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 1.50 by root, Thu Feb 25 02:29:22 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_VALIDATE_UTF8 0x00000020UL
64 106
65#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
66 108
67#define SB do { 109#define SB do {
68#define SE } while (0) 110#define SE } while (0)
80# define CBOR_SLOW 0 122# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
82#endif 124#endif
83 125
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 128
87typedef struct { 129typedef struct {
88 U32 flags; 130 U32 flags;
89 U32 max_depth; 131 U32 max_depth;
90 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
134
135 // for the incremental parser
136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 139} CBOR;
92 140
93ecb_inline void 141ecb_inline void
94cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
95{ 143{
96 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 145 cbor->max_depth = 512;
146}
147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
98} 153}
99 154
100///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
101// utility functions 156// utility functions
102 157
124 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 180#endif
126 } 181 }
127} 182}
128 183
184// minimum length of a string to be registered for stringref
185ecb_inline int
186minimum_string_length (UV idx)
187{
188 return idx > 23
189 ? idx > 0xffU
190 ? idx > 0xffffU
191 ? idx > 0xffffffffU
192 ? 11
193 : 7
194 : 5
195 : 4
196 : 3;
197}
198
129///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
130// encoder 200// encoder
131 201
132// structure used for encoding CBOR 202// structure used for encoding CBOR
133typedef struct 203typedef struct
135 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
137 SV *sv; // result scalar 207 SV *sv; // result scalar
138 CBOR cbor; 208 CBOR cbor;
139 U32 depth; // recursion level 209 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 210 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
211 UV stringref_idx;
141 HV *sharable; // ptr => index, or 0 212 HV *shareable; // ptr => index, or 0
142 HV *sharable_idx; 213 UV shareable_idx;
143} enc_t; 214} enc_t;
144 215
145ecb_inline void 216ecb_inline void
146need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
147{ 218{
164static void 235static void
165encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
166{ 237{
167 need (enc, 9); 238 need (enc, 9);
168 239
169 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
171 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
172 { 243 {
173 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 245 *enc->cur++ = len;
175 } 246 }
176 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
177 { 248 {
178 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 251 *enc->cur++ = len;
181 } 252 }
182 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
183 { 254 {
184 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 259 *enc->cur++ = len;
189 } 260 }
190 else 261 else
191 { 262 {
192 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 271 *enc->cur++ = len;
201 } 272 }
202} 273}
203 274
204static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 283{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 285 need (enc, len);
209 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
210 enc->cur += len; 287 enc->cur += len;
211} 288}
212 289
213ecb_inline void
214encode_tag (enc_t *enc, UV tag)
215{
216 encode_uint (enc, 0xc0, tag);
217}
218
219static int 290static void
220encode_sharable2 (enc_t *enc, SV *sv) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
221{ 292{
222 if (!enc->sharable) 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
223 enc->sharable = (HV *)sv_2mortal ((SV *)newHV ()); 294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
224 296
225 SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227 if (SvOK (*svp)) 297 if (SvOK (*svp))
228 { 298 {
299 // already registered, use stringref
229 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
230 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232 return 1; 302 return;
233 } 303 }
234 else 304 else if (len >= minimum_string_length (enc->stringref_idx))
235 { 305 {
306 // register only
236 sv_setuv (*svp, enc->sharable_idx++); 307 sv_setuv (*svp, enc->stringref_idx);
237 encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); 308 ++enc->stringref_idx;
238 309 }
239 return 0;
240 } 310 }
241}
242 311
243ecb_inline int 312 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} 313}
252 314
253static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
254 316
255static void 317static void
260 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
262 324
263 ++enc->depth; 325 ++enc->depth;
264 326
265 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 328
329 if (SvMAGICAL (av))
267 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
268 { 331 {
269 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
272 341
273 --enc->depth; 342 --enc->depth;
274} 343}
275 344
276static void 345static void
285 354
286 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
288 357
289 if (mg) 358 if (mg)
290 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 360 else
292 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
293 362
294 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
295 { 364 {
296 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
298 else 367 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 369
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 371 }
303 372
304 if (mg) 373 if (mg)
305 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 375
307 --enc->depth; 376 --enc->depth;
308} 377}
309 378
310// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
311static void 380static void
312encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
313{ 382{
314 svtype svt;
315
316 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
317 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 386
322 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
323 { 388 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 390 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 396 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 398
334 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
335 GV *method;
336 400
337 if (stash == boolean_stash) 401 if (stash == boolean_stash)
338 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
339 else if (stash == error_stash) 406 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
341 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
342 { 412 {
343 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 415
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
348 } 432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
436 }
437 else
438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 451 {
351 dSP; 452 dSP;
352 453
353 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS;
455 PUSHMARK (SP);
354 // we re-bless the reference to get overload and other niceties right 456 // we re-bless the reference to get overload and other niceties right
355 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 457 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
356 458
357 PUTBACK; 459 PUTBACK;
358 // G_SCALAR ensures that return value is 1 460 // G_SCALAR ensures that return value is 1
371 } 473 }
372 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 474 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
373 { 475 {
374 dSP; 476 dSP;
375 477
376 ENTER; SAVETMPS; PUSHMARK (SP); 478 ENTER; SAVETMPS;
479 SAVESTACK_POS ();
480 PUSHMARK (SP);
377 EXTEND (SP, 2); 481 EXTEND (SP, 2);
378 // 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
379 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 483 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
380 PUSHs (sv_cbor); 484 PUSHs (sv_cbor);
381 485
386 // catch this surprisingly common error 490 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 491 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)); 492 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389 493
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 494 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 495 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 496 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 497
394 while (count) 498 while (count)
395 encode_sv (enc, SP[1 - count--]); 499 encode_sv (enc, SP[1 - count--]);
396 500
397 PUTBACK; 501 PUTBACK;
418{ 522{
419 double nv = SvNVX (sv); 523 double nv = SvNVX (sv);
420 524
421 need (enc, 9); 525 need (enc, 9);
422 526
423 if (ecb_expect_false (nv == (U32)nv)) 527 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 528 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 529 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 530 else if (ecb_expect_false (nv == (float)nv))
427 { 531 {
428 uint32_t fp = ecb_float_to_binary32 (nv); 532 uint32_t fp = ecb_float_to_binary32 (nv);
429 533
430 *enc->cur++ = 0xe0 | 26; 534 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
431 535
432 if (!ecb_big_endian ()) 536 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp); 537 fp = ecb_bswap32 (fp);
434 538
435 memcpy (enc->cur, &fp, 4); 539 memcpy (enc->cur, &fp, 4);
437 } 541 }
438 else 542 else
439 { 543 {
440 uint64_t fp = ecb_double_to_binary64 (nv); 544 uint64_t fp = ecb_double_to_binary64 (nv);
441 545
442 *enc->cur++ = 0xe0 | 27; 546 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
443 547
444 if (!ecb_big_endian ()) 548 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp); 549 fp = ecb_bswap64 (fp);
446 550
447 memcpy (enc->cur, &fp, 8); 551 memcpy (enc->cur, &fp, 8);
452static void 556static void
453encode_sv (enc_t *enc, SV *sv) 557encode_sv (enc_t *enc, SV *sv)
454{ 558{
455 SvGETMAGIC (sv); 559 SvGETMAGIC (sv);
456 560
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 561 if (SvPOKp (sv))
461 { 562 {
462 STRLEN len; 563 STRLEN len;
463 char *str = SvPV (sv, len); 564 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 565 encode_strref (enc, SvUTF8 (sv), str, len);
465 } 566 }
466 else if (SvNOKp (sv)) 567 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 568 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 569 else if (SvIOKp (sv))
469 { 570 {
470 if (SvIsUV (sv)) 571 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv)); 572 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
472 else if (SvIVX (sv) >= 0) 573 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv)); 574 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
474 else 575 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 576 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
476 } 577 }
477 else if (SvROK (sv)) 578 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 579 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 580 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 581 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 582 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 583 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 584 else
484 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 585 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 586 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 587}
487 588
488static SV * 589static SV *
489encode_cbor (SV *scalar, CBOR *cbor) 590encode_cbor (SV *scalar, CBOR *cbor)
490{ 591{
491 enc_t enc = { }; 592 enc_t enc = { 0 };
492 593
493 enc.cbor = *cbor; 594 enc.cbor = *cbor;
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 595 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 596 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 597 enc.end = SvEND (enc.sv);
497 598
498 SvPOK_only (enc.sv); 599 SvPOK_only (enc.sv);
600
601 if (cbor->flags & F_PACK_STRINGS)
602 {
603 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
604 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
605 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
606 }
607
499 encode_sv (&enc, scalar); 608 encode_sv (&enc, scalar);
500 609
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 610 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 611 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503 612
517 U8 *end; // end of input string 626 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 627 const char *err; // parse error, if != 0
519 CBOR cbor; 628 CBOR cbor;
520 U32 depth; // recursion depth 629 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 630 U32 maxdepth; // recursion depth limit
522 AV *sharable; 631 AV *shareable;
632 AV *stringref;
633 SV *decode_tagged;
523} dec_t; 634} dec_t;
524 635
525#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 636#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
526 637
527#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 638#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 641#define DEC_DEC_DEPTH --dec->depth
531 642
532static UV 643static UV
533decode_uint (dec_t *dec) 644decode_uint (dec_t *dec)
534{ 645{
535 switch (*dec->cur & 31) 646 U8 m = *dec->cur & MINOR_MASK;
536 { 647 ++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;
541 648
542 case 24: 649 if (ecb_expect_true (m < LENGTH_EXT1))
650 return m;
651 else if (ecb_expect_true (m == LENGTH_EXT1))
652 {
543 WANT (2); 653 WANT (1);
544 dec->cur += 2; 654 dec->cur += 1;
545 return dec->cur[-1]; 655 return dec->cur[-1];
546 656 }
547 case 25: 657 else if (ecb_expect_true (m == LENGTH_EXT2))
658 {
548 WANT (3); 659 WANT (2);
549 dec->cur += 3; 660 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 661 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 662 | ((UV)dec->cur[-1]);
552 663 }
553 case 26: 664 else if (ecb_expect_true (m == LENGTH_EXT4))
665 {
554 WANT (5); 666 WANT (4);
555 dec->cur += 5; 667 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 668 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
560 672 }
561 case 27: 673 else if (ecb_expect_true (m == LENGTH_EXT8))
674 {
562 WANT (9); 675 WANT (8);
563 dec->cur += 9; 676 dec->cur += 8;
677
678 return
679#if UVSIZE < 8
680 0
681#else
564 return (((UV)dec->cur[-8]) << 56) 682 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 683 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 684 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 685 | (((UV)dec->cur[-5]) << 32)
686#endif
568 | (((UV)dec->cur[-4]) << 24) 687 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 688 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 689 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 690 | ((UV)dec->cur[-1]);
572 691 }
573 default: 692 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 693 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 694
577fail: 695fail:
578 return 0; 696 return 0;
579} 697}
580 698
585{ 703{
586 AV *av = newAV (); 704 AV *av = newAV ();
587 705
588 DEC_INC_DEPTH; 706 DEC_INC_DEPTH;
589 707
590 if ((*dec->cur & 31) == 31) 708 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 709 {
592 ++dec->cur; 710 ++dec->cur;
593 711
594 for (;;) 712 for (;;)
595 { 713 {
596 WANT (1); 714 WANT (1);
597 715
598 if (*dec->cur == (0xe0 | 31)) 716 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 717 {
600 ++dec->cur; 718 ++dec->cur;
601 break; 719 break;
602 } 720 }
603 721
606 } 724 }
607 else 725 else
608 { 726 {
609 int i, len = decode_uint (dec); 727 int i, len = decode_uint (dec);
610 728
729 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); 730 av_fill (av, len - 1);
612 731
613 for (i = 0; i < len; ++i) 732 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 733 AvARRAY (av)[i] = decode_sv (dec);
615 } 734 }
625 744
626static void 745static void
627decode_he (dec_t *dec, HV *hv) 746decode_he (dec_t *dec, HV *hv)
628{ 747{
629 // for speed reasons, we specialcase single-string 748 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 749 // byte or utf-8 strings as keys, but only when !stringref
631 750
632 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 751 if (ecb_expect_true (!dec->stringref))
752 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633 { 753 {
634 I32 len = decode_uint (dec); 754 I32 len = decode_uint (dec);
635 char *key = (char *)dec->cur; 755 char *key = (char *)dec->cur;
636 756
757 WANT (len);
637 dec->cur += len; 758 dec->cur += len;
638 759
639 hv_store (hv, key, len, decode_sv (dec), 0); 760 hv_store (hv, key, len, decode_sv (dec), 0);
761
762 return;
640 } 763 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 764 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 765 {
643 I32 len = decode_uint (dec); 766 I32 len = decode_uint (dec);
644 char *key = (char *)dec->cur; 767 char *key = (char *)dec->cur;
645 768
769 WANT (len);
646 dec->cur += len; 770 dec->cur += len;
647 771
772 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
773 if (!is_utf8_string (key, len))
774 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
775
648 hv_store (hv, key, -len, decode_sv (dec), 0); 776 hv_store (hv, key, -len, decode_sv (dec), 0);
777
778 return;
649 } 779 }
650 else 780
651 {
652 SV *k = decode_sv (dec); 781 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 782 SV *v = decode_sv (dec);
654 783
655 hv_store_ent (hv, k, v, 0); 784 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 785 SvREFCNT_dec (k);
657 } 786
787fail:
788 ;
658} 789}
659 790
660static SV * 791static SV *
661decode_hv (dec_t *dec) 792decode_hv (dec_t *dec)
662{ 793{
663 HV *hv = newHV (); 794 HV *hv = newHV ();
664 795
665 DEC_INC_DEPTH; 796 DEC_INC_DEPTH;
666 797
667 if ((*dec->cur & 31) == 31) 798 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 799 {
669 ++dec->cur; 800 ++dec->cur;
670 801
671 for (;;) 802 for (;;)
672 { 803 {
673 WANT (1); 804 WANT (1);
674 805
675 if (*dec->cur == (0xe0 | 31)) 806 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
676 { 807 {
677 ++dec->cur; 808 ++dec->cur;
678 break; 809 break;
679 } 810 }
680 811
701static SV * 832static SV *
702decode_str (dec_t *dec, int utf8) 833decode_str (dec_t *dec, int utf8)
703{ 834{
704 SV *sv = 0; 835 SV *sv = 0;
705 836
706 if ((*dec->cur & 31) == 31) 837 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
707 { 838 {
839 // indefinite length strings
708 ++dec->cur; 840 ++dec->cur;
709 841
842 U8 major = *dec->cur & MAJOR_MISC;
843
710 sv = newSVpvn ("", 0); 844 sv = newSVpvn ("", 0);
711 845
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 846 for (;;)
714 { 847 {
715 WANT (1); 848 WANT (1);
716 849
717 if (*dec->cur == (0xe0 | 31)) 850 if ((*dec->cur - major) > LENGTH_EXT8)
851 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 852 {
719 ++dec->cur; 853 ++dec->cur;
720 break; 854 break;
721 } 855 }
856 else
857 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 858
723 sv_catsv (sv, decode_sv (dec)); 859 STRLEN len = decode_uint (dec);
860
861 WANT (len);
862 sv_catpvn (sv, dec->cur, len);
863 dec->cur += len;
724 } 864 }
725 } 865 }
726 else 866 else
727 { 867 {
728 STRLEN len = decode_uint (dec); 868 STRLEN len = decode_uint (dec);
729 869
730 WANT (len); 870 WANT (len);
731 sv = newSVpvn (dec->cur, len); 871 sv = newSVpvn (dec->cur, len);
732 dec->cur += len; 872 dec->cur += len;
873
874 if (ecb_expect_false (dec->stringref)
875 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
876 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733 } 877 }
734 878
735 if (utf8) 879 if (utf8)
880 {
881 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
882 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
883 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
884
736 SvUTF8_on (sv); 885 SvUTF8_on (sv);
886 }
737 887
738 return sv; 888 return sv;
739 889
740fail: 890fail:
741 SvREFCNT_dec (sv); 891 SvREFCNT_dec (sv);
743} 893}
744 894
745static SV * 895static SV *
746decode_tagged (dec_t *dec) 896decode_tagged (dec_t *dec)
747{ 897{
898 SV *sv = 0;
748 UV tag = decode_uint (dec); 899 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 900
901 WANT (1);
750 902
751 switch (tag) 903 switch (tag)
752 { 904 {
753 case CBOR_TAG_MAGIC: 905 case CBOR_TAG_MAGIC:
754 return sv; 906 sv = decode_sv (dec);
907 break;
755 908
756 case CBOR_TAG_INDIRECTION: 909 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 910 sv = newRV_noinc (decode_sv (dec));
911 break;
758 912
913 case CBOR_TAG_STRINGREF_NAMESPACE:
914 {
915 ENTER; SAVETMPS;
916
917 SAVESPTR (dec->stringref);
918 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
919
920 sv = decode_sv (dec);
921
922 FREETMPS; LEAVE;
923 }
924 break;
925
926 case CBOR_TAG_STRINGREF:
927 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
930
931 UV idx = decode_uint (dec);
932
933 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
934 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
935
936 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
937 }
938 break;
939
759 case CBOR_TAG_VALUE_SHARABLE: 940 case CBOR_TAG_VALUE_SHAREABLE:
941 {
760 if (ecb_expect_false (!dec->sharable)) 942 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 943 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 944
945 if (dec->cbor.flags & F_ALLOW_CYCLES)
946 {
947 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 948 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 949
765 return sv; 950 SV *osv = decode_sv (dec);
951 sv_setsv (sv, osv);
952 SvREFCNT_dec_NN (osv);
953 }
954 else
955 {
956 av_push (dec->shareable, &PL_sv_undef);
957 int idx = AvFILLp (dec->shareable);
958 sv = decode_sv (dec);
959 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
960 }
961 }
962 break;
766 963
767 case CBOR_TAG_VALUE_SHAREDREF: 964 case CBOR_TAG_VALUE_SHAREDREF:
768 { 965 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 966 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 967 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 968
969 UV idx = decode_uint (dec);
970
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 971 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 972 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 973
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); 974 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
975
976 if (sv == &PL_sv_undef)
977 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
778 } 978 }
979 break;
779 980
780 case CBOR_TAG_PERL_OBJECT: 981 case CBOR_TAG_PERL_OBJECT:
781 { 982 {
983 sv = decode_sv (dec);
984
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 985 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 986 ERR ("corrupted CBOR data (non-array perl object)");
784 987
785 AV *av = (AV *)SvRV (sv); 988 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 989 int len = av_len (av) + 1;
794 if (!method) 997 if (!method)
795 ERR ("cannot decode perl-object (package does not have a THAW method)"); 998 ERR ("cannot decode perl-object (package does not have a THAW method)");
796 999
797 dSP; 1000 dSP;
798 1001
799 ENTER; SAVETMPS; PUSHMARK (SP); 1002 ENTER; SAVETMPS;
1003 PUSHMARK (SP);
800 EXTEND (SP, len + 1); 1004 EXTEND (SP, len + 1);
801 // we re-bless the reference to get overload and other niceties right 1005 // we re-bless the reference to get overload and other niceties right
802 PUSHs (*av_fetch (av, 0, 1)); 1006 PUSHs (*av_fetch (av, 0, 1));
803 PUSHs (sv_cbor); 1007 PUSHs (sv_cbor);
804 1008
821 sv = SvREFCNT_inc (POPs); 1025 sv = SvREFCNT_inc (POPs);
822 1026
823 PUTBACK; 1027 PUTBACK;
824 1028
825 FREETMPS; LEAVE; 1029 FREETMPS; LEAVE;
826
827 return sv;
828 } 1030 }
1031 break;
829 1032
830 default: 1033 default:
831 { 1034 {
1035 sv = decode_sv (dec);
1036
1037 dSP;
1038 ENTER; SAVETMPS;
1039 SAVESTACK_POS ();
1040 PUSHMARK (SP);
1041 EXTEND (SP, 2);
1042 PUSHs (newSVuv (tag));
1043 PUSHs (sv);
1044
1045 PUTBACK;
1046 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1047 SPAGAIN;
1048
1049 if (SvTRUE (ERRSV))
1050 {
1051 FREETMPS; LEAVE;
1052 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1053 }
1054
1055 if (count)
1056 {
1057 SvREFCNT_dec (sv);
1058 sv = SvREFCNT_inc (POPs);
1059 }
1060 else
1061 {
832 AV *av = newAV (); 1062 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1063 av_push (av, newSVuv (tag));
834 av_push (av, sv); 1064 av_push (av, sv);
835 1065
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1066 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1067 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1068 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1069 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1070 }
1071
1072 PUTBACK;
1073
1074 FREETMPS; LEAVE;
1075 }
1076 break;
842 } 1077 }
1078
1079 return sv;
843 1080
844fail: 1081fail:
845 SvREFCNT_dec (sv); 1082 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1083 return &PL_sv_undef;
847} 1084}
849static SV * 1086static SV *
850decode_sv (dec_t *dec) 1087decode_sv (dec_t *dec)
851{ 1088{
852 WANT (1); 1089 WANT (1);
853 1090
854 switch (*dec->cur >> 5) 1091 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1092 {
856 case 0: // unsigned int 1093 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1094 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1095 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1096 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1097 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1098 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1099 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1100
864 case 4: // array 1101 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) 1102 switch (*dec->cur++ & MINOR_MASK)
872 { 1103 {
873 case 20: 1104 case SIMPLE_FALSE:
874#if CBOR_SLOW 1105#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1106 types_false = get_bool ("Types::Serialiser::false");
876#endif 1107#endif
877 return newSVsv (types_false); 1108 return newSVsv (types_false);
878 case 21: 1109 case SIMPLE_TRUE:
879#if CBOR_SLOW 1110#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1111 types_true = get_bool ("Types::Serialiser::true");
881#endif 1112#endif
882 return newSVsv (types_true); 1113 return newSVsv (types_true);
883 case 22: 1114 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1115 return newSVsv (&PL_sv_undef);
885 case 23: 1116 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1117#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1118 types_error = get_bool ("Types::Serialiser::error");
888#endif 1119#endif
889 return newSVsv (types_error); 1120 return newSVsv (types_error);
890 1121
891 case 25: 1122 case MISC_FLOAT16:
892 { 1123 {
893 WANT (2); 1124 WANT (2);
894 1125
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1126 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1127 dec->cur += 2;
897 1128
898 return newSVnv (ecb_binary16_to_float (fp)); 1129 return newSVnv (ecb_binary16_to_float (fp));
899 } 1130 }
900 1131
901 case 26: 1132 case MISC_FLOAT32:
902 { 1133 {
903 uint32_t fp; 1134 uint32_t fp;
904 WANT (4); 1135 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1136 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1137 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1140 fp = ecb_bswap32 (fp);
910 1141
911 return newSVnv (ecb_binary32_to_float (fp)); 1142 return newSVnv (ecb_binary32_to_float (fp));
912 } 1143 }
913 1144
914 case 27: 1145 case MISC_FLOAT64:
915 { 1146 {
916 uint64_t fp; 1147 uint64_t fp;
917 WANT (8); 1148 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1149 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1150 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1153 fp = ecb_bswap64 (fp);
923 1154
924 return newSVnv (ecb_binary64_to_double (fp)); 1155 return newSVnv (ecb_binary64_to_double (fp));
925 } 1156 }
926 1157
927 // 0..19 unassigned 1158 // 0..19 unassigned simple
928 // 24 reserved + unassigned (reserved values are not encodable) 1159 // 24 reserved + unassigned simple (reserved values are not encodable)
1160 // 28-30 unassigned misc
1161 // 31 break code
929 default: 1162 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1163 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1164 }
932 1165
933 break; 1166 break;
934 } 1167 }
935 1168
938} 1171}
939 1172
940static SV * 1173static SV *
941decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1174decode_cbor (SV *string, CBOR *cbor, char **offset_return)
942{ 1175{
943 dec_t dec = { }; 1176 dec_t dec = { 0 };
944 SV *sv; 1177 SV *sv;
945 STRLEN len; 1178 STRLEN len;
946 char *data = SvPVbyte (string, len); 1179 char *data = SvPVbyte (string, len);
947 1180
948 if (len > cbor->max_size && cbor->max_size) 1181 if (len > cbor->max_size && cbor->max_size)
962 if (dec.cur != dec.end && !dec.err) 1195 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1196 dec.err = "garbage after CBOR object";
964 1197
965 if (dec.err) 1198 if (dec.err)
966 { 1199 {
1200 if (dec.shareable)
1201 {
1202 // need to break cyclic links, which whould all be in shareable
1203 int i;
1204 SV **svp;
1205
1206 for (i = av_len (dec.shareable) + 1; i--; )
1207 if ((svp = av_fetch (dec.shareable, i, 0)))
1208 sv_setsv (*svp, &PL_sv_undef);
1209 }
1210
967 SvREFCNT_dec (sv); 1211 SvREFCNT_dec (sv);
968 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1212 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
969 } 1213 }
970 1214
971 sv = sv_2mortal (sv); 1215 sv = sv_2mortal (sv);
972 1216
973 return sv; 1217 return sv;
974} 1218}
975 1219
1220/////////////////////////////////////////////////////////////////////////////
1221// incremental parser
1222
1223#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1224
1225// returns 0 for notyet, 1 for success or error
1226static int
1227incr_parse (CBOR *self, SV *cborstr)
1228{
1229 STRLEN cur;
1230 SvPV (cborstr, cur);
1231
1232 while (ecb_expect_true (self->incr_need <= cur))
1233 {
1234 // table of integer count bytes
1235 static I8 incr_len[MINOR_MASK + 1] = {
1236 0, 0, 0, 0, 0, 0, 0, 0,
1237 0, 0, 0, 0, 0, 0, 0, 0,
1238 0, 0, 0, 0, 0, 0, 0, 0,
1239 1, 2, 4, 8,-1,-1,-1,-2
1240 };
1241
1242 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1243 U8 m = *p & MINOR_MASK;
1244 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1245 I8 ilen = incr_len[m];
1246
1247 self->incr_need = self->incr_pos + 1;
1248
1249 if (ecb_expect_false (ilen < 0))
1250 {
1251 if (m != MINOR_INDEF)
1252 return 1; // error
1253
1254 if (*p == (MAJOR_MISC | MINOR_INDEF))
1255 {
1256 if (count >= 0)
1257 return 1; // error
1258
1259 count = 1;
1260 }
1261 else
1262 {
1263 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1264 count = -1;
1265 }
1266 }
1267 else
1268 {
1269 self->incr_need += ilen;
1270 if (ecb_expect_false (self->incr_need > cur))
1271 return 0;
1272
1273 int major = *p >> MAJOR_SHIFT;
1274
1275 switch (major)
1276 {
1277 case MAJOR_TAG >> MAJOR_SHIFT:
1278 ++count; // tags merely prefix another value
1279 break;
1280
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 case MAJOR_ARRAY >> MAJOR_SHIFT:
1284 case MAJOR_MAP >> MAJOR_SHIFT:
1285 {
1286 UV len;
1287
1288 if (ecb_expect_false (ilen))
1289 {
1290 len = 0;
1291
1292 do {
1293 len = (len << 8) | *++p;
1294 } while (--ilen);
1295 }
1296 else
1297 len = m;
1298
1299 switch (major)
1300 {
1301 case MAJOR_BYTES >> MAJOR_SHIFT:
1302 case MAJOR_TEXT >> MAJOR_SHIFT:
1303 self->incr_need += len;
1304 if (ecb_expect_false (self->incr_need > cur))
1305 return 0;
1306
1307 break;
1308
1309 case MAJOR_MAP >> MAJOR_SHIFT:
1310 len <<= 1;
1311 case MAJOR_ARRAY >> MAJOR_SHIFT:
1312 if (len)
1313 {
1314 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1315 count = len + 1;
1316 }
1317 break;
1318 }
1319 }
1320 }
1321 }
1322
1323 self->incr_pos = self->incr_need;
1324
1325 if (count > 0)
1326 {
1327 while (!--count)
1328 {
1329 if (!AvFILLp (self->incr_count))
1330 return 1; // done
1331
1332 SvREFCNT_dec_NN (av_pop (self->incr_count));
1333 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1334 }
1335
1336 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1337 }
1338 }
1339
1340 return 0;
1341}
1342
1343
976///////////////////////////////////////////////////////////////////////////// 1344/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1345// XS interface functions
978 1346
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1347MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1348
987 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1355 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
988 1356
989 types_true = get_bool ("Types::Serialiser::true" ); 1357 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1358 types_false = get_bool ("Types::Serialiser::false");
991 types_error = get_bool ("Types::Serialiser::error"); 1359 types_error = get_bool ("Types::Serialiser::error");
1360
1361 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
992 1362
993 sv_cbor = newSVpv ("CBOR", 0); 1363 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1364 SvREADONLY_on (sv_cbor);
995} 1365}
996 1366
1018void shrink (CBOR *self, int enable = 1) 1388void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1389 ALIAS:
1020 shrink = F_SHRINK 1390 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1391 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1392 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1393 allow_cycles = F_ALLOW_CYCLES
1024 dedup_strings = F_DEDUP_STRINGS 1394 pack_strings = F_PACK_STRINGS
1395 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1396 PPCODE:
1026{ 1397{
1027 if (enable) 1398 if (enable)
1028 self->flags |= ix; 1399 self->flags |= ix;
1029 else 1400 else
1035void get_shrink (CBOR *self) 1406void get_shrink (CBOR *self)
1036 ALIAS: 1407 ALIAS:
1037 get_shrink = F_SHRINK 1408 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1409 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1410 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1411 get_allow_cycles = F_ALLOW_CYCLES
1041 get_dedup_strings = F_DEDUP_STRINGS 1412 get_pack_strings = F_PACK_STRINGS
1413 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1414 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1415 XPUSHs (boolSV (self->flags & ix));
1044 1416
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1417void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1418 PPCODE:
1059 XPUSHs (ST (0)); 1431 XPUSHs (ST (0));
1060 1432
1061int get_max_size (CBOR *self) 1433int get_max_size (CBOR *self)
1062 CODE: 1434 CODE:
1063 RETVAL = self->max_size; 1435 RETVAL = self->max_size;
1436 OUTPUT:
1437 RETVAL
1438
1439void filter (CBOR *self, SV *filter = 0)
1440 PPCODE:
1441 SvREFCNT_dec (self->filter);
1442 self->filter = filter ? newSVsv (filter) : filter;
1443 XPUSHs (ST (0));
1444
1445SV *get_filter (CBOR *self)
1446 CODE:
1447 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064 OUTPUT: 1448 OUTPUT:
1065 RETVAL 1449 RETVAL
1066 1450
1067void encode (CBOR *self, SV *scalar) 1451void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1452 PPCODE:
1083 EXTEND (SP, 2); 1467 EXTEND (SP, 2);
1084 PUSHs (sv); 1468 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1469 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1470}
1087 1471
1472void incr_parse (CBOR *self, SV *cborstr)
1473 ALIAS:
1474 incr_parse_multiple = 1
1475 PPCODE:
1476{
1477 if (SvUTF8 (cborstr))
1478 sv_utf8_downgrade (cborstr, 0);
1479
1480 if (!self->incr_count)
1481 {
1482 self->incr_count = newAV ();
1483 self->incr_pos = 0;
1484 self->incr_need = 1;
1485
1486 av_push (self->incr_count, newSViv (1));
1487 }
1488
1489 do
1490 {
1491 if (!incr_parse (self, cborstr))
1492 {
1493 if (self->incr_need > self->max_size && self->max_size)
1494 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1495 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1496
1497 break;
1498 }
1499
1500 SV *sv;
1501 char *offset;
1502
1503 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1504 XPUSHs (sv);
1505
1506 sv_chop (cborstr, offset);
1507
1508 av_clear (self->incr_count);
1509 av_push (self->incr_count, newSViv (1));
1510
1511 self->incr_pos = 0;
1512 self->incr_need = self->incr_pos + 1;
1513 }
1514 while (ix);
1515}
1516
1517void incr_reset (CBOR *self)
1518 CODE:
1519{
1520 SvREFCNT_dec (self->incr_count);
1521 self->incr_count = 0;
1522}
1523
1524void DESTROY (CBOR *self)
1525 PPCODE:
1526 cbor_free (self);
1527
1088PROTOTYPES: ENABLE 1528PROTOTYPES: ENABLE
1089 1529
1090void encode_cbor (SV *scalar) 1530void encode_cbor (SV *scalar)
1531 ALIAS:
1532 encode_cbor = 0
1533 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1534 PPCODE:
1092{ 1535{
1093 CBOR cbor; 1536 CBOR cbor;
1094 cbor_init (&cbor); 1537 cbor_init (&cbor);
1538 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1539 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1540 XPUSHs (scalar);
1097} 1541}
1098 1542
1099void decode_cbor (SV *cborstr) 1543void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines