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 1.32 by root, Wed May 9 16:41:12 2007 UTC vs.
Revision 1.75 by root, Wed Mar 19 15:37:54 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
35#define F_RELAXED 0x00001000UL
36
23#define F_MAXDEPTH 0xf8000000UL 37#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27 38#define S_MAXDEPTH 27
39#define F_MAXSIZE 0x01f00000UL
40#define S_MAXSIZE 20
41#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25 42
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 43#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27 44#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 45
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 46#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH) 47#define F_DEFAULT (9UL << S_MAXDEPTH)
33 48
34#define INIT_SIZE 32 // initial scalar size to be allocated 49#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 50#define INDENT_STEP 3 // spaces per indentation level
36 51
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size 52#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38 53
39#define SB do { 54#define SB do {
40#define SE } while (0) 55#define SE } while (0)
41 56
57#if __GNUC__ >= 3
58# define expect(expr,value) __builtin_expect ((expr), (value))
59# define INLINE static inline
60#else
61# define expect(expr,value) (expr)
62# define INLINE static
63#endif
64
65#define expect_false(expr) expect ((expr) != 0, 0)
66#define expect_true(expr) expect ((expr) != 0, 1)
67
68#define IN_RANGE_INC(type,val,beg,end) \
69 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71
72#ifdef USE_ITHREADS
73# define JSON_SLOW 1
74# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
75#else
76# define JSON_SLOW 0
77# define JSON_STASH json_stash
78#endif
79
42static HV *json_stash; // JSON::XS:: 80static HV *json_stash, *json_boolean_stash; // JSON::XS::
81static SV *json_true, *json_false;
82
83typedef struct {
84 U32 flags;
85 SV *cb_object;
86 HV *cb_sk_object;
87} JSON;
43 88
44///////////////////////////////////////////////////////////////////////////// 89/////////////////////////////////////////////////////////////////////////////
45// utility functions 90// utility functions
46 91
47static UV * 92INLINE void
48SvJSON (SV *sv)
49{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
51 croak ("object is not of type JSON::XS");
52
53 return &SvUVX (SvRV (sv));
54}
55
56static void
57shrink (SV *sv) 93shrink (SV *sv)
58{ 94{
59 sv_utf8_downgrade (sv, 1); 95 sv_utf8_downgrade (sv, 1);
60 if (SvLEN (sv) > SvCUR (sv) + 1) 96 if (SvLEN (sv) > SvCUR (sv) + 1)
61 { 97 {
… …
70// decode an utf-8 character and return it, or (UV)-1 in 106// decode an utf-8 character and return it, or (UV)-1 in
71// case of an error. 107// case of an error.
72// we special-case "safe" characters from U+80 .. U+7FF, 108// we special-case "safe" characters from U+80 .. U+7FF,
73// but use the very good perl function to parse anything else. 109// but use the very good perl function to parse anything else.
74// note that we never call this function for a ascii codepoints 110// note that we never call this function for a ascii codepoints
75static UV 111INLINE UV
76decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 112decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
77{ 113{
78 if (s[0] > 0xdf || s[0] < 0xc2) 114 if (expect_true (len >= 2
79 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 115 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
80 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 116 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
81 { 117 {
82 *clen = 2; 118 *clen = 2;
83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 119 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
84 } 120 }
85 else 121 else
86 { 122 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
87 *clen = (STRLEN)-1; 123}
88 return (UV)-1; 124
89 } 125// likewise for encoding, also never called for ascii codepoints
126// this function takes advantage of this fact, although current gccs
127// seem to optimise the check for >= 0x80 away anyways
128INLINE unsigned char *
129encode_utf8 (unsigned char *s, UV ch)
130{
131 if (expect_false (ch < 0x000080))
132 *s++ = ch;
133 else if (expect_true (ch < 0x000800))
134 *s++ = 0xc0 | ( ch >> 6),
135 *s++ = 0x80 | ( ch & 0x3f);
136 else if ( ch < 0x010000)
137 *s++ = 0xe0 | ( ch >> 12),
138 *s++ = 0x80 | ((ch >> 6) & 0x3f),
139 *s++ = 0x80 | ( ch & 0x3f);
140 else if ( ch < 0x110000)
141 *s++ = 0xf0 | ( ch >> 18),
142 *s++ = 0x80 | ((ch >> 12) & 0x3f),
143 *s++ = 0x80 | ((ch >> 6) & 0x3f),
144 *s++ = 0x80 | ( ch & 0x3f);
145
146 return s;
90} 147}
91 148
92///////////////////////////////////////////////////////////////////////////// 149/////////////////////////////////////////////////////////////////////////////
93// encoder 150// encoder
94 151
… …
96typedef struct 153typedef struct
97{ 154{
98 char *cur; // SvPVX (sv) + current output position 155 char *cur; // SvPVX (sv) + current output position
99 char *end; // SvEND (sv) 156 char *end; // SvEND (sv)
100 SV *sv; // result scalar 157 SV *sv; // result scalar
101 U32 flags; // F_* 158 JSON json;
102 U32 indent; // indentation level 159 U32 indent; // indentation level
103 U32 maxdepth; // max. indentation/recursion level 160 U32 maxdepth; // max. indentation/recursion level
161 UV limit; // escape character values >= this value when encoding
104} enc_t; 162} enc_t;
105 163
106static void 164INLINE void
107need (enc_t *enc, STRLEN len) 165need (enc_t *enc, STRLEN len)
108{ 166{
109 if (enc->cur + len >= enc->end) 167 if (expect_false (enc->cur + len >= enc->end))
110 { 168 {
111 STRLEN cur = enc->cur - SvPVX (enc->sv); 169 STRLEN cur = enc->cur - SvPVX (enc->sv);
112 SvGROW (enc->sv, cur + len + 1); 170 SvGROW (enc->sv, cur + len + 1);
113 enc->cur = SvPVX (enc->sv) + cur; 171 enc->cur = SvPVX (enc->sv) + cur;
114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 172 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
115 } 173 }
116} 174}
117 175
118static void 176INLINE void
119encode_ch (enc_t *enc, char ch) 177encode_ch (enc_t *enc, char ch)
120{ 178{
121 need (enc, 1); 179 need (enc, 1);
122 *enc->cur++ = ch; 180 *enc->cur++ = ch;
123} 181}
… …
131 189
132 while (str < end) 190 while (str < end)
133 { 191 {
134 unsigned char ch = *(unsigned char *)str; 192 unsigned char ch = *(unsigned char *)str;
135 193
136 if (ch >= 0x20 && ch < 0x80) // most common case 194 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
137 { 195 {
138 if (ch == '"') // but with slow exceptions 196 if (expect_false (ch == '"')) // but with slow exceptions
139 { 197 {
140 need (enc, len += 1); 198 need (enc, len += 1);
141 *enc->cur++ = '\\'; 199 *enc->cur++ = '\\';
142 *enc->cur++ = '"'; 200 *enc->cur++ = '"';
143 } 201 }
144 else if (ch == '\\') 202 else if (expect_false (ch == '\\'))
145 { 203 {
146 need (enc, len += 1); 204 need (enc, len += 1);
147 *enc->cur++ = '\\'; 205 *enc->cur++ = '\\';
148 *enc->cur++ = '\\'; 206 *enc->cur++ = '\\';
149 } 207 }
… …
167 STRLEN clen; 225 STRLEN clen;
168 UV uch; 226 UV uch;
169 227
170 if (is_utf8) 228 if (is_utf8)
171 { 229 {
172 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173 uch = decode_utf8 (str, end - str, &clen); 230 uch = decode_utf8 (str, end - str, &clen);
174 if (clen == (STRLEN)-1) 231 if (clen == (STRLEN)-1)
175 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 232 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
176 } 233 }
177 else 234 else
178 { 235 {
179 uch = ch; 236 uch = ch;
180 clen = 1; 237 clen = 1;
181 } 238 }
182 239
183 if (uch > 0x10FFFFUL) 240 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 { 241 {
188 if (uch > 0xFFFFUL) 242 if (uch >= 0x10000UL)
189 { 243 {
244 if (uch >= 0x110000UL)
245 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
246
190 need (enc, len += 11); 247 need (enc, len += 11);
191 sprintf (enc->cur, "\\u%04x\\u%04x", 248 sprintf (enc->cur, "\\u%04x\\u%04x",
192 (int)((uch - 0x10000) / 0x400 + 0xD800), 249 (int)((uch - 0x10000) / 0x400 + 0xD800),
193 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 250 (int)((uch - 0x10000) % 0x400 + 0xDC00));
194 enc->cur += 12; 251 enc->cur += 12;
… …
205 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 262 *enc->cur++ = hexdigit [(uch >> 0) & 15];
206 } 263 }
207 264
208 str += clen; 265 str += clen;
209 } 266 }
210 else if (enc->flags & F_LATIN1) 267 else if (enc->json.flags & F_LATIN1)
211 { 268 {
212 *enc->cur++ = uch; 269 *enc->cur++ = uch;
213 str += clen; 270 str += clen;
214 } 271 }
215 else if (is_utf8) 272 else if (is_utf8)
… …
222 while (--clen); 279 while (--clen);
223 } 280 }
224 else 281 else
225 { 282 {
226 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 283 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
227 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 284 enc->cur = encode_utf8 (enc->cur, uch);
228 ++str; 285 ++str;
229 } 286 }
230 } 287 }
231 } 288 }
232 } 289 }
233 290
234 --len; 291 --len;
235 } 292 }
236} 293}
237 294
238static void 295INLINE void
239encode_indent (enc_t *enc) 296encode_indent (enc_t *enc)
240{ 297{
241 if (enc->flags & F_INDENT) 298 if (enc->json.flags & F_INDENT)
242 { 299 {
243 int spaces = enc->indent * INDENT_STEP; 300 int spaces = enc->indent * INDENT_STEP;
244 301
245 need (enc, spaces); 302 need (enc, spaces);
246 memset (enc->cur, ' ', spaces); 303 memset (enc->cur, ' ', spaces);
247 enc->cur += spaces; 304 enc->cur += spaces;
248 } 305 }
249} 306}
250 307
251static void 308INLINE void
252encode_space (enc_t *enc) 309encode_space (enc_t *enc)
253{ 310{
254 need (enc, 1); 311 need (enc, 1);
255 encode_ch (enc, ' '); 312 encode_ch (enc, ' ');
256} 313}
257 314
258static void 315INLINE void
259encode_nl (enc_t *enc) 316encode_nl (enc_t *enc)
260{ 317{
261 if (enc->flags & F_INDENT) 318 if (enc->json.flags & F_INDENT)
262 { 319 {
263 need (enc, 1); 320 need (enc, 1);
264 encode_ch (enc, '\n'); 321 encode_ch (enc, '\n');
265 } 322 }
266} 323}
267 324
268static void 325INLINE void
269encode_comma (enc_t *enc) 326encode_comma (enc_t *enc)
270{ 327{
271 encode_ch (enc, ','); 328 encode_ch (enc, ',');
272 329
273 if (enc->flags & F_INDENT) 330 if (enc->json.flags & F_INDENT)
274 encode_nl (enc); 331 encode_nl (enc);
275 else if (enc->flags & F_SPACE_AFTER) 332 else if (enc->json.flags & F_SPACE_AFTER)
276 encode_space (enc); 333 encode_space (enc);
277} 334}
278 335
279static void encode_sv (enc_t *enc, SV *sv); 336static void encode_sv (enc_t *enc, SV *sv);
280 337
… …
284 int i, len = av_len (av); 341 int i, len = av_len (av);
285 342
286 if (enc->indent >= enc->maxdepth) 343 if (enc->indent >= enc->maxdepth)
287 croak ("data structure too deep (hit recursion limit)"); 344 croak ("data structure too deep (hit recursion limit)");
288 345
289 encode_ch (enc, '['); encode_nl (enc); 346 encode_ch (enc, '[');
290 ++enc->indent; 347
348 if (len >= 0)
349 {
350 encode_nl (enc); ++enc->indent;
291 351
292 for (i = 0; i <= len; ++i) 352 for (i = 0; i <= len; ++i)
293 { 353 {
354 SV **svp = av_fetch (av, i, 0);
355
294 encode_indent (enc); 356 encode_indent (enc);
295 encode_sv (enc, *av_fetch (av, i, 0));
296 357
358 if (svp)
359 encode_sv (enc, *svp);
360 else
361 encode_str (enc, "null", 4, 0);
362
297 if (i < len) 363 if (i < len)
298 encode_comma (enc); 364 encode_comma (enc);
299 } 365 }
300 366
367 encode_nl (enc); --enc->indent; encode_indent (enc);
368 }
369
301 encode_nl (enc); 370 encode_ch (enc, ']');
302
303 --enc->indent;
304 encode_indent (enc); encode_ch (enc, ']');
305} 371}
306 372
307static void 373static void
308encode_he (enc_t *enc, HE *he) 374encode_hk (enc_t *enc, HE *he)
309{ 375{
310 encode_ch (enc, '"'); 376 encode_ch (enc, '"');
311 377
312 if (HeKLEN (he) == HEf_SVKEY) 378 if (HeKLEN (he) == HEf_SVKEY)
313 { 379 {
… …
323 else 389 else
324 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 390 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
325 391
326 encode_ch (enc, '"'); 392 encode_ch (enc, '"');
327 393
328 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 394 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
329 encode_ch (enc, ':'); 395 encode_ch (enc, ':');
330 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 396 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
331 encode_sv (enc, HeVAL (he));
332} 397}
333 398
334// compare hash entries, used when all keys are bytestrings 399// compare hash entries, used when all keys are bytestrings
335static int 400static int
336he_cmp_fast (const void *a_, const void *b_) 401he_cmp_fast (const void *a_, const void *b_)
… …
341 HE *b = *(HE **)b_; 406 HE *b = *(HE **)b_;
342 407
343 STRLEN la = HeKLEN (a); 408 STRLEN la = HeKLEN (a);
344 STRLEN lb = HeKLEN (b); 409 STRLEN lb = HeKLEN (b);
345 410
346 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 411 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
347 cmp = la - lb; 412 cmp = lb - la;
348 413
349 return cmp; 414 return cmp;
350} 415}
351 416
352// compare hash entries, used when some keys are sv's or utf-x 417// compare hash entries, used when some keys are sv's or utf-x
353static int 418static int
354he_cmp_slow (const void *a, const void *b) 419he_cmp_slow (const void *a, const void *b)
355{ 420{
356 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 421 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
357} 422}
358 423
359static void 424static void
360encode_hv (enc_t *enc, HV *hv) 425encode_hv (enc_t *enc, HV *hv)
361{ 426{
427 HE *he;
362 int count, i; 428 int count;
363 429
364 if (enc->indent >= enc->maxdepth) 430 if (enc->indent >= enc->maxdepth)
365 croak ("data structure too deep (hit recursion limit)"); 431 croak ("data structure too deep (hit recursion limit)");
366 432
367 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 433 encode_ch (enc, '{');
368 434
369 if ((count = hv_iterinit (hv)))
370 {
371 // for canonical output we have to sort by keys first 435 // for canonical output we have to sort by keys first
372 // actually, this is mostly due to the stupid so-called 436 // actually, this is mostly due to the stupid so-called
373 // security workaround added somewhere in 5.8.x. 437 // security workaround added somewhere in 5.8.x.
374 // that randomises hash orderings 438 // that randomises hash orderings
375 if (enc->flags & F_CANONICAL) 439 if (enc->json.flags & F_CANONICAL)
440 {
441 int count = hv_iterinit (hv);
442
443 if (SvMAGICAL (hv))
376 { 444 {
445 // need to count by iterating. could improve by dynamically building the vector below
446 // but I don't care for the speed of this special case.
447 // note also that we will run into undefined behaviour when the two iterations
448 // do not result in the same count, something I might care for in some later release.
449
450 count = 0;
451 while (hv_iternext (hv))
452 ++count;
453
454 hv_iterinit (hv);
455 }
456
457 if (count)
458 {
459 int i, fast = 1;
460#if defined(__BORLANDC__) || defined(_MSC_VER)
461 HE **hes = _alloca (count * sizeof (HE));
462#else
377 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 463 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
378 int fast = 1; 464#endif
379 465
380 i = 0; 466 i = 0;
381 while ((he = hv_iternext (hv))) 467 while ((he = hv_iternext (hv)))
382 { 468 {
383 hes [i++] = he; 469 hes [i++] = he;
… …
405 491
406 FREETMPS; 492 FREETMPS;
407 LEAVE; 493 LEAVE;
408 } 494 }
409 495
410 for (i = 0; i < count; ++i) 496 encode_nl (enc); ++enc->indent;
497
498 while (count--)
411 { 499 {
412 encode_indent (enc); 500 encode_indent (enc);
501 he = hes [count];
413 encode_he (enc, hes [i]); 502 encode_hk (enc, he);
503 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
414 504
415 if (i < count - 1) 505 if (count)
416 encode_comma (enc); 506 encode_comma (enc);
417 } 507 }
418 508
509 encode_nl (enc); --enc->indent; encode_indent (enc);
510 }
511 }
512 else
513 {
514 if (hv_iterinit (hv) || SvMAGICAL (hv))
515 if ((he = hv_iternext (hv)))
516 {
517 encode_nl (enc); ++enc->indent;
518
519 for (;;)
520 {
419 encode_nl (enc); 521 encode_indent (enc);
522 encode_hk (enc, he);
523 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
524
525 if (!(he = hv_iternext (hv)))
526 break;
527
528 encode_comma (enc);
529 }
530
531 encode_nl (enc); --enc->indent; encode_indent (enc);
532 }
533 }
534
535 encode_ch (enc, '}');
536}
537
538// encode objects, arrays and special \0=false and \1=true values.
539static void
540encode_rv (enc_t *enc, SV *sv)
541{
542 svtype svt;
543
544 SvGETMAGIC (sv);
545 svt = SvTYPE (sv);
546
547 if (expect_false (SvOBJECT (sv)))
548 {
549 HV *stash = !JSON_SLOW || json_boolean_stash
550 ? json_boolean_stash
551 : gv_stashpv ("JSON::XS::Boolean", 1);
552
553 if (SvSTASH (sv) == stash)
554 {
555 if (SvIV (sv))
556 encode_str (enc, "true", 4, 0);
557 else
558 encode_str (enc, "false", 5, 0);
420 } 559 }
421 else 560 else
422 { 561 {
423 HE *he = hv_iternext (hv); 562#if 0
424 563 if (0 && sv_derived_from (rv, "JSON::Literal"))
425 for (;;)
426 { 564 {
427 encode_indent (enc); 565 // not yet
428 encode_he (enc, he);
429
430 if (!(he = hv_iternext (hv)))
431 break;
432
433 encode_comma (enc);
434 } 566 }
567#endif
568 if (enc->json.flags & F_CONV_BLESSED)
569 {
570 // we re-bless the reference to get overload and other niceties right
571 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
435 572
573 if (to_json)
574 {
575 dSP;
576
577 ENTER; SAVETMPS; PUSHMARK (SP);
578 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
579
580 // calling with G_SCALAR ensures that we always get a 1 return value
581 PUTBACK;
582 call_sv ((SV *)GvCV (to_json), G_SCALAR);
583 SPAGAIN;
584
585 // catch this surprisingly common error
586 if (SvROK (TOPs) && SvRV (TOPs) == sv)
587 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
588
589 sv = POPs;
590 PUTBACK;
591
436 encode_nl (enc); 592 encode_sv (enc, sv);
593
594 FREETMPS; LEAVE;
595 }
596 else if (enc->json.flags & F_ALLOW_BLESSED)
597 encode_str (enc, "null", 4, 0);
598 else
599 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
600 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
601 }
602 else if (enc->json.flags & F_ALLOW_BLESSED)
603 encode_str (enc, "null", 4, 0);
604 else
605 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
606 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
437 } 607 }
438 } 608 }
439
440 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
441}
442
443// encode objects, arrays and special \0=false and \1=true values.
444static void
445encode_rv (enc_t *enc, SV *sv)
446{
447 svtype svt;
448
449 SvGETMAGIC (sv);
450 svt = SvTYPE (sv);
451
452 if (svt == SVt_PVHV) 609 else if (svt == SVt_PVHV)
453 encode_hv (enc, (HV *)sv); 610 encode_hv (enc, (HV *)sv);
454 else if (svt == SVt_PVAV) 611 else if (svt == SVt_PVAV)
455 encode_av (enc, (AV *)sv); 612 encode_av (enc, (AV *)sv);
456 else if (svt < SVt_PVAV) 613 else if (svt < SVt_PVAV)
457 { 614 {
458 if (SvNIOK (sv) && SvIV (sv) == 0) 615 STRLEN len = 0;
616 char *pv = svt ? SvPV (sv, len) : 0;
617
618 if (len == 1 && *pv == '1')
619 encode_str (enc, "true", 4, 0);
620 else if (len == 1 && *pv == '0')
459 encode_str (enc, "false", 5, 0); 621 encode_str (enc, "false", 5, 0);
460 else if (SvNIOK (sv) && SvIV (sv) == 1)
461 encode_str (enc, "true", 4, 0);
462 else 622 else
463 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 623 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
464 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 624 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
465 } 625 }
466 else 626 else
… …
481 encode_str (enc, str, len, SvUTF8 (sv)); 641 encode_str (enc, str, len, SvUTF8 (sv));
482 encode_ch (enc, '"'); 642 encode_ch (enc, '"');
483 } 643 }
484 else if (SvNOKp (sv)) 644 else if (SvNOKp (sv))
485 { 645 {
646 // trust that perl will do the right thing w.r.t. JSON syntax.
486 need (enc, NV_DIG + 32); 647 need (enc, NV_DIG + 32);
487 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 648 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
488 enc->cur += strlen (enc->cur); 649 enc->cur += strlen (enc->cur);
489 } 650 }
490 else if (SvIOKp (sv)) 651 else if (SvIOKp (sv))
491 { 652 {
653 // we assume we can always read an IV as a UV and vice versa
654 // we assume two's complement
655 // we assume no aliasing issues in the union
656 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
657 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
658 {
659 // optimise the "small number case"
660 // code will likely be branchless and use only a single multiplication
661 // works for numbers up to 59074
662 I32 i = SvIVX (sv);
663 U32 u;
664 char digit, nz = 0;
665
492 need (enc, 64); 666 need (enc, 6);
667
668 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
669 u = i < 0 ? -i : i;
670
671 // convert to 4.28 fixed-point representation
672 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
673
674 // now output digit by digit, each time masking out the integer part
675 // and multiplying by 5 while moving the decimal point one to the right,
676 // resulting in a net multiplication by 10.
677 // we always write the digit to memory but conditionally increment
678 // the pointer, to enable the use of conditional move instructions.
679 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
680 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
681 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
682 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
683 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
684 }
685 else
686 {
687 // large integer, use the (rather slow) snprintf way.
688 need (enc, IVUV_MAXCHARS);
493 enc->cur += 689 enc->cur +=
494 SvIsUV(sv) 690 SvIsUV(sv)
495 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 691 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
496 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 692 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
693 }
497 } 694 }
498 else if (SvROK (sv)) 695 else if (SvROK (sv))
499 encode_rv (enc, SvRV (sv)); 696 encode_rv (enc, SvRV (sv));
500 else if (!SvOK (sv)) 697 else if (!SvOK (sv))
501 encode_str (enc, "null", 4, 0); 698 encode_str (enc, "null", 4, 0);
… …
503 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 700 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
504 SvPV_nolen (sv), SvFLAGS (sv)); 701 SvPV_nolen (sv), SvFLAGS (sv));
505} 702}
506 703
507static SV * 704static SV *
508encode_json (SV *scalar, U32 flags) 705encode_json (SV *scalar, JSON *json)
509{ 706{
510 enc_t enc; 707 enc_t enc;
511 708
512 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 709 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
513 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 710 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
514 711
515 enc.flags = flags; 712 enc.json = *json;
516 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 713 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517 enc.cur = SvPVX (enc.sv); 714 enc.cur = SvPVX (enc.sv);
518 enc.end = SvEND (enc.sv); 715 enc.end = SvEND (enc.sv);
519 enc.indent = 0; 716 enc.indent = 0;
520 enc.maxdepth = DEC_DEPTH (flags); 717 enc.maxdepth = DEC_DEPTH (enc.json.flags);
718 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
719 : enc.json.flags & F_LATIN1 ? 0x000100UL
720 : 0x110000UL;
521 721
522 SvPOK_only (enc.sv); 722 SvPOK_only (enc.sv);
523 encode_sv (&enc, scalar); 723 encode_sv (&enc, scalar);
524 724
525 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 725 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
526 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 726 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
527 727
528 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 728 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
529 SvUTF8_on (enc.sv); 729 SvUTF8_on (enc.sv);
530 730
531 if (enc.flags & F_SHRINK) 731 if (enc.json.flags & F_SHRINK)
532 shrink (enc.sv); 732 shrink (enc.sv);
533 733
534 return enc.sv; 734 return enc.sv;
535} 735}
536 736
… …
541typedef struct 741typedef struct
542{ 742{
543 char *cur; // current parser pointer 743 char *cur; // current parser pointer
544 char *end; // end of input string 744 char *end; // end of input string
545 const char *err; // parse error, if != 0 745 const char *err; // parse error, if != 0
546 U32 flags; // F_* 746 JSON json;
547 U32 depth; // recursion depth 747 U32 depth; // recursion depth
548 U32 maxdepth; // recursion depth limit 748 U32 maxdepth; // recursion depth limit
549} dec_t; 749} dec_t;
550 750
551static void 751INLINE void
752decode_comment (dec_t *dec)
753{
754 // only '#'-style comments allowed a.t.m.
755
756 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
757 ++dec->cur;
758}
759
760INLINE void
552decode_ws (dec_t *dec) 761decode_ws (dec_t *dec)
553{ 762{
554 for (;;) 763 for (;;)
555 { 764 {
556 char ch = *dec->cur; 765 char ch = *dec->cur;
557 766
558 if (ch > 0x20 767 if (ch > 0x20)
768 {
769 if (expect_false (ch == '#'))
770 {
771 if (dec->json.flags & F_RELAXED)
772 decode_comment (dec);
773 else
774 break;
775 }
776 else
777 break;
778 }
559 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 779 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
560 break; 780 break; // parse error, but let higher level handle it, gives better error messages
561 781
562 ++dec->cur; 782 ++dec->cur;
563 } 783 }
564} 784}
565 785
… …
582decode_4hex (dec_t *dec) 802decode_4hex (dec_t *dec)
583{ 803{
584 signed char d1, d2, d3, d4; 804 signed char d1, d2, d3, d4;
585 unsigned char *cur = (unsigned char *)dec->cur; 805 unsigned char *cur = (unsigned char *)dec->cur;
586 806
587 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 807 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
588 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 808 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
589 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 809 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
590 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 810 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
591 811
592 dec->cur += 4; 812 dec->cur += 4;
593 813
594 return ((UV)d1) << 12 814 return ((UV)d1) << 12
595 | ((UV)d2) << 8 815 | ((UV)d2) << 8
… …
603static SV * 823static SV *
604decode_str (dec_t *dec) 824decode_str (dec_t *dec)
605{ 825{
606 SV *sv = 0; 826 SV *sv = 0;
607 int utf8 = 0; 827 int utf8 = 0;
828 char *dec_cur = dec->cur;
608 829
609 do 830 do
610 { 831 {
611 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 832 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
612 char *cur = buf; 833 char *cur = buf;
613 834
614 do 835 do
615 { 836 {
616 unsigned char ch = *(unsigned char *)dec->cur++; 837 unsigned char ch = *(unsigned char *)dec_cur++;
617 838
618 if (ch == '"') 839 if (expect_false (ch == '"'))
619 { 840 {
620 --dec->cur; 841 --dec_cur;
621 break; 842 break;
622 } 843 }
623 else if (ch == '\\') 844 else if (expect_false (ch == '\\'))
624 { 845 {
625 switch (*dec->cur) 846 switch (*dec_cur)
626 { 847 {
627 case '\\': 848 case '\\':
628 case '/': 849 case '/':
629 case '"': *cur++ = *dec->cur++; break; 850 case '"': *cur++ = *dec_cur++; break;
630 851
631 case 'b': ++dec->cur; *cur++ = '\010'; break; 852 case 'b': ++dec_cur; *cur++ = '\010'; break;
632 case 't': ++dec->cur; *cur++ = '\011'; break; 853 case 't': ++dec_cur; *cur++ = '\011'; break;
633 case 'n': ++dec->cur; *cur++ = '\012'; break; 854 case 'n': ++dec_cur; *cur++ = '\012'; break;
634 case 'f': ++dec->cur; *cur++ = '\014'; break; 855 case 'f': ++dec_cur; *cur++ = '\014'; break;
635 case 'r': ++dec->cur; *cur++ = '\015'; break; 856 case 'r': ++dec_cur; *cur++ = '\015'; break;
636 857
637 case 'u': 858 case 'u':
638 { 859 {
639 UV lo, hi; 860 UV lo, hi;
640 ++dec->cur; 861 ++dec_cur;
641 862
863 dec->cur = dec_cur;
642 hi = decode_4hex (dec); 864 hi = decode_4hex (dec);
865 dec_cur = dec->cur;
643 if (hi == (UV)-1) 866 if (hi == (UV)-1)
644 goto fail; 867 goto fail;
645 868
646 // possibly a surrogate pair 869 // possibly a surrogate pair
647 if (hi >= 0xd800) 870 if (hi >= 0xd800)
648 if (hi < 0xdc00) 871 if (hi < 0xdc00)
649 { 872 {
650 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 873 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
651 ERR ("missing low surrogate character in surrogate pair"); 874 ERR ("missing low surrogate character in surrogate pair");
652 875
653 dec->cur += 2; 876 dec_cur += 2;
654 877
878 dec->cur = dec_cur;
655 lo = decode_4hex (dec); 879 lo = decode_4hex (dec);
880 dec_cur = dec->cur;
656 if (lo == (UV)-1) 881 if (lo == (UV)-1)
657 goto fail; 882 goto fail;
658 883
659 if (lo < 0xdc00 || lo >= 0xe000) 884 if (lo < 0xdc00 || lo >= 0xe000)
660 ERR ("surrogate pair expected"); 885 ERR ("surrogate pair expected");
… …
666 891
667 if (hi >= 0x80) 892 if (hi >= 0x80)
668 { 893 {
669 utf8 = 1; 894 utf8 = 1;
670 895
671 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 896 cur = encode_utf8 (cur, hi);
672 } 897 }
673 else 898 else
674 *cur++ = hi; 899 *cur++ = hi;
675 } 900 }
676 break; 901 break;
677 902
678 default: 903 default:
679 --dec->cur; 904 --dec_cur;
680 ERR ("illegal backslash escape sequence in string"); 905 ERR ("illegal backslash escape sequence in string");
681 } 906 }
682 } 907 }
683 else if (ch >= 0x20 && ch <= 0x7f) 908 else if (expect_true (ch >= 0x20 && ch < 0x80))
684 *cur++ = ch; 909 *cur++ = ch;
685 else if (ch >= 0x80) 910 else if (ch >= 0x80)
686 { 911 {
687 STRLEN clen; 912 STRLEN clen;
688 UV uch; 913 UV uch;
689 914
690 --dec->cur; 915 --dec_cur;
691 916
692 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 917 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
693 if (clen == (STRLEN)-1) 918 if (clen == (STRLEN)-1)
694 ERR ("malformed UTF-8 character in JSON string"); 919 ERR ("malformed UTF-8 character in JSON string");
695 920
696 do 921 do
697 *cur++ = *dec->cur++; 922 *cur++ = *dec_cur++;
698 while (--clen); 923 while (--clen);
699 924
700 utf8 = 1; 925 utf8 = 1;
701 } 926 }
702 else 927 else
703 { 928 {
704 --dec->cur; 929 --dec_cur;
705 930
706 if (!ch) 931 if (!ch)
707 ERR ("unexpected end of string while parsing JSON string"); 932 ERR ("unexpected end of string while parsing JSON string");
708 else 933 else
709 ERR ("invalid character encountered while parsing JSON string"); 934 ERR ("invalid character encountered while parsing JSON string");
… …
722 } 947 }
723 else 948 else
724 sv = newSVpvn (buf, len); 949 sv = newSVpvn (buf, len);
725 } 950 }
726 } 951 }
727 while (*dec->cur != '"'); 952 while (*dec_cur != '"');
728 953
729 ++dec->cur; 954 ++dec_cur;
730 955
731 if (sv) 956 if (sv)
732 { 957 {
733 SvPOK_only (sv); 958 SvPOK_only (sv);
734 *SvEND (sv) = 0; 959 *SvEND (sv) = 0;
… …
737 SvUTF8_on (sv); 962 SvUTF8_on (sv);
738 } 963 }
739 else 964 else
740 sv = newSVpvn ("", 0); 965 sv = newSVpvn ("", 0);
741 966
967 dec->cur = dec_cur;
742 return sv; 968 return sv;
743 969
744fail: 970fail:
971 dec->cur = dec_cur;
745 return 0; 972 return 0;
746} 973}
747 974
748static SV * 975static SV *
749decode_num (dec_t *dec) 976decode_num (dec_t *dec)
… …
807 is_nv = 1; 1034 is_nv = 1;
808 } 1035 }
809 1036
810 if (!is_nv) 1037 if (!is_nv)
811 { 1038 {
812 UV uv; 1039 int len = dec->cur - start;
813 int numtype = grok_number (start, dec->cur - start, &uv); 1040
814 if (numtype & IS_NUMBER_IN_UV) 1041 // special case the rather common 1..5-digit-int case
815 if (numtype & IS_NUMBER_NEG) 1042 if (*start == '-')
1043 switch (len)
816 { 1044 {
817 if (uv < (UV)IV_MIN) 1045 case 2: return newSViv (-( start [1] - '0' * 1));
818 return newSViv (-(IV)uv); 1046 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1047 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1048 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1049 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
819 } 1050 }
1051 else
1052 switch (len)
1053 {
1054 case 1: return newSViv ( start [0] - '0' * 1);
1055 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1056 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1057 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1058 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1059 }
1060
1061 {
1062 UV uv;
1063 int numtype = grok_number (start, len, &uv);
1064 if (numtype & IS_NUMBER_IN_UV)
1065 if (numtype & IS_NUMBER_NEG)
1066 {
1067 if (uv < (UV)IV_MIN)
1068 return newSViv (-(IV)uv);
1069 }
820 else 1070 else
821 return newSVuv (uv); 1071 return newSVuv (uv);
822 } 1072 }
823 1073
1074 len -= *start == '-' ? 1 : 0;
1075
1076 // does not fit into IV or UV, try NV
1077 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1078 #if defined (LDBL_DIG)
1079 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1080 #endif
1081 )
1082 // fits into NV without loss of precision
1083 return newSVnv (Atof (start));
1084
1085 // everything else fails, convert it to a string
1086 return newSVpvn (start, dec->cur - start);
1087 }
1088
1089 // loss of precision here
824 return newSVnv (Atof (start)); 1090 return newSVnv (Atof (start));
825 1091
826fail: 1092fail:
827 return 0; 1093 return 0;
828} 1094}
… …
858 1124
859 if (*dec->cur != ',') 1125 if (*dec->cur != ',')
860 ERR (", or ] expected while parsing array"); 1126 ERR (", or ] expected while parsing array");
861 1127
862 ++dec->cur; 1128 ++dec->cur;
1129
1130 decode_ws (dec);
1131
1132 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1133 {
1134 ++dec->cur;
1135 break;
1136 }
863 } 1137 }
864 1138
865 DEC_DEC_DEPTH; 1139 DEC_DEC_DEPTH;
866 return newRV_noinc ((SV *)av); 1140 return newRV_noinc ((SV *)av);
867 1141
… …
872} 1146}
873 1147
874static SV * 1148static SV *
875decode_hv (dec_t *dec) 1149decode_hv (dec_t *dec)
876{ 1150{
1151 SV *sv;
877 HV *hv = newHV (); 1152 HV *hv = newHV ();
878 1153
879 DEC_INC_DEPTH; 1154 DEC_INC_DEPTH;
880 decode_ws (dec); 1155 decode_ws (dec);
881 1156
882 if (*dec->cur == '}') 1157 if (*dec->cur == '}')
883 ++dec->cur; 1158 ++dec->cur;
884 else 1159 else
885 for (;;) 1160 for (;;)
886 { 1161 {
1162 EXPECT_CH ('"');
1163
1164 // heuristic: assume that
1165 // a) decode_str + hv_store_ent are abysmally slow.
1166 // b) most hash keys are short, simple ascii text.
1167 // => try to "fast-match" such strings to avoid
1168 // the overhead of decode_str + hv_store_ent.
1169 {
887 SV *key, *value; 1170 SV *value;
1171 char *p = dec->cur;
1172 char *e = p + 24; // only try up to 24 bytes
888 1173
889 decode_ws (dec); EXPECT_CH ('"'); 1174 for (;;)
890
891 key = decode_str (dec);
892 if (!key)
893 goto fail;
894
895 decode_ws (dec); EXPECT_CH (':');
896
897 value = decode_sv (dec);
898 if (!value)
899 { 1175 {
1176 // the >= 0x80 is true on most architectures
1177 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1178 {
1179 // slow path, back up and use decode_str
1180 SV *key = decode_str (dec);
1181 if (!key)
1182 goto fail;
1183
1184 decode_ws (dec); EXPECT_CH (':');
1185
1186 decode_ws (dec);
1187 value = decode_sv (dec);
1188 if (!value)
1189 {
1190 SvREFCNT_dec (key);
1191 goto fail;
1192 }
1193
1194 hv_store_ent (hv, key, value, 0);
900 SvREFCNT_dec (key); 1195 SvREFCNT_dec (key);
1196
1197 break;
1198 }
1199 else if (*p == '"')
1200 {
1201 // fast path, got a simple key
1202 char *key = dec->cur;
1203 int len = p - key;
1204 dec->cur = p + 1;
1205
1206 decode_ws (dec); EXPECT_CH (':');
1207
1208 decode_ws (dec);
1209 value = decode_sv (dec);
1210 if (!value)
901 goto fail; 1211 goto fail;
1212
1213 hv_store (hv, key, len, value, 0);
1214
1215 break;
1216 }
1217
1218 ++p;
902 } 1219 }
903 1220 }
904 hv_store_ent (hv, key, value, 0);
905 SvREFCNT_dec (key);
906 1221
907 decode_ws (dec); 1222 decode_ws (dec);
908 1223
909 if (*dec->cur == '}') 1224 if (*dec->cur == '}')
910 { 1225 {
… …
914 1229
915 if (*dec->cur != ',') 1230 if (*dec->cur != ',')
916 ERR (", or } expected while parsing object/hash"); 1231 ERR (", or } expected while parsing object/hash");
917 1232
918 ++dec->cur; 1233 ++dec->cur;
1234
1235 decode_ws (dec);
1236
1237 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1238 {
1239 ++dec->cur;
1240 break;
1241 }
919 } 1242 }
920 1243
921 DEC_DEC_DEPTH; 1244 DEC_DEC_DEPTH;
922 return newRV_noinc ((SV *)hv); 1245 sv = newRV_noinc ((SV *)hv);
1246
1247 // check filter callbacks
1248 if (dec->json.flags & F_HOOK)
1249 {
1250 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1251 {
1252 HE *cb, *he;
1253
1254 hv_iterinit (hv);
1255 he = hv_iternext (hv);
1256 hv_iterinit (hv);
1257
1258 // the next line creates a mortal sv each time its called.
1259 // might want to optimise this for common cases.
1260 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1261
1262 if (cb)
1263 {
1264 dSP;
1265 int count;
1266
1267 ENTER; SAVETMPS; PUSHMARK (SP);
1268 XPUSHs (HeVAL (he));
1269
1270 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1271
1272 if (count == 1)
1273 {
1274 sv = newSVsv (POPs);
1275 FREETMPS; LEAVE;
1276 return sv;
1277 }
1278
1279 FREETMPS; LEAVE;
1280 }
1281 }
1282
1283 if (dec->json.cb_object)
1284 {
1285 dSP;
1286 int count;
1287
1288 ENTER; SAVETMPS; PUSHMARK (SP);
1289 XPUSHs (sv_2mortal (sv));
1290
1291 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1292
1293 if (count == 1)
1294 {
1295 sv = newSVsv (POPs);
1296 FREETMPS; LEAVE;
1297 return sv;
1298 }
1299
1300 SvREFCNT_inc (sv);
1301 FREETMPS; LEAVE;
1302 }
1303 }
1304
1305 return sv;
923 1306
924fail: 1307fail:
925 SvREFCNT_dec (hv); 1308 SvREFCNT_dec (hv);
926 DEC_DEC_DEPTH; 1309 DEC_DEC_DEPTH;
927 return 0; 1310 return 0;
928} 1311}
929 1312
930static SV * 1313static SV *
931decode_sv (dec_t *dec) 1314decode_sv (dec_t *dec)
932{ 1315{
933 decode_ws (dec); 1316 // the beauty of JSON: you need exactly one character lookahead
1317 // to parse everything.
934 switch (*dec->cur) 1318 switch (*dec->cur)
935 { 1319 {
936 case '"': ++dec->cur; return decode_str (dec); 1320 case '"': ++dec->cur; return decode_str (dec);
937 case '[': ++dec->cur; return decode_av (dec); 1321 case '[': ++dec->cur; return decode_av (dec);
938 case '{': ++dec->cur; return decode_hv (dec); 1322 case '{': ++dec->cur; return decode_hv (dec);
939 1323
940 case '-': 1324 case '-':
941 case '0': case '1': case '2': case '3': case '4': 1325 case '0': case '1': case '2': case '3': case '4':
942 case '5': case '6': case '7': case '8': case '9': 1326 case '5': case '6': case '7': case '8': case '9':
943 return decode_num (dec); 1327 return decode_num (dec);
944 1328
945 case 't': 1329 case 't':
946 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1330 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
947 { 1331 {
948 dec->cur += 4; 1332 dec->cur += 4;
949 return newSViv (1); 1333#if JSON_SLOW
1334 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1335#endif
1336 return SvREFCNT_inc (json_true);
950 } 1337 }
951 else 1338 else
952 ERR ("'true' expected"); 1339 ERR ("'true' expected");
953 1340
954 break; 1341 break;
955 1342
956 case 'f': 1343 case 'f':
957 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1344 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
958 { 1345 {
959 dec->cur += 5; 1346 dec->cur += 5;
960 return newSViv (0); 1347#if JSON_SLOW
1348 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1349#endif
1350 return SvREFCNT_inc (json_false);
961 } 1351 }
962 else 1352 else
963 ERR ("'false' expected"); 1353 ERR ("'false' expected");
964 1354
965 break; 1355 break;
… …
983fail: 1373fail:
984 return 0; 1374 return 0;
985} 1375}
986 1376
987static SV * 1377static SV *
988decode_json (SV *string, U32 flags, UV *offset_return) 1378decode_json (SV *string, JSON *json, UV *offset_return)
989{ 1379{
990 dec_t dec; 1380 dec_t dec;
991 UV offset; 1381 UV offset;
992 SV *sv; 1382 SV *sv;
993 1383
994 SvGETMAGIC (string); 1384 SvGETMAGIC (string);
995 SvUPGRADE (string, SVt_PV); 1385 SvUPGRADE (string, SVt_PV);
996 1386
1387 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1388 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1389 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1390
997 if (flags & F_UTF8) 1391 if (json->flags & F_UTF8)
998 sv_utf8_downgrade (string, 0); 1392 sv_utf8_downgrade (string, 0);
999 else 1393 else
1000 sv_utf8_upgrade (string); 1394 sv_utf8_upgrade (string);
1001 1395
1002 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1396 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1003 1397
1004 dec.flags = flags; 1398 dec.json = *json;
1005 dec.cur = SvPVX (string); 1399 dec.cur = SvPVX (string);
1006 dec.end = SvEND (string); 1400 dec.end = SvEND (string);
1007 dec.err = 0; 1401 dec.err = 0;
1008 dec.depth = 0; 1402 dec.depth = 0;
1009 dec.maxdepth = DEC_DEPTH (dec.flags); 1403 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1404
1405 if (dec.json.cb_object || dec.json.cb_sk_object)
1406 dec.json.flags |= F_HOOK;
1010 1407
1011 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1408 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1409
1410 decode_ws (&dec);
1012 sv = decode_sv (&dec); 1411 sv = decode_sv (&dec);
1013 1412
1014 if (!(offset_return || !sv)) 1413 if (!(offset_return || !sv))
1015 { 1414 {
1016 // check for trailing garbage 1415 // check for trailing garbage
… …
1024 } 1423 }
1025 } 1424 }
1026 1425
1027 if (offset_return || !sv) 1426 if (offset_return || !sv)
1028 { 1427 {
1029 offset = dec.flags & F_UTF8 1428 offset = dec.json.flags & F_UTF8
1030 ? dec.cur - SvPVX (string) 1429 ? dec.cur - SvPVX (string)
1031 : utf8_distance (dec.cur, SvPVX (string)); 1430 : utf8_distance (dec.cur, SvPVX (string));
1032 1431
1033 if (offset_return) 1432 if (offset_return)
1034 *offset_return = offset; 1433 *offset_return = offset;
… …
1053 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1452 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1054 } 1453 }
1055 1454
1056 sv = sv_2mortal (sv); 1455 sv = sv_2mortal (sv);
1057 1456
1058 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1457 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1059 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1458 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1060 1459
1061 return sv; 1460 return sv;
1062} 1461}
1063 1462
… …
1067MODULE = JSON::XS PACKAGE = JSON::XS 1466MODULE = JSON::XS PACKAGE = JSON::XS
1068 1467
1069BOOT: 1468BOOT:
1070{ 1469{
1071 int i; 1470 int i;
1072
1073 memset (decode_hexdigit, 0xff, 256);
1074 1471
1075 for (i = 0; i < 256; ++i) 1472 for (i = 0; i < 256; ++i)
1076 decode_hexdigit [i] = 1473 decode_hexdigit [i] =
1077 i >= '0' && i <= '9' ? i - '0' 1474 i >= '0' && i <= '9' ? i - '0'
1078 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1475 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1079 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1476 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1080 : -1; 1477 : -1;
1081 1478
1082 json_stash = gv_stashpv ("JSON::XS", 1); 1479 json_stash = gv_stashpv ("JSON::XS" , 1);
1480 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1481
1482 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1483 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1083} 1484}
1084 1485
1085PROTOTYPES: DISABLE 1486PROTOTYPES: DISABLE
1086 1487
1087SV *new (char *dummy) 1488void CLONE (...)
1088 CODE: 1489 CODE:
1089 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1490 json_stash = 0;
1491 json_boolean_stash = 0;
1492
1493void new (char *klass)
1494 PPCODE:
1495{
1496 SV *pv = NEWSV (0, sizeof (JSON));
1497 SvPOK_only (pv);
1498 Zero (SvPVX (pv), 1, JSON);
1499 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1500 XPUSHs (sv_2mortal (sv_bless (
1501 newRV_noinc (pv),
1502 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1503 )));
1504}
1505
1506void ascii (JSON *self, int enable = 1)
1507 ALIAS:
1508 ascii = F_ASCII
1509 latin1 = F_LATIN1
1510 utf8 = F_UTF8
1511 indent = F_INDENT
1512 canonical = F_CANONICAL
1513 space_before = F_SPACE_BEFORE
1514 space_after = F_SPACE_AFTER
1515 pretty = F_PRETTY
1516 allow_nonref = F_ALLOW_NONREF
1517 shrink = F_SHRINK
1518 allow_blessed = F_ALLOW_BLESSED
1519 convert_blessed = F_CONV_BLESSED
1520 relaxed = F_RELAXED
1521 PPCODE:
1522{
1523 if (enable)
1524 self->flags |= ix;
1525 else
1526 self->flags &= ~ix;
1527
1528 XPUSHs (ST (0));
1529}
1530
1531void get_ascii (JSON *self)
1532 ALIAS:
1533 get_ascii = F_ASCII
1534 get_latin1 = F_LATIN1
1535 get_utf8 = F_UTF8
1536 get_indent = F_INDENT
1537 get_canonical = F_CANONICAL
1538 get_space_before = F_SPACE_BEFORE
1539 get_space_after = F_SPACE_AFTER
1540 get_allow_nonref = F_ALLOW_NONREF
1541 get_shrink = F_SHRINK
1542 get_allow_blessed = F_ALLOW_BLESSED
1543 get_convert_blessed = F_CONV_BLESSED
1544 get_relaxed = F_RELAXED
1545 PPCODE:
1546 XPUSHs (boolSV (self->flags & ix));
1547
1548void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1549 PPCODE:
1550{
1551 UV log2 = 0;
1552
1553 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1554
1555 while ((1UL << log2) < max_depth)
1556 ++log2;
1557
1558 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1559
1560 XPUSHs (ST (0));
1561}
1562
1563U32 get_max_depth (JSON *self)
1564 CODE:
1565 RETVAL = DEC_DEPTH (self->flags);
1090 OUTPUT: 1566 OUTPUT:
1091 RETVAL 1567 RETVAL
1092 1568
1093SV *ascii (SV *self, int enable = 1) 1569void max_size (JSON *self, UV max_size = 0)
1094 ALIAS: 1570 PPCODE:
1095 ascii = F_ASCII 1571{
1096 latin1 = F_LATIN1 1572 UV log2 = 0;
1097 utf8 = F_UTF8 1573
1098 indent = F_INDENT 1574 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1099 canonical = F_CANONICAL 1575 if (max_size == 1) max_size = 2;
1100 space_before = F_SPACE_BEFORE 1576
1101 space_after = F_SPACE_AFTER 1577 while ((1UL << log2) < max_size)
1102 pretty = F_PRETTY 1578 ++log2;
1103 allow_nonref = F_ALLOW_NONREF 1579
1104 shrink = F_SHRINK 1580 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1581
1582 XPUSHs (ST (0));
1583}
1584
1585int get_max_size (JSON *self)
1105 CODE: 1586 CODE:
1106{ 1587 RETVAL = DEC_SIZE (self->flags);
1107 UV *uv = SvJSON (self);
1108 if (enable)
1109 *uv |= ix;
1110 else
1111 *uv &= ~ix;
1112
1113 RETVAL = newSVsv (self);
1114}
1115 OUTPUT: 1588 OUTPUT:
1116 RETVAL 1589 RETVAL
1117 1590
1118SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1591void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1119 CODE:
1120{
1121 UV *uv = SvJSON (self);
1122 UV log2 = 0;
1123
1124 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1125
1126 while ((1UL << log2) < max_depth)
1127 ++log2;
1128
1129 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1130
1131 RETVAL = newSVsv (self);
1132}
1133 OUTPUT:
1134 RETVAL
1135
1136void encode (SV *self, SV *scalar)
1137 PPCODE: 1592 PPCODE:
1138 XPUSHs (encode_json (scalar, *SvJSON (self))); 1593{
1594 SvREFCNT_dec (self->cb_object);
1595 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1139 1596
1140void decode (SV *self, SV *jsonstr) 1597 XPUSHs (ST (0));
1598}
1599
1600void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1141 PPCODE: 1601 PPCODE:
1602{
1603 if (!self->cb_sk_object)
1604 self->cb_sk_object = newHV ();
1605
1606 if (SvOK (cb))
1607 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1608 else
1609 {
1610 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1611
1612 if (!HvKEYS (self->cb_sk_object))
1613 {
1614 SvREFCNT_dec (self->cb_sk_object);
1615 self->cb_sk_object = 0;
1616 }
1617 }
1618
1619 XPUSHs (ST (0));
1620}
1621
1622void encode (JSON *self, SV *scalar)
1623 PPCODE:
1624 XPUSHs (encode_json (scalar, self));
1625
1626void decode (JSON *self, SV *jsonstr)
1627 PPCODE:
1142 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1628 XPUSHs (decode_json (jsonstr, self, 0));
1143 1629
1144void decode_prefix (SV *self, SV *jsonstr) 1630void decode_prefix (JSON *self, SV *jsonstr)
1145 PPCODE: 1631 PPCODE:
1146{ 1632{
1147 UV offset; 1633 UV offset;
1148 EXTEND (SP, 2); 1634 EXTEND (SP, 2);
1149 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1635 PUSHs (decode_json (jsonstr, self, &offset));
1150 PUSHs (sv_2mortal (newSVuv (offset))); 1636 PUSHs (sv_2mortal (newSVuv (offset)));
1151} 1637}
1152 1638
1639void DESTROY (JSON *self)
1640 CODE:
1641 SvREFCNT_dec (self->cb_sk_object);
1642 SvREFCNT_dec (self->cb_object);
1643
1153PROTOTYPES: ENABLE 1644PROTOTYPES: ENABLE
1154 1645
1155void to_json (SV *scalar) 1646void encode_json (SV *scalar)
1156 ALIAS:
1157 objToJson = 0
1158 PPCODE: 1647 PPCODE:
1648{
1649 JSON json = { F_DEFAULT | F_UTF8 };
1159 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1650 XPUSHs (encode_json (scalar, &json));
1651}
1160 1652
1161void from_json (SV *jsonstr) 1653void decode_json (SV *jsonstr)
1162 ALIAS:
1163 jsonToObj = 0
1164 PPCODE: 1654 PPCODE:
1655{
1656 JSON json = { F_DEFAULT | F_UTF8 };
1165 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1657 XPUSHs (decode_json (jsonstr, &json, 0));
1658}
1166 1659

Diff Legend

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