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 1.36 by root, Wed Jun 6 17:27:33 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" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 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
14#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 28#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 29#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 30#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 35#define F_RELAXED 0x00001000UL
24#define S_MAXDEPTH 27 36#define F_ALLOW_UNKNOWN 0x00002000UL
25 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 38
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH)
33 40
34#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
36 43
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size 44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38 45
39#define SB do { 46#define SB do {
40#define SE } while (0) 47#define SE } while (0)
41 48
42#if __GNUC__ >= 3 49#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 50# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 51# define INLINE static inline
45#else 52#else
46# define expect(expr,value) (expr) 53# define expect(expr,value) (expr)
47# define inline static 54# define INLINE static
48#endif 55#endif
49 56
50#define expect_false(expr) expect ((expr) != 0, 0) 57#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 58#define expect_true(expr) expect ((expr) != 0, 1)
52 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
53static HV *json_stash; // JSON::XS:: 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}
54 107
55///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
56// utility functions 109// utility functions
57 110
58static UV * 111INLINE SV *
59SvJSON (SV *sv) 112get_bool (const char *name)
60{ 113{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 114 SV *sv = get_sv (name, 1);
62 croak ("object is not of type JSON::XS");
63 115
64 return &SvUVX (SvRV (sv)); 116 SvREADONLY_on (sv);
65} 117 SvREADONLY_on (SvRV (sv));
66 118
67static void 119 return sv;
120}
121
122INLINE void
68shrink (SV *sv) 123shrink (SV *sv)
69{ 124{
70 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
71 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 128 {
73#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
… …
81// decode an utf-8 character and return it, or (UV)-1 in 137// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error. 138// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
86inline UV 142INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 144{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 145 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 148 {
93 *clen = 2; 149 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 151 }
96 else 152 else
97 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 154}
99 return (UV)-1; 155
100 } 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;
101} 178}
102 179
103///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
104// encoder 181// encoder
105 182
… …
107typedef struct 184typedef struct
108{ 185{
109 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
111 SV *sv; // result scalar 188 SV *sv; // result scalar
112 U32 flags; // F_* 189 JSON json;
113 U32 indent; // indentation level 190 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
115} enc_t; 192} enc_t;
116 193
117inline void 194INLINE void
118need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
119{ 196{
120 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
121 { 198 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + len + 1); 200 SvGROW (enc->sv, cur + len + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 203 }
127} 204}
128 205
129inline void 206INLINE void
130encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
131{ 208{
132 need (enc, 1); 209 need (enc, 1);
133 *enc->cur++ = ch; 210 *enc->cur++ = ch;
134} 211}
… …
178 STRLEN clen; 255 STRLEN clen;
179 UV uch; 256 UV uch;
180 257
181 if (is_utf8) 258 if (is_utf8)
182 { 259 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 260 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 261 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 262 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 263 }
188 else 264 else
189 { 265 {
190 uch = ch; 266 uch = ch;
191 clen = 1; 267 clen = 1;
192 } 268 }
193 269
194 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 { 271 {
199 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
200 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
201 need (enc, len += 11); 277 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
203 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
204 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
205 enc->cur += 12; 281 enc->cur += 12;
206 } 282 }
207 else 283 else
208 { 284 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5); 285 need (enc, len += 5);
211 *enc->cur++ = '\\'; 286 *enc->cur++ = '\\';
212 *enc->cur++ = 'u'; 287 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ]; 288 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 289 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 290 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 291 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
217 } 292 }
218 293
219 str += clen; 294 str += clen;
220 } 295 }
221 else if (enc->flags & F_LATIN1) 296 else if (enc->json.flags & F_LATIN1)
222 { 297 {
223 *enc->cur++ = uch; 298 *enc->cur++ = uch;
224 str += clen; 299 str += clen;
225 } 300 }
226 else if (is_utf8) 301 else if (is_utf8)
… …
233 while (--clen); 308 while (--clen);
234 } 309 }
235 else 310 else
236 { 311 {
237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 312 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 313 enc->cur = encode_utf8 (enc->cur, uch);
239 ++str; 314 ++str;
240 } 315 }
241 } 316 }
242 } 317 }
243 } 318 }
244 319
245 --len; 320 --len;
246 } 321 }
247} 322}
248 323
249inline void 324INLINE void
250encode_indent (enc_t *enc) 325encode_indent (enc_t *enc)
251{ 326{
252 if (enc->flags & F_INDENT) 327 if (enc->json.flags & F_INDENT)
253 { 328 {
254 int spaces = enc->indent * INDENT_STEP; 329 int spaces = enc->indent * INDENT_STEP;
255 330
256 need (enc, spaces); 331 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 332 memset (enc->cur, ' ', spaces);
258 enc->cur += spaces; 333 enc->cur += spaces;
259 } 334 }
260} 335}
261 336
262inline void 337INLINE void
263encode_space (enc_t *enc) 338encode_space (enc_t *enc)
264{ 339{
265 need (enc, 1); 340 need (enc, 1);
266 encode_ch (enc, ' '); 341 encode_ch (enc, ' ');
267} 342}
268 343
269inline void 344INLINE void
270encode_nl (enc_t *enc) 345encode_nl (enc_t *enc)
271{ 346{
272 if (enc->flags & F_INDENT) 347 if (enc->json.flags & F_INDENT)
273 { 348 {
274 need (enc, 1); 349 need (enc, 1);
275 encode_ch (enc, '\n'); 350 encode_ch (enc, '\n');
276 } 351 }
277} 352}
278 353
279inline void 354INLINE void
280encode_comma (enc_t *enc) 355encode_comma (enc_t *enc)
281{ 356{
282 encode_ch (enc, ','); 357 encode_ch (enc, ',');
283 358
284 if (enc->flags & F_INDENT) 359 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 360 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 361 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 362 encode_space (enc);
288} 363}
289 364
290static void encode_sv (enc_t *enc, SV *sv); 365static void encode_sv (enc_t *enc, SV *sv);
291 366
292static void 367static void
293encode_av (enc_t *enc, AV *av) 368encode_av (enc_t *enc, AV *av)
294{ 369{
295 int i, len = av_len (av); 370 int i, len = av_len (av);
296 371
297 if (enc->indent >= enc->maxdepth) 372 if (enc->indent >= enc->json.max_depth)
298 croak ("data structure too deep (hit recursion limit)"); 373 croak (ERR_NESTING_EXCEEDED);
299 374
300 encode_ch (enc, '['); encode_nl (enc); 375 encode_ch (enc, '[');
301 ++enc->indent; 376
377 if (len >= 0)
378 {
379 encode_nl (enc); ++enc->indent;
302 380
303 for (i = 0; i <= len; ++i) 381 for (i = 0; i <= len; ++i)
304 { 382 {
383 SV **svp = av_fetch (av, i, 0);
384
305 encode_indent (enc); 385 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0));
307 386
387 if (svp)
388 encode_sv (enc, *svp);
389 else
390 encode_str (enc, "null", 4, 0);
391
308 if (i < len) 392 if (i < len)
309 encode_comma (enc); 393 encode_comma (enc);
310 } 394 }
311 395
396 encode_nl (enc); --enc->indent; encode_indent (enc);
397 }
398
312 encode_nl (enc); 399 encode_ch (enc, ']');
313
314 --enc->indent;
315 encode_indent (enc); encode_ch (enc, ']');
316} 400}
317 401
318static void 402static void
319encode_he (enc_t *enc, HE *he) 403encode_hk (enc_t *enc, HE *he)
320{ 404{
321 encode_ch (enc, '"'); 405 encode_ch (enc, '"');
322 406
323 if (HeKLEN (he) == HEf_SVKEY) 407 if (HeKLEN (he) == HEf_SVKEY)
324 { 408 {
… …
334 else 418 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 419 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 420
337 encode_ch (enc, '"'); 421 encode_ch (enc, '"');
338 422
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 423 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 424 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 425 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he));
343} 426}
344 427
345// compare hash entries, used when all keys are bytestrings 428// compare hash entries, used when all keys are bytestrings
346static int 429static int
347he_cmp_fast (const void *a_, const void *b_) 430he_cmp_fast (const void *a_, const void *b_)
… …
352 HE *b = *(HE **)b_; 435 HE *b = *(HE **)b_;
353 436
354 STRLEN la = HeKLEN (a); 437 STRLEN la = HeKLEN (a);
355 STRLEN lb = HeKLEN (b); 438 STRLEN lb = HeKLEN (b);
356 439
357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 440 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
358 cmp = la - lb; 441 cmp = lb - la;
359 442
360 return cmp; 443 return cmp;
361} 444}
362 445
363// 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
364static int 447static int
365he_cmp_slow (const void *a, const void *b) 448he_cmp_slow (const void *a, const void *b)
366{ 449{
367 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 450 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
368} 451}
369 452
370static void 453static void
371encode_hv (enc_t *enc, HV *hv) 454encode_hv (enc_t *enc, HV *hv)
372{ 455{
373 int count, i; 456 HE *he;
374 457
375 if (enc->indent >= enc->maxdepth) 458 if (enc->indent >= enc->json.max_depth)
376 croak ("data structure too deep (hit recursion limit)"); 459 croak (ERR_NESTING_EXCEEDED);
377 460
378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 461 encode_ch (enc, '{');
379 462
380 if ((count = hv_iterinit (hv)))
381 {
382 // for canonical output we have to sort by keys first 463 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 464 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 465 // security workaround added somewhere in 5.8.x
385 // that randomises hash orderings 466 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 467 if (enc->json.flags & F_CANONICAL)
468 {
469 int count = hv_iterinit (hv);
470
471 if (SvMAGICAL (hv))
387 { 472 {
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 {
388 int fast = 1; 487 int i, fast = 1;
389 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER) 488#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE)); 489 HE **hes = _alloca (count * sizeof (HE));
392#else 490#else
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 491 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
394#endif 492#endif
… …
421 519
422 FREETMPS; 520 FREETMPS;
423 LEAVE; 521 LEAVE;
424 } 522 }
425 523
426 for (i = 0; i < count; ++i) 524 encode_nl (enc); ++enc->indent;
525
526 while (count--)
427 { 527 {
428 encode_indent (enc); 528 encode_indent (enc);
529 he = hes [count];
429 encode_he (enc, hes [i]); 530 encode_hk (enc, he);
531 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
430 532
431 if (i < count - 1) 533 if (count)
432 encode_comma (enc); 534 encode_comma (enc);
433 } 535 }
434 536
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)))
544 {
545 encode_nl (enc); ++enc->indent;
546
547 for (;;)
548 {
435 encode_nl (enc); 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);
436 } 587 }
437 else 588 else
438 { 589 {
439 HE *he = hv_iternext (hv); 590#if 0
440 591 if (0 && sv_derived_from (rv, "JSON::Literal"))
441 for (;;)
442 { 592 {
443 encode_indent (enc); 593 // not yet
444 encode_he (enc, he);
445
446 if (!(he = hv_iternext (hv)))
447 break;
448
449 encode_comma (enc);
450 } 594 }
595#endif
596 if (enc->json.flags & F_CONV_BLESSED)
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);
451 600
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
452 encode_nl (enc); 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))));
453 } 635 }
454 } 636 }
455
456 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
457}
458
459// encode objects, arrays and special \0=false and \1=true values.
460static void
461encode_rv (enc_t *enc, SV *sv)
462{
463 svtype svt;
464
465 SvGETMAGIC (sv);
466 svt = SvTYPE (sv);
467
468 if (svt == SVt_PVHV) 637 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 638 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 639 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 640 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 641 else if (svt < SVt_PVAV)
473 { 642 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 643 STRLEN len = 0;
644 char *pv = svt ? SvPV (sv, len) : 0;
645
646 if (len == 1 && *pv == '1')
647 encode_str (enc, "true", 4, 0);
648 else if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 649 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1) 650 else if (enc->json.flags & F_ALLOW_UNKNOWN)
477 encode_str (enc, "true", 4, 0); 651 encode_str (enc, "null", 4, 0);
478 else 652 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 653 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 654 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 655 }
656 else if (enc->json.flags & F_ALLOW_UNKNOWN)
657 encode_str (enc, "null", 4, 0);
482 else 658 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 659 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
485} 661}
486 662
… …
497 encode_str (enc, str, len, SvUTF8 (sv)); 673 encode_str (enc, str, len, SvUTF8 (sv));
498 encode_ch (enc, '"'); 674 encode_ch (enc, '"');
499 } 675 }
500 else if (SvNOKp (sv)) 676 else if (SvNOKp (sv))
501 { 677 {
678 // trust that perl will do the right thing w.r.t. JSON syntax.
502 need (enc, NV_DIG + 32); 679 need (enc, NV_DIG + 32);
503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 680 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504 enc->cur += strlen (enc->cur); 681 enc->cur += strlen (enc->cur);
505 } 682 }
506 else if (SvIOKp (sv)) 683 else if (SvIOKp (sv))
507 { 684 {
508 // we assume we can always read an IV as a UV 685 // we assume we can always read an IV as a UV and vice versa
509 if (SvUV (sv) & ~(UV)0x7fff) 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)
510 { 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
511 need (enc, 64); 698 need (enc, 6);
512 enc->cur += 699
513 SvIsUV(sv) 700 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
514 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 701 u = i < 0 ? -i : i;
515 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 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'
516 } 716 }
517 else 717 else
518 { 718 {
519 // optimise the "small number case" 719 // large integer, use the (rather slow) snprintf way.
520 // code will likely be branchless and use only a single multiplication
521 I32 i = SvIV (sv);
522 U32 u;
523
524 need (enc, 6); 720 need (enc, IVUV_MAXCHARS);
525 721 enc->cur +=
526 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; 722 SvIsUV(sv)
527 u = i < 0 ? -i : i; 723 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
528 724 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
529 // convert to 4.28 fixed-point representation
530 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531
532 char digit, nz = 0;
533
534 digit = u >> 28; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0xfffffff) * 5;
535 digit = u >> 27; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x7ffffff) * 5;
536 digit = u >> 26; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x3ffffff) * 5;
537 digit = u >> 25; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x1ffffff) * 5;
538 digit = u >> 24; *enc->cur = digit + '0'; nz |= digit; enc->cur += 1;
539 } 725 }
540 } 726 }
541 else if (SvROK (sv)) 727 else if (SvROK (sv))
542 encode_rv (enc, SvRV (sv)); 728 encode_rv (enc, SvRV (sv));
543 else if (!SvOK (sv)) 729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
544 encode_str (enc, "null", 4, 0); 730 encode_str (enc, "null", 4, 0);
545 else 731 else
546 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
547 SvPV_nolen (sv), SvFLAGS (sv)); 733 SvPV_nolen (sv), SvFLAGS (sv));
548} 734}
549 735
550static SV * 736static SV *
551encode_json (SV *scalar, U32 flags) 737encode_json (SV *scalar, JSON *json)
552{ 738{
553 enc_t enc; 739 enc_t enc;
554 740
555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 741 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 742 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557 743
558 enc.flags = flags; 744 enc.json = *json;
559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 745 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
560 enc.cur = SvPVX (enc.sv); 746 enc.cur = SvPVX (enc.sv);
561 enc.end = SvEND (enc.sv); 747 enc.end = SvEND (enc.sv);
562 enc.indent = 0; 748 enc.indent = 0;
563 enc.maxdepth = DEC_DEPTH (flags); 749 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
750 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL;
564 752
565 SvPOK_only (enc.sv); 753 SvPOK_only (enc.sv);
566 encode_sv (&enc, scalar); 754 encode_sv (&enc, scalar);
567 755
568 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570 758
571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
572 SvUTF8_on (enc.sv); 760 SvUTF8_on (enc.sv);
573 761
574 if (enc.flags & F_SHRINK) 762 if (enc.json.flags & F_SHRINK)
575 shrink (enc.sv); 763 shrink (enc.sv);
576 764
577 return enc.sv; 765 return enc.sv;
578} 766}
579 767
… …
584typedef struct 772typedef struct
585{ 773{
586 char *cur; // current parser pointer 774 char *cur; // current parser pointer
587 char *end; // end of input string 775 char *end; // end of input string
588 const char *err; // parse error, if != 0 776 const char *err; // parse error, if != 0
589 U32 flags; // F_* 777 JSON json;
590 U32 depth; // recursion depth 778 U32 depth; // recursion depth
591 U32 maxdepth; // recursion depth limit 779 U32 maxdepth; // recursion depth limit
592} dec_t; 780} dec_t;
593 781
594inline void 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
595decode_ws (dec_t *dec) 792decode_ws (dec_t *dec)
596{ 793{
597 for (;;) 794 for (;;)
598 { 795 {
599 char ch = *dec->cur; 796 char ch = *dec->cur;
600 797
601 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 }
602 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 810 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
603 break; 811 break; // parse error, but let higher level handle it, gives better error messages
604 812
605 ++dec->cur; 813 ++dec->cur;
606 } 814 }
607} 815}
608 816
… …
612 if (*dec->cur != ch) \ 820 if (*dec->cur != ch) \
613 ERR (# ch " expected"); \ 821 ERR (# ch " expected"); \
614 ++dec->cur; \ 822 ++dec->cur; \
615 SE 823 SE
616 824
617#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 825#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
618#define DEC_DEC_DEPTH --dec->depth 826#define DEC_DEC_DEPTH --dec->depth
619 827
620static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
621 829
622static signed char decode_hexdigit[256]; 830static signed char decode_hexdigit[256];
… …
625decode_4hex (dec_t *dec) 833decode_4hex (dec_t *dec)
626{ 834{
627 signed char d1, d2, d3, d4; 835 signed char d1, d2, d3, d4;
628 unsigned char *cur = (unsigned char *)dec->cur; 836 unsigned char *cur = (unsigned char *)dec->cur;
629 837
630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 838 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 839 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 840 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 841 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
634 842
635 dec->cur += 4; 843 dec->cur += 4;
636 844
637 return ((UV)d1) << 12 845 return ((UV)d1) << 12
638 | ((UV)d2) << 8 846 | ((UV)d2) << 8
… …
646static SV * 854static SV *
647decode_str (dec_t *dec) 855decode_str (dec_t *dec)
648{ 856{
649 SV *sv = 0; 857 SV *sv = 0;
650 int utf8 = 0; 858 int utf8 = 0;
859 char *dec_cur = dec->cur;
651 860
652 do 861 do
653 { 862 {
654 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 863 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
655 char *cur = buf; 864 char *cur = buf;
656 865
657 do 866 do
658 { 867 {
659 unsigned char ch = *(unsigned char *)dec->cur++; 868 unsigned char ch = *(unsigned char *)dec_cur++;
660 869
661 if (expect_false (ch == '"')) 870 if (expect_false (ch == '"'))
662 { 871 {
663 --dec->cur; 872 --dec_cur;
664 break; 873 break;
665 } 874 }
666 else if (expect_false (ch == '\\')) 875 else if (expect_false (ch == '\\'))
667 { 876 {
668 switch (*dec->cur) 877 switch (*dec_cur)
669 { 878 {
670 case '\\': 879 case '\\':
671 case '/': 880 case '/':
672 case '"': *cur++ = *dec->cur++; break; 881 case '"': *cur++ = *dec_cur++; break;
673 882
674 case 'b': ++dec->cur; *cur++ = '\010'; break; 883 case 'b': ++dec_cur; *cur++ = '\010'; break;
675 case 't': ++dec->cur; *cur++ = '\011'; break; 884 case 't': ++dec_cur; *cur++ = '\011'; break;
676 case 'n': ++dec->cur; *cur++ = '\012'; break; 885 case 'n': ++dec_cur; *cur++ = '\012'; break;
677 case 'f': ++dec->cur; *cur++ = '\014'; break; 886 case 'f': ++dec_cur; *cur++ = '\014'; break;
678 case 'r': ++dec->cur; *cur++ = '\015'; break; 887 case 'r': ++dec_cur; *cur++ = '\015'; break;
679 888
680 case 'u': 889 case 'u':
681 { 890 {
682 UV lo, hi; 891 UV lo, hi;
683 ++dec->cur; 892 ++dec_cur;
684 893
894 dec->cur = dec_cur;
685 hi = decode_4hex (dec); 895 hi = decode_4hex (dec);
896 dec_cur = dec->cur;
686 if (hi == (UV)-1) 897 if (hi == (UV)-1)
687 goto fail; 898 goto fail;
688 899
689 // possibly a surrogate pair 900 // possibly a surrogate pair
690 if (hi >= 0xd800) 901 if (hi >= 0xd800)
691 if (hi < 0xdc00) 902 if (hi < 0xdc00)
692 { 903 {
693 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 904 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
694 ERR ("missing low surrogate character in surrogate pair"); 905 ERR ("missing low surrogate character in surrogate pair");
695 906
696 dec->cur += 2; 907 dec_cur += 2;
697 908
909 dec->cur = dec_cur;
698 lo = decode_4hex (dec); 910 lo = decode_4hex (dec);
911 dec_cur = dec->cur;
699 if (lo == (UV)-1) 912 if (lo == (UV)-1)
700 goto fail; 913 goto fail;
701 914
702 if (lo < 0xdc00 || lo >= 0xe000) 915 if (lo < 0xdc00 || lo >= 0xe000)
703 ERR ("surrogate pair expected"); 916 ERR ("surrogate pair expected");
… …
709 922
710 if (hi >= 0x80) 923 if (hi >= 0x80)
711 { 924 {
712 utf8 = 1; 925 utf8 = 1;
713 926
714 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 927 cur = encode_utf8 (cur, hi);
715 } 928 }
716 else 929 else
717 *cur++ = hi; 930 *cur++ = hi;
718 } 931 }
719 break; 932 break;
720 933
721 default: 934 default:
722 --dec->cur; 935 --dec_cur;
723 ERR ("illegal backslash escape sequence in string"); 936 ERR ("illegal backslash escape sequence in string");
724 } 937 }
725 } 938 }
726 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 939 else if (expect_true (ch >= 0x20 && ch < 0x80))
727 *cur++ = ch; 940 *cur++ = ch;
728 else if (ch >= 0x80) 941 else if (ch >= 0x80)
729 { 942 {
730 STRLEN clen; 943 STRLEN clen;
731 UV uch; 944 UV uch;
732 945
733 --dec->cur; 946 --dec_cur;
734 947
735 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
736 if (clen == (STRLEN)-1) 949 if (clen == (STRLEN)-1)
737 ERR ("malformed UTF-8 character in JSON string"); 950 ERR ("malformed UTF-8 character in JSON string");
738 951
739 do 952 do
740 *cur++ = *dec->cur++; 953 *cur++ = *dec_cur++;
741 while (--clen); 954 while (--clen);
742 955
743 utf8 = 1; 956 utf8 = 1;
744 } 957 }
745 else 958 else
746 { 959 {
747 --dec->cur; 960 --dec_cur;
748 961
749 if (!ch) 962 if (!ch)
750 ERR ("unexpected end of string while parsing JSON string"); 963 ERR ("unexpected end of string while parsing JSON string");
751 else 964 else
752 ERR ("invalid character encountered while parsing JSON string"); 965 ERR ("invalid character encountered while parsing JSON string");
… …
765 } 978 }
766 else 979 else
767 sv = newSVpvn (buf, len); 980 sv = newSVpvn (buf, len);
768 } 981 }
769 } 982 }
770 while (*dec->cur != '"'); 983 while (*dec_cur != '"');
771 984
772 ++dec->cur; 985 ++dec_cur;
773 986
774 if (sv) 987 if (sv)
775 { 988 {
776 SvPOK_only (sv); 989 SvPOK_only (sv);
777 *SvEND (sv) = 0; 990 *SvEND (sv) = 0;
… …
780 SvUTF8_on (sv); 993 SvUTF8_on (sv);
781 } 994 }
782 else 995 else
783 sv = newSVpvn ("", 0); 996 sv = newSVpvn ("", 0);
784 997
998 dec->cur = dec_cur;
785 return sv; 999 return sv;
786 1000
787fail: 1001fail:
1002 dec->cur = dec_cur;
788 return 0; 1003 return 0;
789} 1004}
790 1005
791static SV * 1006static SV *
792decode_num (dec_t *dec) 1007decode_num (dec_t *dec)
… …
850 is_nv = 1; 1065 is_nv = 1;
851 } 1066 }
852 1067
853 if (!is_nv) 1068 if (!is_nv)
854 { 1069 {
1070 int len = dec->cur - start;
1071
855 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1072 // special case the rather common 1..5-digit-int case
856 if (*start == '-') 1073 if (*start == '-')
857 switch (dec->cur - start) 1074 switch (len)
858 { 1075 {
859 case 2: return newSViv (-( start [1] - '0' )); 1076 case 2: return newSViv (-( start [1] - '0' * 1));
860 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
861 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1078 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
862 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 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));
863 } 1081 }
864 else 1082 else
865 switch (dec->cur - start) 1083 switch (len)
866 { 1084 {
867 case 1: return newSViv ( start [0] - '0' ); 1085 case 1: return newSViv ( start [0] - '0' * 1);
868 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
869 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1087 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
870 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 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);
871 } 1090 }
872 1091
873 { 1092 {
874 UV uv; 1093 UV uv;
875 int numtype = grok_number (start, dec->cur - start, &uv); 1094 int numtype = grok_number (start, len, &uv);
876 if (numtype & IS_NUMBER_IN_UV) 1095 if (numtype & IS_NUMBER_IN_UV)
877 if (numtype & IS_NUMBER_NEG) 1096 if (numtype & IS_NUMBER_NEG)
878 { 1097 {
879 if (uv < (UV)IV_MIN) 1098 if (uv < (UV)IV_MIN)
880 return newSViv (-(IV)uv); 1099 return newSViv (-(IV)uv);
881 } 1100 }
882 else 1101 else
883 return newSVuv (uv); 1102 return newSVuv (uv);
884 } 1103 }
885 }
886 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
887 return newSVnv (Atof (start)); 1121 return newSVnv (Atof (start));
888 1122
889fail: 1123fail:
890 return 0; 1124 return 0;
891} 1125}
… …
921 1155
922 if (*dec->cur != ',') 1156 if (*dec->cur != ',')
923 ERR (", or ] expected while parsing array"); 1157 ERR (", or ] expected while parsing array");
924 1158
925 ++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 }
926 } 1168 }
927 1169
928 DEC_DEC_DEPTH; 1170 DEC_DEC_DEPTH;
929 return newRV_noinc ((SV *)av); 1171 return newRV_noinc ((SV *)av);
930 1172
… …
935} 1177}
936 1178
937static SV * 1179static SV *
938decode_hv (dec_t *dec) 1180decode_hv (dec_t *dec)
939{ 1181{
1182 SV *sv;
940 HV *hv = newHV (); 1183 HV *hv = newHV ();
941 1184
942 DEC_INC_DEPTH; 1185 DEC_INC_DEPTH;
943 decode_ws (dec); 1186 decode_ws (dec);
944 1187
945 if (*dec->cur == '}') 1188 if (*dec->cur == '}')
946 ++dec->cur; 1189 ++dec->cur;
947 else 1190 else
948 for (;;) 1191 for (;;)
949 { 1192 {
1193 EXPECT_CH ('"');
1194
1195 // heuristic: assume that
1196 // a) decode_str + hv_store_ent are abysmally slow.
1197 // b) most hash keys are short, simple ascii text.
1198 // => try to "fast-match" such strings to avoid
1199 // the overhead of decode_str + hv_store_ent.
1200 {
950 SV *key, *value; 1201 SV *value;
1202 char *p = dec->cur;
1203 char *e = p + 24; // only try up to 24 bytes
951 1204
952 decode_ws (dec); EXPECT_CH ('"'); 1205 for (;;)
953
954 key = decode_str (dec);
955 if (!key)
956 goto fail;
957
958 decode_ws (dec); EXPECT_CH (':');
959
960 value = decode_sv (dec);
961 if (!value)
962 { 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);
963 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)
964 goto fail; 1242 goto fail;
1243
1244 hv_store (hv, key, len, value, 0);
1245
1246 break;
1247 }
1248
1249 ++p;
965 } 1250 }
966 1251 }
967 hv_store_ent (hv, key, value, 0);
968 SvREFCNT_dec (key);
969 1252
970 decode_ws (dec); 1253 decode_ws (dec);
971 1254
972 if (*dec->cur == '}') 1255 if (*dec->cur == '}')
973 { 1256 {
… …
977 1260
978 if (*dec->cur != ',') 1261 if (*dec->cur != ',')
979 ERR (", or } expected while parsing object/hash"); 1262 ERR (", or } expected while parsing object/hash");
980 1263
981 ++dec->cur; 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 }
982 } 1273 }
983 1274
984 DEC_DEC_DEPTH; 1275 DEC_DEC_DEPTH;
985 return newRV_noinc ((SV *)hv); 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)
1282 {
1283 HE *cb, *he;
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 }
1312 }
1313
1314 if (dec->json.cb_object)
1315 {
1316 dSP;
1317 int count;
1318
1319 ENTER; SAVETMPS; PUSHMARK (SP);
1320 XPUSHs (sv_2mortal (sv));
1321
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;
986 1337
987fail: 1338fail:
988 SvREFCNT_dec (hv); 1339 SvREFCNT_dec (hv);
989 DEC_DEC_DEPTH; 1340 DEC_DEC_DEPTH;
990 return 0; 1341 return 0;
991} 1342}
992 1343
993static SV * 1344static SV *
994decode_sv (dec_t *dec) 1345decode_sv (dec_t *dec)
995{ 1346{
996 decode_ws (dec); 1347 // the beauty of JSON: you need exactly one character lookahead
1348 // to parse everything.
997 switch (*dec->cur) 1349 switch (*dec->cur)
998 { 1350 {
999 case '"': ++dec->cur; return decode_str (dec); 1351 case '"': ++dec->cur; return decode_str (dec);
1000 case '[': ++dec->cur; return decode_av (dec); 1352 case '[': ++dec->cur; return decode_av (dec);
1001 case '{': ++dec->cur; return decode_hv (dec); 1353 case '{': ++dec->cur; return decode_hv (dec);
1002 1354
1003 case '-': 1355 case '-':
1004 case '0': case '1': case '2': case '3': case '4': 1356 case '0': case '1': case '2': case '3': case '4':
1005 case '5': case '6': case '7': case '8': case '9': 1357 case '5': case '6': case '7': case '8': case '9':
1006 return decode_num (dec); 1358 return decode_num (dec);
1007 1359
1008 case 't': 1360 case 't':
1009 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1361 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1010 { 1362 {
1011 dec->cur += 4; 1363 dec->cur += 4;
1364#if JSON_SLOW
1365 json_true = get_bool ("JSON::XS::true");
1366#endif
1012 return newSViv (1); 1367 return newSVsv (json_true);
1013 } 1368 }
1014 else 1369 else
1015 ERR ("'true' expected"); 1370 ERR ("'true' expected");
1016 1371
1017 break; 1372 break;
1018 1373
1019 case 'f': 1374 case 'f':
1020 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1375 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1021 { 1376 {
1022 dec->cur += 5; 1377 dec->cur += 5;
1378#if JSON_SLOW
1379 json_false = get_bool ("JSON::XS::false");
1380#endif
1023 return newSViv (0); 1381 return newSVsv (json_false);
1024 } 1382 }
1025 else 1383 else
1026 ERR ("'false' expected"); 1384 ERR ("'false' expected");
1027 1385
1028 break; 1386 break;
… …
1046fail: 1404fail:
1047 return 0; 1405 return 0;
1048} 1406}
1049 1407
1050static SV * 1408static SV *
1051decode_json (SV *string, U32 flags, UV *offset_return) 1409decode_json (SV *string, JSON *json, STRLEN *offset_return)
1052{ 1410{
1053 dec_t dec; 1411 dec_t dec;
1054 UV offset; 1412 STRLEN offset;
1055 SV *sv; 1413 SV *sv;
1056 1414
1415 /* work around bugs in 5.10 where manipulating magic values
1416 * will perl ignore the magic in subsequent accesses
1417 */
1057 SvGETMAGIC (string); 1418 /*SvGETMAGIC (string);*/
1419 if (SvMAGICAL (string))
1420 string = sv_2mortal (newSVsv (string));
1421
1058 SvUPGRADE (string, SVt_PV); 1422 SvUPGRADE (string, SVt_PV);
1059 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
1060 if (flags & F_UTF8) 1444 if (json->flags & F_UTF8)
1061 sv_utf8_downgrade (string, 0); 1445 sv_utf8_downgrade (string, 0);
1062 else 1446 else
1063 sv_utf8_upgrade (string); 1447 sv_utf8_upgrade (string);
1064 1448
1065 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1449 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1066 1450
1067 dec.flags = flags; 1451 dec.json = *json;
1068 dec.cur = SvPVX (string); 1452 dec.cur = SvPVX (string);
1069 dec.end = SvEND (string); 1453 dec.end = SvEND (string);
1070 dec.err = 0; 1454 dec.err = 0;
1071 dec.depth = 0; 1455 dec.depth = 0;
1072 dec.maxdepth = DEC_DEPTH (dec.flags); 1456
1457 if (dec.json.cb_object || dec.json.cb_sk_object)
1458 dec.json.flags |= F_HOOK;
1073 1459
1074 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1460 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1461
1462 decode_ws (&dec);
1075 sv = decode_sv (&dec); 1463 sv = decode_sv (&dec);
1076 1464
1077 if (!(offset_return || !sv)) 1465 if (!(offset_return || !sv))
1078 { 1466 {
1079 // check for trailing garbage 1467 // check for trailing garbage
… …
1087 } 1475 }
1088 } 1476 }
1089 1477
1090 if (offset_return || !sv) 1478 if (offset_return || !sv)
1091 { 1479 {
1092 offset = dec.flags & F_UTF8 1480 offset = dec.json.flags & F_UTF8
1093 ? dec.cur - SvPVX (string) 1481 ? dec.cur - SvPVX (string)
1094 : utf8_distance (dec.cur, SvPVX (string)); 1482 : utf8_distance (dec.cur, SvPVX (string));
1095 1483
1096 if (offset_return) 1484 if (offset_return)
1097 *offset_return = offset; 1485 *offset_return = offset;
… …
1116 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1504 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1117 } 1505 }
1118 1506
1119 sv = sv_2mortal (sv); 1507 sv = sv_2mortal (sv);
1120 1508
1121 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1509 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1122 croak ("JSON text must be an object or array (but found number, string, true, false or null, 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)");
1123 1511
1124 return sv; 1512 return sv;
1513}
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
1125} 1630}
1126 1631
1127///////////////////////////////////////////////////////////////////////////// 1632/////////////////////////////////////////////////////////////////////////////
1128// XS interface functions 1633// XS interface functions
1129 1634
1130MODULE = JSON::XS PACKAGE = JSON::XS 1635MODULE = JSON::XS PACKAGE = JSON::XS
1131 1636
1132BOOT: 1637BOOT:
1133{ 1638{
1134 int i; 1639 int i;
1135
1136 memset (decode_hexdigit, 0xff, 256);
1137 1640
1138 for (i = 0; i < 256; ++i) 1641 for (i = 0; i < 256; ++i)
1139 decode_hexdigit [i] = 1642 decode_hexdigit [i] =
1140 i >= '0' && i <= '9' ? i - '0' 1643 i >= '0' && i <= '9' ? i - '0'
1141 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1644 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1142 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1645 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1143 : -1; 1646 : -1;
1144 1647
1145 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");
1146} 1653}
1147 1654
1148PROTOTYPES: DISABLE 1655PROTOTYPES: DISABLE
1149 1656
1150SV *new (char *dummy) 1657void CLONE (...)
1151 CODE: 1658 CODE:
1152 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;
1153 OUTPUT: 1726 OUTPUT:
1154 RETVAL 1727 RETVAL
1155 1728
1156SV *ascii (SV *self, int enable = 1) 1729void max_size (JSON *self, U32 max_size = 0)
1157 ALIAS: 1730 PPCODE:
1158 ascii = F_ASCII 1731 self->max_size = max_size;
1159 latin1 = F_LATIN1 1732 XPUSHs (ST (0));
1160 utf8 = F_UTF8 1733
1161 indent = F_INDENT 1734int get_max_size (JSON *self)
1162 canonical = F_CANONICAL
1163 space_before = F_SPACE_BEFORE
1164 space_after = F_SPACE_AFTER
1165 pretty = F_PRETTY
1166 allow_nonref = F_ALLOW_NONREF
1167 shrink = F_SHRINK
1168 CODE: 1735 CODE:
1169{ 1736 RETVAL = self->max_size;
1170 UV *uv = SvJSON (self);
1171 if (enable)
1172 *uv |= ix;
1173 else
1174 *uv &= ~ix;
1175
1176 RETVAL = newSVsv (self);
1177}
1178 OUTPUT: 1737 OUTPUT:
1179 RETVAL 1738 RETVAL
1180 1739
1181SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1740void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1741 PPCODE:
1742{
1743 SvREFCNT_dec (self->cb_object);
1744 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1745
1746 XPUSHs (ST (0));
1747}
1748
1749void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
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:
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
1182 CODE: 1846 CODE:
1183{ 1847{
1184 UV *uv = SvJSON (self); 1848 if (self->incr_pos)
1185 UV log2 = 0; 1849 croak ("incr_text can not be called when the incremental parser already started parsing");
1186 1850
1187 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1851 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1188
1189 while ((1UL << log2) < max_depth)
1190 ++log2;
1191
1192 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1193
1194 RETVAL = newSVsv (self);
1195} 1852}
1196 OUTPUT: 1853 OUTPUT:
1197 RETVAL 1854 RETVAL
1198 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);
1883
1884PROTOTYPES: ENABLE
1885
1199void encode (SV *self, SV *scalar) 1886void encode_json (SV *scalar)
1887 ALIAS:
1888 to_json_ = 0
1889 encode_json = F_UTF8
1200 PPCODE: 1890 PPCODE:
1891{
1892 JSON json;
1893 json_init (&json);
1894 json.flags |= ix;
1201 XPUSHs (encode_json (scalar, *SvJSON (self))); 1895 XPUSHs (encode_json (scalar, &json));
1896}
1202 1897
1203void decode (SV *self, SV *jsonstr) 1898void decode_json (SV *jsonstr)
1899 ALIAS:
1900 from_json_ = 0
1901 decode_json = F_UTF8
1204 PPCODE: 1902 PPCODE:
1903{
1904 JSON json;
1905 json_init (&json);
1906 json.flags |= ix;
1205 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1907 XPUSHs (decode_json (jsonstr, &json, 0));
1206
1207void decode_prefix (SV *self, SV *jsonstr)
1208 PPCODE:
1209{
1210 UV offset;
1211 EXTEND (SP, 2);
1212 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1213 PUSHs (sv_2mortal (newSVuv (offset)));
1214} 1908}
1215 1909
1216PROTOTYPES: ENABLE
1217 1910
1218void to_json (SV *scalar)
1219 ALIAS:
1220 objToJson = 0
1221 PPCODE:
1222 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1223
1224void from_json (SV *jsonstr)
1225 ALIAS:
1226 jsonToObj = 0
1227 PPCODE:
1228 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1229

Diff Legend

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