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 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.39 by root, Sun Dec 1 14:45:03 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// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
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
27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
64
14// known tags 65// known tags
15enum cbor_tag 66enum cbor_tag
16{ 67{
17 // inofficial extensions (pending iana registration) 68 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
20 76
21 // rfc7049 77 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
28 84
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 89
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 95
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 97};
42 98
43#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
45 105
46#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
47 107
48#define SB do { 108#define SB do {
49#define SE } while (0) 109#define SE } while (0)
61# define CBOR_SLOW 0 121# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
63#endif 123#endif
64 124
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 127
68typedef struct { 128typedef struct {
69 U32 flags; 129 U32 flags;
70 U32 max_depth; 130 U32 max_depth;
71 STRLEN max_size; 131 STRLEN max_size;
132 SV *filter;
72} CBOR; 133} CBOR;
73 134
74ecb_inline void 135ecb_inline void
75cbor_init (CBOR *cbor) 136cbor_init (CBOR *cbor)
76{ 137{
77 Zero (cbor, 1, CBOR); 138 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 139 cbor->max_depth = 512;
140}
141
142ecb_inline void
143cbor_free (CBOR *cbor)
144{
145 SvREFCNT_dec (cbor->filter);
79} 146}
80 147
81///////////////////////////////////////////////////////////////////////////// 148/////////////////////////////////////////////////////////////////////////////
82// utility functions 149// utility functions
83 150
105 SvPV_renew (sv, SvCUR (sv) + 1); 172 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 173#endif
107 } 174 }
108} 175}
109 176
110///////////////////////////////////////////////////////////////////////////// 177// minimum length of a string to be registered for stringref
111// fp hell 178ecb_inline int
112 179minimum_string_length (UV idx)
113//TODO 180{
181 return idx > 23
182 ? idx > 0xffU
183 ? idx > 0xffffU
184 ? idx > 0xffffffffU
185 ? 11
186 : 7
187 : 5
188 : 4
189 : 3;
190}
114 191
115///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
116// encoder 193// encoder
117 194
118// structure used for encoding CBOR 195// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
123 SV *sv; // result scalar 200 SV *sv; // result scalar
124 CBOR cbor; 201 CBOR cbor;
125 U32 depth; // recursion level 202 U32 depth; // recursion level
203 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
204 UV stringref_idx;
205 HV *shareable; // ptr => index, or 0
206 UV shareable_idx;
126} enc_t; 207} enc_t;
127 208
128ecb_inline void 209ecb_inline void
129need (enc_t *enc, STRLEN len) 210need (enc_t *enc, STRLEN len)
130{ 211{
147static void 228static void
148encode_uint (enc_t *enc, int major, UV len) 229encode_uint (enc_t *enc, int major, UV len)
149{ 230{
150 need (enc, 9); 231 need (enc, 9);
151 232
152 if (len < 24) 233 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
154 else if (len <= 0xff) 235 else if (ecb_expect_true (len <= 0xffU))
155 { 236 {
156 *enc->cur++ = major | 24; 237 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 238 *enc->cur++ = len;
158 } 239 }
159 else if (len <= 0xffff) 240 else if (len <= 0xffffU)
160 { 241 {
161 *enc->cur++ = major | 25; 242 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 243 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 244 *enc->cur++ = len;
164 } 245 }
165 else if (len <= 0xffffffff) 246 else if (len <= 0xffffffffU)
166 { 247 {
167 *enc->cur++ = major | 26; 248 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 249 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 250 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 252 *enc->cur++ = len;
172 } 253 }
173 else 254 else
174 { 255 {
175 *enc->cur++ = major | 27; 256 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 257 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 258 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 259 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 260 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 261 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 263 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 264 *enc->cur++ = len;
184 } 265 }
185} 266}
186 267
187static void 268ecb_inline void
269encode_tag (enc_t *enc, UV tag)
270{
271 encode_uint (enc, MAJOR_TAG, tag);
272}
273
274ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 275encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 276{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 278 need (enc, len);
192 memcpy (enc->cur, str, len); 279 memcpy (enc->cur, str, len);
193 enc->cur += len; 280 enc->cur += len;
194} 281}
195 282
283static void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
285{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289
290 if (SvOK (*svp))
291 {
292 // already registered, use stringref
293 encode_tag (enc, CBOR_TAG_STRINGREF);
294 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
295 return;
296 }
297 else if (len >= minimum_string_length (enc->stringref_idx))
298 {
299 // register only
300 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx;
302 }
303 }
304
305 encode_str (enc, utf8, str, len);
306}
307
196static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
197 309
198static void 310static void
199encode_av (enc_t *enc, AV *av) 311encode_av (enc_t *enc, AV *av)
200{ 312{
203 if (enc->depth >= enc->cbor.max_depth) 315 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 316 croak (ERR_NESTING_EXCEEDED);
205 317
206 ++enc->depth; 318 ++enc->depth;
207 319
208 encode_uint (enc, 0x80, len + 1); 320 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 321
210 for (i = 0; i <= len; ++i) 322 for (i = 0; i <= len; ++i)
211 { 323 {
212 SV **svp = av_fetch (av, i, 0); 324 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 325 encode_sv (enc, svp ? *svp : &PL_sv_undef);
228 340
229 int pairs = hv_iterinit (hv); 341 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 342 int mg = SvMAGICAL (hv);
231 343
232 if (mg) 344 if (mg)
233 encode_ch (enc, 0xa0 | 31); 345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 346 else
235 encode_uint (enc, 0xa0, pairs); 347 encode_uint (enc, MAJOR_MAP, pairs);
236 348
237 while ((he = hv_iternext (hv))) 349 while ((he = hv_iternext (hv)))
238 { 350 {
239 if (HeKLEN (he) == HEf_SVKEY) 351 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 352 encode_sv (enc, HeSVKEY (he));
241 else 353 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 354 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 355
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 356 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 357 }
246 358
247 if (mg) 359 if (mg)
248 encode_ch (enc, 0xe0 | 31); 360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 361
250 --enc->depth; 362 --enc->depth;
251} 363}
252 364
253// encode objects, arrays and special \0=false and \1=true values. 365// encode objects, arrays and special \0=false and \1=true values.
254static void 366static void
255encode_rv (enc_t *enc, SV *sv) 367encode_rv (enc_t *enc, SV *sv)
256{ 368{
257 svtype svt;
258
259 SvGETMAGIC (sv); 369 SvGETMAGIC (sv);
370
260 svt = SvTYPE (sv); 371 svtype svt = SvTYPE (sv);
261 372
262 if (ecb_expect_false (SvOBJECT (sv))) 373 if (ecb_expect_false (SvOBJECT (sv)))
263 { 374 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 375 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 376 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 381 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 382 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 383 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 384
274 HV *stash = SvSTASH (sv); 385 HV *stash = SvSTASH (sv);
275 GV *method;
276 386
277 if (stash == boolean_stash) 387 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 388 {
389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
390 return;
391 }
279 else if (stash == error_stash) 392 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 393 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 return;
396 }
281 else if (stash == tagged_stash) 397 else if (stash == tagged_stash)
282 { 398 {
283 if (svt != SVt_PVAV) 399 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 401
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
404
405 return;
406 }
407 }
408
409 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 {
412 if (!enc->shareable)
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416
417 if (SvOK (*svp))
288 } 418 {
419 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
420 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
421 return;
422 }
423 else
424 {
425 sv_setuv (*svp, enc->shareable_idx);
426 ++enc->shareable_idx;
427 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
428 }
429 }
430
431 if (ecb_expect_false (SvOBJECT (sv)))
432 {
433 HV *stash = SvSTASH (sv);
434 GV *method;
435
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 437 {
291 dSP; 438 dSP;
292 439
293 ENTER; SAVETMPS; PUSHMARK (SP); 440 ENTER; SAVETMPS; PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 441 // we re-bless the reference to get overload and other niceties right
325 472
326 // catch this surprisingly common error 473 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 476
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 477 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 478 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 480
334 while (count) 481 while (count)
335 encode_sv (enc, SP[1 - count--]); 482 encode_sv (enc, SP[1 - count--]);
336 483
337 PUTBACK; 484 PUTBACK;
344 } 491 }
345 else if (svt == SVt_PVHV) 492 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 493 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 494 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 495 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 496 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 497 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 498 encode_tag (enc, CBOR_TAG_INDIRECTION);
499 encode_sv (enc, sv);
500 }
369} 501}
370 502
371static void 503static void
372encode_nv (enc_t *enc, SV *sv) 504encode_nv (enc_t *enc, SV *sv)
373{ 505{
374 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
375 507
376 need (enc, 9); 508 need (enc, 9);
377 509
378 if (ecb_expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 512 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
382 { 514 {
383 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
384 516
385 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
386 518
387 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
389 521
390 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
392 } 524 }
393 else 525 else
394 { 526 {
395 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
396 528
397 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
398 530
399 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
401 533
402 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
411 543
412 if (SvPOKp (sv)) 544 if (SvPOKp (sv))
413 { 545 {
414 STRLEN len; 546 STRLEN len;
415 char *str = SvPV (sv, len); 547 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 548 encode_strref (enc, SvUTF8 (sv), str, len);
417 } 549 }
418 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 551 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
421 { 553 {
422 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
424 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
426 else 558 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
428 } 560 }
429 else if (SvROK (sv)) 561 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 567 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 568 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 570}
439 571
440static SV * 572static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 573encode_cbor (SV *scalar, CBOR *cbor)
442{ 574{
443 enc_t enc; 575 enc_t enc = { };
444 576
445 enc.cbor = *cbor; 577 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 579 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 580 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 581
451 SvPOK_only (enc.sv); 582 SvPOK_only (enc.sv);
583
584 if (cbor->flags & F_PACK_STRINGS)
585 {
586 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
587 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
588 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
589 }
590
452 encode_sv (&enc, scalar); 591 encode_sv (&enc, scalar);
453 592
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 593 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 594 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 595
470 U8 *end; // end of input string 609 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 610 const char *err; // parse error, if != 0
472 CBOR cbor; 611 CBOR cbor;
473 U32 depth; // recursion depth 612 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 613 U32 maxdepth; // recursion depth limit
614 AV *shareable;
615 AV *stringref;
616 SV *decode_tagged;
475} dec_t; 617} dec_t;
476 618
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 619#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
478 620
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 621#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
482#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
483 625
484static UV 626static UV
485decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
486{ 628{
487 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
488 { 630 ++dec->cur;
489 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
490 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
491 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
492 return *dec->cur++ & 31;
493 631
494 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
495 WANT (2); 636 WANT (1);
496 dec->cur += 2; 637 dec->cur += 1;
497 return dec->cur[-1]; 638 return dec->cur[-1];
498 639 }
499 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
500 WANT (3); 642 WANT (2);
501 dec->cur += 3; 643 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
504 646 }
505 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
506 WANT (5); 649 WANT (4);
507 dec->cur += 5; 650 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
512 655 }
513 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
514 WANT (9); 658 WANT (8);
515 dec->cur += 9; 659 dec->cur += 8;
660
661 return
662#if UVSIZE < 8
663 0
664#else
516 return (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
669#endif
520 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
524 674 }
525 default: 675 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 677
529fail: 678fail:
530 return 0; 679 return 0;
531} 680}
532 681
537{ 686{
538 AV *av = newAV (); 687 AV *av = newAV ();
539 688
540 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
541 690
542 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 692 {
544 ++dec->cur; 693 ++dec->cur;
545 694
546 for (;;) 695 for (;;)
547 { 696 {
548 WANT (1); 697 WANT (1);
549 698
550 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
551 { 700 {
552 ++dec->cur; 701 ++dec->cur;
553 break; 702 break;
554 } 703 }
555 704
558 } 707 }
559 else 708 else
560 { 709 {
561 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
562 711
712 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
563 av_fill (av, len - 1); 713 av_fill (av, len - 1);
564 714
565 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
567 } 717 }
573 SvREFCNT_dec (av); 723 SvREFCNT_dec (av);
574 DEC_DEC_DEPTH; 724 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 725 return &PL_sv_undef;
576} 726}
577 727
728static void
729decode_he (dec_t *dec, HV *hv)
730{
731 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref
733
734 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 {
737 I32 len = decode_uint (dec);
738 char *key = (char *)dec->cur;
739
740 dec->cur += len;
741
742 hv_store (hv, key, len, decode_sv (dec), 0);
743
744 return;
745 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756
757 hv_store (hv, key, -len, decode_sv (dec), 0);
758
759 return;
760 }
761
762 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec);
764
765 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k);
767
768fail:
769 ;
770}
771
578static SV * 772static SV *
579decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
580{ 774{
581 HV *hv = newHV (); 775 HV *hv = newHV ();
582 776
583 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
584 778
585 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 780 {
587 ++dec->cur; 781 ++dec->cur;
588 782
589 for (;;) 783 for (;;)
590 { 784 {
591 WANT (1); 785 WANT (1);
592 786
593 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
594 { 788 {
595 ++dec->cur; 789 ++dec->cur;
596 break; 790 break;
597 } 791 }
598 792
599 SV *k = decode_sv (dec); 793 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 794 }
605 } 795 }
606 else 796 else
607 { 797 {
608 int len = decode_uint (dec); 798 int pairs = decode_uint (dec);
609 799
610 while (len--) 800 while (pairs--)
611 { 801 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 802 }
619 803
620 DEC_DEC_DEPTH; 804 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 805 return newRV_noinc ((SV *)hv);
622 806
629static SV * 813static SV *
630decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
631{ 815{
632 SV *sv = 0; 816 SV *sv = 0;
633 817
634 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
635 { 819 {
820 // indefinite length strings
636 ++dec->cur; 821 ++dec->cur;
637 822
823 U8 major = *dec->cur & MAJOR_MISC;
824
638 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
639 826
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 827 for (;;)
642 { 828 {
643 WANT (1); 829 WANT (1);
644 830
645 if (*dec->cur == (0xe0 | 31)) 831 if ((*dec->cur - major) > LENGTH_EXT8)
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 833 {
647 ++dec->cur; 834 ++dec->cur;
648 break; 835 break;
649 } 836 }
837 else
838 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 839
651 sv_catsv (sv, decode_sv (dec)); 840 STRLEN len = decode_uint (dec);
841
842 WANT (len);
843 sv_catpvn (sv, dec->cur, len);
844 dec->cur += len;
652 } 845 }
653 } 846 }
654 else 847 else
655 { 848 {
656 STRLEN len = decode_uint (dec); 849 STRLEN len = decode_uint (dec);
657 850
658 WANT (len); 851 WANT (len);
659 sv = newSVpvn (dec->cur, len); 852 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 853 dec->cur += len;
854
855 if (ecb_expect_false (dec->stringref)
856 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
857 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 858 }
662 859
663 if (utf8) 860 if (utf8)
861 {
862 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
863 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
864 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
865
664 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
665 868
666 return sv; 869 return sv;
667 870
668fail: 871fail:
669 SvREFCNT_dec (sv); 872 SvREFCNT_dec (sv);
671} 874}
672 875
673static SV * 876static SV *
674decode_tagged (dec_t *dec) 877decode_tagged (dec_t *dec)
675{ 878{
879 SV *sv = 0;
676 UV tag = decode_uint (dec); 880 UV tag = decode_uint (dec);
881
882 WANT (1);
883
884 switch (tag)
885 {
886 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 887 sv = decode_sv (dec);
888 break;
678 889
679 if (tag == CBOR_TAG_MAGIC) 890 case CBOR_TAG_INDIRECTION:
680 return sv; 891 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 892 break;
682 { 893
894 case CBOR_TAG_STRINGREF_NAMESPACE:
895 {
896 ENTER; SAVETMPS;
897
898 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900
901 sv = decode_sv (dec);
902
903 FREETMPS; LEAVE;
904 }
905 break;
906
907 case CBOR_TAG_STRINGREF:
908 {
909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
911
912 UV idx = decode_uint (dec);
913
914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
915 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
916
917 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
918 }
919 break;
920
921 case CBOR_TAG_VALUE_SHAREABLE:
922 {
923 if (ecb_expect_false (!dec->shareable))
924 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
925
926 if (dec->cbor.flags & F_ALLOW_CYCLES)
927 {
928 sv = newSV (0);
929 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
930
931 SV *osv = decode_sv (dec);
932 sv_setsv (sv, osv);
933 SvREFCNT_dec_NN (osv);
934 }
935 else
936 {
937 av_push (dec->shareable, &PL_sv_undef);
938 int idx = AvFILLp (dec->shareable);
939 sv = decode_sv (dec);
940 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
941 }
942 }
943 break;
944
945 case CBOR_TAG_VALUE_SHAREDREF:
946 {
947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
949
950 UV idx = decode_uint (dec);
951
952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
953 ERR ("corrupted CBOR data (sharedref index out of bounds)");
954
955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956
957 if (sv == &PL_sv_undef)
958 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
959 }
960 break;
961
962 case CBOR_TAG_PERL_OBJECT:
963 {
964 sv = decode_sv (dec);
965
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 967 ERR ("corrupted CBOR data (non-array perl object)");
685 968
686 AV *av = (AV *)SvRV (sv); 969 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 970 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 971 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 972
690 if (!stash) 973 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 974 ERR ("cannot decode perl-object (package does not exist)");
692 975
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 976 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 977
695 if (!method) 978 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 979 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 980
698 dSP; 981 dSP;
699 982
700 ENTER; SAVETMPS; PUSHMARK (SP); 983 ENTER; SAVETMPS; PUSHMARK (SP);
701 EXTEND (SP, len + 1); 984 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 985 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 986 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 987 PUSHs (sv_cbor);
705 988
706 int i; 989 int i;
707 990
708 for (i = 1; i < len; ++i) 991 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 992 PUSHs (*av_fetch (av, i, 1));
710 993
711 PUTBACK; 994 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 995 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 996 SPAGAIN;
714 997
998 if (SvTRUE (ERRSV))
999 {
1000 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1002 }
1003
1004 SvREFCNT_dec (sv);
715 sv = SvREFCNT_inc (POPs); 1005 sv = SvREFCNT_inc (POPs);
716 1006
717 PUTBACK; 1007 PUTBACK;
718 1008
719 FREETMPS; LEAVE; 1009 FREETMPS; LEAVE;
1010 }
1011 break;
720 1012
721 return sv; 1013 default:
722 } 1014 {
723 else 1015 sv = decode_sv (dec);
724 { 1016
1017 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP);
1019 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag));
1021 PUSHs (sv);
1022
1023 PUTBACK;
1024 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1025 SPAGAIN;
1026
1027 if (SvTRUE (ERRSV))
1028 {
1029 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1031 }
1032
1033 if (count)
1034 {
1035 SvREFCNT_dec (sv);
1036 sv = SvREFCNT_inc (POPs);
1037 }
1038 else
1039 {
725 AV *av = newAV (); 1040 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1041 av_push (av, newSVuv (tag));
727 av_push (av, sv); 1042 av_push (av, sv);
728 1043
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1045 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1046 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1047 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1048 }
1049
1050 PUTBACK;
1051
1052 FREETMPS; LEAVE;
1053 }
1054 break;
734 } 1055 }
1056
1057 return sv;
735 1058
736fail: 1059fail:
737 SvREFCNT_dec (sv); 1060 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1061 return &PL_sv_undef;
739} 1062}
741static SV * 1064static SV *
742decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
743{ 1066{
744 WANT (1); 1067 WANT (1);
745 1068
746 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1070 {
748 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1078
756 case 4: // array 1079 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1080 switch (*dec->cur++ & MINOR_MASK)
764 { 1081 {
765 case 20: 1082 case SIMPLE_FALSE:
766#if CBOR_SLOW 1083#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
768#endif 1085#endif
769 return newSVsv (types_false); 1086 return newSVsv (types_false);
770 case 21: 1087 case SIMPLE_TRUE:
771#if CBOR_SLOW 1088#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
773#endif 1090#endif
774 return newSVsv (types_true); 1091 return newSVsv (types_true);
775 case 22: 1092 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
777 case 23: 1094 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1095#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1096 types_error = get_bool ("Types::Serialiser::error");
780#endif 1097#endif
781 return newSVsv (types_error); 1098 return newSVsv (types_error);
782 1099
783 case 25: 1100 case MISC_FLOAT16:
784 { 1101 {
785 WANT (2); 1102 WANT (2);
786 1103
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1105 dec->cur += 2;
789 1106
790 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
791 } 1108 }
792 1109
793 case 26: 1110 case MISC_FLOAT32:
794 { 1111 {
795 uint32_t fp; 1112 uint32_t fp;
796 WANT (4); 1113 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1115 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
802 1119
803 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
804 } 1121 }
805 1122
806 case 27: 1123 case MISC_FLOAT64:
807 { 1124 {
808 uint64_t fp; 1125 uint64_t fp;
809 WANT (8); 1126 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1128 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
815 1132
816 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
817 } 1134 }
818 1135
819 // 0..19 unassigned 1136 // 0..19 unassigned simple
820 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
821 default: 1138 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
823 } 1140 }
824 1141
830} 1147}
831 1148
832static SV * 1149static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1150decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1151{
835 dec_t dec; 1152 dec_t dec = { };
836 SV *sv; 1153 SV *sv;
1154 STRLEN len;
1155 char *data = SvPVbyte (string, len);
837 1156
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1157 if (len > cbor->max_size && cbor->max_size)
867 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1158 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1159 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1160
873 dec.cbor = *cbor; 1161 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1162 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1163 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1164
879 sv = decode_sv (&dec); 1165 sv = decode_sv (&dec);
880 1166
881 if (offset_return) 1167 if (offset_return)
882 *offset_return = dec.cur; 1168 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1171 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
887 1173
888 if (dec.err) 1174 if (dec.err)
889 { 1175 {
1176 if (dec.shareable)
1177 {
1178 // need to break cyclic links, which whould all be in shareable
1179 int i;
1180 SV **svp;
1181
1182 for (i = av_len (dec.shareable) + 1; i--; )
1183 if ((svp = av_fetch (dec.shareable, i, 0)))
1184 sv_setsv (*svp, &PL_sv_undef);
1185 }
1186
890 SvREFCNT_dec (sv); 1187 SvREFCNT_dec (sv);
891 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1188 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
892 } 1189 }
893 1190
894 sv = sv_2mortal (sv); 1191 sv = sv_2mortal (sv);
895 1192
896 return sv; 1193 return sv;
910 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1207 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
911 1208
912 types_true = get_bool ("Types::Serialiser::true" ); 1209 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1210 types_false = get_bool ("Types::Serialiser::false");
914 types_error = get_bool ("Types::Serialiser::error"); 1211 types_error = get_bool ("Types::Serialiser::error");
1212
1213 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
915 1214
916 sv_cbor = newSVpv ("CBOR", 0); 1215 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1216 SvREADONLY_on (sv_cbor);
918} 1217}
919 1218
940 1239
941void shrink (CBOR *self, int enable = 1) 1240void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1241 ALIAS:
943 shrink = F_SHRINK 1242 shrink = F_SHRINK
944 allow_unknown = F_ALLOW_UNKNOWN 1243 allow_unknown = F_ALLOW_UNKNOWN
1244 allow_sharing = F_ALLOW_SHARING
1245 allow_cycles = F_ALLOW_CYCLES
1246 pack_strings = F_PACK_STRINGS
1247 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1248 PPCODE:
946{ 1249{
947 if (enable) 1250 if (enable)
948 self->flags |= ix; 1251 self->flags |= ix;
949 else 1252 else
954 1257
955void get_shrink (CBOR *self) 1258void get_shrink (CBOR *self)
956 ALIAS: 1259 ALIAS:
957 get_shrink = F_SHRINK 1260 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1261 get_allow_unknown = F_ALLOW_UNKNOWN
1262 get_allow_sharing = F_ALLOW_SHARING
1263 get_allow_cycles = F_ALLOW_CYCLES
1264 get_pack_strings = F_PACK_STRINGS
1265 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1266 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1267 XPUSHs (boolSV (self->flags & ix));
961 1268
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1269void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1270 PPCODE:
979 CODE: 1286 CODE:
980 RETVAL = self->max_size; 1287 RETVAL = self->max_size;
981 OUTPUT: 1288 OUTPUT:
982 RETVAL 1289 RETVAL
983 1290
984#if 0 //TODO 1291void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1292 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1293 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1294 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1295 XPUSHs (ST (0));
993}
994 1296
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1297SV *get_filter (CBOR *self)
996 PPCODE: 1298 CODE:
997{ 1299 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1300 OUTPUT:
999 self->cb_sk_object = newHV (); 1301 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1302
1019void encode (CBOR *self, SV *scalar) 1303void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1304 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1305 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1306 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1319 EXTEND (SP, 2);
1036 PUSHs (sv); 1320 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1321 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1322}
1039 1323
1040#if 0
1041
1042void DESTROY (CBOR *self) 1324void DESTROY (CBOR *self)
1043 CODE: 1325 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1326 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1327
1049PROTOTYPES: ENABLE 1328PROTOTYPES: ENABLE
1050 1329
1051void encode_cbor (SV *scalar) 1330void encode_cbor (SV *scalar)
1331 ALIAS:
1332 encode_cbor = 0
1333 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1334 PPCODE:
1053{ 1335{
1054 CBOR cbor; 1336 CBOR cbor;
1055 cbor_init (&cbor); 1337 cbor_init (&cbor);
1338 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1339 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1340 XPUSHs (scalar);
1058} 1341}
1059 1342
1060void decode_cbor (SV *cborstr) 1343void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines