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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.40 by root, Tue Dec 10 13:54:40 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
64
14// known tags 65// known tags
15enum cbor_tag 66enum cbor_tag
16{ 67{
17 // inofficial extensions (pending iana registration) 68 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
20 76
21 // rfc7049 77 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
28 84
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 89
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 95
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 97};
42 98
43#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
45 105
46#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
47 107
48#define SB do { 108#define SB do {
49#define SE } while (0) 109#define SE } while (0)
61# define CBOR_SLOW 0 121# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
63#endif 123#endif
64 124
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 127
68typedef struct { 128typedef struct {
69 U32 flags; 129 U32 flags;
70 U32 max_depth; 130 U32 max_depth;
71 STRLEN max_size; 131 STRLEN max_size;
132 SV *filter;
133
134 // for the incremental parser
135 STRLEN incr_pos; // the current offset into the text
136 STRLEN incr_need; // minimum bytes needed to decode
137 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 138} CBOR;
73 139
74ecb_inline void 140ecb_inline void
75cbor_init (CBOR *cbor) 141cbor_init (CBOR *cbor)
76{ 142{
77 Zero (cbor, 1, CBOR); 143 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 144 cbor->max_depth = 512;
145}
146
147ecb_inline void
148cbor_free (CBOR *cbor)
149{
150 SvREFCNT_dec (cbor->filter);
151 SvREFCNT_dec (cbor->incr_count);
79} 152}
80 153
81///////////////////////////////////////////////////////////////////////////// 154/////////////////////////////////////////////////////////////////////////////
82// utility functions 155// utility functions
83 156
105 SvPV_renew (sv, SvCUR (sv) + 1); 178 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 179#endif
107 } 180 }
108} 181}
109 182
110///////////////////////////////////////////////////////////////////////////// 183// minimum length of a string to be registered for stringref
111// fp hell 184ecb_inline int
112 185minimum_string_length (UV idx)
113//TODO 186{
187 return idx > 23
188 ? idx > 0xffU
189 ? idx > 0xffffU
190 ? idx > 0xffffffffU
191 ? 11
192 : 7
193 : 5
194 : 4
195 : 3;
196}
114 197
115///////////////////////////////////////////////////////////////////////////// 198/////////////////////////////////////////////////////////////////////////////
116// encoder 199// encoder
117 200
118// structure used for encoding CBOR 201// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 204 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 205 char *end; // SvEND (sv)
123 SV *sv; // result scalar 206 SV *sv; // result scalar
124 CBOR cbor; 207 CBOR cbor;
125 U32 depth; // recursion level 208 U32 depth; // recursion level
209 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
210 UV stringref_idx;
211 HV *shareable; // ptr => index, or 0
212 UV shareable_idx;
126} enc_t; 213} enc_t;
127 214
128ecb_inline void 215ecb_inline void
129need (enc_t *enc, STRLEN len) 216need (enc_t *enc, STRLEN len)
130{ 217{
147static void 234static void
148encode_uint (enc_t *enc, int major, UV len) 235encode_uint (enc_t *enc, int major, UV len)
149{ 236{
150 need (enc, 9); 237 need (enc, 9);
151 238
152 if (len < 24) 239 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 240 *enc->cur++ = major | len;
154 else if (len <= 0xff) 241 else if (ecb_expect_true (len <= 0xffU))
155 { 242 {
156 *enc->cur++ = major | 24; 243 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 244 *enc->cur++ = len;
158 } 245 }
159 else if (len <= 0xffff) 246 else if (len <= 0xffffU)
160 { 247 {
161 *enc->cur++ = major | 25; 248 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 250 *enc->cur++ = len;
164 } 251 }
165 else if (len <= 0xffffffff) 252 else if (len <= 0xffffffffU)
166 { 253 {
167 *enc->cur++ = major | 26; 254 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 255 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 256 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 257 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 258 *enc->cur++ = len;
172 } 259 }
173 else 260 else
174 { 261 {
175 *enc->cur++ = major | 27; 262 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 263 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 264 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 265 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 266 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 267 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 269 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 270 *enc->cur++ = len;
184 } 271 }
185} 272}
186 273
187static void 274ecb_inline void
275encode_tag (enc_t *enc, UV tag)
276{
277 encode_uint (enc, MAJOR_TAG, tag);
278}
279
280ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 281encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 282{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 283 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 284 need (enc, len);
192 memcpy (enc->cur, str, len); 285 memcpy (enc->cur, str, len);
193 enc->cur += len; 286 enc->cur += len;
194} 287}
195 288
289static void
290encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
291{
292 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
293 {
294 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
295
296 if (SvOK (*svp))
297 {
298 // already registered, use stringref
299 encode_tag (enc, CBOR_TAG_STRINGREF);
300 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
301 return;
302 }
303 else if (len >= minimum_string_length (enc->stringref_idx))
304 {
305 // register only
306 sv_setuv (*svp, enc->stringref_idx);
307 ++enc->stringref_idx;
308 }
309 }
310
311 encode_str (enc, utf8, str, len);
312}
313
196static void encode_sv (enc_t *enc, SV *sv); 314static void encode_sv (enc_t *enc, SV *sv);
197 315
198static void 316static void
199encode_av (enc_t *enc, AV *av) 317encode_av (enc_t *enc, AV *av)
200{ 318{
203 if (enc->depth >= enc->cbor.max_depth) 321 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 322 croak (ERR_NESTING_EXCEEDED);
205 323
206 ++enc->depth; 324 ++enc->depth;
207 325
208 encode_uint (enc, 0x80, len + 1); 326 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 327
210 for (i = 0; i <= len; ++i) 328 for (i = 0; i <= len; ++i)
211 { 329 {
212 SV **svp = av_fetch (av, i, 0); 330 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 331 encode_sv (enc, svp ? *svp : &PL_sv_undef);
228 346
229 int pairs = hv_iterinit (hv); 347 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 348 int mg = SvMAGICAL (hv);
231 349
232 if (mg) 350 if (mg)
233 encode_ch (enc, 0xa0 | 31); 351 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 352 else
235 encode_uint (enc, 0xa0, pairs); 353 encode_uint (enc, MAJOR_MAP, pairs);
236 354
237 while ((he = hv_iternext (hv))) 355 while ((he = hv_iternext (hv)))
238 { 356 {
239 if (HeKLEN (he) == HEf_SVKEY) 357 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 358 encode_sv (enc, HeSVKEY (he));
241 else 359 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 360 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 361
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 362 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 363 }
246 364
247 if (mg) 365 if (mg)
248 encode_ch (enc, 0xe0 | 31); 366 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 367
250 --enc->depth; 368 --enc->depth;
251} 369}
252 370
253// encode objects, arrays and special \0=false and \1=true values. 371// encode objects, arrays and special \0=false and \1=true values.
254static void 372static void
255encode_rv (enc_t *enc, SV *sv) 373encode_rv (enc_t *enc, SV *sv)
256{ 374{
257 svtype svt;
258
259 SvGETMAGIC (sv); 375 SvGETMAGIC (sv);
376
260 svt = SvTYPE (sv); 377 svtype svt = SvTYPE (sv);
261 378
262 if (ecb_expect_false (SvOBJECT (sv))) 379 if (ecb_expect_false (SvOBJECT (sv)))
263 { 380 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 381 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 382 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 387 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 388 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 389 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 390
274 HV *stash = SvSTASH (sv); 391 HV *stash = SvSTASH (sv);
275 GV *method;
276 392
277 if (stash == boolean_stash) 393 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 394 {
395 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
396 return;
397 }
279 else if (stash == error_stash) 398 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 399 {
400 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
401 return;
402 }
281 else if (stash == tagged_stash) 403 else if (stash == tagged_stash)
282 { 404 {
283 if (svt != SVt_PVAV) 405 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 406 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 407
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 408 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 409 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
410
411 return;
412 }
413 }
414
415 if (ecb_expect_false (SvREFCNT (sv) > 1)
416 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
417 {
418 if (!enc->shareable)
419 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
420
421 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
422
423 if (SvOK (*svp))
288 } 424 {
425 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
426 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
427 return;
428 }
429 else
430 {
431 sv_setuv (*svp, enc->shareable_idx);
432 ++enc->shareable_idx;
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
434 }
435 }
436
437 if (ecb_expect_false (SvOBJECT (sv)))
438 {
439 HV *stash = SvSTASH (sv);
440 GV *method;
441
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 442 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 443 {
291 dSP; 444 dSP;
292 445
293 ENTER; SAVETMPS; PUSHMARK (SP); 446 ENTER; SAVETMPS; PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 447 // we re-bless the reference to get overload and other niceties right
325 478
326 // catch this surprisingly common error 479 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 480 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 481 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 482
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 483 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 484 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 485 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 486
334 while (count) 487 while (count)
335 encode_sv (enc, SP[1 - count--]); 488 encode_sv (enc, SP[1 - count--]);
336 489
337 PUTBACK; 490 PUTBACK;
344 } 497 }
345 else if (svt == SVt_PVHV) 498 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 499 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 500 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 501 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 502 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 503 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 504 encode_tag (enc, CBOR_TAG_INDIRECTION);
505 encode_sv (enc, sv);
506 }
369} 507}
370 508
371static void 509static void
372encode_nv (enc_t *enc, SV *sv) 510encode_nv (enc_t *enc, SV *sv)
373{ 511{
374 double nv = SvNVX (sv); 512 double nv = SvNVX (sv);
375 513
376 need (enc, 9); 514 need (enc, 9);
377 515
378 if (ecb_expect_false (nv == (U32)nv)) 516 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 517 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 518 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 519 else if (ecb_expect_false (nv == (float)nv))
382 { 520 {
383 uint32_t fp = ecb_float_to_binary32 (nv); 521 uint32_t fp = ecb_float_to_binary32 (nv);
384 522
385 *enc->cur++ = 0xe0 | 26; 523 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
386 524
387 if (!ecb_big_endian ()) 525 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp); 526 fp = ecb_bswap32 (fp);
389 527
390 memcpy (enc->cur, &fp, 4); 528 memcpy (enc->cur, &fp, 4);
392 } 530 }
393 else 531 else
394 { 532 {
395 uint64_t fp = ecb_double_to_binary64 (nv); 533 uint64_t fp = ecb_double_to_binary64 (nv);
396 534
397 *enc->cur++ = 0xe0 | 27; 535 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
398 536
399 if (!ecb_big_endian ()) 537 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp); 538 fp = ecb_bswap64 (fp);
401 539
402 memcpy (enc->cur, &fp, 8); 540 memcpy (enc->cur, &fp, 8);
411 549
412 if (SvPOKp (sv)) 550 if (SvPOKp (sv))
413 { 551 {
414 STRLEN len; 552 STRLEN len;
415 char *str = SvPV (sv, len); 553 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 554 encode_strref (enc, SvUTF8 (sv), str, len);
417 } 555 }
418 else if (SvNOKp (sv)) 556 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 557 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 558 else if (SvIOKp (sv))
421 { 559 {
422 if (SvIsUV (sv)) 560 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv)); 561 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
424 else if (SvIVX (sv) >= 0) 562 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv)); 563 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
426 else 564 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 565 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
428 } 566 }
429 else if (SvROK (sv)) 567 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 568 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 569 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 570 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 571 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 572 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 573 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 574 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 575 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 576}
439 577
440static SV * 578static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 579encode_cbor (SV *scalar, CBOR *cbor)
442{ 580{
443 enc_t enc; 581 enc_t enc = { };
444 582
445 enc.cbor = *cbor; 583 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 584 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 585 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 586 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 587
451 SvPOK_only (enc.sv); 588 SvPOK_only (enc.sv);
589
590 if (cbor->flags & F_PACK_STRINGS)
591 {
592 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
593 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
594 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
595 }
596
452 encode_sv (&enc, scalar); 597 encode_sv (&enc, scalar);
453 598
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 599 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 600 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 601
470 U8 *end; // end of input string 615 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 616 const char *err; // parse error, if != 0
472 CBOR cbor; 617 CBOR cbor;
473 U32 depth; // recursion depth 618 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 619 U32 maxdepth; // recursion depth limit
620 AV *shareable;
621 AV *stringref;
622 SV *decode_tagged;
475} dec_t; 623} dec_t;
476 624
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 625#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
478 626
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 627#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
482#define DEC_DEC_DEPTH --dec->depth 630#define DEC_DEC_DEPTH --dec->depth
483 631
484static UV 632static UV
485decode_uint (dec_t *dec) 633decode_uint (dec_t *dec)
486{ 634{
487 switch (*dec->cur & 31) 635 U8 m = *dec->cur & MINOR_MASK;
488 { 636 ++dec->cur;
489 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
490 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
491 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
492 return *dec->cur++ & 31;
493 637
494 case 24: 638 if (ecb_expect_true (m < LENGTH_EXT1))
639 return m;
640 else if (ecb_expect_true (m == LENGTH_EXT1))
641 {
495 WANT (2); 642 WANT (1);
496 dec->cur += 2; 643 dec->cur += 1;
497 return dec->cur[-1]; 644 return dec->cur[-1];
498 645 }
499 case 25: 646 else if (ecb_expect_true (m == LENGTH_EXT2))
647 {
500 WANT (3); 648 WANT (2);
501 dec->cur += 3; 649 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 650 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 651 | ((UV)dec->cur[-1]);
504 652 }
505 case 26: 653 else if (ecb_expect_true (m == LENGTH_EXT4))
654 {
506 WANT (5); 655 WANT (4);
507 dec->cur += 5; 656 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 657 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 658 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 659 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 660 | ((UV)dec->cur[-1]);
512 661 }
513 case 27: 662 else if (ecb_expect_true (m == LENGTH_EXT8))
663 {
514 WANT (9); 664 WANT (8);
515 dec->cur += 9; 665 dec->cur += 8;
666
667 return
668#if UVSIZE < 8
669 0
670#else
516 return (((UV)dec->cur[-8]) << 56) 671 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 672 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 673 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 674 | (((UV)dec->cur[-5]) << 32)
675#endif
520 | (((UV)dec->cur[-4]) << 24) 676 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 677 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 678 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 679 | ((UV)dec->cur[-1]);
524 680 }
525 default: 681 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 682 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 683
529fail: 684fail:
530 return 0; 685 return 0;
531} 686}
532 687
537{ 692{
538 AV *av = newAV (); 693 AV *av = newAV ();
539 694
540 DEC_INC_DEPTH; 695 DEC_INC_DEPTH;
541 696
542 if ((*dec->cur & 31) == 31) 697 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 698 {
544 ++dec->cur; 699 ++dec->cur;
545 700
546 for (;;) 701 for (;;)
547 { 702 {
548 WANT (1); 703 WANT (1);
549 704
550 if (*dec->cur == (0xe0 | 31)) 705 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
551 { 706 {
552 ++dec->cur; 707 ++dec->cur;
553 break; 708 break;
554 } 709 }
555 710
558 } 713 }
559 else 714 else
560 { 715 {
561 int i, len = decode_uint (dec); 716 int i, len = decode_uint (dec);
562 717
718 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
563 av_fill (av, len - 1); 719 av_fill (av, len - 1);
564 720
565 for (i = 0; i < len; ++i) 721 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 722 AvARRAY (av)[i] = decode_sv (dec);
567 } 723 }
573 SvREFCNT_dec (av); 729 SvREFCNT_dec (av);
574 DEC_DEC_DEPTH; 730 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 731 return &PL_sv_undef;
576} 732}
577 733
734static void
735decode_he (dec_t *dec, HV *hv)
736{
737 // for speed reasons, we specialcase single-string
738 // byte or utf-8 strings as keys, but only when !stringref
739
740 if (ecb_expect_true (!dec->stringref))
741 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
742 {
743 I32 len = decode_uint (dec);
744 char *key = (char *)dec->cur;
745
746 dec->cur += len;
747
748 hv_store (hv, key, len, decode_sv (dec), 0);
749
750 return;
751 }
752 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
753 {
754 I32 len = decode_uint (dec);
755 char *key = (char *)dec->cur;
756
757 dec->cur += len;
758
759 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
760 if (!is_utf8_string (key, len))
761 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
762
763 hv_store (hv, key, -len, decode_sv (dec), 0);
764
765 return;
766 }
767
768 SV *k = decode_sv (dec);
769 SV *v = decode_sv (dec);
770
771 hv_store_ent (hv, k, v, 0);
772 SvREFCNT_dec (k);
773
774fail:
775 ;
776}
777
578static SV * 778static SV *
579decode_hv (dec_t *dec) 779decode_hv (dec_t *dec)
580{ 780{
581 HV *hv = newHV (); 781 HV *hv = newHV ();
582 782
583 DEC_INC_DEPTH; 783 DEC_INC_DEPTH;
584 784
585 if ((*dec->cur & 31) == 31) 785 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 786 {
587 ++dec->cur; 787 ++dec->cur;
588 788
589 for (;;) 789 for (;;)
590 { 790 {
591 WANT (1); 791 WANT (1);
592 792
593 if (*dec->cur == (0xe0 | 31)) 793 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
594 { 794 {
595 ++dec->cur; 795 ++dec->cur;
596 break; 796 break;
597 } 797 }
598 798
599 SV *k = decode_sv (dec); 799 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 800 }
605 } 801 }
606 else 802 else
607 { 803 {
608 int len = decode_uint (dec); 804 int pairs = decode_uint (dec);
609 805
610 while (len--) 806 while (pairs--)
611 { 807 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 808 }
619 809
620 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 811 return newRV_noinc ((SV *)hv);
622 812
629static SV * 819static SV *
630decode_str (dec_t *dec, int utf8) 820decode_str (dec_t *dec, int utf8)
631{ 821{
632 SV *sv = 0; 822 SV *sv = 0;
633 823
634 if ((*dec->cur & 31) == 31) 824 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
635 { 825 {
826 // indefinite length strings
636 ++dec->cur; 827 ++dec->cur;
637 828
829 U8 major = *dec->cur & MAJOR_MISC;
830
638 sv = newSVpvn ("", 0); 831 sv = newSVpvn ("", 0);
639 832
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 833 for (;;)
642 { 834 {
643 WANT (1); 835 WANT (1);
644 836
645 if (*dec->cur == (0xe0 | 31)) 837 if ((*dec->cur - major) > LENGTH_EXT8)
838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 839 {
647 ++dec->cur; 840 ++dec->cur;
648 break; 841 break;
649 } 842 }
843 else
844 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 845
651 sv_catsv (sv, decode_sv (dec)); 846 STRLEN len = decode_uint (dec);
847
848 WANT (len);
849 sv_catpvn (sv, dec->cur, len);
850 dec->cur += len;
652 } 851 }
653 } 852 }
654 else 853 else
655 { 854 {
656 STRLEN len = decode_uint (dec); 855 STRLEN len = decode_uint (dec);
657 856
658 WANT (len); 857 WANT (len);
659 sv = newSVpvn (dec->cur, len); 858 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 859 dec->cur += len;
860
861 if (ecb_expect_false (dec->stringref)
862 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
863 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 864 }
662 865
663 if (utf8) 866 if (utf8)
867 {
868 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
869 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
870 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
871
664 SvUTF8_on (sv); 872 SvUTF8_on (sv);
873 }
665 874
666 return sv; 875 return sv;
667 876
668fail: 877fail:
669 SvREFCNT_dec (sv); 878 SvREFCNT_dec (sv);
671} 880}
672 881
673static SV * 882static SV *
674decode_tagged (dec_t *dec) 883decode_tagged (dec_t *dec)
675{ 884{
885 SV *sv = 0;
676 UV tag = decode_uint (dec); 886 UV tag = decode_uint (dec);
887
888 WANT (1);
889
890 switch (tag)
891 {
892 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 893 sv = decode_sv (dec);
894 break;
678 895
679 if (tag == CBOR_TAG_MAGIC) 896 case CBOR_TAG_INDIRECTION:
680 return sv; 897 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 898 break;
682 { 899
900 case CBOR_TAG_STRINGREF_NAMESPACE:
901 {
902 ENTER; SAVETMPS;
903
904 SAVESPTR (dec->stringref);
905 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
906
907 sv = decode_sv (dec);
908
909 FREETMPS; LEAVE;
910 }
911 break;
912
913 case CBOR_TAG_STRINGREF:
914 {
915 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
916 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
917
918 UV idx = decode_uint (dec);
919
920 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
921 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
922
923 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
924 }
925 break;
926
927 case CBOR_TAG_VALUE_SHAREABLE:
928 {
929 if (ecb_expect_false (!dec->shareable))
930 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
931
932 if (dec->cbor.flags & F_ALLOW_CYCLES)
933 {
934 sv = newSV (0);
935 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
936
937 SV *osv = decode_sv (dec);
938 sv_setsv (sv, osv);
939 SvREFCNT_dec_NN (osv);
940 }
941 else
942 {
943 av_push (dec->shareable, &PL_sv_undef);
944 int idx = AvFILLp (dec->shareable);
945 sv = decode_sv (dec);
946 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
947 }
948 }
949 break;
950
951 case CBOR_TAG_VALUE_SHAREDREF:
952 {
953 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
954 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
955
956 UV idx = decode_uint (dec);
957
958 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
959 ERR ("corrupted CBOR data (sharedref index out of bounds)");
960
961 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
962
963 if (sv == &PL_sv_undef)
964 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
965 }
966 break;
967
968 case CBOR_TAG_PERL_OBJECT:
969 {
970 sv = decode_sv (dec);
971
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 972 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 973 ERR ("corrupted CBOR data (non-array perl object)");
685 974
686 AV *av = (AV *)SvRV (sv); 975 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 976 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 977 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 978
690 if (!stash) 979 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 980 ERR ("cannot decode perl-object (package does not exist)");
692 981
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 982 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 983
695 if (!method) 984 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 985 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 986
698 dSP; 987 dSP;
699 988
700 ENTER; SAVETMPS; PUSHMARK (SP); 989 ENTER; SAVETMPS; PUSHMARK (SP);
701 EXTEND (SP, len + 1); 990 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 991 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 992 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 993 PUSHs (sv_cbor);
705 994
706 int i; 995 int i;
707 996
708 for (i = 1; i < len; ++i) 997 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 998 PUSHs (*av_fetch (av, i, 1));
710 999
711 PUTBACK; 1000 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1001 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1002 SPAGAIN;
714 1003
1004 if (SvTRUE (ERRSV))
1005 {
1006 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1008 }
1009
1010 SvREFCNT_dec (sv);
715 sv = SvREFCNT_inc (POPs); 1011 sv = SvREFCNT_inc (POPs);
716 1012
717 PUTBACK; 1013 PUTBACK;
718 1014
719 FREETMPS; LEAVE; 1015 FREETMPS; LEAVE;
1016 }
1017 break;
720 1018
721 return sv; 1019 default:
722 } 1020 {
723 else 1021 sv = decode_sv (dec);
724 { 1022
1023 dSP;
1024 ENTER; SAVETMPS; PUSHMARK (SP);
1025 EXTEND (SP, 2);
1026 PUSHs (newSVuv (tag));
1027 PUSHs (sv);
1028
1029 PUTBACK;
1030 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1031 SPAGAIN;
1032
1033 if (SvTRUE (ERRSV))
1034 {
1035 FREETMPS; LEAVE;
1036 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1037 }
1038
1039 if (count)
1040 {
1041 SvREFCNT_dec (sv);
1042 sv = SvREFCNT_inc (POPs);
1043 }
1044 else
1045 {
725 AV *av = newAV (); 1046 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1047 av_push (av, newSVuv (tag));
727 av_push (av, sv); 1048 av_push (av, sv);
728 1049
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1050 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1051 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1052 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1053 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1054 }
1055
1056 PUTBACK;
1057
1058 FREETMPS; LEAVE;
1059 }
1060 break;
734 } 1061 }
1062
1063 return sv;
735 1064
736fail: 1065fail:
737 SvREFCNT_dec (sv); 1066 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1067 return &PL_sv_undef;
739} 1068}
741static SV * 1070static SV *
742decode_sv (dec_t *dec) 1071decode_sv (dec_t *dec)
743{ 1072{
744 WANT (1); 1073 WANT (1);
745 1074
746 switch (*dec->cur >> 5) 1075 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1076 {
748 case 0: // unsigned int 1077 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1078 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1079 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1080 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1081 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1082 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1083 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1084
756 case 4: // array 1085 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1086 switch (*dec->cur++ & MINOR_MASK)
764 { 1087 {
765 case 20: 1088 case SIMPLE_FALSE:
766#if CBOR_SLOW 1089#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1090 types_false = get_bool ("Types::Serialiser::false");
768#endif 1091#endif
769 return newSVsv (types_false); 1092 return newSVsv (types_false);
770 case 21: 1093 case SIMPLE_TRUE:
771#if CBOR_SLOW 1094#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1095 types_true = get_bool ("Types::Serialiser::true");
773#endif 1096#endif
774 return newSVsv (types_true); 1097 return newSVsv (types_true);
775 case 22: 1098 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1099 return newSVsv (&PL_sv_undef);
777 case 23: 1100 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1101#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1102 types_error = get_bool ("Types::Serialiser::error");
780#endif 1103#endif
781 return newSVsv (types_error); 1104 return newSVsv (types_error);
782 1105
783 case 25: 1106 case MISC_FLOAT16:
784 { 1107 {
785 WANT (2); 1108 WANT (2);
786 1109
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1110 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1111 dec->cur += 2;
789 1112
790 return newSVnv (ecb_binary16_to_float (fp)); 1113 return newSVnv (ecb_binary16_to_float (fp));
791 } 1114 }
792 1115
793 case 26: 1116 case MISC_FLOAT32:
794 { 1117 {
795 uint32_t fp; 1118 uint32_t fp;
796 WANT (4); 1119 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1120 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1121 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1124 fp = ecb_bswap32 (fp);
802 1125
803 return newSVnv (ecb_binary32_to_float (fp)); 1126 return newSVnv (ecb_binary32_to_float (fp));
804 } 1127 }
805 1128
806 case 27: 1129 case MISC_FLOAT64:
807 { 1130 {
808 uint64_t fp; 1131 uint64_t fp;
809 WANT (8); 1132 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1133 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1134 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1137 fp = ecb_bswap64 (fp);
815 1138
816 return newSVnv (ecb_binary64_to_double (fp)); 1139 return newSVnv (ecb_binary64_to_double (fp));
817 } 1140 }
818 1141
819 // 0..19 unassigned 1142 // 0..19 unassigned simple
820 // 24 reserved + unassigned (reserved values are not encodable) 1143 // 24 reserved + unassigned simple (reserved values are not encodable)
1144 // 28-30 unassigned misc
1145 // 31 break code
821 default: 1146 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1147 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
823 } 1148 }
824 1149
825 break; 1150 break;
826 } 1151 }
827 1152
830} 1155}
831 1156
832static SV * 1157static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1158decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1159{
835 dec_t dec; 1160 dec_t dec = { };
836 SV *sv; 1161 SV *sv;
1162 STRLEN len;
1163 char *data = SvPVbyte (string, len);
837 1164
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1165 if (len > cbor->max_size && cbor->max_size)
867 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1166 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1167 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1168
873 dec.cbor = *cbor; 1169 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1170 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1171 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1172
879 sv = decode_sv (&dec); 1173 sv = decode_sv (&dec);
880 1174
881 if (offset_return) 1175 if (offset_return)
882 *offset_return = dec.cur; 1176 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1179 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1180 dec.err = "garbage after CBOR object";
887 1181
888 if (dec.err) 1182 if (dec.err)
889 { 1183 {
1184 if (dec.shareable)
1185 {
1186 // need to break cyclic links, which whould all be in shareable
1187 int i;
1188 SV **svp;
1189
1190 for (i = av_len (dec.shareable) + 1; i--; )
1191 if ((svp = av_fetch (dec.shareable, i, 0)))
1192 sv_setsv (*svp, &PL_sv_undef);
1193 }
1194
890 SvREFCNT_dec (sv); 1195 SvREFCNT_dec (sv);
891 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1196 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
892 } 1197 }
893 1198
894 sv = sv_2mortal (sv); 1199 sv = sv_2mortal (sv);
895 1200
896 return sv; 1201 return sv;
897} 1202}
898 1203
1204/////////////////////////////////////////////////////////////////////////////
1205// incremental parser
1206
1207#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1208
1209// returns 0 for notyet, 1 for success or error
1210static int
1211incr_parse (CBOR *self, SV *cborstr)
1212{
1213 STRLEN cur;
1214 SvPV (cborstr, cur);
1215
1216 while (ecb_expect_true (self->incr_need <= cur))
1217 {
1218 // table of integer count bytes
1219 static I8 incr_len[MINOR_MASK + 1] = {
1220 0, 0, 0, 0, 0, 0, 0, 0,
1221 0, 0, 0, 0, 0, 0, 0, 0,
1222 0, 0, 0, 0, 0, 0, 0, 0,
1223 1, 2, 4, 8,-1,-1,-1,-2
1224 };
1225
1226 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1227 U8 m = *p & MINOR_MASK;
1228 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1229 I8 ilen = incr_len[m];
1230
1231 self->incr_need = self->incr_pos + 1;
1232
1233 if (ecb_expect_false (ilen < 0))
1234 {
1235 if (m != MINOR_INDEF)
1236 return 1; // error
1237
1238 if (*p == (MAJOR_MISC | MINOR_INDEF))
1239 {
1240 if (count >= 0)
1241 return 1; // error
1242
1243 count = 1;
1244 }
1245 else
1246 {
1247 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1248 count = -1;
1249 }
1250 }
1251 else
1252 {
1253 self->incr_need += ilen;
1254 if (ecb_expect_false (self->incr_need > cur))
1255 return 0;
1256
1257 int major = *p >> MAJOR_SHIFT;
1258
1259 switch (major)
1260 {
1261 case MAJOR_BYTES >> MAJOR_SHIFT:
1262 case MAJOR_TEXT >> MAJOR_SHIFT:
1263 case MAJOR_ARRAY >> MAJOR_SHIFT:
1264 case MAJOR_MAP >> MAJOR_SHIFT:
1265 {
1266 UV len;
1267
1268 if (ecb_expect_false (ilen))
1269 {
1270 len = 0;
1271
1272 do {
1273 len = (len << 8) | *++p;
1274 } while (--ilen);
1275 }
1276 else
1277 len = m;
1278
1279 switch (major)
1280 {
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 self->incr_need += len;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 break;
1288
1289 case MAJOR_MAP >> MAJOR_SHIFT:
1290 len <<= 1;
1291 case MAJOR_ARRAY >> MAJOR_SHIFT:
1292 if (len)
1293 {
1294 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1295 count = len + 1;
1296 }
1297 break;
1298 }
1299 }
1300 }
1301 }
1302
1303 self->incr_pos = self->incr_need;
1304
1305 if (count > 0)
1306 {
1307 while (!--count)
1308 {
1309 if (!AvFILLp (self->incr_count))
1310 return 1; // done
1311
1312 SvREFCNT_dec_NN (av_pop (self->incr_count));
1313 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1314 }
1315
1316 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1317 }
1318 }
1319
1320 return 0;
1321}
1322
1323
899///////////////////////////////////////////////////////////////////////////// 1324/////////////////////////////////////////////////////////////////////////////
900// XS interface functions 1325// XS interface functions
901 1326
902MODULE = CBOR::XS PACKAGE = CBOR::XS 1327MODULE = CBOR::XS PACKAGE = CBOR::XS
903 1328
910 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1335 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
911 1336
912 types_true = get_bool ("Types::Serialiser::true" ); 1337 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1338 types_false = get_bool ("Types::Serialiser::false");
914 types_error = get_bool ("Types::Serialiser::error"); 1339 types_error = get_bool ("Types::Serialiser::error");
1340
1341 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
915 1342
916 sv_cbor = newSVpv ("CBOR", 0); 1343 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1344 SvREADONLY_on (sv_cbor);
918} 1345}
919 1346
940 1367
941void shrink (CBOR *self, int enable = 1) 1368void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1369 ALIAS:
943 shrink = F_SHRINK 1370 shrink = F_SHRINK
944 allow_unknown = F_ALLOW_UNKNOWN 1371 allow_unknown = F_ALLOW_UNKNOWN
1372 allow_sharing = F_ALLOW_SHARING
1373 allow_cycles = F_ALLOW_CYCLES
1374 pack_strings = F_PACK_STRINGS
1375 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1376 PPCODE:
946{ 1377{
947 if (enable) 1378 if (enable)
948 self->flags |= ix; 1379 self->flags |= ix;
949 else 1380 else
954 1385
955void get_shrink (CBOR *self) 1386void get_shrink (CBOR *self)
956 ALIAS: 1387 ALIAS:
957 get_shrink = F_SHRINK 1388 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1389 get_allow_unknown = F_ALLOW_UNKNOWN
1390 get_allow_sharing = F_ALLOW_SHARING
1391 get_allow_cycles = F_ALLOW_CYCLES
1392 get_pack_strings = F_PACK_STRINGS
1393 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1394 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1395 XPUSHs (boolSV (self->flags & ix));
961 1396
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1397void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1398 PPCODE:
979 CODE: 1414 CODE:
980 RETVAL = self->max_size; 1415 RETVAL = self->max_size;
981 OUTPUT: 1416 OUTPUT:
982 RETVAL 1417 RETVAL
983 1418
984#if 0 //TODO 1419void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1420 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1421 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1422 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1423 XPUSHs (ST (0));
993}
994 1424
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1425SV *get_filter (CBOR *self)
996 PPCODE: 1426 CODE:
997{ 1427 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1428 OUTPUT:
999 self->cb_sk_object = newHV (); 1429 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1430
1019void encode (CBOR *self, SV *scalar) 1431void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1432 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1433 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1434 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1447 EXTEND (SP, 2);
1036 PUSHs (sv); 1448 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1449 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1450}
1039 1451
1040#if 0 1452void incr_parse (CBOR *self, SV *cborstr, int chop = 0)
1453 PPCODE:
1454{
1455 if (SvUTF8 (cborstr))
1456 sv_utf8_downgrade (cborstr, 0);
1457
1458 if (!self->incr_count)
1459 {
1460 self->incr_count = newAV ();
1461 self->incr_pos = 0;
1462 self->incr_need = 1;
1463
1464 av_push (self->incr_count, newSViv (1));
1465 }
1466
1467 for (;;)
1468 {
1469 if (!incr_parse (self, cborstr))
1470 {
1471 if (self->incr_need > self->max_size && self->max_size)
1472 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1473 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1474
1475 break;
1476 }
1477
1478 SV *sv;
1479 char *offset;
1480
1481 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1482 XPUSHs (sv);
1483
1484 av_clear (self->incr_count);
1485 av_push (self->incr_count, newSViv (1));
1486
1487 if (chop)
1488 {
1489 self->incr_pos = 0;
1490 sv_chop (cborstr, offset);
1491 }
1492 else
1493 self->incr_pos = offset - SvPVX (cborstr);
1494
1495 self->incr_need = self->incr_pos + 1;
1496 }
1497}
1498
1499void incr_reset (CBOR *self)
1500 CODE:
1501{
1502 SvREFCNT_dec (self->incr_count);
1503 self->incr_count = 0;
1504}
1041 1505
1042void DESTROY (CBOR *self) 1506void DESTROY (CBOR *self)
1043 CODE: 1507 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1508 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1509
1049PROTOTYPES: ENABLE 1510PROTOTYPES: ENABLE
1050 1511
1051void encode_cbor (SV *scalar) 1512void encode_cbor (SV *scalar)
1513 ALIAS:
1514 encode_cbor = 0
1515 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1516 PPCODE:
1053{ 1517{
1054 CBOR cbor; 1518 CBOR cbor;
1055 cbor_init (&cbor); 1519 cbor_init (&cbor);
1520 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1521 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1522 XPUSHs (scalar);
1058} 1523}
1059 1524
1060void decode_cbor (SV *cborstr) 1525void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines