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 1.19 by root, Wed Nov 20 02:03:09 2013 UTC vs.
Revision 1.51 by root, Sun Apr 24 13:15:19 2016 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_SHAREABLE = 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_UTF8_STRINGS 0x00000020UL
106#define F_VALIDATE_UTF8 0x00000040UL
64 107
65#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
66 109
67#define SB do { 110#define SB do {
68#define SE } while (0) 111#define SE } while (0)
80# define CBOR_SLOW 0 123# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 124# define CBOR_STASH cbor_stash
82#endif 125#endif
83 126
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 127static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 128static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 129
87typedef struct { 130typedef struct {
88 U32 flags; 131 U32 flags;
89 U32 max_depth; 132 U32 max_depth;
90 STRLEN max_size; 133 STRLEN max_size;
134 SV *filter;
135
136 // for the incremental parser
137 STRLEN incr_pos; // the current offset into the text
138 STRLEN incr_need; // minimum bytes needed to decode
139 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 140} CBOR;
92 141
93ecb_inline void 142ecb_inline void
94cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
95{ 144{
96 Zero (cbor, 1, CBOR); 145 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 146 cbor->max_depth = 512;
147}
148
149ecb_inline void
150cbor_free (CBOR *cbor)
151{
152 SvREFCNT_dec (cbor->filter);
153 SvREFCNT_dec (cbor->incr_count);
98} 154}
99 155
100///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
101// utility functions 157// utility functions
102 158
124 SvPV_renew (sv, SvCUR (sv) + 1); 180 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 181#endif
126 } 182 }
127} 183}
128 184
185// minimum length of a string to be registered for stringref
186ecb_inline int
187minimum_string_length (UV idx)
188{
189 return idx > 23
190 ? idx > 0xffU
191 ? idx > 0xffffU
192 ? idx > 0xffffffffU
193 ? 11
194 : 7
195 : 5
196 : 4
197 : 3;
198}
199
129///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
130// encoder 201// encoder
131 202
132// structure used for encoding CBOR 203// structure used for encoding CBOR
133typedef struct 204typedef struct
135 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
137 SV *sv; // result scalar 208 SV *sv; // result scalar
138 CBOR cbor; 209 CBOR cbor;
139 U32 depth; // recursion level 210 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 211 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
212 UV stringref_idx;
141 HV *shareable; // ptr => index, or 0 213 HV *shareable; // ptr => index, or 0
142 UV shareable_idx; 214 UV shareable_idx;
143} enc_t; 215} enc_t;
144 216
145ecb_inline void 217ecb_inline void
164static void 236static void
165encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
166{ 238{
167 need (enc, 9); 239 need (enc, 9);
168 240
169 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
171 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
172 { 244 {
173 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 246 *enc->cur++ = len;
175 } 247 }
176 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
177 { 249 {
178 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 252 *enc->cur++ = len;
181 } 253 }
182 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
183 { 255 {
184 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 260 *enc->cur++ = len;
189 } 261 }
190 else 262 else
191 { 263 {
192 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 272 *enc->cur++ = len;
201 } 273 }
202} 274}
203 275
204static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282ecb_inline void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 284{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS))
286 if (!utf8)
287 {
288 SV *sv = sv_newmortal ();
289 char *s; STRLEN l;
290
291 sv_setpvn (sv, str, len);
292
293 s = SvPVutf8 (sv, l);
294 encode_str (enc, 1, s, l);
295 return;
296 }
297
298 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 299 need (enc, len);
209 memcpy (enc->cur, str, len); 300 memcpy (enc->cur, str, len);
210 enc->cur += len; 301 enc->cur += len;
211} 302}
212 303
213ecb_inline void 304static void
214encode_tag (enc_t *enc, UV tag) 305encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
215{ 306{
216 encode_uint (enc, 0xc0, tag); 307 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
308 {
309 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
310
311 if (SvOK (*svp))
312 {
313 // already registered, use stringref
314 encode_tag (enc, CBOR_TAG_STRINGREF);
315 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
316 return;
317 }
318 else if (len >= minimum_string_length (enc->stringref_idx))
319 {
320 // register only
321 sv_setuv (*svp, enc->stringref_idx);
322 ++enc->stringref_idx;
323 }
324 }
325
326 encode_str (enc, utf8, str, len);
217} 327}
218 328
219static void encode_sv (enc_t *enc, SV *sv); 329static void encode_sv (enc_t *enc, SV *sv);
220 330
221static void 331static void
226 if (enc->depth >= enc->cbor.max_depth) 336 if (enc->depth >= enc->cbor.max_depth)
227 croak (ERR_NESTING_EXCEEDED); 337 croak (ERR_NESTING_EXCEEDED);
228 338
229 ++enc->depth; 339 ++enc->depth;
230 340
231 encode_uint (enc, 0x80, len + 1); 341 encode_uint (enc, MAJOR_ARRAY, len + 1);
232 342
343 if (SvMAGICAL (av))
233 for (i = 0; i <= len; ++i) 344 for (i = 0; i <= len; ++i)
234 { 345 {
235 SV **svp = av_fetch (av, i, 0); 346 SV **svp = av_fetch (av, i, 0);
236 encode_sv (enc, svp ? *svp : &PL_sv_undef); 347 encode_sv (enc, svp ? *svp : &PL_sv_undef);
237 } 348 }
349 else
350 for (i = 0; i <= len; ++i)
351 {
352 SV *sv = AvARRAY (av)[i];
353 encode_sv (enc, sv ? sv : &PL_sv_undef);
354 }
238 355
239 --enc->depth; 356 --enc->depth;
240} 357}
241 358
242static void 359static void
251 368
252 int pairs = hv_iterinit (hv); 369 int pairs = hv_iterinit (hv);
253 int mg = SvMAGICAL (hv); 370 int mg = SvMAGICAL (hv);
254 371
255 if (mg) 372 if (mg)
256 encode_ch (enc, 0xa0 | 31); 373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
257 else 374 else
258 encode_uint (enc, 0xa0, pairs); 375 encode_uint (enc, MAJOR_MAP, pairs);
259 376
260 while ((he = hv_iternext (hv))) 377 while ((he = hv_iternext (hv)))
261 { 378 {
262 if (HeKLEN (he) == HEf_SVKEY) 379 if (HeKLEN (he) == HEf_SVKEY)
263 encode_sv (enc, HeSVKEY (he)); 380 encode_sv (enc, HeSVKEY (he));
264 else 381 else
265 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 382 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
266 383
267 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 384 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
268 } 385 }
269 386
270 if (mg) 387 if (mg)
271 encode_ch (enc, 0xe0 | 31); 388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
272 389
273 --enc->depth; 390 --enc->depth;
274} 391}
275 392
276// encode objects, arrays and special \0=false and \1=true values. 393// encode objects, arrays and special \0=false and \1=true values.
277static void 394static void
278encode_rv (enc_t *enc, SV *sv) 395encode_rv (enc_t *enc, SV *sv)
279{ 396{
280 SvGETMAGIC (sv); 397 SvGETMAGIC (sv);
281
282 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
283 && ecb_expect_false (SvREFCNT (sv) > 1))
284 {
285 if (!enc->shareable)
286 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
287
288 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
289
290 if (SvOK (*svp))
291 {
292 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
293 encode_uint (enc, 0x00, SvUV (*svp));
294 return;
295 }
296 else
297 {
298 sv_setuv (*svp, enc->shareable_idx);
299 ++enc->shareable_idx;
300 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
301 }
302 }
303 398
304 svtype svt = SvTYPE (sv); 399 svtype svt = SvTYPE (sv);
305 400
306 if (ecb_expect_false (SvOBJECT (sv))) 401 if (ecb_expect_false (SvOBJECT (sv)))
307 { 402 {
314 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 409 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
315 ? cbor_tagged_stash 410 ? cbor_tagged_stash
316 : gv_stashpv ("CBOR::XS::Tagged" , 1); 411 : gv_stashpv ("CBOR::XS::Tagged" , 1);
317 412
318 HV *stash = SvSTASH (sv); 413 HV *stash = SvSTASH (sv);
319 GV *method;
320 414
321 if (stash == boolean_stash) 415 if (stash == boolean_stash)
322 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 416 {
417 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
418 return;
419 }
323 else if (stash == error_stash) 420 else if (stash == error_stash)
324 encode_ch (enc, 0xe0 | 23); 421 {
422 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
423 return;
424 }
325 else if (stash == tagged_stash) 425 else if (stash == tagged_stash)
326 { 426 {
327 if (svt != SVt_PVAV) 427 if (svt != SVt_PVAV)
328 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 428 croak ("encountered CBOR::XS::Tagged object that isn't an array");
329 429
330 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 430 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
331 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 431 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
432
433 return;
434 }
435 }
436
437 if (ecb_expect_false (SvREFCNT (sv) > 1)
438 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
439 {
440 if (!enc->shareable)
441 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
442
443 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
444
445 if (SvOK (*svp))
332 } 446 {
447 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
448 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
449 return;
450 }
451 else
452 {
453 sv_setuv (*svp, enc->shareable_idx);
454 ++enc->shareable_idx;
455 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
456 }
457 }
458
459 if (ecb_expect_false (SvOBJECT (sv)))
460 {
461 HV *stash = SvSTASH (sv);
462 GV *method;
463
333 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 464 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
334 { 465 {
335 dSP; 466 dSP;
336 467
337 ENTER; SAVETMPS; PUSHMARK (SP); 468 ENTER; SAVETMPS;
469 PUSHMARK (SP);
338 // we re-bless the reference to get overload and other niceties right 470 // we re-bless the reference to get overload and other niceties right
339 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 471 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
340 472
341 PUTBACK; 473 PUTBACK;
342 // G_SCALAR ensures that return value is 1 474 // G_SCALAR ensures that return value is 1
355 } 487 }
356 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 488 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
357 { 489 {
358 dSP; 490 dSP;
359 491
360 ENTER; SAVETMPS; PUSHMARK (SP); 492 ENTER; SAVETMPS;
493 SAVESTACK_POS ();
494 PUSHMARK (SP);
361 EXTEND (SP, 2); 495 EXTEND (SP, 2);
362 // we re-bless the reference to get overload and other niceties right 496 // we re-bless the reference to get overload and other niceties right
363 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 497 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
364 PUSHs (sv_cbor); 498 PUSHs (sv_cbor);
365 499
370 // catch this surprisingly common error 504 // catch this surprisingly common error
371 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 505 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
372 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 506 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
373 507
374 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 508 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
375 encode_uint (enc, 0x80, count + 1); 509 encode_uint (enc, MAJOR_ARRAY, count + 1);
376 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 510 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
377 511
378 while (count) 512 while (count)
379 encode_sv (enc, SP[1 - count--]); 513 encode_sv (enc, SP[1 - count--]);
380 514
381 PUTBACK; 515 PUTBACK;
402{ 536{
403 double nv = SvNVX (sv); 537 double nv = SvNVX (sv);
404 538
405 need (enc, 9); 539 need (enc, 9);
406 540
407 if (ecb_expect_false (nv == (U32)nv)) 541 if (ecb_expect_false (nv == (NV)(U32)nv))
408 encode_uint (enc, 0x00, (U32)nv); 542 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
409 //TODO: maybe I32? 543 //TODO: maybe I32?
410 else if (ecb_expect_false (nv == (float)nv)) 544 else if (ecb_expect_false (nv == (float)nv))
411 { 545 {
412 uint32_t fp = ecb_float_to_binary32 (nv); 546 uint32_t fp = ecb_float_to_binary32 (nv);
413 547
414 *enc->cur++ = 0xe0 | 26; 548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
415 549
416 if (!ecb_big_endian ()) 550 if (!ecb_big_endian ())
417 fp = ecb_bswap32 (fp); 551 fp = ecb_bswap32 (fp);
418 552
419 memcpy (enc->cur, &fp, 4); 553 memcpy (enc->cur, &fp, 4);
421 } 555 }
422 else 556 else
423 { 557 {
424 uint64_t fp = ecb_double_to_binary64 (nv); 558 uint64_t fp = ecb_double_to_binary64 (nv);
425 559
426 *enc->cur++ = 0xe0 | 27; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
427 561
428 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
429 fp = ecb_bswap64 (fp); 563 fp = ecb_bswap64 (fp);
430 564
431 memcpy (enc->cur, &fp, 8); 565 memcpy (enc->cur, &fp, 8);
440 574
441 if (SvPOKp (sv)) 575 if (SvPOKp (sv))
442 { 576 {
443 STRLEN len; 577 STRLEN len;
444 char *str = SvPV (sv, len); 578 char *str = SvPV (sv, len);
445 encode_str (enc, SvUTF8 (sv), str, len); 579 encode_strref (enc, SvUTF8 (sv), str, len);
446 } 580 }
447 else if (SvNOKp (sv)) 581 else if (SvNOKp (sv))
448 encode_nv (enc, sv); 582 encode_nv (enc, sv);
449 else if (SvIOKp (sv)) 583 else if (SvIOKp (sv))
450 { 584 {
451 if (SvIsUV (sv)) 585 if (SvIsUV (sv))
452 encode_uint (enc, 0x00, SvUVX (sv)); 586 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
453 else if (SvIVX (sv) >= 0) 587 else if (SvIVX (sv) >= 0)
454 encode_uint (enc, 0x00, SvIVX (sv)); 588 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
455 else 589 else
456 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 590 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
457 } 591 }
458 else if (SvROK (sv)) 592 else if (SvROK (sv))
459 encode_rv (enc, SvRV (sv)); 593 encode_rv (enc, SvRV (sv));
460 else if (!SvOK (sv)) 594 else if (!SvOK (sv))
461 encode_ch (enc, 0xe0 | 22); 595 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
462 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 596 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
463 encode_ch (enc, 0xe0 | 23); 597 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
464 else 598 else
465 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 599 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
466 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 600 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
467} 601}
468 602
469static SV * 603static SV *
470encode_cbor (SV *scalar, CBOR *cbor) 604encode_cbor (SV *scalar, CBOR *cbor)
471{ 605{
472 enc_t enc = { }; 606 enc_t enc = { 0 };
473 607
474 enc.cbor = *cbor; 608 enc.cbor = *cbor;
475 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
476 enc.cur = SvPVX (enc.sv); 610 enc.cur = SvPVX (enc.sv);
477 enc.end = SvEND (enc.sv); 611 enc.end = SvEND (enc.sv);
478 612
479 SvPOK_only (enc.sv); 613 SvPOK_only (enc.sv);
614
615 if (cbor->flags & F_PACK_STRINGS)
616 {
617 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
618 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
619 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
620 }
621
480 encode_sv (&enc, scalar); 622 encode_sv (&enc, scalar);
481 623
482 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 624 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
483 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 625 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
484 626
499 const char *err; // parse error, if != 0 641 const char *err; // parse error, if != 0
500 CBOR cbor; 642 CBOR cbor;
501 U32 depth; // recursion depth 643 U32 depth; // recursion depth
502 U32 maxdepth; // recursion depth limit 644 U32 maxdepth; // recursion depth limit
503 AV *shareable; 645 AV *shareable;
646 AV *stringref;
647 SV *decode_tagged;
504} dec_t; 648} dec_t;
505 649
506#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 650#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
507 651
508#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 652#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
511#define DEC_DEC_DEPTH --dec->depth 655#define DEC_DEC_DEPTH --dec->depth
512 656
513static UV 657static UV
514decode_uint (dec_t *dec) 658decode_uint (dec_t *dec)
515{ 659{
516 switch (*dec->cur & 31) 660 U8 m = *dec->cur & MINOR_MASK;
517 { 661 ++dec->cur;
518 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
519 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
520 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
521 return *dec->cur++ & 31;
522 662
523 case 24: 663 if (ecb_expect_true (m < LENGTH_EXT1))
664 return m;
665 else if (ecb_expect_true (m == LENGTH_EXT1))
666 {
524 WANT (2); 667 WANT (1);
525 dec->cur += 2; 668 dec->cur += 1;
526 return dec->cur[-1]; 669 return dec->cur[-1];
527 670 }
528 case 25: 671 else if (ecb_expect_true (m == LENGTH_EXT2))
672 {
529 WANT (3); 673 WANT (2);
530 dec->cur += 3; 674 dec->cur += 2;
531 return (((UV)dec->cur[-2]) << 8) 675 return (((UV)dec->cur[-2]) << 8)
532 | ((UV)dec->cur[-1]); 676 | ((UV)dec->cur[-1]);
533 677 }
534 case 26: 678 else if (ecb_expect_true (m == LENGTH_EXT4))
679 {
535 WANT (5); 680 WANT (4);
536 dec->cur += 5; 681 dec->cur += 4;
537 return (((UV)dec->cur[-4]) << 24) 682 return (((UV)dec->cur[-4]) << 24)
538 | (((UV)dec->cur[-3]) << 16) 683 | (((UV)dec->cur[-3]) << 16)
539 | (((UV)dec->cur[-2]) << 8) 684 | (((UV)dec->cur[-2]) << 8)
540 | ((UV)dec->cur[-1]); 685 | ((UV)dec->cur[-1]);
541 686 }
542 case 27: 687 else if (ecb_expect_true (m == LENGTH_EXT8))
688 {
543 WANT (9); 689 WANT (8);
544 dec->cur += 9; 690 dec->cur += 8;
691
692 return
693#if UVSIZE < 8
694 0
695#else
545 return (((UV)dec->cur[-8]) << 56) 696 (((UV)dec->cur[-8]) << 56)
546 | (((UV)dec->cur[-7]) << 48) 697 | (((UV)dec->cur[-7]) << 48)
547 | (((UV)dec->cur[-6]) << 40) 698 | (((UV)dec->cur[-6]) << 40)
548 | (((UV)dec->cur[-5]) << 32) 699 | (((UV)dec->cur[-5]) << 32)
700#endif
549 | (((UV)dec->cur[-4]) << 24) 701 | (((UV)dec->cur[-4]) << 24)
550 | (((UV)dec->cur[-3]) << 16) 702 | (((UV)dec->cur[-3]) << 16)
551 | (((UV)dec->cur[-2]) << 8) 703 | (((UV)dec->cur[-2]) << 8)
552 | ((UV)dec->cur[-1]); 704 | ((UV)dec->cur[-1]);
553 705 }
554 default: 706 else
555 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 707 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
556 }
557 708
558fail: 709fail:
559 return 0; 710 return 0;
560} 711}
561 712
566{ 717{
567 AV *av = newAV (); 718 AV *av = newAV ();
568 719
569 DEC_INC_DEPTH; 720 DEC_INC_DEPTH;
570 721
571 if ((*dec->cur & 31) == 31) 722 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
572 { 723 {
573 ++dec->cur; 724 ++dec->cur;
574 725
575 for (;;) 726 for (;;)
576 { 727 {
577 WANT (1); 728 WANT (1);
578 729
579 if (*dec->cur == (0xe0 | 31)) 730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
580 { 731 {
581 ++dec->cur; 732 ++dec->cur;
582 break; 733 break;
583 } 734 }
584 735
587 } 738 }
588 else 739 else
589 { 740 {
590 int i, len = decode_uint (dec); 741 int i, len = decode_uint (dec);
591 742
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
592 av_fill (av, len - 1); 744 av_fill (av, len - 1);
593 745
594 for (i = 0; i < len; ++i) 746 for (i = 0; i < len; ++i)
595 AvARRAY (av)[i] = decode_sv (dec); 747 AvARRAY (av)[i] = decode_sv (dec);
596 } 748 }
606 758
607static void 759static void
608decode_he (dec_t *dec, HV *hv) 760decode_he (dec_t *dec, HV *hv)
609{ 761{
610 // for speed reasons, we specialcase single-string 762 // for speed reasons, we specialcase single-string
611 // byte or utf-8 strings as keys. 763 // byte or utf-8 strings as keys, but only when !stringref
612 764
613 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 765 if (ecb_expect_true (!dec->stringref))
766 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
614 { 767 {
615 I32 len = decode_uint (dec); 768 I32 len = decode_uint (dec);
616 char *key = (char *)dec->cur; 769 char *key = (char *)dec->cur;
617 770
771 WANT (len);
618 dec->cur += len; 772 dec->cur += len;
619 773
620 hv_store (hv, key, len, decode_sv (dec), 0); 774 hv_store (hv, key, len, decode_sv (dec), 0);
775
776 return;
621 } 777 }
622 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 778 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
623 { 779 {
624 I32 len = decode_uint (dec); 780 I32 len = decode_uint (dec);
625 char *key = (char *)dec->cur; 781 char *key = (char *)dec->cur;
626 782
783 WANT (len);
627 dec->cur += len; 784 dec->cur += len;
628 785
786 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
787 if (!is_utf8_string (key, len))
788 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
789
629 hv_store (hv, key, -len, decode_sv (dec), 0); 790 hv_store (hv, key, -len, decode_sv (dec), 0);
791
792 return;
630 } 793 }
631 else 794
632 {
633 SV *k = decode_sv (dec); 795 SV *k = decode_sv (dec);
634 SV *v = decode_sv (dec); 796 SV *v = decode_sv (dec);
635 797
636 hv_store_ent (hv, k, v, 0); 798 hv_store_ent (hv, k, v, 0);
637 SvREFCNT_dec (k); 799 SvREFCNT_dec (k);
638 } 800
801fail:
802 ;
639} 803}
640 804
641static SV * 805static SV *
642decode_hv (dec_t *dec) 806decode_hv (dec_t *dec)
643{ 807{
644 HV *hv = newHV (); 808 HV *hv = newHV ();
645 809
646 DEC_INC_DEPTH; 810 DEC_INC_DEPTH;
647 811
648 if ((*dec->cur & 31) == 31) 812 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
649 { 813 {
650 ++dec->cur; 814 ++dec->cur;
651 815
652 for (;;) 816 for (;;)
653 { 817 {
654 WANT (1); 818 WANT (1);
655 819
656 if (*dec->cur == (0xe0 | 31)) 820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
657 { 821 {
658 ++dec->cur; 822 ++dec->cur;
659 break; 823 break;
660 } 824 }
661 825
682static SV * 846static SV *
683decode_str (dec_t *dec, int utf8) 847decode_str (dec_t *dec, int utf8)
684{ 848{
685 SV *sv = 0; 849 SV *sv = 0;
686 850
687 if ((*dec->cur & 31) == 31) 851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
688 { 852 {
853 // indefinite length strings
689 ++dec->cur; 854 ++dec->cur;
690 855
856 U8 major = *dec->cur & MAJOR_MISC;
857
691 sv = newSVpvn ("", 0); 858 sv = newSVpvn ("", 0);
692 859
693 // not very fast, and certainly not robust against illegal input
694 for (;;) 860 for (;;)
695 { 861 {
696 WANT (1); 862 WANT (1);
697 863
698 if (*dec->cur == (0xe0 | 31)) 864 if ((*dec->cur - major) > LENGTH_EXT8)
865 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
699 { 866 {
700 ++dec->cur; 867 ++dec->cur;
701 break; 868 break;
702 } 869 }
870 else
871 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
703 872
704 sv_catsv (sv, decode_sv (dec)); 873 STRLEN len = decode_uint (dec);
874
875 WANT (len);
876 sv_catpvn (sv, dec->cur, len);
877 dec->cur += len;
705 } 878 }
706 } 879 }
707 else 880 else
708 { 881 {
709 STRLEN len = decode_uint (dec); 882 STRLEN len = decode_uint (dec);
710 883
711 WANT (len); 884 WANT (len);
712 sv = newSVpvn (dec->cur, len); 885 sv = newSVpvn (dec->cur, len);
713 dec->cur += len; 886 dec->cur += len;
887
888 if (ecb_expect_false (dec->stringref)
889 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
890 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
714 } 891 }
715 892
716 if (utf8) 893 if (utf8)
894 {
895 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
896 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
897 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
898
717 SvUTF8_on (sv); 899 SvUTF8_on (sv);
900 }
718 901
719 return sv; 902 return sv;
720 903
721fail: 904fail:
722 SvREFCNT_dec (sv); 905 SvREFCNT_dec (sv);
732 WANT (1); 915 WANT (1);
733 916
734 switch (tag) 917 switch (tag)
735 { 918 {
736 case CBOR_TAG_MAGIC: 919 case CBOR_TAG_MAGIC:
737 return decode_sv (dec); 920 sv = decode_sv (dec);
921 break;
738 922
739 case CBOR_TAG_INDIRECTION: 923 case CBOR_TAG_INDIRECTION:
740 return newRV_noinc (decode_sv (dec)); 924 sv = newRV_noinc (decode_sv (dec));
925 break;
926
927 case CBOR_TAG_STRINGREF_NAMESPACE:
928 {
929 ENTER; SAVETMPS;
930
931 SAVESPTR (dec->stringref);
932 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
933
934 sv = decode_sv (dec);
935
936 FREETMPS; LEAVE;
937 }
938 break;
939
940 case CBOR_TAG_STRINGREF:
941 {
942 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
943 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
944
945 UV idx = decode_uint (dec);
946
947 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
948 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
949
950 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
951 }
952 break;
741 953
742 case CBOR_TAG_VALUE_SHAREABLE: 954 case CBOR_TAG_VALUE_SHAREABLE:
743 { 955 {
744 if (ecb_expect_false (!dec->shareable)) 956 if (ecb_expect_false (!dec->shareable))
745 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 957 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
746 958
959 if (dec->cbor.flags & F_ALLOW_CYCLES)
960 {
747 sv = newSV (0); 961 sv = newSV (0);
748 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 962 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
749 963
750 SV *osv = decode_sv (dec); 964 SV *osv = decode_sv (dec);
751 sv_setsv (sv, osv); 965 sv_setsv (sv, osv);
752 SvREFCNT_dec_NN (osv); 966 SvREFCNT_dec_NN (osv);
753 } 967 }
754 968 else
755 return sv; 969 {
970 av_push (dec->shareable, &PL_sv_undef);
971 int idx = AvFILLp (dec->shareable);
972 sv = decode_sv (dec);
973 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
974 }
975 }
976 break;
756 977
757 case CBOR_TAG_VALUE_SHAREDREF: 978 case CBOR_TAG_VALUE_SHAREDREF:
758 { 979 {
759 if ((*dec->cur >> 5) != 0) 980 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
760 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 981 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
761 982
762 UV idx = decode_uint (dec); 983 UV idx = decode_uint (dec);
763 984
764 if (!dec->shareable || idx > AvFILLp (dec->shareable)) 985 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
765 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 986 ERR ("corrupted CBOR data (sharedref index out of bounds)");
766 987
767 return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 988 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
989
990 if (sv == &PL_sv_undef)
991 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
768 } 992 }
993 break;
769 994
770 case CBOR_TAG_PERL_OBJECT: 995 case CBOR_TAG_PERL_OBJECT:
771 { 996 {
772 sv = decode_sv (dec); 997 sv = decode_sv (dec);
773 998
786 if (!method) 1011 if (!method)
787 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1012 ERR ("cannot decode perl-object (package does not have a THAW method)");
788 1013
789 dSP; 1014 dSP;
790 1015
791 ENTER; SAVETMPS; PUSHMARK (SP); 1016 ENTER; SAVETMPS;
1017 PUSHMARK (SP);
792 EXTEND (SP, len + 1); 1018 EXTEND (SP, len + 1);
793 // we re-bless the reference to get overload and other niceties right 1019 // we re-bless the reference to get overload and other niceties right
794 PUSHs (*av_fetch (av, 0, 1)); 1020 PUSHs (*av_fetch (av, 0, 1));
795 PUSHs (sv_cbor); 1021 PUSHs (sv_cbor);
796 1022
813 sv = SvREFCNT_inc (POPs); 1039 sv = SvREFCNT_inc (POPs);
814 1040
815 PUTBACK; 1041 PUTBACK;
816 1042
817 FREETMPS; LEAVE; 1043 FREETMPS; LEAVE;
818
819 return sv;
820 } 1044 }
1045 break;
821 1046
822 default: 1047 default:
823 { 1048 {
824 sv = decode_sv (dec); 1049 sv = decode_sv (dec);
825 1050
1051 dSP;
1052 ENTER; SAVETMPS;
1053 SAVESTACK_POS ();
1054 PUSHMARK (SP);
1055 EXTEND (SP, 2);
1056 PUSHs (newSVuv (tag));
1057 PUSHs (sv);
1058
1059 PUTBACK;
1060 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1061 SPAGAIN;
1062
1063 if (SvTRUE (ERRSV))
1064 {
1065 FREETMPS; LEAVE;
1066 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1067 }
1068
1069 if (count)
1070 {
1071 SvREFCNT_dec (sv);
1072 sv = SvREFCNT_inc (POPs);
1073 }
1074 else
1075 {
826 AV *av = newAV (); 1076 AV *av = newAV ();
827 av_push (av, newSVuv (tag)); 1077 av_push (av, newSVuv (tag));
828 av_push (av, sv); 1078 av_push (av, sv);
829 1079
830 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1080 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
831 ? cbor_tagged_stash 1081 ? cbor_tagged_stash
832 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1082 : gv_stashpv ("CBOR::XS::Tagged" , 1);
833
834 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1083 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
835 } 1084 }
1085
1086 PUTBACK;
1087
1088 FREETMPS; LEAVE;
1089 }
1090 break;
836 } 1091 }
1092
1093 return sv;
837 1094
838fail: 1095fail:
839 SvREFCNT_dec (sv); 1096 SvREFCNT_dec (sv);
840 return &PL_sv_undef; 1097 return &PL_sv_undef;
841} 1098}
843static SV * 1100static SV *
844decode_sv (dec_t *dec) 1101decode_sv (dec_t *dec)
845{ 1102{
846 WANT (1); 1103 WANT (1);
847 1104
848 switch (*dec->cur >> 5) 1105 switch (*dec->cur >> MAJOR_SHIFT)
849 { 1106 {
850 case 0: // unsigned int 1107 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
851 return newSVuv (decode_uint (dec)); 1108 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
852 case 1: // negative int 1109 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
853 return newSViv (-1 - (IV)decode_uint (dec)); 1110 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
854 case 2: // octet string 1111 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
855 return decode_str (dec, 0); 1112 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
856 case 3: // utf-8 string 1113 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
857 return decode_str (dec, 1); 1114
858 case 4: // array 1115 case MAJOR_MISC >> MAJOR_SHIFT:
859 return decode_av (dec);
860 case 5: // map
861 return decode_hv (dec);
862 case 6: // tag
863 return decode_tagged (dec);
864 case 7: // misc
865 switch (*dec->cur++ & 31) 1116 switch (*dec->cur++ & MINOR_MASK)
866 { 1117 {
867 case 20: 1118 case SIMPLE_FALSE:
868#if CBOR_SLOW 1119#if CBOR_SLOW
869 types_false = get_bool ("Types::Serialiser::false"); 1120 types_false = get_bool ("Types::Serialiser::false");
870#endif 1121#endif
871 return newSVsv (types_false); 1122 return newSVsv (types_false);
872 case 21: 1123 case SIMPLE_TRUE:
873#if CBOR_SLOW 1124#if CBOR_SLOW
874 types_true = get_bool ("Types::Serialiser::true"); 1125 types_true = get_bool ("Types::Serialiser::true");
875#endif 1126#endif
876 return newSVsv (types_true); 1127 return newSVsv (types_true);
877 case 22: 1128 case SIMPLE_NULL:
878 return newSVsv (&PL_sv_undef); 1129 return newSVsv (&PL_sv_undef);
879 case 23: 1130 case SIMPLE_UNDEF:
880#if CBOR_SLOW 1131#if CBOR_SLOW
881 types_error = get_bool ("Types::Serialiser::error"); 1132 types_error = get_bool ("Types::Serialiser::error");
882#endif 1133#endif
883 return newSVsv (types_error); 1134 return newSVsv (types_error);
884 1135
885 case 25: 1136 case MISC_FLOAT16:
886 { 1137 {
887 WANT (2); 1138 WANT (2);
888 1139
889 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1140 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
890 dec->cur += 2; 1141 dec->cur += 2;
891 1142
892 return newSVnv (ecb_binary16_to_float (fp)); 1143 return newSVnv (ecb_binary16_to_float (fp));
893 } 1144 }
894 1145
895 case 26: 1146 case MISC_FLOAT32:
896 { 1147 {
897 uint32_t fp; 1148 uint32_t fp;
898 WANT (4); 1149 WANT (4);
899 memcpy (&fp, dec->cur, 4); 1150 memcpy (&fp, dec->cur, 4);
900 dec->cur += 4; 1151 dec->cur += 4;
903 fp = ecb_bswap32 (fp); 1154 fp = ecb_bswap32 (fp);
904 1155
905 return newSVnv (ecb_binary32_to_float (fp)); 1156 return newSVnv (ecb_binary32_to_float (fp));
906 } 1157 }
907 1158
908 case 27: 1159 case MISC_FLOAT64:
909 { 1160 {
910 uint64_t fp; 1161 uint64_t fp;
911 WANT (8); 1162 WANT (8);
912 memcpy (&fp, dec->cur, 8); 1163 memcpy (&fp, dec->cur, 8);
913 dec->cur += 8; 1164 dec->cur += 8;
916 fp = ecb_bswap64 (fp); 1167 fp = ecb_bswap64 (fp);
917 1168
918 return newSVnv (ecb_binary64_to_double (fp)); 1169 return newSVnv (ecb_binary64_to_double (fp));
919 } 1170 }
920 1171
921 // 0..19 unassigned 1172 // 0..19 unassigned simple
922 // 24 reserved + unassigned (reserved values are not encodable) 1173 // 24 reserved + unassigned simple (reserved values are not encodable)
1174 // 28-30 unassigned misc
1175 // 31 break code
923 default: 1176 default:
924 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1177 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
925 } 1178 }
926 1179
927 break; 1180 break;
928 } 1181 }
929 1182
932} 1185}
933 1186
934static SV * 1187static SV *
935decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1188decode_cbor (SV *string, CBOR *cbor, char **offset_return)
936{ 1189{
937 dec_t dec = { }; 1190 dec_t dec = { 0 };
938 SV *sv; 1191 SV *sv;
939 STRLEN len; 1192 STRLEN len;
940 char *data = SvPVbyte (string, len); 1193 char *data = SvPVbyte (string, len);
941 1194
942 if (len > cbor->max_size && cbor->max_size) 1195 if (len > cbor->max_size && cbor->max_size)
956 if (dec.cur != dec.end && !dec.err) 1209 if (dec.cur != dec.end && !dec.err)
957 dec.err = "garbage after CBOR object"; 1210 dec.err = "garbage after CBOR object";
958 1211
959 if (dec.err) 1212 if (dec.err)
960 { 1213 {
1214 if (dec.shareable)
1215 {
1216 // need to break cyclic links, which whould all be in shareable
1217 int i;
1218 SV **svp;
1219
1220 for (i = av_len (dec.shareable) + 1; i--; )
1221 if ((svp = av_fetch (dec.shareable, i, 0)))
1222 sv_setsv (*svp, &PL_sv_undef);
1223 }
1224
961 SvREFCNT_dec (sv); 1225 SvREFCNT_dec (sv);
962 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1226 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
963 } 1227 }
964 1228
965 sv = sv_2mortal (sv); 1229 sv = sv_2mortal (sv);
966 1230
967 return sv; 1231 return sv;
968} 1232}
969 1233
1234/////////////////////////////////////////////////////////////////////////////
1235// incremental parser
1236
1237#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1238
1239// returns 0 for notyet, 1 for success or error
1240static int
1241incr_parse (CBOR *self, SV *cborstr)
1242{
1243 STRLEN cur;
1244 SvPV (cborstr, cur);
1245
1246 while (ecb_expect_true (self->incr_need <= cur))
1247 {
1248 // table of integer count bytes
1249 static I8 incr_len[MINOR_MASK + 1] = {
1250 0, 0, 0, 0, 0, 0, 0, 0,
1251 0, 0, 0, 0, 0, 0, 0, 0,
1252 0, 0, 0, 0, 0, 0, 0, 0,
1253 1, 2, 4, 8,-1,-1,-1,-2
1254 };
1255
1256 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1257 U8 m = *p & MINOR_MASK;
1258 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1259 I8 ilen = incr_len[m];
1260
1261 self->incr_need = self->incr_pos + 1;
1262
1263 if (ecb_expect_false (ilen < 0))
1264 {
1265 if (m != MINOR_INDEF)
1266 return 1; // error
1267
1268 if (*p == (MAJOR_MISC | MINOR_INDEF))
1269 {
1270 if (count >= 0)
1271 return 1; // error
1272
1273 count = 1;
1274 }
1275 else
1276 {
1277 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1278 count = -1;
1279 }
1280 }
1281 else
1282 {
1283 self->incr_need += ilen;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 int major = *p >> MAJOR_SHIFT;
1288
1289 switch (major)
1290 {
1291 case MAJOR_TAG >> MAJOR_SHIFT:
1292 ++count; // tags merely prefix another value
1293 break;
1294
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 case MAJOR_ARRAY >> MAJOR_SHIFT:
1298 case MAJOR_MAP >> MAJOR_SHIFT:
1299 {
1300 UV len;
1301
1302 if (ecb_expect_false (ilen))
1303 {
1304 len = 0;
1305
1306 do {
1307 len = (len << 8) | *++p;
1308 } while (--ilen);
1309 }
1310 else
1311 len = m;
1312
1313 switch (major)
1314 {
1315 case MAJOR_BYTES >> MAJOR_SHIFT:
1316 case MAJOR_TEXT >> MAJOR_SHIFT:
1317 self->incr_need += len;
1318 if (ecb_expect_false (self->incr_need > cur))
1319 return 0;
1320
1321 break;
1322
1323 case MAJOR_MAP >> MAJOR_SHIFT:
1324 len <<= 1;
1325 case MAJOR_ARRAY >> MAJOR_SHIFT:
1326 if (len)
1327 {
1328 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1329 count = len + 1;
1330 }
1331 break;
1332 }
1333 }
1334 }
1335 }
1336
1337 self->incr_pos = self->incr_need;
1338
1339 if (count > 0)
1340 {
1341 while (!--count)
1342 {
1343 if (!AvFILLp (self->incr_count))
1344 return 1; // done
1345
1346 SvREFCNT_dec_NN (av_pop (self->incr_count));
1347 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1348 }
1349
1350 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1351 }
1352 }
1353
1354 return 0;
1355}
1356
1357
970///////////////////////////////////////////////////////////////////////////// 1358/////////////////////////////////////////////////////////////////////////////
971// XS interface functions 1359// XS interface functions
972 1360
973MODULE = CBOR::XS PACKAGE = CBOR::XS 1361MODULE = CBOR::XS PACKAGE = CBOR::XS
974 1362
981 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1369 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
982 1370
983 types_true = get_bool ("Types::Serialiser::true" ); 1371 types_true = get_bool ("Types::Serialiser::true" );
984 types_false = get_bool ("Types::Serialiser::false"); 1372 types_false = get_bool ("Types::Serialiser::false");
985 types_error = get_bool ("Types::Serialiser::error"); 1373 types_error = get_bool ("Types::Serialiser::error");
1374
1375 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
986 1376
987 sv_cbor = newSVpv ("CBOR", 0); 1377 sv_cbor = newSVpv ("CBOR", 0);
988 SvREADONLY_on (sv_cbor); 1378 SvREADONLY_on (sv_cbor);
989} 1379}
990 1380
1012void shrink (CBOR *self, int enable = 1) 1402void shrink (CBOR *self, int enable = 1)
1013 ALIAS: 1403 ALIAS:
1014 shrink = F_SHRINK 1404 shrink = F_SHRINK
1015 allow_unknown = F_ALLOW_UNKNOWN 1405 allow_unknown = F_ALLOW_UNKNOWN
1016 allow_sharing = F_ALLOW_SHARING 1406 allow_sharing = F_ALLOW_SHARING
1017 dedup_keys = F_DEDUP_KEYS 1407 allow_cycles = F_ALLOW_CYCLES
1018 dedup_strings = F_DEDUP_STRINGS 1408 pack_strings = F_PACK_STRINGS
1409 utf8_strings = F_UTF8_STRINGS
1410 validate_utf8 = F_VALIDATE_UTF8
1019 PPCODE: 1411 PPCODE:
1020{ 1412{
1021 if (enable) 1413 if (enable)
1022 self->flags |= ix; 1414 self->flags |= ix;
1023 else 1415 else
1029void get_shrink (CBOR *self) 1421void get_shrink (CBOR *self)
1030 ALIAS: 1422 ALIAS:
1031 get_shrink = F_SHRINK 1423 get_shrink = F_SHRINK
1032 get_allow_unknown = F_ALLOW_UNKNOWN 1424 get_allow_unknown = F_ALLOW_UNKNOWN
1033 get_allow_sharing = F_ALLOW_SHARING 1425 get_allow_sharing = F_ALLOW_SHARING
1034 get_dedup_keys = F_DEDUP_KEYS 1426 get_allow_cycles = F_ALLOW_CYCLES
1035 get_dedup_strings = F_DEDUP_STRINGS 1427 get_pack_strings = F_PACK_STRINGS
1428 get_validate_utf8 = F_VALIDATE_UTF8
1036 PPCODE: 1429 PPCODE:
1037 XPUSHs (boolSV (self->flags & ix)); 1430 XPUSHs (boolSV (self->flags & ix));
1038 1431
1039void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1040 PPCODE: 1433 PPCODE:
1053 XPUSHs (ST (0)); 1446 XPUSHs (ST (0));
1054 1447
1055int get_max_size (CBOR *self) 1448int get_max_size (CBOR *self)
1056 CODE: 1449 CODE:
1057 RETVAL = self->max_size; 1450 RETVAL = self->max_size;
1451 OUTPUT:
1452 RETVAL
1453
1454void filter (CBOR *self, SV *filter = 0)
1455 PPCODE:
1456 SvREFCNT_dec (self->filter);
1457 self->filter = filter ? newSVsv (filter) : filter;
1458 XPUSHs (ST (0));
1459
1460SV *get_filter (CBOR *self)
1461 CODE:
1462 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1058 OUTPUT: 1463 OUTPUT:
1059 RETVAL 1464 RETVAL
1060 1465
1061void encode (CBOR *self, SV *scalar) 1466void encode (CBOR *self, SV *scalar)
1062 PPCODE: 1467 PPCODE:
1077 EXTEND (SP, 2); 1482 EXTEND (SP, 2);
1078 PUSHs (sv); 1483 PUSHs (sv);
1079 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1484 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1080} 1485}
1081 1486
1487void incr_parse (CBOR *self, SV *cborstr)
1488 ALIAS:
1489 incr_parse_multiple = 1
1490 PPCODE:
1491{
1492 if (SvUTF8 (cborstr))
1493 sv_utf8_downgrade (cborstr, 0);
1494
1495 if (!self->incr_count)
1496 {
1497 self->incr_count = newAV ();
1498 self->incr_pos = 0;
1499 self->incr_need = 1;
1500
1501 av_push (self->incr_count, newSViv (1));
1502 }
1503
1504 do
1505 {
1506 if (!incr_parse (self, cborstr))
1507 {
1508 if (self->incr_need > self->max_size && self->max_size)
1509 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1510 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1511
1512 break;
1513 }
1514
1515 SV *sv;
1516 char *offset;
1517
1518 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1519 XPUSHs (sv);
1520
1521 sv_chop (cborstr, offset);
1522
1523 av_clear (self->incr_count);
1524 av_push (self->incr_count, newSViv (1));
1525
1526 self->incr_pos = 0;
1527 self->incr_need = self->incr_pos + 1;
1528 }
1529 while (ix);
1530}
1531
1532void incr_reset (CBOR *self)
1533 CODE:
1534{
1535 SvREFCNT_dec (self->incr_count);
1536 self->incr_count = 0;
1537}
1538
1539void DESTROY (CBOR *self)
1540 PPCODE:
1541 cbor_free (self);
1542
1082PROTOTYPES: ENABLE 1543PROTOTYPES: ENABLE
1083 1544
1084void encode_cbor (SV *scalar) 1545void encode_cbor (SV *scalar)
1546 ALIAS:
1547 encode_cbor = 0
1548 encode_cbor_sharing = F_ALLOW_SHARING
1085 PPCODE: 1549 PPCODE:
1086{ 1550{
1087 CBOR cbor; 1551 CBOR cbor;
1088 cbor_init (&cbor); 1552 cbor_init (&cbor);
1553 cbor.flags |= ix;
1089 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1554 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1090 XPUSHs (scalar); 1555 XPUSHs (scalar);
1091} 1556}
1092 1557
1093void decode_cbor (SV *cborstr) 1558void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines