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 1.46 by root, Sun Dec 14 05:57:22 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 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
272 341
273 --enc->depth; 342 --enc->depth;
274} 343}
275 344
276static void 345static void
285 354
286 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
288 357
289 if (mg) 358 if (mg)
290 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 360 else
292 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
293 362
294 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
295 { 364 {
296 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
298 else 367 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 369
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 371 }
303 372
304 if (mg) 373 if (mg)
305 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 375
307 --enc->depth; 376 --enc->depth;
308} 377}
309 378
310// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
311static void 380static void
312encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
313{ 382{
314 svtype svt;
315
316 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
317 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 386
322 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
323 { 388 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 390 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 396 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 398
334 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
335 GV *method;
336 400
337 if (stash == boolean_stash) 401 if (stash == boolean_stash)
338 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
339 else if (stash == error_stash) 406 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
341 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
342 { 412 {
343 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 415
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
348 } 432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
436 }
437 else
438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 451 {
351 dSP; 452 dSP;
352 453
353 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS; PUSHMARK (SP);
354 // we re-bless the reference to get overload and other niceties right 455 // we re-bless the reference to get overload and other niceties right
386 // catch this surprisingly common error 487 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 488 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)); 489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389 490
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 492 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 494
394 while (count) 495 while (count)
395 encode_sv (enc, SP[1 - count--]); 496 encode_sv (enc, SP[1 - count--]);
396 497
397 PUTBACK; 498 PUTBACK;
418{ 519{
419 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
420 521
421 need (enc, 9); 522 need (enc, 9);
422 523
423 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 526 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
427 { 528 {
428 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
429 530
430 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
431 532
432 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
434 535
435 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
437 } 538 }
438 else 539 else
439 { 540 {
440 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
441 542
442 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
443 544
444 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
446 547
447 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
452static void 553static void
453encode_sv (enc_t *enc, SV *sv) 554encode_sv (enc_t *enc, SV *sv)
454{ 555{
455 SvGETMAGIC (sv); 556 SvGETMAGIC (sv);
456 557
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
461 { 559 {
462 STRLEN len; 560 STRLEN len;
463 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
465 } 563 }
466 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 565 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
469 { 567 {
470 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
472 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
474 else 572 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
476 } 574 }
477 else if (SvROK (sv)) 575 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 581 else
484 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 582 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 584}
487 585
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
497 595
498 SvPOK_only (enc.sv); 596 SvPOK_only (enc.sv);
597
598 if (cbor->flags & F_PACK_STRINGS)
599 {
600 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
601 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
602 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
603 }
604
499 encode_sv (&enc, scalar); 605 encode_sv (&enc, scalar);
500 606
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 607 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 608 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503 609
517 U8 *end; // end of input string 623 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 624 const char *err; // parse error, if != 0
519 CBOR cbor; 625 CBOR cbor;
520 U32 depth; // recursion depth 626 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 627 U32 maxdepth; // recursion depth limit
522 AV *sharable; 628 AV *shareable;
629 AV *stringref;
630 SV *decode_tagged;
523} dec_t; 631} dec_t;
524 632
525#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 633#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
526 634
527#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 635#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 638#define DEC_DEC_DEPTH --dec->depth
531 639
532static UV 640static UV
533decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
534{ 642{
535 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
536 { 644 ++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;
541 645
542 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
543 WANT (2); 650 WANT (1);
544 dec->cur += 2; 651 dec->cur += 1;
545 return dec->cur[-1]; 652 return dec->cur[-1];
546 653 }
547 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
548 WANT (3); 656 WANT (2);
549 dec->cur += 3; 657 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
552 660 }
553 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
554 WANT (5); 663 WANT (4);
555 dec->cur += 5; 664 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
560 669 }
561 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
562 WANT (9); 672 WANT (8);
563 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
564 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
568 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
572 688 }
573 default: 689 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 691
577fail: 692fail:
578 return 0; 693 return 0;
579} 694}
580 695
585{ 700{
586 AV *av = newAV (); 701 AV *av = newAV ();
587 702
588 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
589 704
590 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 706 {
592 ++dec->cur; 707 ++dec->cur;
593 708
594 for (;;) 709 for (;;)
595 { 710 {
596 WANT (1); 711 WANT (1);
597 712
598 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 714 {
600 ++dec->cur; 715 ++dec->cur;
601 break; 716 break;
602 } 717 }
603 718
606 } 721 }
607 else 722 else
608 { 723 {
609 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
610 725
726 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); 727 av_fill (av, len - 1);
612 728
613 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
615 } 731 }
625 741
626static void 742static void
627decode_he (dec_t *dec, HV *hv) 743decode_he (dec_t *dec, HV *hv)
628{ 744{
629 // for speed reasons, we specialcase single-string 745 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 746 // byte or utf-8 strings as keys, but only when !stringref
631 747
632 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 748 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633 { 750 {
634 I32 len = decode_uint (dec); 751 I32 len = decode_uint (dec);
635 char *key = (char *)dec->cur; 752 char *key = (char *)dec->cur;
636 753
637 dec->cur += len; 754 dec->cur += len;
638 755
639 hv_store (hv, key, len, decode_sv (dec), 0); 756 hv_store (hv, key, len, decode_sv (dec), 0);
757
758 return;
640 } 759 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 760 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 761 {
643 I32 len = decode_uint (dec); 762 I32 len = decode_uint (dec);
644 char *key = (char *)dec->cur; 763 char *key = (char *)dec->cur;
645 764
646 dec->cur += len; 765 dec->cur += len;
647 766
767 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
768 if (!is_utf8_string (key, len))
769 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
770
648 hv_store (hv, key, -len, decode_sv (dec), 0); 771 hv_store (hv, key, -len, decode_sv (dec), 0);
772
773 return;
649 } 774 }
650 else 775
651 {
652 SV *k = decode_sv (dec); 776 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 777 SV *v = decode_sv (dec);
654 778
655 hv_store_ent (hv, k, v, 0); 779 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 780 SvREFCNT_dec (k);
657 } 781
782fail:
783 ;
658} 784}
659 785
660static SV * 786static SV *
661decode_hv (dec_t *dec) 787decode_hv (dec_t *dec)
662{ 788{
663 HV *hv = newHV (); 789 HV *hv = newHV ();
664 790
665 DEC_INC_DEPTH; 791 DEC_INC_DEPTH;
666 792
667 if ((*dec->cur & 31) == 31) 793 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 794 {
669 ++dec->cur; 795 ++dec->cur;
670 796
671 for (;;) 797 for (;;)
672 { 798 {
673 WANT (1); 799 WANT (1);
674 800
675 if (*dec->cur == (0xe0 | 31)) 801 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
676 { 802 {
677 ++dec->cur; 803 ++dec->cur;
678 break; 804 break;
679 } 805 }
680 806
701static SV * 827static SV *
702decode_str (dec_t *dec, int utf8) 828decode_str (dec_t *dec, int utf8)
703{ 829{
704 SV *sv = 0; 830 SV *sv = 0;
705 831
706 if ((*dec->cur & 31) == 31) 832 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
707 { 833 {
834 // indefinite length strings
708 ++dec->cur; 835 ++dec->cur;
709 836
837 U8 major = *dec->cur & MAJOR_MISC;
838
710 sv = newSVpvn ("", 0); 839 sv = newSVpvn ("", 0);
711 840
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 841 for (;;)
714 { 842 {
715 WANT (1); 843 WANT (1);
716 844
717 if (*dec->cur == (0xe0 | 31)) 845 if ((*dec->cur - major) > LENGTH_EXT8)
846 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 847 {
719 ++dec->cur; 848 ++dec->cur;
720 break; 849 break;
721 } 850 }
851 else
852 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 853
723 sv_catsv (sv, decode_sv (dec)); 854 STRLEN len = decode_uint (dec);
855
856 WANT (len);
857 sv_catpvn (sv, dec->cur, len);
858 dec->cur += len;
724 } 859 }
725 } 860 }
726 else 861 else
727 { 862 {
728 STRLEN len = decode_uint (dec); 863 STRLEN len = decode_uint (dec);
729 864
730 WANT (len); 865 WANT (len);
731 sv = newSVpvn (dec->cur, len); 866 sv = newSVpvn (dec->cur, len);
732 dec->cur += len; 867 dec->cur += len;
868
869 if (ecb_expect_false (dec->stringref)
870 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
871 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733 } 872 }
734 873
735 if (utf8) 874 if (utf8)
875 {
876 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
877 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
878 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
879
736 SvUTF8_on (sv); 880 SvUTF8_on (sv);
881 }
737 882
738 return sv; 883 return sv;
739 884
740fail: 885fail:
741 SvREFCNT_dec (sv); 886 SvREFCNT_dec (sv);
743} 888}
744 889
745static SV * 890static SV *
746decode_tagged (dec_t *dec) 891decode_tagged (dec_t *dec)
747{ 892{
893 SV *sv = 0;
748 UV tag = decode_uint (dec); 894 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 895
896 WANT (1);
750 897
751 switch (tag) 898 switch (tag)
752 { 899 {
753 case CBOR_TAG_MAGIC: 900 case CBOR_TAG_MAGIC:
754 return sv; 901 sv = decode_sv (dec);
902 break;
755 903
756 case CBOR_TAG_INDIRECTION: 904 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 905 sv = newRV_noinc (decode_sv (dec));
906 break;
758 907
908 case CBOR_TAG_STRINGREF_NAMESPACE:
909 {
910 ENTER; SAVETMPS;
911
912 SAVESPTR (dec->stringref);
913 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
914
915 sv = decode_sv (dec);
916
917 FREETMPS; LEAVE;
918 }
919 break;
920
921 case CBOR_TAG_STRINGREF:
922 {
923 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
924 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
925
926 UV idx = decode_uint (dec);
927
928 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
929 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
930
931 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
932 }
933 break;
934
759 case CBOR_TAG_VALUE_SHARABLE: 935 case CBOR_TAG_VALUE_SHAREABLE:
936 {
760 if (ecb_expect_false (!dec->sharable)) 937 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 938 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 939
940 if (dec->cbor.flags & F_ALLOW_CYCLES)
941 {
942 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 943 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 944
765 return sv; 945 SV *osv = decode_sv (dec);
946 sv_setsv (sv, osv);
947 SvREFCNT_dec_NN (osv);
948 }
949 else
950 {
951 av_push (dec->shareable, &PL_sv_undef);
952 int idx = AvFILLp (dec->shareable);
953 sv = decode_sv (dec);
954 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
955 }
956 }
957 break;
766 958
767 case CBOR_TAG_VALUE_SHAREDREF: 959 case CBOR_TAG_VALUE_SHAREDREF:
768 { 960 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 961 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 962 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 963
964 UV idx = decode_uint (dec);
965
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 966 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 967 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 968
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); 969 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
970
971 if (sv == &PL_sv_undef)
972 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
778 } 973 }
974 break;
779 975
780 case CBOR_TAG_PERL_OBJECT: 976 case CBOR_TAG_PERL_OBJECT:
781 { 977 {
978 sv = decode_sv (dec);
979
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 980 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 981 ERR ("corrupted CBOR data (non-array perl object)");
784 982
785 AV *av = (AV *)SvRV (sv); 983 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 984 int len = av_len (av) + 1;
821 sv = SvREFCNT_inc (POPs); 1019 sv = SvREFCNT_inc (POPs);
822 1020
823 PUTBACK; 1021 PUTBACK;
824 1022
825 FREETMPS; LEAVE; 1023 FREETMPS; LEAVE;
826
827 return sv;
828 } 1024 }
1025 break;
829 1026
830 default: 1027 default:
831 { 1028 {
1029 sv = decode_sv (dec);
1030
1031 dSP;
1032 ENTER; SAVETMPS; PUSHMARK (SP);
1033 EXTEND (SP, 2);
1034 PUSHs (newSVuv (tag));
1035 PUSHs (sv);
1036
1037 PUTBACK;
1038 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1039 SPAGAIN;
1040
1041 if (SvTRUE (ERRSV))
1042 {
1043 FREETMPS; LEAVE;
1044 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1045 }
1046
1047 if (count)
1048 {
1049 SvREFCNT_dec (sv);
1050 sv = SvREFCNT_inc (POPs);
1051 }
1052 else
1053 {
832 AV *av = newAV (); 1054 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1055 av_push (av, newSVuv (tag));
834 av_push (av, sv); 1056 av_push (av, sv);
835 1057
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1058 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1059 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1060 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1061 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1062 }
1063
1064 PUTBACK;
1065
1066 FREETMPS; LEAVE;
1067 }
1068 break;
842 } 1069 }
1070
1071 return sv;
843 1072
844fail: 1073fail:
845 SvREFCNT_dec (sv); 1074 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1075 return &PL_sv_undef;
847} 1076}
849static SV * 1078static SV *
850decode_sv (dec_t *dec) 1079decode_sv (dec_t *dec)
851{ 1080{
852 WANT (1); 1081 WANT (1);
853 1082
854 switch (*dec->cur >> 5) 1083 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1084 {
856 case 0: // unsigned int 1085 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1086 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1087 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1088 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1089 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1090 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1091 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1092
864 case 4: // array 1093 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) 1094 switch (*dec->cur++ & MINOR_MASK)
872 { 1095 {
873 case 20: 1096 case SIMPLE_FALSE:
874#if CBOR_SLOW 1097#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1098 types_false = get_bool ("Types::Serialiser::false");
876#endif 1099#endif
877 return newSVsv (types_false); 1100 return newSVsv (types_false);
878 case 21: 1101 case SIMPLE_TRUE:
879#if CBOR_SLOW 1102#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1103 types_true = get_bool ("Types::Serialiser::true");
881#endif 1104#endif
882 return newSVsv (types_true); 1105 return newSVsv (types_true);
883 case 22: 1106 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1107 return newSVsv (&PL_sv_undef);
885 case 23: 1108 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1109#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1110 types_error = get_bool ("Types::Serialiser::error");
888#endif 1111#endif
889 return newSVsv (types_error); 1112 return newSVsv (types_error);
890 1113
891 case 25: 1114 case MISC_FLOAT16:
892 { 1115 {
893 WANT (2); 1116 WANT (2);
894 1117
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1118 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1119 dec->cur += 2;
897 1120
898 return newSVnv (ecb_binary16_to_float (fp)); 1121 return newSVnv (ecb_binary16_to_float (fp));
899 } 1122 }
900 1123
901 case 26: 1124 case MISC_FLOAT32:
902 { 1125 {
903 uint32_t fp; 1126 uint32_t fp;
904 WANT (4); 1127 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1128 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1129 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1132 fp = ecb_bswap32 (fp);
910 1133
911 return newSVnv (ecb_binary32_to_float (fp)); 1134 return newSVnv (ecb_binary32_to_float (fp));
912 } 1135 }
913 1136
914 case 27: 1137 case MISC_FLOAT64:
915 { 1138 {
916 uint64_t fp; 1139 uint64_t fp;
917 WANT (8); 1140 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1141 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1142 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1145 fp = ecb_bswap64 (fp);
923 1146
924 return newSVnv (ecb_binary64_to_double (fp)); 1147 return newSVnv (ecb_binary64_to_double (fp));
925 } 1148 }
926 1149
927 // 0..19 unassigned 1150 // 0..19 unassigned simple
928 // 24 reserved + unassigned (reserved values are not encodable) 1151 // 24 reserved + unassigned simple (reserved values are not encodable)
1152 // 28-30 unassigned misc
1153 // 31 break code
929 default: 1154 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1155 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1156 }
932 1157
933 break; 1158 break;
934 } 1159 }
935 1160
962 if (dec.cur != dec.end && !dec.err) 1187 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1188 dec.err = "garbage after CBOR object";
964 1189
965 if (dec.err) 1190 if (dec.err)
966 { 1191 {
1192 if (dec.shareable)
1193 {
1194 // need to break cyclic links, which whould all be in shareable
1195 int i;
1196 SV **svp;
1197
1198 for (i = av_len (dec.shareable) + 1; i--; )
1199 if ((svp = av_fetch (dec.shareable, i, 0)))
1200 sv_setsv (*svp, &PL_sv_undef);
1201 }
1202
967 SvREFCNT_dec (sv); 1203 SvREFCNT_dec (sv);
968 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1204 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
969 } 1205 }
970 1206
971 sv = sv_2mortal (sv); 1207 sv = sv_2mortal (sv);
972 1208
973 return sv; 1209 return sv;
974} 1210}
975 1211
1212/////////////////////////////////////////////////////////////////////////////
1213// incremental parser
1214
1215#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1216
1217// returns 0 for notyet, 1 for success or error
1218static int
1219incr_parse (CBOR *self, SV *cborstr)
1220{
1221 STRLEN cur;
1222 SvPV (cborstr, cur);
1223
1224 while (ecb_expect_true (self->incr_need <= cur))
1225 {
1226 // table of integer count bytes
1227 static I8 incr_len[MINOR_MASK + 1] = {
1228 0, 0, 0, 0, 0, 0, 0, 0,
1229 0, 0, 0, 0, 0, 0, 0, 0,
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 1, 2, 4, 8,-1,-1,-1,-2
1232 };
1233
1234 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1235 U8 m = *p & MINOR_MASK;
1236 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1237 I8 ilen = incr_len[m];
1238
1239 self->incr_need = self->incr_pos + 1;
1240
1241 if (ecb_expect_false (ilen < 0))
1242 {
1243 if (m != MINOR_INDEF)
1244 return 1; // error
1245
1246 if (*p == (MAJOR_MISC | MINOR_INDEF))
1247 {
1248 if (count >= 0)
1249 return 1; // error
1250
1251 count = 1;
1252 }
1253 else
1254 {
1255 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1256 count = -1;
1257 }
1258 }
1259 else
1260 {
1261 self->incr_need += ilen;
1262 if (ecb_expect_false (self->incr_need > cur))
1263 return 0;
1264
1265 int major = *p >> MAJOR_SHIFT;
1266
1267 switch (major)
1268 {
1269 case MAJOR_BYTES >> MAJOR_SHIFT:
1270 case MAJOR_TEXT >> MAJOR_SHIFT:
1271 case MAJOR_ARRAY >> MAJOR_SHIFT:
1272 case MAJOR_MAP >> MAJOR_SHIFT:
1273 {
1274 UV len;
1275
1276 if (ecb_expect_false (ilen))
1277 {
1278 len = 0;
1279
1280 do {
1281 len = (len << 8) | *++p;
1282 } while (--ilen);
1283 }
1284 else
1285 len = m;
1286
1287 switch (major)
1288 {
1289 case MAJOR_BYTES >> MAJOR_SHIFT:
1290 case MAJOR_TEXT >> MAJOR_SHIFT:
1291 self->incr_need += len;
1292 if (ecb_expect_false (self->incr_need > cur))
1293 return 0;
1294
1295 break;
1296
1297 case MAJOR_MAP >> MAJOR_SHIFT:
1298 len <<= 1;
1299 case MAJOR_ARRAY >> MAJOR_SHIFT:
1300 if (len)
1301 {
1302 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1303 count = len + 1;
1304 }
1305 break;
1306 }
1307 }
1308 }
1309 }
1310
1311 self->incr_pos = self->incr_need;
1312
1313 if (count > 0)
1314 {
1315 while (!--count)
1316 {
1317 if (!AvFILLp (self->incr_count))
1318 return 1; // done
1319
1320 SvREFCNT_dec_NN (av_pop (self->incr_count));
1321 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1322 }
1323
1324 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1325 }
1326 }
1327
1328 return 0;
1329}
1330
1331
976///////////////////////////////////////////////////////////////////////////// 1332/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1333// XS interface functions
978 1334
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1335MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1336
987 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1343 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
988 1344
989 types_true = get_bool ("Types::Serialiser::true" ); 1345 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1346 types_false = get_bool ("Types::Serialiser::false");
991 types_error = get_bool ("Types::Serialiser::error"); 1347 types_error = get_bool ("Types::Serialiser::error");
1348
1349 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
992 1350
993 sv_cbor = newSVpv ("CBOR", 0); 1351 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1352 SvREADONLY_on (sv_cbor);
995} 1353}
996 1354
1018void shrink (CBOR *self, int enable = 1) 1376void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1377 ALIAS:
1020 shrink = F_SHRINK 1378 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1379 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1380 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1381 allow_cycles = F_ALLOW_CYCLES
1024 dedup_strings = F_DEDUP_STRINGS 1382 pack_strings = F_PACK_STRINGS
1383 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1384 PPCODE:
1026{ 1385{
1027 if (enable) 1386 if (enable)
1028 self->flags |= ix; 1387 self->flags |= ix;
1029 else 1388 else
1035void get_shrink (CBOR *self) 1394void get_shrink (CBOR *self)
1036 ALIAS: 1395 ALIAS:
1037 get_shrink = F_SHRINK 1396 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1397 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1398 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1399 get_allow_cycles = F_ALLOW_CYCLES
1041 get_dedup_strings = F_DEDUP_STRINGS 1400 get_pack_strings = F_PACK_STRINGS
1401 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1402 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1403 XPUSHs (boolSV (self->flags & ix));
1044 1404
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1405void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1406 PPCODE:
1059 XPUSHs (ST (0)); 1419 XPUSHs (ST (0));
1060 1420
1061int get_max_size (CBOR *self) 1421int get_max_size (CBOR *self)
1062 CODE: 1422 CODE:
1063 RETVAL = self->max_size; 1423 RETVAL = self->max_size;
1424 OUTPUT:
1425 RETVAL
1426
1427void filter (CBOR *self, SV *filter = 0)
1428 PPCODE:
1429 SvREFCNT_dec (self->filter);
1430 self->filter = filter ? newSVsv (filter) : filter;
1431 XPUSHs (ST (0));
1432
1433SV *get_filter (CBOR *self)
1434 CODE:
1435 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064 OUTPUT: 1436 OUTPUT:
1065 RETVAL 1437 RETVAL
1066 1438
1067void encode (CBOR *self, SV *scalar) 1439void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1440 PPCODE:
1083 EXTEND (SP, 2); 1455 EXTEND (SP, 2);
1084 PUSHs (sv); 1456 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1457 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1458}
1087 1459
1460void incr_parse (CBOR *self, SV *cborstr)
1461 ALIAS:
1462 incr_parse_multiple = 1
1463 PPCODE:
1464{
1465 if (SvUTF8 (cborstr))
1466 sv_utf8_downgrade (cborstr, 0);
1467
1468 if (!self->incr_count)
1469 {
1470 self->incr_count = newAV ();
1471 self->incr_pos = 0;
1472 self->incr_need = 1;
1473
1474 av_push (self->incr_count, newSViv (1));
1475 }
1476
1477 do
1478 {
1479 if (!incr_parse (self, cborstr))
1480 {
1481 if (self->incr_need > self->max_size && self->max_size)
1482 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1483 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1484
1485 break;
1486 }
1487
1488 SV *sv;
1489 char *offset;
1490
1491 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1492 XPUSHs (sv);
1493
1494 sv_chop (cborstr, offset);
1495
1496 av_clear (self->incr_count);
1497 av_push (self->incr_count, newSViv (1));
1498
1499 self->incr_pos = 0;
1500 self->incr_need = self->incr_pos + 1;
1501 }
1502 while (ix);
1503}
1504
1505void incr_reset (CBOR *self)
1506 CODE:
1507{
1508 SvREFCNT_dec (self->incr_count);
1509 self->incr_count = 0;
1510}
1511
1512void DESTROY (CBOR *self)
1513 PPCODE:
1514 cbor_free (self);
1515
1088PROTOTYPES: ENABLE 1516PROTOTYPES: ENABLE
1089 1517
1090void encode_cbor (SV *scalar) 1518void encode_cbor (SV *scalar)
1519 ALIAS:
1520 encode_cbor = 0
1521 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1522 PPCODE:
1092{ 1523{
1093 CBOR cbor; 1524 CBOR cbor;
1094 cbor_init (&cbor); 1525 cbor_init (&cbor);
1526 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1527 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1528 XPUSHs (scalar);
1097} 1529}
1098 1530
1099void decode_cbor (SV *cborstr) 1531void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines