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 1.47 by root, Sun Jul 1 14:08:03 2007 UTC vs.
Revision 1.108 by root, Tue Aug 17 23:27:36 2010 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// 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
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#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)
19 24
20#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
25#define F_SPACE_BEFORE 0x00000020UL 30#define F_SPACE_BEFORE 0x00000020UL
26#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL // NYI 35#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 36#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 37#define F_ALLOW_UNKNOWN 0x00002000UL
35 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
38 39
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40#define F_DEFAULT (9UL << S_MAXDEPTH)
41 41
42#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
44 44
45#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
46 46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
48
47#define SB do { 49#define SB do {
48#define SE } while (0) 50#define SE } while (0)
49 51
50#if __GNUC__ >= 3 52#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value)) 53# define expect(expr,value) __builtin_expect ((expr), (value))
52# define inline inline 54# define INLINE static inline
53#else 55#else
54# define expect(expr,value) (expr) 56# define expect(expr,value) (expr)
55# define inline static 57# define INLINE static
56#endif 58#endif
57 59
58#define expect_false(expr) expect ((expr) != 0, 0) 60#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1) 61#define expect_true(expr) expect ((expr) != 0, 1)
60 62
63#define IN_RANGE_INC(type,val,beg,end) \
64 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
65 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
66
67#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
68
69#ifdef USE_ITHREADS
70# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
72#else
73# define JSON_SLOW 0
74# define JSON_STASH json_stash
75#endif
76
61static HV *json_stash, *json_boolean_stash; // JSON::XS:: 77static HV *json_stash, *json_boolean_stash; // JSON::XS::
62static SV *json_true, *json_false; 78static SV *json_true, *json_false;
63 79
80enum {
81 INCR_M_WS = 0, // initial whitespace skipping, must be 0
82 INCR_M_STR, // inside string
83 INCR_M_BS, // inside backslash
84 INCR_M_C0, // inside comment in initial whitespace sequence
85 INCR_M_C1, // inside comment in other places
86 INCR_M_JSON // outside anything, count nesting
87};
88
89#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
90
64typedef struct json { 91typedef struct {
65 U32 flags; 92 U32 flags;
93 U32 max_depth;
94 STRLEN max_size;
95
96 SV *cb_object;
97 HV *cb_sk_object;
98
99 // for the incremental parser
100 SV *incr_text; // the source text so far
101 STRLEN incr_pos; // the current offset into the text
102 int incr_nest; // {[]}-nesting level
103 unsigned char incr_mode;
66} JSON__XS; 104} JSON;
105
106INLINE void
107json_init (JSON *json)
108{
109 Zero (json, 1, JSON);
110 json->max_depth = 512;
111}
67 112
68///////////////////////////////////////////////////////////////////////////// 113/////////////////////////////////////////////////////////////////////////////
69// utility functions 114// utility functions
70 115
71static UV * 116INLINE SV *
72SvJSON (SV *sv) 117get_bool (const char *name)
73{ 118{
74 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 119 SV *sv = get_sv (name, 1);
75 croak ("object is not of type JSON::XS");
76 120
77 return &SvUVX (SvRV (sv)); 121 SvREADONLY_on (sv);
78} 122 SvREADONLY_on (SvRV (sv));
79 123
80static void 124 return sv;
125}
126
127INLINE void
81shrink (SV *sv) 128shrink (SV *sv)
82{ 129{
83 sv_utf8_downgrade (sv, 1); 130 sv_utf8_downgrade (sv, 1);
131
84 if (SvLEN (sv) > SvCUR (sv) + 1) 132 if (SvLEN (sv) > SvCUR (sv) + 1)
85 { 133 {
86#ifdef SvPV_shrink_to_cur 134#ifdef SvPV_shrink_to_cur
87 SvPV_shrink_to_cur (sv); 135 SvPV_shrink_to_cur (sv);
88#elif defined (SvPV_renew) 136#elif defined (SvPV_renew)
… …
94// decode an utf-8 character and return it, or (UV)-1 in 142// decode an utf-8 character and return it, or (UV)-1 in
95// case of an error. 143// case of an error.
96// we special-case "safe" characters from U+80 .. U+7FF, 144// we special-case "safe" characters from U+80 .. U+7FF,
97// but use the very good perl function to parse anything else. 145// but use the very good perl function to parse anything else.
98// note that we never call this function for a ascii codepoints 146// note that we never call this function for a ascii codepoints
99inline UV 147INLINE UV
100decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 148decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
101{ 149{
102 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 150 if (expect_true (len >= 2
103 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 151 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
104 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 152 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
105 { 153 {
106 *clen = 2; 154 *clen = 2;
107 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 155 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
108 } 156 }
109 else 157 else
110 { 158 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
111 *clen = (STRLEN)-1; 159}
112 return (UV)-1; 160
161// likewise for encoding, also never called for ascii codepoints
162// this function takes advantage of this fact, although current gccs
163// seem to optimise the check for >= 0x80 away anyways
164INLINE unsigned char *
165encode_utf8 (unsigned char *s, UV ch)
166{
167 if (expect_false (ch < 0x000080))
168 *s++ = ch;
169 else if (expect_true (ch < 0x000800))
170 *s++ = 0xc0 | ( ch >> 6),
171 *s++ = 0x80 | ( ch & 0x3f);
172 else if ( ch < 0x010000)
173 *s++ = 0xe0 | ( ch >> 12),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176 else if ( ch < 0x110000)
177 *s++ = 0xf0 | ( ch >> 18),
178 *s++ = 0x80 | ((ch >> 12) & 0x3f),
179 *s++ = 0x80 | ((ch >> 6) & 0x3f),
180 *s++ = 0x80 | ( ch & 0x3f);
181
182 return s;
183}
184
185// convert offset pointer to character index, sv must be string
186static STRLEN
187ptr_to_index (SV *sv, char *offset)
188{
189 return SvUTF8 (sv)
190 ? utf8_distance (offset, SvPVX (sv))
191 : offset - SvPVX (sv);
192}
193
194/////////////////////////////////////////////////////////////////////////////
195// fp hell
196
197// scan a group of digits, and a trailing exponent
198static void
199json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
200{
201 UV uaccum = 0;
202 int eaccum = 0;
203
204 // if we recurse too deep, skip all remaining digits
205 // to avoid a stack overflow attack
206 if (expect_false (--maxdepth <= 0))
207 while (((U8)*s - '0') < 10)
208 ++s;
209
210 for (;;)
113 } 211 {
114} 212 U8 dig = (U8)*s - '0';
115 213
214 if (expect_false (dig >= 10))
215 {
216 if (dig == (U8)((U8)'.' - (U8)'0'))
217 {
218 ++s;
219 json_atof_scan1 (s, accum, expo, 1, maxdepth);
220 }
221 else if ((dig | ' ') == 'e' - '0')
222 {
223 int exp2 = 0;
224 int neg = 0;
225
226 ++s;
227
228 if (*s == '-')
229 {
230 ++s;
231 neg = 1;
232 }
233 else if (*s == '+')
234 ++s;
235
236 while ((dig = (U8)*s - '0') < 10)
237 exp2 = exp2 * 10 + *s++ - '0';
238
239 *expo += neg ? -exp2 : exp2;
240 }
241
242 break;
243 }
244
245 ++s;
246
247 uaccum = uaccum * 10 + dig;
248 ++eaccum;
249
250 // if we have too many digits, then recurse for more
251 // we actually do this for rather few digits
252 if (uaccum >= (UV_MAX - 9) / 10)
253 {
254 if (postdp) *expo -= eaccum;
255 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
256 if (postdp) *expo += eaccum;
257
258 break;
259 }
260 }
261
262 // this relies greatly on the quality of the pow ()
263 // implementation of the platform, but a good
264 // implementation is hard to beat.
265 if (postdp) *expo -= eaccum;
266 *accum += uaccum * Perl_pow (10., *expo);
267 *expo += eaccum;
268}
269
270static NV
271json_atof (const char *s)
272{
273 NV accum = 0.;
274 int expo = 0;
275 int neg = 0;
276
277 if (*s == '-')
278 {
279 ++s;
280 neg = 1;
281 }
282
283 // a recursion depth of ten gives us >>500 bits
284 json_atof_scan1 (s, &accum, &expo, 0, 10);
285
286 return neg ? -accum : accum;
287}
116///////////////////////////////////////////////////////////////////////////// 288/////////////////////////////////////////////////////////////////////////////
117// encoder 289// encoder
118 290
119// structure used for encoding JSON 291// structure used for encoding JSON
120typedef struct 292typedef struct
121{ 293{
122 char *cur; // SvPVX (sv) + current output position 294 char *cur; // SvPVX (sv) + current output position
123 char *end; // SvEND (sv) 295 char *end; // SvEND (sv)
124 SV *sv; // result scalar 296 SV *sv; // result scalar
125 struct json json; 297 JSON json;
126 U32 indent; // indentation level 298 U32 indent; // indentation level
127 U32 maxdepth; // max. indentation/recursion level 299 UV limit; // escape character values >= this value when encoding
128} enc_t; 300} enc_t;
129 301
130inline void 302INLINE void
131need (enc_t *enc, STRLEN len) 303need (enc_t *enc, STRLEN len)
132{ 304{
133 if (expect_false (enc->cur + len >= enc->end)) 305 if (expect_false (enc->cur + len >= enc->end))
134 { 306 {
135 STRLEN cur = enc->cur - SvPVX (enc->sv); 307 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
136 SvGROW (enc->sv, cur + len + 1); 308 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
137 enc->cur = SvPVX (enc->sv) + cur; 309 enc->cur = SvPVX (enc->sv) + cur;
138 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 310 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
139 } 311 }
140} 312}
141 313
142inline void 314INLINE void
143encode_ch (enc_t *enc, char ch) 315encode_ch (enc_t *enc, char ch)
144{ 316{
145 need (enc, 1); 317 need (enc, 1);
146 *enc->cur++ = ch; 318 *enc->cur++ = ch;
147} 319}
… …
201 { 373 {
202 uch = ch; 374 uch = ch;
203 clen = 1; 375 clen = 1;
204 } 376 }
205 377
206 if (uch > 0x10FFFFUL) 378 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
207 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
208
209 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
210 { 379 {
211 if (uch > 0xFFFFUL) 380 if (uch >= 0x10000UL)
212 { 381 {
382 if (uch >= 0x110000UL)
383 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
384
213 need (enc, len += 11); 385 need (enc, len += 11);
214 sprintf (enc->cur, "\\u%04x\\u%04x", 386 sprintf (enc->cur, "\\u%04x\\u%04x",
215 (int)((uch - 0x10000) / 0x400 + 0xD800), 387 (int)((uch - 0x10000) / 0x400 + 0xD800),
216 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 388 (int)((uch - 0x10000) % 0x400 + 0xDC00));
217 enc->cur += 12; 389 enc->cur += 12;
218 } 390 }
219 else 391 else
220 { 392 {
221 static char hexdigit [16] = "0123456789abcdef";
222 need (enc, len += 5); 393 need (enc, len += 5);
223 *enc->cur++ = '\\'; 394 *enc->cur++ = '\\';
224 *enc->cur++ = 'u'; 395 *enc->cur++ = 'u';
225 *enc->cur++ = hexdigit [ uch >> 12 ]; 396 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
226 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 397 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
227 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 398 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
228 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 399 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
229 } 400 }
230 401
231 str += clen; 402 str += clen;
232 } 403 }
233 else if (enc->json.flags & F_LATIN1) 404 else if (enc->json.flags & F_LATIN1)
… …
245 while (--clen); 416 while (--clen);
246 } 417 }
247 else 418 else
248 { 419 {
249 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 420 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
250 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 421 enc->cur = encode_utf8 (enc->cur, uch);
251 ++str; 422 ++str;
252 } 423 }
253 } 424 }
254 } 425 }
255 } 426 }
256 427
257 --len; 428 --len;
258 } 429 }
259} 430}
260 431
261inline void 432INLINE void
262encode_indent (enc_t *enc) 433encode_indent (enc_t *enc)
263{ 434{
264 if (enc->json.flags & F_INDENT) 435 if (enc->json.flags & F_INDENT)
265 { 436 {
266 int spaces = enc->indent * INDENT_STEP; 437 int spaces = enc->indent * INDENT_STEP;
… …
269 memset (enc->cur, ' ', spaces); 440 memset (enc->cur, ' ', spaces);
270 enc->cur += spaces; 441 enc->cur += spaces;
271 } 442 }
272} 443}
273 444
274inline void 445INLINE void
275encode_space (enc_t *enc) 446encode_space (enc_t *enc)
276{ 447{
277 need (enc, 1); 448 need (enc, 1);
278 encode_ch (enc, ' '); 449 encode_ch (enc, ' ');
279} 450}
280 451
281inline void 452INLINE void
282encode_nl (enc_t *enc) 453encode_nl (enc_t *enc)
283{ 454{
284 if (enc->json.flags & F_INDENT) 455 if (enc->json.flags & F_INDENT)
285 { 456 {
286 need (enc, 1); 457 need (enc, 1);
287 encode_ch (enc, '\n'); 458 encode_ch (enc, '\n');
288 } 459 }
289} 460}
290 461
291inline void 462INLINE void
292encode_comma (enc_t *enc) 463encode_comma (enc_t *enc)
293{ 464{
294 encode_ch (enc, ','); 465 encode_ch (enc, ',');
295 466
296 if (enc->json.flags & F_INDENT) 467 if (enc->json.flags & F_INDENT)
… …
304static void 475static void
305encode_av (enc_t *enc, AV *av) 476encode_av (enc_t *enc, AV *av)
306{ 477{
307 int i, len = av_len (av); 478 int i, len = av_len (av);
308 479
309 if (enc->indent >= enc->maxdepth) 480 if (enc->indent >= enc->json.max_depth)
310 croak ("data structure too deep (hit recursion limit)"); 481 croak (ERR_NESTING_EXCEEDED);
311 482
312 encode_ch (enc, '['); encode_nl (enc); 483 encode_ch (enc, '[');
313 ++enc->indent; 484
485 if (len >= 0)
486 {
487 encode_nl (enc); ++enc->indent;
314 488
315 for (i = 0; i <= len; ++i) 489 for (i = 0; i <= len; ++i)
316 { 490 {
491 SV **svp = av_fetch (av, i, 0);
492
317 encode_indent (enc); 493 encode_indent (enc);
318 encode_sv (enc, *av_fetch (av, i, 0));
319 494
495 if (svp)
496 encode_sv (enc, *svp);
497 else
498 encode_str (enc, "null", 4, 0);
499
320 if (i < len) 500 if (i < len)
321 encode_comma (enc); 501 encode_comma (enc);
322 } 502 }
323 503
504 encode_nl (enc); --enc->indent; encode_indent (enc);
505 }
506
324 encode_nl (enc); 507 encode_ch (enc, ']');
325
326 --enc->indent;
327 encode_indent (enc); encode_ch (enc, ']');
328} 508}
329 509
330static void 510static void
331encode_he (enc_t *enc, HE *he) 511encode_hk (enc_t *enc, HE *he)
332{ 512{
333 encode_ch (enc, '"'); 513 encode_ch (enc, '"');
334 514
335 if (HeKLEN (he) == HEf_SVKEY) 515 if (HeKLEN (he) == HEf_SVKEY)
336 { 516 {
… …
349 encode_ch (enc, '"'); 529 encode_ch (enc, '"');
350 530
351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 531 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
352 encode_ch (enc, ':'); 532 encode_ch (enc, ':');
353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 533 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
354 encode_sv (enc, HeVAL (he));
355} 534}
356 535
357// compare hash entries, used when all keys are bytestrings 536// compare hash entries, used when all keys are bytestrings
358static int 537static int
359he_cmp_fast (const void *a_, const void *b_) 538he_cmp_fast (const void *a_, const void *b_)
… …
364 HE *b = *(HE **)b_; 543 HE *b = *(HE **)b_;
365 544
366 STRLEN la = HeKLEN (a); 545 STRLEN la = HeKLEN (a);
367 STRLEN lb = HeKLEN (b); 546 STRLEN lb = HeKLEN (b);
368 547
369 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 548 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
370 cmp = la - lb; 549 cmp = lb - la;
371 550
372 return cmp; 551 return cmp;
373} 552}
374 553
375// compare hash entries, used when some keys are sv's or utf-x 554// compare hash entries, used when some keys are sv's or utf-x
376static int 555static int
377he_cmp_slow (const void *a, const void *b) 556he_cmp_slow (const void *a, const void *b)
378{ 557{
379 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 558 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
380} 559}
381 560
382static void 561static void
383encode_hv (enc_t *enc, HV *hv) 562encode_hv (enc_t *enc, HV *hv)
384{ 563{
385 int count, i; 564 HE *he;
386 565
387 if (enc->indent >= enc->maxdepth) 566 if (enc->indent >= enc->json.max_depth)
388 croak ("data structure too deep (hit recursion limit)"); 567 croak (ERR_NESTING_EXCEEDED);
389 568
390 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 569 encode_ch (enc, '{');
391 570
392 if ((count = hv_iterinit (hv)))
393 {
394 // for canonical output we have to sort by keys first 571 // for canonical output we have to sort by keys first
395 // actually, this is mostly due to the stupid so-called 572 // actually, this is mostly due to the stupid so-called
396 // security workaround added somewhere in 5.8.x. 573 // security workaround added somewhere in 5.8.x
397 // that randomises hash orderings 574 // that randomises hash orderings
398 if (enc->json.flags & F_CANONICAL) 575 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
576 {
577 int count = hv_iterinit (hv);
578
579 if (SvMAGICAL (hv))
399 { 580 {
581 // need to count by iterating. could improve by dynamically building the vector below
582 // but I don't care for the speed of this special case.
583 // note also that we will run into undefined behaviour when the two iterations
584 // do not result in the same count, something I might care for in some later release.
585
586 count = 0;
587 while (hv_iternext (hv))
588 ++count;
589
590 hv_iterinit (hv);
591 }
592
593 if (count)
594 {
400 int fast = 1; 595 int i, fast = 1;
401 HE *he;
402#if defined(__BORLANDC__) || defined(_MSC_VER) 596#if defined(__BORLANDC__) || defined(_MSC_VER)
403 HE **hes = _alloca (count * sizeof (HE)); 597 HE **hes = _alloca (count * sizeof (HE));
404#else 598#else
405 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 599 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
406#endif 600#endif
… …
433 627
434 FREETMPS; 628 FREETMPS;
435 LEAVE; 629 LEAVE;
436 } 630 }
437 631
438 for (i = 0; i < count; ++i) 632 encode_nl (enc); ++enc->indent;
633
634 while (count--)
439 { 635 {
440 encode_indent (enc); 636 encode_indent (enc);
637 he = hes [count];
441 encode_he (enc, hes [i]); 638 encode_hk (enc, he);
639 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
442 640
443 if (i < count - 1) 641 if (count)
444 encode_comma (enc); 642 encode_comma (enc);
445 } 643 }
446 644
447 encode_nl (enc); 645 encode_nl (enc); --enc->indent; encode_indent (enc);
448 } 646 }
647 }
449 else 648 else
649 {
650 if (hv_iterinit (hv) || SvMAGICAL (hv))
651 if ((he = hv_iternext (hv)))
450 { 652 {
451 HE *he = hv_iternext (hv); 653 encode_nl (enc); ++enc->indent;
452 654
453 for (;;) 655 for (;;)
454 { 656 {
455 encode_indent (enc); 657 encode_indent (enc);
456 encode_he (enc, he); 658 encode_hk (enc, he);
659 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
457 660
458 if (!(he = hv_iternext (hv))) 661 if (!(he = hv_iternext (hv)))
459 break; 662 break;
460 663
461 encode_comma (enc); 664 encode_comma (enc);
462 } 665 }
463 666
464 encode_nl (enc); 667 encode_nl (enc); --enc->indent; encode_indent (enc);
465 } 668 }
466 } 669 }
467 670
468 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 671 encode_ch (enc, '}');
469} 672}
470 673
471// encode objects, arrays and special \0=false and \1=true values. 674// encode objects, arrays and special \0=false and \1=true values.
472static void 675static void
473encode_rv (enc_t *enc, SV *sv) 676encode_rv (enc_t *enc, SV *sv)
… …
477 SvGETMAGIC (sv); 680 SvGETMAGIC (sv);
478 svt = SvTYPE (sv); 681 svt = SvTYPE (sv);
479 682
480 if (expect_false (SvOBJECT (sv))) 683 if (expect_false (SvOBJECT (sv)))
481 { 684 {
685 HV *stash = !JSON_SLOW || json_boolean_stash
686 ? json_boolean_stash
687 : gv_stashpv ("JSON::XS::Boolean", 1);
688
482 if (SvSTASH (sv) == json_boolean_stash) 689 if (SvSTASH (sv) == stash)
483 { 690 {
484 if (SvIV (sv) == 0) 691 if (SvIV (sv))
692 encode_str (enc, "true", 4, 0);
693 else
485 encode_str (enc, "false", 5, 0); 694 encode_str (enc, "false", 5, 0);
486 else
487 encode_str (enc, "true", 4, 0);
488 } 695 }
489 else 696 else
490 { 697 {
491#if 0 698#if 0
492 if (0 && sv_derived_from (rv, "JSON::Literal")) 699 if (0 && sv_derived_from (rv, "JSON::Literal"))
… …
495 } 702 }
496#endif 703#endif
497 if (enc->json.flags & F_CONV_BLESSED) 704 if (enc->json.flags & F_CONV_BLESSED)
498 { 705 {
499 // we re-bless the reference to get overload and other niceties right 706 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 707 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
501 708
502 if (to_json) 709 if (to_json)
503 { 710 {
504 dSP; 711 dSP;
505 ENTER; 712
506 SAVETMPS;
507 PUSHMARK (SP); 713 ENTER; SAVETMPS; PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 714 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509 715
510 // calling with G_SCALAR ensures that we always get a 1 reutrn value 716 // calling with G_SCALAR ensures that we always get a 1 return value
511 // check anyways.
512 PUTBACK; 717 PUTBACK;
513 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 718 call_sv ((SV *)GvCV (to_json), G_SCALAR);
514 SPAGAIN; 719 SPAGAIN;
515 720
721 // catch this surprisingly common error
722 if (SvROK (TOPs) && SvRV (TOPs) == sv)
723 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
724
725 sv = POPs;
726 PUTBACK;
727
516 encode_sv (enc, POPs); 728 encode_sv (enc, sv);
517 729
518 FREETMPS; 730 FREETMPS; LEAVE;
519 LEAVE;
520 } 731 }
521 else if (enc->json.flags & F_ALLOW_BLESSED) 732 else if (enc->json.flags & F_ALLOW_BLESSED)
522 encode_str (enc, "null", 4, 0); 733 encode_str (enc, "null", 4, 0);
523 else 734 else
524 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", 735 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
… …
535 encode_hv (enc, (HV *)sv); 746 encode_hv (enc, (HV *)sv);
536 else if (svt == SVt_PVAV) 747 else if (svt == SVt_PVAV)
537 encode_av (enc, (AV *)sv); 748 encode_av (enc, (AV *)sv);
538 else if (svt < SVt_PVAV) 749 else if (svt < SVt_PVAV)
539 { 750 {
540 if (SvNIOK (sv) && SvIV (sv) == 0) 751 STRLEN len = 0;
752 char *pv = svt ? SvPV (sv, len) : 0;
753
754 if (len == 1 && *pv == '1')
755 encode_str (enc, "true", 4, 0);
756 else if (len == 1 && *pv == '0')
541 encode_str (enc, "false", 5, 0); 757 encode_str (enc, "false", 5, 0);
542 else if (SvNIOK (sv) && SvIV (sv) == 1) 758 else if (enc->json.flags & F_ALLOW_UNKNOWN)
543 encode_str (enc, "true", 4, 0); 759 encode_str (enc, "null", 4, 0);
544 else 760 else
545 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 761 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 762 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547 } 763 }
764 else if (enc->json.flags & F_ALLOW_UNKNOWN)
765 encode_str (enc, "null", 4, 0);
548 else 766 else
549 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 767 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
550 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 768 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
551} 769}
552 770
… …
570 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 788 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
571 enc->cur += strlen (enc->cur); 789 enc->cur += strlen (enc->cur);
572 } 790 }
573 else if (SvIOKp (sv)) 791 else if (SvIOKp (sv))
574 { 792 {
575 // we assume we can always read an IV as a UV 793 // we assume we can always read an IV as a UV and vice versa
576 if (SvUV (sv) & ~(UV)0x7fff) 794 // we assume two's complement
577 { 795 // we assume no aliasing issues in the union
578 // large integer, use the (rather slow) snprintf way. 796 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
579 need (enc, sizeof (UV) * 3); 797 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
580 enc->cur +=
581 SvIsUV(sv)
582 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
583 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
584 }
585 else
586 { 798 {
587 // optimise the "small number case" 799 // optimise the "small number case"
588 // code will likely be branchless and use only a single multiplication 800 // code will likely be branchless and use only a single multiplication
801 // works for numbers up to 59074
589 I32 i = SvIV (sv); 802 I32 i = SvIVX (sv);
590 U32 u; 803 U32 u;
591 char digit, nz = 0; 804 char digit, nz = 0;
592 805
593 need (enc, 6); 806 need (enc, 6);
594 807
… …
600 813
601 // now output digit by digit, each time masking out the integer part 814 // now output digit by digit, each time masking out the integer part
602 // and multiplying by 5 while moving the decimal point one to the right, 815 // and multiplying by 5 while moving the decimal point one to the right,
603 // resulting in a net multiplication by 10. 816 // resulting in a net multiplication by 10.
604 // we always write the digit to memory but conditionally increment 817 // we always write the digit to memory but conditionally increment
605 // the pointer, to ease the usage of conditional move instructions. 818 // the pointer, to enable the use of conditional move instructions.
606 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 819 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
607 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 820 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
608 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 821 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
609 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 822 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
610 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 823 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
611 } 824 }
825 else
826 {
827 // large integer, use the (rather slow) snprintf way.
828 need (enc, IVUV_MAXCHARS);
829 enc->cur +=
830 SvIsUV(sv)
831 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
832 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
833 }
612 } 834 }
613 else if (SvROK (sv)) 835 else if (SvROK (sv))
614 encode_rv (enc, SvRV (sv)); 836 encode_rv (enc, SvRV (sv));
615 else if (!SvOK (sv)) 837 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
616 encode_str (enc, "null", 4, 0); 838 encode_str (enc, "null", 4, 0);
617 else 839 else
618 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 840 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
619 SvPV_nolen (sv), SvFLAGS (sv)); 841 SvPV_nolen (sv), SvFLAGS (sv));
620} 842}
621 843
622static SV * 844static SV *
623encode_json (SV *scalar, struct json *json) 845encode_json (SV *scalar, JSON *json)
624{ 846{
625 enc_t enc; 847 enc_t enc;
626 848
627 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 849 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
628 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 850 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
… …
630 enc.json = *json; 852 enc.json = *json;
631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 853 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
632 enc.cur = SvPVX (enc.sv); 854 enc.cur = SvPVX (enc.sv);
633 enc.end = SvEND (enc.sv); 855 enc.end = SvEND (enc.sv);
634 enc.indent = 0; 856 enc.indent = 0;
635 enc.maxdepth = DEC_DEPTH (enc.json.flags); 857 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
858 : enc.json.flags & F_LATIN1 ? 0x000100UL
859 : 0x110000UL;
636 860
637 SvPOK_only (enc.sv); 861 SvPOK_only (enc.sv);
638 encode_sv (&enc, scalar); 862 encode_sv (&enc, scalar);
863 encode_nl (&enc);
639 864
640 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 865 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
641 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 866 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
642 867
643 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 868 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
656typedef struct 881typedef struct
657{ 882{
658 char *cur; // current parser pointer 883 char *cur; // current parser pointer
659 char *end; // end of input string 884 char *end; // end of input string
660 const char *err; // parse error, if != 0 885 const char *err; // parse error, if != 0
661 struct json json; 886 JSON json;
662 U32 depth; // recursion depth 887 U32 depth; // recursion depth
663 U32 maxdepth; // recursion depth limit 888 U32 maxdepth; // recursion depth limit
664} dec_t; 889} dec_t;
665 890
666inline void 891INLINE void
892decode_comment (dec_t *dec)
893{
894 // only '#'-style comments allowed a.t.m.
895
896 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
897 ++dec->cur;
898}
899
900INLINE void
667decode_ws (dec_t *dec) 901decode_ws (dec_t *dec)
668{ 902{
669 for (;;) 903 for (;;)
670 { 904 {
671 char ch = *dec->cur; 905 char ch = *dec->cur;
672 906
673 if (ch > 0x20 907 if (ch > 0x20)
908 {
909 if (expect_false (ch == '#'))
910 {
911 if (dec->json.flags & F_RELAXED)
912 decode_comment (dec);
913 else
914 break;
915 }
916 else
917 break;
918 }
674 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 919 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
675 break; 920 break; // parse error, but let higher level handle it, gives better error messages
676 921
677 ++dec->cur; 922 ++dec->cur;
678 } 923 }
679} 924}
680 925
… …
684 if (*dec->cur != ch) \ 929 if (*dec->cur != ch) \
685 ERR (# ch " expected"); \ 930 ERR (# ch " expected"); \
686 ++dec->cur; \ 931 ++dec->cur; \
687 SE 932 SE
688 933
689#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 934#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
690#define DEC_DEC_DEPTH --dec->depth 935#define DEC_DEC_DEPTH --dec->depth
691 936
692static SV *decode_sv (dec_t *dec); 937static SV *decode_sv (dec_t *dec);
693 938
694static signed char decode_hexdigit[256]; 939static signed char decode_hexdigit[256];
… …
786 1031
787 if (hi >= 0x80) 1032 if (hi >= 0x80)
788 { 1033 {
789 utf8 = 1; 1034 utf8 = 1;
790 1035
791 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1036 cur = encode_utf8 (cur, hi);
792 } 1037 }
793 else 1038 else
794 *cur++ = hi; 1039 *cur++ = hi;
795 } 1040 }
796 break; 1041 break;
… …
798 default: 1043 default:
799 --dec_cur; 1044 --dec_cur;
800 ERR ("illegal backslash escape sequence in string"); 1045 ERR ("illegal backslash escape sequence in string");
801 } 1046 }
802 } 1047 }
803 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1048 else if (expect_true (ch >= 0x20 && ch < 0x80))
804 *cur++ = ch; 1049 *cur++ = ch;
805 else if (ch >= 0x80) 1050 else if (ch >= 0x80)
806 { 1051 {
807 STRLEN clen; 1052 STRLEN clen;
808 UV uch;
809 1053
810 --dec_cur; 1054 --dec_cur;
811 1055
812 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1056 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
813 if (clen == (STRLEN)-1) 1057 if (clen == (STRLEN)-1)
814 ERR ("malformed UTF-8 character in JSON string"); 1058 ERR ("malformed UTF-8 character in JSON string");
815 1059
816 do 1060 do
817 *cur++ = *dec_cur++; 1061 *cur++ = *dec_cur++;
… …
834 { 1078 {
835 STRLEN len = cur - buf; 1079 STRLEN len = cur - buf;
836 1080
837 if (sv) 1081 if (sv)
838 { 1082 {
839 SvGROW (sv, SvCUR (sv) + len + 1); 1083 STRLEN cur = SvCUR (sv);
1084
1085 if (SvLEN (sv) <= cur + len)
1086 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1087
840 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1088 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
841 SvCUR_set (sv, SvCUR (sv) + len); 1089 SvCUR_set (sv, SvCUR (sv) + len);
842 } 1090 }
843 else 1091 else
844 sv = newSVpvn (buf, len); 1092 sv = newSVpvn (buf, len);
… …
929 is_nv = 1; 1177 is_nv = 1;
930 } 1178 }
931 1179
932 if (!is_nv) 1180 if (!is_nv)
933 { 1181 {
1182 int len = dec->cur - start;
1183
934 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1184 // special case the rather common 1..5-digit-int case
935 if (*start == '-') 1185 if (*start == '-')
936 switch (dec->cur - start) 1186 switch (len)
937 { 1187 {
938 case 2: return newSViv (-( start [1] - '0' * 1)); 1188 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
939 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1189 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
940 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1190 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
941 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1191 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1192 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
942 } 1193 }
943 else 1194 else
944 switch (dec->cur - start) 1195 switch (len)
945 { 1196 {
946 case 1: return newSViv ( start [0] - '0' * 1); 1197 case 1: return newSViv ( start [0] - '0' * 1);
947 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1198 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
948 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1199 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
949 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1200 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1201 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
950 } 1202 }
951 1203
952 { 1204 {
953 UV uv; 1205 UV uv;
954 int numtype = grok_number (start, dec->cur - start, &uv); 1206 int numtype = grok_number (start, len, &uv);
955 if (numtype & IS_NUMBER_IN_UV) 1207 if (numtype & IS_NUMBER_IN_UV)
956 if (numtype & IS_NUMBER_NEG) 1208 if (numtype & IS_NUMBER_NEG)
957 { 1209 {
958 if (uv < (UV)IV_MIN) 1210 if (uv < (UV)IV_MIN)
959 return newSViv (-(IV)uv); 1211 return newSViv (-(IV)uv);
960 } 1212 }
961 else 1213 else
962 return newSVuv (uv); 1214 return newSVuv (uv);
963
964 // here would likely be the place for bigint support
965 } 1215 }
966 }
967 1216
968 // if we ever support bigint or bigfloat, this is the place for bigfloat 1217 len -= *start == '-' ? 1 : 0;
1218
1219 // does not fit into IV or UV, try NV
1220 if (len <= NV_DIG)
1221 // fits into NV without loss of precision
1222 return newSVnv (json_atof (start));
1223
1224 // everything else fails, convert it to a string
1225 return newSVpvn (start, dec->cur - start);
1226 }
1227
1228 // loss of precision here
969 return newSVnv (Atof (start)); 1229 return newSVnv (json_atof (start));
970 1230
971fail: 1231fail:
972 return 0; 1232 return 0;
973} 1233}
974 1234
… …
1003 1263
1004 if (*dec->cur != ',') 1264 if (*dec->cur != ',')
1005 ERR (", or ] expected while parsing array"); 1265 ERR (", or ] expected while parsing array");
1006 1266
1007 ++dec->cur; 1267 ++dec->cur;
1268
1269 decode_ws (dec);
1270
1271 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1272 {
1273 ++dec->cur;
1274 break;
1275 }
1008 } 1276 }
1009 1277
1010 DEC_DEC_DEPTH; 1278 DEC_DEC_DEPTH;
1011 return newRV_noinc ((SV *)av); 1279 return newRV_noinc ((SV *)av);
1012 1280
… …
1017} 1285}
1018 1286
1019static SV * 1287static SV *
1020decode_hv (dec_t *dec) 1288decode_hv (dec_t *dec)
1021{ 1289{
1290 SV *sv;
1022 HV *hv = newHV (); 1291 HV *hv = newHV ();
1023 1292
1024 DEC_INC_DEPTH; 1293 DEC_INC_DEPTH;
1025 decode_ws (dec); 1294 decode_ws (dec);
1026 1295
1027 if (*dec->cur == '}') 1296 if (*dec->cur == '}')
1028 ++dec->cur; 1297 ++dec->cur;
1029 else 1298 else
1030 for (;;) 1299 for (;;)
1031 { 1300 {
1032 decode_ws (dec); EXPECT_CH ('"'); 1301 EXPECT_CH ('"');
1033 1302
1034 // heuristic: assume that 1303 // heuristic: assume that
1035 // a) decode_str + hv_store_ent are abysmally slow. 1304 // a) decode_str + hv_store_ent are abysmally slow.
1036 // b) most hash keys are short, simple ascii text. 1305 // b) most hash keys are short, simple ascii text.
1037 // => try to "fast-match" such strings to avoid 1306 // => try to "fast-match" such strings to avoid
… …
1041 char *p = dec->cur; 1310 char *p = dec->cur;
1042 char *e = p + 24; // only try up to 24 bytes 1311 char *e = p + 24; // only try up to 24 bytes
1043 1312
1044 for (;;) 1313 for (;;)
1045 { 1314 {
1046 // the >= 0x80 is true on most architectures 1315 // the >= 0x80 is false on most architectures
1047 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1316 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1048 { 1317 {
1049 // slow path, back up and use decode_str 1318 // slow path, back up and use decode_str
1050 SV *key = decode_str (dec); 1319 SV *key = decode_str (dec);
1051 if (!key) 1320 if (!key)
1052 goto fail; 1321 goto fail;
1053 1322
1054 decode_ws (dec); EXPECT_CH (':'); 1323 decode_ws (dec); EXPECT_CH (':');
1055 1324
1325 decode_ws (dec);
1056 value = decode_sv (dec); 1326 value = decode_sv (dec);
1057 if (!value) 1327 if (!value)
1058 { 1328 {
1059 SvREFCNT_dec (key); 1329 SvREFCNT_dec (key);
1060 goto fail; 1330 goto fail;
… …
1072 int len = p - key; 1342 int len = p - key;
1073 dec->cur = p + 1; 1343 dec->cur = p + 1;
1074 1344
1075 decode_ws (dec); EXPECT_CH (':'); 1345 decode_ws (dec); EXPECT_CH (':');
1076 1346
1347 decode_ws (dec);
1077 value = decode_sv (dec); 1348 value = decode_sv (dec);
1078 if (!value) 1349 if (!value)
1079 goto fail; 1350 goto fail;
1080 1351
1081 hv_store (hv, key, len, value, 0); 1352 hv_store (hv, key, len, value, 0);
… …
1097 1368
1098 if (*dec->cur != ',') 1369 if (*dec->cur != ',')
1099 ERR (", or } expected while parsing object/hash"); 1370 ERR (", or } expected while parsing object/hash");
1100 1371
1101 ++dec->cur; 1372 ++dec->cur;
1373
1374 decode_ws (dec);
1375
1376 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1377 {
1378 ++dec->cur;
1379 break;
1380 }
1102 } 1381 }
1103 1382
1104 DEC_DEC_DEPTH; 1383 DEC_DEC_DEPTH;
1105 return newRV_noinc ((SV *)hv); 1384 sv = newRV_noinc ((SV *)hv);
1385
1386 // check filter callbacks
1387 if (dec->json.flags & F_HOOK)
1388 {
1389 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1390 {
1391 HE *cb, *he;
1392
1393 hv_iterinit (hv);
1394 he = hv_iternext (hv);
1395 hv_iterinit (hv);
1396
1397 // the next line creates a mortal sv each time its called.
1398 // might want to optimise this for common cases.
1399 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1400
1401 if (cb)
1402 {
1403 dSP;
1404 int count;
1405
1406 ENTER; SAVETMPS; PUSHMARK (SP);
1407 XPUSHs (HeVAL (he));
1408 sv_2mortal (sv);
1409
1410 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1411
1412 if (count == 1)
1413 {
1414 sv = newSVsv (POPs);
1415 FREETMPS; LEAVE;
1416 return sv;
1417 }
1418
1419 SvREFCNT_inc (sv);
1420 FREETMPS; LEAVE;
1421 }
1422 }
1423
1424 if (dec->json.cb_object)
1425 {
1426 dSP;
1427 int count;
1428
1429 ENTER; SAVETMPS; PUSHMARK (SP);
1430 XPUSHs (sv_2mortal (sv));
1431
1432 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1433
1434 if (count == 1)
1435 {
1436 sv = newSVsv (POPs);
1437 FREETMPS; LEAVE;
1438 return sv;
1439 }
1440
1441 SvREFCNT_inc (sv);
1442 FREETMPS; LEAVE;
1443 }
1444 }
1445
1446 return sv;
1106 1447
1107fail: 1448fail:
1108 SvREFCNT_dec (hv); 1449 SvREFCNT_dec (hv);
1109 DEC_DEC_DEPTH; 1450 DEC_DEC_DEPTH;
1110 return 0; 1451 return 0;
1111} 1452}
1112 1453
1113static SV * 1454static SV *
1114decode_sv (dec_t *dec) 1455decode_sv (dec_t *dec)
1115{ 1456{
1116 decode_ws (dec);
1117
1118 // the beauty of JSON: you need exactly one character lookahead 1457 // the beauty of JSON: you need exactly one character lookahead
1119 // to parse anything. 1458 // to parse everything.
1120 switch (*dec->cur) 1459 switch (*dec->cur)
1121 { 1460 {
1122 case '"': ++dec->cur; return decode_str (dec); 1461 case '"': ++dec->cur; return decode_str (dec);
1123 case '[': ++dec->cur; return decode_av (dec); 1462 case '[': ++dec->cur; return decode_av (dec);
1124 case '{': ++dec->cur; return decode_hv (dec); 1463 case '{': ++dec->cur; return decode_hv (dec);
1125 1464
1126 case '-': 1465 case '-':
1127 case '0': case '1': case '2': case '3': case '4': 1466 case '0': case '1': case '2': case '3': case '4':
1128 case '5': case '6': case '7': case '8': case '9': 1467 case '5': case '6': case '7': case '8': case '9':
1129 return decode_num (dec); 1468 return decode_num (dec);
1130 1469
1131 case 't': 1470 case 't':
1132 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1471 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1133 { 1472 {
1134 dec->cur += 4; 1473 dec->cur += 4;
1474#if JSON_SLOW
1475 json_true = get_bool ("JSON::XS::true");
1476#endif
1135 return SvREFCNT_inc (json_true); 1477 return newSVsv (json_true);
1136 } 1478 }
1137 else 1479 else
1138 ERR ("'true' expected"); 1480 ERR ("'true' expected");
1139 1481
1140 break; 1482 break;
1141 1483
1142 case 'f': 1484 case 'f':
1143 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1485 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1144 { 1486 {
1145 dec->cur += 5; 1487 dec->cur += 5;
1488#if JSON_SLOW
1489 json_false = get_bool ("JSON::XS::false");
1490#endif
1146 return SvREFCNT_inc (json_false); 1491 return newSVsv (json_false);
1147 } 1492 }
1148 else 1493 else
1149 ERR ("'false' expected"); 1494 ERR ("'false' expected");
1150 1495
1151 break; 1496 break;
… …
1169fail: 1514fail:
1170 return 0; 1515 return 0;
1171} 1516}
1172 1517
1173static SV * 1518static SV *
1174decode_json (SV *string, struct json *json, UV *offset_return) 1519decode_json (SV *string, JSON *json, char **offset_return)
1175{ 1520{
1176 dec_t dec; 1521 dec_t dec;
1177 UV offset;
1178 SV *sv; 1522 SV *sv;
1179 1523
1524 /* work around bugs in 5.10 where manipulating magic values
1525 * will perl ignore the magic in subsequent accesses
1526 */
1180 SvGETMAGIC (string); 1527 /*SvGETMAGIC (string);*/
1528 if (SvMAGICAL (string))
1529 string = sv_2mortal (newSVsv (string));
1530
1181 SvUPGRADE (string, SVt_PV); 1531 SvUPGRADE (string, SVt_PV);
1182 1532
1183 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1533 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1534 * assertion with -DDEBUGGING, although SvCUR is documented to
1535 * return the xpv_cur field which certainly exists after upgrading.
1536 * according to nicholas clark, calling SvPOK fixes this.
1537 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1538 * and hope for the best.
1539 * Damnit, SvPV_nolen still trips over yet another assertion. This
1540 * assertion business is seriously broken, try yet another workaround
1541 * for the broken -DDEBUGGING.
1542 */
1543 {
1544#ifdef DEBUGGING
1545 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1546#else
1547 STRLEN offset = SvCUR (string);
1548#endif
1549
1550 if (offset > json->max_size && json->max_size)
1184 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1551 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1185 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1552 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1553 }
1186 1554
1187 if (json->flags & F_UTF8) 1555 if (DECODE_WANTS_OCTETS (json))
1188 sv_utf8_downgrade (string, 0); 1556 sv_utf8_downgrade (string, 0);
1189 else 1557 else
1190 sv_utf8_upgrade (string); 1558 sv_utf8_upgrade (string);
1191 1559
1192 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1560 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1193 1561
1194 dec.json = *json; 1562 dec.json = *json;
1195 dec.cur = SvPVX (string); 1563 dec.cur = SvPVX (string);
1196 dec.end = SvEND (string); 1564 dec.end = SvEND (string);
1197 dec.err = 0; 1565 dec.err = 0;
1198 dec.depth = 0; 1566 dec.depth = 0;
1199 dec.maxdepth = DEC_DEPTH (dec.json.flags); 1567
1568 if (dec.json.cb_object || dec.json.cb_sk_object)
1569 dec.json.flags |= F_HOOK;
1200 1570
1201 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1571 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1572
1573 decode_ws (&dec);
1202 sv = decode_sv (&dec); 1574 sv = decode_sv (&dec);
1575
1576 if (offset_return)
1577 *offset_return = dec.cur;
1203 1578
1204 if (!(offset_return || !sv)) 1579 if (!(offset_return || !sv))
1205 { 1580 {
1206 // check for trailing garbage 1581 // check for trailing garbage
1207 decode_ws (&dec); 1582 decode_ws (&dec);
… …
1210 { 1585 {
1211 dec.err = "garbage after JSON object"; 1586 dec.err = "garbage after JSON object";
1212 SvREFCNT_dec (sv); 1587 SvREFCNT_dec (sv);
1213 sv = 0; 1588 sv = 0;
1214 } 1589 }
1215 }
1216
1217 if (offset_return || !sv)
1218 {
1219 offset = dec.json.flags & F_UTF8
1220 ? dec.cur - SvPVX (string)
1221 : utf8_distance (dec.cur, SvPVX (string));
1222
1223 if (offset_return)
1224 *offset_return = offset;
1225 } 1590 }
1226 1591
1227 if (!sv) 1592 if (!sv)
1228 { 1593 {
1229 SV *uni = sv_newmortal (); 1594 SV *uni = sv_newmortal ();
… …
1235 SAVEVPTR (PL_curcop); 1600 SAVEVPTR (PL_curcop);
1236 PL_curcop = &cop; 1601 PL_curcop = &cop;
1237 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1602 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1238 LEAVE; 1603 LEAVE;
1239 1604
1240 croak ("%s, at character offset %d [\"%s\"]", 1605 croak ("%s, at character offset %d (before \"%s\")",
1241 dec.err, 1606 dec.err,
1242 (int)offset, 1607 ptr_to_index (string, dec.cur),
1243 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1608 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1244 } 1609 }
1245 1610
1246 sv = sv_2mortal (sv); 1611 sv = sv_2mortal (sv);
1247 1612
1248 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1613 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1249 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1614 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1250 1615
1251 return sv; 1616 return sv;
1617}
1618
1619/////////////////////////////////////////////////////////////////////////////
1620// incremental parser
1621
1622static void
1623incr_parse (JSON *self)
1624{
1625 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1626
1627 // the state machine here is a bit convoluted and could be simplified a lot
1628 // but this would make it slower, so...
1629
1630 for (;;)
1631 {
1632 //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
1633 switch (self->incr_mode)
1634 {
1635 // only used for initial whitespace skipping
1636 case INCR_M_WS:
1637 for (;;)
1638 {
1639 if (*p > 0x20)
1640 {
1641 if (*p == '#')
1642 {
1643 self->incr_mode = INCR_M_C0;
1644 goto incr_m_c;
1645 }
1646 else
1647 {
1648 self->incr_mode = INCR_M_JSON;
1649 goto incr_m_json;
1650 }
1651 }
1652 else if (!*p)
1653 goto interrupt;
1654
1655 ++p;
1656 }
1657
1658 // skip a single char inside a string (for \\-processing)
1659 case INCR_M_BS:
1660 if (!*p)
1661 goto interrupt;
1662
1663 ++p;
1664 self->incr_mode = INCR_M_STR;
1665 goto incr_m_str;
1666
1667 // inside #-style comments
1668 case INCR_M_C0:
1669 case INCR_M_C1:
1670 incr_m_c:
1671 for (;;)
1672 {
1673 if (*p == '\n')
1674 {
1675 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1676 break;
1677 }
1678 else if (!*p)
1679 goto interrupt;
1680
1681 ++p;
1682 }
1683
1684 break;
1685
1686 // inside a string
1687 case INCR_M_STR:
1688 incr_m_str:
1689 for (;;)
1690 {
1691 if (*p == '"')
1692 {
1693 ++p;
1694 self->incr_mode = INCR_M_JSON;
1695
1696 if (!self->incr_nest)
1697 goto interrupt;
1698
1699 goto incr_m_json;
1700 }
1701 else if (*p == '\\')
1702 {
1703 ++p; // "virtually" consumes character after \
1704
1705 if (!*p) // if at end of string we have to switch modes
1706 {
1707 self->incr_mode = INCR_M_BS;
1708 goto interrupt;
1709 }
1710 }
1711 else if (!*p)
1712 goto interrupt;
1713
1714 ++p;
1715 }
1716
1717 // after initial ws, outside string
1718 case INCR_M_JSON:
1719 incr_m_json:
1720 for (;;)
1721 {
1722 switch (*p++)
1723 {
1724 case 0:
1725 --p;
1726 goto interrupt;
1727
1728 case 0x09:
1729 case 0x0a:
1730 case 0x0d:
1731 case 0x20:
1732 if (!self->incr_nest)
1733 {
1734 --p; // do not eat the whitespace, let the next round do it
1735 goto interrupt;
1736 }
1737 break;
1738
1739 case '"':
1740 self->incr_mode = INCR_M_STR;
1741 goto incr_m_str;
1742
1743 case '[':
1744 case '{':
1745 if (++self->incr_nest > self->max_depth)
1746 croak (ERR_NESTING_EXCEEDED);
1747 break;
1748
1749 case ']':
1750 case '}':
1751 if (--self->incr_nest <= 0)
1752 goto interrupt;
1753 break;
1754
1755 case '#':
1756 self->incr_mode = INCR_M_C1;
1757 goto incr_m_c;
1758 }
1759 }
1760 }
1761
1762 modechange:
1763 ;
1764 }
1765
1766interrupt:
1767 self->incr_pos = p - SvPVX (self->incr_text);
1768 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1769 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1252} 1770}
1253 1771
1254///////////////////////////////////////////////////////////////////////////// 1772/////////////////////////////////////////////////////////////////////////////
1255// XS interface functions 1773// XS interface functions
1256 1774
… …
1268 : -1; 1786 : -1;
1269 1787
1270 json_stash = gv_stashpv ("JSON::XS" , 1); 1788 json_stash = gv_stashpv ("JSON::XS" , 1);
1271 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1789 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1272 1790
1273 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1791 json_true = get_bool ("JSON::XS::true");
1274 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1792 json_false = get_bool ("JSON::XS::false");
1793
1794 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1275} 1795}
1276 1796
1277PROTOTYPES: DISABLE 1797PROTOTYPES: DISABLE
1278 1798
1279SV *new (char *dummy) 1799void CLONE (...)
1280 CODE: 1800 CODE:
1281 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1801 json_stash = 0;
1282 OUTPUT: 1802 json_boolean_stash = 0;
1283 RETVAL
1284 1803
1804void new (char *klass)
1805 PPCODE:
1806{
1807 SV *pv = NEWSV (0, sizeof (JSON));
1808 SvPOK_only (pv);
1809 json_init ((JSON *)SvPVX (pv));
1810 XPUSHs (sv_2mortal (sv_bless (
1811 newRV_noinc (pv),
1812 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1813 )));
1814}
1815
1285SV *ascii (SV *self, int enable = 1) 1816void ascii (JSON *self, int enable = 1)
1286 ALIAS: 1817 ALIAS:
1287 ascii = F_ASCII 1818 ascii = F_ASCII
1288 latin1 = F_LATIN1 1819 latin1 = F_LATIN1
1289 utf8 = F_UTF8 1820 utf8 = F_UTF8
1290 indent = F_INDENT 1821 indent = F_INDENT
… …
1294 pretty = F_PRETTY 1825 pretty = F_PRETTY
1295 allow_nonref = F_ALLOW_NONREF 1826 allow_nonref = F_ALLOW_NONREF
1296 shrink = F_SHRINK 1827 shrink = F_SHRINK
1297 allow_blessed = F_ALLOW_BLESSED 1828 allow_blessed = F_ALLOW_BLESSED
1298 convert_blessed = F_CONV_BLESSED 1829 convert_blessed = F_CONV_BLESSED
1830 relaxed = F_RELAXED
1831 allow_unknown = F_ALLOW_UNKNOWN
1832 PPCODE:
1833{
1834 if (enable)
1835 self->flags |= ix;
1836 else
1837 self->flags &= ~ix;
1838
1839 XPUSHs (ST (0));
1840}
1841
1842void get_ascii (JSON *self)
1843 ALIAS:
1844 get_ascii = F_ASCII
1845 get_latin1 = F_LATIN1
1846 get_utf8 = F_UTF8
1847 get_indent = F_INDENT
1848 get_canonical = F_CANONICAL
1849 get_space_before = F_SPACE_BEFORE
1850 get_space_after = F_SPACE_AFTER
1851 get_allow_nonref = F_ALLOW_NONREF
1852 get_shrink = F_SHRINK
1853 get_allow_blessed = F_ALLOW_BLESSED
1854 get_convert_blessed = F_CONV_BLESSED
1855 get_relaxed = F_RELAXED
1856 get_allow_unknown = F_ALLOW_UNKNOWN
1857 PPCODE:
1858 XPUSHs (boolSV (self->flags & ix));
1859
1860void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1861 PPCODE:
1862 self->max_depth = max_depth;
1863 XPUSHs (ST (0));
1864
1865U32 get_max_depth (JSON *self)
1299 CODE: 1866 CODE:
1300{ 1867 RETVAL = self->max_depth;
1301 UV *uv = SvJSON (self);
1302 if (enable)
1303 *uv |= ix;
1304 else
1305 *uv &= ~ix;
1306
1307 RETVAL = newSVsv (self);
1308}
1309 OUTPUT: 1868 OUTPUT:
1310 RETVAL 1869 RETVAL
1311 1870
1312SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1871void max_size (JSON *self, U32 max_size = 0)
1872 PPCODE:
1873 self->max_size = max_size;
1874 XPUSHs (ST (0));
1875
1876int get_max_size (JSON *self)
1313 CODE: 1877 CODE:
1314{ 1878 RETVAL = self->max_size;
1315 UV *uv = SvJSON (self);
1316 UV log2 = 0;
1317
1318 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1319
1320 while ((1UL << log2) < max_depth)
1321 ++log2;
1322
1323 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1324
1325 RETVAL = newSVsv (self);
1326}
1327 OUTPUT: 1879 OUTPUT:
1328 RETVAL 1880 RETVAL
1329 1881
1330SV *max_size (SV *self, UV max_size = 0) 1882void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1883 PPCODE:
1884{
1885 SvREFCNT_dec (self->cb_object);
1886 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1887
1888 XPUSHs (ST (0));
1889}
1890
1891void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1892 PPCODE:
1893{
1894 if (!self->cb_sk_object)
1895 self->cb_sk_object = newHV ();
1896
1897 if (SvOK (cb))
1898 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1899 else
1900 {
1901 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1902
1903 if (!HvKEYS (self->cb_sk_object))
1904 {
1905 SvREFCNT_dec (self->cb_sk_object);
1906 self->cb_sk_object = 0;
1907 }
1908 }
1909
1910 XPUSHs (ST (0));
1911}
1912
1913void encode (JSON *self, SV *scalar)
1914 PPCODE:
1915 XPUSHs (encode_json (scalar, self));
1916
1917void decode (JSON *self, SV *jsonstr)
1918 PPCODE:
1919 XPUSHs (decode_json (jsonstr, self, 0));
1920
1921void decode_prefix (JSON *self, SV *jsonstr)
1922 PPCODE:
1923{
1924 char *offset;
1925 EXTEND (SP, 2);
1926 PUSHs (decode_json (jsonstr, self, &offset));
1927 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1928}
1929
1930void incr_parse (JSON *self, SV *jsonstr = 0)
1931 PPCODE:
1932{
1933 if (!self->incr_text)
1934 self->incr_text = newSVpvn ("", 0);
1935
1936 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1937 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1938 if (DECODE_WANTS_OCTETS (self))
1939 {
1940 if (self->incr_pos)
1941 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1942 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1943
1944 sv_utf8_downgrade (self->incr_text, 0);
1945 }
1946 else
1947 {
1948 sv_utf8_upgrade (self->incr_text);
1949
1950 if (self->incr_pos)
1951 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1952 - (U8 *)SvPVX (self->incr_text);
1953 }
1954
1955 // append data, if any
1956 if (jsonstr)
1957 {
1958 /* make sure both strings have same encoding */
1959 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1960 if (SvUTF8 (jsonstr))
1961 sv_utf8_downgrade (jsonstr, 0);
1962 else
1963 sv_utf8_upgrade (jsonstr);
1964
1965 /* and then just blindly append */
1966 {
1967 STRLEN len;
1968 const char *str = SvPV (jsonstr, len);
1969 STRLEN cur = SvCUR (self->incr_text);
1970
1971 if (SvLEN (self->incr_text) <= cur + len)
1972 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1973
1974 Move (str, SvEND (self->incr_text), len, char);
1975 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1976 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1977 }
1978 }
1979
1980 if (GIMME_V != G_VOID)
1981 do
1982 {
1983 char *offset;
1984
1985 if (!INCR_DONE (self))
1986 {
1987 incr_parse (self);
1988
1989 if (self->incr_pos > self->max_size && self->max_size)
1990 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1991 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1992
1993 if (!INCR_DONE (self))
1994 break;
1995 }
1996
1997 XPUSHs (decode_json (self->incr_text, self, &offset));
1998
1999 self->incr_pos -= offset - SvPVX (self->incr_text);
2000 self->incr_nest = 0;
2001 self->incr_mode = 0;
2002
2003 sv_chop (self->incr_text, offset);
2004 }
2005 while (GIMME_V == G_ARRAY);
2006}
2007
2008SV *incr_text (JSON *self)
2009 ATTRS: lvalue
1331 CODE: 2010 CODE:
1332{ 2011{
1333 UV *uv = SvJSON (self); 2012 if (self->incr_pos)
1334 UV log2 = 0; 2013 croak ("incr_text can not be called when the incremental parser already started parsing");
1335 2014
1336 if (max_size > 0x80000000UL) max_size = 0x80000000UL; 2015 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1337 if (max_size == 1) max_size = 2;
1338
1339 while ((1UL << log2) < max_size)
1340 ++log2;
1341
1342 *uv = *uv & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1343
1344 RETVAL = newSVsv (self);
1345} 2016}
1346 OUTPUT: 2017 OUTPUT:
1347 RETVAL 2018 RETVAL
1348 2019
2020void incr_skip (JSON *self)
2021 CODE:
2022{
2023 if (self->incr_pos)
2024 {
2025 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2026 self->incr_pos = 0;
2027 self->incr_nest = 0;
2028 self->incr_mode = 0;
2029 }
2030}
2031
2032void incr_reset (JSON *self)
2033 CODE:
2034{
2035 SvREFCNT_dec (self->incr_text);
2036 self->incr_text = 0;
2037 self->incr_pos = 0;
2038 self->incr_nest = 0;
2039 self->incr_mode = 0;
2040}
2041
2042void DESTROY (JSON *self)
2043 CODE:
2044 SvREFCNT_dec (self->cb_sk_object);
2045 SvREFCNT_dec (self->cb_object);
2046 SvREFCNT_dec (self->incr_text);
2047
2048PROTOTYPES: ENABLE
2049
1349void encode (SV *self, SV *scalar) 2050void encode_json (SV *scalar)
2051 ALIAS:
2052 to_json_ = 0
2053 encode_json = F_UTF8
1350 PPCODE: 2054 PPCODE:
1351{ 2055{
1352 struct json json = { *SvJSON (self) }; 2056 JSON json;
2057 json_init (&json);
2058 json.flags |= ix;
1353 XPUSHs (encode_json (scalar, &json)); 2059 XPUSHs (encode_json (scalar, &json));
1354} 2060}
1355 2061
1356void decode (SV *self, SV *jsonstr) 2062void decode_json (SV *jsonstr)
2063 ALIAS:
2064 from_json_ = 0
2065 decode_json = F_UTF8
1357 PPCODE: 2066 PPCODE:
1358{ 2067{
1359 struct json json = { *SvJSON (self) }; 2068 JSON json;
2069 json_init (&json);
2070 json.flags |= ix;
1360 XPUSHs (decode_json (jsonstr, &json, 0)); 2071 XPUSHs (decode_json (jsonstr, &json, 0));
1361} 2072}
1362 2073
1363void decode_prefix (SV *self, SV *jsonstr)
1364 PPCODE:
1365{
1366 UV offset;
1367 struct json json = { *SvJSON (self) };
1368 EXTEND (SP, 2);
1369 PUSHs (decode_json (jsonstr, &json, &offset));
1370 PUSHs (sv_2mortal (newSVuv (offset)));
1371}
1372
1373PROTOTYPES: ENABLE
1374
1375void to_json (SV *scalar)
1376 PPCODE:
1377{
1378 struct json json = { F_DEFAULT | F_UTF8 };
1379 XPUSHs (encode_json (scalar, &json));
1380}
1381
1382void from_json (SV *jsonstr)
1383 PPCODE:
1384{
1385 struct json json = { F_DEFAULT | F_UTF8 };
1386 XPUSHs (decode_json (jsonstr, &json, 0));
1387}
1388

Diff Legend

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