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 1.33 by root, Wed May 23 20:26:40 2007 UTC vs.
Revision 1.89 by root, Sat Jul 19 04:21:32 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
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
42static 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}
43 107
44///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
45// utility functions 109// utility functions
46 110
47static UV * 111INLINE SV *
48SvJSON (SV *sv) 112get_bool (const char *name)
49{ 113{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 114 SV *sv = get_sv (name, 1);
51 croak ("object is not of type JSON::XS");
52 115
53 return &SvUVX (SvRV (sv)); 116 SvREADONLY_on (sv);
54} 117 SvREADONLY_on (SvRV (sv));
55 118
56static void 119 return sv;
120}
121
122INLINE void
57shrink (SV *sv) 123shrink (SV *sv)
58{ 124{
59 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
60 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
61 { 128 {
62#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
63 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
64#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
… …
70// 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
71// case of an error. 138// case of an error.
72// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
73// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
74// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
75static UV 142INLINE UV
76decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
77{ 144{
78 if (s[0] > 0xdf || s[0] < 0xc2) 145 if (expect_true (len >= 2
79 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
80 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
81 { 148 {
82 *clen = 2; 149 *clen = 2;
83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
84 } 151 }
85 else 152 else
86 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
87 *clen = (STRLEN)-1; 154}
88 return (UV)-1; 155
89 } 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;
90} 178}
91 179
92///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
93// encoder 181// encoder
94 182
… …
96typedef struct 184typedef struct
97{ 185{
98 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
99 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
100 SV *sv; // result scalar 188 SV *sv; // result scalar
101 U32 flags; // F_* 189 JSON json;
102 U32 indent; // indentation level 190 U32 indent; // indentation level
103 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
104} enc_t; 192} enc_t;
105 193
106static void 194INLINE void
107need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
108{ 196{
109 if (enc->cur + len >= enc->end) 197 if (expect_false (enc->cur + len >= enc->end))
110 { 198 {
111 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - SvPVX (enc->sv);
112 SvGROW (enc->sv, cur + len + 1); 200 SvGROW (enc->sv, cur + len + 1);
113 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
115 } 203 }
116} 204}
117 205
118static void 206INLINE void
119encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
120{ 208{
121 need (enc, 1); 209 need (enc, 1);
122 *enc->cur++ = ch; 210 *enc->cur++ = ch;
123} 211}
… …
131 219
132 while (str < end) 220 while (str < end)
133 { 221 {
134 unsigned char ch = *(unsigned char *)str; 222 unsigned char ch = *(unsigned char *)str;
135 223
136 if (ch >= 0x20 && ch < 0x80) // most common case 224 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
137 { 225 {
138 if (ch == '"') // but with slow exceptions 226 if (expect_false (ch == '"')) // but with slow exceptions
139 { 227 {
140 need (enc, len += 1); 228 need (enc, len += 1);
141 *enc->cur++ = '\\'; 229 *enc->cur++ = '\\';
142 *enc->cur++ = '"'; 230 *enc->cur++ = '"';
143 } 231 }
144 else if (ch == '\\') 232 else if (expect_false (ch == '\\'))
145 { 233 {
146 need (enc, len += 1); 234 need (enc, len += 1);
147 *enc->cur++ = '\\'; 235 *enc->cur++ = '\\';
148 *enc->cur++ = '\\'; 236 *enc->cur++ = '\\';
149 } 237 }
… …
167 STRLEN clen; 255 STRLEN clen;
168 UV uch; 256 UV uch;
169 257
170 if (is_utf8) 258 if (is_utf8)
171 { 259 {
172 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173 uch = decode_utf8 (str, end - str, &clen); 260 uch = decode_utf8 (str, end - str, &clen);
174 if (clen == (STRLEN)-1) 261 if (clen == (STRLEN)-1)
175 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);
176 } 263 }
177 else 264 else
178 { 265 {
179 uch = ch; 266 uch = ch;
180 clen = 1; 267 clen = 1;
181 } 268 }
182 269
183 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
184 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
185
186 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
187 { 271 {
188 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
189 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
190 need (enc, len += 11); 277 need (enc, len += 11);
191 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
192 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
193 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
194 enc->cur += 12; 281 enc->cur += 12;
… …
205 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 292 *enc->cur++ = hexdigit [(uch >> 0) & 15];
206 } 293 }
207 294
208 str += clen; 295 str += clen;
209 } 296 }
210 else if (enc->flags & F_LATIN1) 297 else if (enc->json.flags & F_LATIN1)
211 { 298 {
212 *enc->cur++ = uch; 299 *enc->cur++ = uch;
213 str += clen; 300 str += clen;
214 } 301 }
215 else if (is_utf8) 302 else if (is_utf8)
… …
222 while (--clen); 309 while (--clen);
223 } 310 }
224 else 311 else
225 { 312 {
226 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 313 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
227 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 314 enc->cur = encode_utf8 (enc->cur, uch);
228 ++str; 315 ++str;
229 } 316 }
230 } 317 }
231 } 318 }
232 } 319 }
233 320
234 --len; 321 --len;
235 } 322 }
236} 323}
237 324
238static void 325INLINE void
239encode_indent (enc_t *enc) 326encode_indent (enc_t *enc)
240{ 327{
241 if (enc->flags & F_INDENT) 328 if (enc->json.flags & F_INDENT)
242 { 329 {
243 int spaces = enc->indent * INDENT_STEP; 330 int spaces = enc->indent * INDENT_STEP;
244 331
245 need (enc, spaces); 332 need (enc, spaces);
246 memset (enc->cur, ' ', spaces); 333 memset (enc->cur, ' ', spaces);
247 enc->cur += spaces; 334 enc->cur += spaces;
248 } 335 }
249} 336}
250 337
251static void 338INLINE void
252encode_space (enc_t *enc) 339encode_space (enc_t *enc)
253{ 340{
254 need (enc, 1); 341 need (enc, 1);
255 encode_ch (enc, ' '); 342 encode_ch (enc, ' ');
256} 343}
257 344
258static void 345INLINE void
259encode_nl (enc_t *enc) 346encode_nl (enc_t *enc)
260{ 347{
261 if (enc->flags & F_INDENT) 348 if (enc->json.flags & F_INDENT)
262 { 349 {
263 need (enc, 1); 350 need (enc, 1);
264 encode_ch (enc, '\n'); 351 encode_ch (enc, '\n');
265 } 352 }
266} 353}
267 354
268static void 355INLINE void
269encode_comma (enc_t *enc) 356encode_comma (enc_t *enc)
270{ 357{
271 encode_ch (enc, ','); 358 encode_ch (enc, ',');
272 359
273 if (enc->flags & F_INDENT) 360 if (enc->json.flags & F_INDENT)
274 encode_nl (enc); 361 encode_nl (enc);
275 else if (enc->flags & F_SPACE_AFTER) 362 else if (enc->json.flags & F_SPACE_AFTER)
276 encode_space (enc); 363 encode_space (enc);
277} 364}
278 365
279static void encode_sv (enc_t *enc, SV *sv); 366static void encode_sv (enc_t *enc, SV *sv);
280 367
281static void 368static void
282encode_av (enc_t *enc, AV *av) 369encode_av (enc_t *enc, AV *av)
283{ 370{
284 int i, len = av_len (av); 371 int i, len = av_len (av);
285 372
286 if (enc->indent >= enc->maxdepth) 373 if (enc->indent >= enc->json.max_depth)
287 croak ("data structure too deep (hit recursion limit)"); 374 croak (ERR_NESTING_EXCEEDED);
288 375
289 encode_ch (enc, '['); encode_nl (enc); 376 encode_ch (enc, '[');
290 ++enc->indent; 377
378 if (len >= 0)
379 {
380 encode_nl (enc); ++enc->indent;
291 381
292 for (i = 0; i <= len; ++i) 382 for (i = 0; i <= len; ++i)
293 { 383 {
384 SV **svp = av_fetch (av, i, 0);
385
294 encode_indent (enc); 386 encode_indent (enc);
295 encode_sv (enc, *av_fetch (av, i, 0));
296 387
388 if (svp)
389 encode_sv (enc, *svp);
390 else
391 encode_str (enc, "null", 4, 0);
392
297 if (i < len) 393 if (i < len)
298 encode_comma (enc); 394 encode_comma (enc);
299 } 395 }
300 396
397 encode_nl (enc); --enc->indent; encode_indent (enc);
398 }
399
301 encode_nl (enc); 400 encode_ch (enc, ']');
302
303 --enc->indent;
304 encode_indent (enc); encode_ch (enc, ']');
305} 401}
306 402
307static void 403static void
308encode_he (enc_t *enc, HE *he) 404encode_hk (enc_t *enc, HE *he)
309{ 405{
310 encode_ch (enc, '"'); 406 encode_ch (enc, '"');
311 407
312 if (HeKLEN (he) == HEf_SVKEY) 408 if (HeKLEN (he) == HEf_SVKEY)
313 { 409 {
… …
323 else 419 else
324 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 420 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
325 421
326 encode_ch (enc, '"'); 422 encode_ch (enc, '"');
327 423
328 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 424 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
329 encode_ch (enc, ':'); 425 encode_ch (enc, ':');
330 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 426 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
331 encode_sv (enc, HeVAL (he));
332} 427}
333 428
334// compare hash entries, used when all keys are bytestrings 429// compare hash entries, used when all keys are bytestrings
335static int 430static int
336he_cmp_fast (const void *a_, const void *b_) 431he_cmp_fast (const void *a_, const void *b_)
… …
341 HE *b = *(HE **)b_; 436 HE *b = *(HE **)b_;
342 437
343 STRLEN la = HeKLEN (a); 438 STRLEN la = HeKLEN (a);
344 STRLEN lb = HeKLEN (b); 439 STRLEN lb = HeKLEN (b);
345 440
346 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 441 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
347 cmp = la - lb; 442 cmp = lb - la;
348 443
349 return cmp; 444 return cmp;
350} 445}
351 446
352// compare hash entries, used when some keys are sv's or utf-x 447// compare hash entries, used when some keys are sv's or utf-x
353static int 448static int
354he_cmp_slow (const void *a, const void *b) 449he_cmp_slow (const void *a, const void *b)
355{ 450{
356 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 451 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
357} 452}
358 453
359static void 454static void
360encode_hv (enc_t *enc, HV *hv) 455encode_hv (enc_t *enc, HV *hv)
361{ 456{
362 int count, i; 457 HE *he;
363 458
364 if (enc->indent >= enc->maxdepth) 459 if (enc->indent >= enc->json.max_depth)
365 croak ("data structure too deep (hit recursion limit)"); 460 croak (ERR_NESTING_EXCEEDED);
366 461
367 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 462 encode_ch (enc, '{');
368 463
369 if ((count = hv_iterinit (hv)))
370 {
371 // for canonical output we have to sort by keys first 464 // for canonical output we have to sort by keys first
372 // actually, this is mostly due to the stupid so-called 465 // actually, this is mostly due to the stupid so-called
373 // security workaround added somewhere in 5.8.x. 466 // security workaround added somewhere in 5.8.x.
374 // that randomises hash orderings 467 // that randomises hash orderings
375 if (enc->flags & F_CANONICAL) 468 if (enc->json.flags & F_CANONICAL)
469 {
470 int count = hv_iterinit (hv);
471
472 if (SvMAGICAL (hv))
376 { 473 {
474 // need to count by iterating. could improve by dynamically building the vector below
475 // but I don't care for the speed of this special case.
476 // note also that we will run into undefined behaviour when the two iterations
477 // do not result in the same count, something I might care for in some later release.
478
479 count = 0;
480 while (hv_iternext (hv))
481 ++count;
482
483 hv_iterinit (hv);
484 }
485
486 if (count)
487 {
377 int fast = 1; 488 int i, fast = 1;
378 HE *he; 489#if defined(__BORLANDC__) || defined(_MSC_VER)
379#if WIN32
380 HE **hes = _alloca (count * sizeof (HE)); 490 HE **hes = _alloca (count * sizeof (HE));
381#else 491#else
382 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 492 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
383#endif 493#endif
384 494
… …
410 520
411 FREETMPS; 521 FREETMPS;
412 LEAVE; 522 LEAVE;
413 } 523 }
414 524
415 for (i = 0; i < count; ++i) 525 encode_nl (enc); ++enc->indent;
526
527 while (count--)
416 { 528 {
417 encode_indent (enc); 529 encode_indent (enc);
530 he = hes [count];
418 encode_he (enc, hes [i]); 531 encode_hk (enc, he);
532 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
419 533
420 if (i < count - 1) 534 if (count)
421 encode_comma (enc); 535 encode_comma (enc);
422 } 536 }
423 537
538 encode_nl (enc); --enc->indent; encode_indent (enc);
539 }
540 }
541 else
542 {
543 if (hv_iterinit (hv) || SvMAGICAL (hv))
544 if ((he = hv_iternext (hv)))
545 {
546 encode_nl (enc); ++enc->indent;
547
548 for (;;)
549 {
424 encode_nl (enc); 550 encode_indent (enc);
551 encode_hk (enc, he);
552 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
553
554 if (!(he = hv_iternext (hv)))
555 break;
556
557 encode_comma (enc);
558 }
559
560 encode_nl (enc); --enc->indent; encode_indent (enc);
561 }
562 }
563
564 encode_ch (enc, '}');
565}
566
567// encode objects, arrays and special \0=false and \1=true values.
568static void
569encode_rv (enc_t *enc, SV *sv)
570{
571 svtype svt;
572
573 SvGETMAGIC (sv);
574 svt = SvTYPE (sv);
575
576 if (expect_false (SvOBJECT (sv)))
577 {
578 HV *stash = !JSON_SLOW || json_boolean_stash
579 ? json_boolean_stash
580 : gv_stashpv ("JSON::XS::Boolean", 1);
581
582 if (SvSTASH (sv) == stash)
583 {
584 if (SvIV (sv))
585 encode_str (enc, "true", 4, 0);
586 else
587 encode_str (enc, "false", 5, 0);
425 } 588 }
426 else 589 else
427 { 590 {
428 HE *he = hv_iternext (hv); 591#if 0
429 592 if (0 && sv_derived_from (rv, "JSON::Literal"))
430 for (;;)
431 { 593 {
432 encode_indent (enc); 594 // not yet
433 encode_he (enc, he);
434
435 if (!(he = hv_iternext (hv)))
436 break;
437
438 encode_comma (enc);
439 } 595 }
596#endif
597 if (enc->json.flags & F_CONV_BLESSED)
598 {
599 // we re-bless the reference to get overload and other niceties right
600 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
440 601
602 if (to_json)
603 {
604 dSP;
605
606 ENTER; SAVETMPS; PUSHMARK (SP);
607 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
608
609 // calling with G_SCALAR ensures that we always get a 1 return value
610 PUTBACK;
611 call_sv ((SV *)GvCV (to_json), G_SCALAR);
612 SPAGAIN;
613
614 // catch this surprisingly common error
615 if (SvROK (TOPs) && SvRV (TOPs) == sv)
616 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
617
618 sv = POPs;
619 PUTBACK;
620
441 encode_nl (enc); 621 encode_sv (enc, sv);
622
623 FREETMPS; LEAVE;
624 }
625 else if (enc->json.flags & F_ALLOW_BLESSED)
626 encode_str (enc, "null", 4, 0);
627 else
628 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
629 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
630 }
631 else if (enc->json.flags & F_ALLOW_BLESSED)
632 encode_str (enc, "null", 4, 0);
633 else
634 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
635 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
442 } 636 }
443 } 637 }
444
445 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
446}
447
448// encode objects, arrays and special \0=false and \1=true values.
449static void
450encode_rv (enc_t *enc, SV *sv)
451{
452 svtype svt;
453
454 SvGETMAGIC (sv);
455 svt = SvTYPE (sv);
456
457 if (svt == SVt_PVHV) 638 else if (svt == SVt_PVHV)
458 encode_hv (enc, (HV *)sv); 639 encode_hv (enc, (HV *)sv);
459 else if (svt == SVt_PVAV) 640 else if (svt == SVt_PVAV)
460 encode_av (enc, (AV *)sv); 641 encode_av (enc, (AV *)sv);
461 else if (svt < SVt_PVAV) 642 else if (svt < SVt_PVAV)
462 { 643 {
463 if (SvNIOK (sv) && SvIV (sv) == 0) 644 STRLEN len = 0;
645 char *pv = svt ? SvPV (sv, len) : 0;
646
647 if (len == 1 && *pv == '1')
648 encode_str (enc, "true", 4, 0);
649 else if (len == 1 && *pv == '0')
464 encode_str (enc, "false", 5, 0); 650 encode_str (enc, "false", 5, 0);
465 else if (SvNIOK (sv) && SvIV (sv) == 1) 651 else if (enc->json.flags & F_ALLOW_UNKNOWN)
466 encode_str (enc, "true", 4, 0); 652 encode_str (enc, "null", 4, 0);
467 else 653 else
468 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 654 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
469 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 655 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
470 } 656 }
657 else if (enc->json.flags & F_ALLOW_UNKNOWN)
658 encode_str (enc, "null", 4, 0);
471 else 659 else
472 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 660 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
473 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 661 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
474} 662}
475 663
… …
486 encode_str (enc, str, len, SvUTF8 (sv)); 674 encode_str (enc, str, len, SvUTF8 (sv));
487 encode_ch (enc, '"'); 675 encode_ch (enc, '"');
488 } 676 }
489 else if (SvNOKp (sv)) 677 else if (SvNOKp (sv))
490 { 678 {
679 // trust that perl will do the right thing w.r.t. JSON syntax.
491 need (enc, NV_DIG + 32); 680 need (enc, NV_DIG + 32);
492 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 681 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
493 enc->cur += strlen (enc->cur); 682 enc->cur += strlen (enc->cur);
494 } 683 }
495 else if (SvIOKp (sv)) 684 else if (SvIOKp (sv))
496 { 685 {
686 // we assume we can always read an IV as a UV and vice versa
687 // we assume two's complement
688 // we assume no aliasing issues in the union
689 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
690 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
691 {
692 // optimise the "small number case"
693 // code will likely be branchless and use only a single multiplication
694 // works for numbers up to 59074
695 I32 i = SvIVX (sv);
696 U32 u;
697 char digit, nz = 0;
698
497 need (enc, 64); 699 need (enc, 6);
700
701 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
702 u = i < 0 ? -i : i;
703
704 // convert to 4.28 fixed-point representation
705 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
706
707 // now output digit by digit, each time masking out the integer part
708 // and multiplying by 5 while moving the decimal point one to the right,
709 // resulting in a net multiplication by 10.
710 // we always write the digit to memory but conditionally increment
711 // the pointer, to enable the use of conditional move instructions.
712 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
713 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
714 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
715 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
716 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
717 }
718 else
719 {
720 // large integer, use the (rather slow) snprintf way.
721 need (enc, IVUV_MAXCHARS);
498 enc->cur += 722 enc->cur +=
499 SvIsUV(sv) 723 SvIsUV(sv)
500 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 724 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
501 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 725 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
726 }
502 } 727 }
503 else if (SvROK (sv)) 728 else if (SvROK (sv))
504 encode_rv (enc, SvRV (sv)); 729 encode_rv (enc, SvRV (sv));
505 else if (!SvOK (sv)) 730 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
506 encode_str (enc, "null", 4, 0); 731 encode_str (enc, "null", 4, 0);
507 else 732 else
508 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 733 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
509 SvPV_nolen (sv), SvFLAGS (sv)); 734 SvPV_nolen (sv), SvFLAGS (sv));
510} 735}
511 736
512static SV * 737static SV *
513encode_json (SV *scalar, U32 flags) 738encode_json (SV *scalar, JSON *json)
514{ 739{
515 enc_t enc; 740 enc_t enc;
516 741
517 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 742 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
518 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 743 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
519 744
520 enc.flags = flags; 745 enc.json = *json;
521 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
522 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
523 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
524 enc.indent = 0; 749 enc.indent = 0;
525 enc.maxdepth = DEC_DEPTH (flags); 750 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
751 : enc.json.flags & F_LATIN1 ? 0x000100UL
752 : 0x110000UL;
526 753
527 SvPOK_only (enc.sv); 754 SvPOK_only (enc.sv);
528 encode_sv (&enc, scalar); 755 encode_sv (&enc, scalar);
529 756
530 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 757 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
531 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 758 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
532 759
533 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 760 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
534 SvUTF8_on (enc.sv); 761 SvUTF8_on (enc.sv);
535 762
536 if (enc.flags & F_SHRINK) 763 if (enc.json.flags & F_SHRINK)
537 shrink (enc.sv); 764 shrink (enc.sv);
538 765
539 return enc.sv; 766 return enc.sv;
540} 767}
541 768
… …
546typedef struct 773typedef struct
547{ 774{
548 char *cur; // current parser pointer 775 char *cur; // current parser pointer
549 char *end; // end of input string 776 char *end; // end of input string
550 const char *err; // parse error, if != 0 777 const char *err; // parse error, if != 0
551 U32 flags; // F_* 778 JSON json;
552 U32 depth; // recursion depth 779 U32 depth; // recursion depth
553 U32 maxdepth; // recursion depth limit 780 U32 maxdepth; // recursion depth limit
554} dec_t; 781} dec_t;
555 782
556static void 783INLINE void
784decode_comment (dec_t *dec)
785{
786 // only '#'-style comments allowed a.t.m.
787
788 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
789 ++dec->cur;
790}
791
792INLINE void
557decode_ws (dec_t *dec) 793decode_ws (dec_t *dec)
558{ 794{
559 for (;;) 795 for (;;)
560 { 796 {
561 char ch = *dec->cur; 797 char ch = *dec->cur;
562 798
563 if (ch > 0x20 799 if (ch > 0x20)
800 {
801 if (expect_false (ch == '#'))
802 {
803 if (dec->json.flags & F_RELAXED)
804 decode_comment (dec);
805 else
806 break;
807 }
808 else
809 break;
810 }
564 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 811 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
565 break; 812 break; // parse error, but let higher level handle it, gives better error messages
566 813
567 ++dec->cur; 814 ++dec->cur;
568 } 815 }
569} 816}
570 817
… …
574 if (*dec->cur != ch) \ 821 if (*dec->cur != ch) \
575 ERR (# ch " expected"); \ 822 ERR (# ch " expected"); \
576 ++dec->cur; \ 823 ++dec->cur; \
577 SE 824 SE
578 825
579#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 826#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
580#define DEC_DEC_DEPTH --dec->depth 827#define DEC_DEC_DEPTH --dec->depth
581 828
582static SV *decode_sv (dec_t *dec); 829static SV *decode_sv (dec_t *dec);
583 830
584static signed char decode_hexdigit[256]; 831static signed char decode_hexdigit[256];
… …
587decode_4hex (dec_t *dec) 834decode_4hex (dec_t *dec)
588{ 835{
589 signed char d1, d2, d3, d4; 836 signed char d1, d2, d3, d4;
590 unsigned char *cur = (unsigned char *)dec->cur; 837 unsigned char *cur = (unsigned char *)dec->cur;
591 838
592 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 839 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
593 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 840 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
594 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 841 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
595 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 842 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
596 843
597 dec->cur += 4; 844 dec->cur += 4;
598 845
599 return ((UV)d1) << 12 846 return ((UV)d1) << 12
600 | ((UV)d2) << 8 847 | ((UV)d2) << 8
… …
608static SV * 855static SV *
609decode_str (dec_t *dec) 856decode_str (dec_t *dec)
610{ 857{
611 SV *sv = 0; 858 SV *sv = 0;
612 int utf8 = 0; 859 int utf8 = 0;
860 char *dec_cur = dec->cur;
613 861
614 do 862 do
615 { 863 {
616 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 864 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
617 char *cur = buf; 865 char *cur = buf;
618 866
619 do 867 do
620 { 868 {
621 unsigned char ch = *(unsigned char *)dec->cur++; 869 unsigned char ch = *(unsigned char *)dec_cur++;
622 870
623 if (ch == '"') 871 if (expect_false (ch == '"'))
624 { 872 {
625 --dec->cur; 873 --dec_cur;
626 break; 874 break;
627 } 875 }
628 else if (ch == '\\') 876 else if (expect_false (ch == '\\'))
629 { 877 {
630 switch (*dec->cur) 878 switch (*dec_cur)
631 { 879 {
632 case '\\': 880 case '\\':
633 case '/': 881 case '/':
634 case '"': *cur++ = *dec->cur++; break; 882 case '"': *cur++ = *dec_cur++; break;
635 883
636 case 'b': ++dec->cur; *cur++ = '\010'; break; 884 case 'b': ++dec_cur; *cur++ = '\010'; break;
637 case 't': ++dec->cur; *cur++ = '\011'; break; 885 case 't': ++dec_cur; *cur++ = '\011'; break;
638 case 'n': ++dec->cur; *cur++ = '\012'; break; 886 case 'n': ++dec_cur; *cur++ = '\012'; break;
639 case 'f': ++dec->cur; *cur++ = '\014'; break; 887 case 'f': ++dec_cur; *cur++ = '\014'; break;
640 case 'r': ++dec->cur; *cur++ = '\015'; break; 888 case 'r': ++dec_cur; *cur++ = '\015'; break;
641 889
642 case 'u': 890 case 'u':
643 { 891 {
644 UV lo, hi; 892 UV lo, hi;
645 ++dec->cur; 893 ++dec_cur;
646 894
895 dec->cur = dec_cur;
647 hi = decode_4hex (dec); 896 hi = decode_4hex (dec);
897 dec_cur = dec->cur;
648 if (hi == (UV)-1) 898 if (hi == (UV)-1)
649 goto fail; 899 goto fail;
650 900
651 // possibly a surrogate pair 901 // possibly a surrogate pair
652 if (hi >= 0xd800) 902 if (hi >= 0xd800)
653 if (hi < 0xdc00) 903 if (hi < 0xdc00)
654 { 904 {
655 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 905 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
656 ERR ("missing low surrogate character in surrogate pair"); 906 ERR ("missing low surrogate character in surrogate pair");
657 907
658 dec->cur += 2; 908 dec_cur += 2;
659 909
910 dec->cur = dec_cur;
660 lo = decode_4hex (dec); 911 lo = decode_4hex (dec);
912 dec_cur = dec->cur;
661 if (lo == (UV)-1) 913 if (lo == (UV)-1)
662 goto fail; 914 goto fail;
663 915
664 if (lo < 0xdc00 || lo >= 0xe000) 916 if (lo < 0xdc00 || lo >= 0xe000)
665 ERR ("surrogate pair expected"); 917 ERR ("surrogate pair expected");
… …
671 923
672 if (hi >= 0x80) 924 if (hi >= 0x80)
673 { 925 {
674 utf8 = 1; 926 utf8 = 1;
675 927
676 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 928 cur = encode_utf8 (cur, hi);
677 } 929 }
678 else 930 else
679 *cur++ = hi; 931 *cur++ = hi;
680 } 932 }
681 break; 933 break;
682 934
683 default: 935 default:
684 --dec->cur; 936 --dec_cur;
685 ERR ("illegal backslash escape sequence in string"); 937 ERR ("illegal backslash escape sequence in string");
686 } 938 }
687 } 939 }
688 else if (ch >= 0x20 && ch <= 0x7f) 940 else if (expect_true (ch >= 0x20 && ch < 0x80))
689 *cur++ = ch; 941 *cur++ = ch;
690 else if (ch >= 0x80) 942 else if (ch >= 0x80)
691 { 943 {
692 STRLEN clen; 944 STRLEN clen;
693 UV uch; 945 UV uch;
694 946
695 --dec->cur; 947 --dec_cur;
696 948
697 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 949 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
698 if (clen == (STRLEN)-1) 950 if (clen == (STRLEN)-1)
699 ERR ("malformed UTF-8 character in JSON string"); 951 ERR ("malformed UTF-8 character in JSON string");
700 952
701 do 953 do
702 *cur++ = *dec->cur++; 954 *cur++ = *dec_cur++;
703 while (--clen); 955 while (--clen);
704 956
705 utf8 = 1; 957 utf8 = 1;
706 } 958 }
707 else 959 else
708 { 960 {
709 --dec->cur; 961 --dec_cur;
710 962
711 if (!ch) 963 if (!ch)
712 ERR ("unexpected end of string while parsing JSON string"); 964 ERR ("unexpected end of string while parsing JSON string");
713 else 965 else
714 ERR ("invalid character encountered while parsing JSON string"); 966 ERR ("invalid character encountered while parsing JSON string");
… …
727 } 979 }
728 else 980 else
729 sv = newSVpvn (buf, len); 981 sv = newSVpvn (buf, len);
730 } 982 }
731 } 983 }
732 while (*dec->cur != '"'); 984 while (*dec_cur != '"');
733 985
734 ++dec->cur; 986 ++dec_cur;
735 987
736 if (sv) 988 if (sv)
737 { 989 {
738 SvPOK_only (sv); 990 SvPOK_only (sv);
739 *SvEND (sv) = 0; 991 *SvEND (sv) = 0;
… …
742 SvUTF8_on (sv); 994 SvUTF8_on (sv);
743 } 995 }
744 else 996 else
745 sv = newSVpvn ("", 0); 997 sv = newSVpvn ("", 0);
746 998
999 dec->cur = dec_cur;
747 return sv; 1000 return sv;
748 1001
749fail: 1002fail:
1003 dec->cur = dec_cur;
750 return 0; 1004 return 0;
751} 1005}
752 1006
753static SV * 1007static SV *
754decode_num (dec_t *dec) 1008decode_num (dec_t *dec)
… …
812 is_nv = 1; 1066 is_nv = 1;
813 } 1067 }
814 1068
815 if (!is_nv) 1069 if (!is_nv)
816 { 1070 {
817 UV uv; 1071 int len = dec->cur - start;
818 int numtype = grok_number (start, dec->cur - start, &uv); 1072
819 if (numtype & IS_NUMBER_IN_UV) 1073 // special case the rather common 1..5-digit-int case
820 if (numtype & IS_NUMBER_NEG) 1074 if (*start == '-')
1075 switch (len)
821 { 1076 {
822 if (uv < (UV)IV_MIN) 1077 case 2: return newSViv (-( start [1] - '0' * 1));
823 return newSViv (-(IV)uv); 1078 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1079 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1080 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1081 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
824 } 1082 }
1083 else
1084 switch (len)
1085 {
1086 case 1: return newSViv ( start [0] - '0' * 1);
1087 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1088 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1089 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1090 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1091 }
1092
1093 {
1094 UV uv;
1095 int numtype = grok_number (start, len, &uv);
1096 if (numtype & IS_NUMBER_IN_UV)
1097 if (numtype & IS_NUMBER_NEG)
1098 {
1099 if (uv < (UV)IV_MIN)
1100 return newSViv (-(IV)uv);
1101 }
825 else 1102 else
826 return newSVuv (uv); 1103 return newSVuv (uv);
827 } 1104 }
828 1105
1106 len -= *start == '-' ? 1 : 0;
1107
1108 // does not fit into IV or UV, try NV
1109 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1110 #if defined (LDBL_DIG)
1111 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1112 #endif
1113 )
1114 // fits into NV without loss of precision
1115 return newSVnv (Atof (start));
1116
1117 // everything else fails, convert it to a string
1118 return newSVpvn (start, dec->cur - start);
1119 }
1120
1121 // loss of precision here
829 return newSVnv (Atof (start)); 1122 return newSVnv (Atof (start));
830 1123
831fail: 1124fail:
832 return 0; 1125 return 0;
833} 1126}
… …
863 1156
864 if (*dec->cur != ',') 1157 if (*dec->cur != ',')
865 ERR (", or ] expected while parsing array"); 1158 ERR (", or ] expected while parsing array");
866 1159
867 ++dec->cur; 1160 ++dec->cur;
1161
1162 decode_ws (dec);
1163
1164 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1165 {
1166 ++dec->cur;
1167 break;
1168 }
868 } 1169 }
869 1170
870 DEC_DEC_DEPTH; 1171 DEC_DEC_DEPTH;
871 return newRV_noinc ((SV *)av); 1172 return newRV_noinc ((SV *)av);
872 1173
… …
877} 1178}
878 1179
879static SV * 1180static SV *
880decode_hv (dec_t *dec) 1181decode_hv (dec_t *dec)
881{ 1182{
1183 SV *sv;
882 HV *hv = newHV (); 1184 HV *hv = newHV ();
883 1185
884 DEC_INC_DEPTH; 1186 DEC_INC_DEPTH;
885 decode_ws (dec); 1187 decode_ws (dec);
886 1188
887 if (*dec->cur == '}') 1189 if (*dec->cur == '}')
888 ++dec->cur; 1190 ++dec->cur;
889 else 1191 else
890 for (;;) 1192 for (;;)
891 { 1193 {
1194 EXPECT_CH ('"');
1195
1196 // heuristic: assume that
1197 // a) decode_str + hv_store_ent are abysmally slow.
1198 // b) most hash keys are short, simple ascii text.
1199 // => try to "fast-match" such strings to avoid
1200 // the overhead of decode_str + hv_store_ent.
1201 {
892 SV *key, *value; 1202 SV *value;
1203 char *p = dec->cur;
1204 char *e = p + 24; // only try up to 24 bytes
893 1205
894 decode_ws (dec); EXPECT_CH ('"'); 1206 for (;;)
895
896 key = decode_str (dec);
897 if (!key)
898 goto fail;
899
900 decode_ws (dec); EXPECT_CH (':');
901
902 value = decode_sv (dec);
903 if (!value)
904 { 1207 {
1208 // the >= 0x80 is false on most architectures
1209 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1210 {
1211 // slow path, back up and use decode_str
1212 SV *key = decode_str (dec);
1213 if (!key)
1214 goto fail;
1215
1216 decode_ws (dec); EXPECT_CH (':');
1217
1218 decode_ws (dec);
1219 value = decode_sv (dec);
1220 if (!value)
1221 {
1222 SvREFCNT_dec (key);
1223 goto fail;
1224 }
1225
1226 hv_store_ent (hv, key, value, 0);
905 SvREFCNT_dec (key); 1227 SvREFCNT_dec (key);
1228
1229 break;
1230 }
1231 else if (*p == '"')
1232 {
1233 // fast path, got a simple key
1234 char *key = dec->cur;
1235 int len = p - key;
1236 dec->cur = p + 1;
1237
1238 decode_ws (dec); EXPECT_CH (':');
1239
1240 decode_ws (dec);
1241 value = decode_sv (dec);
1242 if (!value)
906 goto fail; 1243 goto fail;
1244
1245 hv_store (hv, key, len, value, 0);
1246
1247 break;
1248 }
1249
1250 ++p;
907 } 1251 }
908 1252 }
909 hv_store_ent (hv, key, value, 0);
910 SvREFCNT_dec (key);
911 1253
912 decode_ws (dec); 1254 decode_ws (dec);
913 1255
914 if (*dec->cur == '}') 1256 if (*dec->cur == '}')
915 { 1257 {
… …
919 1261
920 if (*dec->cur != ',') 1262 if (*dec->cur != ',')
921 ERR (", or } expected while parsing object/hash"); 1263 ERR (", or } expected while parsing object/hash");
922 1264
923 ++dec->cur; 1265 ++dec->cur;
1266
1267 decode_ws (dec);
1268
1269 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1270 {
1271 ++dec->cur;
1272 break;
1273 }
924 } 1274 }
925 1275
926 DEC_DEC_DEPTH; 1276 DEC_DEC_DEPTH;
927 return newRV_noinc ((SV *)hv); 1277 sv = newRV_noinc ((SV *)hv);
1278
1279 // check filter callbacks
1280 if (dec->json.flags & F_HOOK)
1281 {
1282 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1283 {
1284 HE *cb, *he;
1285
1286 hv_iterinit (hv);
1287 he = hv_iternext (hv);
1288 hv_iterinit (hv);
1289
1290 // the next line creates a mortal sv each time its called.
1291 // might want to optimise this for common cases.
1292 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1293
1294 if (cb)
1295 {
1296 dSP;
1297 int count;
1298
1299 ENTER; SAVETMPS; PUSHMARK (SP);
1300 XPUSHs (HeVAL (he));
1301
1302 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1303
1304 if (count == 1)
1305 {
1306 sv = newSVsv (POPs);
1307 FREETMPS; LEAVE;
1308 return sv;
1309 }
1310
1311 FREETMPS; LEAVE;
1312 }
1313 }
1314
1315 if (dec->json.cb_object)
1316 {
1317 dSP;
1318 int count;
1319
1320 ENTER; SAVETMPS; PUSHMARK (SP);
1321 XPUSHs (sv_2mortal (sv));
1322
1323 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1324
1325 if (count == 1)
1326 {
1327 sv = newSVsv (POPs);
1328 FREETMPS; LEAVE;
1329 return sv;
1330 }
1331
1332 SvREFCNT_inc (sv);
1333 FREETMPS; LEAVE;
1334 }
1335 }
1336
1337 return sv;
928 1338
929fail: 1339fail:
930 SvREFCNT_dec (hv); 1340 SvREFCNT_dec (hv);
931 DEC_DEC_DEPTH; 1341 DEC_DEC_DEPTH;
932 return 0; 1342 return 0;
933} 1343}
934 1344
935static SV * 1345static SV *
936decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
937{ 1347{
938 decode_ws (dec); 1348 // the beauty of JSON: you need exactly one character lookahead
1349 // to parse everything.
939 switch (*dec->cur) 1350 switch (*dec->cur)
940 { 1351 {
941 case '"': ++dec->cur; return decode_str (dec); 1352 case '"': ++dec->cur; return decode_str (dec);
942 case '[': ++dec->cur; return decode_av (dec); 1353 case '[': ++dec->cur; return decode_av (dec);
943 case '{': ++dec->cur; return decode_hv (dec); 1354 case '{': ++dec->cur; return decode_hv (dec);
944 1355
945 case '-': 1356 case '-':
946 case '0': case '1': case '2': case '3': case '4': 1357 case '0': case '1': case '2': case '3': case '4':
947 case '5': case '6': case '7': case '8': case '9': 1358 case '5': case '6': case '7': case '8': case '9':
948 return decode_num (dec); 1359 return decode_num (dec);
949 1360
950 case 't': 1361 case 't':
951 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1362 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
952 { 1363 {
953 dec->cur += 4; 1364 dec->cur += 4;
1365#if JSON_SLOW
1366 json_true = get_bool ("JSON::XS::true");
1367#endif
954 return newSViv (1); 1368 return newSVsv (json_true);
955 } 1369 }
956 else 1370 else
957 ERR ("'true' expected"); 1371 ERR ("'true' expected");
958 1372
959 break; 1373 break;
960 1374
961 case 'f': 1375 case 'f':
962 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1376 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
963 { 1377 {
964 dec->cur += 5; 1378 dec->cur += 5;
1379#if JSON_SLOW
1380 json_false = get_bool ("JSON::XS::false");
1381#endif
965 return newSViv (0); 1382 return newSVsv (json_false);
966 } 1383 }
967 else 1384 else
968 ERR ("'false' expected"); 1385 ERR ("'false' expected");
969 1386
970 break; 1387 break;
… …
988fail: 1405fail:
989 return 0; 1406 return 0;
990} 1407}
991 1408
992static SV * 1409static SV *
993decode_json (SV *string, U32 flags, UV *offset_return) 1410decode_json (SV *string, JSON *json, STRLEN *offset_return)
994{ 1411{
995 dec_t dec; 1412 dec_t dec;
996 UV offset; 1413 STRLEN offset;
997 SV *sv; 1414 SV *sv;
998 1415
999 SvGETMAGIC (string); 1416 SvGETMAGIC (string);
1000 SvUPGRADE (string, SVt_PV); 1417 SvUPGRADE (string, SVt_PV);
1001 1418
1419 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1420 * assertion with -DDEBUGGING, although SvCUR is documented to
1421 * return the xpv_cur field which certainly exists after upgrading.
1422 * according to nicholas clark, calling SvPOK fixes this.
1423 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1424 * and hope for the best.
1425 */
1426#ifdef DEBUGGING
1427 SvPV_nolen (string);
1428#endif
1429
1430 if (SvCUR (string) > json->max_size && json->max_size)
1431 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1432 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1433
1002 if (flags & F_UTF8) 1434 if (json->flags & F_UTF8)
1003 sv_utf8_downgrade (string, 0); 1435 sv_utf8_downgrade (string, 0);
1004 else 1436 else
1005 sv_utf8_upgrade (string); 1437 sv_utf8_upgrade (string);
1006 1438
1007 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1439 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1008 1440
1009 dec.flags = flags; 1441 dec.json = *json;
1010 dec.cur = SvPVX (string); 1442 dec.cur = SvPVX (string);
1011 dec.end = SvEND (string); 1443 dec.end = SvEND (string);
1012 dec.err = 0; 1444 dec.err = 0;
1013 dec.depth = 0; 1445 dec.depth = 0;
1014 dec.maxdepth = DEC_DEPTH (dec.flags); 1446
1447 if (dec.json.cb_object || dec.json.cb_sk_object)
1448 dec.json.flags |= F_HOOK;
1015 1449
1016 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1450 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1451
1452 decode_ws (&dec);
1017 sv = decode_sv (&dec); 1453 sv = decode_sv (&dec);
1018 1454
1019 if (!(offset_return || !sv)) 1455 if (!(offset_return || !sv))
1020 { 1456 {
1021 // check for trailing garbage 1457 // check for trailing garbage
… …
1029 } 1465 }
1030 } 1466 }
1031 1467
1032 if (offset_return || !sv) 1468 if (offset_return || !sv)
1033 { 1469 {
1034 offset = dec.flags & F_UTF8 1470 offset = dec.json.flags & F_UTF8
1035 ? dec.cur - SvPVX (string) 1471 ? dec.cur - SvPVX (string)
1036 : utf8_distance (dec.cur, SvPVX (string)); 1472 : utf8_distance (dec.cur, SvPVX (string));
1037 1473
1038 if (offset_return) 1474 if (offset_return)
1039 *offset_return = offset; 1475 *offset_return = offset;
… …
1058 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1494 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1059 } 1495 }
1060 1496
1061 sv = sv_2mortal (sv); 1497 sv = sv_2mortal (sv);
1062 1498
1063 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1499 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1064 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1500 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1065 1501
1066 return sv; 1502 return sv;
1503}
1504
1505/////////////////////////////////////////////////////////////////////////////
1506// incremental parser
1507
1508static void
1509incr_parse (JSON *self)
1510{
1511 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1512
1513 for (;;)
1514 {
1515 //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
1516 switch (self->incr_mode)
1517 {
1518 // only used for intiial whitespace skipping
1519 case INCR_M_WS:
1520 for (;;)
1521 {
1522 if (*p > 0x20)
1523 {
1524 self->incr_mode = INCR_M_JSON;
1525 goto incr_m_json;
1526 }
1527 else if (!*p)
1528 goto interrupt;
1529
1530 ++p;
1531 }
1532
1533 // skip a single char inside a string (for \\-processing)
1534 case INCR_M_BS:
1535 if (!*p)
1536 goto interrupt;
1537
1538 ++p;
1539 self->incr_mode = INCR_M_STR;
1540 goto incr_m_str;
1541
1542 // inside a string
1543 case INCR_M_STR:
1544 incr_m_str:
1545 for (;;)
1546 {
1547 if (*p == '"')
1548 {
1549 ++p;
1550 self->incr_mode = INCR_M_JSON;
1551
1552 if (!self->incr_nest)
1553 goto interrupt;
1554
1555 goto incr_m_json;
1556 }
1557 else if (*p == '\\')
1558 {
1559 ++p; // "virtually" consumes character after \
1560
1561 if (!*p) // if at end of string we have to switch modes
1562 {
1563 self->incr_mode = INCR_M_BS;
1564 goto interrupt;
1565 }
1566 }
1567 else if (!*p)
1568 goto interrupt;
1569
1570 ++p;
1571 }
1572
1573 // after initial ws, outside string
1574 case INCR_M_JSON:
1575 incr_m_json:
1576 for (;;)
1577 {
1578 switch (*p++)
1579 {
1580 case 0:
1581 --p;
1582 goto interrupt;
1583
1584 case 0x09:
1585 case 0x0a:
1586 case 0x0d:
1587 case 0x20:
1588 if (!self->incr_nest)
1589 {
1590 --p; // do not eat the whitespace, let the next round do it
1591 goto interrupt;
1592 }
1593 break;
1594
1595 case '"':
1596 self->incr_mode = INCR_M_STR;
1597 goto incr_m_str;
1598
1599 case '[':
1600 case '{':
1601 if (++self->incr_nest > self->max_depth)
1602 croak (ERR_NESTING_EXCEEDED);
1603 break;
1604
1605 case ']':
1606 case '}':
1607 if (--self->incr_nest <= 0)
1608 goto interrupt;
1609 }
1610 }
1611 }
1612
1613 modechange:
1614 ;
1615 }
1616
1617interrupt:
1618 self->incr_pos = p - SvPVX (self->incr_text);
1619 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1067} 1620}
1068 1621
1069///////////////////////////////////////////////////////////////////////////// 1622/////////////////////////////////////////////////////////////////////////////
1070// XS interface functions 1623// XS interface functions
1071 1624
1072MODULE = JSON::XS PACKAGE = JSON::XS 1625MODULE = JSON::XS PACKAGE = JSON::XS
1073 1626
1074BOOT: 1627BOOT:
1075{ 1628{
1076 int i; 1629 int i;
1077
1078 memset (decode_hexdigit, 0xff, 256);
1079 1630
1080 for (i = 0; i < 256; ++i) 1631 for (i = 0; i < 256; ++i)
1081 decode_hexdigit [i] = 1632 decode_hexdigit [i] =
1082 i >= '0' && i <= '9' ? i - '0' 1633 i >= '0' && i <= '9' ? i - '0'
1083 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1634 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1084 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1635 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1085 : -1; 1636 : -1;
1086 1637
1087 json_stash = gv_stashpv ("JSON::XS", 1); 1638 json_stash = gv_stashpv ("JSON::XS" , 1);
1639 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1640
1641 json_true = get_bool ("JSON::XS::true");
1642 json_false = get_bool ("JSON::XS::false");
1088} 1643}
1089 1644
1090PROTOTYPES: DISABLE 1645PROTOTYPES: DISABLE
1091 1646
1092SV *new (char *dummy) 1647void CLONE (...)
1093 CODE: 1648 CODE:
1094 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1649 json_stash = 0;
1650 json_boolean_stash = 0;
1651
1652void new (char *klass)
1653 PPCODE:
1654{
1655 SV *pv = NEWSV (0, sizeof (JSON));
1656 SvPOK_only (pv);
1657 json_init ((JSON *)SvPVX (pv));
1658 XPUSHs (sv_2mortal (sv_bless (
1659 newRV_noinc (pv),
1660 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1661 )));
1662}
1663
1664void ascii (JSON *self, int enable = 1)
1665 ALIAS:
1666 ascii = F_ASCII
1667 latin1 = F_LATIN1
1668 utf8 = F_UTF8
1669 indent = F_INDENT
1670 canonical = F_CANONICAL
1671 space_before = F_SPACE_BEFORE
1672 space_after = F_SPACE_AFTER
1673 pretty = F_PRETTY
1674 allow_nonref = F_ALLOW_NONREF
1675 shrink = F_SHRINK
1676 allow_blessed = F_ALLOW_BLESSED
1677 convert_blessed = F_CONV_BLESSED
1678 relaxed = F_RELAXED
1679 allow_unknown = F_ALLOW_UNKNOWN
1680 PPCODE:
1681{
1682 if (enable)
1683 self->flags |= ix;
1684 else
1685 self->flags &= ~ix;
1686
1687 XPUSHs (ST (0));
1688}
1689
1690void get_ascii (JSON *self)
1691 ALIAS:
1692 get_ascii = F_ASCII
1693 get_latin1 = F_LATIN1
1694 get_utf8 = F_UTF8
1695 get_indent = F_INDENT
1696 get_canonical = F_CANONICAL
1697 get_space_before = F_SPACE_BEFORE
1698 get_space_after = F_SPACE_AFTER
1699 get_allow_nonref = F_ALLOW_NONREF
1700 get_shrink = F_SHRINK
1701 get_allow_blessed = F_ALLOW_BLESSED
1702 get_convert_blessed = F_CONV_BLESSED
1703 get_relaxed = F_RELAXED
1704 get_allow_unknown = F_ALLOW_UNKNOWN
1705 PPCODE:
1706 XPUSHs (boolSV (self->flags & ix));
1707
1708void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1709 PPCODE:
1710 self->max_depth = max_depth;
1711 XPUSHs (ST (0));
1712
1713U32 get_max_depth (JSON *self)
1714 CODE:
1715 RETVAL = self->max_depth;
1095 OUTPUT: 1716 OUTPUT:
1096 RETVAL 1717 RETVAL
1097 1718
1098SV *ascii (SV *self, int enable = 1) 1719void max_size (JSON *self, U32 max_size = 0)
1099 ALIAS: 1720 PPCODE:
1100 ascii = F_ASCII 1721 self->max_size = max_size;
1101 latin1 = F_LATIN1 1722 XPUSHs (ST (0));
1102 utf8 = F_UTF8 1723
1103 indent = F_INDENT 1724int get_max_size (JSON *self)
1104 canonical = F_CANONICAL
1105 space_before = F_SPACE_BEFORE
1106 space_after = F_SPACE_AFTER
1107 pretty = F_PRETTY
1108 allow_nonref = F_ALLOW_NONREF
1109 shrink = F_SHRINK
1110 CODE: 1725 CODE:
1111{ 1726 RETVAL = self->max_size;
1112 UV *uv = SvJSON (self);
1113 if (enable)
1114 *uv |= ix;
1115 else
1116 *uv &= ~ix;
1117
1118 RETVAL = newSVsv (self);
1119}
1120 OUTPUT: 1727 OUTPUT:
1121 RETVAL 1728 RETVAL
1122 1729
1123SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1730void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1731 PPCODE:
1732{
1733 SvREFCNT_dec (self->cb_object);
1734 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1735
1736 XPUSHs (ST (0));
1737}
1738
1739void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1740 PPCODE:
1741{
1742 if (!self->cb_sk_object)
1743 self->cb_sk_object = newHV ();
1744
1745 if (SvOK (cb))
1746 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1747 else
1748 {
1749 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1750
1751 if (!HvKEYS (self->cb_sk_object))
1752 {
1753 SvREFCNT_dec (self->cb_sk_object);
1754 self->cb_sk_object = 0;
1755 }
1756 }
1757
1758 XPUSHs (ST (0));
1759}
1760
1761void encode (JSON *self, SV *scalar)
1762 PPCODE:
1763 XPUSHs (encode_json (scalar, self));
1764
1765void decode (JSON *self, SV *jsonstr)
1766 PPCODE:
1767 XPUSHs (decode_json (jsonstr, self, 0));
1768
1769void decode_prefix (JSON *self, SV *jsonstr)
1770 PPCODE:
1771{
1772 STRLEN offset;
1773 EXTEND (SP, 2);
1774 PUSHs (decode_json (jsonstr, self, &offset));
1775 PUSHs (sv_2mortal (newSVuv (offset)));
1776}
1777
1778void incr_parse (JSON *self, SV *jsonstr = 0)
1779 PPCODE:
1780{
1781 if (!self->incr_text)
1782 self->incr_text = newSVpvn ("", 0);
1783
1784 // append data, if any
1785 if (jsonstr)
1786 {
1787 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1788 {
1789 /* utf-8-ness differs, need to upgrade */
1790 sv_utf8_upgrade (self->incr_text);
1791
1792 if (self->incr_pos)
1793 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1794 - (U8 *)SvPVX (self->incr_text);
1795 }
1796
1797 {
1798 STRLEN len;
1799 const char *str = SvPV (jsonstr, len);
1800 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1801 Move (str, SvEND (self->incr_text), len, char);
1802 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1803 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1804 }
1805 }
1806
1807 if (GIMME_V != G_VOID)
1808 do
1809 {
1810 STRLEN offset;
1811
1812 if (!INCR_DONE (self))
1813 {
1814 incr_parse (self);
1815
1816 if (self->incr_pos > self->max_size && self->max_size)
1817 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1818 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1819
1820 if (!INCR_DONE (self))
1821 break;
1822 }
1823
1824 XPUSHs (decode_json (self->incr_text, self, &offset));
1825
1826 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1827 self->incr_pos -= offset;
1828 self->incr_nest = 0;
1829 self->incr_mode = 0;
1830 }
1831 while (GIMME_V == G_ARRAY);
1832}
1833
1834SV *incr_text (JSON *self)
1835 ATTRS: lvalue
1124 CODE: 1836 CODE:
1125{ 1837{
1126 UV *uv = SvJSON (self); 1838 if (self->incr_pos)
1127 UV log2 = 0; 1839 croak ("incr_text can not be called when the incremental parser already started parsing");
1128 1840
1129 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1841 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1130
1131 while ((1UL << log2) < max_depth)
1132 ++log2;
1133
1134 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1135
1136 RETVAL = newSVsv (self);
1137} 1842}
1138 OUTPUT: 1843 OUTPUT:
1139 RETVAL 1844 RETVAL
1140 1845
1846void incr_skip (JSON *self)
1847 CODE:
1848{
1849 if (self->incr_pos)
1850 {
1851 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1852 self->incr_pos = 0;
1853 self->incr_nest = 0;
1854 self->incr_mode = 0;
1855 }
1856}
1857
1858void incr_reset (JSON *self)
1859 CODE:
1860{
1861 SvREFCNT_dec (self->incr_text);
1862 self->incr_text = 0;
1863 self->incr_pos = 0;
1864 self->incr_nest = 0;
1865 self->incr_mode = 0;
1866}
1867
1868void DESTROY (JSON *self)
1869 CODE:
1870 SvREFCNT_dec (self->cb_sk_object);
1871 SvREFCNT_dec (self->cb_object);
1872 SvREFCNT_dec (self->incr_text);
1873
1874PROTOTYPES: ENABLE
1875
1141void encode (SV *self, SV *scalar) 1876void encode_json (SV *scalar)
1877 ALIAS:
1878 to_json_ = 0
1879 encode_json = F_UTF8
1142 PPCODE: 1880 PPCODE:
1881{
1882 JSON json;
1883 json_init (&json);
1884 json.flags |= ix;
1143 XPUSHs (encode_json (scalar, *SvJSON (self))); 1885 XPUSHs (encode_json (scalar, &json));
1886}
1144 1887
1145void decode (SV *self, SV *jsonstr) 1888void decode_json (SV *jsonstr)
1889 ALIAS:
1890 from_json_ = 0
1891 decode_json = F_UTF8
1146 PPCODE: 1892 PPCODE:
1893{
1894 JSON json;
1895 json_init (&json);
1896 json.flags |= ix;
1147 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1897 XPUSHs (decode_json (jsonstr, &json, 0));
1148
1149void decode_prefix (SV *self, SV *jsonstr)
1150 PPCODE:
1151{
1152 UV offset;
1153 EXTEND (SP, 2);
1154 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1155 PUSHs (sv_2mortal (newSVuv (offset)));
1156} 1898}
1157 1899
1158PROTOTYPES: ENABLE
1159 1900
1160void to_json (SV *scalar)
1161 ALIAS:
1162 objToJson = 0
1163 PPCODE:
1164 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1165
1166void from_json (SV *jsonstr)
1167 ALIAS:
1168 jsonToObj = 0
1169 PPCODE:
1170 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1171

Diff Legend

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