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 1.39 by root, Mon Jun 11 03:42:57 2007 UTC vs.
Revision 1.116 by root, Thu May 23 09:31:32 2013 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
16// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24
14#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 29#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 30#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 36#define F_RELAXED 0x00001000UL
24#define S_MAXDEPTH 27 37#define F_ALLOW_UNKNOWN 0x00002000UL
25 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 39
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH)
33 41
34#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
36 44
37#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
38 46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
48
39#define SB do { 49#define SB do {
40#define SE } while (0) 50#define SE } while (0)
41 51
42#if __GNUC__ >= 3 52#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 53# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 54# define INLINE static inline
45#else 55#else
46# define expect(expr,value) (expr) 56# define expect(expr,value) (expr)
47# define inline static 57# define INLINE static
48#endif 58#endif
49 59
50#define expect_false(expr) expect ((expr) != 0, 0) 60#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 61#define expect_true(expr) expect ((expr) != 0, 1)
52 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
77// the amount of HEs to allocate on the stack, when sorting keys
78#define STACK_HES 64
79
53static HV *json_stash; // JSON::XS:: 80static HV *json_stash, *json_boolean_stash; // JSON::XS::
81static SV *json_true, *json_false;
82
83enum {
84 INCR_M_WS = 0, // initial whitespace skipping, must be 0
85 INCR_M_STR, // inside string
86 INCR_M_BS, // inside backslash
87 INCR_M_C0, // inside comment in initial whitespace sequence
88 INCR_M_C1, // inside comment in other places
89 INCR_M_JSON // outside anything, count nesting
90};
91
92#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
93
94typedef struct {
95 U32 flags;
96 U32 max_depth;
97 STRLEN max_size;
98
99 SV *cb_object;
100 HV *cb_sk_object;
101
102 // for the incremental parser
103 SV *incr_text; // the source text so far
104 STRLEN incr_pos; // the current offset into the text
105 int incr_nest; // {[]}-nesting level
106 unsigned char incr_mode;
107} JSON;
108
109INLINE void
110json_init (JSON *json)
111{
112 Zero (json, 1, JSON);
113 json->max_depth = 512;
114}
54 115
55///////////////////////////////////////////////////////////////////////////// 116/////////////////////////////////////////////////////////////////////////////
56// utility functions 117// utility functions
57 118
58static UV * 119INLINE SV *
59SvJSON (SV *sv) 120get_bool (const char *name)
60{ 121{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 122 SV *sv = get_sv (name, 1);
62 croak ("object is not of type JSON::XS");
63 123
64 return &SvUVX (SvRV (sv)); 124 SvREADONLY_on (sv);
65} 125 SvREADONLY_on (SvRV (sv));
66 126
67static void 127 return sv;
128}
129
130INLINE void
68shrink (SV *sv) 131shrink (SV *sv)
69{ 132{
70 sv_utf8_downgrade (sv, 1); 133 sv_utf8_downgrade (sv, 1);
134
71 if (SvLEN (sv) > SvCUR (sv) + 1) 135 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 136 {
73#ifdef SvPV_shrink_to_cur 137#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv); 138 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew) 139#elif defined (SvPV_renew)
… …
81// decode an utf-8 character and return it, or (UV)-1 in 145// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error. 146// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 147// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 148// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 149// note that we never call this function for a ascii codepoints
86inline UV 150INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 151decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 152{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 153 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 154 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 155 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 156 {
93 *clen = 2; 157 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 158 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 159 }
96 else 160 else
97 { 161 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 162}
99 return (UV)-1; 163
164// likewise for encoding, also never called for ascii codepoints
165// this function takes advantage of this fact, although current gccs
166// seem to optimise the check for >= 0x80 away anyways
167INLINE unsigned char *
168encode_utf8 (unsigned char *s, UV ch)
169{
170 if (expect_false (ch < 0x000080))
171 *s++ = ch;
172 else if (expect_true (ch < 0x000800))
173 *s++ = 0xc0 | ( ch >> 6),
174 *s++ = 0x80 | ( ch & 0x3f);
175 else if ( ch < 0x010000)
176 *s++ = 0xe0 | ( ch >> 12),
177 *s++ = 0x80 | ((ch >> 6) & 0x3f),
178 *s++ = 0x80 | ( ch & 0x3f);
179 else if ( ch < 0x110000)
180 *s++ = 0xf0 | ( ch >> 18),
181 *s++ = 0x80 | ((ch >> 12) & 0x3f),
182 *s++ = 0x80 | ((ch >> 6) & 0x3f),
183 *s++ = 0x80 | ( ch & 0x3f);
184
185 return s;
186}
187
188// convert offset pointer to character index, sv must be string
189static STRLEN
190ptr_to_index (SV *sv, char *offset)
191{
192 return SvUTF8 (sv)
193 ? utf8_distance (offset, SvPVX (sv))
194 : offset - SvPVX (sv);
195}
196
197/////////////////////////////////////////////////////////////////////////////
198// fp hell
199
200// scan a group of digits, and a trailing exponent
201static void
202json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
203{
204 UV uaccum = 0;
205 int eaccum = 0;
206
207 // if we recurse too deep, skip all remaining digits
208 // to avoid a stack overflow attack
209 if (expect_false (--maxdepth <= 0))
210 while (((U8)*s - '0') < 10)
211 ++s;
212
213 for (;;)
100 } 214 {
101} 215 U8 dig = (U8)*s - '0';
102 216
217 if (expect_false (dig >= 10))
218 {
219 if (dig == (U8)((U8)'.' - (U8)'0'))
220 {
221 ++s;
222 json_atof_scan1 (s, accum, expo, 1, maxdepth);
223 }
224 else if ((dig | ' ') == 'e' - '0')
225 {
226 int exp2 = 0;
227 int neg = 0;
228
229 ++s;
230
231 if (*s == '-')
232 {
233 ++s;
234 neg = 1;
235 }
236 else if (*s == '+')
237 ++s;
238
239 while ((dig = (U8)*s - '0') < 10)
240 exp2 = exp2 * 10 + *s++ - '0';
241
242 *expo += neg ? -exp2 : exp2;
243 }
244
245 break;
246 }
247
248 ++s;
249
250 uaccum = uaccum * 10 + dig;
251 ++eaccum;
252
253 // if we have too many digits, then recurse for more
254 // we actually do this for rather few digits
255 if (uaccum >= (UV_MAX - 9) / 10)
256 {
257 if (postdp) *expo -= eaccum;
258 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
259 if (postdp) *expo += eaccum;
260
261 break;
262 }
263 }
264
265 // this relies greatly on the quality of the pow ()
266 // implementation of the platform, but a good
267 // implementation is hard to beat.
268 // (IEEE 754 conformant ones are required to be exact)
269 if (postdp) *expo -= eaccum;
270 *accum += uaccum * Perl_pow (10., *expo);
271 *expo += eaccum;
272}
273
274static NV
275json_atof (const char *s)
276{
277 NV accum = 0.;
278 int expo = 0;
279 int neg = 0;
280
281 if (*s == '-')
282 {
283 ++s;
284 neg = 1;
285 }
286
287 // a recursion depth of ten gives us >>500 bits
288 json_atof_scan1 (s, &accum, &expo, 0, 10);
289
290 return neg ? -accum : accum;
291}
103///////////////////////////////////////////////////////////////////////////// 292/////////////////////////////////////////////////////////////////////////////
104// encoder 293// encoder
105 294
106// structure used for encoding JSON 295// structure used for encoding JSON
107typedef struct 296typedef struct
108{ 297{
109 char *cur; // SvPVX (sv) + current output position 298 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 299 char *end; // SvEND (sv)
111 SV *sv; // result scalar 300 SV *sv; // result scalar
112 U32 flags; // F_* 301 JSON json;
113 U32 indent; // indentation level 302 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 303 UV limit; // escape character values >= this value when encoding
115} enc_t; 304} enc_t;
116 305
117inline void 306INLINE void
118need (enc_t *enc, STRLEN len) 307need (enc_t *enc, STRLEN len)
119{ 308{
120 if (expect_false (enc->cur + len >= enc->end)) 309 if (expect_false (enc->cur + len >= enc->end))
121 { 310 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 311 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + len + 1); 312 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 313 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 314 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 315 }
127} 316}
128 317
129inline void 318INLINE void
130encode_ch (enc_t *enc, char ch) 319encode_ch (enc_t *enc, char ch)
131{ 320{
132 need (enc, 1); 321 need (enc, 1);
133 *enc->cur++ = ch; 322 *enc->cur++ = ch;
134} 323}
… …
178 STRLEN clen; 367 STRLEN clen;
179 UV uch; 368 UV uch;
180 369
181 if (is_utf8) 370 if (is_utf8)
182 { 371 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 372 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 373 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 374 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 375 }
188 else 376 else
189 { 377 {
190 uch = ch; 378 uch = ch;
191 clen = 1; 379 clen = 1;
192 } 380 }
193 381
194 if (uch > 0x10FFFFUL) 382 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 { 383 {
199 if (uch > 0xFFFFUL) 384 if (uch >= 0x10000UL)
200 { 385 {
386 if (uch >= 0x110000UL)
387 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
388
201 need (enc, len += 11); 389 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 390 sprintf (enc->cur, "\\u%04x\\u%04x",
203 (int)((uch - 0x10000) / 0x400 + 0xD800), 391 (int)((uch - 0x10000) / 0x400 + 0xD800),
204 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 392 (int)((uch - 0x10000) % 0x400 + 0xDC00));
205 enc->cur += 12; 393 enc->cur += 12;
206 } 394 }
207 else 395 else
208 { 396 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5); 397 need (enc, len += 5);
211 *enc->cur++ = '\\'; 398 *enc->cur++ = '\\';
212 *enc->cur++ = 'u'; 399 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ]; 400 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 401 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 402 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 403 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
217 } 404 }
218 405
219 str += clen; 406 str += clen;
220 } 407 }
221 else if (enc->flags & F_LATIN1) 408 else if (enc->json.flags & F_LATIN1)
222 { 409 {
223 *enc->cur++ = uch; 410 *enc->cur++ = uch;
224 str += clen; 411 str += clen;
225 } 412 }
226 else if (is_utf8) 413 else if (is_utf8)
… …
233 while (--clen); 420 while (--clen);
234 } 421 }
235 else 422 else
236 { 423 {
237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 424 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 425 enc->cur = encode_utf8 (enc->cur, uch);
239 ++str; 426 ++str;
240 } 427 }
241 } 428 }
242 } 429 }
243 } 430 }
244 431
245 --len; 432 --len;
246 } 433 }
247} 434}
248 435
249inline void 436INLINE void
250encode_indent (enc_t *enc) 437encode_indent (enc_t *enc)
251{ 438{
252 if (enc->flags & F_INDENT) 439 if (enc->json.flags & F_INDENT)
253 { 440 {
254 int spaces = enc->indent * INDENT_STEP; 441 int spaces = enc->indent * INDENT_STEP;
255 442
256 need (enc, spaces); 443 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 444 memset (enc->cur, ' ', spaces);
258 enc->cur += spaces; 445 enc->cur += spaces;
259 } 446 }
260} 447}
261 448
262inline void 449INLINE void
263encode_space (enc_t *enc) 450encode_space (enc_t *enc)
264{ 451{
265 need (enc, 1); 452 need (enc, 1);
266 encode_ch (enc, ' '); 453 encode_ch (enc, ' ');
267} 454}
268 455
269inline void 456INLINE void
270encode_nl (enc_t *enc) 457encode_nl (enc_t *enc)
271{ 458{
272 if (enc->flags & F_INDENT) 459 if (enc->json.flags & F_INDENT)
273 { 460 {
274 need (enc, 1); 461 need (enc, 1);
275 encode_ch (enc, '\n'); 462 encode_ch (enc, '\n');
276 } 463 }
277} 464}
278 465
279inline void 466INLINE void
280encode_comma (enc_t *enc) 467encode_comma (enc_t *enc)
281{ 468{
282 encode_ch (enc, ','); 469 encode_ch (enc, ',');
283 470
284 if (enc->flags & F_INDENT) 471 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 472 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 473 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 474 encode_space (enc);
288} 475}
289 476
290static void encode_sv (enc_t *enc, SV *sv); 477static void encode_sv (enc_t *enc, SV *sv);
291 478
292static void 479static void
293encode_av (enc_t *enc, AV *av) 480encode_av (enc_t *enc, AV *av)
294{ 481{
295 int i, len = av_len (av); 482 int i, len = av_len (av);
296 483
297 if (enc->indent >= enc->maxdepth) 484 if (enc->indent >= enc->json.max_depth)
298 croak ("data structure too deep (hit recursion limit)"); 485 croak (ERR_NESTING_EXCEEDED);
299 486
300 encode_ch (enc, '['); encode_nl (enc); 487 encode_ch (enc, '[');
301 ++enc->indent; 488
489 if (len >= 0)
490 {
491 encode_nl (enc); ++enc->indent;
302 492
303 for (i = 0; i <= len; ++i) 493 for (i = 0; i <= len; ++i)
304 { 494 {
495 SV **svp = av_fetch (av, i, 0);
496
305 encode_indent (enc); 497 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0));
307 498
499 if (svp)
500 encode_sv (enc, *svp);
501 else
502 encode_str (enc, "null", 4, 0);
503
308 if (i < len) 504 if (i < len)
309 encode_comma (enc); 505 encode_comma (enc);
310 } 506 }
311 507
508 encode_nl (enc); --enc->indent; encode_indent (enc);
509 }
510
312 encode_nl (enc); 511 encode_ch (enc, ']');
313
314 --enc->indent;
315 encode_indent (enc); encode_ch (enc, ']');
316} 512}
317 513
318static void 514static void
319encode_he (enc_t *enc, HE *he) 515encode_hk (enc_t *enc, HE *he)
320{ 516{
321 encode_ch (enc, '"'); 517 encode_ch (enc, '"');
322 518
323 if (HeKLEN (he) == HEf_SVKEY) 519 if (HeKLEN (he) == HEf_SVKEY)
324 { 520 {
… …
334 else 530 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 531 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 532
337 encode_ch (enc, '"'); 533 encode_ch (enc, '"');
338 534
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 535 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 536 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 537 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he));
343} 538}
344 539
345// compare hash entries, used when all keys are bytestrings 540// compare hash entries, used when all keys are bytestrings
346static int 541static int
347he_cmp_fast (const void *a_, const void *b_) 542he_cmp_fast (const void *a_, const void *b_)
… …
352 HE *b = *(HE **)b_; 547 HE *b = *(HE **)b_;
353 548
354 STRLEN la = HeKLEN (a); 549 STRLEN la = HeKLEN (a);
355 STRLEN lb = HeKLEN (b); 550 STRLEN lb = HeKLEN (b);
356 551
357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 552 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
358 cmp = la - lb; 553 cmp = lb - la;
359 554
360 return cmp; 555 return cmp;
361} 556}
362 557
363// compare hash entries, used when some keys are sv's or utf-x 558// compare hash entries, used when some keys are sv's or utf-x
364static int 559static int
365he_cmp_slow (const void *a, const void *b) 560he_cmp_slow (const void *a, const void *b)
366{ 561{
367 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 562 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
368} 563}
369 564
370static void 565static void
371encode_hv (enc_t *enc, HV *hv) 566encode_hv (enc_t *enc, HV *hv)
372{ 567{
373 int count, i; 568 HE *he;
374 569
375 if (enc->indent >= enc->maxdepth) 570 if (enc->indent >= enc->json.max_depth)
376 croak ("data structure too deep (hit recursion limit)"); 571 croak (ERR_NESTING_EXCEEDED);
377 572
378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 573 encode_ch (enc, '{');
379 574
380 if ((count = hv_iterinit (hv)))
381 {
382 // for canonical output we have to sort by keys first 575 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 576 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 577 // security workaround added somewhere in 5.8.x
385 // that randomises hash orderings 578 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 579 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
580 {
581 int count = hv_iterinit (hv);
582
583 if (SvMAGICAL (hv))
387 { 584 {
585 // need to count by iterating. could improve by dynamically building the vector below
586 // but I don't care for the speed of this special case.
587 // note also that we will run into undefined behaviour when the two iterations
588 // do not result in the same count, something I might care for in some later release.
589
590 count = 0;
591 while (hv_iternext (hv))
592 ++count;
593
594 hv_iterinit (hv);
595 }
596
597 if (count)
598 {
388 int fast = 1; 599 int i, fast = 1;
389 HE *he; 600 HE *hes_stack [STACK_HES];
390#if defined(__BORLANDC__) || defined(_MSC_VER) 601 HE **hes = hes_stack;
391 HE **hes = _alloca (count * sizeof (HE)); 602
392#else 603 // allocate larger arrays on the heap
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 604 if (count > STACK_HES)
394#endif 605 {
606 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
607 hes = (HE **)SvPVX (sv);
608 }
395 609
396 i = 0; 610 i = 0;
397 while ((he = hv_iternext (hv))) 611 while ((he = hv_iternext (hv)))
398 { 612 {
399 hes [i++] = he; 613 hes [i++] = he;
… …
421 635
422 FREETMPS; 636 FREETMPS;
423 LEAVE; 637 LEAVE;
424 } 638 }
425 639
426 for (i = 0; i < count; ++i) 640 encode_nl (enc); ++enc->indent;
641
642 while (count--)
427 { 643 {
428 encode_indent (enc); 644 encode_indent (enc);
645 he = hes [count];
429 encode_he (enc, hes [i]); 646 encode_hk (enc, he);
647 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
430 648
431 if (i < count - 1) 649 if (count)
432 encode_comma (enc); 650 encode_comma (enc);
433 } 651 }
434 652
653 encode_nl (enc); --enc->indent; encode_indent (enc);
654 }
655 }
656 else
657 {
658 if (hv_iterinit (hv) || SvMAGICAL (hv))
659 if ((he = hv_iternext (hv)))
660 {
661 encode_nl (enc); ++enc->indent;
662
663 for (;;)
664 {
435 encode_nl (enc); 665 encode_indent (enc);
666 encode_hk (enc, he);
667 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
668
669 if (!(he = hv_iternext (hv)))
670 break;
671
672 encode_comma (enc);
673 }
674
675 encode_nl (enc); --enc->indent; encode_indent (enc);
676 }
677 }
678
679 encode_ch (enc, '}');
680}
681
682// encode objects, arrays and special \0=false and \1=true values.
683static void
684encode_rv (enc_t *enc, SV *sv)
685{
686 svtype svt;
687
688 SvGETMAGIC (sv);
689 svt = SvTYPE (sv);
690
691 if (expect_false (SvOBJECT (sv)))
692 {
693 HV *stash = !JSON_SLOW || json_boolean_stash
694 ? json_boolean_stash
695 : gv_stashpv ("JSON::XS::Boolean", 1);
696
697 if (SvSTASH (sv) == stash)
698 {
699 if (SvIV (sv))
700 encode_str (enc, "true", 4, 0);
701 else
702 encode_str (enc, "false", 5, 0);
436 } 703 }
437 else 704 else
438 { 705 {
439 HE *he = hv_iternext (hv); 706#if 0
440 707 if (0 && sv_derived_from (rv, "JSON::Literal"))
441 for (;;)
442 {
443 encode_indent (enc);
444 encode_he (enc, he);
445
446 if (!(he = hv_iternext (hv)))
447 break;
448
449 encode_comma (enc);
450 } 708 {
709 // not yet
710 }
711#endif
712 if (enc->json.flags & F_CONV_BLESSED)
713 {
714 // we re-bless the reference to get overload and other niceties right
715 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
451 716
717 if (to_json)
718 {
719 dSP;
720
721 ENTER; SAVETMPS; PUSHMARK (SP);
722 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
723
724 // calling with G_SCALAR ensures that we always get a 1 return value
725 PUTBACK;
726 call_sv ((SV *)GvCV (to_json), G_SCALAR);
727 SPAGAIN;
728
729 // catch this surprisingly common error
730 if (SvROK (TOPs) && SvRV (TOPs) == sv)
731 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
732
733 sv = POPs;
734 PUTBACK;
735
452 encode_nl (enc); 736 encode_sv (enc, sv);
737
738 FREETMPS; LEAVE;
739 }
740 else if (enc->json.flags & F_ALLOW_BLESSED)
741 encode_str (enc, "null", 4, 0);
742 else
743 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
744 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
745 }
746 else if (enc->json.flags & F_ALLOW_BLESSED)
747 encode_str (enc, "null", 4, 0);
748 else
749 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
750 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
453 } 751 }
454 } 752 }
455
456 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
457}
458
459// encode objects, arrays and special \0=false and \1=true values.
460static void
461encode_rv (enc_t *enc, SV *sv)
462{
463 svtype svt;
464
465 SvGETMAGIC (sv);
466 svt = SvTYPE (sv);
467
468 if (svt == SVt_PVHV) 753 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 754 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 755 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 756 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 757 else if (svt < SVt_PVAV)
473 { 758 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 759 STRLEN len = 0;
760 char *pv = svt ? SvPV (sv, len) : 0;
761
762 if (len == 1 && *pv == '1')
763 encode_str (enc, "true", 4, 0);
764 else if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 765 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1) 766 else if (enc->json.flags & F_ALLOW_UNKNOWN)
477 encode_str (enc, "true", 4, 0); 767 encode_str (enc, "null", 4, 0);
478 else 768 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 769 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 770 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 771 }
772 else if (enc->json.flags & F_ALLOW_UNKNOWN)
773 encode_str (enc, "null", 4, 0);
482 else 774 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 775 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 776 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
485} 777}
486 778
… …
504 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 796 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
505 enc->cur += strlen (enc->cur); 797 enc->cur += strlen (enc->cur);
506 } 798 }
507 else if (SvIOKp (sv)) 799 else if (SvIOKp (sv))
508 { 800 {
509 // we assume we can always read an IV as a UV 801 // we assume we can always read an IV as a UV and vice versa
510 if (SvUV (sv) & ~(UV)0x7fff) 802 // we assume two's complement
511 { 803 // we assume no aliasing issues in the union
512 // large integer, use the (rather slow) snprintf way. 804 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
513 need (enc, sizeof (UV) * 3); 805 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
514 enc->cur +=
515 SvIsUV(sv)
516 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
517 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
518 }
519 else
520 { 806 {
521 // optimise the "small number case" 807 // optimise the "small number case"
522 // code will likely be branchless and use only a single multiplication 808 // code will likely be branchless and use only a single multiplication
809 // works for numbers up to 59074
523 I32 i = SvIV (sv); 810 I32 i = SvIVX (sv);
524 U32 u; 811 U32 u;
525 char digit, nz = 0; 812 char digit, nz = 0;
526 813
527 need (enc, 6); 814 need (enc, 6);
528 815
… …
534 821
535 // now output digit by digit, each time masking out the integer part 822 // now output digit by digit, each time masking out the integer part
536 // and multiplying by 5 while moving the decimal point one to the right, 823 // and multiplying by 5 while moving the decimal point one to the right,
537 // resulting in a net multiplication by 10. 824 // resulting in a net multiplication by 10.
538 // we always write the digit to memory but conditionally increment 825 // we always write the digit to memory but conditionally increment
539 // the pointer, to ease the usage of conditional move instructions. 826 // the pointer, to enable the use of conditional move instructions.
540 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 827 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
541 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 828 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
542 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 829 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
543 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 830 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
544 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 831 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
545 } 832 }
833 else
834 {
835 // large integer, use the (rather slow) snprintf way.
836 need (enc, IVUV_MAXCHARS);
837 enc->cur +=
838 SvIsUV(sv)
839 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
840 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
841 }
546 } 842 }
547 else if (SvROK (sv)) 843 else if (SvROK (sv))
548 encode_rv (enc, SvRV (sv)); 844 encode_rv (enc, SvRV (sv));
549 else if (!SvOK (sv)) 845 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
550 encode_str (enc, "null", 4, 0); 846 encode_str (enc, "null", 4, 0);
551 else 847 else
552 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 848 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
553 SvPV_nolen (sv), SvFLAGS (sv)); 849 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
554} 850}
555 851
556static SV * 852static SV *
557encode_json (SV *scalar, U32 flags) 853encode_json (SV *scalar, JSON *json)
558{ 854{
559 enc_t enc; 855 enc_t enc;
560 856
561 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 857 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
562 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 858 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
563 859
564 enc.flags = flags; 860 enc.json = *json;
565 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 861 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
566 enc.cur = SvPVX (enc.sv); 862 enc.cur = SvPVX (enc.sv);
567 enc.end = SvEND (enc.sv); 863 enc.end = SvEND (enc.sv);
568 enc.indent = 0; 864 enc.indent = 0;
569 enc.maxdepth = DEC_DEPTH (flags); 865 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
866 : enc.json.flags & F_LATIN1 ? 0x000100UL
867 : 0x110000UL;
570 868
571 SvPOK_only (enc.sv); 869 SvPOK_only (enc.sv);
572 encode_sv (&enc, scalar); 870 encode_sv (&enc, scalar);
871 encode_nl (&enc);
573 872
574 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 873 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
575 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 874 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
576 875
577 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 876 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
578 SvUTF8_on (enc.sv); 877 SvUTF8_on (enc.sv);
579 878
580 if (enc.flags & F_SHRINK) 879 if (enc.json.flags & F_SHRINK)
581 shrink (enc.sv); 880 shrink (enc.sv);
582 881
583 return enc.sv; 882 return enc.sv;
584} 883}
585 884
… …
590typedef struct 889typedef struct
591{ 890{
592 char *cur; // current parser pointer 891 char *cur; // current parser pointer
593 char *end; // end of input string 892 char *end; // end of input string
594 const char *err; // parse error, if != 0 893 const char *err; // parse error, if != 0
595 U32 flags; // F_* 894 JSON json;
596 U32 depth; // recursion depth 895 U32 depth; // recursion depth
597 U32 maxdepth; // recursion depth limit 896 U32 maxdepth; // recursion depth limit
598} dec_t; 897} dec_t;
599 898
600inline void 899INLINE void
900decode_comment (dec_t *dec)
901{
902 // only '#'-style comments allowed a.t.m.
903
904 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
905 ++dec->cur;
906}
907
908INLINE void
601decode_ws (dec_t *dec) 909decode_ws (dec_t *dec)
602{ 910{
603 for (;;) 911 for (;;)
604 { 912 {
605 char ch = *dec->cur; 913 char ch = *dec->cur;
606 914
607 if (ch > 0x20 915 if (ch > 0x20)
916 {
917 if (expect_false (ch == '#'))
918 {
919 if (dec->json.flags & F_RELAXED)
920 decode_comment (dec);
921 else
922 break;
923 }
924 else
925 break;
926 }
608 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 927 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
609 break; 928 break; // parse error, but let higher level handle it, gives better error messages
610 929
611 ++dec->cur; 930 ++dec->cur;
612 } 931 }
613} 932}
614 933
… …
618 if (*dec->cur != ch) \ 937 if (*dec->cur != ch) \
619 ERR (# ch " expected"); \ 938 ERR (# ch " expected"); \
620 ++dec->cur; \ 939 ++dec->cur; \
621 SE 940 SE
622 941
623#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 942#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
624#define DEC_DEC_DEPTH --dec->depth 943#define DEC_DEC_DEPTH --dec->depth
625 944
626static SV *decode_sv (dec_t *dec); 945static SV *decode_sv (dec_t *dec);
627 946
628static signed char decode_hexdigit[256]; 947static signed char decode_hexdigit[256];
… …
631decode_4hex (dec_t *dec) 950decode_4hex (dec_t *dec)
632{ 951{
633 signed char d1, d2, d3, d4; 952 signed char d1, d2, d3, d4;
634 unsigned char *cur = (unsigned char *)dec->cur; 953 unsigned char *cur = (unsigned char *)dec->cur;
635 954
636 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 955 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
637 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 956 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
638 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 957 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
639 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 958 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
640 959
641 dec->cur += 4; 960 dec->cur += 4;
642 961
643 return ((UV)d1) << 12 962 return ((UV)d1) << 12
644 | ((UV)d2) << 8 963 | ((UV)d2) << 8
… …
720 1039
721 if (hi >= 0x80) 1040 if (hi >= 0x80)
722 { 1041 {
723 utf8 = 1; 1042 utf8 = 1;
724 1043
725 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1044 cur = encode_utf8 (cur, hi);
726 } 1045 }
727 else 1046 else
728 *cur++ = hi; 1047 *cur++ = hi;
729 } 1048 }
730 break; 1049 break;
… …
732 default: 1051 default:
733 --dec_cur; 1052 --dec_cur;
734 ERR ("illegal backslash escape sequence in string"); 1053 ERR ("illegal backslash escape sequence in string");
735 } 1054 }
736 } 1055 }
737 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1056 else if (expect_true (ch >= 0x20 && ch < 0x80))
738 *cur++ = ch; 1057 *cur++ = ch;
739 else if (ch >= 0x80) 1058 else if (ch >= 0x80)
740 { 1059 {
741 STRLEN clen; 1060 STRLEN clen;
742 UV uch;
743 1061
744 --dec_cur; 1062 --dec_cur;
745 1063
746 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1064 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
747 if (clen == (STRLEN)-1) 1065 if (clen == (STRLEN)-1)
748 ERR ("malformed UTF-8 character in JSON string"); 1066 ERR ("malformed UTF-8 character in JSON string");
749 1067
750 do 1068 do
751 *cur++ = *dec_cur++; 1069 *cur++ = *dec_cur++;
… …
768 { 1086 {
769 STRLEN len = cur - buf; 1087 STRLEN len = cur - buf;
770 1088
771 if (sv) 1089 if (sv)
772 { 1090 {
773 SvGROW (sv, SvCUR (sv) + len + 1); 1091 STRLEN cur = SvCUR (sv);
1092
1093 if (SvLEN (sv) <= cur + len)
1094 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1095
774 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1096 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
775 SvCUR_set (sv, SvCUR (sv) + len); 1097 SvCUR_set (sv, SvCUR (sv) + len);
776 } 1098 }
777 else 1099 else
778 sv = newSVpvn (buf, len); 1100 sv = newSVpvn (buf, len);
… …
863 is_nv = 1; 1185 is_nv = 1;
864 } 1186 }
865 1187
866 if (!is_nv) 1188 if (!is_nv)
867 { 1189 {
1190 int len = dec->cur - start;
1191
868 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1192 // special case the rather common 1..5-digit-int case
869 if (*start == '-') 1193 if (*start == '-')
870 switch (dec->cur - start) 1194 switch (len)
871 { 1195 {
872 case 2: return newSViv (-( start [1] - '0' )); 1196 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
873 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1197 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
874 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1198 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
875 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1199 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1200 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
876 } 1201 }
877 else 1202 else
878 switch (dec->cur - start) 1203 switch (len)
879 { 1204 {
880 case 1: return newSViv ( start [0] - '0' ); 1205 case 1: return newSViv ( start [0] - '0' * 1);
881 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1206 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
882 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1207 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
883 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1208 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1209 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
884 } 1210 }
885 1211
886 { 1212 {
887 UV uv; 1213 UV uv;
888 int numtype = grok_number (start, dec->cur - start, &uv); 1214 int numtype = grok_number (start, len, &uv);
889 if (numtype & IS_NUMBER_IN_UV) 1215 if (numtype & IS_NUMBER_IN_UV)
890 if (numtype & IS_NUMBER_NEG) 1216 if (numtype & IS_NUMBER_NEG)
891 { 1217 {
892 if (uv < (UV)IV_MIN) 1218 if (uv < (UV)IV_MIN)
893 return newSViv (-(IV)uv); 1219 return newSViv (-(IV)uv);
894 } 1220 }
895 else 1221 else
896 return newSVuv (uv); 1222 return newSVuv (uv);
897 } 1223 }
898 }
899 1224
1225 len -= *start == '-' ? 1 : 0;
1226
1227 // does not fit into IV or UV, try NV
1228 if (len <= NV_DIG)
1229 // fits into NV without loss of precision
1230 return newSVnv (json_atof (start));
1231
1232 // everything else fails, convert it to a string
1233 return newSVpvn (start, dec->cur - start);
1234 }
1235
1236 // loss of precision here
900 return newSVnv (Atof (start)); 1237 return newSVnv (json_atof (start));
901 1238
902fail: 1239fail:
903 return 0; 1240 return 0;
904} 1241}
905 1242
… …
934 1271
935 if (*dec->cur != ',') 1272 if (*dec->cur != ',')
936 ERR (", or ] expected while parsing array"); 1273 ERR (", or ] expected while parsing array");
937 1274
938 ++dec->cur; 1275 ++dec->cur;
1276
1277 decode_ws (dec);
1278
1279 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1280 {
1281 ++dec->cur;
1282 break;
1283 }
939 } 1284 }
940 1285
941 DEC_DEC_DEPTH; 1286 DEC_DEC_DEPTH;
942 return newRV_noinc ((SV *)av); 1287 return newRV_noinc ((SV *)av);
943 1288
… …
948} 1293}
949 1294
950static SV * 1295static SV *
951decode_hv (dec_t *dec) 1296decode_hv (dec_t *dec)
952{ 1297{
1298 SV *sv;
953 HV *hv = newHV (); 1299 HV *hv = newHV ();
954 1300
955 DEC_INC_DEPTH; 1301 DEC_INC_DEPTH;
956 decode_ws (dec); 1302 decode_ws (dec);
957 1303
958 if (*dec->cur == '}') 1304 if (*dec->cur == '}')
959 ++dec->cur; 1305 ++dec->cur;
960 else 1306 else
961 for (;;) 1307 for (;;)
962 { 1308 {
1309 EXPECT_CH ('"');
1310
1311 // heuristic: assume that
1312 // a) decode_str + hv_store_ent are abysmally slow.
1313 // b) most hash keys are short, simple ascii text.
1314 // => try to "fast-match" such strings to avoid
1315 // the overhead of decode_str + hv_store_ent.
1316 {
963 SV *key, *value; 1317 SV *value;
1318 char *p = dec->cur;
1319 char *e = p + 24; // only try up to 24 bytes
964 1320
965 decode_ws (dec); EXPECT_CH ('"'); 1321 for (;;)
966
967 key = decode_str (dec);
968 if (!key)
969 goto fail;
970
971 decode_ws (dec); EXPECT_CH (':');
972
973 value = decode_sv (dec);
974 if (!value)
975 { 1322 {
1323 // the >= 0x80 is false on most architectures
1324 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1325 {
1326 // slow path, back up and use decode_str
1327 SV *key = decode_str (dec);
1328 if (!key)
1329 goto fail;
1330
1331 decode_ws (dec); EXPECT_CH (':');
1332
1333 decode_ws (dec);
1334 value = decode_sv (dec);
1335 if (!value)
1336 {
1337 SvREFCNT_dec (key);
1338 goto fail;
1339 }
1340
1341 hv_store_ent (hv, key, value, 0);
976 SvREFCNT_dec (key); 1342 SvREFCNT_dec (key);
1343
1344 break;
1345 }
1346 else if (*p == '"')
1347 {
1348 // fast path, got a simple key
1349 char *key = dec->cur;
1350 int len = p - key;
1351 dec->cur = p + 1;
1352
1353 decode_ws (dec); EXPECT_CH (':');
1354
1355 decode_ws (dec);
1356 value = decode_sv (dec);
1357 if (!value)
977 goto fail; 1358 goto fail;
1359
1360 hv_store (hv, key, len, value, 0);
1361
1362 break;
1363 }
1364
1365 ++p;
978 } 1366 }
979 1367 }
980 hv_store_ent (hv, key, value, 0);
981 SvREFCNT_dec (key);
982 1368
983 decode_ws (dec); 1369 decode_ws (dec);
984 1370
985 if (*dec->cur == '}') 1371 if (*dec->cur == '}')
986 { 1372 {
… …
990 1376
991 if (*dec->cur != ',') 1377 if (*dec->cur != ',')
992 ERR (", or } expected while parsing object/hash"); 1378 ERR (", or } expected while parsing object/hash");
993 1379
994 ++dec->cur; 1380 ++dec->cur;
1381
1382 decode_ws (dec);
1383
1384 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1385 {
1386 ++dec->cur;
1387 break;
1388 }
995 } 1389 }
996 1390
997 DEC_DEC_DEPTH; 1391 DEC_DEC_DEPTH;
998 return newRV_noinc ((SV *)hv); 1392 sv = newRV_noinc ((SV *)hv);
1393
1394 // check filter callbacks
1395 if (dec->json.flags & F_HOOK)
1396 {
1397 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1398 {
1399 HE *cb, *he;
1400
1401 hv_iterinit (hv);
1402 he = hv_iternext (hv);
1403 hv_iterinit (hv);
1404
1405 // the next line creates a mortal sv each time its called.
1406 // might want to optimise this for common cases.
1407 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1408
1409 if (cb)
1410 {
1411 dSP;
1412 int count;
1413
1414 ENTER; SAVETMPS; PUSHMARK (SP);
1415 XPUSHs (HeVAL (he));
1416 sv_2mortal (sv);
1417
1418 PUTBACK; count = call_sv (HeVAL (cb), 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 if (dec->json.cb_object)
1433 {
1434 dSP;
1435 int count;
1436
1437 ENTER; SAVETMPS; PUSHMARK (SP);
1438 XPUSHs (sv_2mortal (sv));
1439
1440 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1441
1442 if (count == 1)
1443 {
1444 sv = newSVsv (POPs);
1445 FREETMPS; LEAVE;
1446 return sv;
1447 }
1448
1449 SvREFCNT_inc (sv);
1450 FREETMPS; LEAVE;
1451 }
1452 }
1453
1454 return sv;
999 1455
1000fail: 1456fail:
1001 SvREFCNT_dec (hv); 1457 SvREFCNT_dec (hv);
1002 DEC_DEC_DEPTH; 1458 DEC_DEC_DEPTH;
1003 return 0; 1459 return 0;
1004} 1460}
1005 1461
1006static SV * 1462static SV *
1007decode_sv (dec_t *dec) 1463decode_sv (dec_t *dec)
1008{ 1464{
1009 decode_ws (dec); 1465 // the beauty of JSON: you need exactly one character lookahead
1466 // to parse everything.
1010 switch (*dec->cur) 1467 switch (*dec->cur)
1011 { 1468 {
1012 case '"': ++dec->cur; return decode_str (dec); 1469 case '"': ++dec->cur; return decode_str (dec);
1013 case '[': ++dec->cur; return decode_av (dec); 1470 case '[': ++dec->cur; return decode_av (dec);
1014 case '{': ++dec->cur; return decode_hv (dec); 1471 case '{': ++dec->cur; return decode_hv (dec);
1015 1472
1016 case '-': 1473 case '-':
1017 case '0': case '1': case '2': case '3': case '4': 1474 case '0': case '1': case '2': case '3': case '4':
1018 case '5': case '6': case '7': case '8': case '9': 1475 case '5': case '6': case '7': case '8': case '9':
1019 return decode_num (dec); 1476 return decode_num (dec);
1020 1477
1021 case 't': 1478 case 't':
1022 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1479 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1023 { 1480 {
1024 dec->cur += 4; 1481 dec->cur += 4;
1482#if JSON_SLOW
1483 json_true = get_bool ("JSON::XS::true");
1484#endif
1025 return newSViv (1); 1485 return newSVsv (json_true);
1026 } 1486 }
1027 else 1487 else
1028 ERR ("'true' expected"); 1488 ERR ("'true' expected");
1029 1489
1030 break; 1490 break;
1031 1491
1032 case 'f': 1492 case 'f':
1033 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1493 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1034 { 1494 {
1035 dec->cur += 5; 1495 dec->cur += 5;
1496#if JSON_SLOW
1497 json_false = get_bool ("JSON::XS::false");
1498#endif
1036 return newSViv (0); 1499 return newSVsv (json_false);
1037 } 1500 }
1038 else 1501 else
1039 ERR ("'false' expected"); 1502 ERR ("'false' expected");
1040 1503
1041 break; 1504 break;
… …
1059fail: 1522fail:
1060 return 0; 1523 return 0;
1061} 1524}
1062 1525
1063static SV * 1526static SV *
1064decode_json (SV *string, U32 flags, UV *offset_return) 1527decode_json (SV *string, JSON *json, char **offset_return)
1065{ 1528{
1066 dec_t dec; 1529 dec_t dec;
1067 UV offset;
1068 SV *sv; 1530 SV *sv;
1069 1531
1532 /* work around bugs in 5.10 where manipulating magic values
1533 * makes perl ignore the magic in subsequent accesses.
1534 * also make a copy of non-PV values, to get them into a clean
1535 * state (SvPV should do that, but it's buggy, see below).
1536 */
1070 SvGETMAGIC (string); 1537 /*SvGETMAGIC (string);*/
1538 if (SvMAGICAL (string) || !SvPOK (string))
1539 string = sv_2mortal (newSVsv (string));
1540
1071 SvUPGRADE (string, SVt_PV); 1541 SvUPGRADE (string, SVt_PV);
1072 1542
1073 if (flags & F_UTF8) 1543 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1544 * assertion with -DDEBUGGING, although SvCUR is documented to
1545 * return the xpv_cur field which certainly exists after upgrading.
1546 * according to nicholas clark, calling SvPOK fixes this.
1547 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1548 * and hope for the best.
1549 * Damnit, SvPV_nolen still trips over yet another assertion. This
1550 * assertion business is seriously broken, try yet another workaround
1551 * for the broken -DDEBUGGING.
1552 */
1553 {
1554#ifdef DEBUGGING
1555 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1556#else
1557 STRLEN offset = SvCUR (string);
1558#endif
1559
1560 if (offset > json->max_size && json->max_size)
1561 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1562 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1563 }
1564
1565 if (DECODE_WANTS_OCTETS (json))
1074 sv_utf8_downgrade (string, 0); 1566 sv_utf8_downgrade (string, 0);
1075 else 1567 else
1076 sv_utf8_upgrade (string); 1568 sv_utf8_upgrade (string);
1077 1569
1078 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1570 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1079 1571
1080 dec.flags = flags; 1572 dec.json = *json;
1081 dec.cur = SvPVX (string); 1573 dec.cur = SvPVX (string);
1082 dec.end = SvEND (string); 1574 dec.end = SvEND (string);
1083 dec.err = 0; 1575 dec.err = 0;
1084 dec.depth = 0; 1576 dec.depth = 0;
1085 dec.maxdepth = DEC_DEPTH (dec.flags); 1577
1578 if (dec.json.cb_object || dec.json.cb_sk_object)
1579 dec.json.flags |= F_HOOK;
1086 1580
1087 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1581 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1582
1583 decode_ws (&dec);
1088 sv = decode_sv (&dec); 1584 sv = decode_sv (&dec);
1585
1586 if (offset_return)
1587 *offset_return = dec.cur;
1089 1588
1090 if (!(offset_return || !sv)) 1589 if (!(offset_return || !sv))
1091 { 1590 {
1092 // check for trailing garbage 1591 // check for trailing garbage
1093 decode_ws (&dec); 1592 decode_ws (&dec);
… …
1096 { 1595 {
1097 dec.err = "garbage after JSON object"; 1596 dec.err = "garbage after JSON object";
1098 SvREFCNT_dec (sv); 1597 SvREFCNT_dec (sv);
1099 sv = 0; 1598 sv = 0;
1100 } 1599 }
1101 }
1102
1103 if (offset_return || !sv)
1104 {
1105 offset = dec.flags & F_UTF8
1106 ? dec.cur - SvPVX (string)
1107 : utf8_distance (dec.cur, SvPVX (string));
1108
1109 if (offset_return)
1110 *offset_return = offset;
1111 } 1600 }
1112 1601
1113 if (!sv) 1602 if (!sv)
1114 { 1603 {
1115 SV *uni = sv_newmortal (); 1604 SV *uni = sv_newmortal ();
… …
1121 SAVEVPTR (PL_curcop); 1610 SAVEVPTR (PL_curcop);
1122 PL_curcop = &cop; 1611 PL_curcop = &cop;
1123 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1612 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1124 LEAVE; 1613 LEAVE;
1125 1614
1126 croak ("%s, at character offset %d [\"%s\"]", 1615 croak ("%s, at character offset %d (before \"%s\")",
1127 dec.err, 1616 dec.err,
1128 (int)offset, 1617 (int)ptr_to_index (string, dec.cur),
1129 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1618 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1130 } 1619 }
1131 1620
1132 sv = sv_2mortal (sv); 1621 sv = sv_2mortal (sv);
1133 1622
1134 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1623 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1135 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1624 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1136 1625
1137 return sv; 1626 return sv;
1627}
1628
1629/////////////////////////////////////////////////////////////////////////////
1630// incremental parser
1631
1632static void
1633incr_parse (JSON *self)
1634{
1635 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1636
1637 // the state machine here is a bit convoluted and could be simplified a lot
1638 // but this would make it slower, so...
1639
1640 for (;;)
1641 {
1642 //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
1643 switch (self->incr_mode)
1644 {
1645 // only used for initial whitespace skipping
1646 case INCR_M_WS:
1647 for (;;)
1648 {
1649 if (*p > 0x20)
1650 {
1651 if (*p == '#')
1652 {
1653 self->incr_mode = INCR_M_C0;
1654 goto incr_m_c;
1655 }
1656 else
1657 {
1658 self->incr_mode = INCR_M_JSON;
1659 goto incr_m_json;
1660 }
1661 }
1662 else if (!*p)
1663 goto interrupt;
1664
1665 ++p;
1666 }
1667
1668 // skip a single char inside a string (for \\-processing)
1669 case INCR_M_BS:
1670 if (!*p)
1671 goto interrupt;
1672
1673 ++p;
1674 self->incr_mode = INCR_M_STR;
1675 goto incr_m_str;
1676
1677 // inside #-style comments
1678 case INCR_M_C0:
1679 case INCR_M_C1:
1680 incr_m_c:
1681 for (;;)
1682 {
1683 if (*p == '\n')
1684 {
1685 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1686 break;
1687 }
1688 else if (!*p)
1689 goto interrupt;
1690
1691 ++p;
1692 }
1693
1694 break;
1695
1696 // inside a string
1697 case INCR_M_STR:
1698 incr_m_str:
1699 for (;;)
1700 {
1701 if (*p == '"')
1702 {
1703 ++p;
1704 self->incr_mode = INCR_M_JSON;
1705
1706 if (!self->incr_nest)
1707 goto interrupt;
1708
1709 goto incr_m_json;
1710 }
1711 else if (*p == '\\')
1712 {
1713 ++p; // "virtually" consumes character after \
1714
1715 if (!*p) // if at end of string we have to switch modes
1716 {
1717 self->incr_mode = INCR_M_BS;
1718 goto interrupt;
1719 }
1720 }
1721 else if (!*p)
1722 goto interrupt;
1723
1724 ++p;
1725 }
1726
1727 // after initial ws, outside string
1728 case INCR_M_JSON:
1729 incr_m_json:
1730 for (;;)
1731 {
1732 switch (*p++)
1733 {
1734 case 0:
1735 --p;
1736 goto interrupt;
1737
1738 case 0x09:
1739 case 0x0a:
1740 case 0x0d:
1741 case 0x20:
1742 if (!self->incr_nest)
1743 {
1744 --p; // do not eat the whitespace, let the next round do it
1745 goto interrupt;
1746 }
1747 break;
1748
1749 case '"':
1750 self->incr_mode = INCR_M_STR;
1751 goto incr_m_str;
1752
1753 case '[':
1754 case '{':
1755 if (++self->incr_nest > self->max_depth)
1756 croak (ERR_NESTING_EXCEEDED);
1757 break;
1758
1759 case ']':
1760 case '}':
1761 if (--self->incr_nest <= 0)
1762 goto interrupt;
1763 break;
1764
1765 case '#':
1766 self->incr_mode = INCR_M_C1;
1767 goto incr_m_c;
1768 }
1769 }
1770 }
1771
1772 modechange:
1773 ;
1774 }
1775
1776interrupt:
1777 self->incr_pos = p - SvPVX (self->incr_text);
1778 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1779 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1138} 1780}
1139 1781
1140///////////////////////////////////////////////////////////////////////////// 1782/////////////////////////////////////////////////////////////////////////////
1141// XS interface functions 1783// XS interface functions
1142 1784
1143MODULE = JSON::XS PACKAGE = JSON::XS 1785MODULE = JSON::XS PACKAGE = JSON::XS
1144 1786
1145BOOT: 1787BOOT:
1146{ 1788{
1147 int i; 1789 int i;
1148
1149 memset (decode_hexdigit, 0xff, 256);
1150 1790
1151 for (i = 0; i < 256; ++i) 1791 for (i = 0; i < 256; ++i)
1152 decode_hexdigit [i] = 1792 decode_hexdigit [i] =
1153 i >= '0' && i <= '9' ? i - '0' 1793 i >= '0' && i <= '9' ? i - '0'
1154 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1794 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1155 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1795 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1156 : -1; 1796 : -1;
1157 1797
1158 json_stash = gv_stashpv ("JSON::XS", 1); 1798 json_stash = gv_stashpv ("JSON::XS" , 1);
1799 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1800
1801 json_true = get_bool ("JSON::XS::true");
1802 json_false = get_bool ("JSON::XS::false");
1803
1804 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1159} 1805}
1160 1806
1161PROTOTYPES: DISABLE 1807PROTOTYPES: DISABLE
1162 1808
1163SV *new (char *dummy) 1809void CLONE (...)
1164 CODE: 1810 CODE:
1165 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1811 json_stash = 0;
1812 json_boolean_stash = 0;
1813
1814void new (char *klass)
1815 PPCODE:
1816{
1817 SV *pv = NEWSV (0, sizeof (JSON));
1818 SvPOK_only (pv);
1819 json_init ((JSON *)SvPVX (pv));
1820 XPUSHs (sv_2mortal (sv_bless (
1821 newRV_noinc (pv),
1822 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1823 )));
1824}
1825
1826void ascii (JSON *self, int enable = 1)
1827 ALIAS:
1828 ascii = F_ASCII
1829 latin1 = F_LATIN1
1830 utf8 = F_UTF8
1831 indent = F_INDENT
1832 canonical = F_CANONICAL
1833 space_before = F_SPACE_BEFORE
1834 space_after = F_SPACE_AFTER
1835 pretty = F_PRETTY
1836 allow_nonref = F_ALLOW_NONREF
1837 shrink = F_SHRINK
1838 allow_blessed = F_ALLOW_BLESSED
1839 convert_blessed = F_CONV_BLESSED
1840 relaxed = F_RELAXED
1841 allow_unknown = F_ALLOW_UNKNOWN
1842 PPCODE:
1843{
1844 if (enable)
1845 self->flags |= ix;
1846 else
1847 self->flags &= ~ix;
1848
1849 XPUSHs (ST (0));
1850}
1851
1852void get_ascii (JSON *self)
1853 ALIAS:
1854 get_ascii = F_ASCII
1855 get_latin1 = F_LATIN1
1856 get_utf8 = F_UTF8
1857 get_indent = F_INDENT
1858 get_canonical = F_CANONICAL
1859 get_space_before = F_SPACE_BEFORE
1860 get_space_after = F_SPACE_AFTER
1861 get_allow_nonref = F_ALLOW_NONREF
1862 get_shrink = F_SHRINK
1863 get_allow_blessed = F_ALLOW_BLESSED
1864 get_convert_blessed = F_CONV_BLESSED
1865 get_relaxed = F_RELAXED
1866 get_allow_unknown = F_ALLOW_UNKNOWN
1867 PPCODE:
1868 XPUSHs (boolSV (self->flags & ix));
1869
1870void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1871 PPCODE:
1872 self->max_depth = max_depth;
1873 XPUSHs (ST (0));
1874
1875U32 get_max_depth (JSON *self)
1876 CODE:
1877 RETVAL = self->max_depth;
1166 OUTPUT: 1878 OUTPUT:
1167 RETVAL 1879 RETVAL
1168 1880
1169SV *ascii (SV *self, int enable = 1) 1881void max_size (JSON *self, U32 max_size = 0)
1170 ALIAS: 1882 PPCODE:
1171 ascii = F_ASCII 1883 self->max_size = max_size;
1172 latin1 = F_LATIN1 1884 XPUSHs (ST (0));
1173 utf8 = F_UTF8 1885
1174 indent = F_INDENT 1886int get_max_size (JSON *self)
1175 canonical = F_CANONICAL
1176 space_before = F_SPACE_BEFORE
1177 space_after = F_SPACE_AFTER
1178 pretty = F_PRETTY
1179 allow_nonref = F_ALLOW_NONREF
1180 shrink = F_SHRINK
1181 CODE: 1887 CODE:
1182{ 1888 RETVAL = self->max_size;
1183 UV *uv = SvJSON (self);
1184 if (enable)
1185 *uv |= ix;
1186 else
1187 *uv &= ~ix;
1188
1189 RETVAL = newSVsv (self);
1190}
1191 OUTPUT: 1889 OUTPUT:
1192 RETVAL 1890 RETVAL
1193 1891
1194SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1892void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1893 PPCODE:
1894{
1895 SvREFCNT_dec (self->cb_object);
1896 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1897
1898 XPUSHs (ST (0));
1899}
1900
1901void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1902 PPCODE:
1903{
1904 if (!self->cb_sk_object)
1905 self->cb_sk_object = newHV ();
1906
1907 if (SvOK (cb))
1908 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1909 else
1910 {
1911 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1912
1913 if (!HvKEYS (self->cb_sk_object))
1914 {
1915 SvREFCNT_dec (self->cb_sk_object);
1916 self->cb_sk_object = 0;
1917 }
1918 }
1919
1920 XPUSHs (ST (0));
1921}
1922
1923void encode (JSON *self, SV *scalar)
1924 PPCODE:
1925 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
1926 XPUSHs (scalar);
1927
1928void decode (JSON *self, SV *jsonstr)
1929 PPCODE:
1930 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
1931 XPUSHs (jsonstr);
1932
1933void decode_prefix (JSON *self, SV *jsonstr)
1934 PPCODE:
1935{
1936 SV *sv;
1937 char *offset;
1938 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1939 EXTEND (SP, 2);
1940 PUSHs (sv);
1941 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1942}
1943
1944void incr_parse (JSON *self, SV *jsonstr = 0)
1945 PPCODE:
1946{
1947 if (!self->incr_text)
1948 self->incr_text = newSVpvn ("", 0);
1949
1950 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1951 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1952 if (DECODE_WANTS_OCTETS (self))
1953 {
1954 if (self->incr_pos)
1955 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1956 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1957
1958 sv_utf8_downgrade (self->incr_text, 0);
1959 }
1960 else
1961 {
1962 sv_utf8_upgrade (self->incr_text);
1963
1964 if (self->incr_pos)
1965 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1966 - (U8 *)SvPVX (self->incr_text);
1967 }
1968
1969 // append data, if any
1970 if (jsonstr)
1971 {
1972 /* make sure both strings have same encoding */
1973 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1974 if (SvUTF8 (jsonstr))
1975 sv_utf8_downgrade (jsonstr, 0);
1976 else
1977 sv_utf8_upgrade (jsonstr);
1978
1979 /* and then just blindly append */
1980 {
1981 STRLEN len;
1982 const char *str = SvPV (jsonstr, len);
1983 STRLEN cur = SvCUR (self->incr_text);
1984
1985 if (SvLEN (self->incr_text) <= cur + len)
1986 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1987
1988 Move (str, SvEND (self->incr_text), len, char);
1989 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1990 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1991 }
1992 }
1993
1994 if (GIMME_V != G_VOID)
1995 do
1996 {
1997 SV *sv;
1998 char *offset;
1999
2000 if (!INCR_DONE (self))
2001 {
2002 incr_parse (self);
2003
2004 if (self->incr_pos > self->max_size && self->max_size)
2005 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2006 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2007
2008 if (!INCR_DONE (self))
2009 {
2010 // as an optimisation, do not accumulate white space in the incr buffer
2011 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2012 {
2013 self->incr_pos = 0;
2014 SvCUR_set (self->incr_text, 0);
2015 }
2016
2017 break;
2018 }
2019 }
2020
2021 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2022 XPUSHs (sv);
2023
2024 self->incr_pos -= offset - SvPVX (self->incr_text);
2025 self->incr_nest = 0;
2026 self->incr_mode = 0;
2027
2028 sv_chop (self->incr_text, offset);
2029 }
2030 while (GIMME_V == G_ARRAY);
2031}
2032
2033SV *incr_text (JSON *self)
2034 ATTRS: lvalue
1195 CODE: 2035 CODE:
1196{ 2036{
1197 UV *uv = SvJSON (self); 2037 if (self->incr_pos)
1198 UV log2 = 0; 2038 croak ("incr_text can not be called when the incremental parser already started parsing");
1199 2039
1200 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2040 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1201
1202 while ((1UL << log2) < max_depth)
1203 ++log2;
1204
1205 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1206
1207 RETVAL = newSVsv (self);
1208} 2041}
1209 OUTPUT: 2042 OUTPUT:
1210 RETVAL 2043 RETVAL
1211 2044
2045void incr_skip (JSON *self)
2046 CODE:
2047{
2048 if (self->incr_pos)
2049 {
2050 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2051 self->incr_pos = 0;
2052 self->incr_nest = 0;
2053 self->incr_mode = 0;
2054 }
2055}
2056
2057void incr_reset (JSON *self)
2058 CODE:
2059{
2060 SvREFCNT_dec (self->incr_text);
2061 self->incr_text = 0;
2062 self->incr_pos = 0;
2063 self->incr_nest = 0;
2064 self->incr_mode = 0;
2065}
2066
2067void DESTROY (JSON *self)
2068 CODE:
2069 SvREFCNT_dec (self->cb_sk_object);
2070 SvREFCNT_dec (self->cb_object);
2071 SvREFCNT_dec (self->incr_text);
2072
2073PROTOTYPES: ENABLE
2074
1212void encode (SV *self, SV *scalar) 2075void encode_json (SV *scalar)
2076 ALIAS:
2077 to_json_ = 0
2078 encode_json = F_UTF8
1213 PPCODE: 2079 PPCODE:
1214 XPUSHs (encode_json (scalar, *SvJSON (self))); 2080{
2081 JSON json;
2082 json_init (&json);
2083 json.flags |= ix;
2084 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2085 XPUSHs (scalar);
2086}
1215 2087
1216void decode (SV *self, SV *jsonstr) 2088void decode_json (SV *jsonstr)
2089 ALIAS:
2090 from_json_ = 0
2091 decode_json = F_UTF8
1217 PPCODE: 2092 PPCODE:
1218 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1219
1220void decode_prefix (SV *self, SV *jsonstr)
1221 PPCODE:
1222{ 2093{
1223 UV offset; 2094 JSON json;
1224 EXTEND (SP, 2); 2095 json_init (&json);
1225 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2096 json.flags |= ix;
1226 PUSHs (sv_2mortal (newSVuv (offset))); 2097 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2098 XPUSHs (jsonstr);
1227} 2099}
1228 2100
1229PROTOTYPES: ENABLE
1230
1231void to_json (SV *scalar)
1232 ALIAS:
1233 objToJson = 0
1234 PPCODE:
1235 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1236
1237void from_json (SV *jsonstr)
1238 ALIAS:
1239 jsonToObj = 0
1240 PPCODE:
1241 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1242

Diff Legend

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