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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.19 by root, Wed Nov 20 02:03:09 2013 UTC vs.
Revision 1.49 by root, Mon Feb 8 04:11:11 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_VALIDATE_UTF8 0x00000020UL
64 106
65#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
66 108
67#define SB do { 109#define SB do {
68#define SE } while (0) 110#define SE } while (0)
80# define CBOR_SLOW 0 122# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
82#endif 124#endif
83 125
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 128
87typedef struct { 129typedef struct {
88 U32 flags; 130 U32 flags;
89 U32 max_depth; 131 U32 max_depth;
90 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
134
135 // for the incremental parser
136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 139} CBOR;
92 140
93ecb_inline void 141ecb_inline void
94cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
95{ 143{
96 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 145 cbor->max_depth = 512;
146}
147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
98} 153}
99 154
100///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
101// utility functions 156// utility functions
102 157
124 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 180#endif
126 } 181 }
127} 182}
128 183
184// minimum length of a string to be registered for stringref
185ecb_inline int
186minimum_string_length (UV idx)
187{
188 return idx > 23
189 ? idx > 0xffU
190 ? idx > 0xffffU
191 ? idx > 0xffffffffU
192 ? 11
193 : 7
194 : 5
195 : 4
196 : 3;
197}
198
129///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
130// encoder 200// encoder
131 201
132// structure used for encoding CBOR 202// structure used for encoding CBOR
133typedef struct 203typedef struct
135 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
137 SV *sv; // result scalar 207 SV *sv; // result scalar
138 CBOR cbor; 208 CBOR cbor;
139 U32 depth; // recursion level 209 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 210 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
211 UV stringref_idx;
141 HV *shareable; // ptr => index, or 0 212 HV *shareable; // ptr => index, or 0
142 UV shareable_idx; 213 UV shareable_idx;
143} enc_t; 214} enc_t;
144 215
145ecb_inline void 216ecb_inline void
164static void 235static void
165encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
166{ 237{
167 need (enc, 9); 238 need (enc, 9);
168 239
169 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
171 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
172 { 243 {
173 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 245 *enc->cur++ = len;
175 } 246 }
176 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
177 { 248 {
178 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 251 *enc->cur++ = len;
181 } 252 }
182 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
183 { 254 {
184 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 259 *enc->cur++ = len;
189 } 260 }
190 else 261 else
191 { 262 {
192 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 271 *enc->cur++ = len;
201 } 272 }
202} 273}
203 274
204static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 283{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 285 need (enc, len);
209 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
210 enc->cur += len; 287 enc->cur += len;
211} 288}
212 289
213ecb_inline void 290static void
214encode_tag (enc_t *enc, UV tag) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
215{ 292{
216 encode_uint (enc, 0xc0, tag); 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296
297 if (SvOK (*svp))
298 {
299 // already registered, use stringref
300 encode_tag (enc, CBOR_TAG_STRINGREF);
301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
302 return;
303 }
304 else if (len >= minimum_string_length (enc->stringref_idx))
305 {
306 // register only
307 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx;
309 }
310 }
311
312 encode_str (enc, utf8, str, len);
217} 313}
218 314
219static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
220 316
221static void 317static void
226 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
227 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
228 324
229 ++enc->depth; 325 ++enc->depth;
230 326
231 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
232 328
329 if (SvMAGICAL (av))
233 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
234 { 331 {
235 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
236 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
237 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
238 341
239 --enc->depth; 342 --enc->depth;
240} 343}
241 344
242static void 345static void
251 354
252 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
253 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
254 357
255 if (mg) 358 if (mg)
256 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
257 else 360 else
258 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
259 362
260 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
261 { 364 {
262 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
263 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
264 else 367 else
265 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
266 369
267 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
268 } 371 }
269 372
270 if (mg) 373 if (mg)
271 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
272 375
273 --enc->depth; 376 --enc->depth;
274} 377}
275 378
276// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
277static void 380static void
278encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
279{ 382{
280 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
281
282 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
283 && ecb_expect_false (SvREFCNT (sv) > 1))
284 {
285 if (!enc->shareable)
286 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
287
288 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
289
290 if (SvOK (*svp))
291 {
292 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
293 encode_uint (enc, 0x00, SvUV (*svp));
294 return;
295 }
296 else
297 {
298 sv_setuv (*svp, enc->shareable_idx);
299 ++enc->shareable_idx;
300 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
301 }
302 }
303 384
304 svtype svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
305 386
306 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
307 { 388 {
314 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
315 ? cbor_tagged_stash 396 ? cbor_tagged_stash
316 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
317 398
318 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
319 GV *method;
320 400
321 if (stash == boolean_stash) 401 if (stash == boolean_stash)
322 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
323 else if (stash == error_stash) 406 else if (stash == error_stash)
324 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
325 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
326 { 412 {
327 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
328 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
329 415
330 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
331 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
332 } 432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
436 }
437 else
438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
333 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
334 { 451 {
335 dSP; 452 dSP;
336 453
337 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS; PUSHMARK (SP);
338 // we re-bless the reference to get overload and other niceties right 455 // we re-bless the reference to get overload and other niceties right
370 // catch this surprisingly common error 487 // catch this surprisingly common error
371 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
372 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
373 490
374 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
375 encode_uint (enc, 0x80, count + 1); 492 encode_uint (enc, MAJOR_ARRAY, count + 1);
376 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
377 494
378 while (count) 495 while (count)
379 encode_sv (enc, SP[1 - count--]); 496 encode_sv (enc, SP[1 - count--]);
380 497
381 PUTBACK; 498 PUTBACK;
402{ 519{
403 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
404 521
405 need (enc, 9); 522 need (enc, 9);
406 523
407 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
408 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
409 //TODO: maybe I32? 526 //TODO: maybe I32?
410 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
411 { 528 {
412 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
413 530
414 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
415 532
416 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
417 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
418 535
419 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
421 } 538 }
422 else 539 else
423 { 540 {
424 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
425 542
426 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
427 544
428 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
429 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
430 547
431 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
440 557
441 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
442 { 559 {
443 STRLEN len; 560 STRLEN len;
444 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
445 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
446 } 563 }
447 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
448 encode_nv (enc, sv); 565 encode_nv (enc, sv);
449 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
450 { 567 {
451 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
452 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
453 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
454 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
455 else 572 else
456 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
457 } 574 }
458 else if (SvROK (sv)) 575 else if (SvROK (sv))
459 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
460 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
461 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
462 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
463 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
464 else 581 else
465 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 582 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
466 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
467} 584}
468 585
469static SV * 586static SV *
470encode_cbor (SV *scalar, CBOR *cbor) 587encode_cbor (SV *scalar, CBOR *cbor)
471{ 588{
472 enc_t enc = { }; 589 enc_t enc = { 0 };
473 590
474 enc.cbor = *cbor; 591 enc.cbor = *cbor;
475 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
476 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
477 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
478 595
479 SvPOK_only (enc.sv); 596 SvPOK_only (enc.sv);
597
598 if (cbor->flags & F_PACK_STRINGS)
599 {
600 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
601 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
602 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
603 }
604
480 encode_sv (&enc, scalar); 605 encode_sv (&enc, scalar);
481 606
482 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 607 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
483 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 608 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
484 609
499 const char *err; // parse error, if != 0 624 const char *err; // parse error, if != 0
500 CBOR cbor; 625 CBOR cbor;
501 U32 depth; // recursion depth 626 U32 depth; // recursion depth
502 U32 maxdepth; // recursion depth limit 627 U32 maxdepth; // recursion depth limit
503 AV *shareable; 628 AV *shareable;
629 AV *stringref;
630 SV *decode_tagged;
504} dec_t; 631} dec_t;
505 632
506#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 633#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
507 634
508#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 635#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
511#define DEC_DEC_DEPTH --dec->depth 638#define DEC_DEC_DEPTH --dec->depth
512 639
513static UV 640static UV
514decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
515{ 642{
516 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
517 { 644 ++dec->cur;
518 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
519 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
520 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
521 return *dec->cur++ & 31;
522 645
523 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
524 WANT (2); 650 WANT (1);
525 dec->cur += 2; 651 dec->cur += 1;
526 return dec->cur[-1]; 652 return dec->cur[-1];
527 653 }
528 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
529 WANT (3); 656 WANT (2);
530 dec->cur += 3; 657 dec->cur += 2;
531 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
532 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
533 660 }
534 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
535 WANT (5); 663 WANT (4);
536 dec->cur += 5; 664 dec->cur += 4;
537 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
538 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
539 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
540 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
541 669 }
542 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
543 WANT (9); 672 WANT (8);
544 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
545 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
546 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
547 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
548 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
549 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
550 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
551 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
552 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
553 688 }
554 default: 689 else
555 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
556 }
557 691
558fail: 692fail:
559 return 0; 693 return 0;
560} 694}
561 695
566{ 700{
567 AV *av = newAV (); 701 AV *av = newAV ();
568 702
569 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
570 704
571 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
572 { 706 {
573 ++dec->cur; 707 ++dec->cur;
574 708
575 for (;;) 709 for (;;)
576 { 710 {
577 WANT (1); 711 WANT (1);
578 712
579 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
580 { 714 {
581 ++dec->cur; 715 ++dec->cur;
582 break; 716 break;
583 } 717 }
584 718
587 } 721 }
588 else 722 else
589 { 723 {
590 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
591 725
726 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
592 av_fill (av, len - 1); 727 av_fill (av, len - 1);
593 728
594 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
595 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
596 } 731 }
606 741
607static void 742static void
608decode_he (dec_t *dec, HV *hv) 743decode_he (dec_t *dec, HV *hv)
609{ 744{
610 // for speed reasons, we specialcase single-string 745 // for speed reasons, we specialcase single-string
611 // byte or utf-8 strings as keys. 746 // byte or utf-8 strings as keys, but only when !stringref
612 747
613 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 748 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
614 { 750 {
615 I32 len = decode_uint (dec); 751 I32 len = decode_uint (dec);
616 char *key = (char *)dec->cur; 752 char *key = (char *)dec->cur;
617 753
754 WANT (len);
618 dec->cur += len; 755 dec->cur += len;
619 756
620 hv_store (hv, key, len, decode_sv (dec), 0); 757 hv_store (hv, key, len, decode_sv (dec), 0);
758
759 return;
621 } 760 }
622 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 761 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
623 { 762 {
624 I32 len = decode_uint (dec); 763 I32 len = decode_uint (dec);
625 char *key = (char *)dec->cur; 764 char *key = (char *)dec->cur;
626 765
766 WANT (len);
627 dec->cur += len; 767 dec->cur += len;
628 768
769 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
770 if (!is_utf8_string (key, len))
771 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
772
629 hv_store (hv, key, -len, decode_sv (dec), 0); 773 hv_store (hv, key, -len, decode_sv (dec), 0);
774
775 return;
630 } 776 }
631 else 777
632 {
633 SV *k = decode_sv (dec); 778 SV *k = decode_sv (dec);
634 SV *v = decode_sv (dec); 779 SV *v = decode_sv (dec);
635 780
636 hv_store_ent (hv, k, v, 0); 781 hv_store_ent (hv, k, v, 0);
637 SvREFCNT_dec (k); 782 SvREFCNT_dec (k);
638 } 783
784fail:
785 ;
639} 786}
640 787
641static SV * 788static SV *
642decode_hv (dec_t *dec) 789decode_hv (dec_t *dec)
643{ 790{
644 HV *hv = newHV (); 791 HV *hv = newHV ();
645 792
646 DEC_INC_DEPTH; 793 DEC_INC_DEPTH;
647 794
648 if ((*dec->cur & 31) == 31) 795 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
649 { 796 {
650 ++dec->cur; 797 ++dec->cur;
651 798
652 for (;;) 799 for (;;)
653 { 800 {
654 WANT (1); 801 WANT (1);
655 802
656 if (*dec->cur == (0xe0 | 31)) 803 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
657 { 804 {
658 ++dec->cur; 805 ++dec->cur;
659 break; 806 break;
660 } 807 }
661 808
682static SV * 829static SV *
683decode_str (dec_t *dec, int utf8) 830decode_str (dec_t *dec, int utf8)
684{ 831{
685 SV *sv = 0; 832 SV *sv = 0;
686 833
687 if ((*dec->cur & 31) == 31) 834 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
688 { 835 {
836 // indefinite length strings
689 ++dec->cur; 837 ++dec->cur;
690 838
839 U8 major = *dec->cur & MAJOR_MISC;
840
691 sv = newSVpvn ("", 0); 841 sv = newSVpvn ("", 0);
692 842
693 // not very fast, and certainly not robust against illegal input
694 for (;;) 843 for (;;)
695 { 844 {
696 WANT (1); 845 WANT (1);
697 846
698 if (*dec->cur == (0xe0 | 31)) 847 if ((*dec->cur - major) > LENGTH_EXT8)
848 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
699 { 849 {
700 ++dec->cur; 850 ++dec->cur;
701 break; 851 break;
702 } 852 }
853 else
854 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
703 855
704 sv_catsv (sv, decode_sv (dec)); 856 STRLEN len = decode_uint (dec);
857
858 WANT (len);
859 sv_catpvn (sv, dec->cur, len);
860 dec->cur += len;
705 } 861 }
706 } 862 }
707 else 863 else
708 { 864 {
709 STRLEN len = decode_uint (dec); 865 STRLEN len = decode_uint (dec);
710 866
711 WANT (len); 867 WANT (len);
712 sv = newSVpvn (dec->cur, len); 868 sv = newSVpvn (dec->cur, len);
713 dec->cur += len; 869 dec->cur += len;
870
871 if (ecb_expect_false (dec->stringref)
872 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
873 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
714 } 874 }
715 875
716 if (utf8) 876 if (utf8)
877 {
878 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
879 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
880 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
881
717 SvUTF8_on (sv); 882 SvUTF8_on (sv);
883 }
718 884
719 return sv; 885 return sv;
720 886
721fail: 887fail:
722 SvREFCNT_dec (sv); 888 SvREFCNT_dec (sv);
732 WANT (1); 898 WANT (1);
733 899
734 switch (tag) 900 switch (tag)
735 { 901 {
736 case CBOR_TAG_MAGIC: 902 case CBOR_TAG_MAGIC:
737 return decode_sv (dec); 903 sv = decode_sv (dec);
904 break;
738 905
739 case CBOR_TAG_INDIRECTION: 906 case CBOR_TAG_INDIRECTION:
740 return newRV_noinc (decode_sv (dec)); 907 sv = newRV_noinc (decode_sv (dec));
908 break;
909
910 case CBOR_TAG_STRINGREF_NAMESPACE:
911 {
912 ENTER; SAVETMPS;
913
914 SAVESPTR (dec->stringref);
915 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
916
917 sv = decode_sv (dec);
918
919 FREETMPS; LEAVE;
920 }
921 break;
922
923 case CBOR_TAG_STRINGREF:
924 {
925 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
926 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
927
928 UV idx = decode_uint (dec);
929
930 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
931 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
932
933 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
934 }
935 break;
741 936
742 case CBOR_TAG_VALUE_SHAREABLE: 937 case CBOR_TAG_VALUE_SHAREABLE:
743 { 938 {
744 if (ecb_expect_false (!dec->shareable)) 939 if (ecb_expect_false (!dec->shareable))
745 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 940 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
746 941
942 if (dec->cbor.flags & F_ALLOW_CYCLES)
943 {
747 sv = newSV (0); 944 sv = newSV (0);
748 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 945 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
749 946
750 SV *osv = decode_sv (dec); 947 SV *osv = decode_sv (dec);
751 sv_setsv (sv, osv); 948 sv_setsv (sv, osv);
752 SvREFCNT_dec_NN (osv); 949 SvREFCNT_dec_NN (osv);
753 } 950 }
754 951 else
755 return sv; 952 {
953 av_push (dec->shareable, &PL_sv_undef);
954 int idx = AvFILLp (dec->shareable);
955 sv = decode_sv (dec);
956 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
957 }
958 }
959 break;
756 960
757 case CBOR_TAG_VALUE_SHAREDREF: 961 case CBOR_TAG_VALUE_SHAREDREF:
758 { 962 {
759 if ((*dec->cur >> 5) != 0) 963 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
760 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 964 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
761 965
762 UV idx = decode_uint (dec); 966 UV idx = decode_uint (dec);
763 967
764 if (!dec->shareable || idx > AvFILLp (dec->shareable)) 968 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
765 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 969 ERR ("corrupted CBOR data (sharedref index out of bounds)");
766 970
767 return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 971 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
972
973 if (sv == &PL_sv_undef)
974 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
768 } 975 }
976 break;
769 977
770 case CBOR_TAG_PERL_OBJECT: 978 case CBOR_TAG_PERL_OBJECT:
771 { 979 {
772 sv = decode_sv (dec); 980 sv = decode_sv (dec);
773 981
813 sv = SvREFCNT_inc (POPs); 1021 sv = SvREFCNT_inc (POPs);
814 1022
815 PUTBACK; 1023 PUTBACK;
816 1024
817 FREETMPS; LEAVE; 1025 FREETMPS; LEAVE;
818
819 return sv;
820 } 1026 }
1027 break;
821 1028
822 default: 1029 default:
823 { 1030 {
824 sv = decode_sv (dec); 1031 sv = decode_sv (dec);
825 1032
1033 dSP;
1034 ENTER; SAVETMPS; PUSHMARK (SP);
1035 EXTEND (SP, 2);
1036 PUSHs (newSVuv (tag));
1037 PUSHs (sv);
1038
1039 PUTBACK;
1040 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1041 SPAGAIN;
1042
1043 if (SvTRUE (ERRSV))
1044 {
1045 FREETMPS; LEAVE;
1046 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1047 }
1048
1049 if (count)
1050 {
1051 SvREFCNT_dec (sv);
1052 sv = SvREFCNT_inc (POPs);
1053 }
1054 else
1055 {
826 AV *av = newAV (); 1056 AV *av = newAV ();
827 av_push (av, newSVuv (tag)); 1057 av_push (av, newSVuv (tag));
828 av_push (av, sv); 1058 av_push (av, sv);
829 1059
830 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1060 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
831 ? cbor_tagged_stash 1061 ? cbor_tagged_stash
832 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1062 : gv_stashpv ("CBOR::XS::Tagged" , 1);
833
834 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1063 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
835 } 1064 }
1065
1066 PUTBACK;
1067
1068 FREETMPS; LEAVE;
1069 }
1070 break;
836 } 1071 }
1072
1073 return sv;
837 1074
838fail: 1075fail:
839 SvREFCNT_dec (sv); 1076 SvREFCNT_dec (sv);
840 return &PL_sv_undef; 1077 return &PL_sv_undef;
841} 1078}
843static SV * 1080static SV *
844decode_sv (dec_t *dec) 1081decode_sv (dec_t *dec)
845{ 1082{
846 WANT (1); 1083 WANT (1);
847 1084
848 switch (*dec->cur >> 5) 1085 switch (*dec->cur >> MAJOR_SHIFT)
849 { 1086 {
850 case 0: // unsigned int 1087 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
851 return newSVuv (decode_uint (dec)); 1088 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
852 case 1: // negative int 1089 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
853 return newSViv (-1 - (IV)decode_uint (dec)); 1090 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
854 case 2: // octet string 1091 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
855 return decode_str (dec, 0); 1092 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
856 case 3: // utf-8 string 1093 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
857 return decode_str (dec, 1); 1094
858 case 4: // array 1095 case MAJOR_MISC >> MAJOR_SHIFT:
859 return decode_av (dec);
860 case 5: // map
861 return decode_hv (dec);
862 case 6: // tag
863 return decode_tagged (dec);
864 case 7: // misc
865 switch (*dec->cur++ & 31) 1096 switch (*dec->cur++ & MINOR_MASK)
866 { 1097 {
867 case 20: 1098 case SIMPLE_FALSE:
868#if CBOR_SLOW 1099#if CBOR_SLOW
869 types_false = get_bool ("Types::Serialiser::false"); 1100 types_false = get_bool ("Types::Serialiser::false");
870#endif 1101#endif
871 return newSVsv (types_false); 1102 return newSVsv (types_false);
872 case 21: 1103 case SIMPLE_TRUE:
873#if CBOR_SLOW 1104#if CBOR_SLOW
874 types_true = get_bool ("Types::Serialiser::true"); 1105 types_true = get_bool ("Types::Serialiser::true");
875#endif 1106#endif
876 return newSVsv (types_true); 1107 return newSVsv (types_true);
877 case 22: 1108 case SIMPLE_NULL:
878 return newSVsv (&PL_sv_undef); 1109 return newSVsv (&PL_sv_undef);
879 case 23: 1110 case SIMPLE_UNDEF:
880#if CBOR_SLOW 1111#if CBOR_SLOW
881 types_error = get_bool ("Types::Serialiser::error"); 1112 types_error = get_bool ("Types::Serialiser::error");
882#endif 1113#endif
883 return newSVsv (types_error); 1114 return newSVsv (types_error);
884 1115
885 case 25: 1116 case MISC_FLOAT16:
886 { 1117 {
887 WANT (2); 1118 WANT (2);
888 1119
889 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1120 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
890 dec->cur += 2; 1121 dec->cur += 2;
891 1122
892 return newSVnv (ecb_binary16_to_float (fp)); 1123 return newSVnv (ecb_binary16_to_float (fp));
893 } 1124 }
894 1125
895 case 26: 1126 case MISC_FLOAT32:
896 { 1127 {
897 uint32_t fp; 1128 uint32_t fp;
898 WANT (4); 1129 WANT (4);
899 memcpy (&fp, dec->cur, 4); 1130 memcpy (&fp, dec->cur, 4);
900 dec->cur += 4; 1131 dec->cur += 4;
903 fp = ecb_bswap32 (fp); 1134 fp = ecb_bswap32 (fp);
904 1135
905 return newSVnv (ecb_binary32_to_float (fp)); 1136 return newSVnv (ecb_binary32_to_float (fp));
906 } 1137 }
907 1138
908 case 27: 1139 case MISC_FLOAT64:
909 { 1140 {
910 uint64_t fp; 1141 uint64_t fp;
911 WANT (8); 1142 WANT (8);
912 memcpy (&fp, dec->cur, 8); 1143 memcpy (&fp, dec->cur, 8);
913 dec->cur += 8; 1144 dec->cur += 8;
916 fp = ecb_bswap64 (fp); 1147 fp = ecb_bswap64 (fp);
917 1148
918 return newSVnv (ecb_binary64_to_double (fp)); 1149 return newSVnv (ecb_binary64_to_double (fp));
919 } 1150 }
920 1151
921 // 0..19 unassigned 1152 // 0..19 unassigned simple
922 // 24 reserved + unassigned (reserved values are not encodable) 1153 // 24 reserved + unassigned simple (reserved values are not encodable)
1154 // 28-30 unassigned misc
1155 // 31 break code
923 default: 1156 default:
924 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1157 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
925 } 1158 }
926 1159
927 break; 1160 break;
928 } 1161 }
929 1162
932} 1165}
933 1166
934static SV * 1167static SV *
935decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1168decode_cbor (SV *string, CBOR *cbor, char **offset_return)
936{ 1169{
937 dec_t dec = { }; 1170 dec_t dec = { 0 };
938 SV *sv; 1171 SV *sv;
939 STRLEN len; 1172 STRLEN len;
940 char *data = SvPVbyte (string, len); 1173 char *data = SvPVbyte (string, len);
941 1174
942 if (len > cbor->max_size && cbor->max_size) 1175 if (len > cbor->max_size && cbor->max_size)
956 if (dec.cur != dec.end && !dec.err) 1189 if (dec.cur != dec.end && !dec.err)
957 dec.err = "garbage after CBOR object"; 1190 dec.err = "garbage after CBOR object";
958 1191
959 if (dec.err) 1192 if (dec.err)
960 { 1193 {
1194 if (dec.shareable)
1195 {
1196 // need to break cyclic links, which whould all be in shareable
1197 int i;
1198 SV **svp;
1199
1200 for (i = av_len (dec.shareable) + 1; i--; )
1201 if ((svp = av_fetch (dec.shareable, i, 0)))
1202 sv_setsv (*svp, &PL_sv_undef);
1203 }
1204
961 SvREFCNT_dec (sv); 1205 SvREFCNT_dec (sv);
962 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1206 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
963 } 1207 }
964 1208
965 sv = sv_2mortal (sv); 1209 sv = sv_2mortal (sv);
966 1210
967 return sv; 1211 return sv;
968} 1212}
969 1213
1214/////////////////////////////////////////////////////////////////////////////
1215// incremental parser
1216
1217#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1218
1219// returns 0 for notyet, 1 for success or error
1220static int
1221incr_parse (CBOR *self, SV *cborstr)
1222{
1223 STRLEN cur;
1224 SvPV (cborstr, cur);
1225
1226 while (ecb_expect_true (self->incr_need <= cur))
1227 {
1228 // table of integer count bytes
1229 static I8 incr_len[MINOR_MASK + 1] = {
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 0, 0, 0, 0, 0, 0, 0, 0,
1232 0, 0, 0, 0, 0, 0, 0, 0,
1233 1, 2, 4, 8,-1,-1,-1,-2
1234 };
1235
1236 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1237 U8 m = *p & MINOR_MASK;
1238 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1239 I8 ilen = incr_len[m];
1240
1241 self->incr_need = self->incr_pos + 1;
1242
1243 if (ecb_expect_false (ilen < 0))
1244 {
1245 if (m != MINOR_INDEF)
1246 return 1; // error
1247
1248 if (*p == (MAJOR_MISC | MINOR_INDEF))
1249 {
1250 if (count >= 0)
1251 return 1; // error
1252
1253 count = 1;
1254 }
1255 else
1256 {
1257 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1258 count = -1;
1259 }
1260 }
1261 else
1262 {
1263 self->incr_need += ilen;
1264 if (ecb_expect_false (self->incr_need > cur))
1265 return 0;
1266
1267 int major = *p >> MAJOR_SHIFT;
1268
1269 switch (major)
1270 {
1271 case MAJOR_TAG >> MAJOR_SHIFT:
1272 ++count; // tags merely prefix another value
1273 break;
1274
1275 case MAJOR_BYTES >> MAJOR_SHIFT:
1276 case MAJOR_TEXT >> MAJOR_SHIFT:
1277 case MAJOR_ARRAY >> MAJOR_SHIFT:
1278 case MAJOR_MAP >> MAJOR_SHIFT:
1279 {
1280 UV len;
1281
1282 if (ecb_expect_false (ilen))
1283 {
1284 len = 0;
1285
1286 do {
1287 len = (len << 8) | *++p;
1288 } while (--ilen);
1289 }
1290 else
1291 len = m;
1292
1293 switch (major)
1294 {
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 self->incr_need += len;
1298 if (ecb_expect_false (self->incr_need > cur))
1299 return 0;
1300
1301 break;
1302
1303 case MAJOR_MAP >> MAJOR_SHIFT:
1304 len <<= 1;
1305 case MAJOR_ARRAY >> MAJOR_SHIFT:
1306 if (len)
1307 {
1308 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1309 count = len + 1;
1310 }
1311 break;
1312 }
1313 }
1314 }
1315 }
1316
1317 self->incr_pos = self->incr_need;
1318
1319 if (count > 0)
1320 {
1321 while (!--count)
1322 {
1323 if (!AvFILLp (self->incr_count))
1324 return 1; // done
1325
1326 SvREFCNT_dec_NN (av_pop (self->incr_count));
1327 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1328 }
1329
1330 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1331 }
1332 }
1333
1334 return 0;
1335}
1336
1337
970///////////////////////////////////////////////////////////////////////////// 1338/////////////////////////////////////////////////////////////////////////////
971// XS interface functions 1339// XS interface functions
972 1340
973MODULE = CBOR::XS PACKAGE = CBOR::XS 1341MODULE = CBOR::XS PACKAGE = CBOR::XS
974 1342
981 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1349 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
982 1350
983 types_true = get_bool ("Types::Serialiser::true" ); 1351 types_true = get_bool ("Types::Serialiser::true" );
984 types_false = get_bool ("Types::Serialiser::false"); 1352 types_false = get_bool ("Types::Serialiser::false");
985 types_error = get_bool ("Types::Serialiser::error"); 1353 types_error = get_bool ("Types::Serialiser::error");
1354
1355 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
986 1356
987 sv_cbor = newSVpv ("CBOR", 0); 1357 sv_cbor = newSVpv ("CBOR", 0);
988 SvREADONLY_on (sv_cbor); 1358 SvREADONLY_on (sv_cbor);
989} 1359}
990 1360
1012void shrink (CBOR *self, int enable = 1) 1382void shrink (CBOR *self, int enable = 1)
1013 ALIAS: 1383 ALIAS:
1014 shrink = F_SHRINK 1384 shrink = F_SHRINK
1015 allow_unknown = F_ALLOW_UNKNOWN 1385 allow_unknown = F_ALLOW_UNKNOWN
1016 allow_sharing = F_ALLOW_SHARING 1386 allow_sharing = F_ALLOW_SHARING
1017 dedup_keys = F_DEDUP_KEYS 1387 allow_cycles = F_ALLOW_CYCLES
1018 dedup_strings = F_DEDUP_STRINGS 1388 pack_strings = F_PACK_STRINGS
1389 validate_utf8 = F_VALIDATE_UTF8
1019 PPCODE: 1390 PPCODE:
1020{ 1391{
1021 if (enable) 1392 if (enable)
1022 self->flags |= ix; 1393 self->flags |= ix;
1023 else 1394 else
1029void get_shrink (CBOR *self) 1400void get_shrink (CBOR *self)
1030 ALIAS: 1401 ALIAS:
1031 get_shrink = F_SHRINK 1402 get_shrink = F_SHRINK
1032 get_allow_unknown = F_ALLOW_UNKNOWN 1403 get_allow_unknown = F_ALLOW_UNKNOWN
1033 get_allow_sharing = F_ALLOW_SHARING 1404 get_allow_sharing = F_ALLOW_SHARING
1034 get_dedup_keys = F_DEDUP_KEYS 1405 get_allow_cycles = F_ALLOW_CYCLES
1035 get_dedup_strings = F_DEDUP_STRINGS 1406 get_pack_strings = F_PACK_STRINGS
1407 get_validate_utf8 = F_VALIDATE_UTF8
1036 PPCODE: 1408 PPCODE:
1037 XPUSHs (boolSV (self->flags & ix)); 1409 XPUSHs (boolSV (self->flags & ix));
1038 1410
1039void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1411void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1040 PPCODE: 1412 PPCODE:
1053 XPUSHs (ST (0)); 1425 XPUSHs (ST (0));
1054 1426
1055int get_max_size (CBOR *self) 1427int get_max_size (CBOR *self)
1056 CODE: 1428 CODE:
1057 RETVAL = self->max_size; 1429 RETVAL = self->max_size;
1430 OUTPUT:
1431 RETVAL
1432
1433void filter (CBOR *self, SV *filter = 0)
1434 PPCODE:
1435 SvREFCNT_dec (self->filter);
1436 self->filter = filter ? newSVsv (filter) : filter;
1437 XPUSHs (ST (0));
1438
1439SV *get_filter (CBOR *self)
1440 CODE:
1441 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1058 OUTPUT: 1442 OUTPUT:
1059 RETVAL 1443 RETVAL
1060 1444
1061void encode (CBOR *self, SV *scalar) 1445void encode (CBOR *self, SV *scalar)
1062 PPCODE: 1446 PPCODE:
1077 EXTEND (SP, 2); 1461 EXTEND (SP, 2);
1078 PUSHs (sv); 1462 PUSHs (sv);
1079 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1463 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1080} 1464}
1081 1465
1466void incr_parse (CBOR *self, SV *cborstr)
1467 ALIAS:
1468 incr_parse_multiple = 1
1469 PPCODE:
1470{
1471 if (SvUTF8 (cborstr))
1472 sv_utf8_downgrade (cborstr, 0);
1473
1474 if (!self->incr_count)
1475 {
1476 self->incr_count = newAV ();
1477 self->incr_pos = 0;
1478 self->incr_need = 1;
1479
1480 av_push (self->incr_count, newSViv (1));
1481 }
1482
1483 do
1484 {
1485 if (!incr_parse (self, cborstr))
1486 {
1487 if (self->incr_need > self->max_size && self->max_size)
1488 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1489 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1490
1491 break;
1492 }
1493
1494 SV *sv;
1495 char *offset;
1496
1497 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1498 XPUSHs (sv);
1499
1500 sv_chop (cborstr, offset);
1501
1502 av_clear (self->incr_count);
1503 av_push (self->incr_count, newSViv (1));
1504
1505 self->incr_pos = 0;
1506 self->incr_need = self->incr_pos + 1;
1507 }
1508 while (ix);
1509}
1510
1511void incr_reset (CBOR *self)
1512 CODE:
1513{
1514 SvREFCNT_dec (self->incr_count);
1515 self->incr_count = 0;
1516}
1517
1518void DESTROY (CBOR *self)
1519 PPCODE:
1520 cbor_free (self);
1521
1082PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
1083 1523
1084void encode_cbor (SV *scalar) 1524void encode_cbor (SV *scalar)
1525 ALIAS:
1526 encode_cbor = 0
1527 encode_cbor_sharing = F_ALLOW_SHARING
1085 PPCODE: 1528 PPCODE:
1086{ 1529{
1087 CBOR cbor; 1530 CBOR cbor;
1088 cbor_init (&cbor); 1531 cbor_init (&cbor);
1532 cbor.flags |= ix;
1089 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1533 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1090 XPUSHs (scalar); 1534 XPUSHs (scalar);
1091} 1535}
1092 1536
1093void decode_cbor (SV *cborstr) 1537void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines