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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.18 by root, Wed Nov 20 01:09:46 2013 UTC vs.
Revision 1.57 by root, Fri Nov 25 12:16:12 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_SHARABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 104#define F_PACK_STRINGS 0x00000010UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 105#define F_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
64 108
65#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
66 110
67#define SB do { 111#define SB do {
68#define SE } while (0) 112#define SE } while (0)
80# define CBOR_SLOW 0 124# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 125# define CBOR_STASH cbor_stash
82#endif 126#endif
83 127
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 128static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 129static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 130
87typedef struct { 131typedef struct {
88 U32 flags; 132 U32 flags;
89 U32 max_depth; 133 U32 max_depth;
90 STRLEN max_size; 134 STRLEN max_size;
135 SV *filter;
136
137 // for the incremental parser
138 STRLEN incr_pos; // the current offset into the text
139 STRLEN incr_need; // minimum bytes needed to decode
140 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 141} CBOR;
92 142
93ecb_inline void 143ecb_inline void
94cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
95{ 145{
96 Zero (cbor, 1, CBOR); 146 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 147 cbor->max_depth = 512;
148}
149
150ecb_inline void
151cbor_free (CBOR *cbor)
152{
153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
98} 155}
99 156
100///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
101// utility functions 158// utility functions
102 159
124 SvPV_renew (sv, SvCUR (sv) + 1); 181 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 182#endif
126 } 183 }
127} 184}
128 185
186// minimum length of a string to be registered for stringref
187ecb_inline int
188minimum_string_length (UV idx)
189{
190 return idx > 23
191 ? idx > 0xffU
192 ? idx > 0xffffU
193 ? idx > 0xffffffffU
194 ? 11
195 : 7
196 : 5
197 : 4
198 : 3;
199}
200
129///////////////////////////////////////////////////////////////////////////// 201/////////////////////////////////////////////////////////////////////////////
130// encoder 202// encoder
131 203
132// structure used for encoding CBOR 204// structure used for encoding CBOR
133typedef struct 205typedef struct
135 char *cur; // SvPVX (sv) + current output position 207 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 208 char *end; // SvEND (sv)
137 SV *sv; // result scalar 209 SV *sv; // result scalar
138 CBOR cbor; 210 CBOR cbor;
139 U32 depth; // recursion level 211 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 212 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
213 UV stringref_idx;
141 HV *sharable; // ptr => index, or 0 214 HV *shareable; // ptr => index, or 0
142 HV *sharable_idx; 215 UV shareable_idx;
143} enc_t; 216} enc_t;
144 217
145ecb_inline void 218ecb_inline void
146need (enc_t *enc, STRLEN len) 219need (enc_t *enc, STRLEN len)
147{ 220{
164static void 237static void
165encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
166{ 239{
167 need (enc, 9); 240 need (enc, 9);
168 241
169 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
171 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
172 { 245 {
173 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 247 *enc->cur++ = len;
175 } 248 }
176 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
177 { 250 {
178 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 253 *enc->cur++ = len;
181 } 254 }
182 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
183 { 256 {
184 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 261 *enc->cur++ = len;
189 } 262 }
190 else 263 else
191 { 264 {
192 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 272 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 273 *enc->cur++ = len;
201 } 274 }
202} 275}
203 276
204static void 277ecb_inline void
278encode_tag (enc_t *enc, UV tag)
279{
280 encode_uint (enc, MAJOR_TAG, tag);
281}
282
283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 286{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 287 STRLEN ulen = len;
288 U8 *p, *pend = (U8 *)str + len;
289
290 for (p = (U8 *)str; p < pend; ++p)
291 ulen += *p >> 7; // count set high bits
292
293 encode_uint (enc, MAJOR_TEXT, ulen);
294
295 need (enc, ulen);
296 for (p = (U8 *)str; p < pend; ++p)
297 if (*p < 0x80)
298 *enc->cur++ = *p;
299 else
300 {
301 *enc->cur++ = 0xc0 + (*p >> 6);
302 *enc->cur++ = 0x80 + (*p & 63);
303 }
304}
305
306ecb_inline void
307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 317 need (enc, len);
209 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
210 enc->cur += len; 319 enc->cur += len;
211} 320}
212 321
213ecb_inline void 322ecb_inline void
214encode_tag (enc_t *enc, UV tag) 323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
215{ 324{
216 encode_uint (enc, 0xc0, tag); 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
217} 326 {
327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
218 328
219static int
220encode_sharable2 (enc_t *enc, SV *sv)
221{
222 if (!enc->sharable)
223 enc->sharable = (HV *)sv_2mortal ((SV *)newHV ());
224
225 SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227 if (SvOK (*svp)) 329 if (SvOK (*svp))
228 { 330 {
331 // already registered, use stringref
229 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 332 encode_tag (enc, CBOR_TAG_STRINGREF);
230 encode_uint (enc, 0x00, SvUV (*svp)); 333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232 return 1; 334 return;
233 } 335 }
234 else 336 else if (len >= minimum_string_length (enc->stringref_idx))
235 { 337 {
338 // register only
236 sv_setuv (*svp, enc->sharable_idx++); 339 sv_setuv (*svp, enc->stringref_idx);
237 encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); 340 ++enc->stringref_idx;
238 341 }
239 return 0;
240 } 342 }
241}
242 343
243ecb_inline int 344 encode_str (enc, upgrade_utf8, utf8, str, len);
244encode_sharable (enc_t *enc, SV *sv)
245{
246 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
247 && ecb_expect_false (SvREFCNT (sv) > 1))
248 return encode_sharable2 (enc, sv);
249
250 return 0;
251} 345}
252 346
253static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
254 348
255static void 349static void
260 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
262 356
263 ++enc->depth; 357 ++enc->depth;
264 358
265 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 360
361 if (SvMAGICAL (av))
267 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
268 { 363 {
269 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 366 }
367 else
368 for (i = 0; i <= len; ++i)
369 {
370 SV *sv = AvARRAY (av)[i];
371 encode_sv (enc, sv ? sv : &PL_sv_undef);
372 }
272 373
273 --enc->depth; 374 --enc->depth;
274} 375}
275 376
276static void 377static void
285 386
286 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
288 389
289 if (mg) 390 if (mg)
290 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 392 else
292 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
293 394
294 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
295 { 396 {
296 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
298 else 399 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 401
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 403 }
303 404
304 if (mg) 405 if (mg)
305 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 407
307 --enc->depth; 408 --enc->depth;
308} 409}
309 410
310// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
311static void 412static void
312encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
313{ 414{
314 svtype svt;
315
316 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
416
317 svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 418
322 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
323 { 420 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 421 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 422 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 428 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 430
334 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
335 GV *method;
336 432
337 if (stash == boolean_stash) 433 if (stash == boolean_stash)
338 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 434 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
436 return;
437 }
339 else if (stash == error_stash) 438 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
341 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
342 { 444 {
343 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 447
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 449 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
450
451 return;
452 }
453 }
454
455 if (ecb_expect_false (SvREFCNT (sv) > 1)
456 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
457 {
458 if (!enc->shareable)
459 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
460
461 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
462
463 if (SvOK (*svp))
348 } 464 {
465 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
466 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
467 return;
468 }
469 else
470 {
471 sv_setuv (*svp, enc->shareable_idx);
472 ++enc->shareable_idx;
473 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
474 }
475 }
476
477 if (ecb_expect_false (SvOBJECT (sv)))
478 {
479 HV *stash = SvSTASH (sv);
480 GV *method;
481
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 483 {
351 dSP; 484 dSP;
352 485
353 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
354 // we re-bless the reference to get overload and other niceties right 488 // we re-bless the reference to get overload and other niceties right
355 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
356 490
357 PUTBACK; 491 PUTBACK;
358 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
371 } 505 }
372 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
373 { 507 {
374 dSP; 508 dSP;
375 509
376 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
377 EXTEND (SP, 2); 513 EXTEND (SP, 2);
378 // we re-bless the reference to get overload and other niceties right 514 // we re-bless the reference to get overload and other niceties right
379 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
380 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
381 517
386 // catch this surprisingly common error 522 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
388 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389 525
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 529
394 while (count) 530 while (count)
395 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
396 532
397 PUTBACK; 533 PUTBACK;
418{ 554{
419 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
420 556
421 need (enc, 9); 557 need (enc, 9);
422 558
423 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 561 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
427 { 563 {
428 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
429 565
430 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
431 567
432 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
434 570
435 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
437 } 573 }
438 else 574 else
439 { 575 {
440 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
441 577
442 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
443 579
444 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
446 582
447 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
452static void 588static void
453encode_sv (enc_t *enc, SV *sv) 589encode_sv (enc_t *enc, SV *sv)
454{ 590{
455 SvGETMAGIC (sv); 591 SvGETMAGIC (sv);
456 592
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
461 { 594 {
462 STRLEN len; 595 STRLEN len;
463 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
465 } 598 }
466 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 600 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
469 { 602 {
470 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
472 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
474 else 607 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
476 } 609 }
477 else if (SvROK (sv)) 610 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 616 else
484 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 617 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 619}
487 620
488static SV * 621static SV *
489encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
490{ 623{
491 enc_t enc = { }; 624 enc_t enc = { 0 };
492 625
493 enc.cbor = *cbor; 626 enc.cbor = *cbor;
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
497 630
498 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
632
633 if (cbor->flags & F_PACK_STRINGS)
634 {
635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
638 }
639
499 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
500 641
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 643 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503 644
517 U8 *end; // end of input string 658 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 659 const char *err; // parse error, if != 0
519 CBOR cbor; 660 CBOR cbor;
520 U32 depth; // recursion depth 661 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
522 AV *sharable; 663 AV *shareable;
664 AV *stringref;
665 SV *decode_tagged;
523} dec_t; 666} dec_t;
524 667
525#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
526 669
527#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 670#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR ("unexpected end of CBOR data")
528 671
529#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 672#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
530#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
531 674
532static UV 675static UV
533decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
534{ 677{
535 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
536 { 679 ++dec->cur;
537 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
538 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
539 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
540 return *dec->cur++ & 31;
541 680
542 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
543 WANT (2); 685 WANT (1);
544 dec->cur += 2; 686 dec->cur += 1;
545 return dec->cur[-1]; 687 return dec->cur[-1];
546 688 }
547 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
548 WANT (3); 691 WANT (2);
549 dec->cur += 3; 692 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
552 695 }
553 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
554 WANT (5); 698 WANT (4);
555 dec->cur += 5; 699 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
560 704 }
561 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
562 WANT (9); 707 WANT (8);
563 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
564 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
568 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
572 723 }
573 default: 724 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 726
577fail: 727fail:
578 return 0; 728 return 0;
579} 729}
580 730
585{ 735{
586 AV *av = newAV (); 736 AV *av = newAV ();
587 737
588 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
589 739
590 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 741 {
592 ++dec->cur; 742 ++dec->cur;
593 743
594 for (;;) 744 for (;;)
595 { 745 {
596 WANT (1); 746 WANT (1);
597 747
598 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 749 {
600 ++dec->cur; 750 ++dec->cur;
601 break; 751 break;
602 } 752 }
603 753
604 av_push (av, decode_sv (dec)); 754 av_push (av, decode_sv (dec));
605 } 755 }
606 } 756 }
607 else 757 else
608 { 758 {
609 int i, len = decode_uint (dec); 759 UV i, len = decode_uint (dec);
610 760
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
611 av_fill (av, len - 1); 762 av_fill (av, len - 1);
612 763
613 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
615 } 766 }
625 776
626static void 777static void
627decode_he (dec_t *dec, HV *hv) 778decode_he (dec_t *dec, HV *hv)
628{ 779{
629 // for speed reasons, we specialcase single-string 780 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 781 // byte or utf-8 strings as keys, but only when !stringref
631 782
632 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 783 if (ecb_expect_true (!dec->stringref))
784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633 { 785 {
634 I32 len = decode_uint (dec); 786 STRLEN len = decode_uint (dec);
635 char *key = (char *)dec->cur; 787 char *key = (char *)dec->cur;
636 788
789 WANT (len);
637 dec->cur += len; 790 dec->cur += len;
638 791
639 hv_store (hv, key, len, decode_sv (dec), 0); 792 hv_store (hv, key, len, decode_sv (dec), 0);
793
794 return;
640 } 795 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 797 {
643 I32 len = decode_uint (dec); 798 STRLEN len = decode_uint (dec);
644 char *key = (char *)dec->cur; 799 char *key = (char *)dec->cur;
645 800
801 WANT (len);
646 dec->cur += len; 802 dec->cur += len;
647 803
804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
805 if (!is_utf8_string (key, len))
806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
807
648 hv_store (hv, key, -len, decode_sv (dec), 0); 808 hv_store (hv, key, -len, decode_sv (dec), 0);
809
810 return;
649 } 811 }
650 else 812
651 {
652 SV *k = decode_sv (dec); 813 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 814 SV *v = decode_sv (dec);
654 815
655 hv_store_ent (hv, k, v, 0); 816 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 817 SvREFCNT_dec (k);
657 } 818
819fail:
820 ;
658} 821}
659 822
660static SV * 823static SV *
661decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
662{ 825{
663 HV *hv = newHV (); 826 HV *hv = newHV ();
664 827
665 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
666 829
667 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 831 {
669 ++dec->cur; 832 ++dec->cur;
670 833
671 for (;;) 834 for (;;)
672 { 835 {
673 WANT (1); 836 WANT (1);
674 837
675 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
676 { 839 {
677 ++dec->cur; 840 ++dec->cur;
678 break; 841 break;
679 } 842 }
680 843
681 decode_he (dec, hv); 844 decode_he (dec, hv);
682 } 845 }
683 } 846 }
684 else 847 else
685 { 848 {
686 int pairs = decode_uint (dec); 849 UV pairs = decode_uint (dec);
850
851 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
687 852
688 while (pairs--) 853 while (pairs--)
689 decode_he (dec, hv); 854 decode_he (dec, hv);
690 } 855 }
691 856
701static SV * 866static SV *
702decode_str (dec_t *dec, int utf8) 867decode_str (dec_t *dec, int utf8)
703{ 868{
704 SV *sv = 0; 869 SV *sv = 0;
705 870
706 if ((*dec->cur & 31) == 31) 871 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
707 { 872 {
873 // indefinite length strings
708 ++dec->cur; 874 ++dec->cur;
709 875
876 U8 major = *dec->cur & MAJOR_MISC;
877
710 sv = newSVpvn ("", 0); 878 sv = newSVpvn ("", 0);
711 879
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 880 for (;;)
714 { 881 {
715 WANT (1); 882 WANT (1);
716 883
717 if (*dec->cur == (0xe0 | 31)) 884 if ((*dec->cur - major) > LENGTH_EXT8)
885 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 886 {
719 ++dec->cur; 887 ++dec->cur;
720 break; 888 break;
721 } 889 }
890 else
891 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 892
723 sv_catsv (sv, decode_sv (dec)); 893 STRLEN len = decode_uint (dec);
894
895 WANT (len);
896 sv_catpvn (sv, dec->cur, len);
897 dec->cur += len;
724 } 898 }
725 } 899 }
726 else 900 else
727 { 901 {
728 STRLEN len = decode_uint (dec); 902 STRLEN len = decode_uint (dec);
729 903
730 WANT (len); 904 WANT (len);
731 sv = newSVpvn (dec->cur, len); 905 sv = newSVpvn (dec->cur, len);
732 dec->cur += len; 906 dec->cur += len;
907
908 if (ecb_expect_false (dec->stringref)
909 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
910 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733 } 911 }
734 912
735 if (utf8) 913 if (utf8)
914 {
915 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
916 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
917 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
918
736 SvUTF8_on (sv); 919 SvUTF8_on (sv);
920 }
737 921
738 return sv; 922 return sv;
739 923
740fail: 924fail:
741 SvREFCNT_dec (sv); 925 SvREFCNT_dec (sv);
743} 927}
744 928
745static SV * 929static SV *
746decode_tagged (dec_t *dec) 930decode_tagged (dec_t *dec)
747{ 931{
932 SV *sv = 0;
748 UV tag = decode_uint (dec); 933 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 934
935 WANT (1);
750 936
751 switch (tag) 937 switch (tag)
752 { 938 {
753 case CBOR_TAG_MAGIC: 939 case CBOR_TAG_MAGIC:
754 return sv; 940 sv = decode_sv (dec);
941 break;
755 942
756 case CBOR_TAG_INDIRECTION: 943 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 944 sv = newRV_noinc (decode_sv (dec));
945 break;
758 946
947 case CBOR_TAG_STRINGREF_NAMESPACE:
948 {
949 // do nmot use SAVETMPS/FREETMPS, as these will
950 // erase mortalised caches, e.g. "shareable"
951 ENTER;
952
953 SAVESPTR (dec->stringref);
954 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
955
956 sv = decode_sv (dec);
957
958 LEAVE;
959 }
960 break;
961
962 case CBOR_TAG_STRINGREF:
963 {
964 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
965 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
966
967 UV idx = decode_uint (dec);
968
969 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
970 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
971
972 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
973 }
974 break;
975
759 case CBOR_TAG_VALUE_SHARABLE: 976 case CBOR_TAG_VALUE_SHAREABLE:
977 {
760 if (ecb_expect_false (!dec->sharable)) 978 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 979 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 980
981 if (dec->cbor.flags & F_ALLOW_CYCLES)
982 {
983 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 984 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 985
765 return sv; 986 SV *osv = decode_sv (dec);
987 sv_setsv (sv, osv);
988 SvREFCNT_dec_NN (osv);
989 }
990 else
991 {
992 av_push (dec->shareable, &PL_sv_undef);
993 int idx = AvFILLp (dec->shareable);
994 sv = decode_sv (dec);
995 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
996 }
997 }
998 break;
766 999
767 case CBOR_TAG_VALUE_SHAREDREF: 1000 case CBOR_TAG_VALUE_SHAREDREF:
768 { 1001 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 1002 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 1003 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 1004
1005 UV idx = decode_uint (dec);
1006
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 1007 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1008 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 1009
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); 1010 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1011
1012 if (sv == &PL_sv_undef)
1013 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
778 } 1014 }
1015 break;
779 1016
780 case CBOR_TAG_PERL_OBJECT: 1017 case CBOR_TAG_PERL_OBJECT:
781 { 1018 {
1019 sv = decode_sv (dec);
1020
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1021 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 1022 ERR ("corrupted CBOR data (non-array perl object)");
784 1023
785 AV *av = (AV *)SvRV (sv); 1024 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 1025 int len = av_len (av) + 1;
794 if (!method) 1033 if (!method)
795 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1034 ERR ("cannot decode perl-object (package does not have a THAW method)");
796 1035
797 dSP; 1036 dSP;
798 1037
799 ENTER; SAVETMPS; PUSHMARK (SP); 1038 ENTER; SAVETMPS;
1039 PUSHMARK (SP);
800 EXTEND (SP, len + 1); 1040 EXTEND (SP, len + 1);
801 // we re-bless the reference to get overload and other niceties right 1041 // we re-bless the reference to get overload and other niceties right
802 PUSHs (*av_fetch (av, 0, 1)); 1042 PUSHs (*av_fetch (av, 0, 1));
803 PUSHs (sv_cbor); 1043 PUSHs (sv_cbor);
804 1044
821 sv = SvREFCNT_inc (POPs); 1061 sv = SvREFCNT_inc (POPs);
822 1062
823 PUTBACK; 1063 PUTBACK;
824 1064
825 FREETMPS; LEAVE; 1065 FREETMPS; LEAVE;
826
827 return sv;
828 } 1066 }
1067 break;
829 1068
830 default: 1069 default:
831 { 1070 {
1071 sv = decode_sv (dec);
1072
1073 dSP;
1074 ENTER; SAVETMPS;
1075 SAVESTACK_POS ();
1076 PUSHMARK (SP);
1077 EXTEND (SP, 2);
1078 PUSHs (newSVuv (tag));
1079 PUSHs (sv);
1080
1081 PUTBACK;
1082 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1083 SPAGAIN;
1084
1085 if (SvTRUE (ERRSV))
1086 {
1087 FREETMPS; LEAVE;
1088 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1089 }
1090
1091 if (count)
1092 {
1093 SvREFCNT_dec (sv);
1094 sv = SvREFCNT_inc (POPs);
1095 }
1096 else
1097 {
832 AV *av = newAV (); 1098 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1099 av_push (av, newSVuv (tag));
834 av_push (av, sv); 1100 av_push (av, sv);
835 1101
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1102 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1103 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1104 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1105 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1106 }
1107
1108 PUTBACK;
1109
1110 FREETMPS; LEAVE;
1111 }
1112 break;
842 } 1113 }
1114
1115 return sv;
843 1116
844fail: 1117fail:
845 SvREFCNT_dec (sv); 1118 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1119 return &PL_sv_undef;
847} 1120}
849static SV * 1122static SV *
850decode_sv (dec_t *dec) 1123decode_sv (dec_t *dec)
851{ 1124{
852 WANT (1); 1125 WANT (1);
853 1126
854 switch (*dec->cur >> 5) 1127 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1128 {
856 case 0: // unsigned int 1129 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1130 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1131 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1132 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1133 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1134 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1135 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1136
864 case 4: // array 1137 case MAJOR_MISC >> MAJOR_SHIFT:
865 return decode_av (dec);
866 case 5: // map
867 return decode_hv (dec);
868 case 6: // tag
869 return decode_tagged (dec);
870 case 7: // misc
871 switch (*dec->cur++ & 31) 1138 switch (*dec->cur++ & MINOR_MASK)
872 { 1139 {
873 case 20: 1140 case SIMPLE_FALSE:
874#if CBOR_SLOW 1141#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1142 types_false = get_bool ("Types::Serialiser::false");
876#endif 1143#endif
877 return newSVsv (types_false); 1144 return newSVsv (types_false);
878 case 21: 1145 case SIMPLE_TRUE:
879#if CBOR_SLOW 1146#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1147 types_true = get_bool ("Types::Serialiser::true");
881#endif 1148#endif
882 return newSVsv (types_true); 1149 return newSVsv (types_true);
883 case 22: 1150 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1151 return newSVsv (&PL_sv_undef);
885 case 23: 1152 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1153#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1154 types_error = get_bool ("Types::Serialiser::error");
888#endif 1155#endif
889 return newSVsv (types_error); 1156 return newSVsv (types_error);
890 1157
891 case 25: 1158 case MISC_FLOAT16:
892 { 1159 {
893 WANT (2); 1160 WANT (2);
894 1161
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1162 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1163 dec->cur += 2;
897 1164
898 return newSVnv (ecb_binary16_to_float (fp)); 1165 return newSVnv (ecb_binary16_to_float (fp));
899 } 1166 }
900 1167
901 case 26: 1168 case MISC_FLOAT32:
902 { 1169 {
903 uint32_t fp; 1170 uint32_t fp;
904 WANT (4); 1171 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1172 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1173 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1176 fp = ecb_bswap32 (fp);
910 1177
911 return newSVnv (ecb_binary32_to_float (fp)); 1178 return newSVnv (ecb_binary32_to_float (fp));
912 } 1179 }
913 1180
914 case 27: 1181 case MISC_FLOAT64:
915 { 1182 {
916 uint64_t fp; 1183 uint64_t fp;
917 WANT (8); 1184 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1185 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1186 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1189 fp = ecb_bswap64 (fp);
923 1190
924 return newSVnv (ecb_binary64_to_double (fp)); 1191 return newSVnv (ecb_binary64_to_double (fp));
925 } 1192 }
926 1193
927 // 0..19 unassigned 1194 // 0..19 unassigned simple
928 // 24 reserved + unassigned (reserved values are not encodable) 1195 // 24 reserved + unassigned simple (reserved values are not encodable)
1196 // 28-30 unassigned misc
1197 // 31 break code
929 default: 1198 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1199 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1200 }
932 1201
933 break; 1202 break;
934 } 1203 }
935 1204
938} 1207}
939 1208
940static SV * 1209static SV *
941decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1210decode_cbor (SV *string, CBOR *cbor, char **offset_return)
942{ 1211{
943 dec_t dec = { }; 1212 dec_t dec = { 0 };
944 SV *sv; 1213 SV *sv;
945 STRLEN len; 1214 STRLEN len;
946 char *data = SvPVbyte (string, len); 1215 char *data = SvPVbyte (string, len);
947 1216
948 if (len > cbor->max_size && cbor->max_size) 1217 if (len > cbor->max_size && cbor->max_size)
962 if (dec.cur != dec.end && !dec.err) 1231 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1232 dec.err = "garbage after CBOR object";
964 1233
965 if (dec.err) 1234 if (dec.err)
966 { 1235 {
1236 if (dec.shareable)
1237 {
1238 // need to break cyclic links, which whould all be in shareable
1239 int i;
1240 SV **svp;
1241
1242 for (i = av_len (dec.shareable) + 1; i--; )
1243 if ((svp = av_fetch (dec.shareable, i, 0)))
1244 sv_setsv (*svp, &PL_sv_undef);
1245 }
1246
967 SvREFCNT_dec (sv); 1247 SvREFCNT_dec (sv);
968 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1248 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
969 } 1249 }
970 1250
971 sv = sv_2mortal (sv); 1251 sv = sv_2mortal (sv);
972 1252
973 return sv; 1253 return sv;
974} 1254}
975 1255
1256/////////////////////////////////////////////////////////////////////////////
1257// incremental parser
1258
1259#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1260
1261// returns 0 for notyet, 1 for success or error
1262static int
1263incr_parse (CBOR *self, SV *cborstr)
1264{
1265 STRLEN cur;
1266 SvPV (cborstr, cur);
1267
1268 while (ecb_expect_true (self->incr_need <= cur))
1269 {
1270 // table of integer count bytes
1271 static I8 incr_len[MINOR_MASK + 1] = {
1272 0, 0, 0, 0, 0, 0, 0, 0,
1273 0, 0, 0, 0, 0, 0, 0, 0,
1274 0, 0, 0, 0, 0, 0, 0, 0,
1275 1, 2, 4, 8,-1,-1,-1,-2
1276 };
1277
1278 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1279 U8 m = *p & MINOR_MASK;
1280 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1281 I8 ilen = incr_len[m];
1282
1283 self->incr_need = self->incr_pos + 1;
1284
1285 if (ecb_expect_false (ilen < 0))
1286 {
1287 if (m != MINOR_INDEF)
1288 return 1; // error
1289
1290 if (*p == (MAJOR_MISC | MINOR_INDEF))
1291 {
1292 if (count >= 0)
1293 return 1; // error
1294
1295 count = 1;
1296 }
1297 else
1298 {
1299 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1300 count = -1;
1301 }
1302 }
1303 else
1304 {
1305 self->incr_need += ilen;
1306 if (ecb_expect_false (self->incr_need > cur))
1307 return 0;
1308
1309 int major = *p >> MAJOR_SHIFT;
1310
1311 switch (major)
1312 {
1313 case MAJOR_TAG >> MAJOR_SHIFT:
1314 ++count; // tags merely prefix another value
1315 break;
1316
1317 case MAJOR_BYTES >> MAJOR_SHIFT:
1318 case MAJOR_TEXT >> MAJOR_SHIFT:
1319 case MAJOR_ARRAY >> MAJOR_SHIFT:
1320 case MAJOR_MAP >> MAJOR_SHIFT:
1321 {
1322 UV len;
1323
1324 if (ecb_expect_false (ilen))
1325 {
1326 len = 0;
1327
1328 do {
1329 len = (len << 8) | *++p;
1330 } while (--ilen);
1331 }
1332 else
1333 len = m;
1334
1335 switch (major)
1336 {
1337 case MAJOR_BYTES >> MAJOR_SHIFT:
1338 case MAJOR_TEXT >> MAJOR_SHIFT:
1339 self->incr_need += len;
1340 if (ecb_expect_false (self->incr_need > cur))
1341 return 0;
1342
1343 break;
1344
1345 case MAJOR_MAP >> MAJOR_SHIFT:
1346 len <<= 1;
1347 case MAJOR_ARRAY >> MAJOR_SHIFT:
1348 if (len)
1349 {
1350 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1351 count = len + 1;
1352 }
1353 break;
1354 }
1355 }
1356 }
1357 }
1358
1359 self->incr_pos = self->incr_need;
1360
1361 if (count > 0)
1362 {
1363 while (!--count)
1364 {
1365 if (!AvFILLp (self->incr_count))
1366 return 1; // done
1367
1368 SvREFCNT_dec_NN (av_pop (self->incr_count));
1369 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1370 }
1371
1372 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1373 }
1374 }
1375
1376 return 0;
1377}
1378
1379
976///////////////////////////////////////////////////////////////////////////// 1380/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1381// XS interface functions
978 1382
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1383MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1384
988 1392
989 types_true = get_bool ("Types::Serialiser::true" ); 1393 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1394 types_false = get_bool ("Types::Serialiser::false");
991 types_error = get_bool ("Types::Serialiser::error"); 1395 types_error = get_bool ("Types::Serialiser::error");
992 1396
1397 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1398
993 sv_cbor = newSVpv ("CBOR", 0); 1399 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1400 SvREADONLY_on (sv_cbor);
1401
1402 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
995} 1403}
996 1404
997PROTOTYPES: DISABLE 1405PROTOTYPES: DISABLE
998 1406
999void CLONE (...) 1407void CLONE (...)
1018void shrink (CBOR *self, int enable = 1) 1426void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1427 ALIAS:
1020 shrink = F_SHRINK 1428 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1429 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1430 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1431 allow_cycles = F_ALLOW_CYCLES
1024 dedup_strings = F_DEDUP_STRINGS 1432 pack_strings = F_PACK_STRINGS
1433 text_keys = F_TEXT_KEYS
1434 text_strings = F_TEXT_STRINGS
1435 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1436 PPCODE:
1026{ 1437{
1027 if (enable) 1438 if (enable)
1028 self->flags |= ix; 1439 self->flags |= ix;
1029 else 1440 else
1035void get_shrink (CBOR *self) 1446void get_shrink (CBOR *self)
1036 ALIAS: 1447 ALIAS:
1037 get_shrink = F_SHRINK 1448 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1449 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1450 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1451 get_allow_cycles = F_ALLOW_CYCLES
1041 get_dedup_strings = F_DEDUP_STRINGS 1452 get_pack_strings = F_PACK_STRINGS
1453 get_text_keys = F_TEXT_KEYS
1454 get_text_strings = F_TEXT_STRINGS
1455 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1456 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1457 XPUSHs (boolSV (self->flags & ix));
1044 1458
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1459void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1460 PPCODE:
1059 XPUSHs (ST (0)); 1473 XPUSHs (ST (0));
1060 1474
1061int get_max_size (CBOR *self) 1475int get_max_size (CBOR *self)
1062 CODE: 1476 CODE:
1063 RETVAL = self->max_size; 1477 RETVAL = self->max_size;
1478 OUTPUT:
1479 RETVAL
1480
1481void filter (CBOR *self, SV *filter = 0)
1482 PPCODE:
1483 SvREFCNT_dec (self->filter);
1484 self->filter = filter ? newSVsv (filter) : filter;
1485 XPUSHs (ST (0));
1486
1487SV *get_filter (CBOR *self)
1488 CODE:
1489 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064 OUTPUT: 1490 OUTPUT:
1065 RETVAL 1491 RETVAL
1066 1492
1067void encode (CBOR *self, SV *scalar) 1493void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1494 PPCODE:
1083 EXTEND (SP, 2); 1509 EXTEND (SP, 2);
1084 PUSHs (sv); 1510 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1511 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1512}
1087 1513
1514void incr_parse (CBOR *self, SV *cborstr)
1515 ALIAS:
1516 incr_parse_multiple = 1
1517 PPCODE:
1518{
1519 if (SvUTF8 (cborstr))
1520 sv_utf8_downgrade (cborstr, 0);
1521
1522 if (!self->incr_count)
1523 {
1524 self->incr_count = newAV ();
1525 self->incr_pos = 0;
1526 self->incr_need = 1;
1527
1528 av_push (self->incr_count, newSViv (1));
1529 }
1530
1531 do
1532 {
1533 if (!incr_parse (self, cborstr))
1534 {
1535 if (self->incr_need > self->max_size && self->max_size)
1536 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1537 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1538
1539 break;
1540 }
1541
1542 SV *sv;
1543 char *offset;
1544
1545 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1546 XPUSHs (sv);
1547
1548 sv_chop (cborstr, offset);
1549
1550 av_clear (self->incr_count);
1551 av_push (self->incr_count, newSViv (1));
1552
1553 self->incr_pos = 0;
1554 self->incr_need = self->incr_pos + 1;
1555 }
1556 while (ix);
1557}
1558
1559void incr_reset (CBOR *self)
1560 CODE:
1561{
1562 SvREFCNT_dec (self->incr_count);
1563 self->incr_count = 0;
1564}
1565
1566void DESTROY (CBOR *self)
1567 PPCODE:
1568 cbor_free (self);
1569
1088PROTOTYPES: ENABLE 1570PROTOTYPES: ENABLE
1089 1571
1090void encode_cbor (SV *scalar) 1572void encode_cbor (SV *scalar)
1573 ALIAS:
1574 encode_cbor = 0
1575 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1576 PPCODE:
1092{ 1577{
1093 CBOR cbor; 1578 CBOR cbor;
1094 cbor_init (&cbor); 1579 cbor_init (&cbor);
1580 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1581 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1582 XPUSHs (scalar);
1097} 1583}
1098 1584
1099void decode_cbor (SV *cborstr) 1585void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines