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 1.58 by root, Mon Aug 13 16:06:25 2007 UTC vs.
Revision 1.106 by root, Tue Jan 19 01:36:34 2010 UTC

… …
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>
9#include <float.h> 10#include <float.h>
10 11
11#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 14#endif
14 15
15// some old perls do not have this, try to make it work, no 16// 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// guarantees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
19#endif 20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
20 24
21#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
27#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
28#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
32#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL 36#define F_RELAXED 0x00001000UL
35#define S_MAXSIZE 20 37#define F_ALLOW_UNKNOWN 0x00002000UL
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37 39
38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
40
41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
42#define F_DEFAULT (9UL << S_MAXDEPTH)
43 41
44#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
45#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
46 44
47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
48 46
49#define SB do { 47#define SB do {
50#define SE } while (0) 48#define SE } while (0)
51 49
52#if __GNUC__ >= 3 50#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value)) 51# define expect(expr,value) __builtin_expect ((expr), (value))
54# define inline inline 52# define INLINE static inline
55#else 53#else
56# define expect(expr,value) (expr) 54# define expect(expr,value) (expr)
57# define inline static 55# define INLINE static
58#endif 56#endif
59 57
60#define expect_false(expr) expect ((expr) != 0, 0) 58#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1) 59#define expect_true(expr) expect ((expr) != 0, 1)
62 60
61#define IN_RANGE_INC(type,val,beg,end) \
62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
63#ifdef USE_ITHREADS 67#ifdef USE_ITHREADS
64# define JSON_SLOW 1 68# define JSON_SLOW 1
69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
65#else 70#else
66# define JSON_SLOW 0 71# define JSON_SLOW 0
72# define JSON_STASH json_stash
67#endif 73#endif
68 74
69static HV *json_stash, *json_boolean_stash; // JSON::XS:: 75static HV *json_stash, *json_boolean_stash; // JSON::XS::
70static SV *json_true, *json_false; 76static SV *json_true, *json_false;
71 77
78enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash
82 INCR_M_C0, // inside comment in initial whitespace sequence
83 INCR_M_C1, // inside comment in other places
84 INCR_M_JSON // outside anything, count nesting
85};
86
87#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
88
72typedef struct { 89typedef struct {
73 U32 flags; 90 U32 flags;
91 U32 max_depth;
92 STRLEN max_size;
93
74 SV *cb_object; 94 SV *cb_object;
75 HV *cb_sk_object; 95 HV *cb_sk_object;
96
97 // for the incremental parser
98 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level
101 unsigned char incr_mode;
76} JSON; 102} JSON;
103
104INLINE void
105json_init (JSON *json)
106{
107 Zero (json, 1, JSON);
108 json->max_depth = 512;
109}
77 110
78///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
79// utility functions 112// utility functions
80 113
81inline void 114INLINE SV *
115get_bool (const char *name)
116{
117 SV *sv = get_sv (name, 1);
118
119 SvREADONLY_on (sv);
120 SvREADONLY_on (SvRV (sv));
121
122 return sv;
123}
124
125INLINE void
82shrink (SV *sv) 126shrink (SV *sv)
83{ 127{
84 sv_utf8_downgrade (sv, 1); 128 sv_utf8_downgrade (sv, 1);
129
85 if (SvLEN (sv) > SvCUR (sv) + 1) 130 if (SvLEN (sv) > SvCUR (sv) + 1)
86 { 131 {
87#ifdef SvPV_shrink_to_cur 132#ifdef SvPV_shrink_to_cur
88 SvPV_shrink_to_cur (sv); 133 SvPV_shrink_to_cur (sv);
89#elif defined (SvPV_renew) 134#elif defined (SvPV_renew)
… …
95// decode an utf-8 character and return it, or (UV)-1 in 140// decode an utf-8 character and return it, or (UV)-1 in
96// case of an error. 141// case of an error.
97// we special-case "safe" characters from U+80 .. U+7FF, 142// we special-case "safe" characters from U+80 .. U+7FF,
98// but use the very good perl function to parse anything else. 143// but use the very good perl function to parse anything else.
99// note that we never call this function for a ascii codepoints 144// note that we never call this function for a ascii codepoints
100inline UV 145INLINE UV
101decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 146decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
102{ 147{
103 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 148 if (expect_true (len >= 2
104 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 149 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
105 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 150 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
106 { 151 {
107 *clen = 2; 152 *clen = 2;
108 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 153 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
109 } 154 }
110 else 155 else
111 { 156 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
112 *clen = (STRLEN)-1; 157}
113 return (UV)-1; 158
159// likewise for encoding, also never called for ascii codepoints
160// this function takes advantage of this fact, although current gccs
161// seem to optimise the check for >= 0x80 away anyways
162INLINE unsigned char *
163encode_utf8 (unsigned char *s, UV ch)
164{
165 if (expect_false (ch < 0x000080))
166 *s++ = ch;
167 else if (expect_true (ch < 0x000800))
168 *s++ = 0xc0 | ( ch >> 6),
169 *s++ = 0x80 | ( ch & 0x3f);
170 else if ( ch < 0x010000)
171 *s++ = 0xe0 | ( ch >> 12),
172 *s++ = 0x80 | ((ch >> 6) & 0x3f),
173 *s++ = 0x80 | ( ch & 0x3f);
174 else if ( ch < 0x110000)
175 *s++ = 0xf0 | ( ch >> 18),
176 *s++ = 0x80 | ((ch >> 12) & 0x3f),
177 *s++ = 0x80 | ((ch >> 6) & 0x3f),
178 *s++ = 0x80 | ( ch & 0x3f);
179
180 return s;
181}
182
183// convert offset pointer to character index, sv must be string
184static STRLEN
185ptr_to_index (SV *sv, char *offset)
186{
187 return SvUTF8 (sv)
188 ? utf8_distance (offset, SvPVX (sv))
189 : offset - SvPVX (sv);
190}
191
192/////////////////////////////////////////////////////////////////////////////
193// fp hell
194
195// scan a group of digits, and a trailing exponent
196static void
197json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
198{
199 UV uaccum = 0;
200 int eaccum = 0;
201
202 // if we recurse too deep, skip all remaining digits
203 // to avoid a stack overflow attack
204 if (expect_false (--maxdepth <= 0))
205 while (((U8)*s - '0') < 10)
206 ++s;
207
208 for (;;)
114 } 209 {
115} 210 U8 dig = (U8)*s - '0';
116 211
212 if (expect_false (dig >= 10))
213 {
214 if (dig == (U8)((U8)'.' - (U8)'0'))
215 {
216 ++s;
217 json_atof_scan1 (s, accum, expo, 1, maxdepth);
218 }
219 else if ((dig | ' ') == 'e' - '0')
220 {
221 int exp2 = 0;
222 int neg = 0;
223
224 ++s;
225
226 if (*s == '-')
227 {
228 ++s;
229 neg = 1;
230 }
231 else if (*s == '+')
232 ++s;
233
234 while ((dig = (U8)*s - '0') < 10)
235 exp2 = exp2 * 10 + *s++ - '0';
236
237 *expo += neg ? -exp2 : exp2;
238 }
239
240 break;
241 }
242
243 ++s;
244
245 uaccum = uaccum * 10 + dig;
246 ++eaccum;
247
248 // if we have too many digits, then recurse for more
249 // we actually do this for rather few digits
250 if (uaccum >= (UV_MAX - 9) / 10)
251 {
252 if (postdp) *expo -= eaccum;
253 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
254 if (postdp) *expo += eaccum;
255
256 break;
257 }
258 }
259
260 // this relies greatly on the quality of the pow ()
261 // implementation of the platform, but a good
262 // implementation is hard to beat.
263 if (postdp) *expo -= eaccum;
264 *accum += uaccum * Perl_pow (10., *expo);
265 *expo += eaccum;
266}
267
268static NV
269json_atof (const char *s)
270{
271 NV accum = 0.;
272 int expo = 0;
273 int neg = 0;
274
275 if (*s == '-')
276 {
277 ++s;
278 neg = 1;
279 }
280
281 // a recursion depth of ten gives us >>500 bits
282 json_atof_scan1 (s, &accum, &expo, 0, 10);
283
284 return neg ? -accum : accum;
285}
117///////////////////////////////////////////////////////////////////////////// 286/////////////////////////////////////////////////////////////////////////////
118// encoder 287// encoder
119 288
120// structure used for encoding JSON 289// structure used for encoding JSON
121typedef struct 290typedef struct
… …
123 char *cur; // SvPVX (sv) + current output position 292 char *cur; // SvPVX (sv) + current output position
124 char *end; // SvEND (sv) 293 char *end; // SvEND (sv)
125 SV *sv; // result scalar 294 SV *sv; // result scalar
126 JSON json; 295 JSON json;
127 U32 indent; // indentation level 296 U32 indent; // indentation level
128 U32 maxdepth; // max. indentation/recursion level 297 UV limit; // escape character values >= this value when encoding
129} enc_t; 298} enc_t;
130 299
131inline void 300INLINE void
132need (enc_t *enc, STRLEN len) 301need (enc_t *enc, STRLEN len)
133{ 302{
134 if (expect_false (enc->cur + len >= enc->end)) 303 if (expect_false (enc->cur + len >= enc->end))
135 { 304 {
136 STRLEN cur = enc->cur - SvPVX (enc->sv); 305 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
137 SvGROW (enc->sv, cur + len + 1); 306 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
138 enc->cur = SvPVX (enc->sv) + cur; 307 enc->cur = SvPVX (enc->sv) + cur;
139 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 308 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
140 } 309 }
141} 310}
142 311
143inline void 312INLINE void
144encode_ch (enc_t *enc, char ch) 313encode_ch (enc_t *enc, char ch)
145{ 314{
146 need (enc, 1); 315 need (enc, 1);
147 *enc->cur++ = ch; 316 *enc->cur++ = ch;
148} 317}
… …
202 { 371 {
203 uch = ch; 372 uch = ch;
204 clen = 1; 373 clen = 1;
205 } 374 }
206 375
207 if (uch > 0x10FFFFUL) 376 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
208 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
209
210 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
211 { 377 {
212 if (uch > 0xFFFFUL) 378 if (uch >= 0x10000UL)
213 { 379 {
380 if (uch >= 0x110000UL)
381 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
382
214 need (enc, len += 11); 383 need (enc, len += 11);
215 sprintf (enc->cur, "\\u%04x\\u%04x", 384 sprintf (enc->cur, "\\u%04x\\u%04x",
216 (int)((uch - 0x10000) / 0x400 + 0xD800), 385 (int)((uch - 0x10000) / 0x400 + 0xD800),
217 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 386 (int)((uch - 0x10000) % 0x400 + 0xDC00));
218 enc->cur += 12; 387 enc->cur += 12;
219 } 388 }
220 else 389 else
221 { 390 {
222 static char hexdigit [16] = "0123456789abcdef";
223 need (enc, len += 5); 391 need (enc, len += 5);
224 *enc->cur++ = '\\'; 392 *enc->cur++ = '\\';
225 *enc->cur++ = 'u'; 393 *enc->cur++ = 'u';
226 *enc->cur++ = hexdigit [ uch >> 12 ]; 394 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
227 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 395 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
228 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 396 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
229 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 397 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
230 } 398 }
231 399
232 str += clen; 400 str += clen;
233 } 401 }
234 else if (enc->json.flags & F_LATIN1) 402 else if (enc->json.flags & F_LATIN1)
… …
246 while (--clen); 414 while (--clen);
247 } 415 }
248 else 416 else
249 { 417 {
250 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 418 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
251 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 419 enc->cur = encode_utf8 (enc->cur, uch);
252 ++str; 420 ++str;
253 } 421 }
254 } 422 }
255 } 423 }
256 } 424 }
257 425
258 --len; 426 --len;
259 } 427 }
260} 428}
261 429
262inline void 430INLINE void
263encode_indent (enc_t *enc) 431encode_indent (enc_t *enc)
264{ 432{
265 if (enc->json.flags & F_INDENT) 433 if (enc->json.flags & F_INDENT)
266 { 434 {
267 int spaces = enc->indent * INDENT_STEP; 435 int spaces = enc->indent * INDENT_STEP;
… …
270 memset (enc->cur, ' ', spaces); 438 memset (enc->cur, ' ', spaces);
271 enc->cur += spaces; 439 enc->cur += spaces;
272 } 440 }
273} 441}
274 442
275inline void 443INLINE void
276encode_space (enc_t *enc) 444encode_space (enc_t *enc)
277{ 445{
278 need (enc, 1); 446 need (enc, 1);
279 encode_ch (enc, ' '); 447 encode_ch (enc, ' ');
280} 448}
281 449
282inline void 450INLINE void
283encode_nl (enc_t *enc) 451encode_nl (enc_t *enc)
284{ 452{
285 if (enc->json.flags & F_INDENT) 453 if (enc->json.flags & F_INDENT)
286 { 454 {
287 need (enc, 1); 455 need (enc, 1);
288 encode_ch (enc, '\n'); 456 encode_ch (enc, '\n');
289 } 457 }
290} 458}
291 459
292inline void 460INLINE void
293encode_comma (enc_t *enc) 461encode_comma (enc_t *enc)
294{ 462{
295 encode_ch (enc, ','); 463 encode_ch (enc, ',');
296 464
297 if (enc->json.flags & F_INDENT) 465 if (enc->json.flags & F_INDENT)
… …
305static void 473static void
306encode_av (enc_t *enc, AV *av) 474encode_av (enc_t *enc, AV *av)
307{ 475{
308 int i, len = av_len (av); 476 int i, len = av_len (av);
309 477
310 if (enc->indent >= enc->maxdepth) 478 if (enc->indent >= enc->json.max_depth)
311 croak ("data structure too deep (hit recursion limit)"); 479 croak (ERR_NESTING_EXCEEDED);
312 480
313 encode_ch (enc, '['); encode_nl (enc); 481 encode_ch (enc, '[');
314 ++enc->indent; 482
483 if (len >= 0)
484 {
485 encode_nl (enc); ++enc->indent;
315 486
316 for (i = 0; i <= len; ++i) 487 for (i = 0; i <= len; ++i)
317 { 488 {
318 SV **svp = av_fetch (av, i, 0); 489 SV **svp = av_fetch (av, i, 0);
319 490
320 encode_indent (enc); 491 encode_indent (enc);
321 492
322 if (svp) 493 if (svp)
323 encode_sv (enc, *svp); 494 encode_sv (enc, *svp);
324 else 495 else
325 encode_str (enc, "null", 4, 0); 496 encode_str (enc, "null", 4, 0);
326 497
327 if (i < len) 498 if (i < len)
328 encode_comma (enc); 499 encode_comma (enc);
329 } 500 }
330 501
502 encode_nl (enc); --enc->indent; encode_indent (enc);
503 }
504
331 encode_nl (enc); 505 encode_ch (enc, ']');
332
333 --enc->indent;
334 encode_indent (enc); encode_ch (enc, ']');
335} 506}
336 507
337static void 508static void
338encode_he (enc_t *enc, HE *he) 509encode_hk (enc_t *enc, HE *he)
339{ 510{
340 encode_ch (enc, '"'); 511 encode_ch (enc, '"');
341 512
342 if (HeKLEN (he) == HEf_SVKEY) 513 if (HeKLEN (he) == HEf_SVKEY)
343 { 514 {
… …
356 encode_ch (enc, '"'); 527 encode_ch (enc, '"');
357 528
358 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 529 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
359 encode_ch (enc, ':'); 530 encode_ch (enc, ':');
360 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 531 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
361 encode_sv (enc, HeVAL (he));
362} 532}
363 533
364// compare hash entries, used when all keys are bytestrings 534// compare hash entries, used when all keys are bytestrings
365static int 535static int
366he_cmp_fast (const void *a_, const void *b_) 536he_cmp_fast (const void *a_, const void *b_)
… …
371 HE *b = *(HE **)b_; 541 HE *b = *(HE **)b_;
372 542
373 STRLEN la = HeKLEN (a); 543 STRLEN la = HeKLEN (a);
374 STRLEN lb = HeKLEN (b); 544 STRLEN lb = HeKLEN (b);
375 545
376 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 546 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
377 cmp = la - lb; 547 cmp = lb - la;
378 548
379 return cmp; 549 return cmp;
380} 550}
381 551
382// compare hash entries, used when some keys are sv's or utf-x 552// compare hash entries, used when some keys are sv's or utf-x
383static int 553static int
384he_cmp_slow (const void *a, const void *b) 554he_cmp_slow (const void *a, const void *b)
385{ 555{
386 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 556 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
387} 557}
388 558
389static void 559static void
390encode_hv (enc_t *enc, HV *hv) 560encode_hv (enc_t *enc, HV *hv)
391{ 561{
392 int count, i; 562 HE *he;
393 563
394 if (enc->indent >= enc->maxdepth) 564 if (enc->indent >= enc->json.max_depth)
395 croak ("data structure too deep (hit recursion limit)"); 565 croak (ERR_NESTING_EXCEEDED);
396 566
397 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 567 encode_ch (enc, '{');
398 568
399 if ((count = hv_iterinit (hv)))
400 {
401 // for canonical output we have to sort by keys first 569 // for canonical output we have to sort by keys first
402 // actually, this is mostly due to the stupid so-called 570 // actually, this is mostly due to the stupid so-called
403 // security workaround added somewhere in 5.8.x. 571 // security workaround added somewhere in 5.8.x
404 // that randomises hash orderings 572 // that randomises hash orderings
405 if (enc->json.flags & F_CANONICAL) 573 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
574 {
575 int count = hv_iterinit (hv);
576
577 if (SvMAGICAL (hv))
406 { 578 {
579 // need to count by iterating. could improve by dynamically building the vector below
580 // but I don't care for the speed of this special case.
581 // note also that we will run into undefined behaviour when the two iterations
582 // do not result in the same count, something I might care for in some later release.
583
584 count = 0;
585 while (hv_iternext (hv))
586 ++count;
587
588 hv_iterinit (hv);
589 }
590
591 if (count)
592 {
407 int fast = 1; 593 int i, fast = 1;
408 HE *he;
409#if defined(__BORLANDC__) || defined(_MSC_VER) 594#if defined(__BORLANDC__) || defined(_MSC_VER)
410 HE **hes = _alloca (count * sizeof (HE)); 595 HE **hes = _alloca (count * sizeof (HE));
411#else 596#else
412 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 597 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
413#endif 598#endif
… …
440 625
441 FREETMPS; 626 FREETMPS;
442 LEAVE; 627 LEAVE;
443 } 628 }
444 629
445 for (i = 0; i < count; ++i) 630 encode_nl (enc); ++enc->indent;
631
632 while (count--)
446 { 633 {
447 encode_indent (enc); 634 encode_indent (enc);
635 he = hes [count];
448 encode_he (enc, hes [i]); 636 encode_hk (enc, he);
637 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
449 638
450 if (i < count - 1) 639 if (count)
451 encode_comma (enc); 640 encode_comma (enc);
452 } 641 }
453 642
454 encode_nl (enc); 643 encode_nl (enc); --enc->indent; encode_indent (enc);
455 } 644 }
645 }
456 else 646 else
647 {
648 if (hv_iterinit (hv) || SvMAGICAL (hv))
649 if ((he = hv_iternext (hv)))
457 { 650 {
458 HE *he = hv_iternext (hv); 651 encode_nl (enc); ++enc->indent;
459 652
460 for (;;) 653 for (;;)
461 { 654 {
462 encode_indent (enc); 655 encode_indent (enc);
463 encode_he (enc, he); 656 encode_hk (enc, he);
657 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
464 658
465 if (!(he = hv_iternext (hv))) 659 if (!(he = hv_iternext (hv)))
466 break; 660 break;
467 661
468 encode_comma (enc); 662 encode_comma (enc);
469 } 663 }
470 664
471 encode_nl (enc); 665 encode_nl (enc); --enc->indent; encode_indent (enc);
472 } 666 }
473 } 667 }
474 668
475 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 669 encode_ch (enc, '}');
476} 670}
477 671
478// encode objects, arrays and special \0=false and \1=true values. 672// encode objects, arrays and special \0=false and \1=true values.
479static void 673static void
480encode_rv (enc_t *enc, SV *sv) 674encode_rv (enc_t *enc, SV *sv)
… …
557 751
558 if (len == 1 && *pv == '1') 752 if (len == 1 && *pv == '1')
559 encode_str (enc, "true", 4, 0); 753 encode_str (enc, "true", 4, 0);
560 else if (len == 1 && *pv == '0') 754 else if (len == 1 && *pv == '0')
561 encode_str (enc, "false", 5, 0); 755 encode_str (enc, "false", 5, 0);
756 else if (enc->json.flags & F_ALLOW_UNKNOWN)
757 encode_str (enc, "null", 4, 0);
562 else 758 else
563 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 759 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
564 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 760 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
565 } 761 }
762 else if (enc->json.flags & F_ALLOW_UNKNOWN)
763 encode_str (enc, "null", 4, 0);
566 else 764 else
567 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 765 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
568 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 766 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
569} 767}
570 768
… …
588 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 786 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
589 enc->cur += strlen (enc->cur); 787 enc->cur += strlen (enc->cur);
590 } 788 }
591 else if (SvIOKp (sv)) 789 else if (SvIOKp (sv))
592 { 790 {
593 // we assume we can always read an IV as a UV 791 // we assume we can always read an IV as a UV and vice versa
594 if (SvUV (sv) & ~(UV)0x7fff) 792 // we assume two's complement
595 { 793 // we assume no aliasing issues in the union
596 // large integer, use the (rather slow) snprintf way. 794 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
597 need (enc, sizeof (UV) * 3); 795 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
598 enc->cur +=
599 SvIsUV(sv)
600 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
601 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
602 }
603 else
604 { 796 {
605 // optimise the "small number case" 797 // optimise the "small number case"
606 // code will likely be branchless and use only a single multiplication 798 // code will likely be branchless and use only a single multiplication
799 // works for numbers up to 59074
607 I32 i = SvIV (sv); 800 I32 i = SvIVX (sv);
608 U32 u; 801 U32 u;
609 char digit, nz = 0; 802 char digit, nz = 0;
610 803
611 need (enc, 6); 804 need (enc, 6);
612 805
… …
618 811
619 // now output digit by digit, each time masking out the integer part 812 // now output digit by digit, each time masking out the integer part
620 // and multiplying by 5 while moving the decimal point one to the right, 813 // and multiplying by 5 while moving the decimal point one to the right,
621 // resulting in a net multiplication by 10. 814 // resulting in a net multiplication by 10.
622 // we always write the digit to memory but conditionally increment 815 // we always write the digit to memory but conditionally increment
623 // the pointer, to ease the usage of conditional move instructions. 816 // the pointer, to enable the use of conditional move instructions.
624 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 817 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
625 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 818 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
626 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 819 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
627 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 820 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
628 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 821 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
629 } 822 }
823 else
824 {
825 // large integer, use the (rather slow) snprintf way.
826 need (enc, IVUV_MAXCHARS);
827 enc->cur +=
828 SvIsUV(sv)
829 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
830 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
831 }
630 } 832 }
631 else if (SvROK (sv)) 833 else if (SvROK (sv))
632 encode_rv (enc, SvRV (sv)); 834 encode_rv (enc, SvRV (sv));
633 else if (!SvOK (sv)) 835 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
634 encode_str (enc, "null", 4, 0); 836 encode_str (enc, "null", 4, 0);
635 else 837 else
636 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 838 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
637 SvPV_nolen (sv), SvFLAGS (sv)); 839 SvPV_nolen (sv), SvFLAGS (sv));
638} 840}
… …
648 enc.json = *json; 850 enc.json = *json;
649 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 851 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
650 enc.cur = SvPVX (enc.sv); 852 enc.cur = SvPVX (enc.sv);
651 enc.end = SvEND (enc.sv); 853 enc.end = SvEND (enc.sv);
652 enc.indent = 0; 854 enc.indent = 0;
653 enc.maxdepth = DEC_DEPTH (enc.json.flags); 855 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
856 : enc.json.flags & F_LATIN1 ? 0x000100UL
857 : 0x110000UL;
654 858
655 SvPOK_only (enc.sv); 859 SvPOK_only (enc.sv);
656 encode_sv (&enc, scalar); 860 encode_sv (&enc, scalar);
861 encode_nl (&enc);
657 862
658 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 863 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
659 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 864 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
660 865
661 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 866 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
679 JSON json; 884 JSON json;
680 U32 depth; // recursion depth 885 U32 depth; // recursion depth
681 U32 maxdepth; // recursion depth limit 886 U32 maxdepth; // recursion depth limit
682} dec_t; 887} dec_t;
683 888
684inline void 889INLINE void
890decode_comment (dec_t *dec)
891{
892 // only '#'-style comments allowed a.t.m.
893
894 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
895 ++dec->cur;
896}
897
898INLINE void
685decode_ws (dec_t *dec) 899decode_ws (dec_t *dec)
686{ 900{
687 for (;;) 901 for (;;)
688 { 902 {
689 char ch = *dec->cur; 903 char ch = *dec->cur;
690 904
691 if (ch > 0x20 905 if (ch > 0x20)
906 {
907 if (expect_false (ch == '#'))
908 {
909 if (dec->json.flags & F_RELAXED)
910 decode_comment (dec);
911 else
912 break;
913 }
914 else
915 break;
916 }
692 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 917 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
693 break; 918 break; // parse error, but let higher level handle it, gives better error messages
694 919
695 ++dec->cur; 920 ++dec->cur;
696 } 921 }
697} 922}
698 923
… …
702 if (*dec->cur != ch) \ 927 if (*dec->cur != ch) \
703 ERR (# ch " expected"); \ 928 ERR (# ch " expected"); \
704 ++dec->cur; \ 929 ++dec->cur; \
705 SE 930 SE
706 931
707#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 932#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
708#define DEC_DEC_DEPTH --dec->depth 933#define DEC_DEC_DEPTH --dec->depth
709 934
710static SV *decode_sv (dec_t *dec); 935static SV *decode_sv (dec_t *dec);
711 936
712static signed char decode_hexdigit[256]; 937static signed char decode_hexdigit[256];
… …
804 1029
805 if (hi >= 0x80) 1030 if (hi >= 0x80)
806 { 1031 {
807 utf8 = 1; 1032 utf8 = 1;
808 1033
809 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1034 cur = encode_utf8 (cur, hi);
810 } 1035 }
811 else 1036 else
812 *cur++ = hi; 1037 *cur++ = hi;
813 } 1038 }
814 break; 1039 break;
… …
816 default: 1041 default:
817 --dec_cur; 1042 --dec_cur;
818 ERR ("illegal backslash escape sequence in string"); 1043 ERR ("illegal backslash escape sequence in string");
819 } 1044 }
820 } 1045 }
821 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1046 else if (expect_true (ch >= 0x20 && ch < 0x80))
822 *cur++ = ch; 1047 *cur++ = ch;
823 else if (ch >= 0x80) 1048 else if (ch >= 0x80)
824 { 1049 {
825 STRLEN clen; 1050 STRLEN clen;
826 UV uch;
827 1051
828 --dec_cur; 1052 --dec_cur;
829 1053
830 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1054 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
831 if (clen == (STRLEN)-1) 1055 if (clen == (STRLEN)-1)
832 ERR ("malformed UTF-8 character in JSON string"); 1056 ERR ("malformed UTF-8 character in JSON string");
833 1057
834 do 1058 do
835 *cur++ = *dec_cur++; 1059 *cur++ = *dec_cur++;
… …
852 { 1076 {
853 STRLEN len = cur - buf; 1077 STRLEN len = cur - buf;
854 1078
855 if (sv) 1079 if (sv)
856 { 1080 {
857 SvGROW (sv, SvCUR (sv) + len + 1); 1081 STRLEN cur = SvCUR (sv);
1082
1083 if (SvLEN (sv) <= cur + len)
1084 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1085
858 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1086 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
859 SvCUR_set (sv, SvCUR (sv) + len); 1087 SvCUR_set (sv, SvCUR (sv) + len);
860 } 1088 }
861 else 1089 else
862 sv = newSVpvn (buf, len); 1090 sv = newSVpvn (buf, len);
… …
949 1177
950 if (!is_nv) 1178 if (!is_nv)
951 { 1179 {
952 int len = dec->cur - start; 1180 int len = dec->cur - start;
953 1181
954 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1182 // special case the rather common 1..5-digit-int case
955 if (*start == '-') 1183 if (*start == '-')
956 switch (len) 1184 switch (len)
957 { 1185 {
958 case 2: return newSViv (-( start [1] - '0' * 1)); 1186 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
959 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1187 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
960 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1188 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
961 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1189 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1190 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
962 } 1191 }
963 else 1192 else
964 switch (len) 1193 switch (len)
965 { 1194 {
966 case 1: return newSViv ( start [0] - '0' * 1); 1195 case 1: return newSViv ( start [0] - '0' * 1);
967 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1196 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
968 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1197 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
969 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1198 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1199 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
970 } 1200 }
971 1201
972 { 1202 {
973 UV uv; 1203 UV uv;
974 int numtype = grok_number (start, len, &uv); 1204 int numtype = grok_number (start, len, &uv);
… …
983 } 1213 }
984 1214
985 len -= *start == '-' ? 1 : 0; 1215 len -= *start == '-' ? 1 : 0;
986 1216
987 // does not fit into IV or UV, try NV 1217 // does not fit into IV or UV, try NV
988 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1218 if (len <= NV_DIG)
989 #if defined (LDBL_DIG)
990 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
991 #endif
992 )
993 // fits into NV without loss of precision 1219 // fits into NV without loss of precision
994 return newSVnv (Atof (start)); 1220 return newSVnv (json_atof (start));
995 1221
996 // everything else fails, convert it to a string 1222 // everything else fails, convert it to a string
997 return newSVpvn (start, dec->cur - start); 1223 return newSVpvn (start, dec->cur - start);
998 } 1224 }
999 1225
1000 // loss of precision here 1226 // loss of precision here
1001 return newSVnv (Atof (start)); 1227 return newSVnv (json_atof (start));
1002 1228
1003fail: 1229fail:
1004 return 0; 1230 return 0;
1005} 1231}
1006 1232
… …
1035 1261
1036 if (*dec->cur != ',') 1262 if (*dec->cur != ',')
1037 ERR (", or ] expected while parsing array"); 1263 ERR (", or ] expected while parsing array");
1038 1264
1039 ++dec->cur; 1265 ++dec->cur;
1266
1267 decode_ws (dec);
1268
1269 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1270 {
1271 ++dec->cur;
1272 break;
1273 }
1040 } 1274 }
1041 1275
1042 DEC_DEC_DEPTH; 1276 DEC_DEC_DEPTH;
1043 return newRV_noinc ((SV *)av); 1277 return newRV_noinc ((SV *)av);
1044 1278
… …
1060 if (*dec->cur == '}') 1294 if (*dec->cur == '}')
1061 ++dec->cur; 1295 ++dec->cur;
1062 else 1296 else
1063 for (;;) 1297 for (;;)
1064 { 1298 {
1065 decode_ws (dec); EXPECT_CH ('"'); 1299 EXPECT_CH ('"');
1066 1300
1067 // heuristic: assume that 1301 // heuristic: assume that
1068 // a) decode_str + hv_store_ent are abysmally slow. 1302 // a) decode_str + hv_store_ent are abysmally slow.
1069 // b) most hash keys are short, simple ascii text. 1303 // b) most hash keys are short, simple ascii text.
1070 // => try to "fast-match" such strings to avoid 1304 // => try to "fast-match" such strings to avoid
… …
1074 char *p = dec->cur; 1308 char *p = dec->cur;
1075 char *e = p + 24; // only try up to 24 bytes 1309 char *e = p + 24; // only try up to 24 bytes
1076 1310
1077 for (;;) 1311 for (;;)
1078 { 1312 {
1079 // the >= 0x80 is true on most architectures 1313 // the >= 0x80 is false on most architectures
1080 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1314 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1081 { 1315 {
1082 // slow path, back up and use decode_str 1316 // slow path, back up and use decode_str
1083 SV *key = decode_str (dec); 1317 SV *key = decode_str (dec);
1084 if (!key) 1318 if (!key)
1085 goto fail; 1319 goto fail;
1086 1320
1087 decode_ws (dec); EXPECT_CH (':'); 1321 decode_ws (dec); EXPECT_CH (':');
1088 1322
1323 decode_ws (dec);
1089 value = decode_sv (dec); 1324 value = decode_sv (dec);
1090 if (!value) 1325 if (!value)
1091 { 1326 {
1092 SvREFCNT_dec (key); 1327 SvREFCNT_dec (key);
1093 goto fail; 1328 goto fail;
… …
1105 int len = p - key; 1340 int len = p - key;
1106 dec->cur = p + 1; 1341 dec->cur = p + 1;
1107 1342
1108 decode_ws (dec); EXPECT_CH (':'); 1343 decode_ws (dec); EXPECT_CH (':');
1109 1344
1345 decode_ws (dec);
1110 value = decode_sv (dec); 1346 value = decode_sv (dec);
1111 if (!value) 1347 if (!value)
1112 goto fail; 1348 goto fail;
1113 1349
1114 hv_store (hv, key, len, value, 0); 1350 hv_store (hv, key, len, value, 0);
… …
1130 1366
1131 if (*dec->cur != ',') 1367 if (*dec->cur != ',')
1132 ERR (", or } expected while parsing object/hash"); 1368 ERR (", or } expected while parsing object/hash");
1133 1369
1134 ++dec->cur; 1370 ++dec->cur;
1371
1372 decode_ws (dec);
1373
1374 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1375 {
1376 ++dec->cur;
1377 break;
1378 }
1135 } 1379 }
1136 1380
1137 DEC_DEC_DEPTH; 1381 DEC_DEC_DEPTH;
1138 sv = newRV_noinc ((SV *)hv); 1382 sv = newRV_noinc ((SV *)hv);
1139 1383
… …
1204} 1448}
1205 1449
1206static SV * 1450static SV *
1207decode_sv (dec_t *dec) 1451decode_sv (dec_t *dec)
1208{ 1452{
1209 decode_ws (dec);
1210
1211 // the beauty of JSON: you need exactly one character lookahead 1453 // the beauty of JSON: you need exactly one character lookahead
1212 // to parse anything. 1454 // to parse everything.
1213 switch (*dec->cur) 1455 switch (*dec->cur)
1214 { 1456 {
1215 case '"': ++dec->cur; return decode_str (dec); 1457 case '"': ++dec->cur; return decode_str (dec);
1216 case '[': ++dec->cur; return decode_av (dec); 1458 case '[': ++dec->cur; return decode_av (dec);
1217 case '{': ++dec->cur; return decode_hv (dec); 1459 case '{': ++dec->cur; return decode_hv (dec);
1218 1460
1219 case '-': 1461 case '-':
1220 case '0': case '1': case '2': case '3': case '4': 1462 case '0': case '1': case '2': case '3': case '4':
1221 case '5': case '6': case '7': case '8': case '9': 1463 case '5': case '6': case '7': case '8': case '9':
1222 return decode_num (dec); 1464 return decode_num (dec);
1223 1465
1224 case 't': 1466 case 't':
1225 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1467 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1226 { 1468 {
1227 dec->cur += 4; 1469 dec->cur += 4;
1470#if JSON_SLOW
1471 json_true = get_bool ("JSON::XS::true");
1472#endif
1228 return SvREFCNT_inc (json_true); 1473 return newSVsv (json_true);
1229 } 1474 }
1230 else 1475 else
1231 ERR ("'true' expected"); 1476 ERR ("'true' expected");
1232 1477
1233 break; 1478 break;
1234 1479
1235 case 'f': 1480 case 'f':
1236 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1481 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1237 { 1482 {
1238 dec->cur += 5; 1483 dec->cur += 5;
1484#if JSON_SLOW
1485 json_false = get_bool ("JSON::XS::false");
1486#endif
1239 return SvREFCNT_inc (json_false); 1487 return newSVsv (json_false);
1240 } 1488 }
1241 else 1489 else
1242 ERR ("'false' expected"); 1490 ERR ("'false' expected");
1243 1491
1244 break; 1492 break;
… …
1262fail: 1510fail:
1263 return 0; 1511 return 0;
1264} 1512}
1265 1513
1266static SV * 1514static SV *
1267decode_json (SV *string, JSON *json, UV *offset_return) 1515decode_json (SV *string, JSON *json, char **offset_return)
1268{ 1516{
1269 dec_t dec; 1517 dec_t dec;
1270 UV offset;
1271 SV *sv; 1518 SV *sv;
1272 1519
1520 /* work around bugs in 5.10 where manipulating magic values
1521 * will perl ignore the magic in subsequent accesses
1522 */
1273 SvGETMAGIC (string); 1523 /*SvGETMAGIC (string);*/
1524 if (SvMAGICAL (string))
1525 string = sv_2mortal (newSVsv (string));
1526
1274 SvUPGRADE (string, SVt_PV); 1527 SvUPGRADE (string, SVt_PV);
1275 1528
1276 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1529 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1530 * assertion with -DDEBUGGING, although SvCUR is documented to
1531 * return the xpv_cur field which certainly exists after upgrading.
1532 * according to nicholas clark, calling SvPOK fixes this.
1533 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1534 * and hope for the best.
1535 * Damnit, SvPV_nolen still trips over yet another assertion. This
1536 * assertion business is seriously broken, try yet another workaround
1537 * for the broken -DDEBUGGING.
1538 */
1539 {
1540#ifdef DEBUGGING
1541 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1542#else
1543 STRLEN offset = SvCUR (string);
1544#endif
1545
1546 if (offset > json->max_size && json->max_size)
1277 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1547 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1278 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1548 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1549 }
1279 1550
1280 if (json->flags & F_UTF8) 1551 if (json->flags & F_UTF8)
1281 sv_utf8_downgrade (string, 0); 1552 sv_utf8_downgrade (string, 0);
1282 else 1553 else
1283 sv_utf8_upgrade (string); 1554 sv_utf8_upgrade (string);
1284 1555
1285 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1556 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1286 1557
1287 dec.json = *json; 1558 dec.json = *json;
1288 dec.cur = SvPVX (string); 1559 dec.cur = SvPVX (string);
1289 dec.end = SvEND (string); 1560 dec.end = SvEND (string);
1290 dec.err = 0; 1561 dec.err = 0;
1291 dec.depth = 0; 1562 dec.depth = 0;
1292 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1293 1563
1294 if (dec.json.cb_object || dec.json.cb_sk_object) 1564 if (dec.json.cb_object || dec.json.cb_sk_object)
1295 dec.json.flags |= F_HOOK; 1565 dec.json.flags |= F_HOOK;
1296 1566
1297 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1567 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1568
1569 decode_ws (&dec);
1298 sv = decode_sv (&dec); 1570 sv = decode_sv (&dec);
1571
1572 if (offset_return)
1573 *offset_return = dec.cur;
1299 1574
1300 if (!(offset_return || !sv)) 1575 if (!(offset_return || !sv))
1301 { 1576 {
1302 // check for trailing garbage 1577 // check for trailing garbage
1303 decode_ws (&dec); 1578 decode_ws (&dec);
… …
1306 { 1581 {
1307 dec.err = "garbage after JSON object"; 1582 dec.err = "garbage after JSON object";
1308 SvREFCNT_dec (sv); 1583 SvREFCNT_dec (sv);
1309 sv = 0; 1584 sv = 0;
1310 } 1585 }
1311 }
1312
1313 if (offset_return || !sv)
1314 {
1315 offset = dec.json.flags & F_UTF8
1316 ? dec.cur - SvPVX (string)
1317 : utf8_distance (dec.cur, SvPVX (string));
1318
1319 if (offset_return)
1320 *offset_return = offset;
1321 } 1586 }
1322 1587
1323 if (!sv) 1588 if (!sv)
1324 { 1589 {
1325 SV *uni = sv_newmortal (); 1590 SV *uni = sv_newmortal ();
… …
1331 SAVEVPTR (PL_curcop); 1596 SAVEVPTR (PL_curcop);
1332 PL_curcop = &cop; 1597 PL_curcop = &cop;
1333 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1598 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1334 LEAVE; 1599 LEAVE;
1335 1600
1336 croak ("%s, at character offset %d [\"%s\"]", 1601 croak ("%s, at character offset %d (before \"%s\")",
1337 dec.err, 1602 dec.err,
1338 (int)offset, 1603 ptr_to_index (string, dec.cur),
1339 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1604 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1340 } 1605 }
1341 1606
1342 sv = sv_2mortal (sv); 1607 sv = sv_2mortal (sv);
1343 1608
1344 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1609 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1345 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1610 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1346 1611
1347 return sv; 1612 return sv;
1613}
1614
1615/////////////////////////////////////////////////////////////////////////////
1616// incremental parser
1617
1618static void
1619incr_parse (JSON *self)
1620{
1621 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1622
1623 // the state machine here is a bit convoluted and could be simplified a lot
1624 // but this would make it slower, so...
1625
1626 for (;;)
1627 {
1628 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1629 switch (self->incr_mode)
1630 {
1631 // only used for initial whitespace skipping
1632 case INCR_M_WS:
1633 for (;;)
1634 {
1635 if (*p > 0x20)
1636 {
1637 if (*p == '#')
1638 {
1639 self->incr_mode = INCR_M_C0;
1640 goto incr_m_c;
1641 }
1642 else
1643 {
1644 self->incr_mode = INCR_M_JSON;
1645 goto incr_m_json;
1646 }
1647 }
1648 else if (!*p)
1649 goto interrupt;
1650
1651 ++p;
1652 }
1653
1654 // skip a single char inside a string (for \\-processing)
1655 case INCR_M_BS:
1656 if (!*p)
1657 goto interrupt;
1658
1659 ++p;
1660 self->incr_mode = INCR_M_STR;
1661 goto incr_m_str;
1662
1663 // inside #-style comments
1664 case INCR_M_C0:
1665 case INCR_M_C1:
1666 incr_m_c:
1667 for (;;)
1668 {
1669 if (*p == '\n')
1670 {
1671 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1672 break;
1673 }
1674 else if (!*p)
1675 goto interrupt;
1676
1677 ++p;
1678 }
1679
1680 break;
1681
1682 // inside a string
1683 case INCR_M_STR:
1684 incr_m_str:
1685 for (;;)
1686 {
1687 if (*p == '"')
1688 {
1689 ++p;
1690 self->incr_mode = INCR_M_JSON;
1691
1692 if (!self->incr_nest)
1693 goto interrupt;
1694
1695 goto incr_m_json;
1696 }
1697 else if (*p == '\\')
1698 {
1699 ++p; // "virtually" consumes character after \
1700
1701 if (!*p) // if at end of string we have to switch modes
1702 {
1703 self->incr_mode = INCR_M_BS;
1704 goto interrupt;
1705 }
1706 }
1707 else if (!*p)
1708 goto interrupt;
1709
1710 ++p;
1711 }
1712
1713 // after initial ws, outside string
1714 case INCR_M_JSON:
1715 incr_m_json:
1716 for (;;)
1717 {
1718 switch (*p++)
1719 {
1720 case 0:
1721 --p;
1722 goto interrupt;
1723
1724 case 0x09:
1725 case 0x0a:
1726 case 0x0d:
1727 case 0x20:
1728 if (!self->incr_nest)
1729 {
1730 --p; // do not eat the whitespace, let the next round do it
1731 goto interrupt;
1732 }
1733 break;
1734
1735 case '"':
1736 self->incr_mode = INCR_M_STR;
1737 goto incr_m_str;
1738
1739 case '[':
1740 case '{':
1741 if (++self->incr_nest > self->max_depth)
1742 croak (ERR_NESTING_EXCEEDED);
1743 break;
1744
1745 case ']':
1746 case '}':
1747 if (--self->incr_nest <= 0)
1748 goto interrupt;
1749 break;
1750
1751 case '#':
1752 self->incr_mode = INCR_M_C1;
1753 goto incr_m_c;
1754 }
1755 }
1756 }
1757
1758 modechange:
1759 ;
1760 }
1761
1762interrupt:
1763 self->incr_pos = p - SvPVX (self->incr_text);
1764 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1765 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1348} 1766}
1349 1767
1350///////////////////////////////////////////////////////////////////////////// 1768/////////////////////////////////////////////////////////////////////////////
1351// XS interface functions 1769// XS interface functions
1352 1770
… …
1364 : -1; 1782 : -1;
1365 1783
1366 json_stash = gv_stashpv ("JSON::XS" , 1); 1784 json_stash = gv_stashpv ("JSON::XS" , 1);
1367 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1785 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1368 1786
1369 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1787 json_true = get_bool ("JSON::XS::true");
1370 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1788 json_false = get_bool ("JSON::XS::false");
1789
1790 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1371} 1791}
1372 1792
1373PROTOTYPES: DISABLE 1793PROTOTYPES: DISABLE
1374 1794
1375void CLONE (...) 1795void CLONE (...)
… …
1378 json_boolean_stash = 0; 1798 json_boolean_stash = 0;
1379 1799
1380void new (char *klass) 1800void new (char *klass)
1381 PPCODE: 1801 PPCODE:
1382{ 1802{
1383 HV *stash = !JSON_SLOW || json_stash
1384 ? json_stash
1385 : gv_stashpv ("JSON::XS", 1);
1386 SV *pv = NEWSV (0, sizeof (JSON)); 1803 SV *pv = NEWSV (0, sizeof (JSON));
1387 SvPOK_only (pv); 1804 SvPOK_only (pv);
1388 Zero (SvPVX (pv), 1, JSON); 1805 json_init ((JSON *)SvPVX (pv));
1389 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1806 XPUSHs (sv_2mortal (sv_bless (
1390 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), stash))); 1807 newRV_noinc (pv),
1808 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1809 )));
1391} 1810}
1392 1811
1393void ascii (JSON *self, int enable = 1) 1812void ascii (JSON *self, int enable = 1)
1394 ALIAS: 1813 ALIAS:
1395 ascii = F_ASCII 1814 ascii = F_ASCII
… …
1402 pretty = F_PRETTY 1821 pretty = F_PRETTY
1403 allow_nonref = F_ALLOW_NONREF 1822 allow_nonref = F_ALLOW_NONREF
1404 shrink = F_SHRINK 1823 shrink = F_SHRINK
1405 allow_blessed = F_ALLOW_BLESSED 1824 allow_blessed = F_ALLOW_BLESSED
1406 convert_blessed = F_CONV_BLESSED 1825 convert_blessed = F_CONV_BLESSED
1826 relaxed = F_RELAXED
1827 allow_unknown = F_ALLOW_UNKNOWN
1407 PPCODE: 1828 PPCODE:
1408{ 1829{
1409 if (enable) 1830 if (enable)
1410 self->flags |= ix; 1831 self->flags |= ix;
1411 else 1832 else
1412 self->flags &= ~ix; 1833 self->flags &= ~ix;
1413 1834
1414 XPUSHs (ST (0)); 1835 XPUSHs (ST (0));
1415} 1836}
1416 1837
1417void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1838void get_ascii (JSON *self)
1839 ALIAS:
1840 get_ascii = F_ASCII
1841 get_latin1 = F_LATIN1
1842 get_utf8 = F_UTF8
1843 get_indent = F_INDENT
1844 get_canonical = F_CANONICAL
1845 get_space_before = F_SPACE_BEFORE
1846 get_space_after = F_SPACE_AFTER
1847 get_allow_nonref = F_ALLOW_NONREF
1848 get_shrink = F_SHRINK
1849 get_allow_blessed = F_ALLOW_BLESSED
1850 get_convert_blessed = F_CONV_BLESSED
1851 get_relaxed = F_RELAXED
1852 get_allow_unknown = F_ALLOW_UNKNOWN
1418 PPCODE: 1853 PPCODE:
1419{ 1854 XPUSHs (boolSV (self->flags & ix));
1420 UV log2 = 0;
1421 1855
1422 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1856void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1423 1857 PPCODE:
1424 while ((1UL << log2) < max_depth) 1858 self->max_depth = max_depth;
1425 ++log2;
1426
1427 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1428
1429 XPUSHs (ST (0)); 1859 XPUSHs (ST (0));
1430}
1431 1860
1861U32 get_max_depth (JSON *self)
1862 CODE:
1863 RETVAL = self->max_depth;
1864 OUTPUT:
1865 RETVAL
1866
1432void max_size (JSON *self, UV max_size = 0) 1867void max_size (JSON *self, U32 max_size = 0)
1433 PPCODE: 1868 PPCODE:
1434{ 1869 self->max_size = max_size;
1435 UV log2 = 0;
1436
1437 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1438 if (max_size == 1) max_size = 2;
1439
1440 while ((1UL << log2) < max_size)
1441 ++log2;
1442
1443 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1444
1445 XPUSHs (ST (0)); 1870 XPUSHs (ST (0));
1446} 1871
1872int get_max_size (JSON *self)
1873 CODE:
1874 RETVAL = self->max_size;
1875 OUTPUT:
1876 RETVAL
1447 1877
1448void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1878void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1449 PPCODE: 1879 PPCODE:
1450{ 1880{
1451 SvREFCNT_dec (self->cb_object); 1881 SvREFCNT_dec (self->cb_object);
… …
1485 XPUSHs (decode_json (jsonstr, self, 0)); 1915 XPUSHs (decode_json (jsonstr, self, 0));
1486 1916
1487void decode_prefix (JSON *self, SV *jsonstr) 1917void decode_prefix (JSON *self, SV *jsonstr)
1488 PPCODE: 1918 PPCODE:
1489{ 1919{
1490 UV offset; 1920 char *offset;
1491 EXTEND (SP, 2); 1921 EXTEND (SP, 2);
1492 PUSHs (decode_json (jsonstr, self, &offset)); 1922 PUSHs (decode_json (jsonstr, self, &offset));
1493 PUSHs (sv_2mortal (newSVuv (offset))); 1923 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1924}
1925
1926void incr_parse (JSON *self, SV *jsonstr = 0)
1927 PPCODE:
1928{
1929 if (!self->incr_text)
1930 self->incr_text = newSVpvn ("", 0);
1931
1932 // append data, if any
1933 if (jsonstr)
1934 {
1935 if (SvUTF8 (jsonstr))
1936 {
1937 if (!SvUTF8 (self->incr_text))
1938 {
1939 /* utf-8-ness differs, need to upgrade */
1940 sv_utf8_upgrade (self->incr_text);
1941
1942 if (self->incr_pos)
1943 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1944 - (U8 *)SvPVX (self->incr_text);
1945 }
1946 }
1947 else if (SvUTF8 (self->incr_text))
1948 sv_utf8_upgrade (jsonstr);
1949
1950 {
1951 STRLEN len;
1952 const char *str = SvPV (jsonstr, len);
1953 STRLEN cur = SvCUR (self->incr_text);
1954
1955 if (SvLEN (self->incr_text) <= cur + len)
1956 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1957
1958 Move (str, SvEND (self->incr_text), len, char);
1959 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1960 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1961 }
1962 }
1963
1964 if (GIMME_V != G_VOID)
1965 do
1966 {
1967 char *offset;
1968
1969 if (!INCR_DONE (self))
1970 {
1971 incr_parse (self);
1972
1973 if (self->incr_pos > self->max_size && self->max_size)
1974 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1975 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1976
1977 if (!INCR_DONE (self))
1978 break;
1979 }
1980
1981 XPUSHs (decode_json (self->incr_text, self, &offset));
1982
1983 self->incr_pos -= offset - SvPVX (self->incr_text);
1984 self->incr_nest = 0;
1985 self->incr_mode = 0;
1986
1987 sv_chop (self->incr_text, offset);
1988 }
1989 while (GIMME_V == G_ARRAY);
1990}
1991
1992SV *incr_text (JSON *self)
1993 ATTRS: lvalue
1994 CODE:
1995{
1996 if (self->incr_pos)
1997 croak ("incr_text can not be called when the incremental parser already started parsing");
1998
1999 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2000}
2001 OUTPUT:
2002 RETVAL
2003
2004void incr_skip (JSON *self)
2005 CODE:
2006{
2007 if (self->incr_pos)
2008 {
2009 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2010 self->incr_pos = 0;
2011 self->incr_nest = 0;
2012 self->incr_mode = 0;
2013 }
2014}
2015
2016void incr_reset (JSON *self)
2017 CODE:
2018{
2019 SvREFCNT_dec (self->incr_text);
2020 self->incr_text = 0;
2021 self->incr_pos = 0;
2022 self->incr_nest = 0;
2023 self->incr_mode = 0;
1494} 2024}
1495 2025
1496void DESTROY (JSON *self) 2026void DESTROY (JSON *self)
1497 CODE: 2027 CODE:
1498 SvREFCNT_dec (self->cb_sk_object); 2028 SvREFCNT_dec (self->cb_sk_object);
1499 SvREFCNT_dec (self->cb_object); 2029 SvREFCNT_dec (self->cb_object);
2030 SvREFCNT_dec (self->incr_text);
1500 2031
1501PROTOTYPES: ENABLE 2032PROTOTYPES: ENABLE
1502 2033
1503void to_json (SV *scalar) 2034void encode_json (SV *scalar)
2035 ALIAS:
2036 to_json_ = 0
2037 encode_json = F_UTF8
1504 PPCODE: 2038 PPCODE:
1505{ 2039{
1506 JSON json = { F_DEFAULT | F_UTF8 }; 2040 JSON json;
2041 json_init (&json);
2042 json.flags |= ix;
1507 XPUSHs (encode_json (scalar, &json)); 2043 XPUSHs (encode_json (scalar, &json));
1508} 2044}
1509 2045
1510void from_json (SV *jsonstr) 2046void decode_json (SV *jsonstr)
2047 ALIAS:
2048 from_json_ = 0
2049 decode_json = F_UTF8
1511 PPCODE: 2050 PPCODE:
1512{ 2051{
1513 JSON json = { F_DEFAULT | F_UTF8 }; 2052 JSON json;
2053 json_init (&json);
2054 json.flags |= ix;
1514 XPUSHs (decode_json (jsonstr, &json, 0)); 2055 XPUSHs (decode_json (jsonstr, &json, 0));
1515} 2056}
1516 2057

Diff Legend

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