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 1.18 by root, Wed Nov 20 01:09:46 2013 UTC vs.
Revision 1.45 by root, Sun Dec 14 05:48:39 2014 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 21#endif
21#ifndef HvNAMEUTF8 22#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 23# define HvNAMEUTF8(hv) 0
23#endif 24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
24 65
25// known tags 66// known tags
26enum cbor_tag 67enum cbor_tag
27{ 68{
28 // inofficial extensions (pending iana registration) 69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHARABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 104#define F_PACK_STRINGS 0x00000010UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 105#define F_VALIDATE_UTF8 0x00000020UL
64 106
65#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
66 108
67#define SB do { 109#define SB do {
68#define SE } while (0) 110#define SE } while (0)
80# define CBOR_SLOW 0 122# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
82#endif 124#endif
83 125
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 128
87typedef struct { 129typedef struct {
88 U32 flags; 130 U32 flags;
89 U32 max_depth; 131 U32 max_depth;
90 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
134
135 // for the incremental parser
136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 139} CBOR;
92 140
93ecb_inline void 141ecb_inline void
94cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
95{ 143{
96 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 145 cbor->max_depth = 512;
146}
147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
98} 153}
99 154
100///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
101// utility functions 156// utility functions
102 157
124 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 180#endif
126 } 181 }
127} 182}
128 183
184// minimum length of a string to be registered for stringref
185ecb_inline int
186minimum_string_length (UV idx)
187{
188 return idx > 23
189 ? idx > 0xffU
190 ? idx > 0xffffU
191 ? idx > 0xffffffffU
192 ? 11
193 : 7
194 : 5
195 : 4
196 : 3;
197}
198
129///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
130// encoder 200// encoder
131 201
132// structure used for encoding CBOR 202// structure used for encoding CBOR
133typedef struct 203typedef struct
135 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
137 SV *sv; // result scalar 207 SV *sv; // result scalar
138 CBOR cbor; 208 CBOR cbor;
139 U32 depth; // recursion level 209 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 210 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
211 UV stringref_idx;
141 HV *sharable; // ptr => index, or 0 212 HV *shareable; // ptr => index, or 0
142 HV *sharable_idx; 213 UV shareable_idx;
143} enc_t; 214} enc_t;
144 215
145ecb_inline void 216ecb_inline void
146need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
147{ 218{
164static void 235static void
165encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
166{ 237{
167 need (enc, 9); 238 need (enc, 9);
168 239
169 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
171 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
172 { 243 {
173 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 245 *enc->cur++ = len;
175 } 246 }
176 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
177 { 248 {
178 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 251 *enc->cur++ = len;
181 } 252 }
182 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
183 { 254 {
184 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 259 *enc->cur++ = len;
189 } 260 }
190 else 261 else
191 { 262 {
192 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 271 *enc->cur++ = len;
201 } 272 }
202} 273}
203 274
204static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 283{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 285 need (enc, len);
209 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
210 enc->cur += len; 287 enc->cur += len;
211} 288}
212 289
213ecb_inline void
214encode_tag (enc_t *enc, UV tag)
215{
216 encode_uint (enc, 0xc0, tag);
217}
218
219static int 290static void
220encode_sharable2 (enc_t *enc, SV *sv) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
221{ 292{
222 if (!enc->sharable) 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
223 enc->sharable = (HV *)sv_2mortal ((SV *)newHV ()); 294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
224 296
225 SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227 if (SvOK (*svp)) 297 if (SvOK (*svp))
228 { 298 {
299 // already registered, use stringref
229 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
230 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232 return 1; 302 return;
233 } 303 }
234 else 304 else if (len >= minimum_string_length (enc->stringref_idx))
235 { 305 {
306 // register only
236 sv_setuv (*svp, enc->sharable_idx++); 307 sv_setuv (*svp, enc->stringref_idx);
237 encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); 308 ++enc->stringref_idx;
238 309 }
239 return 0;
240 } 310 }
241}
242 311
243ecb_inline int 312 encode_str (enc, utf8, str, len);
244encode_sharable (enc_t *enc, SV *sv)
245{
246 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
247 && ecb_expect_false (SvREFCNT (sv) > 1))
248 return encode_sharable2 (enc, sv);
249
250 return 0;
251} 313}
252 314
253static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
254 316
255static void 317static void
260 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
262 324
263 ++enc->depth; 325 ++enc->depth;
264 326
265 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 328
329 if (SvMAGICAL (av))
267 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
268 { 331 {
269 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 encode_sv (enc, AvARRAY (av)[i]);
272 338
273 --enc->depth; 339 --enc->depth;
274} 340}
275 341
276static void 342static void
285 351
286 int pairs = hv_iterinit (hv); 352 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 353 int mg = SvMAGICAL (hv);
288 354
289 if (mg) 355 if (mg)
290 encode_ch (enc, 0xa0 | 31); 356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 357 else
292 encode_uint (enc, 0xa0, pairs); 358 encode_uint (enc, MAJOR_MAP, pairs);
293 359
294 while ((he = hv_iternext (hv))) 360 while ((he = hv_iternext (hv)))
295 { 361 {
296 if (HeKLEN (he) == HEf_SVKEY) 362 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 363 encode_sv (enc, HeSVKEY (he));
298 else 364 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 366
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 367 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 368 }
303 369
304 if (mg) 370 if (mg)
305 encode_ch (enc, 0xe0 | 31); 371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 372
307 --enc->depth; 373 --enc->depth;
308} 374}
309 375
310// encode objects, arrays and special \0=false and \1=true values. 376// encode objects, arrays and special \0=false and \1=true values.
311static void 377static void
312encode_rv (enc_t *enc, SV *sv) 378encode_rv (enc_t *enc, SV *sv)
313{ 379{
314 svtype svt;
315
316 SvGETMAGIC (sv); 380 SvGETMAGIC (sv);
381
317 svt = SvTYPE (sv); 382 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 383
322 if (ecb_expect_false (SvOBJECT (sv))) 384 if (ecb_expect_false (SvOBJECT (sv)))
323 { 385 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 386 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 387 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 393 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 395
334 HV *stash = SvSTASH (sv); 396 HV *stash = SvSTASH (sv);
335 GV *method;
336 397
337 if (stash == boolean_stash) 398 if (stash == boolean_stash)
338 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 399 {
400 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
401 return;
402 }
339 else if (stash == error_stash) 403 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
341 else if (stash == tagged_stash) 408 else if (stash == tagged_stash)
342 { 409 {
343 if (svt != SVt_PVAV) 410 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 412
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 414 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
415
416 return;
417 }
418 }
419
420 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 {
423 if (!enc->shareable)
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427
428 if (SvOK (*svp))
348 } 429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
433 }
434 else
435 {
436 sv_setuv (*svp, enc->shareable_idx);
437 ++enc->shareable_idx;
438 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
439 }
440 }
441
442 if (ecb_expect_false (SvOBJECT (sv)))
443 {
444 HV *stash = SvSTASH (sv);
445 GV *method;
446
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 448 {
351 dSP; 449 dSP;
352 450
353 ENTER; SAVETMPS; PUSHMARK (SP); 451 ENTER; SAVETMPS; PUSHMARK (SP);
354 // we re-bless the reference to get overload and other niceties right 452 // we re-bless the reference to get overload and other niceties right
386 // catch this surprisingly common error 484 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
388 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389 487
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 489 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 491
394 while (count) 492 while (count)
395 encode_sv (enc, SP[1 - count--]); 493 encode_sv (enc, SP[1 - count--]);
396 494
397 PUTBACK; 495 PUTBACK;
418{ 516{
419 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
420 518
421 need (enc, 9); 519 need (enc, 9);
422 520
423 if (ecb_expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 523 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
427 { 525 {
428 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
429 527
430 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
431 529
432 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
434 532
435 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
437 } 535 }
438 else 536 else
439 { 537 {
440 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
441 539
442 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
443 541
444 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
446 544
447 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
452static void 550static void
453encode_sv (enc_t *enc, SV *sv) 551encode_sv (enc_t *enc, SV *sv)
454{ 552{
455 SvGETMAGIC (sv); 553 SvGETMAGIC (sv);
456 554
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
461 { 556 {
462 STRLEN len; 557 STRLEN len;
463 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
465 } 560 }
466 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 562 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
469 { 564 {
470 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
472 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
474 else 569 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
476 } 571 }
477 else if (SvROK (sv)) 572 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 578 else
484 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 579 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 581}
487 582
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
497 592
498 SvPOK_only (enc.sv); 593 SvPOK_only (enc.sv);
594
595 if (cbor->flags & F_PACK_STRINGS)
596 {
597 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
598 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
599 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
600 }
601
499 encode_sv (&enc, scalar); 602 encode_sv (&enc, scalar);
500 603
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 604 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 605 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503 606
517 U8 *end; // end of input string 620 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 621 const char *err; // parse error, if != 0
519 CBOR cbor; 622 CBOR cbor;
520 U32 depth; // recursion depth 623 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 624 U32 maxdepth; // recursion depth limit
522 AV *sharable; 625 AV *shareable;
626 AV *stringref;
627 SV *decode_tagged;
523} dec_t; 628} dec_t;
524 629
525#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 630#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
526 631
527#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 632#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
530#define DEC_DEC_DEPTH --dec->depth 635#define DEC_DEC_DEPTH --dec->depth
531 636
532static UV 637static UV
533decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
534{ 639{
535 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
536 { 641 ++dec->cur;
537 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
538 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
539 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
540 return *dec->cur++ & 31;
541 642
542 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
543 WANT (2); 647 WANT (1);
544 dec->cur += 2; 648 dec->cur += 1;
545 return dec->cur[-1]; 649 return dec->cur[-1];
546 650 }
547 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
548 WANT (3); 653 WANT (2);
549 dec->cur += 3; 654 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
552 657 }
553 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
554 WANT (5); 660 WANT (4);
555 dec->cur += 5; 661 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
560 666 }
561 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
562 WANT (9); 669 WANT (8);
563 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
564 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
568 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
572 685 }
573 default: 686 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 688
577fail: 689fail:
578 return 0; 690 return 0;
579} 691}
580 692
585{ 697{
586 AV *av = newAV (); 698 AV *av = newAV ();
587 699
588 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
589 701
590 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 703 {
592 ++dec->cur; 704 ++dec->cur;
593 705
594 for (;;) 706 for (;;)
595 { 707 {
596 WANT (1); 708 WANT (1);
597 709
598 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 711 {
600 ++dec->cur; 712 ++dec->cur;
601 break; 713 break;
602 } 714 }
603 715
606 } 718 }
607 else 719 else
608 { 720 {
609 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
610 722
723 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
611 av_fill (av, len - 1); 724 av_fill (av, len - 1);
612 725
613 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
615 } 728 }
625 738
626static void 739static void
627decode_he (dec_t *dec, HV *hv) 740decode_he (dec_t *dec, HV *hv)
628{ 741{
629 // for speed reasons, we specialcase single-string 742 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 743 // byte or utf-8 strings as keys, but only when !stringref
631 744
632 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 745 if (ecb_expect_true (!dec->stringref))
746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633 { 747 {
634 I32 len = decode_uint (dec); 748 I32 len = decode_uint (dec);
635 char *key = (char *)dec->cur; 749 char *key = (char *)dec->cur;
636 750
637 dec->cur += len; 751 dec->cur += len;
638 752
639 hv_store (hv, key, len, decode_sv (dec), 0); 753 hv_store (hv, key, len, decode_sv (dec), 0);
754
755 return;
640 } 756 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 758 {
643 I32 len = decode_uint (dec); 759 I32 len = decode_uint (dec);
644 char *key = (char *)dec->cur; 760 char *key = (char *)dec->cur;
645 761
646 dec->cur += len; 762 dec->cur += len;
647 763
764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
765 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
767
648 hv_store (hv, key, -len, decode_sv (dec), 0); 768 hv_store (hv, key, -len, decode_sv (dec), 0);
769
770 return;
649 } 771 }
650 else 772
651 {
652 SV *k = decode_sv (dec); 773 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 774 SV *v = decode_sv (dec);
654 775
655 hv_store_ent (hv, k, v, 0); 776 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 777 SvREFCNT_dec (k);
657 } 778
779fail:
780 ;
658} 781}
659 782
660static SV * 783static SV *
661decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
662{ 785{
663 HV *hv = newHV (); 786 HV *hv = newHV ();
664 787
665 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
666 789
667 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 791 {
669 ++dec->cur; 792 ++dec->cur;
670 793
671 for (;;) 794 for (;;)
672 { 795 {
673 WANT (1); 796 WANT (1);
674 797
675 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
676 { 799 {
677 ++dec->cur; 800 ++dec->cur;
678 break; 801 break;
679 } 802 }
680 803
701static SV * 824static SV *
702decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
703{ 826{
704 SV *sv = 0; 827 SV *sv = 0;
705 828
706 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
707 { 830 {
831 // indefinite length strings
708 ++dec->cur; 832 ++dec->cur;
709 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
710 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
711 837
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 838 for (;;)
714 { 839 {
715 WANT (1); 840 WANT (1);
716 841
717 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 844 {
719 ++dec->cur; 845 ++dec->cur;
720 break; 846 break;
721 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 850
723 sv_catsv (sv, decode_sv (dec)); 851 STRLEN len = decode_uint (dec);
852
853 WANT (len);
854 sv_catpvn (sv, dec->cur, len);
855 dec->cur += len;
724 } 856 }
725 } 857 }
726 else 858 else
727 { 859 {
728 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
729 861
730 WANT (len); 862 WANT (len);
731 sv = newSVpvn (dec->cur, len); 863 sv = newSVpvn (dec->cur, len);
732 dec->cur += len; 864 dec->cur += len;
865
866 if (ecb_expect_false (dec->stringref)
867 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
868 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733 } 869 }
734 870
735 if (utf8) 871 if (utf8)
872 {
873 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
874 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
875 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
876
736 SvUTF8_on (sv); 877 SvUTF8_on (sv);
878 }
737 879
738 return sv; 880 return sv;
739 881
740fail: 882fail:
741 SvREFCNT_dec (sv); 883 SvREFCNT_dec (sv);
743} 885}
744 886
745static SV * 887static SV *
746decode_tagged (dec_t *dec) 888decode_tagged (dec_t *dec)
747{ 889{
890 SV *sv = 0;
748 UV tag = decode_uint (dec); 891 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 892
893 WANT (1);
750 894
751 switch (tag) 895 switch (tag)
752 { 896 {
753 case CBOR_TAG_MAGIC: 897 case CBOR_TAG_MAGIC:
754 return sv; 898 sv = decode_sv (dec);
899 break;
755 900
756 case CBOR_TAG_INDIRECTION: 901 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 902 sv = newRV_noinc (decode_sv (dec));
903 break;
758 904
905 case CBOR_TAG_STRINGREF_NAMESPACE:
906 {
907 ENTER; SAVETMPS;
908
909 SAVESPTR (dec->stringref);
910 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
911
912 sv = decode_sv (dec);
913
914 FREETMPS; LEAVE;
915 }
916 break;
917
918 case CBOR_TAG_STRINGREF:
919 {
920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
922
923 UV idx = decode_uint (dec);
924
925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
926 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
927
928 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
929 }
930 break;
931
759 case CBOR_TAG_VALUE_SHARABLE: 932 case CBOR_TAG_VALUE_SHAREABLE:
933 {
760 if (ecb_expect_false (!dec->sharable)) 934 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 935 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 936
937 if (dec->cbor.flags & F_ALLOW_CYCLES)
938 {
939 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 940 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 941
765 return sv; 942 SV *osv = decode_sv (dec);
943 sv_setsv (sv, osv);
944 SvREFCNT_dec_NN (osv);
945 }
946 else
947 {
948 av_push (dec->shareable, &PL_sv_undef);
949 int idx = AvFILLp (dec->shareable);
950 sv = decode_sv (dec);
951 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
952 }
953 }
954 break;
766 955
767 case CBOR_TAG_VALUE_SHAREDREF: 956 case CBOR_TAG_VALUE_SHAREDREF:
768 { 957 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 960
961 UV idx = decode_uint (dec);
962
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 964 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 965
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); 966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967
968 if (sv == &PL_sv_undef)
969 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
778 } 970 }
971 break;
779 972
780 case CBOR_TAG_PERL_OBJECT: 973 case CBOR_TAG_PERL_OBJECT:
781 { 974 {
975 sv = decode_sv (dec);
976
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 978 ERR ("corrupted CBOR data (non-array perl object)");
784 979
785 AV *av = (AV *)SvRV (sv); 980 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 981 int len = av_len (av) + 1;
821 sv = SvREFCNT_inc (POPs); 1016 sv = SvREFCNT_inc (POPs);
822 1017
823 PUTBACK; 1018 PUTBACK;
824 1019
825 FREETMPS; LEAVE; 1020 FREETMPS; LEAVE;
826
827 return sv;
828 } 1021 }
1022 break;
829 1023
830 default: 1024 default:
831 { 1025 {
1026 sv = decode_sv (dec);
1027
1028 dSP;
1029 ENTER; SAVETMPS; PUSHMARK (SP);
1030 EXTEND (SP, 2);
1031 PUSHs (newSVuv (tag));
1032 PUSHs (sv);
1033
1034 PUTBACK;
1035 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1036 SPAGAIN;
1037
1038 if (SvTRUE (ERRSV))
1039 {
1040 FREETMPS; LEAVE;
1041 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1042 }
1043
1044 if (count)
1045 {
1046 SvREFCNT_dec (sv);
1047 sv = SvREFCNT_inc (POPs);
1048 }
1049 else
1050 {
832 AV *av = newAV (); 1051 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1052 av_push (av, newSVuv (tag));
834 av_push (av, sv); 1053 av_push (av, sv);
835 1054
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1056 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1057 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1058 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1059 }
1060
1061 PUTBACK;
1062
1063 FREETMPS; LEAVE;
1064 }
1065 break;
842 } 1066 }
1067
1068 return sv;
843 1069
844fail: 1070fail:
845 SvREFCNT_dec (sv); 1071 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1072 return &PL_sv_undef;
847} 1073}
849static SV * 1075static SV *
850decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
851{ 1077{
852 WANT (1); 1078 WANT (1);
853 1079
854 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1081 {
856 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1089
864 case 4: // array 1090 case MAJOR_MISC >> MAJOR_SHIFT:
865 return decode_av (dec);
866 case 5: // map
867 return decode_hv (dec);
868 case 6: // tag
869 return decode_tagged (dec);
870 case 7: // misc
871 switch (*dec->cur++ & 31) 1091 switch (*dec->cur++ & MINOR_MASK)
872 { 1092 {
873 case 20: 1093 case SIMPLE_FALSE:
874#if CBOR_SLOW 1094#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
876#endif 1096#endif
877 return newSVsv (types_false); 1097 return newSVsv (types_false);
878 case 21: 1098 case SIMPLE_TRUE:
879#if CBOR_SLOW 1099#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
881#endif 1101#endif
882 return newSVsv (types_true); 1102 return newSVsv (types_true);
883 case 22: 1103 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
885 case 23: 1105 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1106#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1107 types_error = get_bool ("Types::Serialiser::error");
888#endif 1108#endif
889 return newSVsv (types_error); 1109 return newSVsv (types_error);
890 1110
891 case 25: 1111 case MISC_FLOAT16:
892 { 1112 {
893 WANT (2); 1113 WANT (2);
894 1114
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1116 dec->cur += 2;
897 1117
898 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
899 } 1119 }
900 1120
901 case 26: 1121 case MISC_FLOAT32:
902 { 1122 {
903 uint32_t fp; 1123 uint32_t fp;
904 WANT (4); 1124 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1126 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
910 1130
911 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
912 } 1132 }
913 1133
914 case 27: 1134 case MISC_FLOAT64:
915 { 1135 {
916 uint64_t fp; 1136 uint64_t fp;
917 WANT (8); 1137 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1139 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
923 1143
924 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
925 } 1145 }
926 1146
927 // 0..19 unassigned 1147 // 0..19 unassigned simple
928 // 24 reserved + unassigned (reserved values are not encodable) 1148 // 24 reserved + unassigned simple (reserved values are not encodable)
1149 // 28-30 unassigned misc
1150 // 31 break code
929 default: 1151 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1153 }
932 1154
933 break; 1155 break;
934 } 1156 }
935 1157
962 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
964 1186
965 if (dec.err) 1187 if (dec.err)
966 { 1188 {
1189 if (dec.shareable)
1190 {
1191 // need to break cyclic links, which whould all be in shareable
1192 int i;
1193 SV **svp;
1194
1195 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef);
1198 }
1199
967 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
968 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1201 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
969 } 1202 }
970 1203
971 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
972 1205
973 return sv; 1206 return sv;
974} 1207}
975 1208
1209/////////////////////////////////////////////////////////////////////////////
1210// incremental parser
1211
1212#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1213
1214// returns 0 for notyet, 1 for success or error
1215static int
1216incr_parse (CBOR *self, SV *cborstr)
1217{
1218 STRLEN cur;
1219 SvPV (cborstr, cur);
1220
1221 while (ecb_expect_true (self->incr_need <= cur))
1222 {
1223 // table of integer count bytes
1224 static I8 incr_len[MINOR_MASK + 1] = {
1225 0, 0, 0, 0, 0, 0, 0, 0,
1226 0, 0, 0, 0, 0, 0, 0, 0,
1227 0, 0, 0, 0, 0, 0, 0, 0,
1228 1, 2, 4, 8,-1,-1,-1,-2
1229 };
1230
1231 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1232 U8 m = *p & MINOR_MASK;
1233 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1234 I8 ilen = incr_len[m];
1235
1236 self->incr_need = self->incr_pos + 1;
1237
1238 if (ecb_expect_false (ilen < 0))
1239 {
1240 if (m != MINOR_INDEF)
1241 return 1; // error
1242
1243 if (*p == (MAJOR_MISC | MINOR_INDEF))
1244 {
1245 if (count >= 0)
1246 return 1; // error
1247
1248 count = 1;
1249 }
1250 else
1251 {
1252 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1253 count = -1;
1254 }
1255 }
1256 else
1257 {
1258 self->incr_need += ilen;
1259 if (ecb_expect_false (self->incr_need > cur))
1260 return 0;
1261
1262 int major = *p >> MAJOR_SHIFT;
1263
1264 switch (major)
1265 {
1266 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT:
1270 {
1271 UV len;
1272
1273 if (ecb_expect_false (ilen))
1274 {
1275 len = 0;
1276
1277 do {
1278 len = (len << 8) | *++p;
1279 } while (--ilen);
1280 }
1281 else
1282 len = m;
1283
1284 switch (major)
1285 {
1286 case MAJOR_BYTES >> MAJOR_SHIFT:
1287 case MAJOR_TEXT >> MAJOR_SHIFT:
1288 self->incr_need += len;
1289 if (ecb_expect_false (self->incr_need > cur))
1290 return 0;
1291
1292 break;
1293
1294 case MAJOR_MAP >> MAJOR_SHIFT:
1295 len <<= 1;
1296 case MAJOR_ARRAY >> MAJOR_SHIFT:
1297 if (len)
1298 {
1299 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1300 count = len + 1;
1301 }
1302 break;
1303 }
1304 }
1305 }
1306 }
1307
1308 self->incr_pos = self->incr_need;
1309
1310 if (count > 0)
1311 {
1312 while (!--count)
1313 {
1314 if (!AvFILLp (self->incr_count))
1315 return 1; // done
1316
1317 SvREFCNT_dec_NN (av_pop (self->incr_count));
1318 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1319 }
1320
1321 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1322 }
1323 }
1324
1325 return 0;
1326}
1327
1328
976///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1330// XS interface functions
978 1331
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1333
987 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1340 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
988 1341
989 types_true = get_bool ("Types::Serialiser::true" ); 1342 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1343 types_false = get_bool ("Types::Serialiser::false");
991 types_error = get_bool ("Types::Serialiser::error"); 1344 types_error = get_bool ("Types::Serialiser::error");
1345
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
992 1347
993 sv_cbor = newSVpv ("CBOR", 0); 1348 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1349 SvREADONLY_on (sv_cbor);
995} 1350}
996 1351
1018void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1374 ALIAS:
1020 shrink = F_SHRINK 1375 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1376 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1377 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1378 allow_cycles = F_ALLOW_CYCLES
1024 dedup_strings = F_DEDUP_STRINGS 1379 pack_strings = F_PACK_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1381 PPCODE:
1026{ 1382{
1027 if (enable) 1383 if (enable)
1028 self->flags |= ix; 1384 self->flags |= ix;
1029 else 1385 else
1035void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
1036 ALIAS: 1392 ALIAS:
1037 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1394 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1395 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1396 get_allow_cycles = F_ALLOW_CYCLES
1041 get_dedup_strings = F_DEDUP_STRINGS 1397 get_pack_strings = F_PACK_STRINGS
1398 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1399 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
1044 1401
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1403 PPCODE:
1059 XPUSHs (ST (0)); 1416 XPUSHs (ST (0));
1060 1417
1061int get_max_size (CBOR *self) 1418int get_max_size (CBOR *self)
1062 CODE: 1419 CODE:
1063 RETVAL = self->max_size; 1420 RETVAL = self->max_size;
1421 OUTPUT:
1422 RETVAL
1423
1424void filter (CBOR *self, SV *filter = 0)
1425 PPCODE:
1426 SvREFCNT_dec (self->filter);
1427 self->filter = filter ? newSVsv (filter) : filter;
1428 XPUSHs (ST (0));
1429
1430SV *get_filter (CBOR *self)
1431 CODE:
1432 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064 OUTPUT: 1433 OUTPUT:
1065 RETVAL 1434 RETVAL
1066 1435
1067void encode (CBOR *self, SV *scalar) 1436void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1437 PPCODE:
1083 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
1084 PUSHs (sv); 1453 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1455}
1087 1456
1457void incr_parse (CBOR *self, SV *cborstr)
1458 ALIAS:
1459 incr_parse_multiple = 1
1460 PPCODE:
1461{
1462 if (SvUTF8 (cborstr))
1463 sv_utf8_downgrade (cborstr, 0);
1464
1465 if (!self->incr_count)
1466 {
1467 self->incr_count = newAV ();
1468 self->incr_pos = 0;
1469 self->incr_need = 1;
1470
1471 av_push (self->incr_count, newSViv (1));
1472 }
1473
1474 do
1475 {
1476 if (!incr_parse (self, cborstr))
1477 {
1478 if (self->incr_need > self->max_size && self->max_size)
1479 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1480 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1481
1482 break;
1483 }
1484
1485 SV *sv;
1486 char *offset;
1487
1488 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1489 XPUSHs (sv);
1490
1491 sv_chop (cborstr, offset);
1492
1493 av_clear (self->incr_count);
1494 av_push (self->incr_count, newSViv (1));
1495
1496 self->incr_pos = 0;
1497 self->incr_need = self->incr_pos + 1;
1498 }
1499 while (ix);
1500}
1501
1502void incr_reset (CBOR *self)
1503 CODE:
1504{
1505 SvREFCNT_dec (self->incr_count);
1506 self->incr_count = 0;
1507}
1508
1509void DESTROY (CBOR *self)
1510 PPCODE:
1511 cbor_free (self);
1512
1088PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
1089 1514
1090void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1519 PPCODE:
1092{ 1520{
1093 CBOR cbor; 1521 CBOR cbor;
1094 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1525 XPUSHs (scalar);
1097} 1526}
1098 1527
1099void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines