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 1.29 by root, Wed May 9 15:34:04 2007 UTC vs.
Revision 1.74 by root, Wed Mar 19 15:17:53 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 <float.h>
9 10
10#if defined(__BORLANDC__) || defined(_MSC_VER) 11#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 12# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 13#endif
13 14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
14#define F_ASCII 0x00000001UL 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
15#define F_UTF8 0x00000002UL 23#define F_UTF8 0x00000004UL
16#define F_INDENT 0x00000004UL 24#define F_INDENT 0x00000008UL
17#define F_CANONICAL 0x00000008UL 25#define F_CANONICAL 0x00000010UL
18#define F_SPACE_BEFORE 0x00000010UL 26#define F_SPACE_BEFORE 0x00000020UL
19#define F_SPACE_AFTER 0x00000020UL 27#define F_SPACE_AFTER 0x00000040UL
20#define F_ALLOW_NONREF 0x00000080UL 28#define F_ALLOW_NONREF 0x00000100UL
21#define F_SHRINK 0x00000100UL 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL
33
22#define F_MAXDEPTH 0xf8000000UL 34#define F_MAXDEPTH 0xf8000000UL
23#define S_MAXDEPTH 27 35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
24 39
25#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
26 41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
27// F_SELFCONVERT? <=> to_json/toJson
28// F_BLESSED? <=> { $__class__$ => }
29 42
30#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
31#define F_DEFAULT (9UL << S_MAXDEPTH) 44#define F_DEFAULT (9UL << S_MAXDEPTH)
32 45
33#define INIT_SIZE 32 // initial scalar size to be allocated 46#define INIT_SIZE 32 // initial scalar size to be allocated
34#define INDENT_STEP 3 // spaces per indentation level 47#define INDENT_STEP 3 // spaces per indentation level
35 48
36#define SHORT_STRING_LEN 512 // special-case strings of up to this size 49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
37 50
38#define SB do { 51#define SB do {
39#define SE } while (0) 52#define SE } while (0)
40 53
54#if __GNUC__ >= 3
55# define expect(expr,value) __builtin_expect ((expr), (value))
56# define INLINE static inline
57#else
58# define expect(expr,value) (expr)
59# define INLINE static
60#endif
61
62#define expect_false(expr) expect ((expr) != 0, 0)
63#define expect_true(expr) expect ((expr) != 0, 1)
64
65#define IN_RANGE_INC(type,val,beg,end) \
66 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
67 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
68
69#ifdef USE_ITHREADS
70# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
72#else
73# define JSON_SLOW 0
74# define JSON_STASH json_stash
75#endif
76
41static HV *json_stash; // JSON::XS:: 77static HV *json_stash, *json_boolean_stash; // JSON::XS::
78static SV *json_true, *json_false;
79
80typedef struct {
81 U32 flags;
82 SV *cb_object;
83 HV *cb_sk_object;
84} JSON;
42 85
43///////////////////////////////////////////////////////////////////////////// 86/////////////////////////////////////////////////////////////////////////////
44// utility functions 87// utility functions
45 88
46static UV * 89INLINE void
47SvJSON (SV *sv)
48{
49 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
50 croak ("object is not of type JSON::XS");
51
52 return &SvUVX (SvRV (sv));
53}
54
55static void
56shrink (SV *sv) 90shrink (SV *sv)
57{ 91{
58 sv_utf8_downgrade (sv, 1); 92 sv_utf8_downgrade (sv, 1);
59 if (SvLEN (sv) > SvCUR (sv) + 1) 93 if (SvLEN (sv) > SvCUR (sv) + 1)
60 { 94 {
… …
69// decode an utf-8 character and return it, or (UV)-1 in 103// decode an utf-8 character and return it, or (UV)-1 in
70// case of an error. 104// case of an error.
71// we special-case "safe" characters from U+80 .. U+7FF, 105// we special-case "safe" characters from U+80 .. U+7FF,
72// but use the very good perl function to parse anything else. 106// but use the very good perl function to parse anything else.
73// note that we never call this function for a ascii codepoints 107// note that we never call this function for a ascii codepoints
74static UV 108INLINE UV
75decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 109decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
76{ 110{
77 if (s[0] > 0xdf || s[0] < 0xc2) 111 if (expect_true (len >= 2
78 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 112 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
79 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 113 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
80 { 114 {
81 *clen = 2; 115 *clen = 2;
82 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 116 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
83 } 117 }
84 else 118 else
85 { 119 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
86 *clen = (STRLEN)-1; 120}
87 return (UV)-1; 121
88 } 122// likewise for encoding, also never called for ascii codepoints
123// this function takes advantage of this fact, although current gccs
124// seem to optimise the check for >= 0x80 away anyways
125INLINE unsigned char *
126encode_utf8 (unsigned char *s, UV ch)
127{
128 if (expect_false (ch < 0x000080))
129 *s++ = ch;
130 else if (expect_true (ch < 0x000800))
131 *s++ = 0xc0 | ( ch >> 6),
132 *s++ = 0x80 | ( ch & 0x3f);
133 else if ( ch < 0x010000)
134 *s++ = 0xe0 | ( ch >> 12),
135 *s++ = 0x80 | ((ch >> 6) & 0x3f),
136 *s++ = 0x80 | ( ch & 0x3f);
137 else if ( ch < 0x110000)
138 *s++ = 0xf0 | ( ch >> 18),
139 *s++ = 0x80 | ((ch >> 12) & 0x3f),
140 *s++ = 0x80 | ((ch >> 6) & 0x3f),
141 *s++ = 0x80 | ( ch & 0x3f);
142
143 return s;
89} 144}
90 145
91///////////////////////////////////////////////////////////////////////////// 146/////////////////////////////////////////////////////////////////////////////
92// encoder 147// encoder
93 148
… …
95typedef struct 150typedef struct
96{ 151{
97 char *cur; // SvPVX (sv) + current output position 152 char *cur; // SvPVX (sv) + current output position
98 char *end; // SvEND (sv) 153 char *end; // SvEND (sv)
99 SV *sv; // result scalar 154 SV *sv; // result scalar
100 U32 flags; // F_* 155 JSON json;
101 U32 indent; // indentation level 156 U32 indent; // indentation level
102 U32 maxdepth; // max. indentation/recursion level 157 U32 maxdepth; // max. indentation/recursion level
158 UV limit; // escape character values >= this value when encoding
103} enc_t; 159} enc_t;
104 160
105static void 161INLINE void
106need (enc_t *enc, STRLEN len) 162need (enc_t *enc, STRLEN len)
107{ 163{
108 if (enc->cur + len >= enc->end) 164 if (expect_false (enc->cur + len >= enc->end))
109 { 165 {
110 STRLEN cur = enc->cur - SvPVX (enc->sv); 166 STRLEN cur = enc->cur - SvPVX (enc->sv);
111 SvGROW (enc->sv, cur + len + 1); 167 SvGROW (enc->sv, cur + len + 1);
112 enc->cur = SvPVX (enc->sv) + cur; 168 enc->cur = SvPVX (enc->sv) + cur;
113 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 169 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
114 } 170 }
115} 171}
116 172
117static void 173INLINE void
118encode_ch (enc_t *enc, char ch) 174encode_ch (enc_t *enc, char ch)
119{ 175{
120 need (enc, 1); 176 need (enc, 1);
121 *enc->cur++ = ch; 177 *enc->cur++ = ch;
122} 178}
… …
130 186
131 while (str < end) 187 while (str < end)
132 { 188 {
133 unsigned char ch = *(unsigned char *)str; 189 unsigned char ch = *(unsigned char *)str;
134 190
135 if (ch >= 0x20 && ch < 0x80) // most common case 191 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
136 { 192 {
137 if (ch == '"') // but with slow exceptions 193 if (expect_false (ch == '"')) // but with slow exceptions
138 { 194 {
139 need (enc, len += 1); 195 need (enc, len += 1);
140 *enc->cur++ = '\\'; 196 *enc->cur++ = '\\';
141 *enc->cur++ = '"'; 197 *enc->cur++ = '"';
142 } 198 }
143 else if (ch == '\\') 199 else if (expect_false (ch == '\\'))
144 { 200 {
145 need (enc, len += 1); 201 need (enc, len += 1);
146 *enc->cur++ = '\\'; 202 *enc->cur++ = '\\';
147 *enc->cur++ = '\\'; 203 *enc->cur++ = '\\';
148 } 204 }
… …
166 STRLEN clen; 222 STRLEN clen;
167 UV uch; 223 UV uch;
168 224
169 if (is_utf8) 225 if (is_utf8)
170 { 226 {
171 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
172 uch = decode_utf8 (str, end - str, &clen); 227 uch = decode_utf8 (str, end - str, &clen);
173 if (clen == (STRLEN)-1) 228 if (clen == (STRLEN)-1)
174 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 229 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
175 } 230 }
176 else 231 else
177 { 232 {
178 uch = ch; 233 uch = ch;
179 clen = 1; 234 clen = 1;
180 } 235 }
181 236
182 if (uch > 0x10FFFFUL) 237 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
183 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
184
185 if (uch < 0x80 || enc->flags & F_ASCII)
186 { 238 {
187 if (uch > 0xFFFFUL) 239 if (uch >= 0x10000UL)
188 { 240 {
241 if (uch >= 0x110000UL)
242 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
243
189 need (enc, len += 11); 244 need (enc, len += 11);
190 sprintf (enc->cur, "\\u%04x\\u%04x", 245 sprintf (enc->cur, "\\u%04x\\u%04x",
191 (int)((uch - 0x10000) / 0x400 + 0xD800), 246 (int)((uch - 0x10000) / 0x400 + 0xD800),
192 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 247 (int)((uch - 0x10000) % 0x400 + 0xDC00));
193 enc->cur += 12; 248 enc->cur += 12;
… …
204 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 259 *enc->cur++ = hexdigit [(uch >> 0) & 15];
205 } 260 }
206 261
207 str += clen; 262 str += clen;
208 } 263 }
264 else if (enc->json.flags & F_LATIN1)
265 {
266 *enc->cur++ = uch;
267 str += clen;
268 }
209 else if (is_utf8) 269 else if (is_utf8)
210 { 270 {
211 need (enc, len += clen); 271 need (enc, len += clen);
212 do 272 do
213 { 273 {
… …
216 while (--clen); 276 while (--clen);
217 } 277 }
218 else 278 else
219 { 279 {
220 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 280 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
221 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 281 enc->cur = encode_utf8 (enc->cur, uch);
222 ++str; 282 ++str;
223 } 283 }
224 } 284 }
225 } 285 }
226 } 286 }
227 287
228 --len; 288 --len;
229 } 289 }
230} 290}
231 291
232static void 292INLINE void
233encode_indent (enc_t *enc) 293encode_indent (enc_t *enc)
234{ 294{
235 if (enc->flags & F_INDENT) 295 if (enc->json.flags & F_INDENT)
236 { 296 {
237 int spaces = enc->indent * INDENT_STEP; 297 int spaces = enc->indent * INDENT_STEP;
238 298
239 need (enc, spaces); 299 need (enc, spaces);
240 memset (enc->cur, ' ', spaces); 300 memset (enc->cur, ' ', spaces);
241 enc->cur += spaces; 301 enc->cur += spaces;
242 } 302 }
243} 303}
244 304
245static void 305INLINE void
246encode_space (enc_t *enc) 306encode_space (enc_t *enc)
247{ 307{
248 need (enc, 1); 308 need (enc, 1);
249 encode_ch (enc, ' '); 309 encode_ch (enc, ' ');
250} 310}
251 311
252static void 312INLINE void
253encode_nl (enc_t *enc) 313encode_nl (enc_t *enc)
254{ 314{
255 if (enc->flags & F_INDENT) 315 if (enc->json.flags & F_INDENT)
256 { 316 {
257 need (enc, 1); 317 need (enc, 1);
258 encode_ch (enc, '\n'); 318 encode_ch (enc, '\n');
259 } 319 }
260} 320}
261 321
262static void 322INLINE void
263encode_comma (enc_t *enc) 323encode_comma (enc_t *enc)
264{ 324{
265 encode_ch (enc, ','); 325 encode_ch (enc, ',');
266 326
267 if (enc->flags & F_INDENT) 327 if (enc->json.flags & F_INDENT)
268 encode_nl (enc); 328 encode_nl (enc);
269 else if (enc->flags & F_SPACE_AFTER) 329 else if (enc->json.flags & F_SPACE_AFTER)
270 encode_space (enc); 330 encode_space (enc);
271} 331}
272 332
273static void encode_sv (enc_t *enc, SV *sv); 333static void encode_sv (enc_t *enc, SV *sv);
274 334
… …
278 int i, len = av_len (av); 338 int i, len = av_len (av);
279 339
280 if (enc->indent >= enc->maxdepth) 340 if (enc->indent >= enc->maxdepth)
281 croak ("data structure too deep (hit recursion limit)"); 341 croak ("data structure too deep (hit recursion limit)");
282 342
283 encode_ch (enc, '['); encode_nl (enc); 343 encode_ch (enc, '[');
284 ++enc->indent; 344
345 if (len >= 0)
346 {
347 encode_nl (enc); ++enc->indent;
285 348
286 for (i = 0; i <= len; ++i) 349 for (i = 0; i <= len; ++i)
287 { 350 {
351 SV **svp = av_fetch (av, i, 0);
352
288 encode_indent (enc); 353 encode_indent (enc);
289 encode_sv (enc, *av_fetch (av, i, 0));
290 354
355 if (svp)
356 encode_sv (enc, *svp);
357 else
358 encode_str (enc, "null", 4, 0);
359
291 if (i < len) 360 if (i < len)
292 encode_comma (enc); 361 encode_comma (enc);
293 } 362 }
294 363
364 encode_nl (enc); --enc->indent; encode_indent (enc);
365 }
366
295 encode_nl (enc); 367 encode_ch (enc, ']');
296
297 --enc->indent;
298 encode_indent (enc); encode_ch (enc, ']');
299} 368}
300 369
301static void 370static void
302encode_he (enc_t *enc, HE *he) 371encode_hk (enc_t *enc, HE *he)
303{ 372{
304 encode_ch (enc, '"'); 373 encode_ch (enc, '"');
305 374
306 if (HeKLEN (he) == HEf_SVKEY) 375 if (HeKLEN (he) == HEf_SVKEY)
307 { 376 {
… …
317 else 386 else
318 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 387 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
319 388
320 encode_ch (enc, '"'); 389 encode_ch (enc, '"');
321 390
322 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 391 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
323 encode_ch (enc, ':'); 392 encode_ch (enc, ':');
324 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 393 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
325 encode_sv (enc, HeVAL (he));
326} 394}
327 395
328// compare hash entries, used when all keys are bytestrings 396// compare hash entries, used when all keys are bytestrings
329static int 397static int
330he_cmp_fast (const void *a_, const void *b_) 398he_cmp_fast (const void *a_, const void *b_)
… …
335 HE *b = *(HE **)b_; 403 HE *b = *(HE **)b_;
336 404
337 STRLEN la = HeKLEN (a); 405 STRLEN la = HeKLEN (a);
338 STRLEN lb = HeKLEN (b); 406 STRLEN lb = HeKLEN (b);
339 407
340 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 408 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
341 cmp = la - lb; 409 cmp = lb - la;
342 410
343 return cmp; 411 return cmp;
344} 412}
345 413
346// compare hash entries, used when some keys are sv's or utf-x 414// compare hash entries, used when some keys are sv's or utf-x
347static int 415static int
348he_cmp_slow (const void *a, const void *b) 416he_cmp_slow (const void *a, const void *b)
349{ 417{
350 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 418 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
351} 419}
352 420
353static void 421static void
354encode_hv (enc_t *enc, HV *hv) 422encode_hv (enc_t *enc, HV *hv)
355{ 423{
424 HE *he;
356 int count, i; 425 int count;
357 426
358 if (enc->indent >= enc->maxdepth) 427 if (enc->indent >= enc->maxdepth)
359 croak ("data structure too deep (hit recursion limit)"); 428 croak ("data structure too deep (hit recursion limit)");
360 429
361 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 430 encode_ch (enc, '{');
362 431
363 if ((count = hv_iterinit (hv)))
364 {
365 // for canonical output we have to sort by keys first 432 // for canonical output we have to sort by keys first
366 // actually, this is mostly due to the stupid so-called 433 // actually, this is mostly due to the stupid so-called
367 // security workaround added somewhere in 5.8.x. 434 // security workaround added somewhere in 5.8.x.
368 // that randomises hash orderings 435 // that randomises hash orderings
369 if (enc->flags & F_CANONICAL) 436 if (enc->json.flags & F_CANONICAL)
437 {
438 int count = hv_iterinit (hv);
439
440 if (SvMAGICAL (hv))
370 { 441 {
442 // need to count by iterating. could improve by dynamically building the vector below
443 // but I don't care for the speed of this special case.
444 // note also that we will run into undefined behaviour when the two iterations
445 // do not result in the same count, something I might care for in some later release.
446
447 count = 0;
448 while (hv_iternext (hv))
449 ++count;
450
451 hv_iterinit (hv);
452 }
453
454 if (count)
455 {
456 int i, fast = 1;
457#if defined(__BORLANDC__) || defined(_MSC_VER)
458 HE **hes = _alloca (count * sizeof (HE));
459#else
371 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 460 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
372 int fast = 1; 461#endif
373 462
374 i = 0; 463 i = 0;
375 while ((he = hv_iternext (hv))) 464 while ((he = hv_iternext (hv)))
376 { 465 {
377 hes [i++] = he; 466 hes [i++] = he;
… …
399 488
400 FREETMPS; 489 FREETMPS;
401 LEAVE; 490 LEAVE;
402 } 491 }
403 492
404 for (i = 0; i < count; ++i) 493 encode_nl (enc); ++enc->indent;
494
495 while (count--)
405 { 496 {
406 encode_indent (enc); 497 encode_indent (enc);
498 he = hes [count];
407 encode_he (enc, hes [i]); 499 encode_hk (enc, he);
500 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
408 501
409 if (i < count - 1) 502 if (count)
410 encode_comma (enc); 503 encode_comma (enc);
411 } 504 }
412 505
506 encode_nl (enc); --enc->indent; encode_indent (enc);
507 }
508 }
509 else
510 {
511 if (hv_iterinit (hv) || SvMAGICAL (hv))
512 if ((he = hv_iternext (hv)))
513 {
514 encode_nl (enc); ++enc->indent;
515
516 for (;;)
517 {
413 encode_nl (enc); 518 encode_indent (enc);
519 encode_hk (enc, he);
520 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
521
522 if (!(he = hv_iternext (hv)))
523 break;
524
525 encode_comma (enc);
526 }
527
528 encode_nl (enc); --enc->indent; encode_indent (enc);
529 }
530 }
531
532 encode_ch (enc, '}');
533}
534
535// encode objects, arrays and special \0=false and \1=true values.
536static void
537encode_rv (enc_t *enc, SV *sv)
538{
539 svtype svt;
540
541 SvGETMAGIC (sv);
542 svt = SvTYPE (sv);
543
544 if (expect_false (SvOBJECT (sv)))
545 {
546 HV *stash = !JSON_SLOW || json_boolean_stash
547 ? json_boolean_stash
548 : gv_stashpv ("JSON::XS::Boolean", 1);
549
550 if (SvSTASH (sv) == stash)
551 {
552 if (SvIV (sv))
553 encode_str (enc, "true", 4, 0);
554 else
555 encode_str (enc, "false", 5, 0);
414 } 556 }
415 else 557 else
416 { 558 {
417 HE *he = hv_iternext (hv); 559#if 0
418 560 if (0 && sv_derived_from (rv, "JSON::Literal"))
419 for (;;)
420 { 561 {
421 encode_indent (enc); 562 // not yet
422 encode_he (enc, he);
423
424 if (!(he = hv_iternext (hv)))
425 break;
426
427 encode_comma (enc);
428 } 563 }
564#endif
565 if (enc->json.flags & F_CONV_BLESSED)
566 {
567 // we re-bless the reference to get overload and other niceties right
568 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
429 569
570 if (to_json)
571 {
572 dSP;
573
574 ENTER; SAVETMPS; PUSHMARK (SP);
575 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
576
577 // calling with G_SCALAR ensures that we always get a 1 return value
578 PUTBACK;
579 call_sv ((SV *)GvCV (to_json), G_SCALAR);
580 SPAGAIN;
581
582 // catch this surprisingly common error
583 if (SvROK (TOPs) && SvRV (TOPs) == sv)
584 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
585
586 sv = POPs;
587 PUTBACK;
588
430 encode_nl (enc); 589 encode_sv (enc, sv);
590
591 FREETMPS; LEAVE;
592 }
593 else if (enc->json.flags & F_ALLOW_BLESSED)
594 encode_str (enc, "null", 4, 0);
595 else
596 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
597 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
598 }
599 else if (enc->json.flags & F_ALLOW_BLESSED)
600 encode_str (enc, "null", 4, 0);
601 else
602 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
603 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
431 } 604 }
432 } 605 }
433
434 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
435}
436
437// encode objects, arrays and special \0=false and \1=true values.
438static void
439encode_rv (enc_t *enc, SV *sv)
440{
441 svtype svt;
442
443 SvGETMAGIC (sv);
444 svt = SvTYPE (sv);
445
446 if (svt == SVt_PVHV) 606 else if (svt == SVt_PVHV)
447 encode_hv (enc, (HV *)sv); 607 encode_hv (enc, (HV *)sv);
448 else if (svt == SVt_PVAV) 608 else if (svt == SVt_PVAV)
449 encode_av (enc, (AV *)sv); 609 encode_av (enc, (AV *)sv);
450 else if (svt < SVt_PVAV) 610 else if (svt < SVt_PVAV)
451 { 611 {
452 if (SvNIOK (sv) && SvIV (sv) == 0) 612 STRLEN len = 0;
613 char *pv = svt ? SvPV (sv, len) : 0;
614
615 if (len == 1 && *pv == '1')
616 encode_str (enc, "true", 4, 0);
617 else if (len == 1 && *pv == '0')
453 encode_str (enc, "false", 5, 0); 618 encode_str (enc, "false", 5, 0);
454 else if (SvNIOK (sv) && SvIV (sv) == 1)
455 encode_str (enc, "true", 4, 0);
456 else 619 else
457 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 620 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
458 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 621 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
459 } 622 }
460 else 623 else
… …
475 encode_str (enc, str, len, SvUTF8 (sv)); 638 encode_str (enc, str, len, SvUTF8 (sv));
476 encode_ch (enc, '"'); 639 encode_ch (enc, '"');
477 } 640 }
478 else if (SvNOKp (sv)) 641 else if (SvNOKp (sv))
479 { 642 {
643 // trust that perl will do the right thing w.r.t. JSON syntax.
480 need (enc, NV_DIG + 32); 644 need (enc, NV_DIG + 32);
481 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 645 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
482 enc->cur += strlen (enc->cur); 646 enc->cur += strlen (enc->cur);
483 } 647 }
484 else if (SvIOKp (sv)) 648 else if (SvIOKp (sv))
485 { 649 {
486 need (enc, 64); 650 // we assume we can always read an IV as a UV and vice versa
651 // we assume two's complement
652 // we assume no aliasing issues in the union
653 if (SvIsUV (sv) ? SvUVX (sv) > 59000
654 : SvIVX (sv) > 59000 || SvIVX (sv) < -59000)
655 {
656 // large integer, use the (rather slow) snprintf way.
657 need (enc, sizeof (UV) * 5 / 2 + 1); // CHAR_BIT is at least 8
487 enc->cur += 658 enc->cur +=
488 SvIsUV(sv) 659 SvIsUV(sv)
489 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 660 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
490 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 661 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
662 }
663 else
664 {
665 // optimise the "small number case"
666 // code will likely be branchless and use only a single multiplication
667 // works for numbers up to 59074
668 I32 i = SvIVX (sv);
669 U32 u;
670 char digit, nz = 0;
671
672 need (enc, 6);
673
674 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
675 u = i < 0 ? -i : i;
676
677 // convert to 4.28 fixed-point representation
678 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
679
680 // now output digit by digit, each time masking out the integer part
681 // and multiplying by 5 while moving the decimal point one to the right,
682 // resulting in a net multiplication by 10.
683 // we always write the digit to memory but conditionally increment
684 // the pointer, to ease the usage of conditional move instructions.
685 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
686 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
687 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
688 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
689 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
690 }
491 } 691 }
492 else if (SvROK (sv)) 692 else if (SvROK (sv))
493 encode_rv (enc, SvRV (sv)); 693 encode_rv (enc, SvRV (sv));
494 else if (!SvOK (sv)) 694 else if (!SvOK (sv))
495 encode_str (enc, "null", 4, 0); 695 encode_str (enc, "null", 4, 0);
… …
497 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 697 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
498 SvPV_nolen (sv), SvFLAGS (sv)); 698 SvPV_nolen (sv), SvFLAGS (sv));
499} 699}
500 700
501static SV * 701static SV *
502encode_json (SV *scalar, U32 flags) 702encode_json (SV *scalar, JSON *json)
503{ 703{
504 enc_t enc; 704 enc_t enc;
505 705
506 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 706 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
507 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 707 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
508 708
509 enc.flags = flags; 709 enc.json = *json;
510 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 710 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
511 enc.cur = SvPVX (enc.sv); 711 enc.cur = SvPVX (enc.sv);
512 enc.end = SvEND (enc.sv); 712 enc.end = SvEND (enc.sv);
513 enc.indent = 0; 713 enc.indent = 0;
514 enc.maxdepth = DEC_DEPTH (flags); 714 enc.maxdepth = DEC_DEPTH (enc.json.flags);
715 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
716 : enc.json.flags & F_LATIN1 ? 0x000100UL
717 : 0x110000UL;
515 718
516 SvPOK_only (enc.sv); 719 SvPOK_only (enc.sv);
517 encode_sv (&enc, scalar); 720 encode_sv (&enc, scalar);
518 721
519 if (!(flags & (F_ASCII | F_UTF8)))
520 SvUTF8_on (enc.sv);
521
522 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 722 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
523 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 723 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
524 724
725 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
726 SvUTF8_on (enc.sv);
727
525 if (enc.flags & F_SHRINK) 728 if (enc.json.flags & F_SHRINK)
526 shrink (enc.sv); 729 shrink (enc.sv);
527 730
528 return enc.sv; 731 return enc.sv;
529} 732}
530 733
… …
535typedef struct 738typedef struct
536{ 739{
537 char *cur; // current parser pointer 740 char *cur; // current parser pointer
538 char *end; // end of input string 741 char *end; // end of input string
539 const char *err; // parse error, if != 0 742 const char *err; // parse error, if != 0
540 U32 flags; // F_* 743 JSON json;
541 U32 depth; // recursion depth 744 U32 depth; // recursion depth
542 U32 maxdepth; // recursion depth limit 745 U32 maxdepth; // recursion depth limit
543} dec_t; 746} dec_t;
544 747
545static void 748INLINE void
749decode_comment (dec_t *dec)
750{
751 // only '#'-style comments allowed a.t.m.
752
753 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
754 ++dec->cur;
755}
756
757INLINE void
546decode_ws (dec_t *dec) 758decode_ws (dec_t *dec)
547{ 759{
548 for (;;) 760 for (;;)
549 { 761 {
550 char ch = *dec->cur; 762 char ch = *dec->cur;
551 763
552 if (ch > 0x20 764 if (ch > 0x20)
765 {
766 if (expect_false (ch == '#'))
767 {
768 if (dec->json.flags & F_RELAXED)
769 decode_comment (dec);
770 else
771 break;
772 }
773 else
774 break;
775 }
553 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 776 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
554 break; 777 break; // parse error, but let higher level handle it, gives better error messages
555 778
556 ++dec->cur; 779 ++dec->cur;
557 } 780 }
558} 781}
559 782
… …
576decode_4hex (dec_t *dec) 799decode_4hex (dec_t *dec)
577{ 800{
578 signed char d1, d2, d3, d4; 801 signed char d1, d2, d3, d4;
579 unsigned char *cur = (unsigned char *)dec->cur; 802 unsigned char *cur = (unsigned char *)dec->cur;
580 803
581 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 804 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
582 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 805 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
583 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 806 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
584 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 807 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
585 808
586 dec->cur += 4; 809 dec->cur += 4;
587 810
588 return ((UV)d1) << 12 811 return ((UV)d1) << 12
589 | ((UV)d2) << 8 812 | ((UV)d2) << 8
… …
597static SV * 820static SV *
598decode_str (dec_t *dec) 821decode_str (dec_t *dec)
599{ 822{
600 SV *sv = 0; 823 SV *sv = 0;
601 int utf8 = 0; 824 int utf8 = 0;
825 char *dec_cur = dec->cur;
602 826
603 do 827 do
604 { 828 {
605 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 829 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
606 char *cur = buf; 830 char *cur = buf;
607 831
608 do 832 do
609 { 833 {
610 unsigned char ch = *(unsigned char *)dec->cur++; 834 unsigned char ch = *(unsigned char *)dec_cur++;
611 835
612 if (ch == '"') 836 if (expect_false (ch == '"'))
613 { 837 {
614 --dec->cur; 838 --dec_cur;
615 break; 839 break;
616 } 840 }
617 else if (ch == '\\') 841 else if (expect_false (ch == '\\'))
618 { 842 {
619 switch (*dec->cur) 843 switch (*dec_cur)
620 { 844 {
621 case '\\': 845 case '\\':
622 case '/': 846 case '/':
623 case '"': *cur++ = *dec->cur++; break; 847 case '"': *cur++ = *dec_cur++; break;
624 848
625 case 'b': ++dec->cur; *cur++ = '\010'; break; 849 case 'b': ++dec_cur; *cur++ = '\010'; break;
626 case 't': ++dec->cur; *cur++ = '\011'; break; 850 case 't': ++dec_cur; *cur++ = '\011'; break;
627 case 'n': ++dec->cur; *cur++ = '\012'; break; 851 case 'n': ++dec_cur; *cur++ = '\012'; break;
628 case 'f': ++dec->cur; *cur++ = '\014'; break; 852 case 'f': ++dec_cur; *cur++ = '\014'; break;
629 case 'r': ++dec->cur; *cur++ = '\015'; break; 853 case 'r': ++dec_cur; *cur++ = '\015'; break;
630 854
631 case 'u': 855 case 'u':
632 { 856 {
633 UV lo, hi; 857 UV lo, hi;
634 ++dec->cur; 858 ++dec_cur;
635 859
860 dec->cur = dec_cur;
636 hi = decode_4hex (dec); 861 hi = decode_4hex (dec);
862 dec_cur = dec->cur;
637 if (hi == (UV)-1) 863 if (hi == (UV)-1)
638 goto fail; 864 goto fail;
639 865
640 // possibly a surrogate pair 866 // possibly a surrogate pair
641 if (hi >= 0xd800) 867 if (hi >= 0xd800)
642 if (hi < 0xdc00) 868 if (hi < 0xdc00)
643 { 869 {
644 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 870 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
645 ERR ("missing low surrogate character in surrogate pair"); 871 ERR ("missing low surrogate character in surrogate pair");
646 872
647 dec->cur += 2; 873 dec_cur += 2;
648 874
875 dec->cur = dec_cur;
649 lo = decode_4hex (dec); 876 lo = decode_4hex (dec);
877 dec_cur = dec->cur;
650 if (lo == (UV)-1) 878 if (lo == (UV)-1)
651 goto fail; 879 goto fail;
652 880
653 if (lo < 0xdc00 || lo >= 0xe000) 881 if (lo < 0xdc00 || lo >= 0xe000)
654 ERR ("surrogate pair expected"); 882 ERR ("surrogate pair expected");
… …
660 888
661 if (hi >= 0x80) 889 if (hi >= 0x80)
662 { 890 {
663 utf8 = 1; 891 utf8 = 1;
664 892
665 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 893 cur = encode_utf8 (cur, hi);
666 } 894 }
667 else 895 else
668 *cur++ = hi; 896 *cur++ = hi;
669 } 897 }
670 break; 898 break;
671 899
672 default: 900 default:
673 --dec->cur; 901 --dec_cur;
674 ERR ("illegal backslash escape sequence in string"); 902 ERR ("illegal backslash escape sequence in string");
675 } 903 }
676 } 904 }
677 else if (ch >= 0x20 && ch <= 0x7f) 905 else if (expect_true (ch >= 0x20 && ch < 0x80))
678 *cur++ = ch; 906 *cur++ = ch;
679 else if (ch >= 0x80) 907 else if (ch >= 0x80)
680 { 908 {
681 STRLEN clen; 909 STRLEN clen;
682 UV uch; 910 UV uch;
683 911
684 --dec->cur; 912 --dec_cur;
685 913
686 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 914 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
687 if (clen == (STRLEN)-1) 915 if (clen == (STRLEN)-1)
688 ERR ("malformed UTF-8 character in JSON string"); 916 ERR ("malformed UTF-8 character in JSON string");
689 917
690 do 918 do
691 *cur++ = *dec->cur++; 919 *cur++ = *dec_cur++;
692 while (--clen); 920 while (--clen);
693 921
694 utf8 = 1; 922 utf8 = 1;
695 } 923 }
696 else 924 else
697 { 925 {
698 --dec->cur; 926 --dec_cur;
699 927
700 if (!ch) 928 if (!ch)
701 ERR ("unexpected end of string while parsing JSON string"); 929 ERR ("unexpected end of string while parsing JSON string");
702 else 930 else
703 ERR ("invalid character encountered while parsing JSON string"); 931 ERR ("invalid character encountered while parsing JSON string");
… …
716 } 944 }
717 else 945 else
718 sv = newSVpvn (buf, len); 946 sv = newSVpvn (buf, len);
719 } 947 }
720 } 948 }
721 while (*dec->cur != '"'); 949 while (*dec_cur != '"');
722 950
723 ++dec->cur; 951 ++dec_cur;
724 952
725 if (sv) 953 if (sv)
726 { 954 {
727 SvPOK_only (sv); 955 SvPOK_only (sv);
728 *SvEND (sv) = 0; 956 *SvEND (sv) = 0;
… …
731 SvUTF8_on (sv); 959 SvUTF8_on (sv);
732 } 960 }
733 else 961 else
734 sv = newSVpvn ("", 0); 962 sv = newSVpvn ("", 0);
735 963
964 dec->cur = dec_cur;
736 return sv; 965 return sv;
737 966
738fail: 967fail:
968 dec->cur = dec_cur;
739 return 0; 969 return 0;
740} 970}
741 971
742static SV * 972static SV *
743decode_num (dec_t *dec) 973decode_num (dec_t *dec)
… …
801 is_nv = 1; 1031 is_nv = 1;
802 } 1032 }
803 1033
804 if (!is_nv) 1034 if (!is_nv)
805 { 1035 {
806 UV uv; 1036 int len = dec->cur - start;
807 int numtype = grok_number (start, dec->cur - start, &uv); 1037
808 if (numtype & IS_NUMBER_IN_UV) 1038 // special case the rather common 1..5-digit-int case
809 if (numtype & IS_NUMBER_NEG) 1039 if (*start == '-')
1040 switch (len)
810 { 1041 {
811 if (uv < (UV)IV_MIN) 1042 case 2: return newSViv (-( start [1] - '0' * 1));
812 return newSViv (-(IV)uv); 1043 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1044 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1045 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1046 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
813 } 1047 }
1048 else
1049 switch (len)
1050 {
1051 case 1: return newSViv ( start [0] - '0' * 1);
1052 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1053 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1054 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1055 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1056 }
1057
1058 {
1059 UV uv;
1060 int numtype = grok_number (start, len, &uv);
1061 if (numtype & IS_NUMBER_IN_UV)
1062 if (numtype & IS_NUMBER_NEG)
1063 {
1064 if (uv < (UV)IV_MIN)
1065 return newSViv (-(IV)uv);
1066 }
814 else 1067 else
815 return newSVuv (uv); 1068 return newSVuv (uv);
816 } 1069 }
817 1070
1071 len -= *start == '-' ? 1 : 0;
1072
1073 // does not fit into IV or UV, try NV
1074 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1075 #if defined (LDBL_DIG)
1076 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1077 #endif
1078 )
1079 // fits into NV without loss of precision
1080 return newSVnv (Atof (start));
1081
1082 // everything else fails, convert it to a string
1083 return newSVpvn (start, dec->cur - start);
1084 }
1085
1086 // loss of precision here
818 return newSVnv (Atof (start)); 1087 return newSVnv (Atof (start));
819 1088
820fail: 1089fail:
821 return 0; 1090 return 0;
822} 1091}
… …
852 1121
853 if (*dec->cur != ',') 1122 if (*dec->cur != ',')
854 ERR (", or ] expected while parsing array"); 1123 ERR (", or ] expected while parsing array");
855 1124
856 ++dec->cur; 1125 ++dec->cur;
1126
1127 decode_ws (dec);
1128
1129 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1130 {
1131 ++dec->cur;
1132 break;
1133 }
857 } 1134 }
858 1135
859 DEC_DEC_DEPTH; 1136 DEC_DEC_DEPTH;
860 return newRV_noinc ((SV *)av); 1137 return newRV_noinc ((SV *)av);
861 1138
… …
866} 1143}
867 1144
868static SV * 1145static SV *
869decode_hv (dec_t *dec) 1146decode_hv (dec_t *dec)
870{ 1147{
1148 SV *sv;
871 HV *hv = newHV (); 1149 HV *hv = newHV ();
872 1150
873 DEC_INC_DEPTH; 1151 DEC_INC_DEPTH;
874 decode_ws (dec); 1152 decode_ws (dec);
875 1153
876 if (*dec->cur == '}') 1154 if (*dec->cur == '}')
877 ++dec->cur; 1155 ++dec->cur;
878 else 1156 else
879 for (;;) 1157 for (;;)
880 { 1158 {
1159 EXPECT_CH ('"');
1160
1161 // heuristic: assume that
1162 // a) decode_str + hv_store_ent are abysmally slow.
1163 // b) most hash keys are short, simple ascii text.
1164 // => try to "fast-match" such strings to avoid
1165 // the overhead of decode_str + hv_store_ent.
1166 {
881 SV *key, *value; 1167 SV *value;
1168 char *p = dec->cur;
1169 char *e = p + 24; // only try up to 24 bytes
882 1170
883 decode_ws (dec); EXPECT_CH ('"'); 1171 for (;;)
884
885 key = decode_str (dec);
886 if (!key)
887 goto fail;
888
889 decode_ws (dec); EXPECT_CH (':');
890
891 value = decode_sv (dec);
892 if (!value)
893 { 1172 {
1173 // the >= 0x80 is true on most architectures
1174 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1175 {
1176 // slow path, back up and use decode_str
1177 SV *key = decode_str (dec);
1178 if (!key)
1179 goto fail;
1180
1181 decode_ws (dec); EXPECT_CH (':');
1182
1183 decode_ws (dec);
1184 value = decode_sv (dec);
1185 if (!value)
1186 {
1187 SvREFCNT_dec (key);
1188 goto fail;
1189 }
1190
1191 hv_store_ent (hv, key, value, 0);
894 SvREFCNT_dec (key); 1192 SvREFCNT_dec (key);
1193
1194 break;
1195 }
1196 else if (*p == '"')
1197 {
1198 // fast path, got a simple key
1199 char *key = dec->cur;
1200 int len = p - key;
1201 dec->cur = p + 1;
1202
1203 decode_ws (dec); EXPECT_CH (':');
1204
1205 decode_ws (dec);
1206 value = decode_sv (dec);
1207 if (!value)
895 goto fail; 1208 goto fail;
1209
1210 hv_store (hv, key, len, value, 0);
1211
1212 break;
1213 }
1214
1215 ++p;
896 } 1216 }
897 1217 }
898 hv_store_ent (hv, key, value, 0);
899 SvREFCNT_dec (key);
900 1218
901 decode_ws (dec); 1219 decode_ws (dec);
902 1220
903 if (*dec->cur == '}') 1221 if (*dec->cur == '}')
904 { 1222 {
… …
908 1226
909 if (*dec->cur != ',') 1227 if (*dec->cur != ',')
910 ERR (", or } expected while parsing object/hash"); 1228 ERR (", or } expected while parsing object/hash");
911 1229
912 ++dec->cur; 1230 ++dec->cur;
1231
1232 decode_ws (dec);
1233
1234 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1235 {
1236 ++dec->cur;
1237 break;
1238 }
913 } 1239 }
914 1240
915 DEC_DEC_DEPTH; 1241 DEC_DEC_DEPTH;
916 return newRV_noinc ((SV *)hv); 1242 sv = newRV_noinc ((SV *)hv);
1243
1244 // check filter callbacks
1245 if (dec->json.flags & F_HOOK)
1246 {
1247 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1248 {
1249 HE *cb, *he;
1250
1251 hv_iterinit (hv);
1252 he = hv_iternext (hv);
1253 hv_iterinit (hv);
1254
1255 // the next line creates a mortal sv each time its called.
1256 // might want to optimise this for common cases.
1257 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1258
1259 if (cb)
1260 {
1261 dSP;
1262 int count;
1263
1264 ENTER; SAVETMPS; PUSHMARK (SP);
1265 XPUSHs (HeVAL (he));
1266
1267 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1268
1269 if (count == 1)
1270 {
1271 sv = newSVsv (POPs);
1272 FREETMPS; LEAVE;
1273 return sv;
1274 }
1275
1276 FREETMPS; LEAVE;
1277 }
1278 }
1279
1280 if (dec->json.cb_object)
1281 {
1282 dSP;
1283 int count;
1284
1285 ENTER; SAVETMPS; PUSHMARK (SP);
1286 XPUSHs (sv_2mortal (sv));
1287
1288 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1289
1290 if (count == 1)
1291 {
1292 sv = newSVsv (POPs);
1293 FREETMPS; LEAVE;
1294 return sv;
1295 }
1296
1297 SvREFCNT_inc (sv);
1298 FREETMPS; LEAVE;
1299 }
1300 }
1301
1302 return sv;
917 1303
918fail: 1304fail:
919 SvREFCNT_dec (hv); 1305 SvREFCNT_dec (hv);
920 DEC_DEC_DEPTH; 1306 DEC_DEC_DEPTH;
921 return 0; 1307 return 0;
922} 1308}
923 1309
924static SV * 1310static SV *
925decode_sv (dec_t *dec) 1311decode_sv (dec_t *dec)
926{ 1312{
927 decode_ws (dec); 1313 // the beauty of JSON: you need exactly one character lookahead
1314 // to parse everything.
928 switch (*dec->cur) 1315 switch (*dec->cur)
929 { 1316 {
930 case '"': ++dec->cur; return decode_str (dec); 1317 case '"': ++dec->cur; return decode_str (dec);
931 case '[': ++dec->cur; return decode_av (dec); 1318 case '[': ++dec->cur; return decode_av (dec);
932 case '{': ++dec->cur; return decode_hv (dec); 1319 case '{': ++dec->cur; return decode_hv (dec);
933 1320
934 case '-': 1321 case '-':
935 case '0': case '1': case '2': case '3': case '4': 1322 case '0': case '1': case '2': case '3': case '4':
936 case '5': case '6': case '7': case '8': case '9': 1323 case '5': case '6': case '7': case '8': case '9':
937 return decode_num (dec); 1324 return decode_num (dec);
938 1325
939 case 't': 1326 case 't':
940 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1327 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
941 { 1328 {
942 dec->cur += 4; 1329 dec->cur += 4;
943 return newSViv (1); 1330#if JSON_SLOW
1331 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1332#endif
1333 return SvREFCNT_inc (json_true);
944 } 1334 }
945 else 1335 else
946 ERR ("'true' expected"); 1336 ERR ("'true' expected");
947 1337
948 break; 1338 break;
949 1339
950 case 'f': 1340 case 'f':
951 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1341 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
952 { 1342 {
953 dec->cur += 5; 1343 dec->cur += 5;
954 return newSViv (0); 1344#if JSON_SLOW
1345 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1346#endif
1347 return SvREFCNT_inc (json_false);
955 } 1348 }
956 else 1349 else
957 ERR ("'false' expected"); 1350 ERR ("'false' expected");
958 1351
959 break; 1352 break;
… …
977fail: 1370fail:
978 return 0; 1371 return 0;
979} 1372}
980 1373
981static SV * 1374static SV *
982decode_json (SV *string, U32 flags) 1375decode_json (SV *string, JSON *json, UV *offset_return)
983{ 1376{
984 dec_t dec; 1377 dec_t dec;
1378 UV offset;
985 SV *sv; 1379 SV *sv;
986 1380
987 SvGETMAGIC (string); 1381 SvGETMAGIC (string);
988 SvUPGRADE (string, SVt_PV); 1382 SvUPGRADE (string, SVt_PV);
989 1383
1384 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1385 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1386 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1387
990 if (flags & F_UTF8) 1388 if (json->flags & F_UTF8)
991 sv_utf8_downgrade (string, 0); 1389 sv_utf8_downgrade (string, 0);
992 else 1390 else
993 sv_utf8_upgrade (string); 1391 sv_utf8_upgrade (string);
994 1392
995 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1393 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
996 1394
997 dec.flags = flags; 1395 dec.json = *json;
998 dec.cur = SvPVX (string); 1396 dec.cur = SvPVX (string);
999 dec.end = SvEND (string); 1397 dec.end = SvEND (string);
1000 dec.err = 0; 1398 dec.err = 0;
1001 dec.depth = 0; 1399 dec.depth = 0;
1002 dec.maxdepth = DEC_DEPTH (dec.flags); 1400 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1003 1401
1402 if (dec.json.cb_object || dec.json.cb_sk_object)
1403 dec.json.flags |= F_HOOK;
1404
1004 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1405 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1406
1407 decode_ws (&dec);
1005 sv = decode_sv (&dec); 1408 sv = decode_sv (&dec);
1006 1409
1410 if (!(offset_return || !sv))
1411 {
1412 // check for trailing garbage
1413 decode_ws (&dec);
1414
1415 if (*dec.cur)
1416 {
1417 dec.err = "garbage after JSON object";
1418 SvREFCNT_dec (sv);
1419 sv = 0;
1420 }
1421 }
1422
1423 if (offset_return || !sv)
1424 {
1425 offset = dec.json.flags & F_UTF8
1426 ? dec.cur - SvPVX (string)
1427 : utf8_distance (dec.cur, SvPVX (string));
1428
1429 if (offset_return)
1430 *offset_return = offset;
1431 }
1432
1007 if (!sv) 1433 if (!sv)
1008 { 1434 {
1009 IV offset = dec.flags & F_UTF8
1010 ? dec.cur - SvPVX (string)
1011 : utf8_distance (dec.cur, SvPVX (string));
1012 SV *uni = sv_newmortal (); 1435 SV *uni = sv_newmortal ();
1013 1436
1014 // horrible hack to silence warning inside pv_uni_display 1437 // horrible hack to silence warning inside pv_uni_display
1015 COP cop = *PL_curcop; 1438 COP cop = *PL_curcop;
1016 cop.cop_warnings = pWARN_NONE; 1439 cop.cop_warnings = pWARN_NONE;
… …
1026 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1449 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1027 } 1450 }
1028 1451
1029 sv = sv_2mortal (sv); 1452 sv = sv_2mortal (sv);
1030 1453
1031 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1454 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1032 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1455 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1033 1456
1034 return sv; 1457 return sv;
1035} 1458}
1036 1459
… …
1040MODULE = JSON::XS PACKAGE = JSON::XS 1463MODULE = JSON::XS PACKAGE = JSON::XS
1041 1464
1042BOOT: 1465BOOT:
1043{ 1466{
1044 int i; 1467 int i;
1045
1046 memset (decode_hexdigit, 0xff, 256);
1047 1468
1048 for (i = 0; i < 256; ++i) 1469 for (i = 0; i < 256; ++i)
1049 decode_hexdigit [i] = 1470 decode_hexdigit [i] =
1050 i >= '0' && i <= '9' ? i - '0' 1471 i >= '0' && i <= '9' ? i - '0'
1051 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1472 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1052 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1473 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1053 : -1; 1474 : -1;
1054 1475
1055 json_stash = gv_stashpv ("JSON::XS", 1); 1476 json_stash = gv_stashpv ("JSON::XS" , 1);
1477 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1478
1479 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1480 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1056} 1481}
1057 1482
1058PROTOTYPES: DISABLE 1483PROTOTYPES: DISABLE
1059 1484
1060SV *new (char *dummy) 1485void CLONE (...)
1061 CODE: 1486 CODE:
1062 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1487 json_stash = 0;
1488 json_boolean_stash = 0;
1489
1490void new (char *klass)
1491 PPCODE:
1492{
1493 SV *pv = NEWSV (0, sizeof (JSON));
1494 SvPOK_only (pv);
1495 Zero (SvPVX (pv), 1, JSON);
1496 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1497 XPUSHs (sv_2mortal (sv_bless (
1498 newRV_noinc (pv),
1499 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1500 )));
1501}
1502
1503void ascii (JSON *self, int enable = 1)
1504 ALIAS:
1505 ascii = F_ASCII
1506 latin1 = F_LATIN1
1507 utf8 = F_UTF8
1508 indent = F_INDENT
1509 canonical = F_CANONICAL
1510 space_before = F_SPACE_BEFORE
1511 space_after = F_SPACE_AFTER
1512 pretty = F_PRETTY
1513 allow_nonref = F_ALLOW_NONREF
1514 shrink = F_SHRINK
1515 allow_blessed = F_ALLOW_BLESSED
1516 convert_blessed = F_CONV_BLESSED
1517 relaxed = F_RELAXED
1518 PPCODE:
1519{
1520 if (enable)
1521 self->flags |= ix;
1522 else
1523 self->flags &= ~ix;
1524
1525 XPUSHs (ST (0));
1526}
1527
1528void get_ascii (JSON *self)
1529 ALIAS:
1530 get_ascii = F_ASCII
1531 get_latin1 = F_LATIN1
1532 get_utf8 = F_UTF8
1533 get_indent = F_INDENT
1534 get_canonical = F_CANONICAL
1535 get_space_before = F_SPACE_BEFORE
1536 get_space_after = F_SPACE_AFTER
1537 get_allow_nonref = F_ALLOW_NONREF
1538 get_shrink = F_SHRINK
1539 get_allow_blessed = F_ALLOW_BLESSED
1540 get_convert_blessed = F_CONV_BLESSED
1541 get_relaxed = F_RELAXED
1542 PPCODE:
1543 XPUSHs (boolSV (self->flags & ix));
1544
1545void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1546 PPCODE:
1547{
1548 UV log2 = 0;
1549
1550 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1551
1552 while ((1UL << log2) < max_depth)
1553 ++log2;
1554
1555 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1556
1557 XPUSHs (ST (0));
1558}
1559
1560U32 get_max_depth (JSON *self)
1561 CODE:
1562 RETVAL = DEC_DEPTH (self->flags);
1063 OUTPUT: 1563 OUTPUT:
1064 RETVAL 1564 RETVAL
1065 1565
1066SV *ascii (SV *self, int enable = 1) 1566void max_size (JSON *self, UV max_size = 0)
1067 ALIAS: 1567 PPCODE:
1068 ascii = F_ASCII 1568{
1069 utf8 = F_UTF8 1569 UV log2 = 0;
1070 indent = F_INDENT 1570
1071 canonical = F_CANONICAL 1571 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1072 space_before = F_SPACE_BEFORE 1572 if (max_size == 1) max_size = 2;
1073 space_after = F_SPACE_AFTER 1573
1074 pretty = F_PRETTY 1574 while ((1UL << log2) < max_size)
1075 allow_nonref = F_ALLOW_NONREF 1575 ++log2;
1076 shrink = F_SHRINK 1576
1577 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1578
1579 XPUSHs (ST (0));
1580}
1581
1582int get_max_size (JSON *self)
1077 CODE: 1583 CODE:
1078{ 1584 RETVAL = DEC_SIZE (self->flags);
1079 UV *uv = SvJSON (self);
1080 if (enable)
1081 *uv |= ix;
1082 else
1083 *uv &= ~ix;
1084
1085 RETVAL = newSVsv (self);
1086}
1087 OUTPUT: 1585 OUTPUT:
1088 RETVAL 1586 RETVAL
1089 1587
1090SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1588void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1589 PPCODE:
1590{
1591 SvREFCNT_dec (self->cb_object);
1592 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1593
1594 XPUSHs (ST (0));
1595}
1596
1597void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1598 PPCODE:
1599{
1600 if (!self->cb_sk_object)
1601 self->cb_sk_object = newHV ();
1602
1603 if (SvOK (cb))
1604 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1605 else
1606 {
1607 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1608
1609 if (!HvKEYS (self->cb_sk_object))
1610 {
1611 SvREFCNT_dec (self->cb_sk_object);
1612 self->cb_sk_object = 0;
1613 }
1614 }
1615
1616 XPUSHs (ST (0));
1617}
1618
1619void encode (JSON *self, SV *scalar)
1620 PPCODE:
1621 XPUSHs (encode_json (scalar, self));
1622
1623void decode (JSON *self, SV *jsonstr)
1624 PPCODE:
1625 XPUSHs (decode_json (jsonstr, self, 0));
1626
1627void decode_prefix (JSON *self, SV *jsonstr)
1628 PPCODE:
1629{
1630 UV offset;
1631 EXTEND (SP, 2);
1632 PUSHs (decode_json (jsonstr, self, &offset));
1633 PUSHs (sv_2mortal (newSVuv (offset)));
1634}
1635
1636void DESTROY (JSON *self)
1091 CODE: 1637 CODE:
1092{ 1638 SvREFCNT_dec (self->cb_sk_object);
1093 UV *uv = SvJSON (self); 1639 SvREFCNT_dec (self->cb_object);
1094 UV log2 = 0;
1095 1640
1096 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1641PROTOTYPES: ENABLE
1097 1642
1098 while ((1UL << log2) < max_depth)
1099 ++log2;
1100
1101 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1102
1103 RETVAL = newSVsv (self);
1104}
1105 OUTPUT:
1106 RETVAL
1107
1108void encode (SV *self, SV *scalar) 1643void encode_json (SV *scalar)
1109 PPCODE: 1644 PPCODE:
1645{
1646 JSON json = { F_DEFAULT | F_UTF8 };
1110 XPUSHs (encode_json (scalar, *SvJSON (self))); 1647 XPUSHs (encode_json (scalar, &json));
1648}
1111 1649
1112void decode (SV *self, SV *jsonstr) 1650void decode_json (SV *jsonstr)
1113 PPCODE: 1651 PPCODE:
1652{
1653 JSON json = { F_DEFAULT | F_UTF8 };
1114 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1654 XPUSHs (decode_json (jsonstr, &json, 0));
1655}
1115 1656
1116PROTOTYPES: ENABLE
1117
1118void to_json (SV *scalar)
1119 ALIAS:
1120 objToJson = 0
1121 PPCODE:
1122 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1123
1124void from_json (SV *jsonstr)
1125 ALIAS:
1126 jsonToObj = 0
1127 PPCODE:
1128 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1129

Diff Legend

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