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 1.52 by root, Sun Apr 24 19:31:55 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 // exceptional path for bytze strings that need to be utf8-encoded
289 STRLEN ulen = len;
290 U8 *p, *pend = (U8 *)str + len;
291
292 for (p = (U8 *)str; p < pend; ++p)
293 ulen += *p >> 7; // count set high bits
294
295 encode_uint (enc, MAJOR_TEXT, ulen);
296
297 need (enc, ulen);
298 for (p = (U8 *)str; p < pend; ++p)
299 if (*p < 0x80)
300 *enc->cur++ = *p;
301 else
302 {
303 *enc->cur++ = 0xc0 + (*p >> 6);
304 *enc->cur++ = 0x80 + (*p & 63);
305 }
306
307 return;
308 }
309
310 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 311 need (enc, len);
209 memcpy (enc->cur, str, len); 312 memcpy (enc->cur, str, len);
210 enc->cur += len; 313 enc->cur += len;
211} 314}
212 315
213ecb_inline void
214encode_tag (enc_t *enc, UV tag)
215{
216 encode_uint (enc, 0xc0, tag);
217}
218
219static int 316static void
220encode_sharable2 (enc_t *enc, SV *sv) 317encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
221{ 318{
222 if (!enc->sharable) 319 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
223 enc->sharable = (HV *)sv_2mortal ((SV *)newHV ()); 320 {
321 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
224 322
225 SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227 if (SvOK (*svp)) 323 if (SvOK (*svp))
228 { 324 {
325 // already registered, use stringref
229 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 326 encode_tag (enc, CBOR_TAG_STRINGREF);
230 encode_uint (enc, 0x00, SvUV (*svp)); 327 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232 return 1; 328 return;
233 } 329 }
234 else 330 else if (len >= minimum_string_length (enc->stringref_idx))
235 { 331 {
332 // register only
236 sv_setuv (*svp, enc->sharable_idx++); 333 sv_setuv (*svp, enc->stringref_idx);
237 encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); 334 ++enc->stringref_idx;
238 335 }
239 return 0;
240 } 336 }
241}
242 337
243ecb_inline int 338 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} 339}
252 340
253static void encode_sv (enc_t *enc, SV *sv); 341static void encode_sv (enc_t *enc, SV *sv);
254 342
255static void 343static void
260 if (enc->depth >= enc->cbor.max_depth) 348 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 349 croak (ERR_NESTING_EXCEEDED);
262 350
263 ++enc->depth; 351 ++enc->depth;
264 352
265 encode_uint (enc, 0x80, len + 1); 353 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 354
355 if (SvMAGICAL (av))
267 for (i = 0; i <= len; ++i) 356 for (i = 0; i <= len; ++i)
268 { 357 {
269 SV **svp = av_fetch (av, i, 0); 358 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 359 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 360 }
361 else
362 for (i = 0; i <= len; ++i)
363 {
364 SV *sv = AvARRAY (av)[i];
365 encode_sv (enc, sv ? sv : &PL_sv_undef);
366 }
272 367
273 --enc->depth; 368 --enc->depth;
274} 369}
275 370
276static void 371static void
285 380
286 int pairs = hv_iterinit (hv); 381 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 382 int mg = SvMAGICAL (hv);
288 383
289 if (mg) 384 if (mg)
290 encode_ch (enc, 0xa0 | 31); 385 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 386 else
292 encode_uint (enc, 0xa0, pairs); 387 encode_uint (enc, MAJOR_MAP, pairs);
293 388
294 while ((he = hv_iternext (hv))) 389 while ((he = hv_iternext (hv)))
295 { 390 {
296 if (HeKLEN (he) == HEf_SVKEY) 391 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 392 encode_sv (enc, HeSVKEY (he));
298 else 393 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 394 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 395
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 396 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 397 }
303 398
304 if (mg) 399 if (mg)
305 encode_ch (enc, 0xe0 | 31); 400 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 401
307 --enc->depth; 402 --enc->depth;
308} 403}
309 404
310// encode objects, arrays and special \0=false and \1=true values. 405// encode objects, arrays and special \0=false and \1=true values.
311static void 406static void
312encode_rv (enc_t *enc, SV *sv) 407encode_rv (enc_t *enc, SV *sv)
313{ 408{
314 svtype svt;
315
316 SvGETMAGIC (sv); 409 SvGETMAGIC (sv);
410
317 svt = SvTYPE (sv); 411 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 412
322 if (ecb_expect_false (SvOBJECT (sv))) 413 if (ecb_expect_false (SvOBJECT (sv)))
323 { 414 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 415 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 416 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 421 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 422 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 423 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 424
334 HV *stash = SvSTASH (sv); 425 HV *stash = SvSTASH (sv);
335 GV *method;
336 426
337 if (stash == boolean_stash) 427 if (stash == boolean_stash)
338 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 428 {
429 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
430 return;
431 }
339 else if (stash == error_stash) 432 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 433 {
434 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 return;
436 }
341 else if (stash == tagged_stash) 437 else if (stash == tagged_stash)
342 { 438 {
343 if (svt != SVt_PVAV) 439 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 440 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 441
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 442 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 443 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
444
445 return;
446 }
447 }
448
449 if (ecb_expect_false (SvREFCNT (sv) > 1)
450 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
451 {
452 if (!enc->shareable)
453 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
454
455 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
456
457 if (SvOK (*svp))
348 } 458 {
459 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
460 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
461 return;
462 }
463 else
464 {
465 sv_setuv (*svp, enc->shareable_idx);
466 ++enc->shareable_idx;
467 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
468 }
469 }
470
471 if (ecb_expect_false (SvOBJECT (sv)))
472 {
473 HV *stash = SvSTASH (sv);
474 GV *method;
475
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 476 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 477 {
351 dSP; 478 dSP;
352 479
353 ENTER; SAVETMPS; PUSHMARK (SP); 480 ENTER; SAVETMPS;
481 PUSHMARK (SP);
354 // we re-bless the reference to get overload and other niceties right 482 // we re-bless the reference to get overload and other niceties right
355 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 483 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
356 484
357 PUTBACK; 485 PUTBACK;
358 // G_SCALAR ensures that return value is 1 486 // G_SCALAR ensures that return value is 1
371 } 499 }
372 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 500 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
373 { 501 {
374 dSP; 502 dSP;
375 503
376 ENTER; SAVETMPS; PUSHMARK (SP); 504 ENTER; SAVETMPS;
505 SAVESTACK_POS ();
506 PUSHMARK (SP);
377 EXTEND (SP, 2); 507 EXTEND (SP, 2);
378 // we re-bless the reference to get overload and other niceties right 508 // we re-bless the reference to get overload and other niceties right
379 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 509 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
380 PUSHs (sv_cbor); 510 PUSHs (sv_cbor);
381 511
386 // catch this surprisingly common error 516 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 517 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)); 518 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389 519
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 520 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 521 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 522 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 523
394 while (count) 524 while (count)
395 encode_sv (enc, SP[1 - count--]); 525 encode_sv (enc, SP[1 - count--]);
396 526
397 PUTBACK; 527 PUTBACK;
418{ 548{
419 double nv = SvNVX (sv); 549 double nv = SvNVX (sv);
420 550
421 need (enc, 9); 551 need (enc, 9);
422 552
423 if (ecb_expect_false (nv == (U32)nv)) 553 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 554 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 555 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 556 else if (ecb_expect_false (nv == (float)nv))
427 { 557 {
428 uint32_t fp = ecb_float_to_binary32 (nv); 558 uint32_t fp = ecb_float_to_binary32 (nv);
429 559
430 *enc->cur++ = 0xe0 | 26; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
431 561
432 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp); 563 fp = ecb_bswap32 (fp);
434 564
435 memcpy (enc->cur, &fp, 4); 565 memcpy (enc->cur, &fp, 4);
437 } 567 }
438 else 568 else
439 { 569 {
440 uint64_t fp = ecb_double_to_binary64 (nv); 570 uint64_t fp = ecb_double_to_binary64 (nv);
441 571
442 *enc->cur++ = 0xe0 | 27; 572 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
443 573
444 if (!ecb_big_endian ()) 574 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp); 575 fp = ecb_bswap64 (fp);
446 576
447 memcpy (enc->cur, &fp, 8); 577 memcpy (enc->cur, &fp, 8);
452static void 582static void
453encode_sv (enc_t *enc, SV *sv) 583encode_sv (enc_t *enc, SV *sv)
454{ 584{
455 SvGETMAGIC (sv); 585 SvGETMAGIC (sv);
456 586
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 587 if (SvPOKp (sv))
461 { 588 {
462 STRLEN len; 589 STRLEN len;
463 char *str = SvPV (sv, len); 590 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 591 encode_strref (enc, SvUTF8 (sv), str, len);
465 } 592 }
466 else if (SvNOKp (sv)) 593 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 594 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 595 else if (SvIOKp (sv))
469 { 596 {
470 if (SvIsUV (sv)) 597 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv)); 598 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
472 else if (SvIVX (sv) >= 0) 599 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv)); 600 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
474 else 601 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 602 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
476 } 603 }
477 else if (SvROK (sv)) 604 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 605 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 606 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 607 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 608 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 609 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 610 else
484 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 611 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 612 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 613}
487 614
488static SV * 615static SV *
489encode_cbor (SV *scalar, CBOR *cbor) 616encode_cbor (SV *scalar, CBOR *cbor)
490{ 617{
491 enc_t enc = { }; 618 enc_t enc = { 0 };
492 619
493 enc.cbor = *cbor; 620 enc.cbor = *cbor;
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 621 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 622 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 623 enc.end = SvEND (enc.sv);
497 624
498 SvPOK_only (enc.sv); 625 SvPOK_only (enc.sv);
626
627 if (cbor->flags & F_PACK_STRINGS)
628 {
629 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
630 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
631 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
632 }
633
499 encode_sv (&enc, scalar); 634 encode_sv (&enc, scalar);
500 635
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 636 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 637 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503 638
517 U8 *end; // end of input string 652 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 653 const char *err; // parse error, if != 0
519 CBOR cbor; 654 CBOR cbor;
520 U32 depth; // recursion depth 655 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 656 U32 maxdepth; // recursion depth limit
522 AV *sharable; 657 AV *shareable;
658 AV *stringref;
659 SV *decode_tagged;
523} dec_t; 660} dec_t;
524 661
525#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 662#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
526 663
527#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 664#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 667#define DEC_DEC_DEPTH --dec->depth
531 668
532static UV 669static UV
533decode_uint (dec_t *dec) 670decode_uint (dec_t *dec)
534{ 671{
535 switch (*dec->cur & 31) 672 U8 m = *dec->cur & MINOR_MASK;
536 { 673 ++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;
541 674
542 case 24: 675 if (ecb_expect_true (m < LENGTH_EXT1))
676 return m;
677 else if (ecb_expect_true (m == LENGTH_EXT1))
678 {
543 WANT (2); 679 WANT (1);
544 dec->cur += 2; 680 dec->cur += 1;
545 return dec->cur[-1]; 681 return dec->cur[-1];
546 682 }
547 case 25: 683 else if (ecb_expect_true (m == LENGTH_EXT2))
684 {
548 WANT (3); 685 WANT (2);
549 dec->cur += 3; 686 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 687 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 688 | ((UV)dec->cur[-1]);
552 689 }
553 case 26: 690 else if (ecb_expect_true (m == LENGTH_EXT4))
691 {
554 WANT (5); 692 WANT (4);
555 dec->cur += 5; 693 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 694 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 695 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 696 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 697 | ((UV)dec->cur[-1]);
560 698 }
561 case 27: 699 else if (ecb_expect_true (m == LENGTH_EXT8))
700 {
562 WANT (9); 701 WANT (8);
563 dec->cur += 9; 702 dec->cur += 8;
703
704 return
705#if UVSIZE < 8
706 0
707#else
564 return (((UV)dec->cur[-8]) << 56) 708 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 709 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 710 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 711 | (((UV)dec->cur[-5]) << 32)
712#endif
568 | (((UV)dec->cur[-4]) << 24) 713 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 714 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 715 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 716 | ((UV)dec->cur[-1]);
572 717 }
573 default: 718 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 719 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 720
577fail: 721fail:
578 return 0; 722 return 0;
579} 723}
580 724
585{ 729{
586 AV *av = newAV (); 730 AV *av = newAV ();
587 731
588 DEC_INC_DEPTH; 732 DEC_INC_DEPTH;
589 733
590 if ((*dec->cur & 31) == 31) 734 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 735 {
592 ++dec->cur; 736 ++dec->cur;
593 737
594 for (;;) 738 for (;;)
595 { 739 {
596 WANT (1); 740 WANT (1);
597 741
598 if (*dec->cur == (0xe0 | 31)) 742 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 743 {
600 ++dec->cur; 744 ++dec->cur;
601 break; 745 break;
602 } 746 }
603 747
606 } 750 }
607 else 751 else
608 { 752 {
609 int i, len = decode_uint (dec); 753 int i, len = decode_uint (dec);
610 754
755 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); 756 av_fill (av, len - 1);
612 757
613 for (i = 0; i < len; ++i) 758 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 759 AvARRAY (av)[i] = decode_sv (dec);
615 } 760 }
625 770
626static void 771static void
627decode_he (dec_t *dec, HV *hv) 772decode_he (dec_t *dec, HV *hv)
628{ 773{
629 // for speed reasons, we specialcase single-string 774 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 775 // byte or utf-8 strings as keys, but only when !stringref
631 776
632 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 777 if (ecb_expect_true (!dec->stringref))
778 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633 { 779 {
634 I32 len = decode_uint (dec); 780 I32 len = decode_uint (dec);
635 char *key = (char *)dec->cur; 781 char *key = (char *)dec->cur;
636 782
783 WANT (len);
637 dec->cur += len; 784 dec->cur += len;
638 785
639 hv_store (hv, key, len, decode_sv (dec), 0); 786 hv_store (hv, key, len, decode_sv (dec), 0);
787
788 return;
640 } 789 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 790 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 791 {
643 I32 len = decode_uint (dec); 792 I32 len = decode_uint (dec);
644 char *key = (char *)dec->cur; 793 char *key = (char *)dec->cur;
645 794
795 WANT (len);
646 dec->cur += len; 796 dec->cur += len;
647 797
798 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
799 if (!is_utf8_string (key, len))
800 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
801
648 hv_store (hv, key, -len, decode_sv (dec), 0); 802 hv_store (hv, key, -len, decode_sv (dec), 0);
803
804 return;
649 } 805 }
650 else 806
651 {
652 SV *k = decode_sv (dec); 807 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 808 SV *v = decode_sv (dec);
654 809
655 hv_store_ent (hv, k, v, 0); 810 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 811 SvREFCNT_dec (k);
657 } 812
813fail:
814 ;
658} 815}
659 816
660static SV * 817static SV *
661decode_hv (dec_t *dec) 818decode_hv (dec_t *dec)
662{ 819{
663 HV *hv = newHV (); 820 HV *hv = newHV ();
664 821
665 DEC_INC_DEPTH; 822 DEC_INC_DEPTH;
666 823
667 if ((*dec->cur & 31) == 31) 824 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 825 {
669 ++dec->cur; 826 ++dec->cur;
670 827
671 for (;;) 828 for (;;)
672 { 829 {
673 WANT (1); 830 WANT (1);
674 831
675 if (*dec->cur == (0xe0 | 31)) 832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
676 { 833 {
677 ++dec->cur; 834 ++dec->cur;
678 break; 835 break;
679 } 836 }
680 837
701static SV * 858static SV *
702decode_str (dec_t *dec, int utf8) 859decode_str (dec_t *dec, int utf8)
703{ 860{
704 SV *sv = 0; 861 SV *sv = 0;
705 862
706 if ((*dec->cur & 31) == 31) 863 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
707 { 864 {
865 // indefinite length strings
708 ++dec->cur; 866 ++dec->cur;
709 867
868 U8 major = *dec->cur & MAJOR_MISC;
869
710 sv = newSVpvn ("", 0); 870 sv = newSVpvn ("", 0);
711 871
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 872 for (;;)
714 { 873 {
715 WANT (1); 874 WANT (1);
716 875
717 if (*dec->cur == (0xe0 | 31)) 876 if ((*dec->cur - major) > LENGTH_EXT8)
877 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 878 {
719 ++dec->cur; 879 ++dec->cur;
720 break; 880 break;
721 } 881 }
882 else
883 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 884
723 sv_catsv (sv, decode_sv (dec)); 885 STRLEN len = decode_uint (dec);
886
887 WANT (len);
888 sv_catpvn (sv, dec->cur, len);
889 dec->cur += len;
724 } 890 }
725 } 891 }
726 else 892 else
727 { 893 {
728 STRLEN len = decode_uint (dec); 894 STRLEN len = decode_uint (dec);
729 895
730 WANT (len); 896 WANT (len);
731 sv = newSVpvn (dec->cur, len); 897 sv = newSVpvn (dec->cur, len);
732 dec->cur += len; 898 dec->cur += len;
899
900 if (ecb_expect_false (dec->stringref)
901 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
902 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733 } 903 }
734 904
735 if (utf8) 905 if (utf8)
906 {
907 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
908 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
909 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
910
736 SvUTF8_on (sv); 911 SvUTF8_on (sv);
912 }
737 913
738 return sv; 914 return sv;
739 915
740fail: 916fail:
741 SvREFCNT_dec (sv); 917 SvREFCNT_dec (sv);
743} 919}
744 920
745static SV * 921static SV *
746decode_tagged (dec_t *dec) 922decode_tagged (dec_t *dec)
747{ 923{
924 SV *sv = 0;
748 UV tag = decode_uint (dec); 925 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 926
927 WANT (1);
750 928
751 switch (tag) 929 switch (tag)
752 { 930 {
753 case CBOR_TAG_MAGIC: 931 case CBOR_TAG_MAGIC:
754 return sv; 932 sv = decode_sv (dec);
933 break;
755 934
756 case CBOR_TAG_INDIRECTION: 935 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 936 sv = newRV_noinc (decode_sv (dec));
937 break;
758 938
939 case CBOR_TAG_STRINGREF_NAMESPACE:
940 {
941 ENTER; SAVETMPS;
942
943 SAVESPTR (dec->stringref);
944 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
945
946 sv = decode_sv (dec);
947
948 FREETMPS; LEAVE;
949 }
950 break;
951
952 case CBOR_TAG_STRINGREF:
953 {
954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
955 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
956
957 UV idx = decode_uint (dec);
958
959 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
960 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
961
962 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
963 }
964 break;
965
759 case CBOR_TAG_VALUE_SHARABLE: 966 case CBOR_TAG_VALUE_SHAREABLE:
967 {
760 if (ecb_expect_false (!dec->sharable)) 968 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 969 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 970
971 if (dec->cbor.flags & F_ALLOW_CYCLES)
972 {
973 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 974 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 975
765 return sv; 976 SV *osv = decode_sv (dec);
977 sv_setsv (sv, osv);
978 SvREFCNT_dec_NN (osv);
979 }
980 else
981 {
982 av_push (dec->shareable, &PL_sv_undef);
983 int idx = AvFILLp (dec->shareable);
984 sv = decode_sv (dec);
985 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
986 }
987 }
988 break;
766 989
767 case CBOR_TAG_VALUE_SHAREDREF: 990 case CBOR_TAG_VALUE_SHAREDREF:
768 { 991 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 992 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 993 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 994
995 UV idx = decode_uint (dec);
996
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 997 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 998 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 999
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); 1000 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1001
1002 if (sv == &PL_sv_undef)
1003 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
778 } 1004 }
1005 break;
779 1006
780 case CBOR_TAG_PERL_OBJECT: 1007 case CBOR_TAG_PERL_OBJECT:
781 { 1008 {
1009 sv = decode_sv (dec);
1010
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1011 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 1012 ERR ("corrupted CBOR data (non-array perl object)");
784 1013
785 AV *av = (AV *)SvRV (sv); 1014 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 1015 int len = av_len (av) + 1;
794 if (!method) 1023 if (!method)
795 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1024 ERR ("cannot decode perl-object (package does not have a THAW method)");
796 1025
797 dSP; 1026 dSP;
798 1027
799 ENTER; SAVETMPS; PUSHMARK (SP); 1028 ENTER; SAVETMPS;
1029 PUSHMARK (SP);
800 EXTEND (SP, len + 1); 1030 EXTEND (SP, len + 1);
801 // we re-bless the reference to get overload and other niceties right 1031 // we re-bless the reference to get overload and other niceties right
802 PUSHs (*av_fetch (av, 0, 1)); 1032 PUSHs (*av_fetch (av, 0, 1));
803 PUSHs (sv_cbor); 1033 PUSHs (sv_cbor);
804 1034
821 sv = SvREFCNT_inc (POPs); 1051 sv = SvREFCNT_inc (POPs);
822 1052
823 PUTBACK; 1053 PUTBACK;
824 1054
825 FREETMPS; LEAVE; 1055 FREETMPS; LEAVE;
826
827 return sv;
828 } 1056 }
1057 break;
829 1058
830 default: 1059 default:
831 { 1060 {
1061 sv = decode_sv (dec);
1062
1063 dSP;
1064 ENTER; SAVETMPS;
1065 SAVESTACK_POS ();
1066 PUSHMARK (SP);
1067 EXTEND (SP, 2);
1068 PUSHs (newSVuv (tag));
1069 PUSHs (sv);
1070
1071 PUTBACK;
1072 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1073 SPAGAIN;
1074
1075 if (SvTRUE (ERRSV))
1076 {
1077 FREETMPS; LEAVE;
1078 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1079 }
1080
1081 if (count)
1082 {
1083 SvREFCNT_dec (sv);
1084 sv = SvREFCNT_inc (POPs);
1085 }
1086 else
1087 {
832 AV *av = newAV (); 1088 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1089 av_push (av, newSVuv (tag));
834 av_push (av, sv); 1090 av_push (av, sv);
835 1091
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1092 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1093 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1094 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1095 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1096 }
1097
1098 PUTBACK;
1099
1100 FREETMPS; LEAVE;
1101 }
1102 break;
842 } 1103 }
1104
1105 return sv;
843 1106
844fail: 1107fail:
845 SvREFCNT_dec (sv); 1108 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1109 return &PL_sv_undef;
847} 1110}
849static SV * 1112static SV *
850decode_sv (dec_t *dec) 1113decode_sv (dec_t *dec)
851{ 1114{
852 WANT (1); 1115 WANT (1);
853 1116
854 switch (*dec->cur >> 5) 1117 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1118 {
856 case 0: // unsigned int 1119 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1120 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1121 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1122 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1123 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1124 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1125 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1126
864 case 4: // array 1127 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) 1128 switch (*dec->cur++ & MINOR_MASK)
872 { 1129 {
873 case 20: 1130 case SIMPLE_FALSE:
874#if CBOR_SLOW 1131#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1132 types_false = get_bool ("Types::Serialiser::false");
876#endif 1133#endif
877 return newSVsv (types_false); 1134 return newSVsv (types_false);
878 case 21: 1135 case SIMPLE_TRUE:
879#if CBOR_SLOW 1136#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1137 types_true = get_bool ("Types::Serialiser::true");
881#endif 1138#endif
882 return newSVsv (types_true); 1139 return newSVsv (types_true);
883 case 22: 1140 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1141 return newSVsv (&PL_sv_undef);
885 case 23: 1142 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1143#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1144 types_error = get_bool ("Types::Serialiser::error");
888#endif 1145#endif
889 return newSVsv (types_error); 1146 return newSVsv (types_error);
890 1147
891 case 25: 1148 case MISC_FLOAT16:
892 { 1149 {
893 WANT (2); 1150 WANT (2);
894 1151
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1152 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1153 dec->cur += 2;
897 1154
898 return newSVnv (ecb_binary16_to_float (fp)); 1155 return newSVnv (ecb_binary16_to_float (fp));
899 } 1156 }
900 1157
901 case 26: 1158 case MISC_FLOAT32:
902 { 1159 {
903 uint32_t fp; 1160 uint32_t fp;
904 WANT (4); 1161 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1162 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1163 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1166 fp = ecb_bswap32 (fp);
910 1167
911 return newSVnv (ecb_binary32_to_float (fp)); 1168 return newSVnv (ecb_binary32_to_float (fp));
912 } 1169 }
913 1170
914 case 27: 1171 case MISC_FLOAT64:
915 { 1172 {
916 uint64_t fp; 1173 uint64_t fp;
917 WANT (8); 1174 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1175 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1176 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1179 fp = ecb_bswap64 (fp);
923 1180
924 return newSVnv (ecb_binary64_to_double (fp)); 1181 return newSVnv (ecb_binary64_to_double (fp));
925 } 1182 }
926 1183
927 // 0..19 unassigned 1184 // 0..19 unassigned simple
928 // 24 reserved + unassigned (reserved values are not encodable) 1185 // 24 reserved + unassigned simple (reserved values are not encodable)
1186 // 28-30 unassigned misc
1187 // 31 break code
929 default: 1188 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1189 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1190 }
932 1191
933 break; 1192 break;
934 } 1193 }
935 1194
938} 1197}
939 1198
940static SV * 1199static SV *
941decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1200decode_cbor (SV *string, CBOR *cbor, char **offset_return)
942{ 1201{
943 dec_t dec = { }; 1202 dec_t dec = { 0 };
944 SV *sv; 1203 SV *sv;
945 STRLEN len; 1204 STRLEN len;
946 char *data = SvPVbyte (string, len); 1205 char *data = SvPVbyte (string, len);
947 1206
948 if (len > cbor->max_size && cbor->max_size) 1207 if (len > cbor->max_size && cbor->max_size)
962 if (dec.cur != dec.end && !dec.err) 1221 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1222 dec.err = "garbage after CBOR object";
964 1223
965 if (dec.err) 1224 if (dec.err)
966 { 1225 {
1226 if (dec.shareable)
1227 {
1228 // need to break cyclic links, which whould all be in shareable
1229 int i;
1230 SV **svp;
1231
1232 for (i = av_len (dec.shareable) + 1; i--; )
1233 if ((svp = av_fetch (dec.shareable, i, 0)))
1234 sv_setsv (*svp, &PL_sv_undef);
1235 }
1236
967 SvREFCNT_dec (sv); 1237 SvREFCNT_dec (sv);
968 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1238 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
969 } 1239 }
970 1240
971 sv = sv_2mortal (sv); 1241 sv = sv_2mortal (sv);
972 1242
973 return sv; 1243 return sv;
974} 1244}
975 1245
1246/////////////////////////////////////////////////////////////////////////////
1247// incremental parser
1248
1249#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1250
1251// returns 0 for notyet, 1 for success or error
1252static int
1253incr_parse (CBOR *self, SV *cborstr)
1254{
1255 STRLEN cur;
1256 SvPV (cborstr, cur);
1257
1258 while (ecb_expect_true (self->incr_need <= cur))
1259 {
1260 // table of integer count bytes
1261 static I8 incr_len[MINOR_MASK + 1] = {
1262 0, 0, 0, 0, 0, 0, 0, 0,
1263 0, 0, 0, 0, 0, 0, 0, 0,
1264 0, 0, 0, 0, 0, 0, 0, 0,
1265 1, 2, 4, 8,-1,-1,-1,-2
1266 };
1267
1268 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1269 U8 m = *p & MINOR_MASK;
1270 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1271 I8 ilen = incr_len[m];
1272
1273 self->incr_need = self->incr_pos + 1;
1274
1275 if (ecb_expect_false (ilen < 0))
1276 {
1277 if (m != MINOR_INDEF)
1278 return 1; // error
1279
1280 if (*p == (MAJOR_MISC | MINOR_INDEF))
1281 {
1282 if (count >= 0)
1283 return 1; // error
1284
1285 count = 1;
1286 }
1287 else
1288 {
1289 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1290 count = -1;
1291 }
1292 }
1293 else
1294 {
1295 self->incr_need += ilen;
1296 if (ecb_expect_false (self->incr_need > cur))
1297 return 0;
1298
1299 int major = *p >> MAJOR_SHIFT;
1300
1301 switch (major)
1302 {
1303 case MAJOR_TAG >> MAJOR_SHIFT:
1304 ++count; // tags merely prefix another value
1305 break;
1306
1307 case MAJOR_BYTES >> MAJOR_SHIFT:
1308 case MAJOR_TEXT >> MAJOR_SHIFT:
1309 case MAJOR_ARRAY >> MAJOR_SHIFT:
1310 case MAJOR_MAP >> MAJOR_SHIFT:
1311 {
1312 UV len;
1313
1314 if (ecb_expect_false (ilen))
1315 {
1316 len = 0;
1317
1318 do {
1319 len = (len << 8) | *++p;
1320 } while (--ilen);
1321 }
1322 else
1323 len = m;
1324
1325 switch (major)
1326 {
1327 case MAJOR_BYTES >> MAJOR_SHIFT:
1328 case MAJOR_TEXT >> MAJOR_SHIFT:
1329 self->incr_need += len;
1330 if (ecb_expect_false (self->incr_need > cur))
1331 return 0;
1332
1333 break;
1334
1335 case MAJOR_MAP >> MAJOR_SHIFT:
1336 len <<= 1;
1337 case MAJOR_ARRAY >> MAJOR_SHIFT:
1338 if (len)
1339 {
1340 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1341 count = len + 1;
1342 }
1343 break;
1344 }
1345 }
1346 }
1347 }
1348
1349 self->incr_pos = self->incr_need;
1350
1351 if (count > 0)
1352 {
1353 while (!--count)
1354 {
1355 if (!AvFILLp (self->incr_count))
1356 return 1; // done
1357
1358 SvREFCNT_dec_NN (av_pop (self->incr_count));
1359 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1360 }
1361
1362 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1363 }
1364 }
1365
1366 return 0;
1367}
1368
1369
976///////////////////////////////////////////////////////////////////////////// 1370/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1371// XS interface functions
978 1372
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1373MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1374
987 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1381 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
988 1382
989 types_true = get_bool ("Types::Serialiser::true" ); 1383 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1384 types_false = get_bool ("Types::Serialiser::false");
991 types_error = get_bool ("Types::Serialiser::error"); 1385 types_error = get_bool ("Types::Serialiser::error");
1386
1387 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
992 1388
993 sv_cbor = newSVpv ("CBOR", 0); 1389 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1390 SvREADONLY_on (sv_cbor);
995} 1391}
996 1392
1018void shrink (CBOR *self, int enable = 1) 1414void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1415 ALIAS:
1020 shrink = F_SHRINK 1416 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1417 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1418 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1419 allow_cycles = F_ALLOW_CYCLES
1024 dedup_strings = F_DEDUP_STRINGS 1420 pack_strings = F_PACK_STRINGS
1421 utf8_strings = F_UTF8_STRINGS
1422 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1423 PPCODE:
1026{ 1424{
1027 if (enable) 1425 if (enable)
1028 self->flags |= ix; 1426 self->flags |= ix;
1029 else 1427 else
1035void get_shrink (CBOR *self) 1433void get_shrink (CBOR *self)
1036 ALIAS: 1434 ALIAS:
1037 get_shrink = F_SHRINK 1435 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1436 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1437 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1438 get_allow_cycles = F_ALLOW_CYCLES
1041 get_dedup_strings = F_DEDUP_STRINGS 1439 get_pack_strings = F_PACK_STRINGS
1440 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1441 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1442 XPUSHs (boolSV (self->flags & ix));
1044 1443
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1444void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1445 PPCODE:
1059 XPUSHs (ST (0)); 1458 XPUSHs (ST (0));
1060 1459
1061int get_max_size (CBOR *self) 1460int get_max_size (CBOR *self)
1062 CODE: 1461 CODE:
1063 RETVAL = self->max_size; 1462 RETVAL = self->max_size;
1463 OUTPUT:
1464 RETVAL
1465
1466void filter (CBOR *self, SV *filter = 0)
1467 PPCODE:
1468 SvREFCNT_dec (self->filter);
1469 self->filter = filter ? newSVsv (filter) : filter;
1470 XPUSHs (ST (0));
1471
1472SV *get_filter (CBOR *self)
1473 CODE:
1474 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064 OUTPUT: 1475 OUTPUT:
1065 RETVAL 1476 RETVAL
1066 1477
1067void encode (CBOR *self, SV *scalar) 1478void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1479 PPCODE:
1083 EXTEND (SP, 2); 1494 EXTEND (SP, 2);
1084 PUSHs (sv); 1495 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1496 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1497}
1087 1498
1499void incr_parse (CBOR *self, SV *cborstr)
1500 ALIAS:
1501 incr_parse_multiple = 1
1502 PPCODE:
1503{
1504 if (SvUTF8 (cborstr))
1505 sv_utf8_downgrade (cborstr, 0);
1506
1507 if (!self->incr_count)
1508 {
1509 self->incr_count = newAV ();
1510 self->incr_pos = 0;
1511 self->incr_need = 1;
1512
1513 av_push (self->incr_count, newSViv (1));
1514 }
1515
1516 do
1517 {
1518 if (!incr_parse (self, cborstr))
1519 {
1520 if (self->incr_need > self->max_size && self->max_size)
1521 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1522 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1523
1524 break;
1525 }
1526
1527 SV *sv;
1528 char *offset;
1529
1530 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1531 XPUSHs (sv);
1532
1533 sv_chop (cborstr, offset);
1534
1535 av_clear (self->incr_count);
1536 av_push (self->incr_count, newSViv (1));
1537
1538 self->incr_pos = 0;
1539 self->incr_need = self->incr_pos + 1;
1540 }
1541 while (ix);
1542}
1543
1544void incr_reset (CBOR *self)
1545 CODE:
1546{
1547 SvREFCNT_dec (self->incr_count);
1548 self->incr_count = 0;
1549}
1550
1551void DESTROY (CBOR *self)
1552 PPCODE:
1553 cbor_free (self);
1554
1088PROTOTYPES: ENABLE 1555PROTOTYPES: ENABLE
1089 1556
1090void encode_cbor (SV *scalar) 1557void encode_cbor (SV *scalar)
1558 ALIAS:
1559 encode_cbor = 0
1560 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1561 PPCODE:
1092{ 1562{
1093 CBOR cbor; 1563 CBOR cbor;
1094 cbor_init (&cbor); 1564 cbor_init (&cbor);
1565 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1566 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1567 XPUSHs (scalar);
1097} 1568}
1098 1569
1099void decode_cbor (SV *cborstr) 1570void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines