 1451 dec.json = *json;
816 dec.flags = flags;
817 dec.cur = SvPVX (string); 1452 dec.cur = SvPVX (string);
818 dec.end = SvEND (string); 1453 dec.end = SvEND (string);
819 dec.err = 0; 1454 dec.err = 0;
1455 dec.depth = 0;
820 1456
821 *dec.end = 1; // invalid anywhere 1457 if (dec.json.cb_object || dec.json.cb_sk_object)
1458 dec.json.flags |= F_HOOK;
1459
1460 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1461
1462 decode_ws (&dec);
822 sv = decode_sv (&dec); 1463 sv = decode_sv (&dec);
823 *dec.end = 0; 1464
1465 if (!(offset_return || !sv))
1466 {
1467 // check for trailing garbage
1468 decode_ws (&dec);
1469
1470 if (*dec.cur)
1471 {
1472 dec.err = "garbage after JSON object";
1473 SvREFCNT_dec (sv);
1474 sv = 0;
1475 }
1476 }
1477
1478 if (offset_return || !sv)
1479 {
1480 offset = dec.json.flags & F_UTF8
1481 ? dec.cur - SvPVX (string)
1482 : utf8_distance (dec.cur, SvPVX (string));
1483
1484 if (offset_return)
1485 *offset_return = offset;
1486 }
824 1487
825 if (!sv) 1488 if (!sv)
826 { 1489 {
827 IV offset = utf8_distance (dec.cur, SvPVX (string));
828 SV *uni = sv_newmortal (); 1490 SV *uni = sv_newmortal ();
829 1491
1492 // horrible hack to silence warning inside pv_uni_display
1493 COP cop = *PL_curcop;
1494 cop.cop_warnings = pWARN_NONE;
1495 ENTER;
1496 SAVEVPTR (PL_curcop);
1497 PL_curcop = &cop;
830 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1498 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1499 LEAVE;
1500
831 croak ("%s, at character %d (%s)", 1501 croak ("%s, at character offset %d [\"%s\"]",
832 dec.err, 1502 dec.err,
833 (int)offset, 1503 (int)offset,
834 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1504 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
835 } 1505 }
836 1506
837 sv = sv_2mortal (sv); 1507 sv = sv_2mortal (sv);
838 1508
839 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1509 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)"); 1510 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
841 1511
842 return sv; 1512 return sv;
843} 1513}
844 1514
1515/////////////////////////////////////////////////////////////////////////////
1516// incremental parser
1517
1518static void
1519incr_parse (JSON *self)
1520{
1521 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1522
1523 for (;;)
1524 {
1525 //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
1526 switch (self->incr_mode)
1527 {
1528 // only used for intiial whitespace skipping
1529 case INCR_M_WS:
1530 for (;;)
1531 {
1532 if (*p > 0x20)
1533 {
1534 self->incr_mode = INCR_M_JSON;
1535 goto incr_m_json;
1536 }
1537 else if (!*p)
1538 goto interrupt;
1539
1540 ++p;
1541 }
1542
1543 // skip a single char inside a string (for \\-processing)
1544 case INCR_M_BS:
1545 if (!*p)
1546 goto interrupt;
1547
1548 ++p;
1549 self->incr_mode = INCR_M_STR;
1550 goto incr_m_str;
1551
1552 // inside a string
1553 case INCR_M_STR:
1554 incr_m_str:
1555 for (;;)
1556 {
1557 if (*p == '"')
1558 {
1559 ++p;
1560 self->incr_mode = INCR_M_JSON;
1561
1562 if (!self->incr_nest)
1563 goto interrupt;
1564
1565 goto incr_m_json;
1566 }
1567 else if (*p == '\\')
1568 {
1569 ++p; // "virtually" consumes character after \
1570
1571 if (!*p) // if at end of string we have to switch modes
1572 {
1573 self->incr_mode = INCR_M_BS;
1574 goto interrupt;
1575 }
1576 }
1577 else if (!*p)
1578 goto interrupt;
1579
1580 ++p;
1581 }
1582
1583 // after initial ws, outside string
1584 case INCR_M_JSON:
1585 incr_m_json:
1586 for (;;)
1587 {
1588 switch (*p++)
1589 {
1590 case 0:
1591 --p;
1592 goto interrupt;
1593
1594 case 0x09:
1595 case 0x0a:
1596 case 0x0d:
1597 case 0x20:
1598 if (!self->incr_nest)
1599 {
1600 --p; // do not eat the whitespace, let the next round do it
1601 goto interrupt;
1602 }
1603 break;
1604
1605 case '"':
1606 self->incr_mode = INCR_M_STR;
1607 goto incr_m_str;
1608
1609 case '[':
1610 case '{':
1611 if (++self->incr_nest > self->max_depth)
1612 croak (ERR_NESTING_EXCEEDED);
1613 break;
1614
1615 case ']':
1616 case '}':
1617 if (--self->incr_nest <= 0)
1618 goto interrupt;
1619 }
1620 }
1621 }
1622
1623 modechange:
1624 ;
1625 }
1626
1627interrupt:
1628 self->incr_pos = p - SvPVX (self->incr_text);
1629 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1630}
1631
1632/////////////////////////////////////////////////////////////////////////////
1633// XS interface functions
1634
845MODULE = JSON::XS PACKAGE = JSON::XS 1635MODULE = JSON::XS PACKAGE = JSON::XS
846 1636
847BOOT: 1637BOOT:
848{ 1638{
849 int i; 1639 int i;
850 1640
851 memset (decode_hexdigit, 0xff, 256);
852 for (i = 10; i--; ) 1641 for (i = 0; i < 256; ++i)
853 decode_hexdigit ['0' + i] = i; 1642 decode_hexdigit [i] =
1643 i >= '0' && i <= '9' ? i - '0'
1644 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1645 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1646 : -1;
854 1647
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); 1648 json_stash = gv_stashpv ("JSON::XS" , 1);
1649 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1650
1651 json_true = get_bool ("JSON::XS::true");
1652 json_false = get_bool ("JSON::XS::false");
862} 1653}
863 1654
864PROTOTYPES: DISABLE 1655PROTOTYPES: DISABLE
865 1656
866SV *new (char *dummy) 1657void CLONE (...)
867 CODE: 1658 CODE:
868 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1659 json_stash = 0;
1660 json_boolean_stash = 0;
1661
1662void new (char *klass)
1663 PPCODE:
1664{
1665 SV *pv = NEWSV (0, sizeof (JSON));
1666 SvPOK_only (pv);
1667 json_init ((JSON *)SvPVX (pv));
1668 XPUSHs (sv_2mortal (sv_bless (
1669 newRV_noinc (pv),
1670 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1671 )));
1672}
1673
1674void ascii (JSON *self, int enable = 1)
1675 ALIAS:
1676 ascii = F_ASCII
1677 latin1 = F_LATIN1
1678 utf8 = F_UTF8
1679 indent = F_INDENT
1680 canonical = F_CANONICAL
1681 space_before = F_SPACE_BEFORE
1682 space_after = F_SPACE_AFTER
1683 pretty = F_PRETTY
1684 allow_nonref = F_ALLOW_NONREF
1685 shrink = F_SHRINK
1686 allow_blessed = F_ALLOW_BLESSED
1687 convert_blessed = F_CONV_BLESSED
1688 relaxed = F_RELAXED
1689 allow_unknown = F_ALLOW_UNKNOWN
1690 PPCODE:
1691{
1692 if (enable)
1693 self->flags |= ix;
1694 else
1695 self->flags &= ~ix;
1696
1697 XPUSHs (ST (0));
1698}
1699
1700void get_ascii (JSON *self)
1701 ALIAS:
1702 get_ascii = F_ASCII
1703 get_latin1 = F_LATIN1
1704 get_utf8 = F_UTF8
1705 get_indent = F_INDENT
1706 get_canonical = F_CANONICAL
1707 get_space_before = F_SPACE_BEFORE
1708 get_space_after = F_SPACE_AFTER
1709 get_allow_nonref = F_ALLOW_NONREF
1710 get_shrink = F_SHRINK
1711 get_allow_blessed = F_ALLOW_BLESSED
1712 get_convert_blessed = F_CONV_BLESSED
1713 get_relaxed = F_RELAXED
1714 get_allow_unknown = F_ALLOW_UNKNOWN
1715 PPCODE:
1716 XPUSHs (boolSV (self->flags & ix));
1717
1718void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1719 PPCODE:
1720 self->max_depth = max_depth;
1721 XPUSHs (ST (0));
1722
1723U32 get_max_depth (JSON *self)
1724 CODE:
1725 RETVAL = self->max_depth;
869 OUTPUT: 1726 OUTPUT:
870 RETVAL 1727 RETVAL
871 1728
872SV *ascii (SV *self, int enable) 1729void max_size (JSON *self, U32 max_size = 0)
873 ALIAS: 1730 PPCODE:
874 ascii = F_ASCII 1731 self->max_size = max_size;
875 utf8 = F_UTF8 1732 XPUSHs (ST (0));
876 indent = F_INDENT 1733
877 canonical = F_CANONICAL 1734int 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: 1735 CODE:
884{ 1736 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: 1737 OUTPUT:
894 RETVAL 1738 RETVAL
895 1739
896void encode (SV *self, SV *scalar) 1740void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
897 PPCODE: 1741 PPCODE:
898 XPUSHs (encode_json (scalar, *SvJSON (self))); 1742{
1743 SvREFCNT_dec (self->cb_object);
1744 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
899 1745
900void decode (SV *self, SV *jsonstr) 1746 XPUSHs (ST (0));
1747}
1748
1749void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
901 PPCODE: 1750 PPCODE:
1751{
1752 if (!self->cb_sk_object)
1753 self->cb_sk_object = newHV ();
1754
1755 if (SvOK (cb))
1756 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1757 else
1758 {
1759 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1760
1761 if (!HvKEYS (self->cb_sk_object))
1762 {
1763 SvREFCNT_dec (self->cb_sk_object);
1764 self->cb_sk_object = 0;
1765 }
1766 }
1767
1768 XPUSHs (ST (0));
1769}
1770
1771void encode (JSON *self, SV *scalar)
1772 PPCODE:
1773 XPUSHs (encode_json (scalar, self));
1774
1775void decode (JSON *self, SV *jsonstr)
1776 PPCODE:
902 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1777 XPUSHs (decode_json (jsonstr, self, 0));
