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 1.41 by root, Sat Jun 23 22:53:16 2007 UTC vs.
Revision 1.104 by root, Tue Jan 19 00:31:13 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. 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
19 21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
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
24#define F_CANONICAL 0x00000010UL 29#define F_CANONICAL 0x00000010UL
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
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
29#define F_MAXDEPTH 0xf8000000UL 36#define F_RELAXED 0x00001000UL
30#define S_MAXDEPTH 27 37#define F_ALLOW_UNKNOWN 0x00002000UL
31 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
32#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
33
34// F_SELFCONVERT? <=> to_json/toJson
35// F_BLESSED? <=> { $__class__$ => }
36 39
37#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
38#define F_DEFAULT (9UL << S_MAXDEPTH)
39 41
40#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
41#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
42 44
43#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
44 46
45#define SB do { 47#define SB do {
46#define SE } while (0) 48#define SE } while (0)
47 49
48#if __GNUC__ >= 3 50#if __GNUC__ >= 3
49# define expect(expr,value) __builtin_expect ((expr),(value)) 51# define expect(expr,value) __builtin_expect ((expr), (value))
50# define inline inline 52# define INLINE static inline
51#else 53#else
52# define expect(expr,value) (expr) 54# define expect(expr,value) (expr)
53# define inline static 55# define INLINE static
54#endif 56#endif
55 57
56#define expect_false(expr) expect ((expr) != 0, 0) 58#define expect_false(expr) expect ((expr) != 0, 0)
57#define expect_true(expr) expect ((expr) != 0, 1) 59#define expect_true(expr) expect ((expr) != 0, 1)
58 60
61#define IN_RANGE_INC(type,val,beg,end) \
62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
67#ifdef USE_ITHREADS
68# define JSON_SLOW 1
69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
70#else
71# define JSON_SLOW 0
72# define JSON_STASH json_stash
73#endif
74
59static HV *json_stash; // JSON::XS:: 75static HV *json_stash, *json_boolean_stash; // JSON::XS::
76static SV *json_true, *json_false;
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
89typedef struct {
90 U32 flags;
91 U32 max_depth;
92 STRLEN max_size;
93
94 SV *cb_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;
102} JSON;
103
104INLINE void
105json_init (JSON *json)
106{
107 Zero (json, 1, JSON);
108 json->max_depth = 512;
109}
60 110
61///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
62// utility functions 112// utility functions
63 113
64static UV * 114INLINE SV *
65SvJSON (SV *sv) 115get_bool (const char *name)
66{ 116{
67 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 117 SV *sv = get_sv (name, 1);
68 croak ("object is not of type JSON::XS");
69 118
70 return &SvUVX (SvRV (sv)); 119 SvREADONLY_on (sv);
71} 120 SvREADONLY_on (SvRV (sv));
72 121
73static void 122 return sv;
123}
124
125INLINE void
74shrink (SV *sv) 126shrink (SV *sv)
75{ 127{
76 sv_utf8_downgrade (sv, 1); 128 sv_utf8_downgrade (sv, 1);
129
77 if (SvLEN (sv) > SvCUR (sv) + 1) 130 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 131 {
79#ifdef SvPV_shrink_to_cur 132#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv); 133 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew) 134#elif defined (SvPV_renew)
… …
87// 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
88// case of an error. 141// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF, 142// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else. 143// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints 144// note that we never call this function for a ascii codepoints
92inline UV 145INLINE UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 146decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{ 147{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 148 if (expect_true (len >= 2
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 149 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 150 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
98 { 151 {
99 *clen = 2; 152 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 153 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 } 154 }
102 else 155 else
103 { 156 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
104 *clen = (STRLEN)-1; 157}
105 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)
198{
199 UV uaccum = 0;
200 int eaccum = 0;
201
202 for (;;)
106 } 203 {
107} 204 U8 dig = (U8)*s - '0';
108 205
206 if (expect_false (dig >= 10))
207 {
208 if (dig == (U8)((U8)'.' - (U8)'0'))
209 {
210 ++s;
211 json_atof_scan1 (s, accum, expo, 1);
212 }
213 else if ((dig | ' ') == 'e' - '0')
214 {
215 int exp2 = 0;
216 int neg = 0;
217
218 ++s;
219
220 if (*s == '-')
221 {
222 ++s;
223 neg = 1;
224 }
225 else if (*s == '+')
226 ++s;
227
228 while ((dig = (U8)*s - '0') < 10)
229 exp2 = exp2 * 10 + *s++ - '0';
230
231 *expo += neg ? -exp2 : exp2;
232 }
233
234 break;
235 }
236
237 ++s;
238
239 uaccum = uaccum * 10 + dig;
240 ++eaccum;
241
242 // if we have too many digits, then recurse for more
243 // we actually do this for rather few digits
244 if (uaccum >= (UV_MAX - 9) / 10)
245 {
246 if (postdp) *expo -= eaccum;
247 json_atof_scan1 (s, accum, expo, postdp);
248 if (postdp) *expo += eaccum;
249
250 break;
251 }
252 }
253
254 if (postdp) *expo -= eaccum;
255 *accum += uaccum * pow (10., *expo);
256 *expo += eaccum;
257}
258
259static NV
260json_atof (const char *s)
261{
262 NV accum = 0.;
263 int expo = 0;
264 int neg = 0;
265
266 if (*s == '-')
267 {
268 ++s;
269 neg = 1;
270 }
271
272 json_atof_scan1 (s, &accum, &expo, 0);
273
274 return neg ? -accum : accum;
275}
109///////////////////////////////////////////////////////////////////////////// 276/////////////////////////////////////////////////////////////////////////////
110// encoder 277// encoder
111 278
112// structure used for encoding JSON 279// structure used for encoding JSON
113typedef struct 280typedef struct
114{ 281{
115 char *cur; // SvPVX (sv) + current output position 282 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 283 char *end; // SvEND (sv)
117 SV *sv; // result scalar 284 SV *sv; // result scalar
118 U32 flags; // F_* 285 JSON json;
119 U32 indent; // indentation level 286 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 287 UV limit; // escape character values >= this value when encoding
121} enc_t; 288} enc_t;
122 289
123inline void 290INLINE void
124need (enc_t *enc, STRLEN len) 291need (enc_t *enc, STRLEN len)
125{ 292{
126 if (expect_false (enc->cur + len >= enc->end)) 293 if (expect_false (enc->cur + len >= enc->end))
127 { 294 {
128 STRLEN cur = enc->cur - SvPVX (enc->sv); 295 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
129 SvGROW (enc->sv, cur + len + 1); 296 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
130 enc->cur = SvPVX (enc->sv) + cur; 297 enc->cur = SvPVX (enc->sv) + cur;
131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 298 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
132 } 299 }
133} 300}
134 301
135inline void 302INLINE void
136encode_ch (enc_t *enc, char ch) 303encode_ch (enc_t *enc, char ch)
137{ 304{
138 need (enc, 1); 305 need (enc, 1);
139 *enc->cur++ = ch; 306 *enc->cur++ = ch;
140} 307}
… …
194 { 361 {
195 uch = ch; 362 uch = ch;
196 clen = 1; 363 clen = 1;
197 } 364 }
198 365
199 if (uch > 0x10FFFFUL) 366 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
203 { 367 {
204 if (uch > 0xFFFFUL) 368 if (uch >= 0x10000UL)
205 { 369 {
370 if (uch >= 0x110000UL)
371 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
372
206 need (enc, len += 11); 373 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 374 sprintf (enc->cur, "\\u%04x\\u%04x",
208 (int)((uch - 0x10000) / 0x400 + 0xD800), 375 (int)((uch - 0x10000) / 0x400 + 0xD800),
209 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 376 (int)((uch - 0x10000) % 0x400 + 0xDC00));
210 enc->cur += 12; 377 enc->cur += 12;
211 } 378 }
212 else 379 else
213 { 380 {
214 static char hexdigit [16] = "0123456789abcdef";
215 need (enc, len += 5); 381 need (enc, len += 5);
216 *enc->cur++ = '\\'; 382 *enc->cur++ = '\\';
217 *enc->cur++ = 'u'; 383 *enc->cur++ = 'u';
218 *enc->cur++ = hexdigit [ uch >> 12 ]; 384 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
219 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 385 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
220 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 386 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
221 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 387 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
222 } 388 }
223 389
224 str += clen; 390 str += clen;
225 } 391 }
226 else if (enc->flags & F_LATIN1) 392 else if (enc->json.flags & F_LATIN1)
227 { 393 {
228 *enc->cur++ = uch; 394 *enc->cur++ = uch;
229 str += clen; 395 str += clen;
230 } 396 }
231 else if (is_utf8) 397 else if (is_utf8)
… …
238 while (--clen); 404 while (--clen);
239 } 405 }
240 else 406 else
241 { 407 {
242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 408 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 409 enc->cur = encode_utf8 (enc->cur, uch);
244 ++str; 410 ++str;
245 } 411 }
246 } 412 }
247 } 413 }
248 } 414 }
249 415
250 --len; 416 --len;
251 } 417 }
252} 418}
253 419
254inline void 420INLINE void
255encode_indent (enc_t *enc) 421encode_indent (enc_t *enc)
256{ 422{
257 if (enc->flags & F_INDENT) 423 if (enc->json.flags & F_INDENT)
258 { 424 {
259 int spaces = enc->indent * INDENT_STEP; 425 int spaces = enc->indent * INDENT_STEP;
260 426
261 need (enc, spaces); 427 need (enc, spaces);
262 memset (enc->cur, ' ', spaces); 428 memset (enc->cur, ' ', spaces);
263 enc->cur += spaces; 429 enc->cur += spaces;
264 } 430 }
265} 431}
266 432
267inline void 433INLINE void
268encode_space (enc_t *enc) 434encode_space (enc_t *enc)
269{ 435{
270 need (enc, 1); 436 need (enc, 1);
271 encode_ch (enc, ' '); 437 encode_ch (enc, ' ');
272} 438}
273 439
274inline void 440INLINE void
275encode_nl (enc_t *enc) 441encode_nl (enc_t *enc)
276{ 442{
277 if (enc->flags & F_INDENT) 443 if (enc->json.flags & F_INDENT)
278 { 444 {
279 need (enc, 1); 445 need (enc, 1);
280 encode_ch (enc, '\n'); 446 encode_ch (enc, '\n');
281 } 447 }
282} 448}
283 449
284inline void 450INLINE void
285encode_comma (enc_t *enc) 451encode_comma (enc_t *enc)
286{ 452{
287 encode_ch (enc, ','); 453 encode_ch (enc, ',');
288 454
289 if (enc->flags & F_INDENT) 455 if (enc->json.flags & F_INDENT)
290 encode_nl (enc); 456 encode_nl (enc);
291 else if (enc->flags & F_SPACE_AFTER) 457 else if (enc->json.flags & F_SPACE_AFTER)
292 encode_space (enc); 458 encode_space (enc);
293} 459}
294 460
295static void encode_sv (enc_t *enc, SV *sv); 461static void encode_sv (enc_t *enc, SV *sv);
296 462
297static void 463static void
298encode_av (enc_t *enc, AV *av) 464encode_av (enc_t *enc, AV *av)
299{ 465{
300 int i, len = av_len (av); 466 int i, len = av_len (av);
301 467
302 if (enc->indent >= enc->maxdepth) 468 if (enc->indent >= enc->json.max_depth)
303 croak ("data structure too deep (hit recursion limit)"); 469 croak (ERR_NESTING_EXCEEDED);
304 470
305 encode_ch (enc, '['); encode_nl (enc); 471 encode_ch (enc, '[');
306 ++enc->indent; 472
473 if (len >= 0)
474 {
475 encode_nl (enc); ++enc->indent;
307 476
308 for (i = 0; i <= len; ++i) 477 for (i = 0; i <= len; ++i)
309 { 478 {
479 SV **svp = av_fetch (av, i, 0);
480
310 encode_indent (enc); 481 encode_indent (enc);
311 encode_sv (enc, *av_fetch (av, i, 0));
312 482
483 if (svp)
484 encode_sv (enc, *svp);
485 else
486 encode_str (enc, "null", 4, 0);
487
313 if (i < len) 488 if (i < len)
314 encode_comma (enc); 489 encode_comma (enc);
315 } 490 }
316 491
492 encode_nl (enc); --enc->indent; encode_indent (enc);
493 }
494
317 encode_nl (enc); 495 encode_ch (enc, ']');
318
319 --enc->indent;
320 encode_indent (enc); encode_ch (enc, ']');
321} 496}
322 497
323static void 498static void
324encode_he (enc_t *enc, HE *he) 499encode_hk (enc_t *enc, HE *he)
325{ 500{
326 encode_ch (enc, '"'); 501 encode_ch (enc, '"');
327 502
328 if (HeKLEN (he) == HEf_SVKEY) 503 if (HeKLEN (he) == HEf_SVKEY)
329 { 504 {
… …
339 else 514 else
340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 515 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
341 516
342 encode_ch (enc, '"'); 517 encode_ch (enc, '"');
343 518
344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 519 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
345 encode_ch (enc, ':'); 520 encode_ch (enc, ':');
346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 521 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
347 encode_sv (enc, HeVAL (he));
348} 522}
349 523
350// compare hash entries, used when all keys are bytestrings 524// compare hash entries, used when all keys are bytestrings
351static int 525static int
352he_cmp_fast (const void *a_, const void *b_) 526he_cmp_fast (const void *a_, const void *b_)
… …
357 HE *b = *(HE **)b_; 531 HE *b = *(HE **)b_;
358 532
359 STRLEN la = HeKLEN (a); 533 STRLEN la = HeKLEN (a);
360 STRLEN lb = HeKLEN (b); 534 STRLEN lb = HeKLEN (b);
361 535
362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 536 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
363 cmp = la - lb; 537 cmp = lb - la;
364 538
365 return cmp; 539 return cmp;
366} 540}
367 541
368// compare hash entries, used when some keys are sv's or utf-x 542// compare hash entries, used when some keys are sv's or utf-x
369static int 543static int
370he_cmp_slow (const void *a, const void *b) 544he_cmp_slow (const void *a, const void *b)
371{ 545{
372 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 546 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
373} 547}
374 548
375static void 549static void
376encode_hv (enc_t *enc, HV *hv) 550encode_hv (enc_t *enc, HV *hv)
377{ 551{
378 int count, i; 552 HE *he;
379 553
380 if (enc->indent >= enc->maxdepth) 554 if (enc->indent >= enc->json.max_depth)
381 croak ("data structure too deep (hit recursion limit)"); 555 croak (ERR_NESTING_EXCEEDED);
382 556
383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 557 encode_ch (enc, '{');
384 558
385 if ((count = hv_iterinit (hv)))
386 {
387 // for canonical output we have to sort by keys first 559 // for canonical output we have to sort by keys first
388 // actually, this is mostly due to the stupid so-called 560 // actually, this is mostly due to the stupid so-called
389 // security workaround added somewhere in 5.8.x. 561 // security workaround added somewhere in 5.8.x
390 // that randomises hash orderings 562 // that randomises hash orderings
391 if (enc->flags & F_CANONICAL) 563 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
564 {
565 int count = hv_iterinit (hv);
566
567 if (SvMAGICAL (hv))
392 { 568 {
569 // need to count by iterating. could improve by dynamically building the vector below
570 // but I don't care for the speed of this special case.
571 // note also that we will run into undefined behaviour when the two iterations
572 // do not result in the same count, something I might care for in some later release.
573
574 count = 0;
575 while (hv_iternext (hv))
576 ++count;
577
578 hv_iterinit (hv);
579 }
580
581 if (count)
582 {
393 int fast = 1; 583 int i, fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER) 584#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE)); 585 HE **hes = _alloca (count * sizeof (HE));
397#else 586#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 587 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif 588#endif
… …
426 615
427 FREETMPS; 616 FREETMPS;
428 LEAVE; 617 LEAVE;
429 } 618 }
430 619
431 for (i = 0; i < count; ++i) 620 encode_nl (enc); ++enc->indent;
621
622 while (count--)
432 { 623 {
433 encode_indent (enc); 624 encode_indent (enc);
625 he = hes [count];
434 encode_he (enc, hes [i]); 626 encode_hk (enc, he);
627 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
435 628
436 if (i < count - 1) 629 if (count)
437 encode_comma (enc); 630 encode_comma (enc);
438 } 631 }
439 632
633 encode_nl (enc); --enc->indent; encode_indent (enc);
634 }
635 }
636 else
637 {
638 if (hv_iterinit (hv) || SvMAGICAL (hv))
639 if ((he = hv_iternext (hv)))
640 {
641 encode_nl (enc); ++enc->indent;
642
643 for (;;)
644 {
440 encode_nl (enc); 645 encode_indent (enc);
646 encode_hk (enc, he);
647 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
648
649 if (!(he = hv_iternext (hv)))
650 break;
651
652 encode_comma (enc);
653 }
654
655 encode_nl (enc); --enc->indent; encode_indent (enc);
656 }
657 }
658
659 encode_ch (enc, '}');
660}
661
662// encode objects, arrays and special \0=false and \1=true values.
663static void
664encode_rv (enc_t *enc, SV *sv)
665{
666 svtype svt;
667
668 SvGETMAGIC (sv);
669 svt = SvTYPE (sv);
670
671 if (expect_false (SvOBJECT (sv)))
672 {
673 HV *stash = !JSON_SLOW || json_boolean_stash
674 ? json_boolean_stash
675 : gv_stashpv ("JSON::XS::Boolean", 1);
676
677 if (SvSTASH (sv) == stash)
678 {
679 if (SvIV (sv))
680 encode_str (enc, "true", 4, 0);
681 else
682 encode_str (enc, "false", 5, 0);
441 } 683 }
442 else 684 else
443 { 685 {
444 HE *he = hv_iternext (hv); 686#if 0
445 687 if (0 && sv_derived_from (rv, "JSON::Literal"))
446 for (;;)
447 { 688 {
448 encode_indent (enc); 689 // not yet
449 encode_he (enc, he);
450
451 if (!(he = hv_iternext (hv)))
452 break;
453
454 encode_comma (enc);
455 } 690 }
691#endif
692 if (enc->json.flags & F_CONV_BLESSED)
693 {
694 // we re-bless the reference to get overload and other niceties right
695 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
456 696
697 if (to_json)
698 {
699 dSP;
700
701 ENTER; SAVETMPS; PUSHMARK (SP);
702 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
703
704 // calling with G_SCALAR ensures that we always get a 1 return value
705 PUTBACK;
706 call_sv ((SV *)GvCV (to_json), G_SCALAR);
707 SPAGAIN;
708
709 // catch this surprisingly common error
710 if (SvROK (TOPs) && SvRV (TOPs) == sv)
711 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
712
713 sv = POPs;
714 PUTBACK;
715
457 encode_nl (enc); 716 encode_sv (enc, sv);
717
718 FREETMPS; LEAVE;
719 }
720 else if (enc->json.flags & F_ALLOW_BLESSED)
721 encode_str (enc, "null", 4, 0);
722 else
723 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
724 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
725 }
726 else if (enc->json.flags & F_ALLOW_BLESSED)
727 encode_str (enc, "null", 4, 0);
728 else
729 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
730 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
458 } 731 }
459 } 732 }
460
461 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462}
463
464// encode objects, arrays and special \0=false and \1=true values.
465static void
466encode_rv (enc_t *enc, SV *sv)
467{
468 svtype svt;
469
470 SvGETMAGIC (sv);
471 svt = SvTYPE (sv);
472
473 if (svt == SVt_PVHV) 733 else if (svt == SVt_PVHV)
474 encode_hv (enc, (HV *)sv); 734 encode_hv (enc, (HV *)sv);
475 else if (svt == SVt_PVAV) 735 else if (svt == SVt_PVAV)
476 encode_av (enc, (AV *)sv); 736 encode_av (enc, (AV *)sv);
477 else if (svt < SVt_PVAV) 737 else if (svt < SVt_PVAV)
478 { 738 {
479 if (SvNIOK (sv) && SvIV (sv) == 0) 739 STRLEN len = 0;
740 char *pv = svt ? SvPV (sv, len) : 0;
741
742 if (len == 1 && *pv == '1')
743 encode_str (enc, "true", 4, 0);
744 else if (len == 1 && *pv == '0')
480 encode_str (enc, "false", 5, 0); 745 encode_str (enc, "false", 5, 0);
481 else if (SvNIOK (sv) && SvIV (sv) == 1) 746 else if (enc->json.flags & F_ALLOW_UNKNOWN)
482 encode_str (enc, "true", 4, 0); 747 encode_str (enc, "null", 4, 0);
483 else 748 else
484 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 749 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 750 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
486 } 751 }
752 else if (enc->json.flags & F_ALLOW_UNKNOWN)
753 encode_str (enc, "null", 4, 0);
487 else 754 else
488 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 755 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
489 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 756 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
490} 757}
491 758
… …
509 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 776 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
510 enc->cur += strlen (enc->cur); 777 enc->cur += strlen (enc->cur);
511 } 778 }
512 else if (SvIOKp (sv)) 779 else if (SvIOKp (sv))
513 { 780 {
514 // we assume we can always read an IV as a UV 781 // we assume we can always read an IV as a UV and vice versa
515 if (SvUV (sv) & ~(UV)0x7fff) 782 // we assume two's complement
516 { 783 // we assume no aliasing issues in the union
517 // large integer, use the (rather slow) snprintf way. 784 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
518 need (enc, sizeof (UV) * 3); 785 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
519 enc->cur +=
520 SvIsUV(sv)
521 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
522 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
523 }
524 else
525 { 786 {
526 // optimise the "small number case" 787 // optimise the "small number case"
527 // code will likely be branchless and use only a single multiplication 788 // code will likely be branchless and use only a single multiplication
789 // works for numbers up to 59074
528 I32 i = SvIV (sv); 790 I32 i = SvIVX (sv);
529 U32 u; 791 U32 u;
530 char digit, nz = 0; 792 char digit, nz = 0;
531 793
532 need (enc, 6); 794 need (enc, 6);
533 795
… …
539 801
540 // now output digit by digit, each time masking out the integer part 802 // now output digit by digit, each time masking out the integer part
541 // and multiplying by 5 while moving the decimal point one to the right, 803 // and multiplying by 5 while moving the decimal point one to the right,
542 // resulting in a net multiplication by 10. 804 // resulting in a net multiplication by 10.
543 // we always write the digit to memory but conditionally increment 805 // we always write the digit to memory but conditionally increment
544 // the pointer, to ease the usage of conditional move instructions. 806 // the pointer, to enable the use of conditional move instructions.
545 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 807 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
546 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 808 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
547 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 809 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
548 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 810 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
549 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 811 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
550 } 812 }
813 else
814 {
815 // large integer, use the (rather slow) snprintf way.
816 need (enc, IVUV_MAXCHARS);
817 enc->cur +=
818 SvIsUV(sv)
819 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
820 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
821 }
551 } 822 }
552 else if (SvROK (sv)) 823 else if (SvROK (sv))
553 encode_rv (enc, SvRV (sv)); 824 encode_rv (enc, SvRV (sv));
554 else if (!SvOK (sv)) 825 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
555 encode_str (enc, "null", 4, 0); 826 encode_str (enc, "null", 4, 0);
556 else 827 else
557 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 828 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
558 SvPV_nolen (sv), SvFLAGS (sv)); 829 SvPV_nolen (sv), SvFLAGS (sv));
559} 830}
560 831
561static SV * 832static SV *
562encode_json (SV *scalar, U32 flags) 833encode_json (SV *scalar, JSON *json)
563{ 834{
564 enc_t enc; 835 enc_t enc;
565 836
566 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 837 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
567 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 838 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
568 839
569 enc.flags = flags; 840 enc.json = *json;
570 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 841 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
571 enc.cur = SvPVX (enc.sv); 842 enc.cur = SvPVX (enc.sv);
572 enc.end = SvEND (enc.sv); 843 enc.end = SvEND (enc.sv);
573 enc.indent = 0; 844 enc.indent = 0;
574 enc.maxdepth = DEC_DEPTH (flags); 845 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
846 : enc.json.flags & F_LATIN1 ? 0x000100UL
847 : 0x110000UL;
575 848
576 SvPOK_only (enc.sv); 849 SvPOK_only (enc.sv);
577 encode_sv (&enc, scalar); 850 encode_sv (&enc, scalar);
851 encode_nl (&enc);
578 852
579 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 853 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
580 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 854 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
581 855
582 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 856 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
583 SvUTF8_on (enc.sv); 857 SvUTF8_on (enc.sv);
584 858
585 if (enc.flags & F_SHRINK) 859 if (enc.json.flags & F_SHRINK)
586 shrink (enc.sv); 860 shrink (enc.sv);
587 861
588 return enc.sv; 862 return enc.sv;
589} 863}
590 864
… …
595typedef struct 869typedef struct
596{ 870{
597 char *cur; // current parser pointer 871 char *cur; // current parser pointer
598 char *end; // end of input string 872 char *end; // end of input string
599 const char *err; // parse error, if != 0 873 const char *err; // parse error, if != 0
600 U32 flags; // F_* 874 JSON json;
601 U32 depth; // recursion depth 875 U32 depth; // recursion depth
602 U32 maxdepth; // recursion depth limit 876 U32 maxdepth; // recursion depth limit
603} dec_t; 877} dec_t;
604 878
605inline void 879INLINE void
880decode_comment (dec_t *dec)
881{
882 // only '#'-style comments allowed a.t.m.
883
884 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
885 ++dec->cur;
886}
887
888INLINE void
606decode_ws (dec_t *dec) 889decode_ws (dec_t *dec)
607{ 890{
608 for (;;) 891 for (;;)
609 { 892 {
610 char ch = *dec->cur; 893 char ch = *dec->cur;
611 894
612 if (ch > 0x20 895 if (ch > 0x20)
896 {
897 if (expect_false (ch == '#'))
898 {
899 if (dec->json.flags & F_RELAXED)
900 decode_comment (dec);
901 else
902 break;
903 }
904 else
905 break;
906 }
613 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 907 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
614 break; 908 break; // parse error, but let higher level handle it, gives better error messages
615 909
616 ++dec->cur; 910 ++dec->cur;
617 } 911 }
618} 912}
619 913
… …
623 if (*dec->cur != ch) \ 917 if (*dec->cur != ch) \
624 ERR (# ch " expected"); \ 918 ERR (# ch " expected"); \
625 ++dec->cur; \ 919 ++dec->cur; \
626 SE 920 SE
627 921
628#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 922#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
629#define DEC_DEC_DEPTH --dec->depth 923#define DEC_DEC_DEPTH --dec->depth
630 924
631static SV *decode_sv (dec_t *dec); 925static SV *decode_sv (dec_t *dec);
632 926
633static signed char decode_hexdigit[256]; 927static signed char decode_hexdigit[256];
… …
725 1019
726 if (hi >= 0x80) 1020 if (hi >= 0x80)
727 { 1021 {
728 utf8 = 1; 1022 utf8 = 1;
729 1023
730 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1024 cur = encode_utf8 (cur, hi);
731 } 1025 }
732 else 1026 else
733 *cur++ = hi; 1027 *cur++ = hi;
734 } 1028 }
735 break; 1029 break;
… …
737 default: 1031 default:
738 --dec_cur; 1032 --dec_cur;
739 ERR ("illegal backslash escape sequence in string"); 1033 ERR ("illegal backslash escape sequence in string");
740 } 1034 }
741 } 1035 }
742 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1036 else if (expect_true (ch >= 0x20 && ch < 0x80))
743 *cur++ = ch; 1037 *cur++ = ch;
744 else if (ch >= 0x80) 1038 else if (ch >= 0x80)
745 { 1039 {
746 STRLEN clen; 1040 STRLEN clen;
747 UV uch;
748 1041
749 --dec_cur; 1042 --dec_cur;
750 1043
751 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1044 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
752 if (clen == (STRLEN)-1) 1045 if (clen == (STRLEN)-1)
753 ERR ("malformed UTF-8 character in JSON string"); 1046 ERR ("malformed UTF-8 character in JSON string");
754 1047
755 do 1048 do
756 *cur++ = *dec_cur++; 1049 *cur++ = *dec_cur++;
… …
773 { 1066 {
774 STRLEN len = cur - buf; 1067 STRLEN len = cur - buf;
775 1068
776 if (sv) 1069 if (sv)
777 { 1070 {
778 SvGROW (sv, SvCUR (sv) + len + 1); 1071 STRLEN cur = SvCUR (sv);
1072
1073 if (SvLEN (sv) <= cur + len)
1074 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1075
779 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1076 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
780 SvCUR_set (sv, SvCUR (sv) + len); 1077 SvCUR_set (sv, SvCUR (sv) + len);
781 } 1078 }
782 else 1079 else
783 sv = newSVpvn (buf, len); 1080 sv = newSVpvn (buf, len);
… …
868 is_nv = 1; 1165 is_nv = 1;
869 } 1166 }
870 1167
871 if (!is_nv) 1168 if (!is_nv)
872 { 1169 {
1170 int len = dec->cur - start;
1171
873 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1172 // special case the rather common 1..5-digit-int case
874 if (*start == '-') 1173 if (*start == '-')
875 switch (dec->cur - start) 1174 switch (len)
876 { 1175 {
877 case 2: return newSViv (-( start [1] - '0' * 1)); 1176 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
878 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1177 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
879 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1178 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
880 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1179 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1180 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
881 } 1181 }
882 else 1182 else
883 switch (dec->cur - start) 1183 switch (len)
884 { 1184 {
885 case 1: return newSViv ( start [0] - '0' * 1); 1185 case 1: return newSViv ( start [0] - '0' * 1);
886 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1186 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
887 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1187 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
888 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1188 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1189 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
889 } 1190 }
890 1191
891 { 1192 {
892 UV uv; 1193 UV uv;
893 int numtype = grok_number (start, dec->cur - start, &uv); 1194 int numtype = grok_number (start, len, &uv);
894 if (numtype & IS_NUMBER_IN_UV) 1195 if (numtype & IS_NUMBER_IN_UV)
895 if (numtype & IS_NUMBER_NEG) 1196 if (numtype & IS_NUMBER_NEG)
896 { 1197 {
897 if (uv < (UV)IV_MIN) 1198 if (uv < (UV)IV_MIN)
898 return newSViv (-(IV)uv); 1199 return newSViv (-(IV)uv);
899 } 1200 }
900 else 1201 else
901 return newSVuv (uv); 1202 return newSVuv (uv);
902
903 // here would likely be the place for bigint support
904 } 1203 }
905 }
906 1204
907 // if we ever support bigint or bigfloat, this is the place for bigfloat 1205 len -= *start == '-' ? 1 : 0;
1206
1207 // does not fit into IV or UV, try NV
1208 if (len <= NV_DIG)
1209 // fits into NV without loss of precision
1210 return newSVnv (json_atof (start));
1211
1212 // everything else fails, convert it to a string
1213 return newSVpvn (start, dec->cur - start);
1214 }
1215
1216 // loss of precision here
908 return newSVnv (Atof (start)); 1217 return newSVnv (json_atof (start));
909 1218
910fail: 1219fail:
911 return 0; 1220 return 0;
912} 1221}
913 1222
… …
942 1251
943 if (*dec->cur != ',') 1252 if (*dec->cur != ',')
944 ERR (", or ] expected while parsing array"); 1253 ERR (", or ] expected while parsing array");
945 1254
946 ++dec->cur; 1255 ++dec->cur;
1256
1257 decode_ws (dec);
1258
1259 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1260 {
1261 ++dec->cur;
1262 break;
1263 }
947 } 1264 }
948 1265
949 DEC_DEC_DEPTH; 1266 DEC_DEC_DEPTH;
950 return newRV_noinc ((SV *)av); 1267 return newRV_noinc ((SV *)av);
951 1268
… …
956} 1273}
957 1274
958static SV * 1275static SV *
959decode_hv (dec_t *dec) 1276decode_hv (dec_t *dec)
960{ 1277{
1278 SV *sv;
961 HV *hv = newHV (); 1279 HV *hv = newHV ();
962 1280
963 DEC_INC_DEPTH; 1281 DEC_INC_DEPTH;
964 decode_ws (dec); 1282 decode_ws (dec);
965 1283
966 if (*dec->cur == '}') 1284 if (*dec->cur == '}')
967 ++dec->cur; 1285 ++dec->cur;
968 else 1286 else
969 for (;;) 1287 for (;;)
970 { 1288 {
1289 EXPECT_CH ('"');
1290
1291 // heuristic: assume that
1292 // a) decode_str + hv_store_ent are abysmally slow.
1293 // b) most hash keys are short, simple ascii text.
1294 // => try to "fast-match" such strings to avoid
1295 // the overhead of decode_str + hv_store_ent.
1296 {
971 SV *key, *value; 1297 SV *value;
1298 char *p = dec->cur;
1299 char *e = p + 24; // only try up to 24 bytes
972 1300
973 decode_ws (dec); EXPECT_CH ('"'); 1301 for (;;)
974
975 key = decode_str (dec);
976 if (!key)
977 goto fail;
978
979 decode_ws (dec); EXPECT_CH (':');
980
981 value = decode_sv (dec);
982 if (!value)
983 { 1302 {
1303 // the >= 0x80 is false on most architectures
1304 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1305 {
1306 // slow path, back up and use decode_str
1307 SV *key = decode_str (dec);
1308 if (!key)
1309 goto fail;
1310
1311 decode_ws (dec); EXPECT_CH (':');
1312
1313 decode_ws (dec);
1314 value = decode_sv (dec);
1315 if (!value)
1316 {
1317 SvREFCNT_dec (key);
1318 goto fail;
1319 }
1320
1321 hv_store_ent (hv, key, value, 0);
984 SvREFCNT_dec (key); 1322 SvREFCNT_dec (key);
1323
1324 break;
1325 }
1326 else if (*p == '"')
1327 {
1328 // fast path, got a simple key
1329 char *key = dec->cur;
1330 int len = p - key;
1331 dec->cur = p + 1;
1332
1333 decode_ws (dec); EXPECT_CH (':');
1334
1335 decode_ws (dec);
1336 value = decode_sv (dec);
1337 if (!value)
985 goto fail; 1338 goto fail;
1339
1340 hv_store (hv, key, len, value, 0);
1341
1342 break;
1343 }
1344
1345 ++p;
986 } 1346 }
987 1347 }
988 hv_store_ent (hv, key, value, 0);
989 SvREFCNT_dec (key);
990 1348
991 decode_ws (dec); 1349 decode_ws (dec);
992 1350
993 if (*dec->cur == '}') 1351 if (*dec->cur == '}')
994 { 1352 {
… …
998 1356
999 if (*dec->cur != ',') 1357 if (*dec->cur != ',')
1000 ERR (", or } expected while parsing object/hash"); 1358 ERR (", or } expected while parsing object/hash");
1001 1359
1002 ++dec->cur; 1360 ++dec->cur;
1361
1362 decode_ws (dec);
1363
1364 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1365 {
1366 ++dec->cur;
1367 break;
1368 }
1003 } 1369 }
1004 1370
1005 DEC_DEC_DEPTH; 1371 DEC_DEC_DEPTH;
1006 return newRV_noinc ((SV *)hv); 1372 sv = newRV_noinc ((SV *)hv);
1373
1374 // check filter callbacks
1375 if (dec->json.flags & F_HOOK)
1376 {
1377 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1378 {
1379 HE *cb, *he;
1380
1381 hv_iterinit (hv);
1382 he = hv_iternext (hv);
1383 hv_iterinit (hv);
1384
1385 // the next line creates a mortal sv each time its called.
1386 // might want to optimise this for common cases.
1387 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1388
1389 if (cb)
1390 {
1391 dSP;
1392 int count;
1393
1394 ENTER; SAVETMPS; PUSHMARK (SP);
1395 XPUSHs (HeVAL (he));
1396
1397 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1398
1399 if (count == 1)
1400 {
1401 sv = newSVsv (POPs);
1402 FREETMPS; LEAVE;
1403 return sv;
1404 }
1405
1406 FREETMPS; LEAVE;
1407 }
1408 }
1409
1410 if (dec->json.cb_object)
1411 {
1412 dSP;
1413 int count;
1414
1415 ENTER; SAVETMPS; PUSHMARK (SP);
1416 XPUSHs (sv_2mortal (sv));
1417
1418 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1419
1420 if (count == 1)
1421 {
1422 sv = newSVsv (POPs);
1423 FREETMPS; LEAVE;
1424 return sv;
1425 }
1426
1427 SvREFCNT_inc (sv);
1428 FREETMPS; LEAVE;
1429 }
1430 }
1431
1432 return sv;
1007 1433
1008fail: 1434fail:
1009 SvREFCNT_dec (hv); 1435 SvREFCNT_dec (hv);
1010 DEC_DEC_DEPTH; 1436 DEC_DEC_DEPTH;
1011 return 0; 1437 return 0;
1012} 1438}
1013 1439
1014static SV * 1440static SV *
1015decode_sv (dec_t *dec) 1441decode_sv (dec_t *dec)
1016{ 1442{
1017 decode_ws (dec);
1018
1019 // the beauty of JSON: you need exactly one character lookahead 1443 // the beauty of JSON: you need exactly one character lookahead
1020 // to parse anything. 1444 // to parse everything.
1021 switch (*dec->cur) 1445 switch (*dec->cur)
1022 { 1446 {
1023 case '"': ++dec->cur; return decode_str (dec); 1447 case '"': ++dec->cur; return decode_str (dec);
1024 case '[': ++dec->cur; return decode_av (dec); 1448 case '[': ++dec->cur; return decode_av (dec);
1025 case '{': ++dec->cur; return decode_hv (dec); 1449 case '{': ++dec->cur; return decode_hv (dec);
1026 1450
1027 case '-': 1451 case '-':
1028 case '0': case '1': case '2': case '3': case '4': 1452 case '0': case '1': case '2': case '3': case '4':
1029 case '5': case '6': case '7': case '8': case '9': 1453 case '5': case '6': case '7': case '8': case '9':
1030 return decode_num (dec); 1454 return decode_num (dec);
1031 1455
1032 case 't': 1456 case 't':
1033 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1457 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1034 { 1458 {
1035 dec->cur += 4; 1459 dec->cur += 4;
1460#if JSON_SLOW
1461 json_true = get_bool ("JSON::XS::true");
1462#endif
1036 return newSViv (1); 1463 return newSVsv (json_true);
1037 } 1464 }
1038 else 1465 else
1039 ERR ("'true' expected"); 1466 ERR ("'true' expected");
1040 1467
1041 break; 1468 break;
1042 1469
1043 case 'f': 1470 case 'f':
1044 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1471 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1045 { 1472 {
1046 dec->cur += 5; 1473 dec->cur += 5;
1474#if JSON_SLOW
1475 json_false = get_bool ("JSON::XS::false");
1476#endif
1047 return newSViv (0); 1477 return newSVsv (json_false);
1048 } 1478 }
1049 else 1479 else
1050 ERR ("'false' expected"); 1480 ERR ("'false' expected");
1051 1481
1052 break; 1482 break;
… …
1070fail: 1500fail:
1071 return 0; 1501 return 0;
1072} 1502}
1073 1503
1074static SV * 1504static SV *
1075decode_json (SV *string, U32 flags, UV *offset_return) 1505decode_json (SV *string, JSON *json, char **offset_return)
1076{ 1506{
1077 dec_t dec; 1507 dec_t dec;
1078 UV offset;
1079 SV *sv; 1508 SV *sv;
1080 1509
1510 /* work around bugs in 5.10 where manipulating magic values
1511 * will perl ignore the magic in subsequent accesses
1512 */
1081 SvGETMAGIC (string); 1513 /*SvGETMAGIC (string);*/
1514 if (SvMAGICAL (string))
1515 string = sv_2mortal (newSVsv (string));
1516
1082 SvUPGRADE (string, SVt_PV); 1517 SvUPGRADE (string, SVt_PV);
1083 1518
1519 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1520 * assertion with -DDEBUGGING, although SvCUR is documented to
1521 * return the xpv_cur field which certainly exists after upgrading.
1522 * according to nicholas clark, calling SvPOK fixes this.
1523 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1524 * and hope for the best.
1525 * Damnit, SvPV_nolen still trips over yet another assertion. This
1526 * assertion business is seriously broken, try yet another workaround
1527 * for the broken -DDEBUGGING.
1528 */
1529 {
1530#ifdef DEBUGGING
1531 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1532#else
1533 STRLEN offset = SvCUR (string);
1534#endif
1535
1536 if (offset > json->max_size && json->max_size)
1537 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1538 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1539 }
1540
1084 if (flags & F_UTF8) 1541 if (json->flags & F_UTF8)
1085 sv_utf8_downgrade (string, 0); 1542 sv_utf8_downgrade (string, 0);
1086 else 1543 else
1087 sv_utf8_upgrade (string); 1544 sv_utf8_upgrade (string);
1088 1545
1089 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1546 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1090 1547
1091 dec.flags = flags; 1548 dec.json = *json;
1092 dec.cur = SvPVX (string); 1549 dec.cur = SvPVX (string);
1093 dec.end = SvEND (string); 1550 dec.end = SvEND (string);
1094 dec.err = 0; 1551 dec.err = 0;
1095 dec.depth = 0; 1552 dec.depth = 0;
1096 dec.maxdepth = DEC_DEPTH (dec.flags); 1553
1554 if (dec.json.cb_object || dec.json.cb_sk_object)
1555 dec.json.flags |= F_HOOK;
1097 1556
1098 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1557 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1558
1559 decode_ws (&dec);
1099 sv = decode_sv (&dec); 1560 sv = decode_sv (&dec);
1561
1562 if (offset_return)
1563 *offset_return = dec.cur;
1100 1564
1101 if (!(offset_return || !sv)) 1565 if (!(offset_return || !sv))
1102 { 1566 {
1103 // check for trailing garbage 1567 // check for trailing garbage
1104 decode_ws (&dec); 1568 decode_ws (&dec);
… …
1107 { 1571 {
1108 dec.err = "garbage after JSON object"; 1572 dec.err = "garbage after JSON object";
1109 SvREFCNT_dec (sv); 1573 SvREFCNT_dec (sv);
1110 sv = 0; 1574 sv = 0;
1111 } 1575 }
1112 }
1113
1114 if (offset_return || !sv)
1115 {
1116 offset = dec.flags & F_UTF8
1117 ? dec.cur - SvPVX (string)
1118 : utf8_distance (dec.cur, SvPVX (string));
1119
1120 if (offset_return)
1121 *offset_return = offset;
1122 } 1576 }
1123 1577
1124 if (!sv) 1578 if (!sv)
1125 { 1579 {
1126 SV *uni = sv_newmortal (); 1580 SV *uni = sv_newmortal ();
… …
1132 SAVEVPTR (PL_curcop); 1586 SAVEVPTR (PL_curcop);
1133 PL_curcop = &cop; 1587 PL_curcop = &cop;
1134 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1588 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1135 LEAVE; 1589 LEAVE;
1136 1590
1137 croak ("%s, at character offset %d [\"%s\"]", 1591 croak ("%s, at character offset %d (before \"%s\")",
1138 dec.err, 1592 dec.err,
1139 (int)offset, 1593 ptr_to_index (string, dec.cur),
1140 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1594 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1141 } 1595 }
1142 1596
1143 sv = sv_2mortal (sv); 1597 sv = sv_2mortal (sv);
1144 1598
1145 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1599 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1146 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1600 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1147 1601
1148 return sv; 1602 return sv;
1603}
1604
1605/////////////////////////////////////////////////////////////////////////////
1606// incremental parser
1607
1608static void
1609incr_parse (JSON *self)
1610{
1611 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1612
1613 // the state machine here is a bit convoluted and could be simplified a lot
1614 // but this would make it slower, so...
1615
1616 for (;;)
1617 {
1618 //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
1619 switch (self->incr_mode)
1620 {
1621 // only used for initial whitespace skipping
1622 case INCR_M_WS:
1623 for (;;)
1624 {
1625 if (*p > 0x20)
1626 {
1627 if (*p == '#')
1628 {
1629 self->incr_mode = INCR_M_C0;
1630 goto incr_m_c;
1631 }
1632 else
1633 {
1634 self->incr_mode = INCR_M_JSON;
1635 goto incr_m_json;
1636 }
1637 }
1638 else if (!*p)
1639 goto interrupt;
1640
1641 ++p;
1642 }
1643
1644 // skip a single char inside a string (for \\-processing)
1645 case INCR_M_BS:
1646 if (!*p)
1647 goto interrupt;
1648
1649 ++p;
1650 self->incr_mode = INCR_M_STR;
1651 goto incr_m_str;
1652
1653 // inside #-style comments
1654 case INCR_M_C0:
1655 case INCR_M_C1:
1656 incr_m_c:
1657 for (;;)
1658 {
1659 if (*p == '\n')
1660 {
1661 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1662 break;
1663 }
1664 else if (!*p)
1665 goto interrupt;
1666
1667 ++p;
1668 }
1669
1670 break;
1671
1672 // inside a string
1673 case INCR_M_STR:
1674 incr_m_str:
1675 for (;;)
1676 {
1677 if (*p == '"')
1678 {
1679 ++p;
1680 self->incr_mode = INCR_M_JSON;
1681
1682 if (!self->incr_nest)
1683 goto interrupt;
1684
1685 goto incr_m_json;
1686 }
1687 else if (*p == '\\')
1688 {
1689 ++p; // "virtually" consumes character after \
1690
1691 if (!*p) // if at end of string we have to switch modes
1692 {
1693 self->incr_mode = INCR_M_BS;
1694 goto interrupt;
1695 }
1696 }
1697 else if (!*p)
1698 goto interrupt;
1699
1700 ++p;
1701 }
1702
1703 // after initial ws, outside string
1704 case INCR_M_JSON:
1705 incr_m_json:
1706 for (;;)
1707 {
1708 switch (*p++)
1709 {
1710 case 0:
1711 --p;
1712 goto interrupt;
1713
1714 case 0x09:
1715 case 0x0a:
1716 case 0x0d:
1717 case 0x20:
1718 if (!self->incr_nest)
1719 {
1720 --p; // do not eat the whitespace, let the next round do it
1721 goto interrupt;
1722 }
1723 break;
1724
1725 case '"':
1726 self->incr_mode = INCR_M_STR;
1727 goto incr_m_str;
1728
1729 case '[':
1730 case '{':
1731 if (++self->incr_nest > self->max_depth)
1732 croak (ERR_NESTING_EXCEEDED);
1733 break;
1734
1735 case ']':
1736 case '}':
1737 if (--self->incr_nest <= 0)
1738 goto interrupt;
1739 break;
1740
1741 case '#':
1742 self->incr_mode = INCR_M_C1;
1743 goto incr_m_c;
1744 }
1745 }
1746 }
1747
1748 modechange:
1749 ;
1750 }
1751
1752interrupt:
1753 self->incr_pos = p - SvPVX (self->incr_text);
1754 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1755 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1149} 1756}
1150 1757
1151///////////////////////////////////////////////////////////////////////////// 1758/////////////////////////////////////////////////////////////////////////////
1152// XS interface functions 1759// XS interface functions
1153 1760
… …
1162 i >= '0' && i <= '9' ? i - '0' 1769 i >= '0' && i <= '9' ? i - '0'
1163 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1770 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1164 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1771 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1165 : -1; 1772 : -1;
1166 1773
1167 json_stash = gv_stashpv ("JSON::XS", 1); 1774 json_stash = gv_stashpv ("JSON::XS" , 1);
1775 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1776
1777 json_true = get_bool ("JSON::XS::true");
1778 json_false = get_bool ("JSON::XS::false");
1779
1780 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1168} 1781}
1169 1782
1170PROTOTYPES: DISABLE 1783PROTOTYPES: DISABLE
1171 1784
1172SV *new (char *dummy) 1785void CLONE (...)
1173 CODE: 1786 CODE:
1174 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1787 json_stash = 0;
1788 json_boolean_stash = 0;
1789
1790void new (char *klass)
1791 PPCODE:
1792{
1793 SV *pv = NEWSV (0, sizeof (JSON));
1794 SvPOK_only (pv);
1795 json_init ((JSON *)SvPVX (pv));
1796 XPUSHs (sv_2mortal (sv_bless (
1797 newRV_noinc (pv),
1798 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1799 )));
1800}
1801
1802void ascii (JSON *self, int enable = 1)
1803 ALIAS:
1804 ascii = F_ASCII
1805 latin1 = F_LATIN1
1806 utf8 = F_UTF8
1807 indent = F_INDENT
1808 canonical = F_CANONICAL
1809 space_before = F_SPACE_BEFORE
1810 space_after = F_SPACE_AFTER
1811 pretty = F_PRETTY
1812 allow_nonref = F_ALLOW_NONREF
1813 shrink = F_SHRINK
1814 allow_blessed = F_ALLOW_BLESSED
1815 convert_blessed = F_CONV_BLESSED
1816 relaxed = F_RELAXED
1817 allow_unknown = F_ALLOW_UNKNOWN
1818 PPCODE:
1819{
1820 if (enable)
1821 self->flags |= ix;
1822 else
1823 self->flags &= ~ix;
1824
1825 XPUSHs (ST (0));
1826}
1827
1828void get_ascii (JSON *self)
1829 ALIAS:
1830 get_ascii = F_ASCII
1831 get_latin1 = F_LATIN1
1832 get_utf8 = F_UTF8
1833 get_indent = F_INDENT
1834 get_canonical = F_CANONICAL
1835 get_space_before = F_SPACE_BEFORE
1836 get_space_after = F_SPACE_AFTER
1837 get_allow_nonref = F_ALLOW_NONREF
1838 get_shrink = F_SHRINK
1839 get_allow_blessed = F_ALLOW_BLESSED
1840 get_convert_blessed = F_CONV_BLESSED
1841 get_relaxed = F_RELAXED
1842 get_allow_unknown = F_ALLOW_UNKNOWN
1843 PPCODE:
1844 XPUSHs (boolSV (self->flags & ix));
1845
1846void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1847 PPCODE:
1848 self->max_depth = max_depth;
1849 XPUSHs (ST (0));
1850
1851U32 get_max_depth (JSON *self)
1852 CODE:
1853 RETVAL = self->max_depth;
1175 OUTPUT: 1854 OUTPUT:
1176 RETVAL 1855 RETVAL
1177 1856
1178SV *ascii (SV *self, int enable = 1) 1857void max_size (JSON *self, U32 max_size = 0)
1179 ALIAS: 1858 PPCODE:
1180 ascii = F_ASCII 1859 self->max_size = max_size;
1181 latin1 = F_LATIN1 1860 XPUSHs (ST (0));
1182 utf8 = F_UTF8 1861
1183 indent = F_INDENT 1862int get_max_size (JSON *self)
1184 canonical = F_CANONICAL
1185 space_before = F_SPACE_BEFORE
1186 space_after = F_SPACE_AFTER
1187 pretty = F_PRETTY
1188 allow_nonref = F_ALLOW_NONREF
1189 shrink = F_SHRINK
1190 CODE: 1863 CODE:
1191{ 1864 RETVAL = self->max_size;
1192 UV *uv = SvJSON (self);
1193 if (enable)
1194 *uv |= ix;
1195 else
1196 *uv &= ~ix;
1197
1198 RETVAL = newSVsv (self);
1199}
1200 OUTPUT: 1865 OUTPUT:
1201 RETVAL 1866 RETVAL
1202 1867
1203SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1868void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1869 PPCODE:
1870{
1871 SvREFCNT_dec (self->cb_object);
1872 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1873
1874 XPUSHs (ST (0));
1875}
1876
1877void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1878 PPCODE:
1879{
1880 if (!self->cb_sk_object)
1881 self->cb_sk_object = newHV ();
1882
1883 if (SvOK (cb))
1884 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1885 else
1886 {
1887 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1888
1889 if (!HvKEYS (self->cb_sk_object))
1890 {
1891 SvREFCNT_dec (self->cb_sk_object);
1892 self->cb_sk_object = 0;
1893 }
1894 }
1895
1896 XPUSHs (ST (0));
1897}
1898
1899void encode (JSON *self, SV *scalar)
1900 PPCODE:
1901 XPUSHs (encode_json (scalar, self));
1902
1903void decode (JSON *self, SV *jsonstr)
1904 PPCODE:
1905 XPUSHs (decode_json (jsonstr, self, 0));
1906
1907void decode_prefix (JSON *self, SV *jsonstr)
1908 PPCODE:
1909{
1910 char *offset;
1911 EXTEND (SP, 2);
1912 PUSHs (decode_json (jsonstr, self, &offset));
1913 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1914}
1915
1916void incr_parse (JSON *self, SV *jsonstr = 0)
1917 PPCODE:
1918{
1919 if (!self->incr_text)
1920 self->incr_text = newSVpvn ("", 0);
1921
1922 // append data, if any
1923 if (jsonstr)
1924 {
1925 if (SvUTF8 (jsonstr))
1926 {
1927 if (!SvUTF8 (self->incr_text))
1928 {
1929 /* utf-8-ness differs, need to upgrade */
1930 sv_utf8_upgrade (self->incr_text);
1931
1932 if (self->incr_pos)
1933 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1934 - (U8 *)SvPVX (self->incr_text);
1935 }
1936 }
1937 else if (SvUTF8 (self->incr_text))
1938 sv_utf8_upgrade (jsonstr);
1939
1940 {
1941 STRLEN len;
1942 const char *str = SvPV (jsonstr, len);
1943 STRLEN cur = SvCUR (self->incr_text);
1944
1945 if (SvLEN (self->incr_text) <= cur + len)
1946 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1947
1948 Move (str, SvEND (self->incr_text), len, char);
1949 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1950 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1951 }
1952 }
1953
1954 if (GIMME_V != G_VOID)
1955 do
1956 {
1957 char *offset;
1958
1959 if (!INCR_DONE (self))
1960 {
1961 incr_parse (self);
1962
1963 if (self->incr_pos > self->max_size && self->max_size)
1964 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1965 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1966
1967 if (!INCR_DONE (self))
1968 break;
1969 }
1970
1971 XPUSHs (decode_json (self->incr_text, self, &offset));
1972
1973 self->incr_pos -= offset - SvPVX (self->incr_text);
1974 self->incr_nest = 0;
1975 self->incr_mode = 0;
1976
1977 sv_chop (self->incr_text, offset);
1978 }
1979 while (GIMME_V == G_ARRAY);
1980}
1981
1982SV *incr_text (JSON *self)
1983 ATTRS: lvalue
1204 CODE: 1984 CODE:
1205{ 1985{
1206 UV *uv = SvJSON (self); 1986 if (self->incr_pos)
1207 UV log2 = 0; 1987 croak ("incr_text can not be called when the incremental parser already started parsing");
1208 1988
1209 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1989 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1210
1211 while ((1UL << log2) < max_depth)
1212 ++log2;
1213
1214 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1215
1216 RETVAL = newSVsv (self);
1217} 1990}
1218 OUTPUT: 1991 OUTPUT:
1219 RETVAL 1992 RETVAL
1220 1993
1994void incr_skip (JSON *self)
1995 CODE:
1996{
1997 if (self->incr_pos)
1998 {
1999 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2000 self->incr_pos = 0;
2001 self->incr_nest = 0;
2002 self->incr_mode = 0;
2003 }
2004}
2005
2006void incr_reset (JSON *self)
2007 CODE:
2008{
2009 SvREFCNT_dec (self->incr_text);
2010 self->incr_text = 0;
2011 self->incr_pos = 0;
2012 self->incr_nest = 0;
2013 self->incr_mode = 0;
2014}
2015
2016void DESTROY (JSON *self)
2017 CODE:
2018 SvREFCNT_dec (self->cb_sk_object);
2019 SvREFCNT_dec (self->cb_object);
2020 SvREFCNT_dec (self->incr_text);
2021
2022PROTOTYPES: ENABLE
2023
1221void encode (SV *self, SV *scalar) 2024void encode_json (SV *scalar)
2025 ALIAS:
2026 to_json_ = 0
2027 encode_json = F_UTF8
1222 PPCODE: 2028 PPCODE:
2029{
2030 JSON json;
2031 json_init (&json);
2032 json.flags |= ix;
1223 XPUSHs (encode_json (scalar, *SvJSON (self))); 2033 XPUSHs (encode_json (scalar, &json));
2034}
1224 2035
1225void decode (SV *self, SV *jsonstr) 2036void decode_json (SV *jsonstr)
2037 ALIAS:
2038 from_json_ = 0
2039 decode_json = F_UTF8
1226 PPCODE: 2040 PPCODE:
2041{
2042 JSON json;
2043 json_init (&json);
2044 json.flags |= ix;
1227 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 2045 XPUSHs (decode_json (jsonstr, &json, 0));
1228
1229void decode_prefix (SV *self, SV *jsonstr)
1230 PPCODE:
1231{
1232 UV offset;
1233 EXTEND (SP, 2);
1234 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1235 PUSHs (sv_2mortal (newSVuv (offset)));
1236} 2046}
1237 2047
1238PROTOTYPES: ENABLE
1239
1240void to_json (SV *scalar)
1241 ALIAS:
1242 objToJson = 0
1243 PPCODE:
1244 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1245
1246void from_json (SV *jsonstr)
1247 ALIAS:
1248 jsonToObj = 0
1249 PPCODE:
1250 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1251

Diff Legend

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