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 1.70 by root, Wed Mar 19 00:44:54 2008 UTC vs.
Revision 1.134 by root, Thu Nov 15 20:13:03 2018 UTC

… …
4 4
5#include <assert.h> 5#include <assert.h>
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h>
9#include <float.h> 10#include <float.h>
11#include <inttypes.h>
10 12
11#if defined(__BORLANDC__) || defined(_MSC_VER) 13#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 14# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 15#endif
14 16
15// some old perls do not have this, try to make it work, no 17// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces. 18// guarantees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 19#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 20# define UTF8_MAXBYTES 13
19#endif 21#endif
22
23// compatibility with perl <5.18
24#ifndef HvNAMELEN_get
25# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
26#endif
27#ifndef HvNAMELEN
28# define HvNAMELEN(hv) HvNAMELEN_get (hv)
29#endif
30#ifndef HvNAMEUTF8
31# define HvNAMEUTF8(hv) 0
32#endif
33
34// three extra for rounding, sign, and end of string
35#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
20 36
21#define F_ASCII 0x00000001UL 37#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 38#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 39#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 40#define F_INDENT 0x00000008UL
… …
28#define F_ALLOW_NONREF 0x00000100UL 44#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 45#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 46#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 47#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL 48#define F_RELAXED 0x00001000UL
33 49#define F_ALLOW_UNKNOWN 0x00002000UL
34#define F_MAXDEPTH 0xf8000000UL 50#define F_ALLOW_TAGS 0x00004000UL
35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39 52
40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
42
43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
44#define F_DEFAULT (9UL << S_MAXDEPTH)
45 54
46#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 64 // initial scalar size to be allocated
47#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
48 57
49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 58#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
59
60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
50 61
51#define SB do { 62#define SB do {
52#define SE } while (0) 63#define SE } while (0)
53 64
54#if __GNUC__ >= 3 65#if __GNUC__ >= 3
55# define expect(expr,value) __builtin_expect ((expr),(value)) 66# define expect(expr,value) __builtin_expect ((expr), (value))
56# define inline inline 67# define INLINE static inline
57#else 68#else
58# define expect(expr,value) (expr) 69# define expect(expr,value) (expr)
59# define inline static 70# define INLINE static
60#endif 71#endif
61 72
62#define expect_false(expr) expect ((expr) != 0, 0) 73#define expect_false(expr) expect ((expr) != 0, 0)
63#define expect_true(expr) expect ((expr) != 0, 1) 74#define expect_true(expr) expect ((expr) != 0, 1)
64 75
76#define IN_RANGE_INC(type,val,beg,end) \
77 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
79
80#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
81
65#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
66# define JSON_SLOW 1
67# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 83# define JSON_STASH (expect_true (json_stash) ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (expect_true (bool_stash) ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
85# define GET_BOOL(value) (expect_true (bool_ ## value) ? bool_ ## value : get_bool ("Types::Serialiser::" # value))
68#else 86#else
69# define JSON_SLOW 0
70# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
71#endif 90#endif
72 91
73static HV *json_stash, *json_boolean_stash; // JSON::XS:: 92// the amount of HEs to allocate on the stack, when sorting keys
74static SV *json_true, *json_false; 93#define STACK_HES 64
94
95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
96static SV *bool_false, *bool_true;
97static SV *sv_json;
98
99enum {
100 INCR_M_WS = 0, // initial whitespace skipping, must be 0
101 INCR_M_STR, // inside string
102 INCR_M_BS, // inside backslash
103 INCR_M_C0, // inside comment in initial whitespace sequence
104 INCR_M_C1, // inside comment in other places
105 INCR_M_JSON // outside anything, count nesting
106};
107
108#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
75 109
76typedef struct { 110typedef struct {
77 U32 flags; 111 U32 flags;
112 U32 max_depth;
113 STRLEN max_size;
114
78 SV *cb_object; 115 SV *cb_object;
79 HV *cb_sk_object; 116 HV *cb_sk_object;
117
118 // for the incremental parser
119 SV *incr_text; // the source text so far
120 STRLEN incr_pos; // the current offset into the text
121 int incr_nest; // {[]}-nesting level
122 unsigned char incr_mode;
123
124 SV *v_false, *v_true;
80} JSON; 125} JSON;
126
127INLINE void
128json_init (JSON *json)
129{
130 Zero (json, 1, JSON);
131 json->max_depth = 512;
132}
81 133
82///////////////////////////////////////////////////////////////////////////// 134/////////////////////////////////////////////////////////////////////////////
83// utility functions 135// utility functions
84 136
85inline void 137INLINE SV *
138get_bool (const char *name)
139{
140 SV *sv = get_sv (name, 1);
141
142 SvREADONLY_on (sv);
143 SvREADONLY_on (SvRV (sv));
144
145 return sv;
146}
147
148INLINE void
86shrink (SV *sv) 149shrink (SV *sv)
87{ 150{
88 sv_utf8_downgrade (sv, 1); 151 sv_utf8_downgrade (sv, 1);
152
89 if (SvLEN (sv) > SvCUR (sv) + 1) 153 if (SvLEN (sv) > SvCUR (sv) + 1)
90 { 154 {
91#ifdef SvPV_shrink_to_cur 155#ifdef SvPV_shrink_to_cur
92 SvPV_shrink_to_cur (sv); 156 SvPV_shrink_to_cur (sv);
93#elif defined (SvPV_renew) 157#elif defined (SvPV_renew)
94 SvPV_renew (sv, SvCUR (sv) + 1); 158 SvPV_renew (sv, SvCUR (sv) + 1);
95#endif 159#endif
96 } 160 }
97} 161}
98 162
163/* adds two STRLENs together, slow, and with paranoia */
164STRLEN
165strlen_sum (STRLEN l1, STRLEN l2)
166{
167 size_t sum = l1 + l2;
168
169 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
170 croak ("JSON::XS: string size overflow");
171
172 return sum;
173}
174
175/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
176static char *
177json_sv_grow (SV *sv, size_t len1, size_t len2)
178{
179 len1 = strlen_sum (len1, len2);
180 len1 = strlen_sum (len1, len1 >> 1);
181
182 if (len1 > 4096 - 24)
183 len1 = (len1 | 4095) - 24;
184
185 return SvGROW (sv, len1);
186}
187
99// decode an utf-8 character and return it, or (UV)-1 in 188// decode a utf-8 character and return it, or (UV)-1 in
100// case of an error. 189// case of an error.
101// we special-case "safe" characters from U+80 .. U+7FF, 190// we special-case "safe" characters from U+80 .. U+7FF,
102// but use the very good perl function to parse anything else. 191// but use the very good perl function to parse anything else.
103// note that we never call this function for a ascii codepoints 192// note that we never call this function for a ascii codepoints
104inline UV 193INLINE UV
105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 194decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
106{ 195{
107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 196 if (expect_true (len >= 2
108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 197 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 198 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
110 { 199 {
111 *clen = 2; 200 *clen = 2;
112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 201 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
113 } 202 }
114 else 203 else
204 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
205}
206
207// likewise for encoding, also never called for ascii codepoints
208// this function takes advantage of this fact, although current gccs
209// seem to optimise the check for >= 0x80 away anyways
210INLINE unsigned char *
211encode_utf8 (unsigned char *s, UV ch)
212{
213 if (expect_false (ch < 0x000080))
214 *s++ = ch;
215 else if (expect_true (ch < 0x000800))
216 *s++ = 0xc0 | ( ch >> 6),
217 *s++ = 0x80 | ( ch & 0x3f);
218 else if ( ch < 0x010000)
219 *s++ = 0xe0 | ( ch >> 12),
220 *s++ = 0x80 | ((ch >> 6) & 0x3f),
221 *s++ = 0x80 | ( ch & 0x3f);
222 else if ( ch < 0x110000)
223 *s++ = 0xf0 | ( ch >> 18),
224 *s++ = 0x80 | ((ch >> 12) & 0x3f),
225 *s++ = 0x80 | ((ch >> 6) & 0x3f),
226 *s++ = 0x80 | ( ch & 0x3f);
227
228 return s;
229}
230
231// convert offset pointer to character index, sv must be string
232static STRLEN
233ptr_to_index (SV *sv, char *offset)
234{
235 return SvUTF8 (sv)
236 ? utf8_distance (offset, SvPVX (sv))
237 : offset - SvPVX (sv);
238}
239
240/////////////////////////////////////////////////////////////////////////////
241// fp hell
242
243// scan a group of digits, and a trailing exponent
244static void
245json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
246{
247 UV uaccum = 0;
248 int eaccum = 0;
249
250 // if we recurse too deep, skip all remaining digits
251 // to avoid a stack overflow attack
252 if (expect_false (--maxdepth <= 0))
253 while (((U8)*s - '0') < 10)
254 ++s;
255
256 for (;;)
257 {
258 U8 dig = (U8)*s - '0';
259
260 if (expect_false (dig >= 10))
261 {
262 if (dig == (U8)((U8)'.' - (U8)'0'))
263 {
264 ++s;
265 json_atof_scan1 (s, accum, expo, 1, maxdepth);
266 }
267 else if ((dig | ' ') == 'e' - '0')
268 {
269 int exp2 = 0;
270 int neg = 0;
271
272 ++s;
273
274 if (*s == '-')
275 {
276 ++s;
277 neg = 1;
278 }
279 else if (*s == '+')
280 ++s;
281
282 while ((dig = (U8)*s - '0') < 10)
283 exp2 = exp2 * 10 + *s++ - '0';
284
285 *expo += neg ? -exp2 : exp2;
286 }
287
288 break;
289 }
290
291 ++s;
292
293 uaccum = uaccum * 10 + dig;
294 ++eaccum;
295
296 // if we have too many digits, then recurse for more
297 // we actually do this for rather few digits
298 if (uaccum >= (UV_MAX - 9) / 10)
299 {
300 if (postdp) *expo -= eaccum;
301 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
302 if (postdp) *expo += eaccum;
303
304 break;
305 }
115 { 306 }
116 *clen = (STRLEN)-1; 307
308 // this relies greatly on the quality of the pow ()
309 // implementation of the platform, but a good
310 // implementation is hard to beat.
311 // (IEEE 754 conformant ones are required to be exact)
312 if (postdp) *expo -= eaccum;
313 *accum += uaccum * Perl_pow (10., *expo);
314 *expo += eaccum;
315}
316
317static NV
318json_atof (const char *s)
319{
320 NV accum = 0.;
321 int expo = 0;
322 int neg = 0;
323
324 if (*s == '-')
325 {
326 ++s;
327 neg = 1;
328 }
329
330 // a recursion depth of ten gives us >>500 bits
331 json_atof_scan1 (s, &accum, &expo, 0, 10);
332
333 return neg ? -accum : accum;
334}
335
336// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
337static int
338ref_bool_type (SV *sv)
339{
340 svtype svt = SvTYPE (sv);
341
342 if (svt < SVt_PVAV)
343 {
344 STRLEN len = 0;
345 char *pv = svt ? SvPV (sv, len) : 0;
346
347 if (len == 1)
348 if (*pv == '1')
349 return 1;
350 else if (*pv == '0')
351 return 0;
352 }
353
354 return -1;
355}
356
357// returns whether scalar is not a reference in the sense of allow_nonref
358static int
359json_nonref (SV *scalar)
360{
361 if (!SvROK (scalar))
362 return 1;
363
364 scalar = SvRV (scalar);
365
366 if (SvTYPE (scalar) >= SVt_PVMG)
367 {
368 if (SvSTASH (scalar) == bool_stash)
117 return (UV)-1; 369 return 1;
370
371 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
372 return 1;
118 } 373 }
374
375 return 0;
119} 376}
120 377
121///////////////////////////////////////////////////////////////////////////// 378/////////////////////////////////////////////////////////////////////////////
122// encoder 379// encoder
123 380
… …
127 char *cur; // SvPVX (sv) + current output position 384 char *cur; // SvPVX (sv) + current output position
128 char *end; // SvEND (sv) 385 char *end; // SvEND (sv)
129 SV *sv; // result scalar 386 SV *sv; // result scalar
130 JSON json; 387 JSON json;
131 U32 indent; // indentation level 388 U32 indent; // indentation level
132 U32 maxdepth; // max. indentation/recursion level 389 UV limit; // escape character values >= this value when encoding
133} enc_t; 390} enc_t;
134 391
135inline void 392INLINE void
136need (enc_t *enc, STRLEN len) 393need (enc_t *enc, STRLEN len)
137{ 394{
138 if (expect_false (enc->cur + len >= enc->end)) 395 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
139 { 396 {
140 STRLEN cur = enc->cur - SvPVX (enc->sv); 397 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
141 SvGROW (enc->sv, cur + len + 1); 398 char *buf = json_sv_grow (enc->sv, cur, len);
142 enc->cur = SvPVX (enc->sv) + cur; 399 enc->cur = buf + cur;
143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 400 enc->end = buf + SvLEN (enc->sv) - 1;
144 } 401 }
145} 402}
146 403
147inline void 404INLINE void
148encode_ch (enc_t *enc, char ch) 405encode_ch (enc_t *enc, char ch)
149{ 406{
150 need (enc, 1); 407 need (enc, 1);
151 *enc->cur++ = ch; 408 *enc->cur++ = ch;
152} 409}
… …
164 421
165 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 422 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
166 { 423 {
167 if (expect_false (ch == '"')) // but with slow exceptions 424 if (expect_false (ch == '"')) // but with slow exceptions
168 { 425 {
169 need (enc, len += 1); 426 need (enc, len + 1);
170 *enc->cur++ = '\\'; 427 *enc->cur++ = '\\';
171 *enc->cur++ = '"'; 428 *enc->cur++ = '"';
172 } 429 }
173 else if (expect_false (ch == '\\')) 430 else if (expect_false (ch == '\\'))
174 { 431 {
175 need (enc, len += 1); 432 need (enc, len + 1);
176 *enc->cur++ = '\\'; 433 *enc->cur++ = '\\';
177 *enc->cur++ = '\\'; 434 *enc->cur++ = '\\';
178 } 435 }
179 else 436 else
180 *enc->cur++ = ch; 437 *enc->cur++ = ch;
… …
183 } 440 }
184 else 441 else
185 { 442 {
186 switch (ch) 443 switch (ch)
187 { 444 {
188 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 445 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
189 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 446 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
190 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 447 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
191 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 448 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
192 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 449 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
193 450
194 default: 451 default:
195 { 452 {
196 STRLEN clen; 453 STRLEN clen;
197 UV uch; 454 UV uch;
… …
206 { 463 {
207 uch = ch; 464 uch = ch;
208 clen = 1; 465 clen = 1;
209 } 466 }
210 467
211 if (uch > 0x10FFFFUL) 468 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
212 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
213
214 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
215 { 469 {
216 if (uch > 0xFFFFUL) 470 if (uch >= 0x10000UL)
217 { 471 {
472 if (uch >= 0x110000UL)
473 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
474
218 need (enc, len += 11); 475 need (enc, len + 11);
219 sprintf (enc->cur, "\\u%04x\\u%04x", 476 sprintf (enc->cur, "\\u%04x\\u%04x",
220 (int)((uch - 0x10000) / 0x400 + 0xD800), 477 (int)((uch - 0x10000) / 0x400 + 0xD800),
221 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 478 (int)((uch - 0x10000) % 0x400 + 0xDC00));
222 enc->cur += 12; 479 enc->cur += 12;
223 } 480 }
224 else 481 else
225 { 482 {
226 static char hexdigit [16] = "0123456789abcdef";
227 need (enc, len += 5); 483 need (enc, len + 5);
228 *enc->cur++ = '\\'; 484 *enc->cur++ = '\\';
229 *enc->cur++ = 'u'; 485 *enc->cur++ = 'u';
230 *enc->cur++ = hexdigit [ uch >> 12 ]; 486 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
231 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 487 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
232 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 488 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
233 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 489 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
234 } 490 }
235 491
236 str += clen; 492 str += clen;
237 } 493 }
238 else if (enc->json.flags & F_LATIN1) 494 else if (enc->json.flags & F_LATIN1)
… …
240 *enc->cur++ = uch; 496 *enc->cur++ = uch;
241 str += clen; 497 str += clen;
242 } 498 }
243 else if (is_utf8) 499 else if (is_utf8)
244 { 500 {
245 need (enc, len += clen); 501 need (enc, len + clen);
246 do 502 do
247 { 503 {
248 *enc->cur++ = *str++; 504 *enc->cur++ = *str++;
249 } 505 }
250 while (--clen); 506 while (--clen);
251 } 507 }
252 else 508 else
253 { 509 {
254 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 510 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
255 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 511 enc->cur = encode_utf8 (enc->cur, uch);
256 ++str; 512 ++str;
257 } 513 }
258 } 514 }
259 } 515 }
260 } 516 }
261 517
262 --len; 518 --len;
263 } 519 }
264} 520}
265 521
266inline void 522INLINE void
267encode_indent (enc_t *enc) 523encode_indent (enc_t *enc)
268{ 524{
269 if (enc->json.flags & F_INDENT) 525 if (enc->json.flags & F_INDENT)
270 { 526 {
271 int spaces = enc->indent * INDENT_STEP; 527 int spaces = enc->indent * INDENT_STEP;
… …
274 memset (enc->cur, ' ', spaces); 530 memset (enc->cur, ' ', spaces);
275 enc->cur += spaces; 531 enc->cur += spaces;
276 } 532 }
277} 533}
278 534
279inline void 535INLINE void
280encode_space (enc_t *enc) 536encode_space (enc_t *enc)
281{ 537{
282 need (enc, 1); 538 need (enc, 1);
283 encode_ch (enc, ' '); 539 encode_ch (enc, ' ');
284} 540}
285 541
286inline void 542INLINE void
287encode_nl (enc_t *enc) 543encode_nl (enc_t *enc)
288{ 544{
289 if (enc->json.flags & F_INDENT) 545 if (enc->json.flags & F_INDENT)
290 { 546 {
291 need (enc, 1); 547 need (enc, 1);
292 encode_ch (enc, '\n'); 548 encode_ch (enc, '\n');
293 } 549 }
294} 550}
295 551
296inline void 552INLINE void
297encode_comma (enc_t *enc) 553encode_comma (enc_t *enc)
298{ 554{
299 encode_ch (enc, ','); 555 encode_ch (enc, ',');
300 556
301 if (enc->json.flags & F_INDENT) 557 if (enc->json.flags & F_INDENT)
… …
309static void 565static void
310encode_av (enc_t *enc, AV *av) 566encode_av (enc_t *enc, AV *av)
311{ 567{
312 int i, len = av_len (av); 568 int i, len = av_len (av);
313 569
314 if (enc->indent >= enc->maxdepth) 570 if (enc->indent >= enc->json.max_depth)
315 croak ("data structure too deep (hit recursion limit)"); 571 croak (ERR_NESTING_EXCEEDED);
316 572
317 encode_ch (enc, '['); 573 encode_ch (enc, '[');
318 574
319 if (len >= 0) 575 if (len >= 0)
320 { 576 {
321 encode_nl (enc); ++enc->indent; 577 encode_nl (enc); ++enc->indent;
322 578
323 for (i = 0; i <= len; ++i) 579 for (i = 0; i <= len; ++i)
… …
335 encode_comma (enc); 591 encode_comma (enc);
336 } 592 }
337 593
338 encode_nl (enc); --enc->indent; encode_indent (enc); 594 encode_nl (enc); --enc->indent; encode_indent (enc);
339 } 595 }
340 596
341 encode_ch (enc, ']'); 597 encode_ch (enc, ']');
342} 598}
343 599
344static void 600static void
345encode_hk (enc_t *enc, HE *he) 601encode_hk (enc_t *enc, HE *he)
… …
349 if (HeKLEN (he) == HEf_SVKEY) 605 if (HeKLEN (he) == HEf_SVKEY)
350 { 606 {
351 SV *sv = HeSVKEY (he); 607 SV *sv = HeSVKEY (he);
352 STRLEN len; 608 STRLEN len;
353 char *str; 609 char *str;
354 610
355 SvGETMAGIC (sv); 611 SvGETMAGIC (sv);
356 str = SvPV (sv, len); 612 str = SvPV (sv, len);
357 613
358 encode_str (enc, str, len, SvUTF8 (sv)); 614 encode_str (enc, str, len, SvUTF8 (sv));
359 } 615 }
… …
394 650
395static void 651static void
396encode_hv (enc_t *enc, HV *hv) 652encode_hv (enc_t *enc, HV *hv)
397{ 653{
398 HE *he; 654 HE *he;
399 int count;
400 655
401 if (enc->indent >= enc->maxdepth) 656 if (enc->indent >= enc->json.max_depth)
402 croak ("data structure too deep (hit recursion limit)"); 657 croak (ERR_NESTING_EXCEEDED);
403 658
404 encode_ch (enc, '{'); 659 encode_ch (enc, '{');
405 660
406 // for canonical output we have to sort by keys first 661 // for canonical output we have to sort by keys first
407 // actually, this is mostly due to the stupid so-called 662 // actually, this is mostly due to the stupid so-called
408 // security workaround added somewhere in 5.8.x. 663 // security workaround added somewhere in 5.8.x
409 // that randomises hash orderings 664 // that randomises hash orderings
410 if (enc->json.flags & F_CANONICAL) 665 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
411 { 666 {
412 int count = hv_iterinit (hv); 667 int count = hv_iterinit (hv);
413 668
414 if (SvMAGICAL (hv)) 669 if (SvMAGICAL (hv))
415 { 670 {
… …
426 } 681 }
427 682
428 if (count) 683 if (count)
429 { 684 {
430 int i, fast = 1; 685 int i, fast = 1;
431#if defined(__BORLANDC__) || defined(_MSC_VER) 686 HE *hes_stack [STACK_HES];
432 HE **hes = _alloca (count * sizeof (HE)); 687 HE **hes = hes_stack;
433#else 688
434 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 689 // allocate larger arrays on the heap
435#endif 690 if (count > STACK_HES)
691 {
692 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
693 hes = (HE **)SvPVX (sv);
694 }
436 695
437 i = 0; 696 i = 0;
438 while ((he = hv_iternext (hv))) 697 while ((he = hv_iternext (hv)))
439 { 698 {
440 hes [i++] = he; 699 hes [i++] = he;
… …
509// encode objects, arrays and special \0=false and \1=true values. 768// encode objects, arrays and special \0=false and \1=true values.
510static void 769static void
511encode_rv (enc_t *enc, SV *sv) 770encode_rv (enc_t *enc, SV *sv)
512{ 771{
513 svtype svt; 772 svtype svt;
773 GV *method;
514 774
515 SvGETMAGIC (sv); 775 SvGETMAGIC (sv);
516 svt = SvTYPE (sv); 776 svt = SvTYPE (sv);
517 777
518 if (expect_false (SvOBJECT (sv))) 778 if (expect_false (SvOBJECT (sv)))
519 { 779 {
520 HV *stash = !JSON_SLOW || json_boolean_stash 780 HV *stash = SvSTASH (sv);
521 ? json_boolean_stash
522 : gv_stashpv ("JSON::XS::Boolean", 1);
523 781
524 if (SvSTASH (sv) == stash) 782 if (stash == bool_stash)
525 { 783 {
526 if (SvIV (sv))
527 encode_str (enc, "true", 4, 0); 784 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
528 else
529 encode_str (enc, "false", 5, 0); 785 else encode_str (enc, "false", 5, 0);
530 } 786 }
787 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
788 {
789 int count;
790 dSP;
791
792 ENTER; SAVETMPS;
793 PUSHMARK (SP);
794 EXTEND (SP, 2);
795 // we re-bless the reference to get overload and other niceties right
796 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
797 PUSHs (sv_json);
798
799 PUTBACK;
800 count = call_sv ((SV *)GvCV (method), G_ARRAY);
801 SPAGAIN;
802
803 // catch this surprisingly common error
804 if (SvROK (TOPs) && SvRV (TOPs) == sv)
805 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
806
807 encode_ch (enc, '(');
808 encode_ch (enc, '"');
809 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
810 encode_ch (enc, '"');
811 encode_ch (enc, ')');
812 encode_ch (enc, '[');
813
814 if (count)
815 {
816 int i;
817
818 for (i = 0; i < count - 1; ++i)
819 {
820 encode_sv (enc, SP[i + 1 - count]);
821 encode_ch (enc, ',');
822 }
823
824 encode_sv (enc, TOPs);
825 SP -= count;
826 }
827
828 encode_ch (enc, ']');
829
830 FREETMPS; LEAVE;
831 }
832 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
833 {
834 dSP;
835
836 ENTER; SAVETMPS;
837 PUSHMARK (SP);
838 // we re-bless the reference to get overload and other niceties right
839 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
840
841 // calling with G_SCALAR ensures that we always get a 1 return value
842 PUTBACK;
843 call_sv ((SV *)GvCV (method), G_SCALAR);
844 SPAGAIN;
845
846 // catch this surprisingly common error
847 if (SvROK (TOPs) && SvRV (TOPs) == sv)
848 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
849
850 sv = POPs;
851 PUTBACK;
852
853 encode_sv (enc, sv);
854
855 FREETMPS; LEAVE;
856 }
857 else if (enc->json.flags & F_ALLOW_BLESSED)
858 encode_str (enc, "null", 4, 0);
531 else 859 else
532 {
533#if 0
534 if (0 && sv_derived_from (rv, "JSON::Literal"))
535 {
536 // not yet
537 }
538#endif
539 if (enc->json.flags & F_CONV_BLESSED)
540 {
541 // we re-bless the reference to get overload and other niceties right
542 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
543
544 if (to_json)
545 {
546 dSP;
547
548 ENTER; SAVETMPS; PUSHMARK (SP);
549 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
550
551 // calling with G_SCALAR ensures that we always get a 1 return value
552 PUTBACK;
553 call_sv ((SV *)GvCV (to_json), G_SCALAR);
554 SPAGAIN;
555
556 // catch this surprisingly common error
557 if (SvROK (TOPs) && SvRV (TOPs) == sv)
558 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
559
560 sv = POPs;
561 PUTBACK;
562
563 encode_sv (enc, sv);
564
565 FREETMPS; LEAVE;
566 }
567 else if (enc->json.flags & F_ALLOW_BLESSED)
568 encode_str (enc, "null", 4, 0);
569 else
570 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
571 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
572 }
573 else if (enc->json.flags & F_ALLOW_BLESSED)
574 encode_str (enc, "null", 4, 0);
575 else
576 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 860 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
577 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 861 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
578 }
579 } 862 }
580 else if (svt == SVt_PVHV) 863 else if (svt == SVt_PVHV)
581 encode_hv (enc, (HV *)sv); 864 encode_hv (enc, (HV *)sv);
582 else if (svt == SVt_PVAV) 865 else if (svt == SVt_PVAV)
583 encode_av (enc, (AV *)sv); 866 encode_av (enc, (AV *)sv);
584 else if (svt < SVt_PVAV) 867 else if (svt < SVt_PVAV)
585 { 868 {
586 STRLEN len = 0; 869 int bool_type = ref_bool_type (sv);
587 char *pv = svt ? SvPV (sv, len) : 0;
588 870
589 if (len == 1 && *pv == '1') 871 if (bool_type == 1)
590 encode_str (enc, "true", 4, 0); 872 encode_str (enc, "true", 4, 0);
591 else if (len == 1 && *pv == '0') 873 else if (bool_type == 0)
592 encode_str (enc, "false", 5, 0); 874 encode_str (enc, "false", 5, 0);
875 else if (enc->json.flags & F_ALLOW_UNKNOWN)
876 encode_str (enc, "null", 4, 0);
593 else 877 else
594 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 878 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
595 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 879 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
596 } 880 }
881 else if (enc->json.flags & F_ALLOW_UNKNOWN)
882 encode_str (enc, "null", 4, 0);
597 else 883 else
598 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 884 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
599 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 885 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
600} 886}
601 887
… …
619 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 905 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
620 enc->cur += strlen (enc->cur); 906 enc->cur += strlen (enc->cur);
621 } 907 }
622 else if (SvIOKp (sv)) 908 else if (SvIOKp (sv))
623 { 909 {
624 // we assume we can always read an IV as a UV 910 // we assume we can always read an IV as a UV and vice versa
625 if (SvUV (sv) & ~(UV)0x7fff) 911 // we assume two's complement
626 { 912 // we assume no aliasing issues in the union
627 // large integer, use the (rather slow) snprintf way. 913 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
628 need (enc, sizeof (UV) * 3); 914 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
629 enc->cur +=
630 SvIsUV(sv)
631 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
632 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
633 }
634 else
635 { 915 {
636 // optimise the "small number case" 916 // optimise the "small number case"
637 // code will likely be branchless and use only a single multiplication 917 // code will likely be branchless and use only a single multiplication
918 // works for numbers up to 59074
638 I32 i = SvIV (sv); 919 I32 i = SvIVX (sv);
639 U32 u; 920 U32 u;
640 char digit, nz = 0; 921 char digit, nz = 0;
641 922
642 need (enc, 6); 923 need (enc, 6);
643 924
… …
649 930
650 // now output digit by digit, each time masking out the integer part 931 // now output digit by digit, each time masking out the integer part
651 // and multiplying by 5 while moving the decimal point one to the right, 932 // and multiplying by 5 while moving the decimal point one to the right,
652 // resulting in a net multiplication by 10. 933 // resulting in a net multiplication by 10.
653 // we always write the digit to memory but conditionally increment 934 // we always write the digit to memory but conditionally increment
654 // the pointer, to ease the usage of conditional move instructions. 935 // the pointer, to enable the use of conditional move instructions.
655 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 936 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
656 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 937 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
657 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 938 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
658 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 939 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
659 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 940 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
660 } 941 }
942 else
943 {
944 // large integer, use the (rather slow) snprintf way.
945 need (enc, IVUV_MAXCHARS);
946 enc->cur +=
947 SvIsUV(sv)
948 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
949 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
950 }
661 } 951 }
662 else if (SvROK (sv)) 952 else if (SvROK (sv))
663 encode_rv (enc, SvRV (sv)); 953 encode_rv (enc, SvRV (sv));
664 else if (!SvOK (sv)) 954 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
665 encode_str (enc, "null", 4, 0); 955 encode_str (enc, "null", 4, 0);
666 else 956 else
667 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 957 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
668 SvPV_nolen (sv), SvFLAGS (sv)); 958 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
669} 959}
670 960
671static SV * 961static SV *
672encode_json (SV *scalar, JSON *json) 962encode_json (SV *scalar, JSON *json)
673{ 963{
674 enc_t enc; 964 enc_t enc;
675 965
676 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 966 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
677 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 967 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
678 968
679 enc.json = *json; 969 enc.json = *json;
680 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 970 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
681 enc.cur = SvPVX (enc.sv); 971 enc.cur = SvPVX (enc.sv);
682 enc.end = SvEND (enc.sv); 972 enc.end = SvEND (enc.sv);
683 enc.indent = 0; 973 enc.indent = 0;
684 enc.maxdepth = DEC_DEPTH (enc.json.flags); 974 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
975 : enc.json.flags & F_LATIN1 ? 0x000100UL
976 : 0x110000UL;
685 977
686 SvPOK_only (enc.sv); 978 SvPOK_only (enc.sv);
687 encode_sv (&enc, scalar); 979 encode_sv (&enc, scalar);
980 encode_nl (&enc);
688 981
689 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 982 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
690 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 983 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
691 984
692 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 985 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
710 JSON json; 1003 JSON json;
711 U32 depth; // recursion depth 1004 U32 depth; // recursion depth
712 U32 maxdepth; // recursion depth limit 1005 U32 maxdepth; // recursion depth limit
713} dec_t; 1006} dec_t;
714 1007
715inline void 1008INLINE void
716decode_comment (dec_t *dec) 1009decode_comment (dec_t *dec)
717{ 1010{
718 // only '#'-style comments allowed a.t.m. 1011 // only '#'-style comments allowed a.t.m.
719 1012
720 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) 1013 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
721 ++dec->cur; 1014 ++dec->cur;
722} 1015}
723 1016
724inline void 1017INLINE void
725decode_ws (dec_t *dec) 1018decode_ws (dec_t *dec)
726{ 1019{
727 for (;;) 1020 for (;;)
728 { 1021 {
729 char ch = *dec->cur; 1022 char ch = *dec->cur;
… …
753 if (*dec->cur != ch) \ 1046 if (*dec->cur != ch) \
754 ERR (# ch " expected"); \ 1047 ERR (# ch " expected"); \
755 ++dec->cur; \ 1048 ++dec->cur; \
756 SE 1049 SE
757 1050
758#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 1051#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
759#define DEC_DEC_DEPTH --dec->depth 1052#define DEC_DEC_DEPTH --dec->depth
760 1053
761static SV *decode_sv (dec_t *dec); 1054static SV *decode_sv (dec_t *dec);
762 1055
763static signed char decode_hexdigit[256]; 1056static signed char decode_hexdigit[256];
… …
855 1148
856 if (hi >= 0x80) 1149 if (hi >= 0x80)
857 { 1150 {
858 utf8 = 1; 1151 utf8 = 1;
859 1152
860 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1153 cur = encode_utf8 (cur, hi);
861 } 1154 }
862 else 1155 else
863 *cur++ = hi; 1156 *cur++ = hi;
864 } 1157 }
865 break; 1158 break;
… …
867 default: 1160 default:
868 --dec_cur; 1161 --dec_cur;
869 ERR ("illegal backslash escape sequence in string"); 1162 ERR ("illegal backslash escape sequence in string");
870 } 1163 }
871 } 1164 }
872 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1165 else if (expect_true (ch >= 0x20 && ch < 0x80))
873 *cur++ = ch; 1166 *cur++ = ch;
874 else if (ch >= 0x80) 1167 else if (ch >= 0x80)
875 { 1168 {
876 STRLEN clen; 1169 STRLEN clen;
877 UV uch;
878 1170
879 --dec_cur; 1171 --dec_cur;
880 1172
881 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1173 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
882 if (clen == (STRLEN)-1) 1174 if (clen == (STRLEN)-1)
883 ERR ("malformed UTF-8 character in JSON string"); 1175 ERR ("malformed UTF-8 character in JSON string");
884 1176
885 do 1177 do
886 *cur++ = *dec_cur++; 1178 *cur++ = *dec_cur++;
887 while (--clen); 1179 while (--clen);
888 1180
889 utf8 = 1; 1181 utf8 = 1;
890 } 1182 }
1183 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1184 *cur++ = ch;
891 else 1185 else
892 { 1186 {
893 --dec_cur; 1187 --dec_cur;
894 1188
895 if (!ch) 1189 if (!ch)
… …
903 { 1197 {
904 STRLEN len = cur - buf; 1198 STRLEN len = cur - buf;
905 1199
906 if (sv) 1200 if (sv)
907 { 1201 {
908 SvGROW (sv, SvCUR (sv) + len + 1); 1202 STRLEN cur = SvCUR (sv);
1203
1204 if (SvLEN (sv) - cur <= len)
1205 json_sv_grow (sv, cur, len);
1206
909 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1207 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
910 SvCUR_set (sv, SvCUR (sv) + len); 1208 SvCUR_set (sv, SvCUR (sv) + len);
911 } 1209 }
912 else 1210 else
913 sv = newSVpvn (buf, len); 1211 sv = newSVpvn (buf, len);
… …
1004 1302
1005 // special case the rather common 1..5-digit-int case 1303 // special case the rather common 1..5-digit-int case
1006 if (*start == '-') 1304 if (*start == '-')
1007 switch (len) 1305 switch (len)
1008 { 1306 {
1009 case 2: return newSViv (-( start [1] - '0' * 1)); 1307 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1010 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1308 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1011 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1309 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1012 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1310 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1013 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1311 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1014 } 1312 }
1015 else 1313 else
1016 switch (len) 1314 switch (len)
1017 { 1315 {
1018 case 1: return newSViv ( start [0] - '0' * 1); 1316 case 1: return newSViv ( start [0] - '0' * 1);
1019 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1317 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1020 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1318 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1021 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1319 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1022 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1320 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1023 } 1321 }
1024 1322
1025 { 1323 {
1026 UV uv; 1324 UV uv;
1027 int numtype = grok_number (start, len, &uv); 1325 int numtype = grok_number (start, len, &uv);
… …
1036 } 1334 }
1037 1335
1038 len -= *start == '-' ? 1 : 0; 1336 len -= *start == '-' ? 1 : 0;
1039 1337
1040 // does not fit into IV or UV, try NV 1338 // does not fit into IV or UV, try NV
1041 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1339 if (len <= NV_DIG)
1042 #if defined (LDBL_DIG)
1043 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1044 #endif
1045 )
1046 // fits into NV without loss of precision 1340 // fits into NV without loss of precision
1047 return newSVnv (Atof (start)); 1341 return newSVnv (json_atof (start));
1048 1342
1049 // everything else fails, convert it to a string 1343 // everything else fails, convert it to a string
1050 return newSVpvn (start, dec->cur - start); 1344 return newSVpvn (start, dec->cur - start);
1051 } 1345 }
1052 1346
1053 // loss of precision here 1347 // loss of precision here
1054 return newSVnv (Atof (start)); 1348 return newSVnv (json_atof (start));
1055 1349
1056fail: 1350fail:
1057 return 0; 1351 return 0;
1058} 1352}
1059 1353
… …
1083 if (*dec->cur == ']') 1377 if (*dec->cur == ']')
1084 { 1378 {
1085 ++dec->cur; 1379 ++dec->cur;
1086 break; 1380 break;
1087 } 1381 }
1088 1382
1089 if (*dec->cur != ',') 1383 if (*dec->cur != ',')
1090 ERR (", or ] expected while parsing array"); 1384 ERR (", or ] expected while parsing array");
1091 1385
1092 ++dec->cur; 1386 ++dec->cur;
1093 1387
… …
1135 char *p = dec->cur; 1429 char *p = dec->cur;
1136 char *e = p + 24; // only try up to 24 bytes 1430 char *e = p + 24; // only try up to 24 bytes
1137 1431
1138 for (;;) 1432 for (;;)
1139 { 1433 {
1140 // the >= 0x80 is true on most architectures 1434 // the >= 0x80 is false on most architectures
1141 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1435 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1142 { 1436 {
1143 // slow path, back up and use decode_str 1437 // slow path, back up and use decode_str
1144 SV *key = decode_str (dec); 1438 SV *key = decode_str (dec);
1145 if (!key) 1439 if (!key)
… …
1217 1511
1218 hv_iterinit (hv); 1512 hv_iterinit (hv);
1219 he = hv_iternext (hv); 1513 he = hv_iternext (hv);
1220 hv_iterinit (hv); 1514 hv_iterinit (hv);
1221 1515
1222 // the next line creates a mortal sv each time its called. 1516 // the next line creates a mortal sv each time it's called.
1223 // might want to optimise this for common cases. 1517 // might want to optimise this for common cases.
1224 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1518 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1225 1519
1226 if (cb) 1520 if (cb)
1227 { 1521 {
1228 dSP; 1522 dSP;
1229 int count; 1523 int count;
1230 1524
1231 ENTER; SAVETMPS; PUSHMARK (SP); 1525 ENTER; SAVETMPS;
1526 PUSHMARK (SP);
1232 XPUSHs (HeVAL (he)); 1527 XPUSHs (HeVAL (he));
1528 sv_2mortal (sv);
1233 1529
1234 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1530 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1235 1531
1236 if (count == 1) 1532 if (count == 1)
1237 { 1533 {
1238 sv = newSVsv (POPs); 1534 sv = newSVsv (POPs);
1239 FREETMPS; LEAVE; 1535 FREETMPS; LEAVE;
1240 return sv; 1536 return sv;
1241 } 1537 }
1538 else if (count)
1539 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1242 1540
1541 SvREFCNT_inc (sv);
1243 FREETMPS; LEAVE; 1542 FREETMPS; LEAVE;
1244 } 1543 }
1245 } 1544 }
1246 1545
1247 if (dec->json.cb_object) 1546 if (dec->json.cb_object)
1248 { 1547 {
1249 dSP; 1548 dSP;
1250 int count; 1549 int count;
1251 1550
1252 ENTER; SAVETMPS; PUSHMARK (SP); 1551 ENTER; SAVETMPS;
1552 PUSHMARK (SP);
1253 XPUSHs (sv_2mortal (sv)); 1553 XPUSHs (sv_2mortal (sv));
1254 1554
1255 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1555 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1256 1556
1257 if (count == 1) 1557 if (count == 1)
1258 { 1558 {
1259 sv = newSVsv (POPs); 1559 sv = newSVsv (POPs);
1260 FREETMPS; LEAVE; 1560 FREETMPS; LEAVE;
1261 return sv; 1561 return sv;
1262 } 1562 }
1563 else if (count)
1564 croak ("filter_json_object callbacks must not return more than one scalar");
1263 1565
1264 SvREFCNT_inc (sv); 1566 SvREFCNT_inc (sv);
1265 FREETMPS; LEAVE; 1567 FREETMPS; LEAVE;
1266 } 1568 }
1267 } 1569 }
… …
1273 DEC_DEC_DEPTH; 1575 DEC_DEC_DEPTH;
1274 return 0; 1576 return 0;
1275} 1577}
1276 1578
1277static SV * 1579static SV *
1580decode_tag (dec_t *dec)
1581{
1582 SV *tag = 0;
1583 SV *val = 0;
1584
1585 if (!(dec->json.flags & F_ALLOW_TAGS))
1586 ERR ("malformed JSON string, neither array, object, number, string or atom");
1587
1588 ++dec->cur;
1589
1590 decode_ws (dec);
1591
1592 tag = decode_sv (dec);
1593 if (!tag)
1594 goto fail;
1595
1596 if (!SvPOK (tag))
1597 ERR ("malformed JSON string, (tag) must be a string");
1598
1599 decode_ws (dec);
1600
1601 if (*dec->cur != ')')
1602 ERR (") expected after tag");
1603
1604 ++dec->cur;
1605
1606 decode_ws (dec);
1607
1608 val = decode_sv (dec);
1609 if (!val)
1610 goto fail;
1611
1612 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1613 ERR ("malformed JSON string, tag value must be an array");
1614
1615 {
1616 AV *av = (AV *)SvRV (val);
1617 int i, len = av_len (av) + 1;
1618 HV *stash = gv_stashsv (tag, 0);
1619 SV *sv;
1620
1621 if (!stash)
1622 ERR ("cannot decode perl-object (package does not exist)");
1623
1624 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1625
1626 if (!method)
1627 ERR ("cannot decode perl-object (package does not have a THAW method)");
1628
1629 dSP;
1630
1631 ENTER; SAVETMPS;
1632 PUSHMARK (SP);
1633 EXTEND (SP, len + 2);
1634 // we re-bless the reference to get overload and other niceties right
1635 PUSHs (tag);
1636 PUSHs (sv_json);
1637
1638 for (i = 0; i < len; ++i)
1639 PUSHs (*av_fetch (av, i, 1));
1640
1641 PUTBACK;
1642 call_sv ((SV *)GvCV (method), G_SCALAR);
1643 SPAGAIN;
1644
1645 SvREFCNT_dec (tag);
1646 SvREFCNT_dec (val);
1647 sv = SvREFCNT_inc (POPs);
1648
1649 PUTBACK;
1650
1651 FREETMPS; LEAVE;
1652
1653 return sv;
1654 }
1655
1656fail:
1657 SvREFCNT_dec (tag);
1658 SvREFCNT_dec (val);
1659 return 0;
1660}
1661
1662static SV *
1278decode_sv (dec_t *dec) 1663decode_sv (dec_t *dec)
1279{ 1664{
1280 // the beauty of JSON: you need exactly one character lookahead 1665 // the beauty of JSON: you need exactly one character lookahead
1281 // to parse anything. 1666 // to parse everything.
1282 switch (*dec->cur) 1667 switch (*dec->cur)
1283 { 1668 {
1284 case '"': ++dec->cur; return decode_str (dec); 1669 case '"': ++dec->cur; return decode_str (dec);
1285 case '[': ++dec->cur; return decode_av (dec); 1670 case '[': ++dec->cur; return decode_av (dec);
1286 case '{': ++dec->cur; return decode_hv (dec); 1671 case '{': ++dec->cur; return decode_hv (dec);
1672 case '(': return decode_tag (dec);
1287 1673
1288 case '-': 1674 case '-':
1289 case '0': case '1': case '2': case '3': case '4': 1675 case '0': case '1': case '2': case '3': case '4':
1290 case '5': case '6': case '7': case '8': case '9': 1676 case '5': case '6': case '7': case '8': case '9':
1291 return decode_num (dec); 1677 return decode_num (dec);
1292 1678
1679 case 'f':
1680 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1681 {
1682 dec->cur += 5;
1683
1684 if (expect_false (!dec->json.v_false))
1685 dec->json.v_false = GET_BOOL (false);
1686
1687 return newSVsv (dec->json.v_false);
1688 }
1689 else
1690 ERR ("'false' expected");
1691
1692 break;
1693
1293 case 't': 1694 case 't':
1294 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1695 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1295 { 1696 {
1296 dec->cur += 4; 1697 dec->cur += 4;
1297#if JSON_SLOW 1698
1298 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1699 if (expect_false (!dec->json.v_true))
1299#endif 1700 dec->json.v_true = GET_BOOL (true);
1300 return SvREFCNT_inc (json_true); 1701
1702 return newSVsv (dec->json.v_true);
1301 } 1703 }
1302 else 1704 else
1303 ERR ("'true' expected"); 1705 ERR ("'true' expected");
1304
1305 break;
1306
1307 case 'f':
1308 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1309 {
1310 dec->cur += 5;
1311#if JSON_SLOW
1312 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1313#endif
1314 return SvREFCNT_inc (json_false);
1315 }
1316 else
1317 ERR ("'false' expected");
1318 1706
1319 break; 1707 break;
1320 1708
1321 case 'n': 1709 case 'n':
1322 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1710 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1328 ERR ("'null' expected"); 1716 ERR ("'null' expected");
1329 1717
1330 break; 1718 break;
1331 1719
1332 default: 1720 default:
1333 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1721 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1334 break; 1722 break;
1335 } 1723 }
1336 1724
1337fail: 1725fail:
1338 return 0; 1726 return 0;
1339} 1727}
1340 1728
1341static SV * 1729static SV *
1342decode_json (SV *string, JSON *json, UV *offset_return) 1730decode_json (SV *string, JSON *json, STRLEN *offset_return)
1343{ 1731{
1344 dec_t dec; 1732 dec_t dec;
1345 UV offset;
1346 SV *sv; 1733 SV *sv;
1347 1734
1735 /* work around bugs in 5.10 where manipulating magic values
1736 * makes perl ignore the magic in subsequent accesses.
1737 * also make a copy of non-PV values, to get them into a clean
1738 * state (SvPV should do that, but it's buggy, see below).
1739 *
1740 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1741 * as reported by Reini Urban.
1742 */
1348 SvGETMAGIC (string); 1743 /*SvGETMAGIC (string);*/
1744 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1745 string = sv_2mortal (newSVsv (string));
1746
1349 SvUPGRADE (string, SVt_PV); 1747 SvUPGRADE (string, SVt_PV);
1350 1748
1351 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1749 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1750 * assertion with -DDEBUGGING, although SvCUR is documented to
1751 * return the xpv_cur field which certainly exists after upgrading.
1752 * according to nicholas clark, calling SvPOK fixes this.
1753 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1754 * and hope for the best.
1755 * Damnit, SvPV_nolen still trips over yet another assertion. This
1756 * assertion business is seriously broken, try yet another workaround
1757 * for the broken -DDEBUGGING.
1758 */
1759 {
1760#ifdef DEBUGGING
1761 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1762#else
1763 STRLEN offset = SvCUR (string);
1764#endif
1765
1766 if (offset > json->max_size && json->max_size)
1352 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1767 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1353 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1768 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1769 }
1354 1770
1355 if (json->flags & F_UTF8) 1771 if (DECODE_WANTS_OCTETS (json))
1356 sv_utf8_downgrade (string, 0); 1772 sv_utf8_downgrade (string, 0);
1357 else 1773 else
1358 sv_utf8_upgrade (string); 1774 sv_utf8_upgrade (string);
1359 1775
1360 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1776 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1361 1777
1362 dec.json = *json; 1778 dec.json = *json;
1363 dec.cur = SvPVX (string); 1779 dec.cur = SvPVX (string);
1364 dec.end = SvEND (string); 1780 dec.end = SvEND (string);
1365 dec.err = 0; 1781 dec.err = 0;
1366 dec.depth = 0; 1782 dec.depth = 0;
1367 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1368 1783
1369 if (dec.json.cb_object || dec.json.cb_sk_object) 1784 if (dec.json.cb_object || dec.json.cb_sk_object)
1370 dec.json.flags |= F_HOOK; 1785 dec.json.flags |= F_HOOK;
1371 1786
1372 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1787 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1373 1788
1374 decode_ws (&dec); 1789 decode_ws (&dec);
1375 sv = decode_sv (&dec); 1790 sv = decode_sv (&dec);
1376 1791
1377 if (!(offset_return || !sv)) 1792 if (offset_return)
1793 *offset_return = dec.cur - SvPVX (string);
1794 else if (sv)
1378 { 1795 {
1379 // check for trailing garbage 1796 // check for trailing garbage
1380 decode_ws (&dec); 1797 decode_ws (&dec);
1381 1798
1382 if (*dec.cur) 1799 if (*dec.cur)
1383 { 1800 {
1384 dec.err = "garbage after JSON object"; 1801 dec.err = "garbage after JSON object";
1385 SvREFCNT_dec (sv); 1802 SvREFCNT_dec (sv);
1386 sv = 0; 1803 sv = 0;
1387 } 1804 }
1388 }
1389
1390 if (offset_return || !sv)
1391 {
1392 offset = dec.json.flags & F_UTF8
1393 ? dec.cur - SvPVX (string)
1394 : utf8_distance (dec.cur, SvPVX (string));
1395
1396 if (offset_return)
1397 *offset_return = offset;
1398 } 1805 }
1399 1806
1400 if (!sv) 1807 if (!sv)
1401 { 1808 {
1402 SV *uni = sv_newmortal (); 1809 SV *uni = sv_newmortal ();
… …
1408 SAVEVPTR (PL_curcop); 1815 SAVEVPTR (PL_curcop);
1409 PL_curcop = &cop; 1816 PL_curcop = &cop;
1410 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1817 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1411 LEAVE; 1818 LEAVE;
1412 1819
1413 croak ("%s, at character offset %d [\"%s\"]", 1820 croak ("%s, at character offset %d (before \"%s\")",
1414 dec.err, 1821 dec.err,
1415 (int)offset, 1822 (int)ptr_to_index (string, dec.cur),
1416 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1823 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1417 } 1824 }
1418 1825
1419 sv = sv_2mortal (sv); 1826 sv = sv_2mortal (sv);
1420 1827
1421 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1828 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1422 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1829 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1423 1830
1424 return sv; 1831 return sv;
1832}
1833
1834/////////////////////////////////////////////////////////////////////////////
1835// incremental parser
1836
1837static void
1838incr_parse (JSON *self)
1839{
1840 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1841
1842 // the state machine here is a bit convoluted and could be simplified a lot
1843 // but this would make it slower, so...
1844
1845 for (;;)
1846 {
1847 //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
1848 switch (self->incr_mode)
1849 {
1850 // only used for initial whitespace skipping
1851 case INCR_M_WS:
1852 for (;;)
1853 {
1854 if (*p > 0x20)
1855 {
1856 if (*p == '#')
1857 {
1858 self->incr_mode = INCR_M_C0;
1859 goto incr_m_c;
1860 }
1861 else
1862 {
1863 self->incr_mode = INCR_M_JSON;
1864 goto incr_m_json;
1865 }
1866 }
1867 else if (!*p)
1868 goto interrupt;
1869
1870 ++p;
1871 }
1872
1873 // skip a single char inside a string (for \\-processing)
1874 case INCR_M_BS:
1875 if (!*p)
1876 goto interrupt;
1877
1878 ++p;
1879 self->incr_mode = INCR_M_STR;
1880 goto incr_m_str;
1881
1882 // inside #-style comments
1883 case INCR_M_C0:
1884 case INCR_M_C1:
1885 incr_m_c:
1886 for (;;)
1887 {
1888 if (*p == '\n')
1889 {
1890 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1891 break;
1892 }
1893 else if (!*p)
1894 goto interrupt;
1895
1896 ++p;
1897 }
1898
1899 break;
1900
1901 // inside a string
1902 case INCR_M_STR:
1903 incr_m_str:
1904 for (;;)
1905 {
1906 if (*p == '"')
1907 {
1908 ++p;
1909 self->incr_mode = INCR_M_JSON;
1910
1911 if (!self->incr_nest)
1912 goto interrupt;
1913
1914 goto incr_m_json;
1915 }
1916 else if (*p == '\\')
1917 {
1918 ++p; // "virtually" consumes character after \
1919
1920 if (!*p) // if at end of string we have to switch modes
1921 {
1922 self->incr_mode = INCR_M_BS;
1923 goto interrupt;
1924 }
1925 }
1926 else if (!*p)
1927 goto interrupt;
1928
1929 ++p;
1930 }
1931
1932 // after initial ws, outside string
1933 case INCR_M_JSON:
1934 incr_m_json:
1935 for (;;)
1936 {
1937 switch (*p++)
1938 {
1939 case 0:
1940 --p;
1941 goto interrupt;
1942
1943 case 0x09:
1944 case 0x0a:
1945 case 0x0d:
1946 case 0x20:
1947 if (!self->incr_nest)
1948 {
1949 --p; // do not eat the whitespace, let the next round do it
1950 goto interrupt;
1951 }
1952 break;
1953
1954 case '"':
1955 self->incr_mode = INCR_M_STR;
1956 goto incr_m_str;
1957
1958 case '[':
1959 case '{':
1960 case '(':
1961 if (++self->incr_nest > self->max_depth)
1962 croak (ERR_NESTING_EXCEEDED);
1963 break;
1964
1965 case ']':
1966 case '}':
1967 if (--self->incr_nest <= 0)
1968 goto interrupt;
1969 break;
1970
1971 case ')':
1972 --self->incr_nest;
1973 break;
1974
1975 case '#':
1976 self->incr_mode = INCR_M_C1;
1977 goto incr_m_c;
1978 }
1979 }
1980 }
1981
1982 modechange:
1983 ;
1984 }
1985
1986interrupt:
1987 self->incr_pos = p - SvPVX (self->incr_text);
1988 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1989 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1425} 1990}
1426 1991
1427///////////////////////////////////////////////////////////////////////////// 1992/////////////////////////////////////////////////////////////////////////////
1428// XS interface functions 1993// XS interface functions
1429 1994
… …
1438 i >= '0' && i <= '9' ? i - '0' 2003 i >= '0' && i <= '9' ? i - '0'
1439 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2004 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1440 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2005 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1441 : -1; 2006 : -1;
1442 2007
1443 json_stash = gv_stashpv ("JSON::XS" , 1); 2008 json_stash = gv_stashpv ("JSON::XS" , 1);
1444 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2009 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
2010 bool_false = get_bool ("Types::Serialiser::false");
2011 bool_true = get_bool ("Types::Serialiser::true");
1445 2012
1446 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 2013 sv_json = newSVpv ("JSON", 0);
1447 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 2014 SvREADONLY_on (sv_json);
2015
2016 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1448} 2017}
1449 2018
1450PROTOTYPES: DISABLE 2019PROTOTYPES: DISABLE
1451 2020
1452void CLONE (...) 2021void CLONE (...)
1453 CODE: 2022 CODE:
2023 // as long as these writes are atomic, the race should not matter
2024 // as existing threads either already use 0, or use the old value,
2025 // which is sitll correct for the initial thread.
1454 json_stash = 0; 2026 json_stash = 0;
1455 json_boolean_stash = 0; 2027 bool_stash = 0;
2028 bool_false = 0;
2029 bool_true = 0;
1456 2030
1457void new (char *klass) 2031void new (char *klass)
1458 PPCODE: 2032 PPCODE:
1459{ 2033{
1460 SV *pv = NEWSV (0, sizeof (JSON)); 2034 SV *pv = NEWSV (0, sizeof (JSON));
1461 SvPOK_only (pv); 2035 SvPOK_only (pv);
1462 Zero (SvPVX (pv), 1, JSON); 2036 json_init ((JSON *)SvPVX (pv));
1463 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1464 XPUSHs (sv_2mortal (sv_bless ( 2037 XPUSHs (sv_2mortal (sv_bless (
1465 newRV_noinc (pv), 2038 newRV_noinc (pv),
1466 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2039 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1467 ))); 2040 )));
1468} 2041}
2042
2043void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2044 PPCODE:
2045 self->v_false = newSVsv (v_false);
2046 self->v_true = newSVsv (v_true);
2047 XPUSHs (ST (0));
2048
2049void get_boolean_values (JSON *self)
2050 PPCODE:
2051 if (self->v_false && self->v_true)
2052 {
2053 EXTEND (SP, 2);
2054 PUSHs (self->v_false);
2055 PUSHs (self->v_true);
2056 }
1469 2057
1470void ascii (JSON *self, int enable = 1) 2058void ascii (JSON *self, int enable = 1)
1471 ALIAS: 2059 ALIAS:
1472 ascii = F_ASCII 2060 ascii = F_ASCII
1473 latin1 = F_LATIN1 2061 latin1 = F_LATIN1
… …
1480 allow_nonref = F_ALLOW_NONREF 2068 allow_nonref = F_ALLOW_NONREF
1481 shrink = F_SHRINK 2069 shrink = F_SHRINK
1482 allow_blessed = F_ALLOW_BLESSED 2070 allow_blessed = F_ALLOW_BLESSED
1483 convert_blessed = F_CONV_BLESSED 2071 convert_blessed = F_CONV_BLESSED
1484 relaxed = F_RELAXED 2072 relaxed = F_RELAXED
2073 allow_unknown = F_ALLOW_UNKNOWN
2074 allow_tags = F_ALLOW_TAGS
1485 PPCODE: 2075 PPCODE:
1486{ 2076{
1487 if (enable) 2077 if (enable)
1488 self->flags |= ix; 2078 self->flags |= ix;
1489 else 2079 else
… …
1504 get_allow_nonref = F_ALLOW_NONREF 2094 get_allow_nonref = F_ALLOW_NONREF
1505 get_shrink = F_SHRINK 2095 get_shrink = F_SHRINK
1506 get_allow_blessed = F_ALLOW_BLESSED 2096 get_allow_blessed = F_ALLOW_BLESSED
1507 get_convert_blessed = F_CONV_BLESSED 2097 get_convert_blessed = F_CONV_BLESSED
1508 get_relaxed = F_RELAXED 2098 get_relaxed = F_RELAXED
2099 get_allow_unknown = F_ALLOW_UNKNOWN
2100 get_allow_tags = F_ALLOW_TAGS
1509 PPCODE: 2101 PPCODE:
1510 XPUSHs (boolSV (self->flags & ix)); 2102 XPUSHs (boolSV (self->flags & ix));
1511 2103
1512void max_depth (JSON *self, UV max_depth = 0x80000000UL) 2104void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1513 PPCODE: 2105 PPCODE:
1514{ 2106 self->max_depth = max_depth;
1515 UV log2 = 0;
1516
1517 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1518
1519 while ((1UL << log2) < max_depth)
1520 ++log2;
1521
1522 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1523
1524 XPUSHs (ST (0)); 2107 XPUSHs (ST (0));
1525}
1526 2108
1527U32 get_max_depth (JSON *self) 2109U32 get_max_depth (JSON *self)
1528 CODE: 2110 CODE:
1529 RETVAL = DEC_DEPTH (self->flags); 2111 RETVAL = self->max_depth;
1530 OUTPUT: 2112 OUTPUT:
1531 RETVAL 2113 RETVAL
1532 2114
1533void max_size (JSON *self, UV max_size = 0) 2115void max_size (JSON *self, U32 max_size = 0)
1534 PPCODE: 2116 PPCODE:
1535{ 2117 self->max_size = max_size;
1536 UV log2 = 0;
1537
1538 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1539 if (max_size == 1) max_size = 2;
1540
1541 while ((1UL << log2) < max_size)
1542 ++log2;
1543
1544 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1545
1546 XPUSHs (ST (0)); 2118 XPUSHs (ST (0));
1547}
1548 2119
1549int get_max_size (JSON *self) 2120int get_max_size (JSON *self)
1550 CODE: 2121 CODE:
1551 RETVAL = DEC_SIZE (self->flags); 2122 RETVAL = self->max_size;
1552 OUTPUT: 2123 OUTPUT:
1553 RETVAL 2124 RETVAL
1554 2125
1555void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 2126void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1556 PPCODE: 2127 PPCODE:
… …
1562} 2133}
1563 2134
1564void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2135void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1565 PPCODE: 2136 PPCODE:
1566{ 2137{
1567 if (!self->cb_sk_object) 2138 if (!self->cb_sk_object)
1568 self->cb_sk_object = newHV (); 2139 self->cb_sk_object = newHV ();
1569 2140
1570 if (SvOK (cb)) 2141 if (SvOK (cb))
1571 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2142 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1572 else 2143 else
… …
1583 XPUSHs (ST (0)); 2154 XPUSHs (ST (0));
1584} 2155}
1585 2156
1586void encode (JSON *self, SV *scalar) 2157void encode (JSON *self, SV *scalar)
1587 PPCODE: 2158 PPCODE:
1588 XPUSHs (encode_json (scalar, self)); 2159 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2160 XPUSHs (scalar);
1589 2161
1590void decode (JSON *self, SV *jsonstr) 2162void decode (JSON *self, SV *jsonstr)
1591 PPCODE: 2163 PPCODE:
1592 XPUSHs (decode_json (jsonstr, self, 0)); 2164 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2165 XPUSHs (jsonstr);
1593 2166
1594void decode_prefix (JSON *self, SV *jsonstr) 2167void decode_prefix (JSON *self, SV *jsonstr)
1595 PPCODE: 2168 PPCODE:
1596{ 2169{
1597 UV offset; 2170 SV *sv;
2171 STRLEN offset;
2172 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1598 EXTEND (SP, 2); 2173 EXTEND (SP, 2);
1599 PUSHs (decode_json (jsonstr, self, &offset)); 2174 PUSHs (sv);
1600 PUSHs (sv_2mortal (newSVuv (offset))); 2175 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2176}
2177
2178void incr_parse (JSON *self, SV *jsonstr = 0)
2179 PPCODE:
2180{
2181 if (!self->incr_text)
2182 self->incr_text = newSVpvn ("", 0);
2183
2184 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2185 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2186 if (DECODE_WANTS_OCTETS (self))
2187 {
2188 if (self->incr_pos)
2189 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2190 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2191
2192 sv_utf8_downgrade (self->incr_text, 0);
2193 }
2194 else
2195 {
2196 sv_utf8_upgrade (self->incr_text);
2197
2198 if (self->incr_pos)
2199 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2200 - (U8 *)SvPVX (self->incr_text);
2201 }
2202
2203 // append data, if any
2204 if (jsonstr)
2205 {
2206 /* make sure both strings have same encoding */
2207 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2208 if (SvUTF8 (jsonstr))
2209 sv_utf8_downgrade (jsonstr, 0);
2210 else
2211 sv_utf8_upgrade (jsonstr);
2212
2213 /* and then just blindly append */
2214 {
2215 STRLEN len;
2216 const char *str = SvPV (jsonstr, len);
2217 STRLEN cur = SvCUR (self->incr_text);
2218
2219 if (SvLEN (self->incr_text) - cur <= len)
2220 json_sv_grow (self->incr_text, cur, len);
2221
2222 Move (str, SvEND (self->incr_text), len, char);
2223 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2224 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2225 }
2226 }
2227
2228 if (GIMME_V != G_VOID)
2229 do
2230 {
2231 SV *sv;
2232 STRLEN offset;
2233
2234 if (!INCR_DONE (self))
2235 {
2236 incr_parse (self);
2237
2238 if (self->incr_pos > self->max_size && self->max_size)
2239 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2240 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2241
2242 if (!INCR_DONE (self))
2243 {
2244 // as an optimisation, do not accumulate white space in the incr buffer
2245 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2246 {
2247 self->incr_pos = 0;
2248 SvCUR_set (self->incr_text, 0);
2249 }
2250
2251 break;
2252 }
2253 }
2254
2255 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2256 XPUSHs (sv);
2257
2258 self->incr_pos -= offset;
2259 self->incr_nest = 0;
2260 self->incr_mode = 0;
2261
2262 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2263 }
2264 while (GIMME_V == G_ARRAY);
2265}
2266
2267SV *incr_text (JSON *self)
2268 ATTRS: lvalue
2269 CODE:
2270{
2271 if (self->incr_pos)
2272 croak ("incr_text can not be called when the incremental parser already started parsing");
2273
2274 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2275}
2276 OUTPUT:
2277 RETVAL
2278
2279void incr_skip (JSON *self)
2280 CODE:
2281{
2282 if (self->incr_pos)
2283 {
2284 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2285 self->incr_pos = 0;
2286 self->incr_nest = 0;
2287 self->incr_mode = 0;
2288 }
2289}
2290
2291void incr_reset (JSON *self)
2292 CODE:
2293{
2294 SvREFCNT_dec (self->incr_text);
2295 self->incr_text = 0;
2296 self->incr_pos = 0;
2297 self->incr_nest = 0;
2298 self->incr_mode = 0;
1601} 2299}
1602 2300
1603void DESTROY (JSON *self) 2301void DESTROY (JSON *self)
1604 CODE: 2302 CODE:
2303 SvREFCNT_dec (self->v_false);
2304 SvREFCNT_dec (self->v_true);
1605 SvREFCNT_dec (self->cb_sk_object); 2305 SvREFCNT_dec (self->cb_sk_object);
1606 SvREFCNT_dec (self->cb_object); 2306 SvREFCNT_dec (self->cb_object);
2307 SvREFCNT_dec (self->incr_text);
1607 2308
1608PROTOTYPES: ENABLE 2309PROTOTYPES: ENABLE
1609 2310
1610void encode_json (SV *scalar) 2311void encode_json (SV *scalar)
1611 PPCODE: 2312 PPCODE:
1612{ 2313{
1613 JSON json = { F_DEFAULT | F_UTF8 }; 2314 JSON json;
1614 XPUSHs (encode_json (scalar, &json)); 2315 json_init (&json);
2316 json.flags |= F_UTF8;
2317 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2318 XPUSHs (scalar);
1615} 2319}
1616 2320
1617void decode_json (SV *jsonstr) 2321void decode_json (SV *jsonstr)
1618 PPCODE: 2322 PPCODE:
1619{ 2323{
1620 JSON json = { F_DEFAULT | F_UTF8 }; 2324 JSON json;
1621 XPUSHs (decode_json (jsonstr, &json, 0)); 2325 json_init (&json);
2326 json.flags |= F_UTF8;
2327 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2328 XPUSHs (jsonstr);
1622} 2329}
1623 2330

Diff Legend

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