1778
1779void decode_prefix (JSON *self, SV *jsonstr)
1780 PPCODE:
1781{
1782 STRLEN offset;
1783 EXTEND (SP, 2);
1784 PUSHs (decode_json (jsonstr, self, &offset));
1785 PUSHs (sv_2mortal (newSVuv (offset)));
1786}
1787
1788void incr_parse (JSON *self, SV *jsonstr = 0)
1789 PPCODE:
1790{
1791 if (!self->incr_text)
1792 self->incr_text = newSVpvn ("", 0);
1793
1794 // append data, if any
1795 if (jsonstr)
1796 {
1797 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1798 {
1799 /* utf-8-ness differs, need to upgrade */
1800 sv_utf8_upgrade (self->incr_text);
1801
1802 if (self->incr_pos)
1803 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1804 - (U8 *)SvPVX (self->incr_text);
1805 }
1806
1807 {
1808 STRLEN len;
1809 const char *str = SvPV (jsonstr, len);
1810 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1811 Move (str, SvEND (self->incr_text), len, char);
1812 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1813 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1814 }
1815 }
1816
1817 if (GIMME_V != G_VOID)
1818 do
1819 {
1820 STRLEN offset;
1821
1822 if (!INCR_DONE (self))
1823 {
1824 incr_parse (self);
1825
1826 if (self->incr_pos > self->max_size && self->max_size)
1827 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1828 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1829
1830 if (!INCR_DONE (self))
1831 break;
1832 }
1833
1834 XPUSHs (decode_json (self->incr_text, self, &offset));
1835
1836 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1837 self->incr_pos -= offset;
1838 self->incr_nest = 0;
1839 self->incr_mode = 0;
1840 }
1841 while (GIMME_V == G_ARRAY);
1842}
1843
1844SV *incr_text (JSON *self)
1845 ATTRS: lvalue
1846 CODE:
1847{
1848 if (self->incr_pos)
1849 croak ("incr_text can not be called when the incremental parser already started parsing");
1850
1851 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1852}
1853 OUTPUT:
1854 RETVAL
1855
1856void incr_skip (JSON *self)
1857 CODE:
1858{
1859 if (self->incr_pos)
1860 {
1861 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1862 self->incr_pos = 0;
1863 self->incr_nest = 0;
1864 self->incr_mode = 0;
1865 }
1866}
1867
1868void incr_reset (JSON *self)
1869 CODE:
1870{
1871 SvREFCNT_dec (self->incr_text);
1872 self->incr_text = 0;
1873 self->incr_pos = 0;
1874 self->incr_nest = 0;
1875 self->incr_mode = 0;
1876}
1877
1878void DESTROY (JSON *self)
1879 CODE:
1880 SvREFCNT_dec (self->cb_sk_object);
1881 SvREFCNT_dec (self->cb_object);
1882 SvREFCNT_dec (self->incr_text);
903 1883
904PROTOTYPES: ENABLE 1884PROTOTYPES: ENABLE
905 1885
906void to_json (SV *scalar) 1886void encode_json (SV *scalar)
1887 ALIAS:
1888 to_json_ = 0
1889 encode_json = F_UTF8
907 PPCODE: 1890 PPCODE:
1891{
1892 JSON json;
1893 json_init (&json);
1894 json.flags |= ix;
908 XPUSHs (encode_json (scalar, F_UTF8)); 1895 XPUSHs (encode_json (scalar, &json));
1896}
909 1897
910void from_json (SV *jsonstr) 1898void decode_json (SV *jsonstr)
1899 ALIAS:
1900 from_json_ = 0
1901 decode_json = F_UTF8
911 PPCODE: 1902 PPCODE:
1903{
1904 JSON json;
1905 json_init (&json);
1906 json.flags |= ix;
912 XPUSHs (decode_json (jsonstr, F_UTF8)); 1907 XPUSHs (decode_json (jsonstr, &json, 0));
1908}
913 1909
1910

Diff Legend

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