ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.1 by root, Fri Oct 25 23:09:45 2013 UTC vs.
Revision 1.34 by root, Sat Nov 30 16:19:59 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14#if defined(__BORLANDC__) || defined(_MSC_VER) 14// compatibility with perl <5.18
15# define snprintf _snprintf // C compilers have this in stdio.h 15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
16#endif 17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
17 27
28// known tags
29enum cbor_tag
30{
31 // extensions
32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39
40 // rfc7049
41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43 CBOR_TAG_POS_BIGNUM = 2, // byte string
44 CBOR_TAG_NEG_BIGNUM = 3, // byte string
45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46 CBOR_TAG_BIGFLOAT = 5, // array
47
48 CBOR_TAG_CONV_B64U = 21, // base64url, any
49 CBOR_TAG_CONV_B64 = 22, // base64, any
50 CBOR_TAG_CONV_HEX = 23, // base16, any
51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52
53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58
59 CBOR_TAG_MAGIC = 55799 // self-describe cbor
60};
61
18#define F_SHRINK 0x00000200UL 62#define F_SHRINK 0x00000001UL
19#define F_ALLOW_UNKNOWN 0x00002000UL 63#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL
65#define F_PACK_STRINGS 0x00000008UL
20 66
21#define INIT_SIZE 32 // initial scalar size to be allocated 67#define INIT_SIZE 32 // initial scalar size to be allocated
22 68
23#define SB do { 69#define SB do {
24#define SE } while (0) 70#define SE } while (0)
25
26#if __GNUC__ >= 3
27# define expect(expr,value) __builtin_expect ((expr), (value))
28# define INLINE static inline
29#else
30# define expect(expr,value) (expr)
31# define INLINE static
32#endif
33
34#define expect_false(expr) expect ((expr) != 0, 0)
35#define expect_true(expr) expect ((expr) != 0, 1)
36 71
37#define IN_RANGE_INC(type,val,beg,end) \ 72#define IN_RANGE_INC(type,val,beg,end) \
38 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 73 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
39 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 74 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
40 75
46#else 81#else
47# define CBOR_SLOW 0 82# define CBOR_SLOW 0
48# define CBOR_STASH cbor_stash 83# define CBOR_STASH cbor_stash
49#endif 84#endif
50 85
51static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 86static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
52static SV *cbor_true, *cbor_false; 87static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
53 88
54typedef struct { 89typedef struct {
55 U32 flags; 90 U32 flags;
56 U32 max_depth; 91 U32 max_depth;
57 STRLEN max_size; 92 STRLEN max_size;
58 93 SV *filter;
59 SV *cb_object;
60 HV *cb_sk_object;
61} CBOR; 94} CBOR;
62 95
63INLINE void 96ecb_inline void
64cbor_init (CBOR *cbor) 97cbor_init (CBOR *cbor)
65{ 98{
66 Zero (cbor, 1, CBOR); 99 Zero (cbor, 1, CBOR);
67 cbor->max_depth = 512; 100 cbor->max_depth = 512;
68} 101}
69 102
103ecb_inline void
104cbor_free (CBOR *cbor)
105{
106 SvREFCNT_dec (cbor->filter);
107}
108
70///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
71// utility functions 110// utility functions
72 111
73INLINE SV * 112ecb_inline SV *
74get_bool (const char *name) 113get_bool (const char *name)
75{ 114{
76 SV *sv = get_sv (name, 1); 115 SV *sv = get_sv (name, 1);
77 116
78 SvREADONLY_on (sv); 117 SvREADONLY_on (sv);
79 SvREADONLY_on (SvRV (sv)); 118 SvREADONLY_on (SvRV (sv));
80 119
81 return sv; 120 return sv;
82} 121}
83 122
84INLINE void 123ecb_inline void
85shrink (SV *sv) 124shrink (SV *sv)
86{ 125{
87 sv_utf8_downgrade (sv, 1); 126 sv_utf8_downgrade (sv, 1);
88 127
89 if (SvLEN (sv) > SvCUR (sv) + 1) 128 if (SvLEN (sv) > SvCUR (sv) + 1)
94 SvPV_renew (sv, SvCUR (sv) + 1); 133 SvPV_renew (sv, SvCUR (sv) + 1);
95#endif 134#endif
96 } 135 }
97} 136}
98 137
99///////////////////////////////////////////////////////////////////////////// 138// minimum length of a string to be registered for stringref
100// fp hell 139ecb_inline int
101 140minimum_string_length (UV idx)
102//TODO 141{
142 return idx > 23
143 ? idx > 0xffU
144 ? idx > 0xffffU
145 ? idx > 0xffffffffU
146 ? 11
147 : 7
148 : 5
149 : 4
150 : 3;
151}
103 152
104///////////////////////////////////////////////////////////////////////////// 153/////////////////////////////////////////////////////////////////////////////
105// encoder 154// encoder
106 155
107// structure used for encoding CBOR 156// structure used for encoding CBOR
110 char *cur; // SvPVX (sv) + current output position 159 char *cur; // SvPVX (sv) + current output position
111 char *end; // SvEND (sv) 160 char *end; // SvEND (sv)
112 SV *sv; // result scalar 161 SV *sv; // result scalar
113 CBOR cbor; 162 CBOR cbor;
114 U32 depth; // recursion level 163 U32 depth; // recursion level
164 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
165 UV stringref_idx;
166 HV *shareable; // ptr => index, or 0
167 UV shareable_idx;
115} enc_t; 168} enc_t;
116 169
117INLINE void 170ecb_inline void
118need (enc_t *enc, STRLEN len) 171need (enc_t *enc, STRLEN len)
119{ 172{
120 if (expect_false (enc->cur + len >= enc->end)) 173 if (ecb_expect_false (enc->cur + len >= enc->end))
121 { 174 {
122 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 175 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 176 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 177 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 178 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 179 }
127} 180}
128 181
129INLINE void 182ecb_inline void
130encode_ch (enc_t *enc, char ch) 183encode_ch (enc_t *enc, char ch)
131{ 184{
132 need (enc, 1); 185 need (enc, 1);
133 *enc->cur++ = ch; 186 *enc->cur++ = ch;
134} 187}
136static void 189static void
137encode_uint (enc_t *enc, int major, UV len) 190encode_uint (enc_t *enc, int major, UV len)
138{ 191{
139 need (enc, 9); 192 need (enc, 9);
140 193
141 if (len < 24) 194 if (ecb_expect_true (len < 24))
142 *enc->cur++ = major | len; 195 *enc->cur++ = major | len;
143 else if (len < 0x100) 196 else if (ecb_expect_true (len <= 0xff))
144 { 197 {
145 *enc->cur++ = major | 24; 198 *enc->cur++ = major | 24;
146 *enc->cur++ = len; 199 *enc->cur++ = len;
147 } 200 }
148 else if (len < 0x10000) 201 else if (len <= 0xffff)
149 { 202 {
150 *enc->cur++ = major | 25; 203 *enc->cur++ = major | 25;
151 *enc->cur++ = len >> 8; 204 *enc->cur++ = len >> 8;
152 *enc->cur++ = len; 205 *enc->cur++ = len;
153 } 206 }
154 else if (len < 0x100000000) 207 else if (len <= 0xffffffff)
155 { 208 {
156 *enc->cur++ = major | 26; 209 *enc->cur++ = major | 26;
157 *enc->cur++ = len >> 24; 210 *enc->cur++ = len >> 24;
158 *enc->cur++ = len >> 16; 211 *enc->cur++ = len >> 16;
159 *enc->cur++ = len >> 8; 212 *enc->cur++ = len >> 8;
160 *enc->cur++ = len; 213 *enc->cur++ = len;
161 } 214 }
162 else if (len) 215 else
163 { 216 {
164 *enc->cur++ = major | 27; 217 *enc->cur++ = major | 27;
165 *enc->cur++ = len >> 56; 218 *enc->cur++ = len >> 56;
166 *enc->cur++ = len >> 48; 219 *enc->cur++ = len >> 48;
167 *enc->cur++ = len >> 40; 220 *enc->cur++ = len >> 40;
171 *enc->cur++ = len >> 8; 224 *enc->cur++ = len >> 8;
172 *enc->cur++ = len; 225 *enc->cur++ = len;
173 } 226 }
174} 227}
175 228
176static void 229ecb_inline void
230encode_tag (enc_t *enc, UV tag)
231{
232 encode_uint (enc, 0xc0, tag);
233}
234
235ecb_inline void
177encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 236encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
178{ 237{
179 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 238 encode_uint (enc, utf8 ? 0x60 : 0x40, len);
180 need (enc, len); 239 need (enc, len);
181 memcpy (enc->cur, str, len); 240 memcpy (enc->cur, str, len);
182 enc->cur += len; 241 enc->cur += len;
183} 242}
184 243
244static void
245encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
246{
247 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
248 {
249 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
250
251 if (SvOK (*svp))
252 {
253 // already registered, use stringref
254 encode_tag (enc, CBOR_TAG_STRINGREF);
255 encode_uint (enc, 0x00, SvUV (*svp));
256 return;
257 }
258 else if (len >= minimum_string_length (enc->stringref_idx))
259 {
260 // register only
261 sv_setuv (*svp, enc->stringref_idx);
262 ++enc->stringref_idx;
263 }
264 }
265
266 encode_str (enc, utf8, str, len);
267}
268
185static void encode_sv (enc_t *enc, SV *sv); 269static void encode_sv (enc_t *enc, SV *sv);
186 270
187static void 271static void
188encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
189{ 273{
226 while ((he = hv_iternext (hv))) 310 while ((he = hv_iternext (hv)))
227 { 311 {
228 if (HeKLEN (he) == HEf_SVKEY) 312 if (HeKLEN (he) == HEf_SVKEY)
229 encode_sv (enc, HeSVKEY (he)); 313 encode_sv (enc, HeSVKEY (he));
230 else 314 else
231 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 315 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
232 316
233 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 317 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
234 } 318 }
235 319
236 if (mg) 320 if (mg)
237 encode_ch (enc, 0xe0 | 31); 321 encode_ch (enc, 0xe0 | 31);
238 322
241 325
242// encode objects, arrays and special \0=false and \1=true values. 326// encode objects, arrays and special \0=false and \1=true values.
243static void 327static void
244encode_rv (enc_t *enc, SV *sv) 328encode_rv (enc_t *enc, SV *sv)
245{ 329{
246 svtype svt;
247
248 SvGETMAGIC (sv); 330 SvGETMAGIC (sv);
331
249 svt = SvTYPE (sv); 332 svtype svt = SvTYPE (sv);
250 333
251 if (expect_false (SvOBJECT (sv))) 334 if (ecb_expect_false (SvOBJECT (sv)))
252 { 335 {
253 HV *stash = !CBOR_SLOW || cbor_boolean_stash 336 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
254 ? cbor_boolean_stash 337 ? types_boolean_stash
338 : gv_stashpv ("Types::Serialiser::Boolean", 1);
339 HV *error_stash = !CBOR_SLOW || types_error_stash
340 ? types_error_stash
341 : gv_stashpv ("Types::Serialiser::Error", 1);
342 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
343 ? cbor_tagged_stash
255 : gv_stashpv ("CBOR::XS::Boolean", 1); 344 : gv_stashpv ("CBOR::XS::Tagged" , 1);
256 345
257 if (SvSTASH (sv) == stash) 346 HV *stash = SvSTASH (sv);
347
348 if (stash == boolean_stash)
349 {
258 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 350 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
351 return;
352 }
353 else if (stash == error_stash)
354 {
355 encode_ch (enc, 0xe0 | 23);
356 return;
357 }
358 else if (stash == tagged_stash)
359 {
360 if (svt != SVt_PVAV)
361 croak ("encountered CBOR::XS::Tagged object that isn't an array");
362
363 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
364 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
365
366 return;
367 }
368 }
369
370 if (ecb_expect_false (SvREFCNT (sv) > 1)
371 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
372 {
373 if (!enc->shareable)
374 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
375
376 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
377
378 if (SvOK (*svp))
379 {
380 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
381 encode_uint (enc, 0x00, SvUV (*svp));
382 return;
383 }
259 else 384 else
260 { 385 {
261#if 0 //TODO 386 sv_setuv (*svp, enc->shareable_idx);
262 if (enc->cbor.flags & F_CONV_BLESSED) 387 ++enc->shareable_idx;
388 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
389 }
390 }
391
392 if (ecb_expect_false (SvOBJECT (sv)))
393 {
394 HV *stash = SvSTASH (sv);
395 GV *method;
396
397 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
263 { 398 {
399 dSP;
400
401 ENTER; SAVETMPS; PUSHMARK (SP);
264 // we re-bless the reference to get overload and other niceties right 402 // we re-bless the reference to get overload and other niceties right
265 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
266
267 if (to_cbor)
268 {
269 dSP;
270
271 ENTER; SAVETMPS; PUSHMARK (SP);
272 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 403 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
273 404
274 // calling with G_SCALAR ensures that we always get a 1 return value
275 PUTBACK; 405 PUTBACK;
406 // G_SCALAR ensures that return value is 1
276 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 407 call_sv ((SV *)GvCV (method), G_SCALAR);
277 SPAGAIN; 408 SPAGAIN;
278 409
279 // catch this surprisingly common error 410 // catch this surprisingly common error
280 if (SvROK (TOPs) && SvRV (TOPs) == sv) 411 if (SvROK (TOPs) && SvRV (TOPs) == sv)
281 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 412 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
282 413
283 sv = POPs;
284 PUTBACK;
285
286 encode_sv (enc, sv); 414 encode_sv (enc, POPs);
287 415
416 PUTBACK;
417
288 FREETMPS; LEAVE; 418 FREETMPS; LEAVE;
289 }
290 else if (enc->cbor.flags & F_ALLOW_BLESSED)
291 encode_str (enc, "null", 4, 0);
292 else
293 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
294 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
295 } 419 }
296 else if (enc->cbor.flags & F_ALLOW_BLESSED) 420 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
297 encode_str (enc, "null", 4, 0); 421 {
422 dSP;
423
424 ENTER; SAVETMPS; PUSHMARK (SP);
425 EXTEND (SP, 2);
426 // we re-bless the reference to get overload and other niceties right
427 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
428 PUSHs (sv_cbor);
429
430 PUTBACK;
431 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
432 SPAGAIN;
433
434 // catch this surprisingly common error
435 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
436 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
437
438 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
439 encode_uint (enc, 0x80, count + 1);
440 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
441
442 while (count)
443 encode_sv (enc, SP[1 - count--]);
444
445 PUTBACK;
446
447 FREETMPS; LEAVE;
448 }
298 else 449 else
299 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 450 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
300 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 451 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
301#endif
302 }
303 } 452 }
304 else if (svt == SVt_PVHV) 453 else if (svt == SVt_PVHV)
305 encode_hv (enc, (HV *)sv); 454 encode_hv (enc, (HV *)sv);
306 else if (svt == SVt_PVAV) 455 else if (svt == SVt_PVAV)
307 encode_av (enc, (AV *)sv); 456 encode_av (enc, (AV *)sv);
308 else if (svt < SVt_PVAV)
309 {
310 STRLEN len = 0;
311 char *pv = svt ? SvPV (sv, len) : 0;
312
313 if (len == 1 && *pv == '1')
314 encode_ch (enc, 0xe0 | 21);
315 else if (len == 1 && *pv == '0')
316 encode_ch (enc, 0xe0 | 20);
317 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
318 encode_ch (enc, 0xe0 | 23);
319 else
320 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
321 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
322 }
323 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
324 encode_ch (enc, 0xe0 | 23);
325 else 457 else
326 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 458 {
327 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 459 encode_tag (enc, CBOR_TAG_INDIRECTION);
460 encode_sv (enc, sv);
461 }
328} 462}
329 463
330static void 464static void
331encode_nv (enc_t *enc, SV *sv) 465encode_nv (enc_t *enc, SV *sv)
332{ 466{
333 double nv = SvNVX (sv); 467 double nv = SvNVX (sv);
334 468
335 need (enc, 9); 469 need (enc, 9);
336 470
337 if (expect_false (nv == (U32)nv)) 471 if (ecb_expect_false (nv == (U32)nv))
338 encode_uint (enc, 0x00, (U32)nv); 472 encode_uint (enc, 0x00, (U32)nv);
339 //TODO: maybe I32? 473 //TODO: maybe I32?
340 else if (expect_false (nv == (float)nv)) 474 else if (ecb_expect_false (nv == (float)nv))
341 { 475 {
342 uint32_t fp = ecb_float_to_binary32 (nv); 476 uint32_t fp = ecb_float_to_binary32 (nv);
343 477
344 *enc->cur++ = 0xe0 | 26; 478 *enc->cur++ = 0xe0 | 26;
345 479
370 504
371 if (SvPOKp (sv)) 505 if (SvPOKp (sv))
372 { 506 {
373 STRLEN len; 507 STRLEN len;
374 char *str = SvPV (sv, len); 508 char *str = SvPV (sv, len);
375 encode_str (enc, SvUTF8 (sv), str, len); 509 encode_strref (enc, SvUTF8 (sv), str, len);
376 } 510 }
377 else if (SvNOKp (sv)) 511 else if (SvNOKp (sv))
378 encode_nv (enc, sv); 512 encode_nv (enc, sv);
379 else if (SvIOKp (sv)) 513 else if (SvIOKp (sv))
380 { 514 {
397} 531}
398 532
399static SV * 533static SV *
400encode_cbor (SV *scalar, CBOR *cbor) 534encode_cbor (SV *scalar, CBOR *cbor)
401{ 535{
402 enc_t enc; 536 enc_t enc = { };
403 537
404 enc.cbor = *cbor; 538 enc.cbor = *cbor;
405 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 539 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
406 enc.cur = SvPVX (enc.sv); 540 enc.cur = SvPVX (enc.sv);
407 enc.end = SvEND (enc.sv); 541 enc.end = SvEND (enc.sv);
408 enc.depth = 0;
409 542
410 SvPOK_only (enc.sv); 543 SvPOK_only (enc.sv);
544
545 if (cbor->flags & F_PACK_STRINGS)
546 {
547 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
548 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
549 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
550 }
551
411 encode_sv (&enc, scalar); 552 encode_sv (&enc, scalar);
412 553
413 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 554 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
414 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 555 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
415 556
429 U8 *end; // end of input string 570 U8 *end; // end of input string
430 const char *err; // parse error, if != 0 571 const char *err; // parse error, if != 0
431 CBOR cbor; 572 CBOR cbor;
432 U32 depth; // recursion depth 573 U32 depth; // recursion depth
433 U32 maxdepth; // recursion depth limit 574 U32 maxdepth; // recursion depth limit
575 AV *shareable;
576 AV *stringref;
577 SV *decode_tagged;
434} dec_t; 578} dec_t;
435 579
436#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 580#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
437 581
438#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 582#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
439 583
440#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 584#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
441#define DEC_DEC_DEPTH --dec->depth 585#define DEC_DEC_DEPTH --dec->depth
442 586
443static UV 587static UV
470 | ((UV)dec->cur[-1]); 614 | ((UV)dec->cur[-1]);
471 615
472 case 27: 616 case 27:
473 WANT (9); 617 WANT (9);
474 dec->cur += 9; 618 dec->cur += 9;
619
620 return
621#if UVSIZE < 8
622 0
623#else
475 return (((UV)dec->cur[-8]) << 56) 624 (((UV)dec->cur[-8]) << 56)
476 | (((UV)dec->cur[-7]) << 48) 625 | (((UV)dec->cur[-7]) << 48)
477 | (((UV)dec->cur[-6]) << 40) 626 | (((UV)dec->cur[-6]) << 40)
478 | (((UV)dec->cur[-5]) << 32) 627 | (((UV)dec->cur[-5]) << 32)
628#endif
479 | (((UV)dec->cur[-4]) << 24) 629 | (((UV)dec->cur[-4]) << 24)
480 | (((UV)dec->cur[-3]) << 16) 630 | (((UV)dec->cur[-3]) << 16)
481 | (((UV)dec->cur[-2]) << 8) 631 | (((UV)dec->cur[-2]) << 8)
482 | ((UV)dec->cur[-1]); 632 | ((UV)dec->cur[-1]);
483 633
504 654
505 for (;;) 655 for (;;)
506 { 656 {
507 WANT (1); 657 WANT (1);
508 658
509 if (*dec->cur == 0xe0 | 31) 659 if (*dec->cur == (0xe0 | 31))
510 { 660 {
511 ++dec->cur; 661 ++dec->cur;
512 break; 662 break;
513 } 663 }
514 664
532 SvREFCNT_dec (av); 682 SvREFCNT_dec (av);
533 DEC_DEC_DEPTH; 683 DEC_DEC_DEPTH;
534 return &PL_sv_undef; 684 return &PL_sv_undef;
535} 685}
536 686
687static void
688decode_he (dec_t *dec, HV *hv)
689{
690 // for speed reasons, we specialcase single-string
691 // byte or utf-8 strings as keys, but only when !stringref
692
693 if (ecb_expect_true (!dec->stringref))
694 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
695 {
696 I32 len = decode_uint (dec);
697 char *key = (char *)dec->cur;
698
699 dec->cur += len;
700
701 if (ecb_expect_false (dec->stringref))
702 av_push (dec->stringref, newSVpvn (key, len));
703
704 hv_store (hv, key, len, decode_sv (dec), 0);
705
706 return;
707 }
708 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
709 {
710 I32 len = decode_uint (dec);
711 char *key = (char *)dec->cur;
712
713 dec->cur += len;
714
715 if (ecb_expect_false (dec->stringref))
716 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
717
718 hv_store (hv, key, -len, decode_sv (dec), 0);
719
720 return;
721 }
722
723 SV *k = decode_sv (dec);
724 SV *v = decode_sv (dec);
725
726 hv_store_ent (hv, k, v, 0);
727 SvREFCNT_dec (k);
728}
729
537static SV * 730static SV *
538decode_hv (dec_t *dec) 731decode_hv (dec_t *dec)
539{ 732{
540 HV *hv = newHV (); 733 HV *hv = newHV ();
541 734
547 740
548 for (;;) 741 for (;;)
549 { 742 {
550 WANT (1); 743 WANT (1);
551 744
552 if (*dec->cur == 0xe0 | 31) 745 if (*dec->cur == (0xe0 | 31))
553 { 746 {
554 ++dec->cur; 747 ++dec->cur;
555 break; 748 break;
556 } 749 }
557 750
558 SV *k = decode_sv (dec); 751 decode_he (dec, hv);
559 SV *v = decode_sv (dec);
560
561 hv_store_ent (hv, k, v, 0);
562 } 752 }
563 } 753 }
564 else 754 else
565 { 755 {
566 int len = decode_uint (dec); 756 int pairs = decode_uint (dec);
567 757
568 while (len--) 758 while (pairs--)
569 { 759 decode_he (dec, hv);
570 SV *k = decode_sv (dec);
571 SV *v = decode_sv (dec);
572
573 hv_store_ent (hv, k, v, 0);
574 }
575 } 760 }
576 761
577 DEC_DEC_DEPTH; 762 DEC_DEC_DEPTH;
578 return newRV_noinc ((SV *)hv); 763 return newRV_noinc ((SV *)hv);
579
580#if 0
581 SV *sv;
582 HV *hv = newHV ();
583
584 DEC_INC_DEPTH;
585 decode_ws (dec);
586
587 for (;;)
588 {
589 // heuristic: assume that
590 // a) decode_str + hv_store_ent are abysmally slow.
591 // b) most hash keys are short, simple ascii text.
592 // => try to "fast-match" such strings to avoid
593 // the overhead of decode_str + hv_store_ent.
594 {
595 SV *value;
596 char *p = dec->cur;
597 char *e = p + 24; // only try up to 24 bytes
598
599 for (;;)
600 {
601 // the >= 0x80 is false on most architectures
602 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
603 {
604 // slow path, back up and use decode_str
605 SV *key = decode_str (dec);
606 if (!key)
607 goto fail;
608
609 decode_ws (dec); EXPECT_CH (':');
610
611 decode_ws (dec);
612 value = decode_sv (dec);
613 if (!value)
614 {
615 SvREFCNT_dec (key);
616 goto fail;
617 }
618
619 hv_store_ent (hv, key, value, 0);
620 SvREFCNT_dec (key);
621
622 break;
623 }
624 else if (*p == '"')
625 {
626 // fast path, got a simple key
627 char *key = dec->cur;
628 int len = p - key;
629 dec->cur = p + 1;
630
631 decode_ws (dec); EXPECT_CH (':');
632
633 decode_ws (dec);
634 value = decode_sv (dec);
635 if (!value)
636 goto fail;
637
638 hv_store (hv, key, len, value, 0);
639
640 break;
641 }
642
643 ++p;
644 }
645 }
646
647 decode_ws (dec);
648
649 if (*dec->cur == '}')
650 {
651 ++dec->cur;
652 break;
653 }
654
655 if (*dec->cur != ',')
656 ERR (", or } expected while parsing object/hash");
657
658 ++dec->cur;
659
660 decode_ws (dec);
661
662 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
663 {
664 ++dec->cur;
665 break;
666 }
667 }
668
669 DEC_DEC_DEPTH;
670 sv = newRV_noinc ((SV *)hv);
671
672 // check filter callbacks
673 if (dec->cbor.flags & F_HOOK)
674 {
675 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
676 {
677 HE *cb, *he;
678
679 hv_iterinit (hv);
680 he = hv_iternext (hv);
681 hv_iterinit (hv);
682
683 // the next line creates a mortal sv each time its called.
684 // might want to optimise this for common cases.
685 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
686
687 if (cb)
688 {
689 dSP;
690 int count;
691
692 ENTER; SAVETMPS; PUSHMARK (SP);
693 XPUSHs (HeVAL (he));
694 sv_2mortal (sv);
695
696 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
697
698 if (count == 1)
699 {
700 sv = newSVsv (POPs);
701 FREETMPS; LEAVE;
702 return sv;
703 }
704
705 SvREFCNT_inc (sv);
706 FREETMPS; LEAVE;
707 }
708 }
709
710 if (dec->cbor.cb_object)
711 {
712 dSP;
713 int count;
714
715 ENTER; SAVETMPS; PUSHMARK (SP);
716 XPUSHs (sv_2mortal (sv));
717
718 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
719
720 if (count == 1)
721 {
722 sv = newSVsv (POPs);
723 FREETMPS; LEAVE;
724 return sv;
725 }
726
727 SvREFCNT_inc (sv);
728 FREETMPS; LEAVE;
729 }
730 }
731
732 return sv;
733#endif
734 764
735fail: 765fail:
736 SvREFCNT_dec (hv); 766 SvREFCNT_dec (hv);
737 DEC_DEC_DEPTH; 767 DEC_DEC_DEPTH;
738 return &PL_sv_undef; 768 return &PL_sv_undef;
739} 769}
740 770
741static SV * 771static SV *
742decode_str (dec_t *dec, int utf8) 772decode_str (dec_t *dec, int utf8)
743{ 773{
744 SV *sv; 774 SV *sv = 0;
745 775
746 if ((*dec->cur & 31) == 31) 776 if ((*dec->cur & 31) == 31)
747 { 777 {
778 // indefinite length strings
748 ++dec->cur; 779 ++dec->cur;
749 780
781 unsigned char major = *dec->cur & 0xe0;
782
750 sv = newSVpvn ("", 0); 783 sv = newSVpvn ("", 0);
751 784
752 // not very fast, and certainly not robust against illegal input
753 for (;;) 785 for (;;)
754 { 786 {
755 WANT (1); 787 WANT (1);
756 788
789 if ((*dec->cur ^ major) >= 31)
757 if (*dec->cur == 0xe0 | 31) 790 if (*dec->cur == (0xe0 | 31))
758 { 791 {
759 ++dec->cur; 792 ++dec->cur;
760 break; 793 break;
761 } 794 }
795 else
796 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
762 797
763 SV *sv2 = decode_sv (dec); 798 STRLEN len = decode_uint (dec);
764 sv_catsv (sv, sv2); 799
800 WANT (len);
801 sv_catpvn (sv, dec->cur, len);
802 dec->cur += len;
765 } 803 }
766 } 804 }
767 else 805 else
768 { 806 {
769 STRLEN len = decode_uint (dec); 807 STRLEN len = decode_uint (dec);
770 808
771 WANT (len); 809 WANT (len);
772 sv = newSVpvn (dec->cur, len); 810 sv = newSVpvn (dec->cur, len);
773 dec->cur += len; 811 dec->cur += len;
812
813 if (ecb_expect_false (dec->stringref)
814 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
815 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
774 } 816 }
775 817
776 if (utf8) 818 if (utf8)
777 SvUTF8_on (sv); 819 SvUTF8_on (sv);
778 820
779 return sv; 821 return sv;
780 822
781fail: 823fail:
824 SvREFCNT_dec (sv);
825 return &PL_sv_undef;
826}
827
828static SV *
829decode_tagged (dec_t *dec)
830{
831 SV *sv = 0;
832 UV tag = decode_uint (dec);
833
834 WANT (1);
835
836 switch (tag)
837 {
838 case CBOR_TAG_MAGIC:
839 sv = decode_sv (dec);
840 break;
841
842 case CBOR_TAG_INDIRECTION:
843 sv = newRV_noinc (decode_sv (dec));
844 break;
845
846 case CBOR_TAG_STRINGREF_NAMESPACE:
847 {
848 ENTER; SAVETMPS;
849
850 SAVESPTR (dec->stringref);
851 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
852
853 sv = decode_sv (dec);
854
855 FREETMPS; LEAVE;
856 }
857 break;
858
859 case CBOR_TAG_STRINGREF:
860 {
861 if ((*dec->cur >> 5) != 0)
862 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
863
864 UV idx = decode_uint (dec);
865
866 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
867 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
868
869 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
870 }
871 break;
872
873 case CBOR_TAG_VALUE_SHAREABLE:
874 {
875 if (ecb_expect_false (!dec->shareable))
876 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
877
878 sv = newSV (0);
879 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
880
881 SV *osv = decode_sv (dec);
882 sv_setsv (sv, osv);
883 SvREFCNT_dec_NN (osv);
884 }
885 break;
886
887 case CBOR_TAG_VALUE_SHAREDREF:
888 {
889 if ((*dec->cur >> 5) != 0)
890 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
891
892 UV idx = decode_uint (dec);
893
894 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
895 ERR ("corrupted CBOR data (sharedref index out of bounds)");
896
897 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
898 }
899 break;
900
901 case CBOR_TAG_PERL_OBJECT:
902 {
903 sv = decode_sv (dec);
904
905 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
906 ERR ("corrupted CBOR data (non-array perl object)");
907
908 AV *av = (AV *)SvRV (sv);
909 int len = av_len (av) + 1;
910 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
911
912 if (!stash)
913 ERR ("cannot decode perl-object (package does not exist)");
914
915 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
916
917 if (!method)
918 ERR ("cannot decode perl-object (package does not have a THAW method)");
919
920 dSP;
921
922 ENTER; SAVETMPS; PUSHMARK (SP);
923 EXTEND (SP, len + 1);
924 // we re-bless the reference to get overload and other niceties right
925 PUSHs (*av_fetch (av, 0, 1));
926 PUSHs (sv_cbor);
927
928 int i;
929
930 for (i = 1; i < len; ++i)
931 PUSHs (*av_fetch (av, i, 1));
932
933 PUTBACK;
934 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
935 SPAGAIN;
936
937 if (SvTRUE (ERRSV))
938 {
939 FREETMPS; LEAVE;
940 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
941 }
942
943 SvREFCNT_dec (sv);
944 sv = SvREFCNT_inc (POPs);
945
946 PUTBACK;
947
948 FREETMPS; LEAVE;
949 }
950 break;
951
952 default:
953 {
954 sv = decode_sv (dec);
955
956 dSP;
957 ENTER; SAVETMPS; PUSHMARK (SP);
958 EXTEND (SP, 2);
959 PUSHs (newSVuv (tag));
960 PUSHs (sv);
961
962 PUTBACK;
963 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
964 SPAGAIN;
965
966 if (SvTRUE (ERRSV))
967 {
968 FREETMPS; LEAVE;
969 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
970 }
971
972 if (count)
973 {
974 SvREFCNT_dec (sv);
975 sv = SvREFCNT_inc (POPs);
976 }
977 else
978 {
979 AV *av = newAV ();
980 av_push (av, newSVuv (tag));
981 av_push (av, sv);
982
983 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
984 ? cbor_tagged_stash
985 : gv_stashpv ("CBOR::XS::Tagged" , 1);
986 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
987 }
988
989 PUTBACK;
990
991 FREETMPS; LEAVE;
992 }
993 break;
994 }
995
996 return sv;
997
998fail:
999 SvREFCNT_dec (sv);
782 return &PL_sv_undef; 1000 return &PL_sv_undef;
783} 1001}
784 1002
785static SV * 1003static SV *
786decode_sv (dec_t *dec) 1004decode_sv (dec_t *dec)
788 WANT (1); 1006 WANT (1);
789 1007
790 switch (*dec->cur >> 5) 1008 switch (*dec->cur >> 5)
791 { 1009 {
792 case 0: // unsigned int 1010 case 0: // unsigned int
793 //TODO: 64 bit values on 3 2bit perls
794 return newSVuv (decode_uint (dec)); 1011 return newSVuv (decode_uint (dec));
795 case 1: // negative int 1012 case 1: // negative int
796 return newSViv (-1 - (IV)decode_uint (dec)); 1013 return newSViv (-1 - (IV)decode_uint (dec));
797 case 2: // octet string 1014 case 2: // octet string
798 return decode_str (dec, 0); 1015 return decode_str (dec, 0);
801 case 4: // array 1018 case 4: // array
802 return decode_av (dec); 1019 return decode_av (dec);
803 case 5: // map 1020 case 5: // map
804 return decode_hv (dec); 1021 return decode_hv (dec);
805 case 6: // tag 1022 case 6: // tag
806 abort (); 1023 return decode_tagged (dec);
807 break;
808 case 7: // misc 1024 case 7: // misc
809 switch (*dec->cur++ & 31) 1025 switch (*dec->cur++ & 31)
810 { 1026 {
811 case 20: 1027 case 20:
812#if CBOR_SLOW 1028#if CBOR_SLOW
813 cbor_false = get_bool ("CBOR::XS::false"); 1029 types_false = get_bool ("Types::Serialiser::false");
814#endif 1030#endif
815 return newSVsv (cbor_false); 1031 return newSVsv (types_false);
816 case 21: 1032 case 21:
817#if CBOR_SLOW 1033#if CBOR_SLOW
818 cbor_true = get_bool ("CBOR::XS::true"); 1034 types_true = get_bool ("Types::Serialiser::true");
819#endif 1035#endif
820 return newSVsv (cbor_true); 1036 return newSVsv (types_true);
821 case 22: 1037 case 22:
822 return newSVsv (&PL_sv_undef); 1038 return newSVsv (&PL_sv_undef);
1039 case 23:
1040#if CBOR_SLOW
1041 types_error = get_bool ("Types::Serialiser::error");
1042#endif
1043 return newSVsv (types_error);
823 1044
824 case 25: 1045 case 25:
825 // half float
826 abort ();
827 break; 1046 {
1047 WANT (2);
1048
1049 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1050 dec->cur += 2;
1051
1052 return newSVnv (ecb_binary16_to_float (fp));
1053 }
828 1054
829 case 26: 1055 case 26:
830 { 1056 {
831 uint32_t fp; 1057 uint32_t fp;
832 WANT (4); 1058 WANT (4);
858 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1084 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
859 } 1085 }
860 1086
861 break; 1087 break;
862 } 1088 }
863#if 0
864 switch (*dec->cur)
865 {
866 //case '"': ++dec->cur; return decode_str (dec);
867 case '[': ++dec->cur; return decode_av (dec);
868 case '{': ++dec->cur; return decode_hv (dec);
869
870 case '-':
871 case '0': case '1': case '2': case '3': case '4':
872 case '5': case '6': case '7': case '8': case '9':
873 //TODO return decode_num (dec);
874
875 case 't':
876 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
877 {
878 dec->cur += 4;
879#if CBOR_SLOW
880 cbor_true = get_bool ("CBOR::XS::true");
881#endif
882 return newSVsv (cbor_true);
883 }
884 else
885 ERR ("'true' expected");
886
887 break;
888
889 case 'f':
890 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
891 {
892 dec->cur += 5;
893#if CBOR_SLOW
894 cbor_false = get_bool ("CBOR::XS::false");
895#endif
896 return newSVsv (cbor_false);
897 }
898 else
899 ERR ("'false' expected");
900
901 break;
902
903 case 'n':
904 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
905 {
906 dec->cur += 4;
907 return newSVsv (&PL_sv_undef);
908 }
909 else
910 ERR ("'null' expected");
911
912 break;
913
914 default:
915 ERR ("malformed CBOR string, neither array, object, number, string or atom");
916 break;
917 }
918#endif
919 1089
920fail: 1090fail:
921 return &PL_sv_undef; 1091 return &PL_sv_undef;
922} 1092}
923 1093
924static SV * 1094static SV *
925decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1095decode_cbor (SV *string, CBOR *cbor, char **offset_return)
926{ 1096{
927 dec_t dec; 1097 dec_t dec = { };
928 SV *sv; 1098 SV *sv;
1099 STRLEN len;
1100 char *data = SvPVbyte (string, len);
929 1101
930 /* work around bugs in 5.10 where manipulating magic values
931 * makes perl ignore the magic in subsequent accesses.
932 * also make a copy of non-PV values, to get them into a clean
933 * state (SvPV should do that, but it's buggy, see below).
934 */
935 /*SvGETMAGIC (string);*/
936 if (SvMAGICAL (string) || !SvPOK (string))
937 string = sv_2mortal (newSVsv (string));
938
939 SvUPGRADE (string, SVt_PV);
940
941 /* work around a bug in perl 5.10, which causes SvCUR to fail an
942 * assertion with -DDEBUGGING, although SvCUR is documented to
943 * return the xpv_cur field which certainly exists after upgrading.
944 * according to nicholas clark, calling SvPOK fixes this.
945 * But it doesn't fix it, so try another workaround, call SvPV_nolen
946 * and hope for the best.
947 * Damnit, SvPV_nolen still trips over yet another assertion. This
948 * assertion business is seriously broken, try yet another workaround
949 * for the broken -DDEBUGGING.
950 */
951 {
952#ifdef DEBUGGING
953 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
954#else
955 STRLEN offset = SvCUR (string);
956#endif
957
958 if (offset > cbor->max_size && cbor->max_size) 1102 if (len > cbor->max_size && cbor->max_size)
959 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1103 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
960 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1104 (unsigned long)len, (unsigned long)cbor->max_size);
961 }
962
963 sv_utf8_downgrade (string, 0);
964 1105
965 dec.cbor = *cbor; 1106 dec.cbor = *cbor;
966 dec.cur = (U8 *)SvPVX (string); 1107 dec.cur = (U8 *)data;
967 dec.end = (U8 *)SvEND (string); 1108 dec.end = (U8 *)data + len;
968 dec.err = 0;
969 dec.depth = 0;
970
971 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
972 ;//TODO dec.cbor.flags |= F_HOOK;
973 1109
974 sv = decode_sv (&dec); 1110 sv = decode_sv (&dec);
975 1111
976 if (offset_return) 1112 if (offset_return)
977 *offset_return = dec.cur; 1113 *offset_return = dec.cur;
978 1114
979 if (!(offset_return || !sv)) 1115 if (!(offset_return || !sv))
980 {
981 if (*dec.cur && !dec.err) 1116 if (dec.cur != dec.end && !dec.err)
982 {
983 dec.err = "garbage after CBOR object"; 1117 dec.err = "garbage after CBOR object";
1118
1119 if (dec.err)
1120 {
984 SvREFCNT_dec (sv); 1121 SvREFCNT_dec (sv);
985 sv = 0;
986 }
987 }
988
989 if (!sv)
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1122 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1123 }
991 1124
992 sv = sv_2mortal (sv); 1125 sv = sv_2mortal (sv);
993 1126
994 return sv; 1127 return sv;
995} 1128}
1000MODULE = CBOR::XS PACKAGE = CBOR::XS 1133MODULE = CBOR::XS PACKAGE = CBOR::XS
1001 1134
1002BOOT: 1135BOOT:
1003{ 1136{
1004 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1137 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1005 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1138 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1006 1139
1007 cbor_true = get_bool ("CBOR::XS::true"); 1140 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1008 cbor_false = get_bool ("CBOR::XS::false"); 1141 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1142
1143 types_true = get_bool ("Types::Serialiser::true" );
1144 types_false = get_bool ("Types::Serialiser::false");
1145 types_error = get_bool ("Types::Serialiser::error");
1146
1147 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1148
1149 sv_cbor = newSVpv ("CBOR", 0);
1150 SvREADONLY_on (sv_cbor);
1009} 1151}
1010 1152
1011PROTOTYPES: DISABLE 1153PROTOTYPES: DISABLE
1012 1154
1013void CLONE (...) 1155void CLONE (...)
1014 CODE: 1156 CODE:
1015 cbor_stash = 0; 1157 cbor_stash = 0;
1158 cbor_tagged_stash = 0;
1159 types_error_stash = 0;
1016 cbor_boolean_stash = 0; 1160 types_boolean_stash = 0;
1017 1161
1018void new (char *klass) 1162void new (char *klass)
1019 PPCODE: 1163 PPCODE:
1020{ 1164{
1021 SV *pv = NEWSV (0, sizeof (CBOR)); 1165 SV *pv = NEWSV (0, sizeof (CBOR));
1029 1173
1030void shrink (CBOR *self, int enable = 1) 1174void shrink (CBOR *self, int enable = 1)
1031 ALIAS: 1175 ALIAS:
1032 shrink = F_SHRINK 1176 shrink = F_SHRINK
1033 allow_unknown = F_ALLOW_UNKNOWN 1177 allow_unknown = F_ALLOW_UNKNOWN
1178 allow_sharing = F_ALLOW_SHARING
1179 pack_strings = F_PACK_STRINGS
1034 PPCODE: 1180 PPCODE:
1035{ 1181{
1036 if (enable) 1182 if (enable)
1037 self->flags |= ix; 1183 self->flags |= ix;
1038 else 1184 else
1043 1189
1044void get_shrink (CBOR *self) 1190void get_shrink (CBOR *self)
1045 ALIAS: 1191 ALIAS:
1046 get_shrink = F_SHRINK 1192 get_shrink = F_SHRINK
1047 get_allow_unknown = F_ALLOW_UNKNOWN 1193 get_allow_unknown = F_ALLOW_UNKNOWN
1194 get_allow_sharing = F_ALLOW_SHARING
1195 get_pack_strings = F_PACK_STRINGS
1048 PPCODE: 1196 PPCODE:
1049 XPUSHs (boolSV (self->flags & ix)); 1197 XPUSHs (boolSV (self->flags & ix));
1050 1198
1051void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1199void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1052 PPCODE: 1200 PPCODE:
1068 CODE: 1216 CODE:
1069 RETVAL = self->max_size; 1217 RETVAL = self->max_size;
1070 OUTPUT: 1218 OUTPUT:
1071 RETVAL 1219 RETVAL
1072 1220
1073#if 0 //TODO 1221void filter (CBOR *self, SV *filter = 0)
1074
1075void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1076 PPCODE: 1222 PPCODE:
1077{
1078 SvREFCNT_dec (self->cb_object); 1223 SvREFCNT_dec (self->filter);
1079 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1224 self->filter = filter ? newSVsv (filter) : filter;
1080
1081 XPUSHs (ST (0)); 1225 XPUSHs (ST (0));
1082}
1083 1226
1084void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1227SV *get_filter (CBOR *self)
1085 PPCODE: 1228 CODE:
1086{ 1229 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1087 if (!self->cb_sk_object) 1230 OUTPUT:
1088 self->cb_sk_object = newHV (); 1231 RETVAL
1089
1090 if (SvOK (cb))
1091 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1092 else
1093 {
1094 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1095
1096 if (!HvKEYS (self->cb_sk_object))
1097 {
1098 SvREFCNT_dec (self->cb_sk_object);
1099 self->cb_sk_object = 0;
1100 }
1101 }
1102
1103 XPUSHs (ST (0));
1104}
1105
1106#endif
1107 1232
1108void encode (CBOR *self, SV *scalar) 1233void encode (CBOR *self, SV *scalar)
1109 PPCODE: 1234 PPCODE:
1110 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1235 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1111 XPUSHs (scalar); 1236 XPUSHs (scalar);
1125 PUSHs (sv); 1250 PUSHs (sv);
1126 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1251 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1127} 1252}
1128 1253
1129void DESTROY (CBOR *self) 1254void DESTROY (CBOR *self)
1130 CODE: 1255 PPCODE:
1131 SvREFCNT_dec (self->cb_sk_object); 1256 cbor_free (self);
1132 SvREFCNT_dec (self->cb_object);
1133 1257
1134PROTOTYPES: ENABLE 1258PROTOTYPES: ENABLE
1135 1259
1136void encode_cbor (SV *scalar) 1260void encode_cbor (SV *scalar)
1137 PPCODE: 1261 PPCODE:

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines