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 1.65 by root, Sat Oct 13 01:55:31 2007 UTC vs.
Revision 1.140 by root, Fri Apr 19 17:09:51 2019 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_TFN, // inside true/false/null
102 INCR_M_NUM, // inside number
103 INCR_M_STR, // inside string
104 INCR_M_BS, // inside backslash
105 INCR_M_C0, // inside comment in initial whitespace sequence
106 INCR_M_C1, // inside comment in other places
107 INCR_M_JSON // outside anything, count nesting
108};
109
110#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
75 111
76typedef struct { 112typedef struct {
77 U32 flags; 113 U32 flags;
114 U32 max_depth;
115 STRLEN max_size;
116
78 SV *cb_object; 117 SV *cb_object;
79 HV *cb_sk_object; 118 HV *cb_sk_object;
119
120 // for the incremental parser
121 SV *incr_text; // the source text so far
122 STRLEN incr_pos; // the current offset into the text
123 int incr_nest; // {[]}-nesting level
124 unsigned char incr_mode;
125
126 SV *v_false, *v_true;
80} JSON; 127} JSON;
128
129INLINE void
130json_init (JSON *json)
131{
132 static const JSON init = { F_ALLOW_NONREF, 512 };
133
134 *json = init;
135}
81 136
82///////////////////////////////////////////////////////////////////////////// 137/////////////////////////////////////////////////////////////////////////////
83// utility functions 138// utility functions
84 139
85inline void 140INLINE SV *
141get_bool (const char *name)
142{
143 SV *sv = get_sv (name, 1);
144
145 SvREADONLY_on (sv);
146 SvREADONLY_on (SvRV (sv));
147
148 return sv;
149}
150
151INLINE void
86shrink (SV *sv) 152shrink (SV *sv)
87{ 153{
88 sv_utf8_downgrade (sv, 1); 154 sv_utf8_downgrade (sv, 1);
155
89 if (SvLEN (sv) > SvCUR (sv) + 1) 156 if (SvLEN (sv) > SvCUR (sv) + 1)
90 { 157 {
91#ifdef SvPV_shrink_to_cur 158#ifdef SvPV_shrink_to_cur
92 SvPV_shrink_to_cur (sv); 159 SvPV_shrink_to_cur (sv);
93#elif defined (SvPV_renew) 160#elif defined (SvPV_renew)
94 SvPV_renew (sv, SvCUR (sv) + 1); 161 SvPV_renew (sv, SvCUR (sv) + 1);
95#endif 162#endif
96 } 163 }
97} 164}
98 165
166/* adds two STRLENs together, slow, and with paranoia */
167static STRLEN
168strlen_sum (STRLEN l1, STRLEN l2)
169{
170 size_t sum = l1 + l2;
171
172 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
173 croak ("JSON::XS: string size overflow");
174
175 return sum;
176}
177
178/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
179static char *
180json_sv_grow (SV *sv, size_t len1, size_t len2)
181{
182 len1 = strlen_sum (len1, len2);
183 len1 = strlen_sum (len1, len1 >> 1);
184
185 if (len1 > 4096 - 24)
186 len1 = (len1 | 4095) - 24;
187
188 return SvGROW (sv, len1);
189}
190
99// decode an utf-8 character and return it, or (UV)-1 in 191// decode a utf-8 character and return it, or (UV)-1 in
100// case of an error. 192// case of an error.
101// we special-case "safe" characters from U+80 .. U+7FF, 193// we special-case "safe" characters from U+80 .. U+7FF,
102// but use the very good perl function to parse anything else. 194// but use the very good perl function to parse anything else.
103// note that we never call this function for a ascii codepoints 195// note that we never call this function for a ascii codepoints
104inline UV 196INLINE UV
105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 197decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
106{ 198{
107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 199 if (expect_true (len >= 2
108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 200 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 201 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
110 { 202 {
111 *clen = 2; 203 *clen = 2;
112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 204 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
113 } 205 }
114 else 206 else
207 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
208}
209
210// likewise for encoding, also never called for ascii codepoints
211// this function takes advantage of this fact, although current gccs
212// seem to optimise the check for >= 0x80 away anyways
213INLINE unsigned char *
214encode_utf8 (unsigned char *s, UV ch)
215{
216 if (expect_false (ch < 0x000080))
217 *s++ = ch;
218 else if (expect_true (ch < 0x000800))
219 *s++ = 0xc0 | ( ch >> 6),
220 *s++ = 0x80 | ( ch & 0x3f);
221 else if ( ch < 0x010000)
222 *s++ = 0xe0 | ( ch >> 12),
223 *s++ = 0x80 | ((ch >> 6) & 0x3f),
224 *s++ = 0x80 | ( ch & 0x3f);
225 else if ( ch < 0x110000)
226 *s++ = 0xf0 | ( ch >> 18),
227 *s++ = 0x80 | ((ch >> 12) & 0x3f),
228 *s++ = 0x80 | ((ch >> 6) & 0x3f),
229 *s++ = 0x80 | ( ch & 0x3f);
230
231 return s;
232}
233
234// convert offset pointer to character index, sv must be string
235static STRLEN
236ptr_to_index (SV *sv, char *offset)
237{
238 return SvUTF8 (sv)
239 ? utf8_distance (offset, SvPVX (sv))
240 : offset - SvPVX (sv);
241}
242
243/////////////////////////////////////////////////////////////////////////////
244// fp hell
245
246// scan a group of digits, and a trailing exponent
247static void
248json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
249{
250 UV uaccum = 0;
251 int eaccum = 0;
252
253 // if we recurse too deep, skip all remaining digits
254 // to avoid a stack overflow attack
255 if (expect_false (--maxdepth <= 0))
256 while (((U8)*s - '0') < 10)
257 ++s;
258
259 for (;;)
260 {
261 U8 dig = (U8)*s - '0';
262
263 if (expect_false (dig >= 10))
264 {
265 if (dig == (U8)((U8)'.' - (U8)'0'))
266 {
267 ++s;
268 json_atof_scan1 (s, accum, expo, 1, maxdepth);
269 }
270 else if ((dig | ' ') == 'e' - '0')
271 {
272 int exp2 = 0;
273 int neg = 0;
274
275 ++s;
276
277 if (*s == '-')
278 {
279 ++s;
280 neg = 1;
281 }
282 else if (*s == '+')
283 ++s;
284
285 while ((dig = (U8)*s - '0') < 10)
286 exp2 = exp2 * 10 + *s++ - '0';
287
288 *expo += neg ? -exp2 : exp2;
289 }
290
291 break;
292 }
293
294 ++s;
295
296 uaccum = uaccum * 10 + dig;
297 ++eaccum;
298
299 // if we have too many digits, then recurse for more
300 // we actually do this for rather few digits
301 if (uaccum >= (UV_MAX - 9) / 10)
302 {
303 if (postdp) *expo -= eaccum;
304 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
305 if (postdp) *expo += eaccum;
306
307 break;
308 }
115 { 309 }
116 *clen = (STRLEN)-1; 310
311 // this relies greatly on the quality of the pow ()
312 // implementation of the platform, but a good
313 // implementation is hard to beat.
314 // (IEEE 754 conformant ones are required to be exact)
315 if (postdp) *expo -= eaccum;
316 *accum += uaccum * Perl_pow (10., *expo);
317 *expo += eaccum;
318}
319
320static NV
321json_atof (const char *s)
322{
323 NV accum = 0.;
324 int expo = 0;
325 int neg = 0;
326
327 if (*s == '-')
328 {
329 ++s;
330 neg = 1;
331 }
332
333 // a recursion depth of ten gives us >>500 bits
334 json_atof_scan1 (s, &accum, &expo, 0, 10);
335
336 return neg ? -accum : accum;
337}
338
339// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
340static int
341ref_bool_type (SV *sv)
342{
343 svtype svt = SvTYPE (sv);
344
345 if (svt < SVt_PVAV)
346 {
347 STRLEN len = 0;
348 char *pv = svt ? SvPV (sv, len) : 0;
349
350 if (len == 1)
351 if (*pv == '1')
352 return 1;
353 else if (*pv == '0')
354 return 0;
355 }
356
357 return -1;
358}
359
360// returns whether scalar is not a reference in the sense of allow_nonref
361static int
362json_nonref (SV *scalar)
363{
364 if (!SvROK (scalar))
365 return 1;
366
367 scalar = SvRV (scalar);
368
369 if (SvTYPE (scalar) >= SVt_PVMG)
370 {
371 if (SvSTASH (scalar) == bool_stash)
117 return (UV)-1; 372 return 1;
373
374 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
375 return 1;
118 } 376 }
377
378 return 0;
119} 379}
120 380
121///////////////////////////////////////////////////////////////////////////// 381/////////////////////////////////////////////////////////////////////////////
122// encoder 382// encoder
123 383
… …
127 char *cur; // SvPVX (sv) + current output position 387 char *cur; // SvPVX (sv) + current output position
128 char *end; // SvEND (sv) 388 char *end; // SvEND (sv)
129 SV *sv; // result scalar 389 SV *sv; // result scalar
130 JSON json; 390 JSON json;
131 U32 indent; // indentation level 391 U32 indent; // indentation level
132 U32 maxdepth; // max. indentation/recursion level 392 UV limit; // escape character values >= this value when encoding
133} enc_t; 393} enc_t;
134 394
135inline void 395INLINE void
136need (enc_t *enc, STRLEN len) 396need (enc_t *enc, STRLEN len)
137{ 397{
138 if (expect_false (enc->cur + len >= enc->end)) 398 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
139 { 399 {
140 STRLEN cur = enc->cur - SvPVX (enc->sv); 400 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
141 SvGROW (enc->sv, cur + len + 1); 401 char *buf = json_sv_grow (enc->sv, cur, len);
142 enc->cur = SvPVX (enc->sv) + cur; 402 enc->cur = buf + cur;
143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 403 enc->end = buf + SvLEN (enc->sv) - 1;
144 } 404 }
145} 405}
146 406
147inline void 407INLINE void
148encode_ch (enc_t *enc, char ch) 408encode_ch (enc_t *enc, char ch)
149{ 409{
150 need (enc, 1); 410 need (enc, 1);
151 *enc->cur++ = ch; 411 *enc->cur++ = ch;
152} 412}
… …
164 424
165 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 425 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
166 { 426 {
167 if (expect_false (ch == '"')) // but with slow exceptions 427 if (expect_false (ch == '"')) // but with slow exceptions
168 { 428 {
169 need (enc, len += 1); 429 need (enc, len + 1);
170 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
171 *enc->cur++ = '"'; 431 *enc->cur++ = '"';
172 } 432 }
173 else if (expect_false (ch == '\\')) 433 else if (expect_false (ch == '\\'))
174 { 434 {
175 need (enc, len += 1); 435 need (enc, len + 1);
176 *enc->cur++ = '\\'; 436 *enc->cur++ = '\\';
177 *enc->cur++ = '\\'; 437 *enc->cur++ = '\\';
178 } 438 }
179 else 439 else
180 *enc->cur++ = ch; 440 *enc->cur++ = ch;
… …
183 } 443 }
184 else 444 else
185 { 445 {
186 switch (ch) 446 switch (ch)
187 { 447 {
188 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 448 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; 449 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; 450 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; 451 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; 452 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
193 453
194 default: 454 default:
195 { 455 {
196 STRLEN clen; 456 STRLEN clen;
197 UV uch; 457 UV uch;
… …
206 { 466 {
207 uch = ch; 467 uch = ch;
208 clen = 1; 468 clen = 1;
209 } 469 }
210 470
211 if (uch > 0x10FFFFUL) 471 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 { 472 {
216 if (uch > 0xFFFFUL) 473 if (uch >= 0x10000UL)
217 { 474 {
475 if (uch >= 0x110000UL)
476 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
477
218 need (enc, len += 11); 478 need (enc, len + 11);
219 sprintf (enc->cur, "\\u%04x\\u%04x", 479 sprintf (enc->cur, "\\u%04x\\u%04x",
220 (int)((uch - 0x10000) / 0x400 + 0xD800), 480 (int)((uch - 0x10000) / 0x400 + 0xD800),
221 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 481 (int)((uch - 0x10000) % 0x400 + 0xDC00));
222 enc->cur += 12; 482 enc->cur += 12;
223 } 483 }
224 else 484 else
225 { 485 {
226 static char hexdigit [16] = "0123456789abcdef";
227 need (enc, len += 5); 486 need (enc, len + 5);
228 *enc->cur++ = '\\'; 487 *enc->cur++ = '\\';
229 *enc->cur++ = 'u'; 488 *enc->cur++ = 'u';
230 *enc->cur++ = hexdigit [ uch >> 12 ]; 489 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
231 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
232 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 491 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
233 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 492 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
234 } 493 }
235 494
236 str += clen; 495 str += clen;
237 } 496 }
238 else if (enc->json.flags & F_LATIN1) 497 else if (enc->json.flags & F_LATIN1)
… …
240 *enc->cur++ = uch; 499 *enc->cur++ = uch;
241 str += clen; 500 str += clen;
242 } 501 }
243 else if (is_utf8) 502 else if (is_utf8)
244 { 503 {
245 need (enc, len += clen); 504 need (enc, len + clen);
246 do 505 do
247 { 506 {
248 *enc->cur++ = *str++; 507 *enc->cur++ = *str++;
249 } 508 }
250 while (--clen); 509 while (--clen);
251 } 510 }
252 else 511 else
253 { 512 {
254 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 513 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
255 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 514 enc->cur = encode_utf8 (enc->cur, uch);
256 ++str; 515 ++str;
257 } 516 }
258 } 517 }
259 } 518 }
260 } 519 }
261 520
262 --len; 521 --len;
263 } 522 }
264} 523}
265 524
266inline void 525INLINE void
267encode_indent (enc_t *enc) 526encode_indent (enc_t *enc)
268{ 527{
269 if (enc->json.flags & F_INDENT) 528 if (enc->json.flags & F_INDENT)
270 { 529 {
271 int spaces = enc->indent * INDENT_STEP; 530 int spaces = enc->indent * INDENT_STEP;
… …
274 memset (enc->cur, ' ', spaces); 533 memset (enc->cur, ' ', spaces);
275 enc->cur += spaces; 534 enc->cur += spaces;
276 } 535 }
277} 536}
278 537
279inline void 538INLINE void
280encode_space (enc_t *enc) 539encode_space (enc_t *enc)
281{ 540{
282 need (enc, 1); 541 need (enc, 1);
283 encode_ch (enc, ' '); 542 encode_ch (enc, ' ');
284} 543}
285 544
286inline void 545INLINE void
287encode_nl (enc_t *enc) 546encode_nl (enc_t *enc)
288{ 547{
289 if (enc->json.flags & F_INDENT) 548 if (enc->json.flags & F_INDENT)
290 { 549 {
291 need (enc, 1); 550 need (enc, 1);
292 encode_ch (enc, '\n'); 551 encode_ch (enc, '\n');
293 } 552 }
294} 553}
295 554
296inline void 555INLINE void
297encode_comma (enc_t *enc) 556encode_comma (enc_t *enc)
298{ 557{
299 encode_ch (enc, ','); 558 encode_ch (enc, ',');
300 559
301 if (enc->json.flags & F_INDENT) 560 if (enc->json.flags & F_INDENT)
… …
309static void 568static void
310encode_av (enc_t *enc, AV *av) 569encode_av (enc_t *enc, AV *av)
311{ 570{
312 int i, len = av_len (av); 571 int i, len = av_len (av);
313 572
314 if (enc->indent >= enc->maxdepth) 573 if (enc->indent >= enc->json.max_depth)
315 croak ("data structure too deep (hit recursion limit)"); 574 croak (ERR_NESTING_EXCEEDED);
316 575
317 encode_ch (enc, '['); 576 encode_ch (enc, '[');
318 577
319 if (len >= 0) 578 if (len >= 0)
320 { 579 {
321 encode_nl (enc); ++enc->indent; 580 encode_nl (enc); ++enc->indent;
322 581
323 for (i = 0; i <= len; ++i) 582 for (i = 0; i <= len; ++i)
… …
335 encode_comma (enc); 594 encode_comma (enc);
336 } 595 }
337 596
338 encode_nl (enc); --enc->indent; encode_indent (enc); 597 encode_nl (enc); --enc->indent; encode_indent (enc);
339 } 598 }
340 599
341 encode_ch (enc, ']'); 600 encode_ch (enc, ']');
342} 601}
343 602
344static void 603static void
345encode_hk (enc_t *enc, HE *he) 604encode_hk (enc_t *enc, HE *he)
… …
349 if (HeKLEN (he) == HEf_SVKEY) 608 if (HeKLEN (he) == HEf_SVKEY)
350 { 609 {
351 SV *sv = HeSVKEY (he); 610 SV *sv = HeSVKEY (he);
352 STRLEN len; 611 STRLEN len;
353 char *str; 612 char *str;
354 613
355 SvGETMAGIC (sv); 614 SvGETMAGIC (sv);
356 str = SvPV (sv, len); 615 str = SvPV (sv, len);
357 616
358 encode_str (enc, str, len, SvUTF8 (sv)); 617 encode_str (enc, str, len, SvUTF8 (sv));
359 } 618 }
… …
394 653
395static void 654static void
396encode_hv (enc_t *enc, HV *hv) 655encode_hv (enc_t *enc, HV *hv)
397{ 656{
398 HE *he; 657 HE *he;
399 int count;
400 658
401 if (enc->indent >= enc->maxdepth) 659 if (enc->indent >= enc->json.max_depth)
402 croak ("data structure too deep (hit recursion limit)"); 660 croak (ERR_NESTING_EXCEEDED);
403 661
404 encode_ch (enc, '{'); 662 encode_ch (enc, '{');
405 663
406 // for canonical output we have to sort by keys first 664 // for canonical output we have to sort by keys first
407 // actually, this is mostly due to the stupid so-called 665 // actually, this is mostly due to the stupid so-called
408 // security workaround added somewhere in 5.8.x. 666 // security workaround added somewhere in 5.8.x
409 // that randomises hash orderings 667 // that randomises hash orderings
410 if (enc->json.flags & F_CANONICAL) 668 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
411 { 669 {
412 int count = hv_iterinit (hv); 670 int count = hv_iterinit (hv);
413 671
414 if (SvMAGICAL (hv)) 672 if (SvMAGICAL (hv))
415 { 673 {
… …
426 } 684 }
427 685
428 if (count) 686 if (count)
429 { 687 {
430 int i, fast = 1; 688 int i, fast = 1;
431#if defined(__BORLANDC__) || defined(_MSC_VER) 689 HE *hes_stack [STACK_HES];
432 HE **hes = _alloca (count * sizeof (HE)); 690 HE **hes = hes_stack;
433#else 691
434 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 692 // allocate larger arrays on the heap
435#endif 693 if (count > STACK_HES)
694 {
695 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
696 hes = (HE **)SvPVX (sv);
697 }
436 698
437 i = 0; 699 i = 0;
438 while ((he = hv_iternext (hv))) 700 while ((he = hv_iternext (hv)))
439 { 701 {
440 hes [i++] = he; 702 hes [i++] = he;
… …
509// encode objects, arrays and special \0=false and \1=true values. 771// encode objects, arrays and special \0=false and \1=true values.
510static void 772static void
511encode_rv (enc_t *enc, SV *sv) 773encode_rv (enc_t *enc, SV *sv)
512{ 774{
513 svtype svt; 775 svtype svt;
776 GV *method;
514 777
515 SvGETMAGIC (sv); 778 SvGETMAGIC (sv);
516 svt = SvTYPE (sv); 779 svt = SvTYPE (sv);
517 780
518 if (expect_false (SvOBJECT (sv))) 781 if (expect_false (SvOBJECT (sv)))
519 { 782 {
520 HV *stash = !JSON_SLOW || json_boolean_stash 783 HV *stash = SvSTASH (sv);
521 ? json_boolean_stash
522 : gv_stashpv ("JSON::XS::Boolean", 1);
523 784
524 if (SvSTASH (sv) == stash) 785 if (stash == bool_stash)
525 { 786 {
526 if (SvIV (sv))
527 encode_str (enc, "true", 4, 0); 787 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
528 else
529 encode_str (enc, "false", 5, 0); 788 else encode_str (enc, "false", 5, 0);
530 } 789 }
790 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
791 {
792 int count;
793 dSP;
794
795 ENTER; SAVETMPS;
796 PUSHMARK (SP);
797 EXTEND (SP, 2);
798 // we re-bless the reference to get overload and other niceties right
799 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
800 PUSHs (sv_json);
801
802 PUTBACK;
803 count = call_sv ((SV *)GvCV (method), G_ARRAY);
804 SPAGAIN;
805
806 // catch this surprisingly common error
807 if (SvROK (TOPs) && SvRV (TOPs) == sv)
808 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
809
810 encode_ch (enc, '(');
811 encode_ch (enc, '"');
812 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
813 encode_ch (enc, '"');
814 encode_ch (enc, ')');
815 encode_ch (enc, '[');
816
817 if (count)
818 {
819 int i;
820
821 for (i = 0; i < count - 1; ++i)
822 {
823 encode_sv (enc, SP[i + 1 - count]);
824 encode_ch (enc, ',');
825 }
826
827 encode_sv (enc, TOPs);
828 SP -= count;
829 }
830
831 PUTBACK;
832
833 encode_ch (enc, ']');
834
835 FREETMPS; LEAVE;
836 }
837 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
838 {
839 dSP;
840
841 ENTER; SAVETMPS;
842 PUSHMARK (SP);
843 // we re-bless the reference to get overload and other niceties right
844 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
845
846 // calling with G_SCALAR ensures that we always get a 1 return value
847 PUTBACK;
848 call_sv ((SV *)GvCV (method), G_SCALAR);
849 SPAGAIN;
850
851 // catch this surprisingly common error
852 if (SvROK (TOPs) && SvRV (TOPs) == sv)
853 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
854
855 sv = POPs;
856 PUTBACK;
857
858 encode_sv (enc, sv);
859
860 FREETMPS; LEAVE;
861 }
862 else if (enc->json.flags & F_ALLOW_BLESSED)
863 encode_str (enc, "null", 4, 0);
531 else 864 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", 865 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)))); 866 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
578 }
579 } 867 }
580 else if (svt == SVt_PVHV) 868 else if (svt == SVt_PVHV)
581 encode_hv (enc, (HV *)sv); 869 encode_hv (enc, (HV *)sv);
582 else if (svt == SVt_PVAV) 870 else if (svt == SVt_PVAV)
583 encode_av (enc, (AV *)sv); 871 encode_av (enc, (AV *)sv);
584 else if (svt < SVt_PVAV) 872 else if (svt < SVt_PVAV)
585 { 873 {
586 STRLEN len = 0; 874 int bool_type = ref_bool_type (sv);
587 char *pv = svt ? SvPV (sv, len) : 0;
588 875
589 if (len == 1 && *pv == '1') 876 if (bool_type == 1)
590 encode_str (enc, "true", 4, 0); 877 encode_str (enc, "true", 4, 0);
591 else if (len == 1 && *pv == '0') 878 else if (bool_type == 0)
592 encode_str (enc, "false", 5, 0); 879 encode_str (enc, "false", 5, 0);
880 else if (enc->json.flags & F_ALLOW_UNKNOWN)
881 encode_str (enc, "null", 4, 0);
593 else 882 else
594 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 883 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
595 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 884 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
596 } 885 }
886 else if (enc->json.flags & F_ALLOW_UNKNOWN)
887 encode_str (enc, "null", 4, 0);
597 else 888 else
598 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 889 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
599 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 890 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
600} 891}
601 892
… …
619 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 910 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
620 enc->cur += strlen (enc->cur); 911 enc->cur += strlen (enc->cur);
621 } 912 }
622 else if (SvIOKp (sv)) 913 else if (SvIOKp (sv))
623 { 914 {
624 // we assume we can always read an IV as a UV 915 // we assume we can always read an IV as a UV and vice versa
625 if (SvUV (sv) & ~(UV)0x7fff) 916 // we assume two's complement
626 { 917 // we assume no aliasing issues in the union
627 // large integer, use the (rather slow) snprintf way. 918 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
628 need (enc, sizeof (UV) * 3); 919 : 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 { 920 {
636 // optimise the "small number case" 921 // optimise the "small number case"
637 // code will likely be branchless and use only a single multiplication 922 // code will likely be branchless and use only a single multiplication
923 // works for numbers up to 59074
638 I32 i = SvIV (sv); 924 I32 i = SvIVX (sv);
639 U32 u; 925 U32 u;
640 char digit, nz = 0; 926 char digit, nz = 0;
641 927
642 need (enc, 6); 928 need (enc, 6);
643 929
… …
649 935
650 // now output digit by digit, each time masking out the integer part 936 // 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, 937 // and multiplying by 5 while moving the decimal point one to the right,
652 // resulting in a net multiplication by 10. 938 // resulting in a net multiplication by 10.
653 // we always write the digit to memory but conditionally increment 939 // we always write the digit to memory but conditionally increment
654 // the pointer, to ease the usage of conditional move instructions. 940 // 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; 941 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; 942 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; 943 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; 944 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' 945 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
660 } 946 }
947 else
948 {
949 // large integer, use the (rather slow) snprintf way.
950 need (enc, IVUV_MAXCHARS);
951 enc->cur +=
952 SvIsUV(sv)
953 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
954 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
955 }
661 } 956 }
662 else if (SvROK (sv)) 957 else if (SvROK (sv))
663 encode_rv (enc, SvRV (sv)); 958 encode_rv (enc, SvRV (sv));
664 else if (!SvOK (sv)) 959 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
665 encode_str (enc, "null", 4, 0); 960 encode_str (enc, "null", 4, 0);
666 else 961 else
667 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 962 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
668 SvPV_nolen (sv), SvFLAGS (sv)); 963 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
669} 964}
670 965
671static SV * 966static SV *
672encode_json (SV *scalar, JSON *json) 967encode_json (SV *scalar, JSON *json)
673{ 968{
674 enc_t enc; 969 enc_t enc;
675 970
676 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 971 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)"); 972 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
678 973
679 enc.json = *json; 974 enc.json = *json;
680 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 975 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
681 enc.cur = SvPVX (enc.sv); 976 enc.cur = SvPVX (enc.sv);
682 enc.end = SvEND (enc.sv); 977 enc.end = SvEND (enc.sv);
683 enc.indent = 0; 978 enc.indent = 0;
684 enc.maxdepth = DEC_DEPTH (enc.json.flags); 979 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
980 : enc.json.flags & F_LATIN1 ? 0x000100UL
981 : 0x110000UL;
685 982
686 SvPOK_only (enc.sv); 983 SvPOK_only (enc.sv);
687 encode_sv (&enc, scalar); 984 encode_sv (&enc, scalar);
985 encode_nl (&enc);
688 986
689 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 987 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
690 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 988 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
691 989
692 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 990 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
710 JSON json; 1008 JSON json;
711 U32 depth; // recursion depth 1009 U32 depth; // recursion depth
712 U32 maxdepth; // recursion depth limit 1010 U32 maxdepth; // recursion depth limit
713} dec_t; 1011} dec_t;
714 1012
715inline void 1013INLINE void
716decode_comment (dec_t *dec) 1014decode_comment (dec_t *dec)
717{ 1015{
718 // only '#'-style comments allowed a.t.m. 1016 // only '#'-style comments allowed a.t.m.
719 1017
720 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) 1018 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
721 ++dec->cur; 1019 ++dec->cur;
722} 1020}
723 1021
724inline void 1022INLINE void
725decode_ws (dec_t *dec) 1023decode_ws (dec_t *dec)
726{ 1024{
727 for (;;) 1025 for (;;)
728 { 1026 {
729 char ch = *dec->cur; 1027 char ch = *dec->cur;
… …
753 if (*dec->cur != ch) \ 1051 if (*dec->cur != ch) \
754 ERR (# ch " expected"); \ 1052 ERR (# ch " expected"); \
755 ++dec->cur; \ 1053 ++dec->cur; \
756 SE 1054 SE
757 1055
758#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 1056#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
759#define DEC_DEC_DEPTH --dec->depth 1057#define DEC_DEC_DEPTH --dec->depth
760 1058
761static SV *decode_sv (dec_t *dec); 1059static SV *decode_sv (dec_t *dec);
762 1060
763static signed char decode_hexdigit[256]; 1061static signed char decode_hexdigit[256];
… …
855 1153
856 if (hi >= 0x80) 1154 if (hi >= 0x80)
857 { 1155 {
858 utf8 = 1; 1156 utf8 = 1;
859 1157
860 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1158 cur = encode_utf8 (cur, hi);
861 } 1159 }
862 else 1160 else
863 *cur++ = hi; 1161 *cur++ = hi;
864 } 1162 }
865 break; 1163 break;
… …
867 default: 1165 default:
868 --dec_cur; 1166 --dec_cur;
869 ERR ("illegal backslash escape sequence in string"); 1167 ERR ("illegal backslash escape sequence in string");
870 } 1168 }
871 } 1169 }
872 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1170 else if (expect_true (ch >= 0x20 && ch < 0x80))
873 *cur++ = ch; 1171 *cur++ = ch;
874 else if (ch >= 0x80) 1172 else if (ch >= 0x80)
875 { 1173 {
876 STRLEN clen; 1174 STRLEN clen;
877 UV uch;
878 1175
879 --dec_cur; 1176 --dec_cur;
880 1177
881 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1178 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
882 if (clen == (STRLEN)-1) 1179 if (clen == (STRLEN)-1)
883 ERR ("malformed UTF-8 character in JSON string"); 1180 ERR ("malformed UTF-8 character in JSON string");
884 1181
885 do 1182 do
886 *cur++ = *dec_cur++; 1183 *cur++ = *dec_cur++;
887 while (--clen); 1184 while (--clen);
888 1185
889 utf8 = 1; 1186 utf8 = 1;
890 } 1187 }
1188 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1189 *cur++ = ch;
891 else 1190 else
892 { 1191 {
893 --dec_cur; 1192 --dec_cur;
894 1193
895 if (!ch) 1194 if (!ch)
… …
903 { 1202 {
904 STRLEN len = cur - buf; 1203 STRLEN len = cur - buf;
905 1204
906 if (sv) 1205 if (sv)
907 { 1206 {
908 SvGROW (sv, SvCUR (sv) + len + 1); 1207 STRLEN cur = SvCUR (sv);
1208
1209 if (SvLEN (sv) - cur <= len)
1210 json_sv_grow (sv, cur, len);
1211
909 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1212 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
910 SvCUR_set (sv, SvCUR (sv) + len); 1213 SvCUR_set (sv, SvCUR (sv) + len);
911 } 1214 }
912 else 1215 else
913 sv = newSVpvn (buf, len); 1216 sv = newSVpvn (buf, len);
… …
1000 1303
1001 if (!is_nv) 1304 if (!is_nv)
1002 { 1305 {
1003 int len = dec->cur - start; 1306 int len = dec->cur - start;
1004 1307
1005 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1308 // special case the rather common 1..5-digit-int case
1006 if (*start == '-') 1309 if (*start == '-')
1007 switch (len) 1310 switch (len)
1008 { 1311 {
1009 case 2: return newSViv (-( start [1] - '0' * 1)); 1312 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1010 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1313 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)); 1314 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)); 1315 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1316 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1013 } 1317 }
1014 else 1318 else
1015 switch (len) 1319 switch (len)
1016 { 1320 {
1017 case 1: return newSViv ( start [0] - '0' * 1); 1321 case 1: return newSViv ( start [0] - '0' * 1);
1018 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1322 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1019 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1323 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1020 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1324 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1325 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1021 } 1326 }
1022 1327
1023 { 1328 {
1024 UV uv; 1329 UV uv;
1025 int numtype = grok_number (start, len, &uv); 1330 int numtype = grok_number (start, len, &uv);
… …
1034 } 1339 }
1035 1340
1036 len -= *start == '-' ? 1 : 0; 1341 len -= *start == '-' ? 1 : 0;
1037 1342
1038 // does not fit into IV or UV, try NV 1343 // does not fit into IV or UV, try NV
1039 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1344 if (len <= NV_DIG)
1040 #if defined (LDBL_DIG)
1041 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1042 #endif
1043 )
1044 // fits into NV without loss of precision 1345 // fits into NV without loss of precision
1045 return newSVnv (Atof (start)); 1346 return newSVnv (json_atof (start));
1046 1347
1047 // everything else fails, convert it to a string 1348 // everything else fails, convert it to a string
1048 return newSVpvn (start, dec->cur - start); 1349 return newSVpvn (start, dec->cur - start);
1049 } 1350 }
1050 1351
1051 // loss of precision here 1352 // loss of precision here
1052 return newSVnv (Atof (start)); 1353 return newSVnv (json_atof (start));
1053 1354
1054fail: 1355fail:
1055 return 0; 1356 return 0;
1056} 1357}
1057 1358
… …
1081 if (*dec->cur == ']') 1382 if (*dec->cur == ']')
1082 { 1383 {
1083 ++dec->cur; 1384 ++dec->cur;
1084 break; 1385 break;
1085 } 1386 }
1086 1387
1087 if (*dec->cur != ',') 1388 if (*dec->cur != ',')
1088 ERR (", or ] expected while parsing array"); 1389 ERR (", or ] expected while parsing array");
1089 1390
1090 ++dec->cur; 1391 ++dec->cur;
1091 1392
… …
1133 char *p = dec->cur; 1434 char *p = dec->cur;
1134 char *e = p + 24; // only try up to 24 bytes 1435 char *e = p + 24; // only try up to 24 bytes
1135 1436
1136 for (;;) 1437 for (;;)
1137 { 1438 {
1138 // the >= 0x80 is true on most architectures 1439 // the >= 0x80 is false on most architectures
1139 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1440 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1140 { 1441 {
1141 // slow path, back up and use decode_str 1442 // slow path, back up and use decode_str
1142 SV *key = decode_str (dec); 1443 SV *key = decode_str (dec);
1143 if (!key) 1444 if (!key)
… …
1205 1506
1206 DEC_DEC_DEPTH; 1507 DEC_DEC_DEPTH;
1207 sv = newRV_noinc ((SV *)hv); 1508 sv = newRV_noinc ((SV *)hv);
1208 1509
1209 // check filter callbacks 1510 // check filter callbacks
1210 if (dec->json.flags & F_HOOK) 1511 if (expect_false (dec->json.flags & F_HOOK))
1211 { 1512 {
1212 if (dec->json.cb_sk_object && HvKEYS (hv) == 1) 1513 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1213 { 1514 {
1214 HE *cb, *he; 1515 HE *cb, *he;
1215 1516
1216 hv_iterinit (hv); 1517 hv_iterinit (hv);
1217 he = hv_iternext (hv); 1518 he = hv_iternext (hv);
1218 hv_iterinit (hv); 1519 hv_iterinit (hv);
1219 1520
1220 // the next line creates a mortal sv each time its called. 1521 // the next line creates a mortal sv each time it's called.
1221 // might want to optimise this for common cases. 1522 // might want to optimise this for common cases.
1222 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1523 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1223 1524
1224 if (cb) 1525 if (cb)
1225 { 1526 {
1226 dSP; 1527 dSP;
1227 int count; 1528 int count;
1228 1529
1229 ENTER; SAVETMPS; PUSHMARK (SP); 1530 ENTER; SAVETMPS;
1531 PUSHMARK (SP);
1230 XPUSHs (HeVAL (he)); 1532 XPUSHs (HeVAL (he));
1533 sv_2mortal (sv);
1231 1534
1232 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1535 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1233 1536
1234 if (count == 1) 1537 if (count == 1)
1235 { 1538 {
1236 sv = newSVsv (POPs); 1539 sv = newSVsv (POPs);
1540 PUTBACK;
1237 FREETMPS; LEAVE; 1541 FREETMPS; LEAVE;
1238 return sv; 1542 return sv;
1239 } 1543 }
1544 else if (count)
1545 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1546
1547 PUTBACK;
1548
1549 SvREFCNT_inc (sv);
1240 1550
1241 FREETMPS; LEAVE; 1551 FREETMPS; LEAVE;
1242 } 1552 }
1243 } 1553 }
1244 1554
1245 if (dec->json.cb_object) 1555 if (dec->json.cb_object)
1246 { 1556 {
1247 dSP; 1557 dSP;
1248 int count; 1558 int count;
1249 1559
1250 ENTER; SAVETMPS; PUSHMARK (SP); 1560 ENTER; SAVETMPS;
1561 PUSHMARK (SP);
1251 XPUSHs (sv_2mortal (sv)); 1562 XPUSHs (sv_2mortal (sv));
1252 1563
1253 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1564 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1254 1565
1255 if (count == 1) 1566 if (count == 1)
1256 {
1257 sv = newSVsv (POPs); 1567 sv = newSVsv (POPs);
1258 FREETMPS; LEAVE; 1568 else if (count == 0)
1259 return sv;
1260 }
1261
1262 SvREFCNT_inc (sv); 1569 SvREFCNT_inc (sv);
1570 else
1571 croak ("filter_json_object callbacks must not return more than one scalar");
1572
1573 PUTBACK;
1574
1263 FREETMPS; LEAVE; 1575 FREETMPS; LEAVE;
1264 } 1576 }
1265 } 1577 }
1266 1578
1267 return sv; 1579 return sv;
… …
1271 DEC_DEC_DEPTH; 1583 DEC_DEC_DEPTH;
1272 return 0; 1584 return 0;
1273} 1585}
1274 1586
1275static SV * 1587static SV *
1588decode_tag (dec_t *dec)
1589{
1590 SV *tag = 0;
1591 SV *val = 0;
1592
1593 if (!(dec->json.flags & F_ALLOW_TAGS))
1594 ERR ("malformed JSON string, neither array, object, number, string or atom");
1595
1596 ++dec->cur;
1597
1598 decode_ws (dec);
1599
1600 tag = decode_sv (dec);
1601 if (!tag)
1602 goto fail;
1603
1604 if (!SvPOK (tag))
1605 ERR ("malformed JSON string, (tag) must be a string");
1606
1607 decode_ws (dec);
1608
1609 if (*dec->cur != ')')
1610 ERR (") expected after tag");
1611
1612 ++dec->cur;
1613
1614 decode_ws (dec);
1615
1616 val = decode_sv (dec);
1617 if (!val)
1618 goto fail;
1619
1620 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1621 ERR ("malformed JSON string, tag value must be an array");
1622
1623 {
1624 AV *av = (AV *)SvRV (val);
1625 int i, len = av_len (av) + 1;
1626 HV *stash = gv_stashsv (tag, 0);
1627 SV *sv;
1628
1629 if (!stash)
1630 ERR ("cannot decode perl-object (package does not exist)");
1631
1632 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1633
1634 if (!method)
1635 ERR ("cannot decode perl-object (package does not have a THAW method)");
1636
1637 dSP;
1638
1639 ENTER; SAVETMPS;
1640 PUSHMARK (SP);
1641 EXTEND (SP, len + 2);
1642 // we re-bless the reference to get overload and other niceties right
1643 PUSHs (tag);
1644 PUSHs (sv_json);
1645
1646 for (i = 0; i < len; ++i)
1647 PUSHs (*av_fetch (av, i, 1));
1648
1649 PUTBACK;
1650 call_sv ((SV *)GvCV (method), G_SCALAR);
1651 SPAGAIN;
1652
1653 SvREFCNT_dec (tag);
1654 SvREFCNT_dec (val);
1655 sv = SvREFCNT_inc (POPs);
1656
1657 PUTBACK;
1658
1659 FREETMPS; LEAVE;
1660
1661 return sv;
1662 }
1663
1664fail:
1665 SvREFCNT_dec (tag);
1666 SvREFCNT_dec (val);
1667 return 0;
1668}
1669
1670static SV *
1276decode_sv (dec_t *dec) 1671decode_sv (dec_t *dec)
1277{ 1672{
1278 // the beauty of JSON: you need exactly one character lookahead 1673 // the beauty of JSON: you need exactly one character lookahead
1279 // to parse anything. 1674 // to parse everything.
1280 switch (*dec->cur) 1675 switch (*dec->cur)
1281 { 1676 {
1282 case '"': ++dec->cur; return decode_str (dec); 1677 case '"': ++dec->cur; return decode_str (dec);
1283 case '[': ++dec->cur; return decode_av (dec); 1678 case '[': ++dec->cur; return decode_av (dec);
1284 case '{': ++dec->cur; return decode_hv (dec); 1679 case '{': ++dec->cur; return decode_hv (dec);
1680 case '(': return decode_tag (dec);
1285 1681
1286 case '-': 1682 case '-':
1287 case '0': case '1': case '2': case '3': case '4': 1683 case '0': case '1': case '2': case '3': case '4':
1288 case '5': case '6': case '7': case '8': case '9': 1684 case '5': case '6': case '7': case '8': case '9':
1289 return decode_num (dec); 1685 return decode_num (dec);
1290 1686
1687 case 'f':
1688 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1689 {
1690 dec->cur += 5;
1691
1692 if (expect_false (!dec->json.v_false))
1693 dec->json.v_false = GET_BOOL (false);
1694
1695 return newSVsv (dec->json.v_false);
1696 }
1697 else
1698 ERR ("'false' expected");
1699
1700 break;
1701
1291 case 't': 1702 case 't':
1292 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1293 { 1704 {
1294 dec->cur += 4; 1705 dec->cur += 4;
1295#if JSON_SLOW 1706
1296 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1707 if (expect_false (!dec->json.v_true))
1297#endif 1708 dec->json.v_true = GET_BOOL (true);
1298 return SvREFCNT_inc (json_true); 1709
1710 return newSVsv (dec->json.v_true);
1299 } 1711 }
1300 else 1712 else
1301 ERR ("'true' expected"); 1713 ERR ("'true' expected");
1302
1303 break;
1304
1305 case 'f':
1306 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1307 {
1308 dec->cur += 5;
1309#if JSON_SLOW
1310 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1311#endif
1312 return SvREFCNT_inc (json_false);
1313 }
1314 else
1315 ERR ("'false' expected");
1316 1714
1317 break; 1715 break;
1318 1716
1319 case 'n': 1717 case 'n':
1320 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1718 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1326 ERR ("'null' expected"); 1724 ERR ("'null' expected");
1327 1725
1328 break; 1726 break;
1329 1727
1330 default: 1728 default:
1331 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1729 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1332 break; 1730 break;
1333 } 1731 }
1334 1732
1335fail: 1733fail:
1336 return 0; 1734 return 0;
1337} 1735}
1338 1736
1339static SV * 1737static SV *
1340decode_json (SV *string, JSON *json, UV *offset_return) 1738decode_json (SV *string, JSON *json, STRLEN *offset_return)
1341{ 1739{
1342 dec_t dec; 1740 dec_t dec;
1343 UV offset;
1344 SV *sv; 1741 SV *sv;
1345 1742
1743 /* work around bugs in 5.10 where manipulating magic values
1744 * makes perl ignore the magic in subsequent accesses.
1745 * also make a copy of non-PV values, to get them into a clean
1746 * state (SvPV should do that, but it's buggy, see below).
1747 *
1748 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1749 * as reported by Reini Urban.
1750 */
1346 SvGETMAGIC (string); 1751 /*SvGETMAGIC (string);*/
1752 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1753 string = sv_2mortal (newSVsv (string));
1754
1347 SvUPGRADE (string, SVt_PV); 1755 SvUPGRADE (string, SVt_PV);
1348 1756
1349 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1757 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1758 * assertion with -DDEBUGGING, although SvCUR is documented to
1759 * return the xpv_cur field which certainly exists after upgrading.
1760 * according to nicholas clark, calling SvPOK fixes this.
1761 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1762 * and hope for the best.
1763 * Damnit, SvPV_nolen still trips over yet another assertion. This
1764 * assertion business is seriously broken, try yet another workaround
1765 * for the broken -DDEBUGGING.
1766 */
1767 {
1768#ifdef DEBUGGING
1769 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1770#else
1771 STRLEN offset = SvCUR (string);
1772#endif
1773
1774 if (offset > json->max_size && json->max_size)
1350 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1775 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1351 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1776 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1777 }
1352 1778
1353 if (json->flags & F_UTF8) 1779 if (DECODE_WANTS_OCTETS (json))
1354 sv_utf8_downgrade (string, 0); 1780 sv_utf8_downgrade (string, 0);
1355 else 1781 else
1356 sv_utf8_upgrade (string); 1782 sv_utf8_upgrade (string);
1357 1783
1358 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1784 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1359 1785
1360 dec.json = *json; 1786 dec.json = *json;
1361 dec.cur = SvPVX (string); 1787 dec.cur = SvPVX (string);
1362 dec.end = SvEND (string); 1788 dec.end = SvEND (string);
1363 dec.err = 0; 1789 dec.err = 0;
1364 dec.depth = 0; 1790 dec.depth = 0;
1365 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1366 1791
1367 if (dec.json.cb_object || dec.json.cb_sk_object) 1792 if (dec.json.cb_object || dec.json.cb_sk_object)
1368 dec.json.flags |= F_HOOK; 1793 dec.json.flags |= F_HOOK;
1369 1794
1370 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1795 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1371 1796
1372 decode_ws (&dec); 1797 decode_ws (&dec);
1373 sv = decode_sv (&dec); 1798 sv = decode_sv (&dec);
1374 1799
1375 if (!(offset_return || !sv)) 1800 if (offset_return)
1801 *offset_return = dec.cur - SvPVX (string);
1802 else if (sv)
1376 { 1803 {
1377 // check for trailing garbage 1804 // check for trailing garbage
1378 decode_ws (&dec); 1805 decode_ws (&dec);
1379 1806
1380 if (*dec.cur) 1807 if (dec.cur != dec.end)
1381 { 1808 {
1382 dec.err = "garbage after JSON object"; 1809 dec.err = "garbage after JSON object";
1383 SvREFCNT_dec (sv); 1810 SvREFCNT_dec (sv);
1384 sv = 0; 1811 sv = 0;
1385 } 1812 }
1386 }
1387
1388 if (offset_return || !sv)
1389 {
1390 offset = dec.json.flags & F_UTF8
1391 ? dec.cur - SvPVX (string)
1392 : utf8_distance (dec.cur, SvPVX (string));
1393
1394 if (offset_return)
1395 *offset_return = offset;
1396 } 1813 }
1397 1814
1398 if (!sv) 1815 if (!sv)
1399 { 1816 {
1400 SV *uni = sv_newmortal (); 1817 SV *uni = sv_newmortal ();
… …
1406 SAVEVPTR (PL_curcop); 1823 SAVEVPTR (PL_curcop);
1407 PL_curcop = &cop; 1824 PL_curcop = &cop;
1408 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1825 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1409 LEAVE; 1826 LEAVE;
1410 1827
1411 croak ("%s, at character offset %d [\"%s\"]", 1828 croak ("%s, at character offset %d (before \"%s\")",
1412 dec.err, 1829 dec.err,
1413 (int)offset, 1830 (int)ptr_to_index (string, dec.cur),
1414 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1831 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1415 } 1832 }
1416 1833
1417 sv = sv_2mortal (sv); 1834 sv = sv_2mortal (sv);
1418 1835
1419 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1836 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1420 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1837 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1421 1838
1422 return sv; 1839 return sv;
1840}
1841
1842/////////////////////////////////////////////////////////////////////////////
1843// incremental parser
1844
1845static void
1846incr_parse (JSON *self)
1847{
1848 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1849
1850 // the state machine here is a bit convoluted and could be simplified a lot
1851 // but this would make it slower, so...
1852
1853 for (;;)
1854 {
1855 switch (self->incr_mode)
1856 {
1857 // reached end of a scalar, see if we are inside a nested structure or not
1858 end_of_scalar:
1859 self->incr_mode = INCR_M_JSON;
1860
1861 if (self->incr_nest) // end of a scalar inside array, object or tag
1862 goto incr_m_json;
1863 else // end of scalar outside structure, json text ends here
1864 goto interrupt;
1865
1866 // only used for initial whitespace skipping
1867 case INCR_M_WS:
1868 for (;;)
1869 {
1870 if (*p > 0x20)
1871 {
1872 if (*p == '#')
1873 {
1874 self->incr_mode = INCR_M_C0;
1875 goto incr_m_c;
1876 }
1877 else
1878 {
1879 self->incr_mode = INCR_M_JSON;
1880 goto incr_m_json;
1881 }
1882 }
1883 else if (!*p)
1884 goto interrupt;
1885
1886 ++p;
1887 }
1888
1889 // skip a single char inside a string (for \\-processing)
1890 case INCR_M_BS:
1891 if (!*p)
1892 goto interrupt;
1893
1894 ++p;
1895 self->incr_mode = INCR_M_STR;
1896 goto incr_m_str;
1897
1898 // inside #-style comments
1899 case INCR_M_C0:
1900 case INCR_M_C1:
1901 incr_m_c:
1902 for (;;)
1903 {
1904 if (*p == '\n')
1905 {
1906 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1907 break;
1908 }
1909 else if (!*p)
1910 goto interrupt;
1911
1912 ++p;
1913 }
1914
1915 break;
1916
1917 // inside true/false/null
1918 case INCR_M_TFN:
1919 incr_m_tfn:
1920 for (;;)
1921 switch (*p++)
1922 {
1923 case 'r': case 'u': case 'e': // tRUE, falsE, nUll
1924 case 'a': case 'l': case 's': // fALSe, nuLL
1925 // allowed
1926 break;
1927
1928 default:
1929 --p;
1930 goto end_of_scalar;
1931 }
1932
1933 // inside a number
1934 case INCR_M_NUM:
1935 incr_m_num:
1936 for (;;)
1937 switch (*p++)
1938 {
1939 case 'e': case 'E': case '.': case '+':
1940 case '-':
1941 case '0': case '1': case '2': case '3': case '4':
1942 case '5': case '6': case '7': case '8': case '9':
1943 // allowed
1944 break;
1945
1946 default:
1947 --p;
1948 goto end_of_scalar;
1949 }
1950
1951 // inside a string
1952 case INCR_M_STR:
1953 incr_m_str:
1954 for (;;)
1955 {
1956 if (*p == '"')
1957 {
1958 ++p;
1959 goto end_of_scalar;
1960 }
1961 else if (*p == '\\')
1962 {
1963 ++p; // "virtually" consumes character after \
1964
1965 if (!*p) // if at end of string we have to switch modes
1966 {
1967 self->incr_mode = INCR_M_BS;
1968 goto interrupt;
1969 }
1970 }
1971 else if (!*p)
1972 goto interrupt;
1973
1974 ++p;
1975 }
1976
1977 // after initial ws, outside string
1978 case INCR_M_JSON:
1979 incr_m_json:
1980 for (;;)
1981 {
1982 switch (*p++)
1983 {
1984 case 0:
1985 --p;
1986 goto interrupt;
1987
1988 case 0x09:
1989 case 0x0a:
1990 case 0x0d:
1991 case 0x20:
1992 if (!self->incr_nest)
1993 {
1994 --p; // do not eat the whitespace, let the next round do it
1995 goto interrupt;
1996 }
1997 break;
1998
1999 // the following three blocks handle scalars. this makes the parser
2000 // more strict than required inside arrays or objects, and could
2001 // be moved to a special case on the toplevel (except strings)
2002 case 't':
2003 case 'f':
2004 case 'n':
2005 self->incr_mode = INCR_M_TFN;
2006 goto incr_m_tfn;
2007
2008 case '-':
2009 case '0': case '1': case '2': case '3': case '4':
2010 case '5': case '6': case '7': case '8': case '9':
2011 self->incr_mode = INCR_M_NUM;
2012 goto incr_m_num;
2013
2014 case '"':
2015 self->incr_mode = INCR_M_STR;
2016 goto incr_m_str;
2017
2018 case '[':
2019 case '{':
2020 case '(':
2021 if (++self->incr_nest > self->max_depth)
2022 croak (ERR_NESTING_EXCEEDED);
2023 break;
2024
2025 case ']':
2026 case '}':
2027 if (--self->incr_nest <= 0)
2028 goto interrupt;
2029 break;
2030
2031 case ')':
2032 --self->incr_nest;
2033 break;
2034
2035 case '#':
2036 self->incr_mode = INCR_M_C1;
2037 goto incr_m_c;
2038 }
2039 }
2040 }
2041
2042 modechange:
2043 ;
2044 }
2045
2046interrupt:
2047 self->incr_pos = p - SvPVX (self->incr_text);
2048 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
2049 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1423} 2050}
1424 2051
1425///////////////////////////////////////////////////////////////////////////// 2052/////////////////////////////////////////////////////////////////////////////
1426// XS interface functions 2053// XS interface functions
1427 2054
… …
1436 i >= '0' && i <= '9' ? i - '0' 2063 i >= '0' && i <= '9' ? i - '0'
1437 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2064 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1438 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2065 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1439 : -1; 2066 : -1;
1440 2067
1441 json_stash = gv_stashpv ("JSON::XS" , 1); 2068 json_stash = gv_stashpv ("JSON::XS" , 1);
1442 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2069 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
2070 bool_false = get_bool ("Types::Serialiser::false");
2071 bool_true = get_bool ("Types::Serialiser::true");
1443 2072
1444 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 2073 sv_json = newSVpv ("JSON", 0);
1445 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 2074 SvREADONLY_on (sv_json);
2075
2076 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1446} 2077}
1447 2078
1448PROTOTYPES: DISABLE 2079PROTOTYPES: DISABLE
1449 2080
1450void CLONE (...) 2081void CLONE (...)
1451 CODE: 2082 CODE:
2083 // as long as these writes are atomic, the race should not matter
2084 // as existing threads either already use 0, or use the old value,
2085 // which is sitll correct for the initial thread.
1452 json_stash = 0; 2086 json_stash = 0;
1453 json_boolean_stash = 0; 2087 bool_stash = 0;
2088 bool_false = 0;
2089 bool_true = 0;
1454 2090
1455void new (char *klass) 2091void new (char *klass)
1456 PPCODE: 2092 PPCODE:
1457{ 2093{
1458 SV *pv = NEWSV (0, sizeof (JSON)); 2094 SV *pv = NEWSV (0, sizeof (JSON));
1459 SvPOK_only (pv); 2095 SvPOK_only (pv);
1460 Zero (SvPVX (pv), 1, JSON); 2096 json_init ((JSON *)SvPVX (pv));
1461 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 2097 XPUSHs (sv_2mortal (sv_bless (
1462 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); 2098 newRV_noinc (pv),
2099 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
2100 )));
1463} 2101}
2102
2103void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2104 PPCODE:
2105 self->v_false = newSVsv (v_false);
2106 self->v_true = newSVsv (v_true);
2107 XPUSHs (ST (0));
2108
2109void get_boolean_values (JSON *self)
2110 PPCODE:
2111 if (self->v_false && self->v_true)
2112 {
2113 EXTEND (SP, 2);
2114 PUSHs (self->v_false);
2115 PUSHs (self->v_true);
2116 }
1464 2117
1465void ascii (JSON *self, int enable = 1) 2118void ascii (JSON *self, int enable = 1)
1466 ALIAS: 2119 ALIAS:
1467 ascii = F_ASCII 2120 ascii = F_ASCII
1468 latin1 = F_LATIN1 2121 latin1 = F_LATIN1
… …
1475 allow_nonref = F_ALLOW_NONREF 2128 allow_nonref = F_ALLOW_NONREF
1476 shrink = F_SHRINK 2129 shrink = F_SHRINK
1477 allow_blessed = F_ALLOW_BLESSED 2130 allow_blessed = F_ALLOW_BLESSED
1478 convert_blessed = F_CONV_BLESSED 2131 convert_blessed = F_CONV_BLESSED
1479 relaxed = F_RELAXED 2132 relaxed = F_RELAXED
2133 allow_unknown = F_ALLOW_UNKNOWN
2134 allow_tags = F_ALLOW_TAGS
1480 PPCODE: 2135 PPCODE:
1481{ 2136{
1482 if (enable) 2137 if (enable)
1483 self->flags |= ix; 2138 self->flags |= ix;
1484 else 2139 else
1485 self->flags &= ~ix; 2140 self->flags &= ~ix;
1486 2141
1487 XPUSHs (ST (0)); 2142 XPUSHs (ST (0));
1488} 2143}
1489 2144
1490void max_depth (JSON *self, UV max_depth = 0x80000000UL) 2145void get_ascii (JSON *self)
2146 ALIAS:
2147 get_ascii = F_ASCII
2148 get_latin1 = F_LATIN1
2149 get_utf8 = F_UTF8
2150 get_indent = F_INDENT
2151 get_canonical = F_CANONICAL
2152 get_space_before = F_SPACE_BEFORE
2153 get_space_after = F_SPACE_AFTER
2154 get_allow_nonref = F_ALLOW_NONREF
2155 get_shrink = F_SHRINK
2156 get_allow_blessed = F_ALLOW_BLESSED
2157 get_convert_blessed = F_CONV_BLESSED
2158 get_relaxed = F_RELAXED
2159 get_allow_unknown = F_ALLOW_UNKNOWN
2160 get_allow_tags = F_ALLOW_TAGS
1491 PPCODE: 2161 PPCODE:
1492{ 2162 XPUSHs (boolSV (self->flags & ix));
1493 UV log2 = 0;
1494 2163
1495 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2164void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1496 2165 PPCODE:
1497 while ((1UL << log2) < max_depth) 2166 self->max_depth = max_depth;
1498 ++log2;
1499
1500 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1501
1502 XPUSHs (ST (0)); 2167 XPUSHs (ST (0));
1503}
1504 2168
2169U32 get_max_depth (JSON *self)
2170 CODE:
2171 RETVAL = self->max_depth;
2172 OUTPUT:
2173 RETVAL
2174
1505void max_size (JSON *self, UV max_size = 0) 2175void max_size (JSON *self, U32 max_size = 0)
1506 PPCODE: 2176 PPCODE:
1507{ 2177 self->max_size = max_size;
1508 UV log2 = 0;
1509
1510 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1511 if (max_size == 1) max_size = 2;
1512
1513 while ((1UL << log2) < max_size)
1514 ++log2;
1515
1516 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1517
1518 XPUSHs (ST (0)); 2178 XPUSHs (ST (0));
1519} 2179
2180int get_max_size (JSON *self)
2181 CODE:
2182 RETVAL = self->max_size;
2183 OUTPUT:
2184 RETVAL
1520 2185
1521void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 2186void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1522 PPCODE: 2187 PPCODE:
1523{ 2188{
1524 SvREFCNT_dec (self->cb_object); 2189 SvREFCNT_dec (self->cb_object);
… …
1528} 2193}
1529 2194
1530void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2195void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1531 PPCODE: 2196 PPCODE:
1532{ 2197{
1533 if (!self->cb_sk_object) 2198 if (!self->cb_sk_object)
1534 self->cb_sk_object = newHV (); 2199 self->cb_sk_object = newHV ();
1535 2200
1536 if (SvOK (cb)) 2201 if (SvOK (cb))
1537 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2202 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1538 else 2203 else
… …
1549 XPUSHs (ST (0)); 2214 XPUSHs (ST (0));
1550} 2215}
1551 2216
1552void encode (JSON *self, SV *scalar) 2217void encode (JSON *self, SV *scalar)
1553 PPCODE: 2218 PPCODE:
1554 XPUSHs (encode_json (scalar, self)); 2219 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2220 XPUSHs (scalar);
1555 2221
1556void decode (JSON *self, SV *jsonstr) 2222void decode (JSON *self, SV *jsonstr)
1557 PPCODE: 2223 PPCODE:
1558 XPUSHs (decode_json (jsonstr, self, 0)); 2224 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2225 XPUSHs (jsonstr);
1559 2226
1560void decode_prefix (JSON *self, SV *jsonstr) 2227void decode_prefix (JSON *self, SV *jsonstr)
1561 PPCODE: 2228 PPCODE:
1562{ 2229{
1563 UV offset; 2230 SV *sv;
2231 STRLEN offset;
2232 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1564 EXTEND (SP, 2); 2233 EXTEND (SP, 2);
1565 PUSHs (decode_json (jsonstr, self, &offset)); 2234 PUSHs (sv);
1566 PUSHs (sv_2mortal (newSVuv (offset))); 2235 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2236}
2237
2238void incr_parse (JSON *self, SV *jsonstr = 0)
2239 PPCODE:
2240{
2241 if (!self->incr_text)
2242 self->incr_text = newSVpvn ("", 0);
2243
2244 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2245 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2246 if (DECODE_WANTS_OCTETS (self))
2247 {
2248 if (self->incr_pos)
2249 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2250 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2251
2252 sv_utf8_downgrade (self->incr_text, 0);
2253 }
2254 else
2255 {
2256 sv_utf8_upgrade (self->incr_text);
2257
2258 if (self->incr_pos)
2259 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2260 - (U8 *)SvPVX (self->incr_text);
2261 }
2262
2263 // append data, if any
2264 if (jsonstr)
2265 {
2266 /* make sure both strings have same encoding */
2267 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2268 if (SvUTF8 (jsonstr))
2269 sv_utf8_downgrade (jsonstr, 0);
2270 else
2271 sv_utf8_upgrade (jsonstr);
2272
2273 /* and then just blindly append */
2274 {
2275 STRLEN len;
2276 const char *str = SvPV (jsonstr, len);
2277 STRLEN cur = SvCUR (self->incr_text);
2278
2279 if (SvLEN (self->incr_text) - cur <= len)
2280 json_sv_grow (self->incr_text, cur, len);
2281
2282 Move (str, SvEND (self->incr_text), len, char);
2283 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2284 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2285 }
2286 }
2287
2288 if (GIMME_V != G_VOID)
2289 do
2290 {
2291 SV *sv;
2292 STRLEN offset;
2293
2294 if (!INCR_DONE (self))
2295 {
2296 incr_parse (self);
2297
2298 if (self->incr_pos > self->max_size && self->max_size)
2299 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2300 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2301
2302 if (!INCR_DONE (self))
2303 {
2304 // as an optimisation, do not accumulate white space in the incr buffer
2305 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2306 {
2307 self->incr_pos = 0;
2308 SvCUR_set (self->incr_text, 0);
2309 }
2310
2311 break;
2312 }
2313 }
2314
2315 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2316 XPUSHs (sv);
2317
2318 self->incr_pos -= offset;
2319 self->incr_nest = 0;
2320 self->incr_mode = 0;
2321
2322 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2323 }
2324 while (GIMME_V == G_ARRAY);
2325}
2326
2327SV *incr_text (JSON *self)
2328 ATTRS: lvalue
2329 CODE:
2330{
2331 if (self->incr_pos)
2332 croak ("incr_text can not be called when the incremental parser already started parsing");
2333
2334 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2335}
2336 OUTPUT:
2337 RETVAL
2338
2339void incr_skip (JSON *self)
2340 CODE:
2341{
2342 if (self->incr_pos)
2343 {
2344 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2345 self->incr_pos = 0;
2346 self->incr_nest = 0;
2347 self->incr_mode = 0;
2348 }
2349}
2350
2351void incr_reset (JSON *self)
2352 CODE:
2353{
2354 SvREFCNT_dec (self->incr_text);
2355 self->incr_text = 0;
2356 self->incr_pos = 0;
2357 self->incr_nest = 0;
2358 self->incr_mode = 0;
1567} 2359}
1568 2360
1569void DESTROY (JSON *self) 2361void DESTROY (JSON *self)
1570 CODE: 2362 CODE:
2363 SvREFCNT_dec (self->v_false);
2364 SvREFCNT_dec (self->v_true);
1571 SvREFCNT_dec (self->cb_sk_object); 2365 SvREFCNT_dec (self->cb_sk_object);
1572 SvREFCNT_dec (self->cb_object); 2366 SvREFCNT_dec (self->cb_object);
2367 SvREFCNT_dec (self->incr_text);
1573 2368
1574PROTOTYPES: ENABLE 2369PROTOTYPES: ENABLE
1575 2370
1576void to_json (SV *scalar) 2371void encode_json (SV *scalar)
1577 PPCODE: 2372 PPCODE:
1578{ 2373{
1579 JSON json = { F_DEFAULT | F_UTF8 }; 2374 JSON json;
1580 XPUSHs (encode_json (scalar, &json)); 2375 json_init (&json);
2376 json.flags |= F_UTF8;
2377 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2378 XPUSHs (scalar);
1581} 2379}
1582 2380
1583void from_json (SV *jsonstr) 2381void decode_json (SV *jsonstr)
1584 PPCODE: 2382 PPCODE:
1585{ 2383{
1586 JSON json = { F_DEFAULT | F_UTF8 }; 2384 JSON json;
1587 XPUSHs (decode_json (jsonstr, &json, 0)); 2385 json_init (&json);
2386 json.flags |= F_UTF8;
2387 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2388 XPUSHs (jsonstr);
1588} 2389}
1589 2390

Diff Legend

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