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 1.60 by root, Mon Aug 13 16:19:13 2007 UTC vs.
Revision 1.107 by root, Wed Mar 17 01:45:43 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)
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?)"
62 66
63#ifdef USE_ITHREADS 67#ifdef USE_ITHREADS
64# define JSON_SLOW 1 68# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else 70#else
69#endif 73#endif
70 74
71static HV *json_stash, *json_boolean_stash; // JSON::XS:: 75static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false; 76static SV *json_true, *json_false;
73 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
74typedef struct { 89typedef struct {
75 U32 flags; 90 U32 flags;
91 U32 max_depth;
92 STRLEN max_size;
93
76 SV *cb_object; 94 SV *cb_object;
77 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;
78} JSON; 102} JSON;
103
104INLINE void
105json_init (JSON *json)
106{
107 Zero (json, 1, JSON);
108 json->max_depth = 512;
109}
79 110
80///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
81// utility functions 112// utility functions
82 113
83inline 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
84shrink (SV *sv) 126shrink (SV *sv)
85{ 127{
86 sv_utf8_downgrade (sv, 1); 128 sv_utf8_downgrade (sv, 1);
129
87 if (SvLEN (sv) > SvCUR (sv) + 1) 130 if (SvLEN (sv) > SvCUR (sv) + 1)
88 { 131 {
89#ifdef SvPV_shrink_to_cur 132#ifdef SvPV_shrink_to_cur
90 SvPV_shrink_to_cur (sv); 133 SvPV_shrink_to_cur (sv);
91#elif defined (SvPV_renew) 134#elif defined (SvPV_renew)
97// 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
98// case of an error. 141// case of an error.
99// we special-case "safe" characters from U+80 .. U+7FF, 142// we special-case "safe" characters from U+80 .. U+7FF,
100// but use the very good perl function to parse anything else. 143// but use the very good perl function to parse anything else.
101// note that we never call this function for a ascii codepoints 144// note that we never call this function for a ascii codepoints
102inline UV 145INLINE UV
103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 146decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
104{ 147{
105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 148 if (expect_true (len >= 2
106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 149 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 150 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
108 { 151 {
109 *clen = 2; 152 *clen = 2;
110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 153 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
111 } 154 }
112 else 155 else
113 { 156 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
114 *clen = (STRLEN)-1; 157}
115 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 (;;)
116 } 209 {
117} 210 U8 dig = (U8)*s - '0';
118 211
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}
119///////////////////////////////////////////////////////////////////////////// 286/////////////////////////////////////////////////////////////////////////////
120// encoder 287// encoder
121 288
122// structure used for encoding JSON 289// structure used for encoding JSON
123typedef struct 290typedef struct
125 char *cur; // SvPVX (sv) + current output position 292 char *cur; // SvPVX (sv) + current output position
126 char *end; // SvEND (sv) 293 char *end; // SvEND (sv)
127 SV *sv; // result scalar 294 SV *sv; // result scalar
128 JSON json; 295 JSON json;
129 U32 indent; // indentation level 296 U32 indent; // indentation level
130 U32 maxdepth; // max. indentation/recursion level 297 UV limit; // escape character values >= this value when encoding
131} enc_t; 298} enc_t;
132 299
133inline void 300INLINE void
134need (enc_t *enc, STRLEN len) 301need (enc_t *enc, STRLEN len)
135{ 302{
136 if (expect_false (enc->cur + len >= enc->end)) 303 if (expect_false (enc->cur + len >= enc->end))
137 { 304 {
138 STRLEN cur = enc->cur - SvPVX (enc->sv); 305 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
139 SvGROW (enc->sv, cur + len + 1); 306 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
140 enc->cur = SvPVX (enc->sv) + cur; 307 enc->cur = SvPVX (enc->sv) + cur;
141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 308 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
142 } 309 }
143} 310}
144 311
145inline void 312INLINE void
146encode_ch (enc_t *enc, char ch) 313encode_ch (enc_t *enc, char ch)
147{ 314{
148 need (enc, 1); 315 need (enc, 1);
149 *enc->cur++ = ch; 316 *enc->cur++ = ch;
150} 317}
204 { 371 {
205 uch = ch; 372 uch = ch;
206 clen = 1; 373 clen = 1;
207 } 374 }
208 375
209 if (uch > 0x10FFFFUL) 376 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
211
212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
213 { 377 {
214 if (uch > 0xFFFFUL) 378 if (uch >= 0x10000UL)
215 { 379 {
380 if (uch >= 0x110000UL)
381 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
382
216 need (enc, len += 11); 383 need (enc, len += 11);
217 sprintf (enc->cur, "\\u%04x\\u%04x", 384 sprintf (enc->cur, "\\u%04x\\u%04x",
218 (int)((uch - 0x10000) / 0x400 + 0xD800), 385 (int)((uch - 0x10000) / 0x400 + 0xD800),
219 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 386 (int)((uch - 0x10000) % 0x400 + 0xDC00));
220 enc->cur += 12; 387 enc->cur += 12;
221 } 388 }
222 else 389 else
223 { 390 {
224 static char hexdigit [16] = "0123456789abcdef";
225 need (enc, len += 5); 391 need (enc, len += 5);
226 *enc->cur++ = '\\'; 392 *enc->cur++ = '\\';
227 *enc->cur++ = 'u'; 393 *enc->cur++ = 'u';
228 *enc->cur++ = hexdigit [ uch >> 12 ]; 394 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
229 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 395 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
230 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 396 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
231 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 397 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
232 } 398 }
233 399
234 str += clen; 400 str += clen;
235 } 401 }
236 else if (enc->json.flags & F_LATIN1) 402 else if (enc->json.flags & F_LATIN1)
248 while (--clen); 414 while (--clen);
249 } 415 }
250 else 416 else
251 { 417 {
252 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
253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 419 enc->cur = encode_utf8 (enc->cur, uch);
254 ++str; 420 ++str;
255 } 421 }
256 } 422 }
257 } 423 }
258 } 424 }
259 425
260 --len; 426 --len;
261 } 427 }
262} 428}
263 429
264inline void 430INLINE void
265encode_indent (enc_t *enc) 431encode_indent (enc_t *enc)
266{ 432{
267 if (enc->json.flags & F_INDENT) 433 if (enc->json.flags & F_INDENT)
268 { 434 {
269 int spaces = enc->indent * INDENT_STEP; 435 int spaces = enc->indent * INDENT_STEP;
272 memset (enc->cur, ' ', spaces); 438 memset (enc->cur, ' ', spaces);
273 enc->cur += spaces; 439 enc->cur += spaces;
274 } 440 }
275} 441}
276 442
277inline void 443INLINE void
278encode_space (enc_t *enc) 444encode_space (enc_t *enc)
279{ 445{
280 need (enc, 1); 446 need (enc, 1);
281 encode_ch (enc, ' '); 447 encode_ch (enc, ' ');
282} 448}
283 449
284inline void 450INLINE void
285encode_nl (enc_t *enc) 451encode_nl (enc_t *enc)
286{ 452{
287 if (enc->json.flags & F_INDENT) 453 if (enc->json.flags & F_INDENT)
288 { 454 {
289 need (enc, 1); 455 need (enc, 1);
290 encode_ch (enc, '\n'); 456 encode_ch (enc, '\n');
291 } 457 }
292} 458}
293 459
294inline void 460INLINE void
295encode_comma (enc_t *enc) 461encode_comma (enc_t *enc)
296{ 462{
297 encode_ch (enc, ','); 463 encode_ch (enc, ',');
298 464
299 if (enc->json.flags & F_INDENT) 465 if (enc->json.flags & F_INDENT)
307static void 473static void
308encode_av (enc_t *enc, AV *av) 474encode_av (enc_t *enc, AV *av)
309{ 475{
310 int i, len = av_len (av); 476 int i, len = av_len (av);
311 477
312 if (enc->indent >= enc->maxdepth) 478 if (enc->indent >= enc->json.max_depth)
313 croak ("data structure too deep (hit recursion limit)"); 479 croak (ERR_NESTING_EXCEEDED);
314 480
315 encode_ch (enc, '['); encode_nl (enc); 481 encode_ch (enc, '[');
316 ++enc->indent; 482
483 if (len >= 0)
484 {
485 encode_nl (enc); ++enc->indent;
317 486
318 for (i = 0; i <= len; ++i) 487 for (i = 0; i <= len; ++i)
319 { 488 {
320 SV **svp = av_fetch (av, i, 0); 489 SV **svp = av_fetch (av, i, 0);
321 490
322 encode_indent (enc); 491 encode_indent (enc);
323 492
324 if (svp) 493 if (svp)
325 encode_sv (enc, *svp); 494 encode_sv (enc, *svp);
326 else 495 else
327 encode_str (enc, "null", 4, 0); 496 encode_str (enc, "null", 4, 0);
328 497
329 if (i < len) 498 if (i < len)
330 encode_comma (enc); 499 encode_comma (enc);
331 } 500 }
332 501
502 encode_nl (enc); --enc->indent; encode_indent (enc);
503 }
504
333 encode_nl (enc); 505 encode_ch (enc, ']');
334
335 --enc->indent;
336 encode_indent (enc); encode_ch (enc, ']');
337} 506}
338 507
339static void 508static void
340encode_he (enc_t *enc, HE *he) 509encode_hk (enc_t *enc, HE *he)
341{ 510{
342 encode_ch (enc, '"'); 511 encode_ch (enc, '"');
343 512
344 if (HeKLEN (he) == HEf_SVKEY) 513 if (HeKLEN (he) == HEf_SVKEY)
345 { 514 {
358 encode_ch (enc, '"'); 527 encode_ch (enc, '"');
359 528
360 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 529 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
361 encode_ch (enc, ':'); 530 encode_ch (enc, ':');
362 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 531 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
363 encode_sv (enc, HeVAL (he));
364} 532}
365 533
366// compare hash entries, used when all keys are bytestrings 534// compare hash entries, used when all keys are bytestrings
367static int 535static int
368he_cmp_fast (const void *a_, const void *b_) 536he_cmp_fast (const void *a_, const void *b_)
373 HE *b = *(HE **)b_; 541 HE *b = *(HE **)b_;
374 542
375 STRLEN la = HeKLEN (a); 543 STRLEN la = HeKLEN (a);
376 STRLEN lb = HeKLEN (b); 544 STRLEN lb = HeKLEN (b);
377 545
378 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 546 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
379 cmp = la - lb; 547 cmp = lb - la;
380 548
381 return cmp; 549 return cmp;
382} 550}
383 551
384// 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
385static int 553static int
386he_cmp_slow (const void *a, const void *b) 554he_cmp_slow (const void *a, const void *b)
387{ 555{
388 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 556 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
389} 557}
390 558
391static void 559static void
392encode_hv (enc_t *enc, HV *hv) 560encode_hv (enc_t *enc, HV *hv)
393{ 561{
394 int count, i; 562 HE *he;
395 563
396 if (enc->indent >= enc->maxdepth) 564 if (enc->indent >= enc->json.max_depth)
397 croak ("data structure too deep (hit recursion limit)"); 565 croak (ERR_NESTING_EXCEEDED);
398 566
399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 567 encode_ch (enc, '{');
400 568
401 if ((count = hv_iterinit (hv)))
402 {
403 // for canonical output we have to sort by keys first 569 // for canonical output we have to sort by keys first
404 // actually, this is mostly due to the stupid so-called 570 // actually, this is mostly due to the stupid so-called
405 // security workaround added somewhere in 5.8.x. 571 // security workaround added somewhere in 5.8.x
406 // that randomises hash orderings 572 // that randomises hash orderings
407 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))
408 { 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 {
409 int fast = 1; 593 int i, fast = 1;
410 HE *he;
411#if defined(__BORLANDC__) || defined(_MSC_VER) 594#if defined(__BORLANDC__) || defined(_MSC_VER)
412 HE **hes = _alloca (count * sizeof (HE)); 595 HE **hes = _alloca (count * sizeof (HE));
413#else 596#else
414 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
415#endif 598#endif
442 625
443 FREETMPS; 626 FREETMPS;
444 LEAVE; 627 LEAVE;
445 } 628 }
446 629
447 for (i = 0; i < count; ++i) 630 encode_nl (enc); ++enc->indent;
631
632 while (count--)
448 { 633 {
449 encode_indent (enc); 634 encode_indent (enc);
635 he = hes [count];
450 encode_he (enc, hes [i]); 636 encode_hk (enc, he);
637 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
451 638
452 if (i < count - 1) 639 if (count)
453 encode_comma (enc); 640 encode_comma (enc);
454 } 641 }
455 642
456 encode_nl (enc); 643 encode_nl (enc); --enc->indent; encode_indent (enc);
457 } 644 }
645 }
458 else 646 else
647 {
648 if (hv_iterinit (hv) || SvMAGICAL (hv))
649 if ((he = hv_iternext (hv)))
459 { 650 {
460 HE *he = hv_iternext (hv); 651 encode_nl (enc); ++enc->indent;
461 652
462 for (;;) 653 for (;;)
463 { 654 {
464 encode_indent (enc); 655 encode_indent (enc);
465 encode_he (enc, he); 656 encode_hk (enc, he);
657 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
466 658
467 if (!(he = hv_iternext (hv))) 659 if (!(he = hv_iternext (hv)))
468 break; 660 break;
469 661
470 encode_comma (enc); 662 encode_comma (enc);
471 } 663 }
472 664
473 encode_nl (enc); 665 encode_nl (enc); --enc->indent; encode_indent (enc);
474 } 666 }
475 } 667 }
476 668
477 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 669 encode_ch (enc, '}');
478} 670}
479 671
480// encode objects, arrays and special \0=false and \1=true values. 672// encode objects, arrays and special \0=false and \1=true values.
481static void 673static void
482encode_rv (enc_t *enc, SV *sv) 674encode_rv (enc_t *enc, SV *sv)
559 751
560 if (len == 1 && *pv == '1') 752 if (len == 1 && *pv == '1')
561 encode_str (enc, "true", 4, 0); 753 encode_str (enc, "true", 4, 0);
562 else if (len == 1 && *pv == '0') 754 else if (len == 1 && *pv == '0')
563 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);
564 else 758 else
565 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",
566 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 760 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
567 } 761 }
762 else if (enc->json.flags & F_ALLOW_UNKNOWN)
763 encode_str (enc, "null", 4, 0);
568 else 764 else
569 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",
570 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 766 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
571} 767}
572 768
590 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 786 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
591 enc->cur += strlen (enc->cur); 787 enc->cur += strlen (enc->cur);
592 } 788 }
593 else if (SvIOKp (sv)) 789 else if (SvIOKp (sv))
594 { 790 {
595 // 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
596 if (SvUV (sv) & ~(UV)0x7fff) 792 // we assume two's complement
597 { 793 // we assume no aliasing issues in the union
598 // large integer, use the (rather slow) snprintf way. 794 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
599 need (enc, sizeof (UV) * 3); 795 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
600 enc->cur +=
601 SvIsUV(sv)
602 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
603 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
604 }
605 else
606 { 796 {
607 // optimise the "small number case" 797 // optimise the "small number case"
608 // 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
609 I32 i = SvIV (sv); 800 I32 i = SvIVX (sv);
610 U32 u; 801 U32 u;
611 char digit, nz = 0; 802 char digit, nz = 0;
612 803
613 need (enc, 6); 804 need (enc, 6);
614 805
620 811
621 // 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
622 // 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,
623 // resulting in a net multiplication by 10. 814 // resulting in a net multiplication by 10.
624 // we always write the digit to memory but conditionally increment 815 // we always write the digit to memory but conditionally increment
625 // the pointer, to ease the usage of conditional move instructions. 816 // the pointer, to enable the use of conditional move instructions.
626 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;
627 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;
628 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;
629 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;
630 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'
631 } 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 }
632 } 832 }
633 else if (SvROK (sv)) 833 else if (SvROK (sv))
634 encode_rv (enc, SvRV (sv)); 834 encode_rv (enc, SvRV (sv));
635 else if (!SvOK (sv)) 835 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
636 encode_str (enc, "null", 4, 0); 836 encode_str (enc, "null", 4, 0);
637 else 837 else
638 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",
639 SvPV_nolen (sv), SvFLAGS (sv)); 839 SvPV_nolen (sv), SvFLAGS (sv));
640} 840}
650 enc.json = *json; 850 enc.json = *json;
651 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 851 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
652 enc.cur = SvPVX (enc.sv); 852 enc.cur = SvPVX (enc.sv);
653 enc.end = SvEND (enc.sv); 853 enc.end = SvEND (enc.sv);
654 enc.indent = 0; 854 enc.indent = 0;
655 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;
656 858
657 SvPOK_only (enc.sv); 859 SvPOK_only (enc.sv);
658 encode_sv (&enc, scalar); 860 encode_sv (&enc, scalar);
861 encode_nl (&enc);
659 862
660 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 863 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
661 *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
662 865
663 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 866 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
681 JSON json; 884 JSON json;
682 U32 depth; // recursion depth 885 U32 depth; // recursion depth
683 U32 maxdepth; // recursion depth limit 886 U32 maxdepth; // recursion depth limit
684} dec_t; 887} dec_t;
685 888
686inline 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
687decode_ws (dec_t *dec) 899decode_ws (dec_t *dec)
688{ 900{
689 for (;;) 901 for (;;)
690 { 902 {
691 char ch = *dec->cur; 903 char ch = *dec->cur;
692 904
693 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 }
694 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 917 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
695 break; 918 break; // parse error, but let higher level handle it, gives better error messages
696 919
697 ++dec->cur; 920 ++dec->cur;
698 } 921 }
699} 922}
700 923
704 if (*dec->cur != ch) \ 927 if (*dec->cur != ch) \
705 ERR (# ch " expected"); \ 928 ERR (# ch " expected"); \
706 ++dec->cur; \ 929 ++dec->cur; \
707 SE 930 SE
708 931
709#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)
710#define DEC_DEC_DEPTH --dec->depth 933#define DEC_DEC_DEPTH --dec->depth
711 934
712static SV *decode_sv (dec_t *dec); 935static SV *decode_sv (dec_t *dec);
713 936
714static signed char decode_hexdigit[256]; 937static signed char decode_hexdigit[256];
806 1029
807 if (hi >= 0x80) 1030 if (hi >= 0x80)
808 { 1031 {
809 utf8 = 1; 1032 utf8 = 1;
810 1033
811 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1034 cur = encode_utf8 (cur, hi);
812 } 1035 }
813 else 1036 else
814 *cur++ = hi; 1037 *cur++ = hi;
815 } 1038 }
816 break; 1039 break;
818 default: 1041 default:
819 --dec_cur; 1042 --dec_cur;
820 ERR ("illegal backslash escape sequence in string"); 1043 ERR ("illegal backslash escape sequence in string");
821 } 1044 }
822 } 1045 }
823 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1046 else if (expect_true (ch >= 0x20 && ch < 0x80))
824 *cur++ = ch; 1047 *cur++ = ch;
825 else if (ch >= 0x80) 1048 else if (ch >= 0x80)
826 { 1049 {
827 STRLEN clen; 1050 STRLEN clen;
828 UV uch;
829 1051
830 --dec_cur; 1052 --dec_cur;
831 1053
832 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1054 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
833 if (clen == (STRLEN)-1) 1055 if (clen == (STRLEN)-1)
834 ERR ("malformed UTF-8 character in JSON string"); 1056 ERR ("malformed UTF-8 character in JSON string");
835 1057
836 do 1058 do
837 *cur++ = *dec_cur++; 1059 *cur++ = *dec_cur++;
854 { 1076 {
855 STRLEN len = cur - buf; 1077 STRLEN len = cur - buf;
856 1078
857 if (sv) 1079 if (sv)
858 { 1080 {
859 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
860 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1086 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
861 SvCUR_set (sv, SvCUR (sv) + len); 1087 SvCUR_set (sv, SvCUR (sv) + len);
862 } 1088 }
863 else 1089 else
864 sv = newSVpvn (buf, len); 1090 sv = newSVpvn (buf, len);
951 1177
952 if (!is_nv) 1178 if (!is_nv)
953 { 1179 {
954 int len = dec->cur - start; 1180 int len = dec->cur - start;
955 1181
956 // 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
957 if (*start == '-') 1183 if (*start == '-')
958 switch (len) 1184 switch (len)
959 { 1185 {
960 case 2: return newSViv (-( start [1] - '0' * 1)); 1186 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
961 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1187 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
962 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));
963 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));
964 } 1191 }
965 else 1192 else
966 switch (len) 1193 switch (len)
967 { 1194 {
968 case 1: return newSViv ( start [0] - '0' * 1); 1195 case 1: return newSViv ( start [0] - '0' * 1);
969 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1196 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
970 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);
971 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);
972 } 1200 }
973 1201
974 { 1202 {
975 UV uv; 1203 UV uv;
976 int numtype = grok_number (start, len, &uv); 1204 int numtype = grok_number (start, len, &uv);
985 } 1213 }
986 1214
987 len -= *start == '-' ? 1 : 0; 1215 len -= *start == '-' ? 1 : 0;
988 1216
989 // does not fit into IV or UV, try NV 1217 // does not fit into IV or UV, try NV
990 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1218 if (len <= NV_DIG)
991 #if defined (LDBL_DIG)
992 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
993 #endif
994 )
995 // fits into NV without loss of precision 1219 // fits into NV without loss of precision
996 return newSVnv (Atof (start)); 1220 return newSVnv (json_atof (start));
997 1221
998 // everything else fails, convert it to a string 1222 // everything else fails, convert it to a string
999 return newSVpvn (start, dec->cur - start); 1223 return newSVpvn (start, dec->cur - start);
1000 } 1224 }
1001 1225
1002 // loss of precision here 1226 // loss of precision here
1003 return newSVnv (Atof (start)); 1227 return newSVnv (json_atof (start));
1004 1228
1005fail: 1229fail:
1006 return 0; 1230 return 0;
1007} 1231}
1008 1232
1037 1261
1038 if (*dec->cur != ',') 1262 if (*dec->cur != ',')
1039 ERR (", or ] expected while parsing array"); 1263 ERR (", or ] expected while parsing array");
1040 1264
1041 ++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 }
1042 } 1274 }
1043 1275
1044 DEC_DEC_DEPTH; 1276 DEC_DEC_DEPTH;
1045 return newRV_noinc ((SV *)av); 1277 return newRV_noinc ((SV *)av);
1046 1278
1062 if (*dec->cur == '}') 1294 if (*dec->cur == '}')
1063 ++dec->cur; 1295 ++dec->cur;
1064 else 1296 else
1065 for (;;) 1297 for (;;)
1066 { 1298 {
1067 decode_ws (dec); EXPECT_CH ('"'); 1299 EXPECT_CH ('"');
1068 1300
1069 // heuristic: assume that 1301 // heuristic: assume that
1070 // a) decode_str + hv_store_ent are abysmally slow. 1302 // a) decode_str + hv_store_ent are abysmally slow.
1071 // b) most hash keys are short, simple ascii text. 1303 // b) most hash keys are short, simple ascii text.
1072 // => try to "fast-match" such strings to avoid 1304 // => try to "fast-match" such strings to avoid
1076 char *p = dec->cur; 1308 char *p = dec->cur;
1077 char *e = p + 24; // only try up to 24 bytes 1309 char *e = p + 24; // only try up to 24 bytes
1078 1310
1079 for (;;) 1311 for (;;)
1080 { 1312 {
1081 // the >= 0x80 is true on most architectures 1313 // the >= 0x80 is false on most architectures
1082 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1314 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1083 { 1315 {
1084 // slow path, back up and use decode_str 1316 // slow path, back up and use decode_str
1085 SV *key = decode_str (dec); 1317 SV *key = decode_str (dec);
1086 if (!key) 1318 if (!key)
1087 goto fail; 1319 goto fail;
1088 1320
1089 decode_ws (dec); EXPECT_CH (':'); 1321 decode_ws (dec); EXPECT_CH (':');
1090 1322
1323 decode_ws (dec);
1091 value = decode_sv (dec); 1324 value = decode_sv (dec);
1092 if (!value) 1325 if (!value)
1093 { 1326 {
1094 SvREFCNT_dec (key); 1327 SvREFCNT_dec (key);
1095 goto fail; 1328 goto fail;
1107 int len = p - key; 1340 int len = p - key;
1108 dec->cur = p + 1; 1341 dec->cur = p + 1;
1109 1342
1110 decode_ws (dec); EXPECT_CH (':'); 1343 decode_ws (dec); EXPECT_CH (':');
1111 1344
1345 decode_ws (dec);
1112 value = decode_sv (dec); 1346 value = decode_sv (dec);
1113 if (!value) 1347 if (!value)
1114 goto fail; 1348 goto fail;
1115 1349
1116 hv_store (hv, key, len, value, 0); 1350 hv_store (hv, key, len, value, 0);
1132 1366
1133 if (*dec->cur != ',') 1367 if (*dec->cur != ',')
1134 ERR (", or } expected while parsing object/hash"); 1368 ERR (", or } expected while parsing object/hash");
1135 1369
1136 ++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 }
1137 } 1379 }
1138 1380
1139 DEC_DEC_DEPTH; 1381 DEC_DEC_DEPTH;
1140 sv = newRV_noinc ((SV *)hv); 1382 sv = newRV_noinc ((SV *)hv);
1141 1383
1159 dSP; 1401 dSP;
1160 int count; 1402 int count;
1161 1403
1162 ENTER; SAVETMPS; PUSHMARK (SP); 1404 ENTER; SAVETMPS; PUSHMARK (SP);
1163 XPUSHs (HeVAL (he)); 1405 XPUSHs (HeVAL (he));
1406 sv_2mortal (sv);
1164 1407
1165 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1408 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1166 1409
1167 if (count == 1) 1410 if (count == 1)
1168 { 1411 {
1169 sv = newSVsv (POPs); 1412 sv = newSVsv (POPs);
1170 FREETMPS; LEAVE; 1413 FREETMPS; LEAVE;
1171 return sv; 1414 return sv;
1172 } 1415 }
1173 1416
1417 SvREFCNT_inc (sv);
1174 FREETMPS; LEAVE; 1418 FREETMPS; LEAVE;
1175 } 1419 }
1176 } 1420 }
1177 1421
1178 if (dec->json.cb_object) 1422 if (dec->json.cb_object)
1206} 1450}
1207 1451
1208static SV * 1452static SV *
1209decode_sv (dec_t *dec) 1453decode_sv (dec_t *dec)
1210{ 1454{
1211 decode_ws (dec);
1212
1213 // the beauty of JSON: you need exactly one character lookahead 1455 // the beauty of JSON: you need exactly one character lookahead
1214 // to parse anything. 1456 // to parse everything.
1215 switch (*dec->cur) 1457 switch (*dec->cur)
1216 { 1458 {
1217 case '"': ++dec->cur; return decode_str (dec); 1459 case '"': ++dec->cur; return decode_str (dec);
1218 case '[': ++dec->cur; return decode_av (dec); 1460 case '[': ++dec->cur; return decode_av (dec);
1219 case '{': ++dec->cur; return decode_hv (dec); 1461 case '{': ++dec->cur; return decode_hv (dec);
1220 1462
1221 case '-': 1463 case '-':
1222 case '0': case '1': case '2': case '3': case '4': 1464 case '0': case '1': case '2': case '3': case '4':
1223 case '5': case '6': case '7': case '8': case '9': 1465 case '5': case '6': case '7': case '8': case '9':
1224 return decode_num (dec); 1466 return decode_num (dec);
1226 case 't': 1468 case 't':
1227 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1469 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1228 { 1470 {
1229 dec->cur += 4; 1471 dec->cur += 4;
1230#if JSON_SLOW 1472#if JSON_SLOW
1231 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1473 json_true = get_bool ("JSON::XS::true");
1232#endif 1474#endif
1233 return SvREFCNT_inc (json_true); 1475 return newSVsv (json_true);
1234 } 1476 }
1235 else 1477 else
1236 ERR ("'true' expected"); 1478 ERR ("'true' expected");
1237 1479
1238 break; 1480 break;
1240 case 'f': 1482 case 'f':
1241 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1483 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1242 { 1484 {
1243 dec->cur += 5; 1485 dec->cur += 5;
1244#if JSON_SLOW 1486#if JSON_SLOW
1245 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1487 json_false = get_bool ("JSON::XS::false");
1246#endif 1488#endif
1247 return SvREFCNT_inc (json_false); 1489 return newSVsv (json_false);
1248 } 1490 }
1249 else 1491 else
1250 ERR ("'false' expected"); 1492 ERR ("'false' expected");
1251 1493
1252 break; 1494 break;
1270fail: 1512fail:
1271 return 0; 1513 return 0;
1272} 1514}
1273 1515
1274static SV * 1516static SV *
1275decode_json (SV *string, JSON *json, UV *offset_return) 1517decode_json (SV *string, JSON *json, char **offset_return)
1276{ 1518{
1277 dec_t dec; 1519 dec_t dec;
1278 UV offset;
1279 SV *sv; 1520 SV *sv;
1280 1521
1522 /* work around bugs in 5.10 where manipulating magic values
1523 * will perl ignore the magic in subsequent accesses
1524 */
1281 SvGETMAGIC (string); 1525 /*SvGETMAGIC (string);*/
1526 if (SvMAGICAL (string))
1527 string = sv_2mortal (newSVsv (string));
1528
1282 SvUPGRADE (string, SVt_PV); 1529 SvUPGRADE (string, SVt_PV);
1283 1530
1284 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1531 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1532 * assertion with -DDEBUGGING, although SvCUR is documented to
1533 * return the xpv_cur field which certainly exists after upgrading.
1534 * according to nicholas clark, calling SvPOK fixes this.
1535 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1536 * and hope for the best.
1537 * Damnit, SvPV_nolen still trips over yet another assertion. This
1538 * assertion business is seriously broken, try yet another workaround
1539 * for the broken -DDEBUGGING.
1540 */
1541 {
1542#ifdef DEBUGGING
1543 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1544#else
1545 STRLEN offset = SvCUR (string);
1546#endif
1547
1548 if (offset > json->max_size && json->max_size)
1285 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1549 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1286 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1550 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1551 }
1287 1552
1288 if (json->flags & F_UTF8) 1553 if (json->flags & F_UTF8)
1289 sv_utf8_downgrade (string, 0); 1554 sv_utf8_downgrade (string, 0);
1290 else 1555 else
1291 sv_utf8_upgrade (string); 1556 sv_utf8_upgrade (string);
1292 1557
1293 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1558 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1294 1559
1295 dec.json = *json; 1560 dec.json = *json;
1296 dec.cur = SvPVX (string); 1561 dec.cur = SvPVX (string);
1297 dec.end = SvEND (string); 1562 dec.end = SvEND (string);
1298 dec.err = 0; 1563 dec.err = 0;
1299 dec.depth = 0; 1564 dec.depth = 0;
1300 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1301 1565
1302 if (dec.json.cb_object || dec.json.cb_sk_object) 1566 if (dec.json.cb_object || dec.json.cb_sk_object)
1303 dec.json.flags |= F_HOOK; 1567 dec.json.flags |= F_HOOK;
1304 1568
1305 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1569 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1570
1571 decode_ws (&dec);
1306 sv = decode_sv (&dec); 1572 sv = decode_sv (&dec);
1573
1574 if (offset_return)
1575 *offset_return = dec.cur;
1307 1576
1308 if (!(offset_return || !sv)) 1577 if (!(offset_return || !sv))
1309 { 1578 {
1310 // check for trailing garbage 1579 // check for trailing garbage
1311 decode_ws (&dec); 1580 decode_ws (&dec);
1314 { 1583 {
1315 dec.err = "garbage after JSON object"; 1584 dec.err = "garbage after JSON object";
1316 SvREFCNT_dec (sv); 1585 SvREFCNT_dec (sv);
1317 sv = 0; 1586 sv = 0;
1318 } 1587 }
1319 }
1320
1321 if (offset_return || !sv)
1322 {
1323 offset = dec.json.flags & F_UTF8
1324 ? dec.cur - SvPVX (string)
1325 : utf8_distance (dec.cur, SvPVX (string));
1326
1327 if (offset_return)
1328 *offset_return = offset;
1329 } 1588 }
1330 1589
1331 if (!sv) 1590 if (!sv)
1332 { 1591 {
1333 SV *uni = sv_newmortal (); 1592 SV *uni = sv_newmortal ();
1339 SAVEVPTR (PL_curcop); 1598 SAVEVPTR (PL_curcop);
1340 PL_curcop = &cop; 1599 PL_curcop = &cop;
1341 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1600 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1342 LEAVE; 1601 LEAVE;
1343 1602
1344 croak ("%s, at character offset %d [\"%s\"]", 1603 croak ("%s, at character offset %d (before \"%s\")",
1345 dec.err, 1604 dec.err,
1346 (int)offset, 1605 ptr_to_index (string, dec.cur),
1347 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1606 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1348 } 1607 }
1349 1608
1350 sv = sv_2mortal (sv); 1609 sv = sv_2mortal (sv);
1351 1610
1352 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1611 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1353 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1612 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1354 1613
1355 return sv; 1614 return sv;
1615}
1616
1617/////////////////////////////////////////////////////////////////////////////
1618// incremental parser
1619
1620static void
1621incr_parse (JSON *self)
1622{
1623 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1624
1625 // the state machine here is a bit convoluted and could be simplified a lot
1626 // but this would make it slower, so...
1627
1628 for (;;)
1629 {
1630 //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
1631 switch (self->incr_mode)
1632 {
1633 // only used for initial whitespace skipping
1634 case INCR_M_WS:
1635 for (;;)
1636 {
1637 if (*p > 0x20)
1638 {
1639 if (*p == '#')
1640 {
1641 self->incr_mode = INCR_M_C0;
1642 goto incr_m_c;
1643 }
1644 else
1645 {
1646 self->incr_mode = INCR_M_JSON;
1647 goto incr_m_json;
1648 }
1649 }
1650 else if (!*p)
1651 goto interrupt;
1652
1653 ++p;
1654 }
1655
1656 // skip a single char inside a string (for \\-processing)
1657 case INCR_M_BS:
1658 if (!*p)
1659 goto interrupt;
1660
1661 ++p;
1662 self->incr_mode = INCR_M_STR;
1663 goto incr_m_str;
1664
1665 // inside #-style comments
1666 case INCR_M_C0:
1667 case INCR_M_C1:
1668 incr_m_c:
1669 for (;;)
1670 {
1671 if (*p == '\n')
1672 {
1673 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1674 break;
1675 }
1676 else if (!*p)
1677 goto interrupt;
1678
1679 ++p;
1680 }
1681
1682 break;
1683
1684 // inside a string
1685 case INCR_M_STR:
1686 incr_m_str:
1687 for (;;)
1688 {
1689 if (*p == '"')
1690 {
1691 ++p;
1692 self->incr_mode = INCR_M_JSON;
1693
1694 if (!self->incr_nest)
1695 goto interrupt;
1696
1697 goto incr_m_json;
1698 }
1699 else if (*p == '\\')
1700 {
1701 ++p; // "virtually" consumes character after \
1702
1703 if (!*p) // if at end of string we have to switch modes
1704 {
1705 self->incr_mode = INCR_M_BS;
1706 goto interrupt;
1707 }
1708 }
1709 else if (!*p)
1710 goto interrupt;
1711
1712 ++p;
1713 }
1714
1715 // after initial ws, outside string
1716 case INCR_M_JSON:
1717 incr_m_json:
1718 for (;;)
1719 {
1720 switch (*p++)
1721 {
1722 case 0:
1723 --p;
1724 goto interrupt;
1725
1726 case 0x09:
1727 case 0x0a:
1728 case 0x0d:
1729 case 0x20:
1730 if (!self->incr_nest)
1731 {
1732 --p; // do not eat the whitespace, let the next round do it
1733 goto interrupt;
1734 }
1735 break;
1736
1737 case '"':
1738 self->incr_mode = INCR_M_STR;
1739 goto incr_m_str;
1740
1741 case '[':
1742 case '{':
1743 if (++self->incr_nest > self->max_depth)
1744 croak (ERR_NESTING_EXCEEDED);
1745 break;
1746
1747 case ']':
1748 case '}':
1749 if (--self->incr_nest <= 0)
1750 goto interrupt;
1751 break;
1752
1753 case '#':
1754 self->incr_mode = INCR_M_C1;
1755 goto incr_m_c;
1756 }
1757 }
1758 }
1759
1760 modechange:
1761 ;
1762 }
1763
1764interrupt:
1765 self->incr_pos = p - SvPVX (self->incr_text);
1766 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1767 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1356} 1768}
1357 1769
1358///////////////////////////////////////////////////////////////////////////// 1770/////////////////////////////////////////////////////////////////////////////
1359// XS interface functions 1771// XS interface functions
1360 1772
1372 : -1; 1784 : -1;
1373 1785
1374 json_stash = gv_stashpv ("JSON::XS" , 1); 1786 json_stash = gv_stashpv ("JSON::XS" , 1);
1375 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1787 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1376 1788
1377 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1789 json_true = get_bool ("JSON::XS::true");
1378 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1790 json_false = get_bool ("JSON::XS::false");
1791
1792 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1379} 1793}
1380 1794
1381PROTOTYPES: DISABLE 1795PROTOTYPES: DISABLE
1382 1796
1383void CLONE (...) 1797void CLONE (...)
1388void new (char *klass) 1802void new (char *klass)
1389 PPCODE: 1803 PPCODE:
1390{ 1804{
1391 SV *pv = NEWSV (0, sizeof (JSON)); 1805 SV *pv = NEWSV (0, sizeof (JSON));
1392 SvPOK_only (pv); 1806 SvPOK_only (pv);
1393 Zero (SvPVX (pv), 1, JSON); 1807 json_init ((JSON *)SvPVX (pv));
1394 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1808 XPUSHs (sv_2mortal (sv_bless (
1395 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); 1809 newRV_noinc (pv),
1810 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1811 )));
1396} 1812}
1397 1813
1398void ascii (JSON *self, int enable = 1) 1814void ascii (JSON *self, int enable = 1)
1399 ALIAS: 1815 ALIAS:
1400 ascii = F_ASCII 1816 ascii = F_ASCII
1407 pretty = F_PRETTY 1823 pretty = F_PRETTY
1408 allow_nonref = F_ALLOW_NONREF 1824 allow_nonref = F_ALLOW_NONREF
1409 shrink = F_SHRINK 1825 shrink = F_SHRINK
1410 allow_blessed = F_ALLOW_BLESSED 1826 allow_blessed = F_ALLOW_BLESSED
1411 convert_blessed = F_CONV_BLESSED 1827 convert_blessed = F_CONV_BLESSED
1828 relaxed = F_RELAXED
1829 allow_unknown = F_ALLOW_UNKNOWN
1412 PPCODE: 1830 PPCODE:
1413{ 1831{
1414 if (enable) 1832 if (enable)
1415 self->flags |= ix; 1833 self->flags |= ix;
1416 else 1834 else
1417 self->flags &= ~ix; 1835 self->flags &= ~ix;
1418 1836
1419 XPUSHs (ST (0)); 1837 XPUSHs (ST (0));
1420} 1838}
1421 1839
1422void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1840void get_ascii (JSON *self)
1841 ALIAS:
1842 get_ascii = F_ASCII
1843 get_latin1 = F_LATIN1
1844 get_utf8 = F_UTF8
1845 get_indent = F_INDENT
1846 get_canonical = F_CANONICAL
1847 get_space_before = F_SPACE_BEFORE
1848 get_space_after = F_SPACE_AFTER
1849 get_allow_nonref = F_ALLOW_NONREF
1850 get_shrink = F_SHRINK
1851 get_allow_blessed = F_ALLOW_BLESSED
1852 get_convert_blessed = F_CONV_BLESSED
1853 get_relaxed = F_RELAXED
1854 get_allow_unknown = F_ALLOW_UNKNOWN
1423 PPCODE: 1855 PPCODE:
1424{ 1856 XPUSHs (boolSV (self->flags & ix));
1425 UV log2 = 0;
1426 1857
1427 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1858void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1428 1859 PPCODE:
1429 while ((1UL << log2) < max_depth) 1860 self->max_depth = max_depth;
1430 ++log2;
1431
1432 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1433
1434 XPUSHs (ST (0)); 1861 XPUSHs (ST (0));
1435}
1436 1862
1863U32 get_max_depth (JSON *self)
1864 CODE:
1865 RETVAL = self->max_depth;
1866 OUTPUT:
1867 RETVAL
1868
1437void max_size (JSON *self, UV max_size = 0) 1869void max_size (JSON *self, U32 max_size = 0)
1438 PPCODE: 1870 PPCODE:
1439{ 1871 self->max_size = max_size;
1440 UV log2 = 0;
1441
1442 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1443 if (max_size == 1) max_size = 2;
1444
1445 while ((1UL << log2) < max_size)
1446 ++log2;
1447
1448 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1449
1450 XPUSHs (ST (0)); 1872 XPUSHs (ST (0));
1451} 1873
1874int get_max_size (JSON *self)
1875 CODE:
1876 RETVAL = self->max_size;
1877 OUTPUT:
1878 RETVAL
1452 1879
1453void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1880void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1454 PPCODE: 1881 PPCODE:
1455{ 1882{
1456 SvREFCNT_dec (self->cb_object); 1883 SvREFCNT_dec (self->cb_object);
1490 XPUSHs (decode_json (jsonstr, self, 0)); 1917 XPUSHs (decode_json (jsonstr, self, 0));
1491 1918
1492void decode_prefix (JSON *self, SV *jsonstr) 1919void decode_prefix (JSON *self, SV *jsonstr)
1493 PPCODE: 1920 PPCODE:
1494{ 1921{
1495 UV offset; 1922 char *offset;
1496 EXTEND (SP, 2); 1923 EXTEND (SP, 2);
1497 PUSHs (decode_json (jsonstr, self, &offset)); 1924 PUSHs (decode_json (jsonstr, self, &offset));
1498 PUSHs (sv_2mortal (newSVuv (offset))); 1925 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1926}
1927
1928void incr_parse (JSON *self, SV *jsonstr = 0)
1929 PPCODE:
1930{
1931 if (!self->incr_text)
1932 self->incr_text = newSVpvn ("", 0);
1933
1934 // append data, if any
1935 if (jsonstr)
1936 {
1937 if (SvUTF8 (jsonstr))
1938 {
1939 if (!SvUTF8 (self->incr_text))
1940 {
1941 /* utf-8-ness differs, need to upgrade */
1942 sv_utf8_upgrade (self->incr_text);
1943
1944 if (self->incr_pos)
1945 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1946 - (U8 *)SvPVX (self->incr_text);
1947 }
1948 }
1949 else if (SvUTF8 (self->incr_text))
1950 sv_utf8_upgrade (jsonstr);
1951
1952 {
1953 STRLEN len;
1954 const char *str = SvPV (jsonstr, len);
1955 STRLEN cur = SvCUR (self->incr_text);
1956
1957 if (SvLEN (self->incr_text) <= cur + len)
1958 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1959
1960 Move (str, SvEND (self->incr_text), len, char);
1961 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1962 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1963 }
1964 }
1965
1966 if (GIMME_V != G_VOID)
1967 do
1968 {
1969 char *offset;
1970
1971 if (!INCR_DONE (self))
1972 {
1973 incr_parse (self);
1974
1975 if (self->incr_pos > self->max_size && self->max_size)
1976 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1977 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1978
1979 if (!INCR_DONE (self))
1980 break;
1981 }
1982
1983 XPUSHs (decode_json (self->incr_text, self, &offset));
1984
1985 self->incr_pos -= offset - SvPVX (self->incr_text);
1986 self->incr_nest = 0;
1987 self->incr_mode = 0;
1988
1989 sv_chop (self->incr_text, offset);
1990 }
1991 while (GIMME_V == G_ARRAY);
1992}
1993
1994SV *incr_text (JSON *self)
1995 ATTRS: lvalue
1996 CODE:
1997{
1998 if (self->incr_pos)
1999 croak ("incr_text can not be called when the incremental parser already started parsing");
2000
2001 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2002}
2003 OUTPUT:
2004 RETVAL
2005
2006void incr_skip (JSON *self)
2007 CODE:
2008{
2009 if (self->incr_pos)
2010 {
2011 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2012 self->incr_pos = 0;
2013 self->incr_nest = 0;
2014 self->incr_mode = 0;
2015 }
2016}
2017
2018void incr_reset (JSON *self)
2019 CODE:
2020{
2021 SvREFCNT_dec (self->incr_text);
2022 self->incr_text = 0;
2023 self->incr_pos = 0;
2024 self->incr_nest = 0;
2025 self->incr_mode = 0;
1499} 2026}
1500 2027
1501void DESTROY (JSON *self) 2028void DESTROY (JSON *self)
1502 CODE: 2029 CODE:
1503 SvREFCNT_dec (self->cb_sk_object); 2030 SvREFCNT_dec (self->cb_sk_object);
1504 SvREFCNT_dec (self->cb_object); 2031 SvREFCNT_dec (self->cb_object);
2032 SvREFCNT_dec (self->incr_text);
1505 2033
1506PROTOTYPES: ENABLE 2034PROTOTYPES: ENABLE
1507 2035
1508void to_json (SV *scalar) 2036void encode_json (SV *scalar)
2037 ALIAS:
2038 to_json_ = 0
2039 encode_json = F_UTF8
1509 PPCODE: 2040 PPCODE:
1510{ 2041{
1511 JSON json = { F_DEFAULT | F_UTF8 }; 2042 JSON json;
2043 json_init (&json);
2044 json.flags |= ix;
1512 XPUSHs (encode_json (scalar, &json)); 2045 XPUSHs (encode_json (scalar, &json));
1513} 2046}
1514 2047
1515void from_json (SV *jsonstr) 2048void decode_json (SV *jsonstr)
2049 ALIAS:
2050 from_json_ = 0
2051 decode_json = F_UTF8
1516 PPCODE: 2052 PPCODE:
1517{ 2053{
1518 JSON json = { F_DEFAULT | F_UTF8 }; 2054 JSON json;
2055 json_init (&json);
2056 json.flags |= ix;
1519 XPUSHs (decode_json (jsonstr, &json, 0)); 2057 XPUSHs (decode_json (jsonstr, &json, 0));
1520} 2058}
1521 2059

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines