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 1.55 by root, Mon Jul 23 22:57:40 2007 UTC vs.
Revision 1.118 by root, Fri Oct 25 20:27:57 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
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
19 24
20#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
26#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 36#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 37#define F_ALLOW_UNKNOWN 0x00002000UL
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36 39
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
39
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41#define F_DEFAULT (9UL << S_MAXDEPTH)
42 41
43#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
45 44
46#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
47 46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
48
48#define SB do { 49#define SB do {
49#define SE } while (0) 50#define SE } while (0)
50 51
51#if __GNUC__ >= 3 52#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value)) 53# define expect(expr,value) __builtin_expect ((expr), (value))
53# define inline inline 54# define INLINE static inline
54#else 55#else
55# define expect(expr,value) (expr) 56# define expect(expr,value) (expr)
56# define inline static 57# define INLINE static
57#endif 58#endif
58 59
59#define expect_false(expr) expect ((expr) != 0, 0) 60#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1) 61#define expect_true(expr) expect ((expr) != 0, 1)
61 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
62static HV *json_stash, *json_boolean_stash; // JSON::XS:: 80static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false; 81static SV *json_true, *json_false;
64 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
65typedef struct { 94typedef struct {
66 U32 flags; 95 U32 flags;
96 U32 max_depth;
97 STRLEN max_size;
98
67 SV *cb_object; 99 SV *cb_object;
68 HV *cb_sk_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;
69} JSON; 107} JSON;
108
109INLINE void
110json_init (JSON *json)
111{
112 Zero (json, 1, JSON);
113 json->max_depth = 512;
114}
70 115
71///////////////////////////////////////////////////////////////////////////// 116/////////////////////////////////////////////////////////////////////////////
72// utility functions 117// utility functions
73 118
74inline void 119INLINE SV *
120get_bool (const char *name)
121{
122 SV *sv = get_sv (name, 1);
123
124 SvREADONLY_on (sv);
125 SvREADONLY_on (SvRV (sv));
126
127 return sv;
128}
129
130INLINE void
75shrink (SV *sv) 131shrink (SV *sv)
76{ 132{
77 sv_utf8_downgrade (sv, 1); 133 sv_utf8_downgrade (sv, 1);
134
78 if (SvLEN (sv) > SvCUR (sv) + 1) 135 if (SvLEN (sv) > SvCUR (sv) + 1)
79 { 136 {
80#ifdef SvPV_shrink_to_cur 137#ifdef SvPV_shrink_to_cur
81 SvPV_shrink_to_cur (sv); 138 SvPV_shrink_to_cur (sv);
82#elif defined (SvPV_renew) 139#elif defined (SvPV_renew)
… …
88// 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
89// case of an error. 146// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF, 147// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else. 148// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints 149// note that we never call this function for a ascii codepoints
93inline UV 150INLINE UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 151decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{ 152{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 153 if (expect_true (len >= 2
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 154 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 155 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
99 { 156 {
100 *clen = 2; 157 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 158 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 } 159 }
103 else 160 else
104 { 161 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
105 *clen = (STRLEN)-1; 162}
106 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 (;;)
107 } 214 {
108} 215 U8 dig = (U8)*s - '0';
109 216
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}
110///////////////////////////////////////////////////////////////////////////// 292/////////////////////////////////////////////////////////////////////////////
111// encoder 293// encoder
112 294
113// structure used for encoding JSON 295// structure used for encoding JSON
114typedef struct 296typedef struct
… …
116 char *cur; // SvPVX (sv) + current output position 298 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 299 char *end; // SvEND (sv)
118 SV *sv; // result scalar 300 SV *sv; // result scalar
119 JSON json; 301 JSON json;
120 U32 indent; // indentation level 302 U32 indent; // indentation level
121 U32 maxdepth; // max. indentation/recursion level 303 UV limit; // escape character values >= this value when encoding
122} enc_t; 304} enc_t;
123 305
124inline void 306INLINE void
125need (enc_t *enc, STRLEN len) 307need (enc_t *enc, STRLEN len)
126{ 308{
127 if (expect_false (enc->cur + len >= enc->end)) 309 if (expect_false (enc->cur + len >= enc->end))
128 { 310 {
129 STRLEN cur = enc->cur - SvPVX (enc->sv); 311 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
130 SvGROW (enc->sv, cur + len + 1); 312 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
131 enc->cur = SvPVX (enc->sv) + cur; 313 enc->cur = SvPVX (enc->sv) + cur;
132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 314 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
133 } 315 }
134} 316}
135 317
136inline void 318INLINE void
137encode_ch (enc_t *enc, char ch) 319encode_ch (enc_t *enc, char ch)
138{ 320{
139 need (enc, 1); 321 need (enc, 1);
140 *enc->cur++ = ch; 322 *enc->cur++ = ch;
141} 323}
… …
195 { 377 {
196 uch = ch; 378 uch = ch;
197 clen = 1; 379 clen = 1;
198 } 380 }
199 381
200 if (uch > 0x10FFFFUL) 382 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
202
203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
204 { 383 {
205 if (uch > 0xFFFFUL) 384 if (uch >= 0x10000UL)
206 { 385 {
386 if (uch >= 0x110000UL)
387 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
388
207 need (enc, len += 11); 389 need (enc, len += 11);
208 sprintf (enc->cur, "\\u%04x\\u%04x", 390 sprintf (enc->cur, "\\u%04x\\u%04x",
209 (int)((uch - 0x10000) / 0x400 + 0xD800), 391 (int)((uch - 0x10000) / 0x400 + 0xD800),
210 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 392 (int)((uch - 0x10000) % 0x400 + 0xDC00));
211 enc->cur += 12; 393 enc->cur += 12;
212 } 394 }
213 else 395 else
214 { 396 {
215 static char hexdigit [16] = "0123456789abcdef";
216 need (enc, len += 5); 397 need (enc, len += 5);
217 *enc->cur++ = '\\'; 398 *enc->cur++ = '\\';
218 *enc->cur++ = 'u'; 399 *enc->cur++ = 'u';
219 *enc->cur++ = hexdigit [ uch >> 12 ]; 400 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
220 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 401 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
221 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 402 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
222 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 403 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
223 } 404 }
224 405
225 str += clen; 406 str += clen;
226 } 407 }
227 else if (enc->json.flags & F_LATIN1) 408 else if (enc->json.flags & F_LATIN1)
… …
239 while (--clen); 420 while (--clen);
240 } 421 }
241 else 422 else
242 { 423 {
243 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
244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 425 enc->cur = encode_utf8 (enc->cur, uch);
245 ++str; 426 ++str;
246 } 427 }
247 } 428 }
248 } 429 }
249 } 430 }
250 431
251 --len; 432 --len;
252 } 433 }
253} 434}
254 435
255inline void 436INLINE void
256encode_indent (enc_t *enc) 437encode_indent (enc_t *enc)
257{ 438{
258 if (enc->json.flags & F_INDENT) 439 if (enc->json.flags & F_INDENT)
259 { 440 {
260 int spaces = enc->indent * INDENT_STEP; 441 int spaces = enc->indent * INDENT_STEP;
… …
263 memset (enc->cur, ' ', spaces); 444 memset (enc->cur, ' ', spaces);
264 enc->cur += spaces; 445 enc->cur += spaces;
265 } 446 }
266} 447}
267 448
268inline void 449INLINE void
269encode_space (enc_t *enc) 450encode_space (enc_t *enc)
270{ 451{
271 need (enc, 1); 452 need (enc, 1);
272 encode_ch (enc, ' '); 453 encode_ch (enc, ' ');
273} 454}
274 455
275inline void 456INLINE void
276encode_nl (enc_t *enc) 457encode_nl (enc_t *enc)
277{ 458{
278 if (enc->json.flags & F_INDENT) 459 if (enc->json.flags & F_INDENT)
279 { 460 {
280 need (enc, 1); 461 need (enc, 1);
281 encode_ch (enc, '\n'); 462 encode_ch (enc, '\n');
282 } 463 }
283} 464}
284 465
285inline void 466INLINE void
286encode_comma (enc_t *enc) 467encode_comma (enc_t *enc)
287{ 468{
288 encode_ch (enc, ','); 469 encode_ch (enc, ',');
289 470
290 if (enc->json.flags & F_INDENT) 471 if (enc->json.flags & F_INDENT)
… …
298static void 479static void
299encode_av (enc_t *enc, AV *av) 480encode_av (enc_t *enc, AV *av)
300{ 481{
301 int i, len = av_len (av); 482 int i, len = av_len (av);
302 483
303 if (enc->indent >= enc->maxdepth) 484 if (enc->indent >= enc->json.max_depth)
304 croak ("data structure too deep (hit recursion limit)"); 485 croak (ERR_NESTING_EXCEEDED);
305 486
306 encode_ch (enc, '['); encode_nl (enc); 487 encode_ch (enc, '[');
307 ++enc->indent;
308 488
489 if (len >= 0)
490 {
491 encode_nl (enc); ++enc->indent;
492
309 for (i = 0; i <= len; ++i) 493 for (i = 0; i <= len; ++i)
310 { 494 {
311 SV **svp = av_fetch (av, i, 0); 495 SV **svp = av_fetch (av, i, 0);
312 496
313 encode_indent (enc); 497 encode_indent (enc);
314 498
315 if (svp) 499 if (svp)
316 encode_sv (enc, *svp); 500 encode_sv (enc, *svp);
317 else 501 else
318 encode_str (enc, "null", 4, 0); 502 encode_str (enc, "null", 4, 0);
319 503
320 if (i < len) 504 if (i < len)
321 encode_comma (enc); 505 encode_comma (enc);
322 } 506 }
323 507
508 encode_nl (enc); --enc->indent; encode_indent (enc);
509 }
510
324 encode_nl (enc); 511 encode_ch (enc, ']');
325
326 --enc->indent;
327 encode_indent (enc); encode_ch (enc, ']');
328} 512}
329 513
330static void 514static void
331encode_he (enc_t *enc, HE *he) 515encode_hk (enc_t *enc, HE *he)
332{ 516{
333 encode_ch (enc, '"'); 517 encode_ch (enc, '"');
334 518
335 if (HeKLEN (he) == HEf_SVKEY) 519 if (HeKLEN (he) == HEf_SVKEY)
336 { 520 {
337 SV *sv = HeSVKEY (he); 521 SV *sv = HeSVKEY (he);
338 STRLEN len; 522 STRLEN len;
339 char *str; 523 char *str;
340 524
341 SvGETMAGIC (sv); 525 SvGETMAGIC (sv);
342 str = SvPV (sv, len); 526 str = SvPV (sv, len);
343 527
344 encode_str (enc, str, len, SvUTF8 (sv)); 528 encode_str (enc, str, len, SvUTF8 (sv));
345 } 529 }
… …
349 encode_ch (enc, '"'); 533 encode_ch (enc, '"');
350 534
351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 535 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
352 encode_ch (enc, ':'); 536 encode_ch (enc, ':');
353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 537 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
354 encode_sv (enc, HeVAL (he));
355} 538}
356 539
357// compare hash entries, used when all keys are bytestrings 540// compare hash entries, used when all keys are bytestrings
358static int 541static int
359he_cmp_fast (const void *a_, const void *b_) 542he_cmp_fast (const void *a_, const void *b_)
… …
364 HE *b = *(HE **)b_; 547 HE *b = *(HE **)b_;
365 548
366 STRLEN la = HeKLEN (a); 549 STRLEN la = HeKLEN (a);
367 STRLEN lb = HeKLEN (b); 550 STRLEN lb = HeKLEN (b);
368 551
369 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 552 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
370 cmp = la - lb; 553 cmp = lb - la;
371 554
372 return cmp; 555 return cmp;
373} 556}
374 557
375// 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
376static int 559static int
377he_cmp_slow (const void *a, const void *b) 560he_cmp_slow (const void *a, const void *b)
378{ 561{
379 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 562 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
380} 563}
381 564
382static void 565static void
383encode_hv (enc_t *enc, HV *hv) 566encode_hv (enc_t *enc, HV *hv)
384{ 567{
385 int count, i; 568 HE *he;
386 569
387 if (enc->indent >= enc->maxdepth) 570 if (enc->indent >= enc->json.max_depth)
388 croak ("data structure too deep (hit recursion limit)"); 571 croak (ERR_NESTING_EXCEEDED);
389 572
390 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 573 encode_ch (enc, '{');
391 574
392 if ((count = hv_iterinit (hv)))
393 {
394 // for canonical output we have to sort by keys first 575 // for canonical output we have to sort by keys first
395 // actually, this is mostly due to the stupid so-called 576 // actually, this is mostly due to the stupid so-called
396 // security workaround added somewhere in 5.8.x. 577 // security workaround added somewhere in 5.8.x
397 // that randomises hash orderings 578 // that randomises hash orderings
398 if (enc->json.flags & F_CANONICAL) 579 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
580 {
581 int count = hv_iterinit (hv);
582
583 if (SvMAGICAL (hv))
399 { 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 {
400 int fast = 1; 599 int i, fast = 1;
401 HE *he; 600 HE *hes_stack [STACK_HES];
402#if defined(__BORLANDC__) || defined(_MSC_VER) 601 HE **hes = hes_stack;
403 HE **hes = _alloca (count * sizeof (HE)); 602
404#else 603 // allocate larger arrays on the heap
405 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 604 if (count > STACK_HES)
406#endif 605 {
606 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
607 hes = (HE **)SvPVX (sv);
608 }
407 609
408 i = 0; 610 i = 0;
409 while ((he = hv_iternext (hv))) 611 while ((he = hv_iternext (hv)))
410 { 612 {
411 hes [i++] = he; 613 hes [i++] = he;
… …
433 635
434 FREETMPS; 636 FREETMPS;
435 LEAVE; 637 LEAVE;
436 } 638 }
437 639
438 for (i = 0; i < count; ++i) 640 encode_nl (enc); ++enc->indent;
641
642 while (count--)
439 { 643 {
440 encode_indent (enc); 644 encode_indent (enc);
645 he = hes [count];
441 encode_he (enc, hes [i]); 646 encode_hk (enc, he);
647 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
442 648
443 if (i < count - 1) 649 if (count)
444 encode_comma (enc); 650 encode_comma (enc);
445 } 651 }
446 652
447 encode_nl (enc); 653 encode_nl (enc); --enc->indent; encode_indent (enc);
448 } 654 }
655 }
449 else 656 else
657 {
658 if (hv_iterinit (hv) || SvMAGICAL (hv))
659 if ((he = hv_iternext (hv)))
450 { 660 {
451 HE *he = hv_iternext (hv); 661 encode_nl (enc); ++enc->indent;
452 662
453 for (;;) 663 for (;;)
454 { 664 {
455 encode_indent (enc); 665 encode_indent (enc);
456 encode_he (enc, he); 666 encode_hk (enc, he);
667 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
457 668
458 if (!(he = hv_iternext (hv))) 669 if (!(he = hv_iternext (hv)))
459 break; 670 break;
460 671
461 encode_comma (enc); 672 encode_comma (enc);
462 } 673 }
463 674
464 encode_nl (enc); 675 encode_nl (enc); --enc->indent; encode_indent (enc);
465 } 676 }
466 } 677 }
467 678
468 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 679 encode_ch (enc, '}');
469} 680}
470 681
471// encode objects, arrays and special \0=false and \1=true values. 682// encode objects, arrays and special \0=false and \1=true values.
472static void 683static void
473encode_rv (enc_t *enc, SV *sv) 684encode_rv (enc_t *enc, SV *sv)
… …
477 SvGETMAGIC (sv); 688 SvGETMAGIC (sv);
478 svt = SvTYPE (sv); 689 svt = SvTYPE (sv);
479 690
480 if (expect_false (SvOBJECT (sv))) 691 if (expect_false (SvOBJECT (sv)))
481 { 692 {
693 HV *stash = !JSON_SLOW || json_boolean_stash
694 ? json_boolean_stash
695 : gv_stashpv ("JSON::XS::Boolean", 1);
696
482 if (SvSTASH (sv) == json_boolean_stash) 697 if (SvSTASH (sv) == stash)
483 { 698 {
484 if (SvIV (sv)) 699 if (SvIV (sv))
485 encode_str (enc, "true", 4, 0); 700 encode_str (enc, "true", 4, 0);
486 else 701 else
487 encode_str (enc, "false", 5, 0); 702 encode_str (enc, "false", 5, 0);
… …
499 // we re-bless the reference to get overload and other niceties right 714 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); 715 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
501 716
502 if (to_json) 717 if (to_json)
503 { 718 {
504 int count;
505 dSP; 719 dSP;
506 720
507 ENTER; SAVETMPS; PUSHMARK (SP); 721 ENTER; SAVETMPS; PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 722 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509 723
… …
547 761
548 if (len == 1 && *pv == '1') 762 if (len == 1 && *pv == '1')
549 encode_str (enc, "true", 4, 0); 763 encode_str (enc, "true", 4, 0);
550 else if (len == 1 && *pv == '0') 764 else if (len == 1 && *pv == '0')
551 encode_str (enc, "false", 5, 0); 765 encode_str (enc, "false", 5, 0);
766 else if (enc->json.flags & F_ALLOW_UNKNOWN)
767 encode_str (enc, "null", 4, 0);
552 else 768 else
553 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",
554 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 770 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
555 } 771 }
772 else if (enc->json.flags & F_ALLOW_UNKNOWN)
773 encode_str (enc, "null", 4, 0);
556 else 774 else
557 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",
558 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 776 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
559} 777}
560 778
… …
578 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 796 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
579 enc->cur += strlen (enc->cur); 797 enc->cur += strlen (enc->cur);
580 } 798 }
581 else if (SvIOKp (sv)) 799 else if (SvIOKp (sv))
582 { 800 {
583 // 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
584 if (SvUV (sv) & ~(UV)0x7fff) 802 // we assume two's complement
585 { 803 // we assume no aliasing issues in the union
586 // large integer, use the (rather slow) snprintf way. 804 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
587 need (enc, sizeof (UV) * 3); 805 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
588 enc->cur +=
589 SvIsUV(sv)
590 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
591 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
592 }
593 else
594 { 806 {
595 // optimise the "small number case" 807 // optimise the "small number case"
596 // 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
597 I32 i = SvIV (sv); 810 I32 i = SvIVX (sv);
598 U32 u; 811 U32 u;
599 char digit, nz = 0; 812 char digit, nz = 0;
600 813
601 need (enc, 6); 814 need (enc, 6);
602 815
… …
608 821
609 // 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
610 // 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,
611 // resulting in a net multiplication by 10. 824 // resulting in a net multiplication by 10.
612 // we always write the digit to memory but conditionally increment 825 // we always write the digit to memory but conditionally increment
613 // the pointer, to ease the usage of conditional move instructions. 826 // the pointer, to enable the use of conditional move instructions.
614 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;
615 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;
616 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;
617 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;
618 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'
619 } 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 }
620 } 842 }
621 else if (SvROK (sv)) 843 else if (SvROK (sv))
622 encode_rv (enc, SvRV (sv)); 844 encode_rv (enc, SvRV (sv));
623 else if (!SvOK (sv)) 845 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
624 encode_str (enc, "null", 4, 0); 846 encode_str (enc, "null", 4, 0);
625 else 847 else
626 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, check your input data",
627 SvPV_nolen (sv), SvFLAGS (sv)); 849 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
628} 850}
629 851
630static SV * 852static SV *
631encode_json (SV *scalar, JSON *json) 853encode_json (SV *scalar, JSON *json)
632{ 854{
… …
638 enc.json = *json; 860 enc.json = *json;
639 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 861 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
640 enc.cur = SvPVX (enc.sv); 862 enc.cur = SvPVX (enc.sv);
641 enc.end = SvEND (enc.sv); 863 enc.end = SvEND (enc.sv);
642 enc.indent = 0; 864 enc.indent = 0;
643 enc.maxdepth = DEC_DEPTH (enc.json.flags); 865 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
866 : enc.json.flags & F_LATIN1 ? 0x000100UL
867 : 0x110000UL;
644 868
645 SvPOK_only (enc.sv); 869 SvPOK_only (enc.sv);
646 encode_sv (&enc, scalar); 870 encode_sv (&enc, scalar);
871 encode_nl (&enc);
647 872
648 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 873 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
649 *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
650 875
651 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 876 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
669 JSON json; 894 JSON json;
670 U32 depth; // recursion depth 895 U32 depth; // recursion depth
671 U32 maxdepth; // recursion depth limit 896 U32 maxdepth; // recursion depth limit
672} dec_t; 897} dec_t;
673 898
674inline 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
675decode_ws (dec_t *dec) 909decode_ws (dec_t *dec)
676{ 910{
677 for (;;) 911 for (;;)
678 { 912 {
679 char ch = *dec->cur; 913 char ch = *dec->cur;
680 914
681 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 }
682 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 927 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
683 break; 928 break; // parse error, but let higher level handle it, gives better error messages
684 929
685 ++dec->cur; 930 ++dec->cur;
686 } 931 }
687} 932}
688 933
… …
692 if (*dec->cur != ch) \ 937 if (*dec->cur != ch) \
693 ERR (# ch " expected"); \ 938 ERR (# ch " expected"); \
694 ++dec->cur; \ 939 ++dec->cur; \
695 SE 940 SE
696 941
697#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)
698#define DEC_DEC_DEPTH --dec->depth 943#define DEC_DEC_DEPTH --dec->depth
699 944
700static SV *decode_sv (dec_t *dec); 945static SV *decode_sv (dec_t *dec);
701 946
702static signed char decode_hexdigit[256]; 947static signed char decode_hexdigit[256];
… …
794 1039
795 if (hi >= 0x80) 1040 if (hi >= 0x80)
796 { 1041 {
797 utf8 = 1; 1042 utf8 = 1;
798 1043
799 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1044 cur = encode_utf8 (cur, hi);
800 } 1045 }
801 else 1046 else
802 *cur++ = hi; 1047 *cur++ = hi;
803 } 1048 }
804 break; 1049 break;
… …
806 default: 1051 default:
807 --dec_cur; 1052 --dec_cur;
808 ERR ("illegal backslash escape sequence in string"); 1053 ERR ("illegal backslash escape sequence in string");
809 } 1054 }
810 } 1055 }
811 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1056 else if (expect_true (ch >= 0x20 && ch < 0x80))
812 *cur++ = ch; 1057 *cur++ = ch;
813 else if (ch >= 0x80) 1058 else if (ch >= 0x80)
814 { 1059 {
815 STRLEN clen; 1060 STRLEN clen;
816 UV uch;
817 1061
818 --dec_cur; 1062 --dec_cur;
819 1063
820 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1064 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
821 if (clen == (STRLEN)-1) 1065 if (clen == (STRLEN)-1)
822 ERR ("malformed UTF-8 character in JSON string"); 1066 ERR ("malformed UTF-8 character in JSON string");
823 1067
824 do 1068 do
825 *cur++ = *dec_cur++; 1069 *cur++ = *dec_cur++;
… …
842 { 1086 {
843 STRLEN len = cur - buf; 1087 STRLEN len = cur - buf;
844 1088
845 if (sv) 1089 if (sv)
846 { 1090 {
847 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
848 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1096 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
849 SvCUR_set (sv, SvCUR (sv) + len); 1097 SvCUR_set (sv, SvCUR (sv) + len);
850 } 1098 }
851 else 1099 else
852 sv = newSVpvn (buf, len); 1100 sv = newSVpvn (buf, len);
… …
937 is_nv = 1; 1185 is_nv = 1;
938 } 1186 }
939 1187
940 if (!is_nv) 1188 if (!is_nv)
941 { 1189 {
1190 int len = dec->cur - start;
1191
942 // 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
943 if (*start == '-') 1193 if (*start == '-')
944 switch (dec->cur - start) 1194 switch (len)
945 { 1195 {
946 case 2: return newSViv (-( start [1] - '0' * 1)); 1196 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
947 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1197 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
948 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));
949 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));
950 } 1201 }
951 else 1202 else
952 switch (dec->cur - start) 1203 switch (len)
953 { 1204 {
954 case 1: return newSViv ( start [0] - '0' * 1); 1205 case 1: return newSViv ( start [0] - '0' * 1);
955 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1206 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
956 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);
957 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);
958 } 1210 }
959 1211
960 { 1212 {
961 UV uv; 1213 UV uv;
962 int numtype = grok_number (start, dec->cur - start, &uv); 1214 int numtype = grok_number (start, len, &uv);
963 if (numtype & IS_NUMBER_IN_UV) 1215 if (numtype & IS_NUMBER_IN_UV)
964 if (numtype & IS_NUMBER_NEG) 1216 if (numtype & IS_NUMBER_NEG)
965 { 1217 {
966 if (uv < (UV)IV_MIN) 1218 if (uv < (UV)IV_MIN)
967 return newSViv (-(IV)uv); 1219 return newSViv (-(IV)uv);
968 } 1220 }
969 else 1221 else
970 return newSVuv (uv); 1222 return newSVuv (uv);
971
972 // here would likely be the place for bigint support
973 } 1223 }
974 }
975 1224
976 // if we ever support bigint or bigfloat, this is the place for bigfloat 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
977 return newSVnv (Atof (start)); 1237 return newSVnv (json_atof (start));
978 1238
979fail: 1239fail:
980 return 0; 1240 return 0;
981} 1241}
982 1242
… …
1006 if (*dec->cur == ']') 1266 if (*dec->cur == ']')
1007 { 1267 {
1008 ++dec->cur; 1268 ++dec->cur;
1009 break; 1269 break;
1010 } 1270 }
1011 1271
1012 if (*dec->cur != ',') 1272 if (*dec->cur != ',')
1013 ERR (", or ] expected while parsing array"); 1273 ERR (", or ] expected while parsing array");
1014 1274
1015 ++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 }
1016 } 1284 }
1017 1285
1018 DEC_DEC_DEPTH; 1286 DEC_DEC_DEPTH;
1019 return newRV_noinc ((SV *)av); 1287 return newRV_noinc ((SV *)av);
1020 1288
… …
1036 if (*dec->cur == '}') 1304 if (*dec->cur == '}')
1037 ++dec->cur; 1305 ++dec->cur;
1038 else 1306 else
1039 for (;;) 1307 for (;;)
1040 { 1308 {
1041 decode_ws (dec); EXPECT_CH ('"'); 1309 EXPECT_CH ('"');
1042 1310
1043 // heuristic: assume that 1311 // heuristic: assume that
1044 // a) decode_str + hv_store_ent are abysmally slow. 1312 // a) decode_str + hv_store_ent are abysmally slow.
1045 // b) most hash keys are short, simple ascii text. 1313 // b) most hash keys are short, simple ascii text.
1046 // => try to "fast-match" such strings to avoid 1314 // => try to "fast-match" such strings to avoid
… …
1050 char *p = dec->cur; 1318 char *p = dec->cur;
1051 char *e = p + 24; // only try up to 24 bytes 1319 char *e = p + 24; // only try up to 24 bytes
1052 1320
1053 for (;;) 1321 for (;;)
1054 { 1322 {
1055 // the >= 0x80 is true on most architectures 1323 // the >= 0x80 is false on most architectures
1056 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1324 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1057 { 1325 {
1058 // slow path, back up and use decode_str 1326 // slow path, back up and use decode_str
1059 SV *key = decode_str (dec); 1327 SV *key = decode_str (dec);
1060 if (!key) 1328 if (!key)
1061 goto fail; 1329 goto fail;
1062 1330
1063 decode_ws (dec); EXPECT_CH (':'); 1331 decode_ws (dec); EXPECT_CH (':');
1064 1332
1333 decode_ws (dec);
1065 value = decode_sv (dec); 1334 value = decode_sv (dec);
1066 if (!value) 1335 if (!value)
1067 { 1336 {
1068 SvREFCNT_dec (key); 1337 SvREFCNT_dec (key);
1069 goto fail; 1338 goto fail;
… …
1081 int len = p - key; 1350 int len = p - key;
1082 dec->cur = p + 1; 1351 dec->cur = p + 1;
1083 1352
1084 decode_ws (dec); EXPECT_CH (':'); 1353 decode_ws (dec); EXPECT_CH (':');
1085 1354
1355 decode_ws (dec);
1086 value = decode_sv (dec); 1356 value = decode_sv (dec);
1087 if (!value) 1357 if (!value)
1088 goto fail; 1358 goto fail;
1089 1359
1090 hv_store (hv, key, len, value, 0); 1360 hv_store (hv, key, len, value, 0);
… …
1106 1376
1107 if (*dec->cur != ',') 1377 if (*dec->cur != ',')
1108 ERR (", or } expected while parsing object/hash"); 1378 ERR (", or } expected while parsing object/hash");
1109 1379
1110 ++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 }
1111 } 1389 }
1112 1390
1113 DEC_DEC_DEPTH; 1391 DEC_DEC_DEPTH;
1114 sv = newRV_noinc ((SV *)hv); 1392 sv = newRV_noinc ((SV *)hv);
1115 1393
… …
1133 dSP; 1411 dSP;
1134 int count; 1412 int count;
1135 1413
1136 ENTER; SAVETMPS; PUSHMARK (SP); 1414 ENTER; SAVETMPS; PUSHMARK (SP);
1137 XPUSHs (HeVAL (he)); 1415 XPUSHs (HeVAL (he));
1416 sv_2mortal (sv);
1138 1417
1139 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1418 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1140 1419
1141 if (count == 1) 1420 if (count == 1)
1142 { 1421 {
1143 sv = newSVsv (POPs); 1422 sv = newSVsv (POPs);
1144 FREETMPS; LEAVE; 1423 FREETMPS; LEAVE;
1145 return sv; 1424 return sv;
1146 } 1425 }
1147 1426
1427 SvREFCNT_inc (sv);
1148 FREETMPS; LEAVE; 1428 FREETMPS; LEAVE;
1149 } 1429 }
1150 } 1430 }
1151 1431
1152 if (dec->json.cb_object) 1432 if (dec->json.cb_object)
… …
1180} 1460}
1181 1461
1182static SV * 1462static SV *
1183decode_sv (dec_t *dec) 1463decode_sv (dec_t *dec)
1184{ 1464{
1185 decode_ws (dec);
1186
1187 // the beauty of JSON: you need exactly one character lookahead 1465 // the beauty of JSON: you need exactly one character lookahead
1188 // to parse anything. 1466 // to parse everything.
1189 switch (*dec->cur) 1467 switch (*dec->cur)
1190 { 1468 {
1191 case '"': ++dec->cur; return decode_str (dec); 1469 case '"': ++dec->cur; return decode_str (dec);
1192 case '[': ++dec->cur; return decode_av (dec); 1470 case '[': ++dec->cur; return decode_av (dec);
1193 case '{': ++dec->cur; return decode_hv (dec); 1471 case '{': ++dec->cur; return decode_hv (dec);
1194 1472
1195 case '-': 1473 case '-':
1196 case '0': case '1': case '2': case '3': case '4': 1474 case '0': case '1': case '2': case '3': case '4':
1197 case '5': case '6': case '7': case '8': case '9': 1475 case '5': case '6': case '7': case '8': case '9':
1198 return decode_num (dec); 1476 return decode_num (dec);
1199 1477
1200 case 't': 1478 case 't':
1201 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1479 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1202 { 1480 {
1203 dec->cur += 4; 1481 dec->cur += 4;
1482#if JSON_SLOW
1483 json_true = get_bool ("JSON::XS::true");
1484#endif
1204 return SvREFCNT_inc (json_true); 1485 return newSVsv (json_true);
1205 } 1486 }
1206 else 1487 else
1207 ERR ("'true' expected"); 1488 ERR ("'true' expected");
1208 1489
1209 break; 1490 break;
1210 1491
1211 case 'f': 1492 case 'f':
1212 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1493 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1213 { 1494 {
1214 dec->cur += 5; 1495 dec->cur += 5;
1496#if JSON_SLOW
1497 json_false = get_bool ("JSON::XS::false");
1498#endif
1215 return SvREFCNT_inc (json_false); 1499 return newSVsv (json_false);
1216 } 1500 }
1217 else 1501 else
1218 ERR ("'false' expected"); 1502 ERR ("'false' expected");
1219 1503
1220 break; 1504 break;
… …
1238fail: 1522fail:
1239 return 0; 1523 return 0;
1240} 1524}
1241 1525
1242static SV * 1526static SV *
1243decode_json (SV *string, JSON *json, UV *offset_return) 1527decode_json (SV *string, JSON *json, char **offset_return)
1244{ 1528{
1245 dec_t dec; 1529 dec_t dec;
1246 UV offset;
1247 SV *sv; 1530 SV *sv;
1248 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 */
1249 SvGETMAGIC (string); 1537 /*SvGETMAGIC (string);*/
1538 if (SvMAGICAL (string) || !SvPOK (string))
1539 string = sv_2mortal (newSVsv (string));
1540
1250 SvUPGRADE (string, SVt_PV); 1541 SvUPGRADE (string, SVt_PV);
1251 1542
1252 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 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)
1253 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1561 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1254 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1562 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1563 }
1255 1564
1256 if (json->flags & F_UTF8) 1565 if (DECODE_WANTS_OCTETS (json))
1257 sv_utf8_downgrade (string, 0); 1566 sv_utf8_downgrade (string, 0);
1258 else 1567 else
1259 sv_utf8_upgrade (string); 1568 sv_utf8_upgrade (string);
1260 1569
1261 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1570 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1262 1571
1263 dec.json = *json; 1572 dec.json = *json;
1264 dec.cur = SvPVX (string); 1573 dec.cur = SvPVX (string);
1265 dec.end = SvEND (string); 1574 dec.end = SvEND (string);
1266 dec.err = 0; 1575 dec.err = 0;
1267 dec.depth = 0; 1576 dec.depth = 0;
1268 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1269 1577
1270 if (dec.json.cb_object || dec.json.cb_sk_object) 1578 if (dec.json.cb_object || dec.json.cb_sk_object)
1271 dec.json.flags |= F_HOOK; 1579 dec.json.flags |= F_HOOK;
1272 1580
1273 *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);
1274 sv = decode_sv (&dec); 1584 sv = decode_sv (&dec);
1585
1586 if (offset_return)
1587 *offset_return = dec.cur;
1275 1588
1276 if (!(offset_return || !sv)) 1589 if (!(offset_return || !sv))
1277 { 1590 {
1278 // check for trailing garbage 1591 // check for trailing garbage
1279 decode_ws (&dec); 1592 decode_ws (&dec);
… …
1282 { 1595 {
1283 dec.err = "garbage after JSON object"; 1596 dec.err = "garbage after JSON object";
1284 SvREFCNT_dec (sv); 1597 SvREFCNT_dec (sv);
1285 sv = 0; 1598 sv = 0;
1286 } 1599 }
1287 }
1288
1289 if (offset_return || !sv)
1290 {
1291 offset = dec.json.flags & F_UTF8
1292 ? dec.cur - SvPVX (string)
1293 : utf8_distance (dec.cur, SvPVX (string));
1294
1295 if (offset_return)
1296 *offset_return = offset;
1297 } 1600 }
1298 1601
1299 if (!sv) 1602 if (!sv)
1300 { 1603 {
1301 SV *uni = sv_newmortal (); 1604 SV *uni = sv_newmortal ();
… …
1307 SAVEVPTR (PL_curcop); 1610 SAVEVPTR (PL_curcop);
1308 PL_curcop = &cop; 1611 PL_curcop = &cop;
1309 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);
1310 LEAVE; 1613 LEAVE;
1311 1614
1312 croak ("%s, at character offset %d [\"%s\"]", 1615 croak ("%s, at character offset %d (before \"%s\")",
1313 dec.err, 1616 dec.err,
1314 (int)offset, 1617 (int)ptr_to_index (string, dec.cur),
1315 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1618 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1316 } 1619 }
1317 1620
1318 sv = sv_2mortal (sv); 1621 sv = sv_2mortal (sv);
1319 1622
1320 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1623 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1321 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)");
1322 1625
1323 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
1324} 1780}
1325 1781
1326///////////////////////////////////////////////////////////////////////////// 1782/////////////////////////////////////////////////////////////////////////////
1327// XS interface functions 1783// XS interface functions
1328 1784
… …
1340 : -1; 1796 : -1;
1341 1797
1342 json_stash = gv_stashpv ("JSON::XS" , 1); 1798 json_stash = gv_stashpv ("JSON::XS" , 1);
1343 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1799 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1344 1800
1345 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1801 json_true = get_bool ("JSON::XS::true");
1346 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 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 */
1347} 1805}
1348 1806
1349PROTOTYPES: DISABLE 1807PROTOTYPES: DISABLE
1808
1809void CLONE (...)
1810 CODE:
1811 json_stash = 0;
1812 json_boolean_stash = 0;
1350 1813
1351void new (char *klass) 1814void new (char *klass)
1352 PPCODE: 1815 PPCODE:
1353{ 1816{
1354 SV *pv = NEWSV (0, sizeof (JSON)); 1817 SV *pv = NEWSV (0, sizeof (JSON));
1355 SvPOK_only (pv); 1818 SvPOK_only (pv);
1356 Zero (SvPVX (pv), 1, JSON); 1819 json_init ((JSON *)SvPVX (pv));
1357 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1820 XPUSHs (sv_2mortal (sv_bless (
1358 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1821 newRV_noinc (pv),
1822 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1823 )));
1359} 1824}
1360 1825
1361void ascii (JSON *self, int enable = 1) 1826void ascii (JSON *self, int enable = 1)
1362 ALIAS: 1827 ALIAS:
1363 ascii = F_ASCII 1828 ascii = F_ASCII
… …
1370 pretty = F_PRETTY 1835 pretty = F_PRETTY
1371 allow_nonref = F_ALLOW_NONREF 1836 allow_nonref = F_ALLOW_NONREF
1372 shrink = F_SHRINK 1837 shrink = F_SHRINK
1373 allow_blessed = F_ALLOW_BLESSED 1838 allow_blessed = F_ALLOW_BLESSED
1374 convert_blessed = F_CONV_BLESSED 1839 convert_blessed = F_CONV_BLESSED
1840 relaxed = F_RELAXED
1841 allow_unknown = F_ALLOW_UNKNOWN
1375 PPCODE: 1842 PPCODE:
1376{ 1843{
1377 if (enable) 1844 if (enable)
1378 self->flags |= ix; 1845 self->flags |= ix;
1379 else 1846 else
1380 self->flags &= ~ix; 1847 self->flags &= ~ix;
1381 1848
1382 XPUSHs (ST (0)); 1849 XPUSHs (ST (0));
1383} 1850}
1384 1851
1385void max_depth (JSON *self, UV max_depth = 0x80000000UL) 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
1386 PPCODE: 1867 PPCODE:
1387{ 1868 XPUSHs (boolSV (self->flags & ix));
1388 UV log2 = 0;
1389 1869
1390 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1870void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1391 1871 PPCODE:
1392 while ((1UL << log2) < max_depth) 1872 self->max_depth = max_depth;
1393 ++log2;
1394
1395 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1396
1397 XPUSHs (ST (0)); 1873 XPUSHs (ST (0));
1398}
1399 1874
1875U32 get_max_depth (JSON *self)
1876 CODE:
1877 RETVAL = self->max_depth;
1878 OUTPUT:
1879 RETVAL
1880
1400void max_size (JSON *self, UV max_size = 0) 1881void max_size (JSON *self, U32 max_size = 0)
1401 PPCODE: 1882 PPCODE:
1402{ 1883 self->max_size = max_size;
1403 UV log2 = 0;
1404
1405 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1406 if (max_size == 1) max_size = 2;
1407
1408 while ((1UL << log2) < max_size)
1409 ++log2;
1410
1411 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1412
1413 XPUSHs (ST (0)); 1884 XPUSHs (ST (0));
1414} 1885
1886int get_max_size (JSON *self)
1887 CODE:
1888 RETVAL = self->max_size;
1889 OUTPUT:
1890 RETVAL
1415 1891
1416void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1892void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1417 PPCODE: 1893 PPCODE:
1418{ 1894{
1419 SvREFCNT_dec (self->cb_object); 1895 SvREFCNT_dec (self->cb_object);
… …
1423} 1899}
1424 1900
1425void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 1901void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1426 PPCODE: 1902 PPCODE:
1427{ 1903{
1428 if (!self->cb_sk_object) 1904 if (!self->cb_sk_object)
1429 self->cb_sk_object = newHV (); 1905 self->cb_sk_object = newHV ();
1430 1906
1431 if (SvOK (cb)) 1907 if (SvOK (cb))
1432 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 1908 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1433 else 1909 else
… …
1444 XPUSHs (ST (0)); 1920 XPUSHs (ST (0));
1445} 1921}
1446 1922
1447void encode (JSON *self, SV *scalar) 1923void encode (JSON *self, SV *scalar)
1448 PPCODE: 1924 PPCODE:
1449 XPUSHs (encode_json (scalar, self)); 1925 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
1926 XPUSHs (scalar);
1450 1927
1451void decode (JSON *self, SV *jsonstr) 1928void decode (JSON *self, SV *jsonstr)
1452 PPCODE: 1929 PPCODE:
1453 XPUSHs (decode_json (jsonstr, self, 0)); 1930 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
1931 XPUSHs (jsonstr);
1454 1932
1455void decode_prefix (JSON *self, SV *jsonstr) 1933void decode_prefix (JSON *self, SV *jsonstr)
1456 PPCODE: 1934 PPCODE:
1457{ 1935{
1458 UV offset; 1936 SV *sv;
1937 char *offset;
1938 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1459 EXTEND (SP, 2); 1939 EXTEND (SP, 2);
1460 PUSHs (decode_json (jsonstr, self, &offset)); 1940 PUSHs (sv);
1461 PUSHs (sv_2mortal (newSVuv (offset))); 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
2035 CODE:
2036{
2037 if (self->incr_pos)
2038 croak ("incr_text can not be called when the incremental parser already started parsing");
2039
2040 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2041}
2042 OUTPUT:
2043 RETVAL
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;
1462} 2065}
1463 2066
1464void DESTROY (JSON *self) 2067void DESTROY (JSON *self)
1465 CODE: 2068 CODE:
1466 SvREFCNT_dec (self->cb_sk_object); 2069 SvREFCNT_dec (self->cb_sk_object);
1467 SvREFCNT_dec (self->cb_object); 2070 SvREFCNT_dec (self->cb_object);
2071 SvREFCNT_dec (self->incr_text);
1468 2072
1469PROTOTYPES: ENABLE 2073PROTOTYPES: ENABLE
1470 2074
1471void to_json (SV *scalar) 2075void encode_json (SV *scalar)
1472 PPCODE: 2076 PPCODE:
1473{ 2077{
1474 JSON json = { F_DEFAULT | F_UTF8 }; 2078 JSON json;
1475 XPUSHs (encode_json (scalar, &json)); 2079 json_init (&json);
2080 json.flags |= F_UTF8;
2081 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2082 XPUSHs (scalar);
1476} 2083}
1477 2084
1478void from_json (SV *jsonstr) 2085void decode_json (SV *jsonstr)
1479 PPCODE: 2086 PPCODE:
1480{ 2087{
1481 JSON json = { F_DEFAULT | F_UTF8 }; 2088 JSON json;
1482 XPUSHs (decode_json (jsonstr, &json, 0)); 2089 json_init (&json);
2090 json.flags |= F_UTF8;
2091 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2092 XPUSHs (jsonstr);
1483} 2093}
1484 2094

Diff Legend

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