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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.18 by root, Wed Nov 20 01:09:46 2013 UTC vs.
Revision 1.51 by root, Sun Apr 24 13:15:19 2016 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 21#endif
21#ifndef HvNAMEUTF8 22#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 23# define HvNAMEUTF8(hv) 0
23#endif 24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
24 65
25// known tags 66// known tags
26enum cbor_tag 67enum cbor_tag
27{ 68{
28 // inofficial extensions (pending iana registration) 69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHARABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 104#define F_PACK_STRINGS 0x00000010UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 105#define F_UTF8_STRINGS 0x00000020UL
106#define F_VALIDATE_UTF8 0x00000040UL
64 107
65#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
66 109
67#define SB do { 110#define SB do {
68#define SE } while (0) 111#define SE } while (0)
80# define CBOR_SLOW 0 123# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 124# define CBOR_STASH cbor_stash
82#endif 125#endif
83 126
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 127static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 128static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 129
87typedef struct { 130typedef struct {
88 U32 flags; 131 U32 flags;
89 U32 max_depth; 132 U32 max_depth;
90 STRLEN max_size; 133 STRLEN max_size;
134 SV *filter;
135
136 // for the incremental parser
137 STRLEN incr_pos; // the current offset into the text
138 STRLEN incr_need; // minimum bytes needed to decode
139 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 140} CBOR;
92 141
93ecb_inline void 142ecb_inline void
94cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
95{ 144{
96 Zero (cbor, 1, CBOR); 145 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 146 cbor->max_depth = 512;
147}
148
149ecb_inline void
150cbor_free (CBOR *cbor)
151{
152 SvREFCNT_dec (cbor->filter);
153 SvREFCNT_dec (cbor->incr_count);
98} 154}
99 155
100///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
101// utility functions 157// utility functions
102 158
124 SvPV_renew (sv, SvCUR (sv) + 1); 180 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 181#endif
126 } 182 }
127} 183}
128 184
185// minimum length of a string to be registered for stringref
186ecb_inline int
187minimum_string_length (UV idx)
188{
189 return idx > 23
190 ? idx > 0xffU
191 ? idx > 0xffffU
192 ? idx > 0xffffffffU
193 ? 11
194 : 7
195 : 5
196 : 4
197 : 3;
198}
199
129///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
130// encoder 201// encoder
131 202
132// structure used for encoding CBOR 203// structure used for encoding CBOR
133typedef struct 204typedef struct
135 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
137 SV *sv; // result scalar 208 SV *sv; // result scalar
138 CBOR cbor; 209 CBOR cbor;
139 U32 depth; // recursion level 210 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 211 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
212 UV stringref_idx;
141 HV *sharable; // ptr => index, or 0 213 HV *shareable; // ptr => index, or 0
142 HV *sharable_idx; 214 UV shareable_idx;
143} enc_t; 215} enc_t;
144 216
145ecb_inline void 217ecb_inline void
146need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
147{ 219{
164static void 236static void
165encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
166{ 238{
167 need (enc, 9); 239 need (enc, 9);
168 240
169 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
171 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
172 { 244 {
173 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 246 *enc->cur++ = len;
175 } 247 }
176 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
177 { 249 {
178 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 252 *enc->cur++ = len;
181 } 253 }
182 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
183 { 255 {
184 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 260 *enc->cur++ = len;
189 } 261 }
190 else 262 else
191 { 263 {
192 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 272 *enc->cur++ = len;
201 } 273 }
202} 274}
203 275
204static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282ecb_inline void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 284{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS))
286 if (!utf8)
287 {
288 SV *sv = sv_newmortal ();
289 char *s; STRLEN l;
290
291 sv_setpvn (sv, str, len);
292
293 s = SvPVutf8 (sv, l);
294 encode_str (enc, 1, s, l);
295 return;
296 }
297
298 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 299 need (enc, len);
209 memcpy (enc->cur, str, len); 300 memcpy (enc->cur, str, len);
210 enc->cur += len; 301 enc->cur += len;
211} 302}
212 303
213ecb_inline void
214encode_tag (enc_t *enc, UV tag)
215{
216 encode_uint (enc, 0xc0, tag);
217}
218
219static int 304static void
220encode_sharable2 (enc_t *enc, SV *sv) 305encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
221{ 306{
222 if (!enc->sharable) 307 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
223 enc->sharable = (HV *)sv_2mortal ((SV *)newHV ()); 308 {
309 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
224 310
225 SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227 if (SvOK (*svp)) 311 if (SvOK (*svp))
228 { 312 {
313 // already registered, use stringref
229 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 314 encode_tag (enc, CBOR_TAG_STRINGREF);
230 encode_uint (enc, 0x00, SvUV (*svp)); 315 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232 return 1; 316 return;
233 } 317 }
234 else 318 else if (len >= minimum_string_length (enc->stringref_idx))
235 { 319 {
320 // register only
236 sv_setuv (*svp, enc->sharable_idx++); 321 sv_setuv (*svp, enc->stringref_idx);
237 encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); 322 ++enc->stringref_idx;
238 323 }
239 return 0;
240 } 324 }
241}
242 325
243ecb_inline int 326 encode_str (enc, utf8, str, len);
244encode_sharable (enc_t *enc, SV *sv)
245{
246 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
247 && ecb_expect_false (SvREFCNT (sv) > 1))
248 return encode_sharable2 (enc, sv);
249
250 return 0;
251} 327}
252 328
253static void encode_sv (enc_t *enc, SV *sv); 329static void encode_sv (enc_t *enc, SV *sv);
254 330
255static void 331static void
260 if (enc->depth >= enc->cbor.max_depth) 336 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 337 croak (ERR_NESTING_EXCEEDED);
262 338
263 ++enc->depth; 339 ++enc->depth;
264 340
265 encode_uint (enc, 0x80, len + 1); 341 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 342
343 if (SvMAGICAL (av))
267 for (i = 0; i <= len; ++i) 344 for (i = 0; i <= len; ++i)
268 { 345 {
269 SV **svp = av_fetch (av, i, 0); 346 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 347 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 348 }
349 else
350 for (i = 0; i <= len; ++i)
351 {
352 SV *sv = AvARRAY (av)[i];
353 encode_sv (enc, sv ? sv : &PL_sv_undef);
354 }
272 355
273 --enc->depth; 356 --enc->depth;
274} 357}
275 358
276static void 359static void
285 368
286 int pairs = hv_iterinit (hv); 369 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 370 int mg = SvMAGICAL (hv);
288 371
289 if (mg) 372 if (mg)
290 encode_ch (enc, 0xa0 | 31); 373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 374 else
292 encode_uint (enc, 0xa0, pairs); 375 encode_uint (enc, MAJOR_MAP, pairs);
293 376
294 while ((he = hv_iternext (hv))) 377 while ((he = hv_iternext (hv)))
295 { 378 {
296 if (HeKLEN (he) == HEf_SVKEY) 379 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 380 encode_sv (enc, HeSVKEY (he));
298 else 381 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 382 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 383
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 384 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 385 }
303 386
304 if (mg) 387 if (mg)
305 encode_ch (enc, 0xe0 | 31); 388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 389
307 --enc->depth; 390 --enc->depth;
308} 391}
309 392
310// encode objects, arrays and special \0=false and \1=true values. 393// encode objects, arrays and special \0=false and \1=true values.
311static void 394static void
312encode_rv (enc_t *enc, SV *sv) 395encode_rv (enc_t *enc, SV *sv)
313{ 396{
314 svtype svt;
315
316 SvGETMAGIC (sv); 397 SvGETMAGIC (sv);
398
317 svt = SvTYPE (sv); 399 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 400
322 if (ecb_expect_false (SvOBJECT (sv))) 401 if (ecb_expect_false (SvOBJECT (sv)))
323 { 402 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 403 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 404 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 409 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 410 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 411 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 412
334 HV *stash = SvSTASH (sv); 413 HV *stash = SvSTASH (sv);
335 GV *method;
336 414
337 if (stash == boolean_stash) 415 if (stash == boolean_stash)
338 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 416 {
417 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
418 return;
419 }
339 else if (stash == error_stash) 420 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 421 {
422 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
423 return;
424 }
341 else if (stash == tagged_stash) 425 else if (stash == tagged_stash)
342 { 426 {
343 if (svt != SVt_PVAV) 427 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 428 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 429
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 430 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 431 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
432
433 return;
434 }
435 }
436
437 if (ecb_expect_false (SvREFCNT (sv) > 1)
438 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
439 {
440 if (!enc->shareable)
441 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
442
443 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
444
445 if (SvOK (*svp))
348 } 446 {
447 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
448 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
449 return;
450 }
451 else
452 {
453 sv_setuv (*svp, enc->shareable_idx);
454 ++enc->shareable_idx;
455 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
456 }
457 }
458
459 if (ecb_expect_false (SvOBJECT (sv)))
460 {
461 HV *stash = SvSTASH (sv);
462 GV *method;
463
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 464 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 465 {
351 dSP; 466 dSP;
352 467
353 ENTER; SAVETMPS; PUSHMARK (SP); 468 ENTER; SAVETMPS;
469 PUSHMARK (SP);
354 // we re-bless the reference to get overload and other niceties right 470 // we re-bless the reference to get overload and other niceties right
355 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 471 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
356 472
357 PUTBACK; 473 PUTBACK;
358 // G_SCALAR ensures that return value is 1 474 // G_SCALAR ensures that return value is 1
371 } 487 }
372 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 488 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
373 { 489 {
374 dSP; 490 dSP;
375 491
376 ENTER; SAVETMPS; PUSHMARK (SP); 492 ENTER; SAVETMPS;
493 SAVESTACK_POS ();
494 PUSHMARK (SP);
377 EXTEND (SP, 2); 495 EXTEND (SP, 2);
378 // we re-bless the reference to get overload and other niceties right 496 // we re-bless the reference to get overload and other niceties right
379 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 497 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
380 PUSHs (sv_cbor); 498 PUSHs (sv_cbor);
381 499
386 // catch this surprisingly common error 504 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 505 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
388 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 506 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389 507
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 508 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 509 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 510 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 511
394 while (count) 512 while (count)
395 encode_sv (enc, SP[1 - count--]); 513 encode_sv (enc, SP[1 - count--]);
396 514
397 PUTBACK; 515 PUTBACK;
418{ 536{
419 double nv = SvNVX (sv); 537 double nv = SvNVX (sv);
420 538
421 need (enc, 9); 539 need (enc, 9);
422 540
423 if (ecb_expect_false (nv == (U32)nv)) 541 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 542 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 543 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 544 else if (ecb_expect_false (nv == (float)nv))
427 { 545 {
428 uint32_t fp = ecb_float_to_binary32 (nv); 546 uint32_t fp = ecb_float_to_binary32 (nv);
429 547
430 *enc->cur++ = 0xe0 | 26; 548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
431 549
432 if (!ecb_big_endian ()) 550 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp); 551 fp = ecb_bswap32 (fp);
434 552
435 memcpy (enc->cur, &fp, 4); 553 memcpy (enc->cur, &fp, 4);
437 } 555 }
438 else 556 else
439 { 557 {
440 uint64_t fp = ecb_double_to_binary64 (nv); 558 uint64_t fp = ecb_double_to_binary64 (nv);
441 559
442 *enc->cur++ = 0xe0 | 27; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
443 561
444 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp); 563 fp = ecb_bswap64 (fp);
446 564
447 memcpy (enc->cur, &fp, 8); 565 memcpy (enc->cur, &fp, 8);
452static void 570static void
453encode_sv (enc_t *enc, SV *sv) 571encode_sv (enc_t *enc, SV *sv)
454{ 572{
455 SvGETMAGIC (sv); 573 SvGETMAGIC (sv);
456 574
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 575 if (SvPOKp (sv))
461 { 576 {
462 STRLEN len; 577 STRLEN len;
463 char *str = SvPV (sv, len); 578 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 579 encode_strref (enc, SvUTF8 (sv), str, len);
465 } 580 }
466 else if (SvNOKp (sv)) 581 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 582 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 583 else if (SvIOKp (sv))
469 { 584 {
470 if (SvIsUV (sv)) 585 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv)); 586 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
472 else if (SvIVX (sv) >= 0) 587 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv)); 588 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
474 else 589 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 590 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
476 } 591 }
477 else if (SvROK (sv)) 592 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 593 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 594 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 595 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 596 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 597 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 598 else
484 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 599 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 600 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 601}
487 602
488static SV * 603static SV *
489encode_cbor (SV *scalar, CBOR *cbor) 604encode_cbor (SV *scalar, CBOR *cbor)
490{ 605{
491 enc_t enc = { }; 606 enc_t enc = { 0 };
492 607
493 enc.cbor = *cbor; 608 enc.cbor = *cbor;
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 610 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 611 enc.end = SvEND (enc.sv);
497 612
498 SvPOK_only (enc.sv); 613 SvPOK_only (enc.sv);
614
615 if (cbor->flags & F_PACK_STRINGS)
616 {
617 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
618 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
619 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
620 }
621
499 encode_sv (&enc, scalar); 622 encode_sv (&enc, scalar);
500 623
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 624 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 625 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503 626
517 U8 *end; // end of input string 640 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 641 const char *err; // parse error, if != 0
519 CBOR cbor; 642 CBOR cbor;
520 U32 depth; // recursion depth 643 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 644 U32 maxdepth; // recursion depth limit
522 AV *sharable; 645 AV *shareable;
646 AV *stringref;
647 SV *decode_tagged;
523} dec_t; 648} dec_t;
524 649
525#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 650#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
526 651
527#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 652#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
530#define DEC_DEC_DEPTH --dec->depth 655#define DEC_DEC_DEPTH --dec->depth
531 656
532static UV 657static UV
533decode_uint (dec_t *dec) 658decode_uint (dec_t *dec)
534{ 659{
535 switch (*dec->cur & 31) 660 U8 m = *dec->cur & MINOR_MASK;
536 { 661 ++dec->cur;
537 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
538 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
539 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
540 return *dec->cur++ & 31;
541 662
542 case 24: 663 if (ecb_expect_true (m < LENGTH_EXT1))
664 return m;
665 else if (ecb_expect_true (m == LENGTH_EXT1))
666 {
543 WANT (2); 667 WANT (1);
544 dec->cur += 2; 668 dec->cur += 1;
545 return dec->cur[-1]; 669 return dec->cur[-1];
546 670 }
547 case 25: 671 else if (ecb_expect_true (m == LENGTH_EXT2))
672 {
548 WANT (3); 673 WANT (2);
549 dec->cur += 3; 674 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 675 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 676 | ((UV)dec->cur[-1]);
552 677 }
553 case 26: 678 else if (ecb_expect_true (m == LENGTH_EXT4))
679 {
554 WANT (5); 680 WANT (4);
555 dec->cur += 5; 681 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 682 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 683 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 684 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 685 | ((UV)dec->cur[-1]);
560 686 }
561 case 27: 687 else if (ecb_expect_true (m == LENGTH_EXT8))
688 {
562 WANT (9); 689 WANT (8);
563 dec->cur += 9; 690 dec->cur += 8;
691
692 return
693#if UVSIZE < 8
694 0
695#else
564 return (((UV)dec->cur[-8]) << 56) 696 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 697 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 698 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 699 | (((UV)dec->cur[-5]) << 32)
700#endif
568 | (((UV)dec->cur[-4]) << 24) 701 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 702 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 703 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 704 | ((UV)dec->cur[-1]);
572 705 }
573 default: 706 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 707 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 708
577fail: 709fail:
578 return 0; 710 return 0;
579} 711}
580 712
585{ 717{
586 AV *av = newAV (); 718 AV *av = newAV ();
587 719
588 DEC_INC_DEPTH; 720 DEC_INC_DEPTH;
589 721
590 if ((*dec->cur & 31) == 31) 722 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 723 {
592 ++dec->cur; 724 ++dec->cur;
593 725
594 for (;;) 726 for (;;)
595 { 727 {
596 WANT (1); 728 WANT (1);
597 729
598 if (*dec->cur == (0xe0 | 31)) 730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 731 {
600 ++dec->cur; 732 ++dec->cur;
601 break; 733 break;
602 } 734 }
603 735
606 } 738 }
607 else 739 else
608 { 740 {
609 int i, len = decode_uint (dec); 741 int i, len = decode_uint (dec);
610 742
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
611 av_fill (av, len - 1); 744 av_fill (av, len - 1);
612 745
613 for (i = 0; i < len; ++i) 746 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 747 AvARRAY (av)[i] = decode_sv (dec);
615 } 748 }
625 758
626static void 759static void
627decode_he (dec_t *dec, HV *hv) 760decode_he (dec_t *dec, HV *hv)
628{ 761{
629 // for speed reasons, we specialcase single-string 762 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 763 // byte or utf-8 strings as keys, but only when !stringref
631 764
632 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 765 if (ecb_expect_true (!dec->stringref))
766 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633 { 767 {
634 I32 len = decode_uint (dec); 768 I32 len = decode_uint (dec);
635 char *key = (char *)dec->cur; 769 char *key = (char *)dec->cur;
636 770
771 WANT (len);
637 dec->cur += len; 772 dec->cur += len;
638 773
639 hv_store (hv, key, len, decode_sv (dec), 0); 774 hv_store (hv, key, len, decode_sv (dec), 0);
775
776 return;
640 } 777 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 778 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 779 {
643 I32 len = decode_uint (dec); 780 I32 len = decode_uint (dec);
644 char *key = (char *)dec->cur; 781 char *key = (char *)dec->cur;
645 782
783 WANT (len);
646 dec->cur += len; 784 dec->cur += len;
647 785
786 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
787 if (!is_utf8_string (key, len))
788 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
789
648 hv_store (hv, key, -len, decode_sv (dec), 0); 790 hv_store (hv, key, -len, decode_sv (dec), 0);
791
792 return;
649 } 793 }
650 else 794
651 {
652 SV *k = decode_sv (dec); 795 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 796 SV *v = decode_sv (dec);
654 797
655 hv_store_ent (hv, k, v, 0); 798 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 799 SvREFCNT_dec (k);
657 } 800
801fail:
802 ;
658} 803}
659 804
660static SV * 805static SV *
661decode_hv (dec_t *dec) 806decode_hv (dec_t *dec)
662{ 807{
663 HV *hv = newHV (); 808 HV *hv = newHV ();
664 809
665 DEC_INC_DEPTH; 810 DEC_INC_DEPTH;
666 811
667 if ((*dec->cur & 31) == 31) 812 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 813 {
669 ++dec->cur; 814 ++dec->cur;
670 815
671 for (;;) 816 for (;;)
672 { 817 {
673 WANT (1); 818 WANT (1);
674 819
675 if (*dec->cur == (0xe0 | 31)) 820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
676 { 821 {
677 ++dec->cur; 822 ++dec->cur;
678 break; 823 break;
679 } 824 }
680 825
701static SV * 846static SV *
702decode_str (dec_t *dec, int utf8) 847decode_str (dec_t *dec, int utf8)
703{ 848{
704 SV *sv = 0; 849 SV *sv = 0;
705 850
706 if ((*dec->cur & 31) == 31) 851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
707 { 852 {
853 // indefinite length strings
708 ++dec->cur; 854 ++dec->cur;
709 855
856 U8 major = *dec->cur & MAJOR_MISC;
857
710 sv = newSVpvn ("", 0); 858 sv = newSVpvn ("", 0);
711 859
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 860 for (;;)
714 { 861 {
715 WANT (1); 862 WANT (1);
716 863
717 if (*dec->cur == (0xe0 | 31)) 864 if ((*dec->cur - major) > LENGTH_EXT8)
865 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 866 {
719 ++dec->cur; 867 ++dec->cur;
720 break; 868 break;
721 } 869 }
870 else
871 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 872
723 sv_catsv (sv, decode_sv (dec)); 873 STRLEN len = decode_uint (dec);
874
875 WANT (len);
876 sv_catpvn (sv, dec->cur, len);
877 dec->cur += len;
724 } 878 }
725 } 879 }
726 else 880 else
727 { 881 {
728 STRLEN len = decode_uint (dec); 882 STRLEN len = decode_uint (dec);
729 883
730 WANT (len); 884 WANT (len);
731 sv = newSVpvn (dec->cur, len); 885 sv = newSVpvn (dec->cur, len);
732 dec->cur += len; 886 dec->cur += len;
887
888 if (ecb_expect_false (dec->stringref)
889 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
890 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733 } 891 }
734 892
735 if (utf8) 893 if (utf8)
894 {
895 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
896 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
897 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
898
736 SvUTF8_on (sv); 899 SvUTF8_on (sv);
900 }
737 901
738 return sv; 902 return sv;
739 903
740fail: 904fail:
741 SvREFCNT_dec (sv); 905 SvREFCNT_dec (sv);
743} 907}
744 908
745static SV * 909static SV *
746decode_tagged (dec_t *dec) 910decode_tagged (dec_t *dec)
747{ 911{
912 SV *sv = 0;
748 UV tag = decode_uint (dec); 913 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 914
915 WANT (1);
750 916
751 switch (tag) 917 switch (tag)
752 { 918 {
753 case CBOR_TAG_MAGIC: 919 case CBOR_TAG_MAGIC:
754 return sv; 920 sv = decode_sv (dec);
921 break;
755 922
756 case CBOR_TAG_INDIRECTION: 923 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 924 sv = newRV_noinc (decode_sv (dec));
925 break;
758 926
927 case CBOR_TAG_STRINGREF_NAMESPACE:
928 {
929 ENTER; SAVETMPS;
930
931 SAVESPTR (dec->stringref);
932 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
933
934 sv = decode_sv (dec);
935
936 FREETMPS; LEAVE;
937 }
938 break;
939
940 case CBOR_TAG_STRINGREF:
941 {
942 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
943 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
944
945 UV idx = decode_uint (dec);
946
947 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
948 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
949
950 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
951 }
952 break;
953
759 case CBOR_TAG_VALUE_SHARABLE: 954 case CBOR_TAG_VALUE_SHAREABLE:
955 {
760 if (ecb_expect_false (!dec->sharable)) 956 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 957 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 958
959 if (dec->cbor.flags & F_ALLOW_CYCLES)
960 {
961 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 962 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 963
765 return sv; 964 SV *osv = decode_sv (dec);
965 sv_setsv (sv, osv);
966 SvREFCNT_dec_NN (osv);
967 }
968 else
969 {
970 av_push (dec->shareable, &PL_sv_undef);
971 int idx = AvFILLp (dec->shareable);
972 sv = decode_sv (dec);
973 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
974 }
975 }
976 break;
766 977
767 case CBOR_TAG_VALUE_SHAREDREF: 978 case CBOR_TAG_VALUE_SHAREDREF:
768 { 979 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 980 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 981 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 982
983 UV idx = decode_uint (dec);
984
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 985 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 986 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 987
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); 988 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
989
990 if (sv == &PL_sv_undef)
991 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
778 } 992 }
993 break;
779 994
780 case CBOR_TAG_PERL_OBJECT: 995 case CBOR_TAG_PERL_OBJECT:
781 { 996 {
997 sv = decode_sv (dec);
998
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 999 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 1000 ERR ("corrupted CBOR data (non-array perl object)");
784 1001
785 AV *av = (AV *)SvRV (sv); 1002 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 1003 int len = av_len (av) + 1;
794 if (!method) 1011 if (!method)
795 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1012 ERR ("cannot decode perl-object (package does not have a THAW method)");
796 1013
797 dSP; 1014 dSP;
798 1015
799 ENTER; SAVETMPS; PUSHMARK (SP); 1016 ENTER; SAVETMPS;
1017 PUSHMARK (SP);
800 EXTEND (SP, len + 1); 1018 EXTEND (SP, len + 1);
801 // we re-bless the reference to get overload and other niceties right 1019 // we re-bless the reference to get overload and other niceties right
802 PUSHs (*av_fetch (av, 0, 1)); 1020 PUSHs (*av_fetch (av, 0, 1));
803 PUSHs (sv_cbor); 1021 PUSHs (sv_cbor);
804 1022
821 sv = SvREFCNT_inc (POPs); 1039 sv = SvREFCNT_inc (POPs);
822 1040
823 PUTBACK; 1041 PUTBACK;
824 1042
825 FREETMPS; LEAVE; 1043 FREETMPS; LEAVE;
826
827 return sv;
828 } 1044 }
1045 break;
829 1046
830 default: 1047 default:
831 { 1048 {
1049 sv = decode_sv (dec);
1050
1051 dSP;
1052 ENTER; SAVETMPS;
1053 SAVESTACK_POS ();
1054 PUSHMARK (SP);
1055 EXTEND (SP, 2);
1056 PUSHs (newSVuv (tag));
1057 PUSHs (sv);
1058
1059 PUTBACK;
1060 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1061 SPAGAIN;
1062
1063 if (SvTRUE (ERRSV))
1064 {
1065 FREETMPS; LEAVE;
1066 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1067 }
1068
1069 if (count)
1070 {
1071 SvREFCNT_dec (sv);
1072 sv = SvREFCNT_inc (POPs);
1073 }
1074 else
1075 {
832 AV *av = newAV (); 1076 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1077 av_push (av, newSVuv (tag));
834 av_push (av, sv); 1078 av_push (av, sv);
835 1079
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1080 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1081 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1082 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1083 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1084 }
1085
1086 PUTBACK;
1087
1088 FREETMPS; LEAVE;
1089 }
1090 break;
842 } 1091 }
1092
1093 return sv;
843 1094
844fail: 1095fail:
845 SvREFCNT_dec (sv); 1096 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1097 return &PL_sv_undef;
847} 1098}
849static SV * 1100static SV *
850decode_sv (dec_t *dec) 1101decode_sv (dec_t *dec)
851{ 1102{
852 WANT (1); 1103 WANT (1);
853 1104
854 switch (*dec->cur >> 5) 1105 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1106 {
856 case 0: // unsigned int 1107 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1108 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1109 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1110 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1111 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1112 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1113 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1114
864 case 4: // array 1115 case MAJOR_MISC >> MAJOR_SHIFT:
865 return decode_av (dec);
866 case 5: // map
867 return decode_hv (dec);
868 case 6: // tag
869 return decode_tagged (dec);
870 case 7: // misc
871 switch (*dec->cur++ & 31) 1116 switch (*dec->cur++ & MINOR_MASK)
872 { 1117 {
873 case 20: 1118 case SIMPLE_FALSE:
874#if CBOR_SLOW 1119#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1120 types_false = get_bool ("Types::Serialiser::false");
876#endif 1121#endif
877 return newSVsv (types_false); 1122 return newSVsv (types_false);
878 case 21: 1123 case SIMPLE_TRUE:
879#if CBOR_SLOW 1124#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1125 types_true = get_bool ("Types::Serialiser::true");
881#endif 1126#endif
882 return newSVsv (types_true); 1127 return newSVsv (types_true);
883 case 22: 1128 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1129 return newSVsv (&PL_sv_undef);
885 case 23: 1130 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1131#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1132 types_error = get_bool ("Types::Serialiser::error");
888#endif 1133#endif
889 return newSVsv (types_error); 1134 return newSVsv (types_error);
890 1135
891 case 25: 1136 case MISC_FLOAT16:
892 { 1137 {
893 WANT (2); 1138 WANT (2);
894 1139
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1140 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1141 dec->cur += 2;
897 1142
898 return newSVnv (ecb_binary16_to_float (fp)); 1143 return newSVnv (ecb_binary16_to_float (fp));
899 } 1144 }
900 1145
901 case 26: 1146 case MISC_FLOAT32:
902 { 1147 {
903 uint32_t fp; 1148 uint32_t fp;
904 WANT (4); 1149 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1150 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1151 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1154 fp = ecb_bswap32 (fp);
910 1155
911 return newSVnv (ecb_binary32_to_float (fp)); 1156 return newSVnv (ecb_binary32_to_float (fp));
912 } 1157 }
913 1158
914 case 27: 1159 case MISC_FLOAT64:
915 { 1160 {
916 uint64_t fp; 1161 uint64_t fp;
917 WANT (8); 1162 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1163 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1164 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1167 fp = ecb_bswap64 (fp);
923 1168
924 return newSVnv (ecb_binary64_to_double (fp)); 1169 return newSVnv (ecb_binary64_to_double (fp));
925 } 1170 }
926 1171
927 // 0..19 unassigned 1172 // 0..19 unassigned simple
928 // 24 reserved + unassigned (reserved values are not encodable) 1173 // 24 reserved + unassigned simple (reserved values are not encodable)
1174 // 28-30 unassigned misc
1175 // 31 break code
929 default: 1176 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1177 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1178 }
932 1179
933 break; 1180 break;
934 } 1181 }
935 1182
938} 1185}
939 1186
940static SV * 1187static SV *
941decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1188decode_cbor (SV *string, CBOR *cbor, char **offset_return)
942{ 1189{
943 dec_t dec = { }; 1190 dec_t dec = { 0 };
944 SV *sv; 1191 SV *sv;
945 STRLEN len; 1192 STRLEN len;
946 char *data = SvPVbyte (string, len); 1193 char *data = SvPVbyte (string, len);
947 1194
948 if (len > cbor->max_size && cbor->max_size) 1195 if (len > cbor->max_size && cbor->max_size)
962 if (dec.cur != dec.end && !dec.err) 1209 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1210 dec.err = "garbage after CBOR object";
964 1211
965 if (dec.err) 1212 if (dec.err)
966 { 1213 {
1214 if (dec.shareable)
1215 {
1216 // need to break cyclic links, which whould all be in shareable
1217 int i;
1218 SV **svp;
1219
1220 for (i = av_len (dec.shareable) + 1; i--; )
1221 if ((svp = av_fetch (dec.shareable, i, 0)))
1222 sv_setsv (*svp, &PL_sv_undef);
1223 }
1224
967 SvREFCNT_dec (sv); 1225 SvREFCNT_dec (sv);
968 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1226 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
969 } 1227 }
970 1228
971 sv = sv_2mortal (sv); 1229 sv = sv_2mortal (sv);
972 1230
973 return sv; 1231 return sv;
974} 1232}
975 1233
1234/////////////////////////////////////////////////////////////////////////////
1235// incremental parser
1236
1237#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1238
1239// returns 0 for notyet, 1 for success or error
1240static int
1241incr_parse (CBOR *self, SV *cborstr)
1242{
1243 STRLEN cur;
1244 SvPV (cborstr, cur);
1245
1246 while (ecb_expect_true (self->incr_need <= cur))
1247 {
1248 // table of integer count bytes
1249 static I8 incr_len[MINOR_MASK + 1] = {
1250 0, 0, 0, 0, 0, 0, 0, 0,
1251 0, 0, 0, 0, 0, 0, 0, 0,
1252 0, 0, 0, 0, 0, 0, 0, 0,
1253 1, 2, 4, 8,-1,-1,-1,-2
1254 };
1255
1256 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1257 U8 m = *p & MINOR_MASK;
1258 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1259 I8 ilen = incr_len[m];
1260
1261 self->incr_need = self->incr_pos + 1;
1262
1263 if (ecb_expect_false (ilen < 0))
1264 {
1265 if (m != MINOR_INDEF)
1266 return 1; // error
1267
1268 if (*p == (MAJOR_MISC | MINOR_INDEF))
1269 {
1270 if (count >= 0)
1271 return 1; // error
1272
1273 count = 1;
1274 }
1275 else
1276 {
1277 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1278 count = -1;
1279 }
1280 }
1281 else
1282 {
1283 self->incr_need += ilen;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 int major = *p >> MAJOR_SHIFT;
1288
1289 switch (major)
1290 {
1291 case MAJOR_TAG >> MAJOR_SHIFT:
1292 ++count; // tags merely prefix another value
1293 break;
1294
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 case MAJOR_ARRAY >> MAJOR_SHIFT:
1298 case MAJOR_MAP >> MAJOR_SHIFT:
1299 {
1300 UV len;
1301
1302 if (ecb_expect_false (ilen))
1303 {
1304 len = 0;
1305
1306 do {
1307 len = (len << 8) | *++p;
1308 } while (--ilen);
1309 }
1310 else
1311 len = m;
1312
1313 switch (major)
1314 {
1315 case MAJOR_BYTES >> MAJOR_SHIFT:
1316 case MAJOR_TEXT >> MAJOR_SHIFT:
1317 self->incr_need += len;
1318 if (ecb_expect_false (self->incr_need > cur))
1319 return 0;
1320
1321 break;
1322
1323 case MAJOR_MAP >> MAJOR_SHIFT:
1324 len <<= 1;
1325 case MAJOR_ARRAY >> MAJOR_SHIFT:
1326 if (len)
1327 {
1328 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1329 count = len + 1;
1330 }
1331 break;
1332 }
1333 }
1334 }
1335 }
1336
1337 self->incr_pos = self->incr_need;
1338
1339 if (count > 0)
1340 {
1341 while (!--count)
1342 {
1343 if (!AvFILLp (self->incr_count))
1344 return 1; // done
1345
1346 SvREFCNT_dec_NN (av_pop (self->incr_count));
1347 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1348 }
1349
1350 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1351 }
1352 }
1353
1354 return 0;
1355}
1356
1357
976///////////////////////////////////////////////////////////////////////////// 1358/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1359// XS interface functions
978 1360
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1361MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1362
987 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1369 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
988 1370
989 types_true = get_bool ("Types::Serialiser::true" ); 1371 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1372 types_false = get_bool ("Types::Serialiser::false");
991 types_error = get_bool ("Types::Serialiser::error"); 1373 types_error = get_bool ("Types::Serialiser::error");
1374
1375 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
992 1376
993 sv_cbor = newSVpv ("CBOR", 0); 1377 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1378 SvREADONLY_on (sv_cbor);
995} 1379}
996 1380
1018void shrink (CBOR *self, int enable = 1) 1402void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1403 ALIAS:
1020 shrink = F_SHRINK 1404 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1405 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1406 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1407 allow_cycles = F_ALLOW_CYCLES
1024 dedup_strings = F_DEDUP_STRINGS 1408 pack_strings = F_PACK_STRINGS
1409 utf8_strings = F_UTF8_STRINGS
1410 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1411 PPCODE:
1026{ 1412{
1027 if (enable) 1413 if (enable)
1028 self->flags |= ix; 1414 self->flags |= ix;
1029 else 1415 else
1035void get_shrink (CBOR *self) 1421void get_shrink (CBOR *self)
1036 ALIAS: 1422 ALIAS:
1037 get_shrink = F_SHRINK 1423 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1424 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1425 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1426 get_allow_cycles = F_ALLOW_CYCLES
1041 get_dedup_strings = F_DEDUP_STRINGS 1427 get_pack_strings = F_PACK_STRINGS
1428 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1429 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1430 XPUSHs (boolSV (self->flags & ix));
1044 1431
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1433 PPCODE:
1059 XPUSHs (ST (0)); 1446 XPUSHs (ST (0));
1060 1447
1061int get_max_size (CBOR *self) 1448int get_max_size (CBOR *self)
1062 CODE: 1449 CODE:
1063 RETVAL = self->max_size; 1450 RETVAL = self->max_size;
1451 OUTPUT:
1452 RETVAL
1453
1454void filter (CBOR *self, SV *filter = 0)
1455 PPCODE:
1456 SvREFCNT_dec (self->filter);
1457 self->filter = filter ? newSVsv (filter) : filter;
1458 XPUSHs (ST (0));
1459
1460SV *get_filter (CBOR *self)
1461 CODE:
1462 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064 OUTPUT: 1463 OUTPUT:
1065 RETVAL 1464 RETVAL
1066 1465
1067void encode (CBOR *self, SV *scalar) 1466void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1467 PPCODE:
1083 EXTEND (SP, 2); 1482 EXTEND (SP, 2);
1084 PUSHs (sv); 1483 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1484 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1485}
1087 1486
1487void incr_parse (CBOR *self, SV *cborstr)
1488 ALIAS:
1489 incr_parse_multiple = 1
1490 PPCODE:
1491{
1492 if (SvUTF8 (cborstr))
1493 sv_utf8_downgrade (cborstr, 0);
1494
1495 if (!self->incr_count)
1496 {
1497 self->incr_count = newAV ();
1498 self->incr_pos = 0;
1499 self->incr_need = 1;
1500
1501 av_push (self->incr_count, newSViv (1));
1502 }
1503
1504 do
1505 {
1506 if (!incr_parse (self, cborstr))
1507 {
1508 if (self->incr_need > self->max_size && self->max_size)
1509 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1510 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1511
1512 break;
1513 }
1514
1515 SV *sv;
1516 char *offset;
1517
1518 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1519 XPUSHs (sv);
1520
1521 sv_chop (cborstr, offset);
1522
1523 av_clear (self->incr_count);
1524 av_push (self->incr_count, newSViv (1));
1525
1526 self->incr_pos = 0;
1527 self->incr_need = self->incr_pos + 1;
1528 }
1529 while (ix);
1530}
1531
1532void incr_reset (CBOR *self)
1533 CODE:
1534{
1535 SvREFCNT_dec (self->incr_count);
1536 self->incr_count = 0;
1537}
1538
1539void DESTROY (CBOR *self)
1540 PPCODE:
1541 cbor_free (self);
1542
1088PROTOTYPES: ENABLE 1543PROTOTYPES: ENABLE
1089 1544
1090void encode_cbor (SV *scalar) 1545void encode_cbor (SV *scalar)
1546 ALIAS:
1547 encode_cbor = 0
1548 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1549 PPCODE:
1092{ 1550{
1093 CBOR cbor; 1551 CBOR cbor;
1094 cbor_init (&cbor); 1552 cbor_init (&cbor);
1553 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1554 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1555 XPUSHs (scalar);
1097} 1556}
1098 1557
1099void decode_cbor (SV *cborstr) 1558void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines