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 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs.
Revision 1.58 by root, Fri Nov 25 13:27:29 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_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
122 SvPV_shrink_to_cur (sv); 179 SvPV_shrink_to_cur (sv);
123#elif defined (SvPV_renew) 180#elif defined (SvPV_renew)
124 SvPV_renew (sv, SvCUR (sv) + 1); 181 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 182#endif
126 } 183 }
184}
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;
127} 199}
128 200
129///////////////////////////////////////////////////////////////////////////// 201/////////////////////////////////////////////////////////////////////////////
130// encoder 202// encoder
131 203
165static void 237static void
166encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
167{ 239{
168 need (enc, 9); 240 need (enc, 9);
169 241
170 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
171 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
172 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
173 { 245 {
174 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
175 *enc->cur++ = len; 247 *enc->cur++ = len;
176 } 248 }
177 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
178 { 250 {
179 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
180 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
181 *enc->cur++ = len; 253 *enc->cur++ = len;
182 } 254 }
183 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
184 { 256 {
185 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
186 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
187 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
188 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
189 *enc->cur++ = len; 261 *enc->cur++ = len;
190 } 262 }
191 else 263 else
192 { 264 {
193 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
194 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
195 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
196 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
197 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
198 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
200 *enc->cur++ = len >> 8; 272 *enc->cur++ = len >> 8;
201 *enc->cur++ = len; 273 *enc->cur++ = len;
202 } 274 }
203} 275}
204 276
205static 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
206encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
207{ 286{
208 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);
209 need (enc, len); 317 need (enc, len);
210 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
211 enc->cur += len; 319 enc->cur += len;
212} 320}
213 321
214ecb_inline void 322ecb_inline void
215encode_tag (enc_t *enc, UV tag) 323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
216{ 324{
217 encode_uint (enc, 0xc0, tag); 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
326 {
327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
328
329 if (SvOK (*svp))
330 {
331 // already registered, use stringref
332 encode_tag (enc, CBOR_TAG_STRINGREF);
333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
334 return;
335 }
336 else if (len >= minimum_string_length (enc->stringref_idx))
337 {
338 // register only
339 sv_setuv (*svp, enc->stringref_idx);
340 ++enc->stringref_idx;
341 }
342 }
343
344 encode_str (enc, upgrade_utf8, utf8, str, len);
218} 345}
219 346
220static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
221 348
222static void 349static void
227 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
228 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
229 356
230 ++enc->depth; 357 ++enc->depth;
231 358
232 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
233 360
361 if (SvMAGICAL (av))
234 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
235 { 363 {
236 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
237 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
238 } 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 }
239 373
240 --enc->depth; 374 --enc->depth;
241}
242
243ecb_inline void
244encode_he (enc_t *enc, HE *he)
245{
246 if (HeKLEN (he) == HEf_SVKEY)
247 encode_sv (enc, HeSVKEY (he));
248 else
249 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
250} 375}
251 376
252static void 377static void
253encode_hv (enc_t *enc, HV *hv) 378encode_hv (enc_t *enc, HV *hv)
254{ 379{
261 386
262 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
263 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
264 389
265 if (mg) 390 if (mg)
266 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
267 else 392 else
268 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
269 394
270 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
271 { 396 {
272 if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273 {
274 SV **svp;
275
276 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
277 svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/ 398 encode_sv (enc, HeSVKEY (he));
278 else
279 svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281 if (SvOK (*svp))
282 {
283 encode_tag (enc, CBOR_TAG_STRINGREF);
284 encode_uint (enc, 0x00, SvUV (*svp));
285 }
286 else
287 {
288 sv_setuv (*svp, enc->stringref_idx);
289 ++enc->stringref_idx;
290 encode_he (enc, he);
291 }
292 }
293 else 399 else
294 encode_he (enc, he); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
295 401
296 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));
297 } 403 }
298 404
299 if (mg) 405 if (mg)
300 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
301 407
302 --enc->depth; 408 --enc->depth;
303} 409}
304 410
305// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
306static void 412static void
307encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
308{ 414{
309 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
310
311 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312 && ecb_expect_false (SvREFCNT (sv) > 1))
313 {
314 if (!enc->shareable)
315 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319 if (SvOK (*svp))
320 {
321 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322 encode_uint (enc, 0x00, SvUV (*svp));
323 return;
324 }
325 else
326 {
327 sv_setuv (*svp, enc->shareable_idx);
328 ++enc->shareable_idx;
329 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330 }
331 }
332 416
333 svtype svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
334 418
335 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
336 { 420 {
343 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
344 ? cbor_tagged_stash 428 ? cbor_tagged_stash
345 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
346 430
347 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
348 GV *method;
349 432
350 if (stash == boolean_stash) 433 if (stash == boolean_stash)
351 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 }
352 else if (stash == error_stash) 438 else if (stash == error_stash)
353 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
354 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
355 { 444 {
356 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
357 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
358 447
359 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
360 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))
361 } 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
362 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363 { 483 {
364 dSP; 484 dSP;
365 485
366 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
367 // 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
368 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369 490
370 PUTBACK; 491 PUTBACK;
371 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
384 } 505 }
385 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386 { 507 {
387 dSP; 508 dSP;
388 509
389 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
390 EXTEND (SP, 2); 513 EXTEND (SP, 2);
391 // 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
392 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
394 517
399 // catch this surprisingly common error 522 // catch this surprisingly common error
400 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401 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));
402 525
403 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
405 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406 529
407 while (count) 530 while (count)
408 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
409 532
410 PUTBACK; 533 PUTBACK;
431{ 554{
432 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
433 556
434 need (enc, 9); 557 need (enc, 9);
435 558
436 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
437 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
438 //TODO: maybe I32? 561 //TODO: maybe I32?
439 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
440 { 563 {
441 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
442 565
443 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
444 567
445 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
446 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
447 570
448 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
450 } 573 }
451 else 574 else
452 { 575 {
453 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
454 577
455 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
456 579
457 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
458 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
459 582
460 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
469 592
470 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
471 { 594 {
472 STRLEN len; 595 STRLEN len;
473 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
474 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
475 } 598 }
476 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
477 encode_nv (enc, sv); 600 encode_nv (enc, sv);
478 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
479 { 602 {
480 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
481 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
482 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
483 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
484 else 607 else
485 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
486 } 609 }
487 else if (SvROK (sv)) 610 else if (SvROK (sv))
488 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
489 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
490 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
491 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
492 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
493 else 616 else
494 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",
495 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
496} 619}
497 620
498static SV * 621static SV *
499encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
500{ 623{
501 enc_t enc = { }; 624 enc_t enc = { 0 };
502 625
503 enc.cbor = *cbor; 626 enc.cbor = *cbor;
504 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
505 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
506 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
507 630
508 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
509 632
510 if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)) 633 if (cbor->flags & F_PACK_STRINGS)
511 { 634 {
512 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515 } 638 }
537 CBOR cbor; 660 CBOR cbor;
538 U32 depth; // recursion depth 661 U32 depth; // recursion depth
539 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
540 AV *shareable; 663 AV *shareable;
541 AV *stringref; 664 AV *stringref;
665 SV *decode_tagged;
542} dec_t; 666} dec_t;
543 667
544#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
545 669
546#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")
547 671
548#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)
549#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
550 674
551static UV 675static UV
552decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
553{ 677{
554 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
555 { 679 ++dec->cur;
556 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
557 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
558 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
559 return *dec->cur++ & 31;
560 680
561 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
562 WANT (2); 685 WANT (1);
563 dec->cur += 2; 686 dec->cur += 1;
564 return dec->cur[-1]; 687 return dec->cur[-1];
565 688 }
566 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
567 WANT (3); 691 WANT (2);
568 dec->cur += 3; 692 dec->cur += 2;
569 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
570 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
571 695 }
572 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
573 WANT (5); 698 WANT (4);
574 dec->cur += 5; 699 dec->cur += 4;
575 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
576 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
577 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
578 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
579 704 }
580 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
581 WANT (9); 707 WANT (8);
582 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
583 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
584 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
585 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
586 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
587 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
588 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
589 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
590 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
591 723 }
592 default: 724 else
593 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
594 }
595 726
596fail: 727fail:
597 return 0; 728 return 0;
598} 729}
599 730
604{ 735{
605 AV *av = newAV (); 736 AV *av = newAV ();
606 737
607 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
608 739
609 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
610 { 741 {
611 ++dec->cur; 742 ++dec->cur;
612 743
613 for (;;) 744 for (;;)
614 { 745 {
615 WANT (1); 746 WANT (1);
616 747
617 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
618 { 749 {
619 ++dec->cur; 750 ++dec->cur;
620 break; 751 break;
621 } 752 }
622 753
623 av_push (av, decode_sv (dec)); 754 av_push (av, decode_sv (dec));
624 } 755 }
625 } 756 }
626 else 757 else
627 { 758 {
628 int i, len = decode_uint (dec); 759 UV i, len = decode_uint (dec);
629 760
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
630 av_fill (av, len - 1); 762 av_fill (av, len - 1);
631 763
632 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
633 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
634 } 766 }
644 776
645static void 777static void
646decode_he (dec_t *dec, HV *hv) 778decode_he (dec_t *dec, HV *hv)
647{ 779{
648 // for speed reasons, we specialcase single-string 780 // for speed reasons, we specialcase single-string
649 // byte or utf-8 strings as keys. 781 // byte or utf-8 strings as keys, but only when !stringref
650 782
651 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))
652 { 785 {
653 I32 len = decode_uint (dec); 786 STRLEN len = decode_uint (dec);
654 char *key = (char *)dec->cur; 787 char *key = (char *)dec->cur;
655 788
789 WANT (len);
656 dec->cur += len; 790 dec->cur += len;
657 791
658 if (ecb_expect_false (dec->stringref))
659 av_push (dec->stringref, newSVpvn (key, len));
660
661 hv_store (hv, key, len, decode_sv (dec), 0); 792 hv_store (hv, key, len, decode_sv (dec), 0);
793
794 return;
662 } 795 }
663 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
664 { 797 {
665 I32 len = decode_uint (dec); 798 STRLEN len = decode_uint (dec);
666 char *key = (char *)dec->cur; 799 char *key = (char *)dec->cur;
667 800
801 WANT (len);
668 dec->cur += len; 802 dec->cur += len;
669 803
670 if (ecb_expect_false (dec->stringref)) 804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
671 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 805 if (!is_utf8_string (key, len))
806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
672 807
673 hv_store (hv, key, -len, decode_sv (dec), 0); 808 hv_store (hv, key, -len, decode_sv (dec), 0);
809
810 return;
674 } 811 }
675 else 812
676 {
677 SV *k = decode_sv (dec); 813 SV *k = decode_sv (dec);
678 SV *v = decode_sv (dec); 814 SV *v = decode_sv (dec);
679 815
680 hv_store_ent (hv, k, v, 0); 816 hv_store_ent (hv, k, v, 0);
681 SvREFCNT_dec (k); 817 SvREFCNT_dec (k);
682 } 818
819fail:
820 ;
683} 821}
684 822
685static SV * 823static SV *
686decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
687{ 825{
688 HV *hv = newHV (); 826 HV *hv = newHV ();
689 827
690 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
691 829
692 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
693 { 831 {
694 ++dec->cur; 832 ++dec->cur;
695 833
696 for (;;) 834 for (;;)
697 { 835 {
698 WANT (1); 836 WANT (1);
699 837
700 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
701 { 839 {
702 ++dec->cur; 840 ++dec->cur;
703 break; 841 break;
704 } 842 }
705 843
706 decode_he (dec, hv); 844 decode_he (dec, hv);
707 } 845 }
708 } 846 }
709 else 847 else
710 { 848 {
711 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
712 852
713 while (pairs--) 853 while (pairs--)
714 decode_he (dec, hv); 854 decode_he (dec, hv);
715 } 855 }
716 856
726static SV * 866static SV *
727decode_str (dec_t *dec, int utf8) 867decode_str (dec_t *dec, int utf8)
728{ 868{
729 SV *sv = 0; 869 SV *sv = 0;
730 870
731 if ((*dec->cur & 31) == 31) 871 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
732 { 872 {
873 // indefinite length strings
733 ++dec->cur; 874 ++dec->cur;
734 875
876 U8 major = *dec->cur & MAJOR_MISC;
877
735 sv = newSVpvn ("", 0); 878 sv = newSVpvn ("", 0);
736 879
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 880 for (;;)
739 { 881 {
740 WANT (1); 882 WANT (1);
741 883
742 if (*dec->cur == (0xe0 | 31)) 884 if ((*dec->cur - major) > LENGTH_EXT8)
885 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 886 {
744 ++dec->cur; 887 ++dec->cur;
745 break; 888 break;
746 } 889 }
890 else
891 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 892
748 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;
749 } 898 }
750 } 899 }
751 else 900 else
752 { 901 {
753 STRLEN len = decode_uint (dec); 902 STRLEN len = decode_uint (dec);
754 903
755 WANT (len); 904 WANT (len);
756 sv = newSVpvn (dec->cur, len); 905 sv = newSVpvn (dec->cur, len);
757 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));
758 } 911 }
759 912
760 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
761 SvUTF8_on (sv); 919 SvUTF8_on (sv);
762 920 }
763 if (ecb_expect_false (dec->stringref))
764 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765 921
766 return sv; 922 return sv;
767 923
768fail: 924fail:
769 SvREFCNT_dec (sv); 925 SvREFCNT_dec (sv);
788 sv = newRV_noinc (decode_sv (dec)); 944 sv = newRV_noinc (decode_sv (dec));
789 break; 945 break;
790 946
791 case CBOR_TAG_STRINGREF_NAMESPACE: 947 case CBOR_TAG_STRINGREF_NAMESPACE:
792 { 948 {
949 // do nmot use SAVETMPS/FREETMPS, as these will
950 // erase mortalised caches, e.g. "shareable"
793 ENTER; SAVETMPS; 951 ENTER;
794 952
795 SAVESPTR (dec->stringref); 953 SAVESPTR (dec->stringref);
796 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 954 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797 955
798 sv = decode_sv (dec); 956 sv = decode_sv (dec);
799 957
800 FREETMPS; LEAVE; 958 LEAVE;
801 } 959 }
802 break; 960 break;
803 961
804 case CBOR_TAG_STRINGREF: 962 case CBOR_TAG_STRINGREF:
805 { 963 {
806 if ((*dec->cur >> 5) != 0) 964 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
807 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 965 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808 966
809 UV idx = decode_uint (dec); 967 UV idx = decode_uint (dec);
810 968
811 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 969 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
818 case CBOR_TAG_VALUE_SHAREABLE: 976 case CBOR_TAG_VALUE_SHAREABLE:
819 { 977 {
820 if (ecb_expect_false (!dec->shareable)) 978 if (ecb_expect_false (!dec->shareable))
821 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 979 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822 980
981 if (dec->cbor.flags & F_ALLOW_CYCLES)
982 {
823 sv = newSV (0); 983 sv = newSV (0);
824 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 984 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825 985
826 SV *osv = decode_sv (dec); 986 SV *osv = decode_sv (dec);
827 sv_setsv (sv, osv); 987 sv_setsv (sv, osv);
828 SvREFCNT_dec_NN (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 }
829 } 997 }
830 break; 998 break;
831 999
832 case CBOR_TAG_VALUE_SHAREDREF: 1000 case CBOR_TAG_VALUE_SHAREDREF:
833 { 1001 {
834 if ((*dec->cur >> 5) != 0) 1002 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
835 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1003 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836 1004
837 UV idx = decode_uint (dec); 1005 UV idx = decode_uint (dec);
838 1006
839 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1007 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
840 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1008 ERR ("corrupted CBOR data (sharedref index out of bounds)");
841 1009
842 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[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");
843 } 1014 }
844 break; 1015 break;
845 1016
846 case CBOR_TAG_PERL_OBJECT: 1017 case CBOR_TAG_PERL_OBJECT:
847 { 1018 {
862 if (!method) 1033 if (!method)
863 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)");
864 1035
865 dSP; 1036 dSP;
866 1037
867 ENTER; SAVETMPS; PUSHMARK (SP); 1038 ENTER; SAVETMPS;
1039 PUSHMARK (SP);
868 EXTEND (SP, len + 1); 1040 EXTEND (SP, len + 1);
869 // 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
870 PUSHs (*av_fetch (av, 0, 1)); 1042 PUSHs (*av_fetch (av, 0, 1));
871 PUSHs (sv_cbor); 1043 PUSHs (sv_cbor);
872 1044
894 } 1066 }
895 break; 1067 break;
896 1068
897 default: 1069 default:
898 { 1070 {
1071 SV *tag_sv = newSVuv (tag);
1072
899 sv = decode_sv (dec); 1073 sv = decode_sv (dec);
900 1074
1075 dSP;
1076 ENTER; SAVETMPS;
1077 SAVESTACK_POS ();
1078 PUSHMARK (SP);
1079 EXTEND (SP, 2);
1080 PUSHs (tag_sv);
1081 PUSHs (sv);
1082
1083 PUTBACK;
1084 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1085 SPAGAIN;
1086
1087 if (SvTRUE (ERRSV))
1088 {
1089 SvREFCNT_dec (tag_sv);
1090 FREETMPS; LEAVE;
1091 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1092 }
1093
1094 if (count)
1095 {
1096 SvREFCNT_dec (tag_sv);
1097 SvREFCNT_dec (sv);
1098 sv = SvREFCNT_inc (POPs);
1099 }
1100 else
1101 {
901 AV *av = newAV (); 1102 AV *av = newAV ();
902 av_push (av, newSVuv (tag)); 1103 av_push (av, tag_sv);
903 av_push (av, sv); 1104 av_push (av, sv);
904 1105
905 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1106 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
906 ? cbor_tagged_stash 1107 ? cbor_tagged_stash
907 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1108 : gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1109 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1110 }
1111
1112 PUTBACK;
1113
1114 FREETMPS; LEAVE;
910 } 1115 }
911 break; 1116 break;
912 } 1117 }
913 1118
914 return sv; 1119 return sv;
921static SV * 1126static SV *
922decode_sv (dec_t *dec) 1127decode_sv (dec_t *dec)
923{ 1128{
924 WANT (1); 1129 WANT (1);
925 1130
926 switch (*dec->cur >> 5) 1131 switch (*dec->cur >> MAJOR_SHIFT)
927 { 1132 {
928 case 0: // unsigned int 1133 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
929 return newSVuv (decode_uint (dec)); 1134 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
930 case 1: // negative int 1135 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
931 return newSViv (-1 - (IV)decode_uint (dec)); 1136 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
932 case 2: // octet string 1137 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
933 return decode_str (dec, 0); 1138 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
934 case 3: // utf-8 string 1139 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
935 return decode_str (dec, 1); 1140
936 case 4: // array 1141 case MAJOR_MISC >> MAJOR_SHIFT:
937 return decode_av (dec);
938 case 5: // map
939 return decode_hv (dec);
940 case 6: // tag
941 return decode_tagged (dec);
942 case 7: // misc
943 switch (*dec->cur++ & 31) 1142 switch (*dec->cur++ & MINOR_MASK)
944 { 1143 {
945 case 20: 1144 case SIMPLE_FALSE:
946#if CBOR_SLOW 1145#if CBOR_SLOW
947 types_false = get_bool ("Types::Serialiser::false"); 1146 types_false = get_bool ("Types::Serialiser::false");
948#endif 1147#endif
949 return newSVsv (types_false); 1148 return newSVsv (types_false);
950 case 21: 1149 case SIMPLE_TRUE:
951#if CBOR_SLOW 1150#if CBOR_SLOW
952 types_true = get_bool ("Types::Serialiser::true"); 1151 types_true = get_bool ("Types::Serialiser::true");
953#endif 1152#endif
954 return newSVsv (types_true); 1153 return newSVsv (types_true);
955 case 22: 1154 case SIMPLE_NULL:
956 return newSVsv (&PL_sv_undef); 1155 return newSVsv (&PL_sv_undef);
957 case 23: 1156 case SIMPLE_UNDEF:
958#if CBOR_SLOW 1157#if CBOR_SLOW
959 types_error = get_bool ("Types::Serialiser::error"); 1158 types_error = get_bool ("Types::Serialiser::error");
960#endif 1159#endif
961 return newSVsv (types_error); 1160 return newSVsv (types_error);
962 1161
963 case 25: 1162 case MISC_FLOAT16:
964 { 1163 {
965 WANT (2); 1164 WANT (2);
966 1165
967 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1166 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
968 dec->cur += 2; 1167 dec->cur += 2;
969 1168
970 return newSVnv (ecb_binary16_to_float (fp)); 1169 return newSVnv (ecb_binary16_to_float (fp));
971 } 1170 }
972 1171
973 case 26: 1172 case MISC_FLOAT32:
974 { 1173 {
975 uint32_t fp; 1174 uint32_t fp;
976 WANT (4); 1175 WANT (4);
977 memcpy (&fp, dec->cur, 4); 1176 memcpy (&fp, dec->cur, 4);
978 dec->cur += 4; 1177 dec->cur += 4;
981 fp = ecb_bswap32 (fp); 1180 fp = ecb_bswap32 (fp);
982 1181
983 return newSVnv (ecb_binary32_to_float (fp)); 1182 return newSVnv (ecb_binary32_to_float (fp));
984 } 1183 }
985 1184
986 case 27: 1185 case MISC_FLOAT64:
987 { 1186 {
988 uint64_t fp; 1187 uint64_t fp;
989 WANT (8); 1188 WANT (8);
990 memcpy (&fp, dec->cur, 8); 1189 memcpy (&fp, dec->cur, 8);
991 dec->cur += 8; 1190 dec->cur += 8;
994 fp = ecb_bswap64 (fp); 1193 fp = ecb_bswap64 (fp);
995 1194
996 return newSVnv (ecb_binary64_to_double (fp)); 1195 return newSVnv (ecb_binary64_to_double (fp));
997 } 1196 }
998 1197
999 // 0..19 unassigned 1198 // 0..19 unassigned simple
1000 // 24 reserved + unassigned (reserved values are not encodable) 1199 // 24 reserved + unassigned simple (reserved values are not encodable)
1200 // 28-30 unassigned misc
1201 // 31 break code
1001 default: 1202 default:
1002 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1203 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1003 } 1204 }
1004 1205
1005 break; 1206 break;
1006 } 1207 }
1007 1208
1010} 1211}
1011 1212
1012static SV * 1213static SV *
1013decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1214decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1014{ 1215{
1015 dec_t dec = { }; 1216 dec_t dec = { 0 };
1016 SV *sv; 1217 SV *sv;
1017 STRLEN len; 1218 STRLEN len;
1018 char *data = SvPVbyte (string, len); 1219 char *data = SvPVbyte (string, len);
1019 1220
1020 if (len > cbor->max_size && cbor->max_size) 1221 if (len > cbor->max_size && cbor->max_size)
1034 if (dec.cur != dec.end && !dec.err) 1235 if (dec.cur != dec.end && !dec.err)
1035 dec.err = "garbage after CBOR object"; 1236 dec.err = "garbage after CBOR object";
1036 1237
1037 if (dec.err) 1238 if (dec.err)
1038 { 1239 {
1240 if (dec.shareable)
1241 {
1242 // need to break cyclic links, which would all be in shareable
1243 int i;
1244 SV **svp;
1245
1246 for (i = av_len (dec.shareable) + 1; i--; )
1247 if ((svp = av_fetch (dec.shareable, i, 0)))
1248 sv_setsv (*svp, &PL_sv_undef);
1249 }
1250
1039 SvREFCNT_dec (sv); 1251 SvREFCNT_dec (sv);
1040 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1252 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1041 } 1253 }
1042 1254
1043 sv = sv_2mortal (sv); 1255 sv = sv_2mortal (sv);
1044 1256
1045 return sv; 1257 return sv;
1046} 1258}
1047 1259
1260/////////////////////////////////////////////////////////////////////////////
1261// incremental parser
1262
1263#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1264
1265// returns 0 for notyet, 1 for success or error
1266static int
1267incr_parse (CBOR *self, SV *cborstr)
1268{
1269 STRLEN cur;
1270 SvPV (cborstr, cur);
1271
1272 while (ecb_expect_true (self->incr_need <= cur))
1273 {
1274 // table of integer count bytes
1275 static I8 incr_len[MINOR_MASK + 1] = {
1276 0, 0, 0, 0, 0, 0, 0, 0,
1277 0, 0, 0, 0, 0, 0, 0, 0,
1278 0, 0, 0, 0, 0, 0, 0, 0,
1279 1, 2, 4, 8,-1,-1,-1,-2
1280 };
1281
1282 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1283 U8 m = *p & MINOR_MASK;
1284 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1285 I8 ilen = incr_len[m];
1286
1287 self->incr_need = self->incr_pos + 1;
1288
1289 if (ecb_expect_false (ilen < 0))
1290 {
1291 if (m != MINOR_INDEF)
1292 return 1; // error
1293
1294 if (*p == (MAJOR_MISC | MINOR_INDEF))
1295 {
1296 if (count >= 0)
1297 return 1; // error
1298
1299 count = 1;
1300 }
1301 else
1302 {
1303 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1304 count = -1;
1305 }
1306 }
1307 else
1308 {
1309 self->incr_need += ilen;
1310 if (ecb_expect_false (self->incr_need > cur))
1311 return 0;
1312
1313 int major = *p >> MAJOR_SHIFT;
1314
1315 switch (major)
1316 {
1317 case MAJOR_TAG >> MAJOR_SHIFT:
1318 ++count; // tags merely prefix another value
1319 break;
1320
1321 case MAJOR_BYTES >> MAJOR_SHIFT:
1322 case MAJOR_TEXT >> MAJOR_SHIFT:
1323 case MAJOR_ARRAY >> MAJOR_SHIFT:
1324 case MAJOR_MAP >> MAJOR_SHIFT:
1325 {
1326 UV len;
1327
1328 if (ecb_expect_false (ilen))
1329 {
1330 len = 0;
1331
1332 do {
1333 len = (len << 8) | *++p;
1334 } while (--ilen);
1335 }
1336 else
1337 len = m;
1338
1339 switch (major)
1340 {
1341 case MAJOR_BYTES >> MAJOR_SHIFT:
1342 case MAJOR_TEXT >> MAJOR_SHIFT:
1343 self->incr_need += len;
1344 if (ecb_expect_false (self->incr_need > cur))
1345 return 0;
1346
1347 break;
1348
1349 case MAJOR_MAP >> MAJOR_SHIFT:
1350 len <<= 1;
1351 case MAJOR_ARRAY >> MAJOR_SHIFT:
1352 if (len)
1353 {
1354 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1355 count = len + 1;
1356 }
1357 break;
1358 }
1359 }
1360 }
1361 }
1362
1363 self->incr_pos = self->incr_need;
1364
1365 if (count > 0)
1366 {
1367 while (!--count)
1368 {
1369 if (!AvFILLp (self->incr_count))
1370 return 1; // done
1371
1372 SvREFCNT_dec_NN (av_pop (self->incr_count));
1373 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1374 }
1375
1376 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1377 }
1378 }
1379
1380 return 0;
1381}
1382
1383
1048///////////////////////////////////////////////////////////////////////////// 1384/////////////////////////////////////////////////////////////////////////////
1049// XS interface functions 1385// XS interface functions
1050 1386
1051MODULE = CBOR::XS PACKAGE = CBOR::XS 1387MODULE = CBOR::XS PACKAGE = CBOR::XS
1052 1388
1060 1396
1061 types_true = get_bool ("Types::Serialiser::true" ); 1397 types_true = get_bool ("Types::Serialiser::true" );
1062 types_false = get_bool ("Types::Serialiser::false"); 1398 types_false = get_bool ("Types::Serialiser::false");
1063 types_error = get_bool ("Types::Serialiser::error"); 1399 types_error = get_bool ("Types::Serialiser::error");
1064 1400
1401 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1402
1065 sv_cbor = newSVpv ("CBOR", 0); 1403 sv_cbor = newSVpv ("CBOR", 0);
1066 SvREADONLY_on (sv_cbor); 1404 SvREADONLY_on (sv_cbor);
1405
1406 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1067} 1407}
1068 1408
1069PROTOTYPES: DISABLE 1409PROTOTYPES: DISABLE
1070 1410
1071void CLONE (...) 1411void CLONE (...)
1090void shrink (CBOR *self, int enable = 1) 1430void shrink (CBOR *self, int enable = 1)
1091 ALIAS: 1431 ALIAS:
1092 shrink = F_SHRINK 1432 shrink = F_SHRINK
1093 allow_unknown = F_ALLOW_UNKNOWN 1433 allow_unknown = F_ALLOW_UNKNOWN
1094 allow_sharing = F_ALLOW_SHARING 1434 allow_sharing = F_ALLOW_SHARING
1095 dedup_keys = F_DEDUP_KEYS 1435 allow_cycles = F_ALLOW_CYCLES
1096 dedup_strings = F_DEDUP_STRINGS 1436 pack_strings = F_PACK_STRINGS
1437 text_keys = F_TEXT_KEYS
1438 text_strings = F_TEXT_STRINGS
1439 validate_utf8 = F_VALIDATE_UTF8
1097 PPCODE: 1440 PPCODE:
1098{ 1441{
1099 if (enable) 1442 if (enable)
1100 self->flags |= ix; 1443 self->flags |= ix;
1101 else 1444 else
1107void get_shrink (CBOR *self) 1450void get_shrink (CBOR *self)
1108 ALIAS: 1451 ALIAS:
1109 get_shrink = F_SHRINK 1452 get_shrink = F_SHRINK
1110 get_allow_unknown = F_ALLOW_UNKNOWN 1453 get_allow_unknown = F_ALLOW_UNKNOWN
1111 get_allow_sharing = F_ALLOW_SHARING 1454 get_allow_sharing = F_ALLOW_SHARING
1112 get_dedup_keys = F_DEDUP_KEYS 1455 get_allow_cycles = F_ALLOW_CYCLES
1113 get_dedup_strings = F_DEDUP_STRINGS 1456 get_pack_strings = F_PACK_STRINGS
1457 get_text_keys = F_TEXT_KEYS
1458 get_text_strings = F_TEXT_STRINGS
1459 get_validate_utf8 = F_VALIDATE_UTF8
1114 PPCODE: 1460 PPCODE:
1115 XPUSHs (boolSV (self->flags & ix)); 1461 XPUSHs (boolSV (self->flags & ix));
1116 1462
1117void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1463void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1118 PPCODE: 1464 PPCODE:
1131 XPUSHs (ST (0)); 1477 XPUSHs (ST (0));
1132 1478
1133int get_max_size (CBOR *self) 1479int get_max_size (CBOR *self)
1134 CODE: 1480 CODE:
1135 RETVAL = self->max_size; 1481 RETVAL = self->max_size;
1482 OUTPUT:
1483 RETVAL
1484
1485void filter (CBOR *self, SV *filter = 0)
1486 PPCODE:
1487 SvREFCNT_dec (self->filter);
1488 self->filter = filter ? newSVsv (filter) : filter;
1489 XPUSHs (ST (0));
1490
1491SV *get_filter (CBOR *self)
1492 CODE:
1493 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1136 OUTPUT: 1494 OUTPUT:
1137 RETVAL 1495 RETVAL
1138 1496
1139void encode (CBOR *self, SV *scalar) 1497void encode (CBOR *self, SV *scalar)
1140 PPCODE: 1498 PPCODE:
1155 EXTEND (SP, 2); 1513 EXTEND (SP, 2);
1156 PUSHs (sv); 1514 PUSHs (sv);
1157 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1515 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1158} 1516}
1159 1517
1518void incr_parse (CBOR *self, SV *cborstr)
1519 ALIAS:
1520 incr_parse_multiple = 1
1521 PPCODE:
1522{
1523 if (SvUTF8 (cborstr))
1524 sv_utf8_downgrade (cborstr, 0);
1525
1526 if (!self->incr_count)
1527 {
1528 self->incr_count = newAV ();
1529 self->incr_pos = 0;
1530 self->incr_need = 1;
1531
1532 av_push (self->incr_count, newSViv (1));
1533 }
1534
1535 do
1536 {
1537 if (!incr_parse (self, cborstr))
1538 {
1539 if (self->incr_need > self->max_size && self->max_size)
1540 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1541 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1542
1543 break;
1544 }
1545
1546 SV *sv;
1547 char *offset;
1548
1549 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1550 XPUSHs (sv);
1551
1552 sv_chop (cborstr, offset);
1553
1554 av_clear (self->incr_count);
1555 av_push (self->incr_count, newSViv (1));
1556
1557 self->incr_pos = 0;
1558 self->incr_need = self->incr_pos + 1;
1559 }
1560 while (ix);
1561}
1562
1563void incr_reset (CBOR *self)
1564 CODE:
1565{
1566 SvREFCNT_dec (self->incr_count);
1567 self->incr_count = 0;
1568}
1569
1570void DESTROY (CBOR *self)
1571 PPCODE:
1572 cbor_free (self);
1573
1160PROTOTYPES: ENABLE 1574PROTOTYPES: ENABLE
1161 1575
1162void encode_cbor (SV *scalar) 1576void encode_cbor (SV *scalar)
1577 ALIAS:
1578 encode_cbor = 0
1579 encode_cbor_sharing = F_ALLOW_SHARING
1163 PPCODE: 1580 PPCODE:
1164{ 1581{
1165 CBOR cbor; 1582 CBOR cbor;
1166 cbor_init (&cbor); 1583 cbor_init (&cbor);
1584 cbor.flags |= ix;
1167 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1585 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1168 XPUSHs (scalar); 1586 XPUSHs (scalar);
1169} 1587}
1170 1588
1171void decode_cbor (SV *cborstr) 1589void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines