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 1.17 by root, Tue Oct 29 22:04:52 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
29 CBOR_TAG_PERL_OBJECT = 256, 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30 CBOR_TAG_GENERIC_OBJECT = 257, 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31 77
32 // rfc7049 78 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
39 85
40 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44 90
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50 96
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
52}; 98};
53 99
54#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
55#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_VALIDATE_UTF8 0x00000040UL
56 107
57#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
58 109
59#define SB do { 110#define SB do {
60#define SE } while (0) 111#define SE } while (0)
72# define CBOR_SLOW 0 123# define CBOR_SLOW 0
73# define CBOR_STASH cbor_stash 124# define CBOR_STASH cbor_stash
74#endif 125#endif
75 126
76static 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::
77static SV *types_true, *types_false, *types_error, *sv_cbor; 128static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78 129
79typedef struct { 130typedef struct {
80 U32 flags; 131 U32 flags;
81 U32 max_depth; 132 U32 max_depth;
82 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.
83} CBOR; 140} CBOR;
84 141
85ecb_inline void 142ecb_inline void
86cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
87{ 144{
88 Zero (cbor, 1, CBOR); 145 Zero (cbor, 1, CBOR);
89 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);
90} 154}
91 155
92///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
93// utility functions 157// utility functions
94 158
116 SvPV_renew (sv, SvCUR (sv) + 1); 180 SvPV_renew (sv, SvCUR (sv) + 1);
117#endif 181#endif
118 } 182 }
119} 183}
120 184
121///////////////////////////////////////////////////////////////////////////// 185// minimum length of a string to be registered for stringref
122// fp hell 186ecb_inline int
123 187minimum_string_length (UV idx)
124//TODO 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}
125 199
126///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
127// encoder 201// encoder
128 202
129// structure used for encoding CBOR 203// structure used for encoding CBOR
132 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
134 SV *sv; // result scalar 208 SV *sv; // result scalar
135 CBOR cbor; 209 CBOR cbor;
136 U32 depth; // recursion level 210 U32 depth; // recursion level
211 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
212 UV stringref_idx;
213 HV *shareable; // ptr => index, or 0
214 UV shareable_idx;
137} enc_t; 215} enc_t;
138 216
139ecb_inline void 217ecb_inline void
140need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
141{ 219{
158static void 236static void
159encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
160{ 238{
161 need (enc, 9); 239 need (enc, 9);
162 240
163 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
164 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
165 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
166 { 244 {
167 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
168 *enc->cur++ = len; 246 *enc->cur++ = len;
169 } 247 }
170 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
171 { 249 {
172 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
173 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
174 *enc->cur++ = len; 252 *enc->cur++ = len;
175 } 253 }
176 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
177 { 255 {
178 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
179 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
180 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
181 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
182 *enc->cur++ = len; 260 *enc->cur++ = len;
183 } 261 }
184 else 262 else
185 { 263 {
186 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
187 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
188 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
189 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
190 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
191 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
193 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
194 *enc->cur++ = len; 272 *enc->cur++ = len;
195 } 273 }
196} 274}
197 275
198static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282ecb_inline void
199encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
200{ 284{
201 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);
202 need (enc, len); 311 need (enc, len);
203 memcpy (enc->cur, str, len); 312 memcpy (enc->cur, str, len);
204 enc->cur += len; 313 enc->cur += len;
205} 314}
206 315
316static void
317encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
318{
319 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
320 {
321 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
322
323 if (SvOK (*svp))
324 {
325 // already registered, use stringref
326 encode_tag (enc, CBOR_TAG_STRINGREF);
327 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
328 return;
329 }
330 else if (len >= minimum_string_length (enc->stringref_idx))
331 {
332 // register only
333 sv_setuv (*svp, enc->stringref_idx);
334 ++enc->stringref_idx;
335 }
336 }
337
338 encode_str (enc, utf8, str, len);
339}
340
207static void encode_sv (enc_t *enc, SV *sv); 341static void encode_sv (enc_t *enc, SV *sv);
208 342
209static void 343static void
210encode_av (enc_t *enc, AV *av) 344encode_av (enc_t *enc, AV *av)
211{ 345{
214 if (enc->depth >= enc->cbor.max_depth) 348 if (enc->depth >= enc->cbor.max_depth)
215 croak (ERR_NESTING_EXCEEDED); 349 croak (ERR_NESTING_EXCEEDED);
216 350
217 ++enc->depth; 351 ++enc->depth;
218 352
219 encode_uint (enc, 0x80, len + 1); 353 encode_uint (enc, MAJOR_ARRAY, len + 1);
220 354
355 if (SvMAGICAL (av))
221 for (i = 0; i <= len; ++i) 356 for (i = 0; i <= len; ++i)
222 { 357 {
223 SV **svp = av_fetch (av, i, 0); 358 SV **svp = av_fetch (av, i, 0);
224 encode_sv (enc, svp ? *svp : &PL_sv_undef); 359 encode_sv (enc, svp ? *svp : &PL_sv_undef);
225 } 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 }
226 367
227 --enc->depth; 368 --enc->depth;
228} 369}
229 370
230static void 371static void
239 380
240 int pairs = hv_iterinit (hv); 381 int pairs = hv_iterinit (hv);
241 int mg = SvMAGICAL (hv); 382 int mg = SvMAGICAL (hv);
242 383
243 if (mg) 384 if (mg)
244 encode_ch (enc, 0xa0 | 31); 385 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245 else 386 else
246 encode_uint (enc, 0xa0, pairs); 387 encode_uint (enc, MAJOR_MAP, pairs);
247 388
248 while ((he = hv_iternext (hv))) 389 while ((he = hv_iternext (hv)))
249 { 390 {
250 if (HeKLEN (he) == HEf_SVKEY) 391 if (HeKLEN (he) == HEf_SVKEY)
251 encode_sv (enc, HeSVKEY (he)); 392 encode_sv (enc, HeSVKEY (he));
252 else 393 else
253 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 394 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
254 395
255 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));
256 } 397 }
257 398
258 if (mg) 399 if (mg)
259 encode_ch (enc, 0xe0 | 31); 400 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260 401
261 --enc->depth; 402 --enc->depth;
262} 403}
263 404
264// encode objects, arrays and special \0=false and \1=true values. 405// encode objects, arrays and special \0=false and \1=true values.
265static void 406static void
266encode_rv (enc_t *enc, SV *sv) 407encode_rv (enc_t *enc, SV *sv)
267{ 408{
268 svtype svt;
269
270 SvGETMAGIC (sv); 409 SvGETMAGIC (sv);
410
271 svt = SvTYPE (sv); 411 svtype svt = SvTYPE (sv);
272 412
273 if (ecb_expect_false (SvOBJECT (sv))) 413 if (ecb_expect_false (SvOBJECT (sv)))
274 { 414 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 415 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 416 ? types_boolean_stash
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 421 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash 422 ? cbor_tagged_stash
283 : gv_stashpv ("CBOR::XS::Tagged" , 1); 423 : gv_stashpv ("CBOR::XS::Tagged" , 1);
284 424
285 HV *stash = SvSTASH (sv); 425 HV *stash = SvSTASH (sv);
286 GV *method;
287 426
288 if (stash == boolean_stash) 427 if (stash == boolean_stash)
289 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 }
290 else if (stash == error_stash) 432 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23); 433 {
434 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 return;
436 }
292 else if (stash == tagged_stash) 437 else if (stash == tagged_stash)
293 { 438 {
294 if (svt != SVt_PVAV) 439 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 440 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296 441
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 442 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 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))
299 } 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
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 476 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 { 477 {
302 dSP; 478 dSP;
303 479
304 ENTER; SAVETMPS; PUSHMARK (SP); 480 ENTER; SAVETMPS;
481 PUSHMARK (SP);
305 // 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
306 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 483 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307 484
308 PUTBACK; 485 PUTBACK;
309 // G_SCALAR ensures that return value is 1 486 // G_SCALAR ensures that return value is 1
322 } 499 }
323 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 500 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324 { 501 {
325 dSP; 502 dSP;
326 503
327 ENTER; SAVETMPS; PUSHMARK (SP); 504 ENTER; SAVETMPS;
505 SAVESTACK_POS ();
506 PUSHMARK (SP);
328 EXTEND (SP, 2); 507 EXTEND (SP, 2);
329 // 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
330 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 509 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331 PUSHs (sv_cbor); 510 PUSHs (sv_cbor);
332 511
336 515
337 // catch this surprisingly common error 516 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 517 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 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));
340 519
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 520 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 521 encode_uint (enc, MAJOR_ARRAY, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 522 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 523
345 while (count) 524 while (count)
346 encode_sv (enc, SP[1 - count--]); 525 encode_sv (enc, SP[1 - count--]);
347 526
348 PUTBACK; 527 PUTBACK;
355 } 534 }
356 else if (svt == SVt_PVHV) 535 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 536 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 537 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 538 encode_av (enc, (AV *)sv);
360 else if (svt < SVt_PVAV)
361 {
362 STRLEN len = 0;
363 char *pv = svt ? SvPV (sv, len) : 0;
364
365 if (len == 1 && *pv == '1')
366 encode_ch (enc, 0xe0 | 21);
367 else if (len == 1 && *pv == '0')
368 encode_ch (enc, 0xe0 | 20);
369 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370 encode_ch (enc, 0xe0 | 23);
371 else
372 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374 }
375 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376 encode_ch (enc, 0xe0 | 23);
377 else 539 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 540 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 541 encode_tag (enc, CBOR_TAG_INDIRECTION);
542 encode_sv (enc, sv);
543 }
380} 544}
381 545
382static void 546static void
383encode_nv (enc_t *enc, SV *sv) 547encode_nv (enc_t *enc, SV *sv)
384{ 548{
385 double nv = SvNVX (sv); 549 double nv = SvNVX (sv);
386 550
387 need (enc, 9); 551 need (enc, 9);
388 552
389 if (ecb_expect_false (nv == (U32)nv)) 553 if (ecb_expect_false (nv == (NV)(U32)nv))
390 encode_uint (enc, 0x00, (U32)nv); 554 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391 //TODO: maybe I32? 555 //TODO: maybe I32?
392 else if (ecb_expect_false (nv == (float)nv)) 556 else if (ecb_expect_false (nv == (float)nv))
393 { 557 {
394 uint32_t fp = ecb_float_to_binary32 (nv); 558 uint32_t fp = ecb_float_to_binary32 (nv);
395 559
396 *enc->cur++ = 0xe0 | 26; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
397 561
398 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp); 563 fp = ecb_bswap32 (fp);
400 564
401 memcpy (enc->cur, &fp, 4); 565 memcpy (enc->cur, &fp, 4);
403 } 567 }
404 else 568 else
405 { 569 {
406 uint64_t fp = ecb_double_to_binary64 (nv); 570 uint64_t fp = ecb_double_to_binary64 (nv);
407 571
408 *enc->cur++ = 0xe0 | 27; 572 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
409 573
410 if (!ecb_big_endian ()) 574 if (!ecb_big_endian ())
411 fp = ecb_bswap64 (fp); 575 fp = ecb_bswap64 (fp);
412 576
413 memcpy (enc->cur, &fp, 8); 577 memcpy (enc->cur, &fp, 8);
422 586
423 if (SvPOKp (sv)) 587 if (SvPOKp (sv))
424 { 588 {
425 STRLEN len; 589 STRLEN len;
426 char *str = SvPV (sv, len); 590 char *str = SvPV (sv, len);
427 encode_str (enc, SvUTF8 (sv), str, len); 591 encode_strref (enc, SvUTF8 (sv), str, len);
428 } 592 }
429 else if (SvNOKp (sv)) 593 else if (SvNOKp (sv))
430 encode_nv (enc, sv); 594 encode_nv (enc, sv);
431 else if (SvIOKp (sv)) 595 else if (SvIOKp (sv))
432 { 596 {
433 if (SvIsUV (sv)) 597 if (SvIsUV (sv))
434 encode_uint (enc, 0x00, SvUVX (sv)); 598 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
435 else if (SvIVX (sv) >= 0) 599 else if (SvIVX (sv) >= 0)
436 encode_uint (enc, 0x00, SvIVX (sv)); 600 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
437 else 601 else
438 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 602 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
439 } 603 }
440 else if (SvROK (sv)) 604 else if (SvROK (sv))
441 encode_rv (enc, SvRV (sv)); 605 encode_rv (enc, SvRV (sv));
442 else if (!SvOK (sv)) 606 else if (!SvOK (sv))
443 encode_ch (enc, 0xe0 | 22); 607 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 608 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445 encode_ch (enc, 0xe0 | 23); 609 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446 else 610 else
447 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",
448 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 612 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449} 613}
450 614
451static SV * 615static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 616encode_cbor (SV *scalar, CBOR *cbor)
453{ 617{
454 enc_t enc; 618 enc_t enc = { 0 };
455 619
456 enc.cbor = *cbor; 620 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 621 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 622 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 623 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 624
462 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
463 encode_sv (&enc, scalar); 634 encode_sv (&enc, scalar);
464 635
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 636 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466 *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
467 638
481 U8 *end; // end of input string 652 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 653 const char *err; // parse error, if != 0
483 CBOR cbor; 654 CBOR cbor;
484 U32 depth; // recursion depth 655 U32 depth; // recursion depth
485 U32 maxdepth; // recursion depth limit 656 U32 maxdepth; // recursion depth limit
657 AV *shareable;
658 AV *stringref;
659 SV *decode_tagged;
486} dec_t; 660} dec_t;
487 661
488#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
489 663
490#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")
493#define DEC_DEC_DEPTH --dec->depth 667#define DEC_DEC_DEPTH --dec->depth
494 668
495static UV 669static UV
496decode_uint (dec_t *dec) 670decode_uint (dec_t *dec)
497{ 671{
498 switch (*dec->cur & 31) 672 U8 m = *dec->cur & MINOR_MASK;
499 { 673 ++dec->cur;
500 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503 return *dec->cur++ & 31;
504 674
505 case 24: 675 if (ecb_expect_true (m < LENGTH_EXT1))
676 return m;
677 else if (ecb_expect_true (m == LENGTH_EXT1))
678 {
506 WANT (2); 679 WANT (1);
507 dec->cur += 2; 680 dec->cur += 1;
508 return dec->cur[-1]; 681 return dec->cur[-1];
509 682 }
510 case 25: 683 else if (ecb_expect_true (m == LENGTH_EXT2))
684 {
511 WANT (3); 685 WANT (2);
512 dec->cur += 3; 686 dec->cur += 2;
513 return (((UV)dec->cur[-2]) << 8) 687 return (((UV)dec->cur[-2]) << 8)
514 | ((UV)dec->cur[-1]); 688 | ((UV)dec->cur[-1]);
515 689 }
516 case 26: 690 else if (ecb_expect_true (m == LENGTH_EXT4))
691 {
517 WANT (5); 692 WANT (4);
518 dec->cur += 5; 693 dec->cur += 4;
519 return (((UV)dec->cur[-4]) << 24) 694 return (((UV)dec->cur[-4]) << 24)
520 | (((UV)dec->cur[-3]) << 16) 695 | (((UV)dec->cur[-3]) << 16)
521 | (((UV)dec->cur[-2]) << 8) 696 | (((UV)dec->cur[-2]) << 8)
522 | ((UV)dec->cur[-1]); 697 | ((UV)dec->cur[-1]);
523 698 }
524 case 27: 699 else if (ecb_expect_true (m == LENGTH_EXT8))
700 {
525 WANT (9); 701 WANT (8);
526 dec->cur += 9; 702 dec->cur += 8;
703
704 return
705#if UVSIZE < 8
706 0
707#else
527 return (((UV)dec->cur[-8]) << 56) 708 (((UV)dec->cur[-8]) << 56)
528 | (((UV)dec->cur[-7]) << 48) 709 | (((UV)dec->cur[-7]) << 48)
529 | (((UV)dec->cur[-6]) << 40) 710 | (((UV)dec->cur[-6]) << 40)
530 | (((UV)dec->cur[-5]) << 32) 711 | (((UV)dec->cur[-5]) << 32)
712#endif
531 | (((UV)dec->cur[-4]) << 24) 713 | (((UV)dec->cur[-4]) << 24)
532 | (((UV)dec->cur[-3]) << 16) 714 | (((UV)dec->cur[-3]) << 16)
533 | (((UV)dec->cur[-2]) << 8) 715 | (((UV)dec->cur[-2]) << 8)
534 | ((UV)dec->cur[-1]); 716 | ((UV)dec->cur[-1]);
535 717 }
536 default: 718 else
537 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 719 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538 }
539 720
540fail: 721fail:
541 return 0; 722 return 0;
542} 723}
543 724
548{ 729{
549 AV *av = newAV (); 730 AV *av = newAV ();
550 731
551 DEC_INC_DEPTH; 732 DEC_INC_DEPTH;
552 733
553 if ((*dec->cur & 31) == 31) 734 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554 { 735 {
555 ++dec->cur; 736 ++dec->cur;
556 737
557 for (;;) 738 for (;;)
558 { 739 {
559 WANT (1); 740 WANT (1);
560 741
561 if (*dec->cur == (0xe0 | 31)) 742 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
562 { 743 {
563 ++dec->cur; 744 ++dec->cur;
564 break; 745 break;
565 } 746 }
566 747
569 } 750 }
570 else 751 else
571 { 752 {
572 int i, len = decode_uint (dec); 753 int i, len = decode_uint (dec);
573 754
755 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574 av_fill (av, len - 1); 756 av_fill (av, len - 1);
575 757
576 for (i = 0; i < len; ++i) 758 for (i = 0; i < len; ++i)
577 AvARRAY (av)[i] = decode_sv (dec); 759 AvARRAY (av)[i] = decode_sv (dec);
578 } 760 }
588 770
589static void 771static void
590decode_he (dec_t *dec, HV *hv) 772decode_he (dec_t *dec, HV *hv)
591{ 773{
592 // for speed reasons, we specialcase single-string 774 // for speed reasons, we specialcase single-string
593 // byte or utf-8 strings as keys. 775 // byte or utf-8 strings as keys, but only when !stringref
594 776
595 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))
596 { 779 {
597 I32 len = decode_uint (dec); 780 I32 len = decode_uint (dec);
598 char *key = (char *)dec->cur; 781 char *key = (char *)dec->cur;
599 782
783 WANT (len);
600 dec->cur += len; 784 dec->cur += len;
601 785
602 hv_store (hv, key, len, decode_sv (dec), 0); 786 hv_store (hv, key, len, decode_sv (dec), 0);
787
788 return;
603 } 789 }
604 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 790 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
605 { 791 {
606 I32 len = decode_uint (dec); 792 I32 len = decode_uint (dec);
607 char *key = (char *)dec->cur; 793 char *key = (char *)dec->cur;
608 794
795 WANT (len);
609 dec->cur += len; 796 dec->cur += len;
610 797
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
611 hv_store (hv, key, -len, decode_sv (dec), 0); 802 hv_store (hv, key, -len, decode_sv (dec), 0);
803
804 return;
612 } 805 }
613 else 806
614 {
615 SV *k = decode_sv (dec); 807 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec); 808 SV *v = decode_sv (dec);
617 809
618 hv_store_ent (hv, k, v, 0); 810 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k); 811 SvREFCNT_dec (k);
620 } 812
813fail:
814 ;
621} 815}
622 816
623static SV * 817static SV *
624decode_hv (dec_t *dec) 818decode_hv (dec_t *dec)
625{ 819{
626 HV *hv = newHV (); 820 HV *hv = newHV ();
627 821
628 DEC_INC_DEPTH; 822 DEC_INC_DEPTH;
629 823
630 if ((*dec->cur & 31) == 31) 824 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
631 { 825 {
632 ++dec->cur; 826 ++dec->cur;
633 827
634 for (;;) 828 for (;;)
635 { 829 {
636 WANT (1); 830 WANT (1);
637 831
638 if (*dec->cur == (0xe0 | 31)) 832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
639 { 833 {
640 ++dec->cur; 834 ++dec->cur;
641 break; 835 break;
642 } 836 }
643 837
664static SV * 858static SV *
665decode_str (dec_t *dec, int utf8) 859decode_str (dec_t *dec, int utf8)
666{ 860{
667 SV *sv = 0; 861 SV *sv = 0;
668 862
669 if ((*dec->cur & 31) == 31) 863 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
670 { 864 {
865 // indefinite length strings
671 ++dec->cur; 866 ++dec->cur;
672 867
868 U8 major = *dec->cur & MAJOR_MISC;
869
673 sv = newSVpvn ("", 0); 870 sv = newSVpvn ("", 0);
674 871
675 // not very fast, and certainly not robust against illegal input
676 for (;;) 872 for (;;)
677 { 873 {
678 WANT (1); 874 WANT (1);
679 875
680 if (*dec->cur == (0xe0 | 31)) 876 if ((*dec->cur - major) > LENGTH_EXT8)
877 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
681 { 878 {
682 ++dec->cur; 879 ++dec->cur;
683 break; 880 break;
684 } 881 }
882 else
883 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
685 884
686 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;
687 } 890 }
688 } 891 }
689 else 892 else
690 { 893 {
691 STRLEN len = decode_uint (dec); 894 STRLEN len = decode_uint (dec);
692 895
693 WANT (len); 896 WANT (len);
694 sv = newSVpvn (dec->cur, len); 897 sv = newSVpvn (dec->cur, len);
695 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));
696 } 903 }
697 904
698 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
699 SvUTF8_on (sv); 911 SvUTF8_on (sv);
912 }
700 913
701 return sv; 914 return sv;
702 915
703fail: 916fail:
704 SvREFCNT_dec (sv); 917 SvREFCNT_dec (sv);
706} 919}
707 920
708static SV * 921static SV *
709decode_tagged (dec_t *dec) 922decode_tagged (dec_t *dec)
710{ 923{
924 SV *sv = 0;
711 UV tag = decode_uint (dec); 925 UV tag = decode_uint (dec);
926
927 WANT (1);
928
929 switch (tag)
930 {
931 case CBOR_TAG_MAGIC:
712 SV *sv = decode_sv (dec); 932 sv = decode_sv (dec);
933 break;
713 934
714 if (tag == CBOR_TAG_MAGIC) 935 case CBOR_TAG_INDIRECTION:
715 return sv; 936 sv = newRV_noinc (decode_sv (dec));
716 else if (tag == CBOR_TAG_PERL_OBJECT) 937 break;
717 {
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)");
720 938
721 AV *av = (AV *)SvRV (sv); 939 case CBOR_TAG_STRINGREF_NAMESPACE:
722 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724
725 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)");
727
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729 940 {
730 if (!method) 941 ENTER; SAVETMPS;
731 ERR ("cannot decode perl-object (package does not have a THAW method)");
732
733 dSP;
734 942
735 ENTER; SAVETMPS; PUSHMARK (SP); 943 SAVESPTR (dec->stringref);
736 EXTEND (SP, len + 1); 944 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
737 // we re-bless the reference to get overload and other niceties right
738 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor);
740 945
741 int i; 946 sv = decode_sv (dec);
742 947
743 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1));
745
746 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
748 SPAGAIN;
749
750 if (SvTRUE (ERRSV))
751 {
752 FREETMPS; LEAVE; 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
966 case CBOR_TAG_VALUE_SHAREABLE:
967 {
968 if (ecb_expect_false (!dec->shareable))
969 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
970
971 if (dec->cbor.flags & F_ALLOW_CYCLES)
972 {
973 sv = newSV (0);
974 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
975
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;
989
990 case CBOR_TAG_VALUE_SHAREDREF:
991 {
992 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
993 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
994
995 UV idx = decode_uint (dec);
996
997 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
998 ERR ("corrupted CBOR data (sharedref index out of bounds)");
999
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");
1004 }
1005 break;
1006
1007 case CBOR_TAG_PERL_OBJECT:
1008 {
1009 sv = decode_sv (dec);
1010
1011 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1012 ERR ("corrupted CBOR data (non-array perl object)");
1013
1014 AV *av = (AV *)SvRV (sv);
1015 int len = av_len (av) + 1;
1016 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1017
1018 if (!stash)
1019 ERR ("cannot decode perl-object (package does not exist)");
1020
1021 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1022
1023 if (!method)
1024 ERR ("cannot decode perl-object (package does not have a THAW method)");
1025
1026 dSP;
1027
1028 ENTER; SAVETMPS;
1029 PUSHMARK (SP);
1030 EXTEND (SP, len + 1);
1031 // we re-bless the reference to get overload and other niceties right
1032 PUSHs (*av_fetch (av, 0, 1));
1033 PUSHs (sv_cbor);
1034
1035 int i;
1036
1037 for (i = 1; i < len; ++i)
1038 PUSHs (*av_fetch (av, i, 1));
1039
1040 PUTBACK;
1041 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1042 SPAGAIN;
1043
1044 if (SvTRUE (ERRSV))
1045 {
1046 FREETMPS; LEAVE;
753 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1047 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
754 } 1048 }
755 1049
756 SvREFCNT_dec (sv); 1050 SvREFCNT_dec (sv);
757 sv = SvREFCNT_inc (POPs); 1051 sv = SvREFCNT_inc (POPs);
758 1052
759 PUTBACK; 1053 PUTBACK;
760 1054
761 FREETMPS; LEAVE; 1055 FREETMPS; LEAVE;
1056 }
1057 break;
762 1058
763 return sv; 1059 default:
764 } 1060 {
765 else 1061 sv = decode_sv (dec);
766 { 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 {
767 AV *av = newAV (); 1088 AV *av = newAV ();
768 av_push (av, newSVuv (tag)); 1089 av_push (av, newSVuv (tag));
769 av_push (av, sv); 1090 av_push (av, sv);
770 1091
771 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1092 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
772 ? cbor_tagged_stash 1093 ? cbor_tagged_stash
773 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1094 : gv_stashpv ("CBOR::XS::Tagged" , 1);
774
775 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1095 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1096 }
1097
1098 PUTBACK;
1099
1100 FREETMPS; LEAVE;
1101 }
1102 break;
776 } 1103 }
1104
1105 return sv;
777 1106
778fail: 1107fail:
779 SvREFCNT_dec (sv); 1108 SvREFCNT_dec (sv);
780 return &PL_sv_undef; 1109 return &PL_sv_undef;
781} 1110}
783static SV * 1112static SV *
784decode_sv (dec_t *dec) 1113decode_sv (dec_t *dec)
785{ 1114{
786 WANT (1); 1115 WANT (1);
787 1116
788 switch (*dec->cur >> 5) 1117 switch (*dec->cur >> MAJOR_SHIFT)
789 { 1118 {
790 case 0: // unsigned int 1119 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
791 return newSVuv (decode_uint (dec)); 1120 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
792 case 1: // negative int 1121 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
793 return newSViv (-1 - (IV)decode_uint (dec)); 1122 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
794 case 2: // octet string 1123 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
795 return decode_str (dec, 0); 1124 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
796 case 3: // utf-8 string 1125 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
797 return decode_str (dec, 1); 1126
798 case 4: // array 1127 case MAJOR_MISC >> MAJOR_SHIFT:
799 return decode_av (dec);
800 case 5: // map
801 return decode_hv (dec);
802 case 6: // tag
803 return decode_tagged (dec);
804 case 7: // misc
805 switch (*dec->cur++ & 31) 1128 switch (*dec->cur++ & MINOR_MASK)
806 { 1129 {
807 case 20: 1130 case SIMPLE_FALSE:
808#if CBOR_SLOW 1131#if CBOR_SLOW
809 types_false = get_bool ("Types::Serialiser::false"); 1132 types_false = get_bool ("Types::Serialiser::false");
810#endif 1133#endif
811 return newSVsv (types_false); 1134 return newSVsv (types_false);
812 case 21: 1135 case SIMPLE_TRUE:
813#if CBOR_SLOW 1136#if CBOR_SLOW
814 types_true = get_bool ("Types::Serialiser::true"); 1137 types_true = get_bool ("Types::Serialiser::true");
815#endif 1138#endif
816 return newSVsv (types_true); 1139 return newSVsv (types_true);
817 case 22: 1140 case SIMPLE_NULL:
818 return newSVsv (&PL_sv_undef); 1141 return newSVsv (&PL_sv_undef);
819 case 23: 1142 case SIMPLE_UNDEF:
820#if CBOR_SLOW 1143#if CBOR_SLOW
821 types_error = get_bool ("Types::Serialiser::error"); 1144 types_error = get_bool ("Types::Serialiser::error");
822#endif 1145#endif
823 return newSVsv (types_error); 1146 return newSVsv (types_error);
824 1147
825 case 25: 1148 case MISC_FLOAT16:
826 { 1149 {
827 WANT (2); 1150 WANT (2);
828 1151
829 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1152 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
830 dec->cur += 2; 1153 dec->cur += 2;
831 1154
832 return newSVnv (ecb_binary16_to_float (fp)); 1155 return newSVnv (ecb_binary16_to_float (fp));
833 } 1156 }
834 1157
835 case 26: 1158 case MISC_FLOAT32:
836 { 1159 {
837 uint32_t fp; 1160 uint32_t fp;
838 WANT (4); 1161 WANT (4);
839 memcpy (&fp, dec->cur, 4); 1162 memcpy (&fp, dec->cur, 4);
840 dec->cur += 4; 1163 dec->cur += 4;
843 fp = ecb_bswap32 (fp); 1166 fp = ecb_bswap32 (fp);
844 1167
845 return newSVnv (ecb_binary32_to_float (fp)); 1168 return newSVnv (ecb_binary32_to_float (fp));
846 } 1169 }
847 1170
848 case 27: 1171 case MISC_FLOAT64:
849 { 1172 {
850 uint64_t fp; 1173 uint64_t fp;
851 WANT (8); 1174 WANT (8);
852 memcpy (&fp, dec->cur, 8); 1175 memcpy (&fp, dec->cur, 8);
853 dec->cur += 8; 1176 dec->cur += 8;
856 fp = ecb_bswap64 (fp); 1179 fp = ecb_bswap64 (fp);
857 1180
858 return newSVnv (ecb_binary64_to_double (fp)); 1181 return newSVnv (ecb_binary64_to_double (fp));
859 } 1182 }
860 1183
861 // 0..19 unassigned 1184 // 0..19 unassigned simple
862 // 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
863 default: 1188 default:
864 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1189 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
865 } 1190 }
866 1191
867 break; 1192 break;
868 } 1193 }
869 1194
872} 1197}
873 1198
874static SV * 1199static SV *
875decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1200decode_cbor (SV *string, CBOR *cbor, char **offset_return)
876{ 1201{
877 dec_t dec; 1202 dec_t dec = { 0 };
878 SV *sv; 1203 SV *sv;
879 STRLEN len; 1204 STRLEN len;
880 char *data = SvPVbyte (string, len); 1205 char *data = SvPVbyte (string, len);
881 1206
882 if (len > cbor->max_size && cbor->max_size) 1207 if (len > cbor->max_size && cbor->max_size)
884 (unsigned long)len, (unsigned long)cbor->max_size); 1209 (unsigned long)len, (unsigned long)cbor->max_size);
885 1210
886 dec.cbor = *cbor; 1211 dec.cbor = *cbor;
887 dec.cur = (U8 *)data; 1212 dec.cur = (U8 *)data;
888 dec.end = (U8 *)data + len; 1213 dec.end = (U8 *)data + len;
889 dec.err = 0;
890 dec.depth = 0;
891 1214
892 sv = decode_sv (&dec); 1215 sv = decode_sv (&dec);
893 1216
894 if (offset_return) 1217 if (offset_return)
895 *offset_return = dec.cur; 1218 *offset_return = dec.cur;
898 if (dec.cur != dec.end && !dec.err) 1221 if (dec.cur != dec.end && !dec.err)
899 dec.err = "garbage after CBOR object"; 1222 dec.err = "garbage after CBOR object";
900 1223
901 if (dec.err) 1224 if (dec.err)
902 { 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
903 SvREFCNT_dec (sv); 1237 SvREFCNT_dec (sv);
904 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);
905 } 1239 }
906 1240
907 sv = sv_2mortal (sv); 1241 sv = sv_2mortal (sv);
908 1242
909 return sv; 1243 return sv;
910} 1244}
911 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
912///////////////////////////////////////////////////////////////////////////// 1370/////////////////////////////////////////////////////////////////////////////
913// XS interface functions 1371// XS interface functions
914 1372
915MODULE = CBOR::XS PACKAGE = CBOR::XS 1373MODULE = CBOR::XS PACKAGE = CBOR::XS
916 1374
923 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1381 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
924 1382
925 types_true = get_bool ("Types::Serialiser::true" ); 1383 types_true = get_bool ("Types::Serialiser::true" );
926 types_false = get_bool ("Types::Serialiser::false"); 1384 types_false = get_bool ("Types::Serialiser::false");
927 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);
928 1388
929 sv_cbor = newSVpv ("CBOR", 0); 1389 sv_cbor = newSVpv ("CBOR", 0);
930 SvREADONLY_on (sv_cbor); 1390 SvREADONLY_on (sv_cbor);
931} 1391}
932 1392
953 1413
954void shrink (CBOR *self, int enable = 1) 1414void shrink (CBOR *self, int enable = 1)
955 ALIAS: 1415 ALIAS:
956 shrink = F_SHRINK 1416 shrink = F_SHRINK
957 allow_unknown = F_ALLOW_UNKNOWN 1417 allow_unknown = F_ALLOW_UNKNOWN
1418 allow_sharing = F_ALLOW_SHARING
1419 allow_cycles = F_ALLOW_CYCLES
1420 pack_strings = F_PACK_STRINGS
1421 utf8_strings = F_UTF8_STRINGS
1422 validate_utf8 = F_VALIDATE_UTF8
958 PPCODE: 1423 PPCODE:
959{ 1424{
960 if (enable) 1425 if (enable)
961 self->flags |= ix; 1426 self->flags |= ix;
962 else 1427 else
967 1432
968void get_shrink (CBOR *self) 1433void get_shrink (CBOR *self)
969 ALIAS: 1434 ALIAS:
970 get_shrink = F_SHRINK 1435 get_shrink = F_SHRINK
971 get_allow_unknown = F_ALLOW_UNKNOWN 1436 get_allow_unknown = F_ALLOW_UNKNOWN
1437 get_allow_sharing = F_ALLOW_SHARING
1438 get_allow_cycles = F_ALLOW_CYCLES
1439 get_pack_strings = F_PACK_STRINGS
1440 get_validate_utf8 = F_VALIDATE_UTF8
972 PPCODE: 1441 PPCODE:
973 XPUSHs (boolSV (self->flags & ix)); 1442 XPUSHs (boolSV (self->flags & ix));
974 1443
975void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1444void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
976 PPCODE: 1445 PPCODE:
989 XPUSHs (ST (0)); 1458 XPUSHs (ST (0));
990 1459
991int get_max_size (CBOR *self) 1460int get_max_size (CBOR *self)
992 CODE: 1461 CODE:
993 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);
994 OUTPUT: 1475 OUTPUT:
995 RETVAL 1476 RETVAL
996 1477
997void encode (CBOR *self, SV *scalar) 1478void encode (CBOR *self, SV *scalar)
998 PPCODE: 1479 PPCODE:
1013 EXTEND (SP, 2); 1494 EXTEND (SP, 2);
1014 PUSHs (sv); 1495 PUSHs (sv);
1015 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1496 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1016} 1497}
1017 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
1018PROTOTYPES: ENABLE 1555PROTOTYPES: ENABLE
1019 1556
1020void encode_cbor (SV *scalar) 1557void encode_cbor (SV *scalar)
1558 ALIAS:
1559 encode_cbor = 0
1560 encode_cbor_sharing = F_ALLOW_SHARING
1021 PPCODE: 1561 PPCODE:
1022{ 1562{
1023 CBOR cbor; 1563 CBOR cbor;
1024 cbor_init (&cbor); 1564 cbor_init (&cbor);
1565 cbor.flags |= ix;
1025 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1566 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1026 XPUSHs (scalar); 1567 XPUSHs (scalar);
1027} 1568}
1028 1569
1029void decode_cbor (SV *cborstr) 1570void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines