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 1.15 by root, Tue Oct 29 18:37:31 2013 UTC vs.
Revision 1.54 by root, Sun Apr 24 20:08:41 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_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
56 108
57#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
58 110
59#define SB do { 111#define SB do {
60#define SE } while (0) 112#define SE } while (0)
72# define CBOR_SLOW 0 124# define CBOR_SLOW 0
73# define CBOR_STASH cbor_stash 125# define CBOR_STASH cbor_stash
74#endif 126#endif
75 127
76static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 128static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
77static SV *types_true, *types_false, *types_error, *sv_cbor; 129static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78 130
79typedef struct { 131typedef struct {
80 U32 flags; 132 U32 flags;
81 U32 max_depth; 133 U32 max_depth;
82 STRLEN max_size; 134 STRLEN max_size;
135 SV *filter;
136
137 // for the incremental parser
138 STRLEN incr_pos; // the current offset into the text
139 STRLEN incr_need; // minimum bytes needed to decode
140 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
83} CBOR; 141} CBOR;
84 142
85ecb_inline void 143ecb_inline void
86cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
87{ 145{
88 Zero (cbor, 1, CBOR); 146 Zero (cbor, 1, CBOR);
89 cbor->max_depth = 512; 147 cbor->max_depth = 512;
148}
149
150ecb_inline void
151cbor_free (CBOR *cbor)
152{
153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
90} 155}
91 156
92///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
93// utility functions 158// utility functions
94 159
116 SvPV_renew (sv, SvCUR (sv) + 1); 181 SvPV_renew (sv, SvCUR (sv) + 1);
117#endif 182#endif
118 } 183 }
119} 184}
120 185
121///////////////////////////////////////////////////////////////////////////// 186// minimum length of a string to be registered for stringref
122// fp hell 187ecb_inline int
123 188minimum_string_length (UV idx)
124//TODO 189{
190 return idx > 23
191 ? idx > 0xffU
192 ? idx > 0xffffU
193 ? idx > 0xffffffffU
194 ? 11
195 : 7
196 : 5
197 : 4
198 : 3;
199}
125 200
126///////////////////////////////////////////////////////////////////////////// 201/////////////////////////////////////////////////////////////////////////////
127// encoder 202// encoder
128 203
129// structure used for encoding CBOR 204// structure used for encoding CBOR
132 char *cur; // SvPVX (sv) + current output position 207 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 208 char *end; // SvEND (sv)
134 SV *sv; // result scalar 209 SV *sv; // result scalar
135 CBOR cbor; 210 CBOR cbor;
136 U32 depth; // recursion level 211 U32 depth; // recursion level
212 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
213 UV stringref_idx;
214 HV *shareable; // ptr => index, or 0
215 UV shareable_idx;
137} enc_t; 216} enc_t;
138 217
139ecb_inline void 218ecb_inline void
140need (enc_t *enc, STRLEN len) 219need (enc_t *enc, STRLEN len)
141{ 220{
158static void 237static void
159encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
160{ 239{
161 need (enc, 9); 240 need (enc, 9);
162 241
163 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
164 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
165 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
166 { 245 {
167 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
168 *enc->cur++ = len; 247 *enc->cur++ = len;
169 } 248 }
170 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
171 { 250 {
172 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
173 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
174 *enc->cur++ = len; 253 *enc->cur++ = len;
175 } 254 }
176 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
177 { 256 {
178 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
179 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
180 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
181 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
182 *enc->cur++ = len; 261 *enc->cur++ = len;
183 } 262 }
184 else 263 else
185 { 264 {
186 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
187 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
188 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
189 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
190 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
191 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
193 *enc->cur++ = len >> 8; 272 *enc->cur++ = len >> 8;
194 *enc->cur++ = len; 273 *enc->cur++ = len;
195 } 274 }
196} 275}
197 276
198static void 277ecb_inline void
278encode_tag (enc_t *enc, UV tag)
279{
280 encode_uint (enc, MAJOR_TAG, tag);
281}
282
283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
199encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
200{ 286{
201 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 287 STRLEN ulen = len;
288 U8 *p, *pend = (U8 *)str + len;
289
290 for (p = (U8 *)str; p < pend; ++p)
291 ulen += *p >> 7; // count set high bits
292
293 encode_uint (enc, MAJOR_TEXT, ulen);
294
295 need (enc, ulen);
296 for (p = (U8 *)str; p < pend; ++p)
297 if (*p < 0x80)
298 *enc->cur++ = *p;
299 else
300 {
301 *enc->cur++ = 0xc0 + (*p >> 6);
302 *enc->cur++ = 0x80 + (*p & 63);
303 }
304}
305
306ecb_inline void
307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
202 need (enc, len); 317 need (enc, len);
203 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
204 enc->cur += len; 319 enc->cur += len;
205} 320}
206 321
322ecb_inline void
323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
324{
325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
326 {
327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
328
329 if (SvOK (*svp))
330 {
331 // already registered, use stringref
332 encode_tag (enc, CBOR_TAG_STRINGREF);
333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
334 return;
335 }
336 else if (len >= minimum_string_length (enc->stringref_idx))
337 {
338 // register only
339 sv_setuv (*svp, enc->stringref_idx);
340 ++enc->stringref_idx;
341 }
342 }
343
344 encode_str (enc, upgrade_utf8, utf8, str, len);
345}
346
207static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
208 348
209static void 349static void
210encode_av (enc_t *enc, AV *av) 350encode_av (enc_t *enc, AV *av)
211{ 351{
214 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
215 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
216 356
217 ++enc->depth; 357 ++enc->depth;
218 358
219 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
220 360
361 if (SvMAGICAL (av))
221 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
222 { 363 {
223 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
224 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
225 } 366 }
367 else
368 for (i = 0; i <= len; ++i)
369 {
370 SV *sv = AvARRAY (av)[i];
371 encode_sv (enc, sv ? sv : &PL_sv_undef);
372 }
226 373
227 --enc->depth; 374 --enc->depth;
228} 375}
229 376
230static void 377static void
239 386
240 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
241 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
242 389
243 if (mg) 390 if (mg)
244 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245 else 392 else
246 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
247 394
248 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
249 { 396 {
250 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
251 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
252 else 399 else
253 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
254 401
255 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
256 } 403 }
257 404
258 if (mg) 405 if (mg)
259 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260 407
261 --enc->depth; 408 --enc->depth;
262} 409}
263 410
264// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
265static void 412static void
266encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
267{ 414{
268 svtype svt;
269
270 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
416
271 svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
272 418
273 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
274 { 420 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 421 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 422 ? types_boolean_stash
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash 428 ? cbor_tagged_stash
283 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
284 430
285 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
286 GV *method;
287 432
288 if (stash == boolean_stash) 433 if (stash == boolean_stash)
289 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 434 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
436 return;
437 }
290 else if (stash == error_stash) 438 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
292 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
293 { 444 {
294 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296 447
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 449 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
450
451 return;
452 }
453 }
454
455 if (ecb_expect_false (SvREFCNT (sv) > 1)
456 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
457 {
458 if (!enc->shareable)
459 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
460
461 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
462
463 if (SvOK (*svp))
299 } 464 {
465 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
466 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
467 return;
468 }
469 else
470 {
471 sv_setuv (*svp, enc->shareable_idx);
472 ++enc->shareable_idx;
473 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
474 }
475 }
476
477 if (ecb_expect_false (SvOBJECT (sv)))
478 {
479 HV *stash = SvSTASH (sv);
480 GV *method;
481
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 { 483 {
302 dSP; 484 dSP;
303 485
304 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
305 // we re-bless the reference to get overload and other niceties right 488 // we re-bless the reference to get overload and other niceties right
306 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307 490
308 PUTBACK; 491 PUTBACK;
309 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
322 } 505 }
323 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324 { 507 {
325 dSP; 508 dSP;
326 509
327 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
328 EXTEND (SP, 2); 513 EXTEND (SP, 2);
329 // we re-bless the reference to get overload and other niceties right 514 // we re-bless the reference to get overload and other niceties right
330 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
332 517
336 521
337 // catch this surprisingly common error 522 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 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)); 524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340 525
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 529
345 while (count) 530 while (count)
346 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
347 532
348 PUTBACK; 533 PUTBACK;
355 } 540 }
356 else if (svt == SVt_PVHV) 541 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 542 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 543 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 544 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 545 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 546 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 547 encode_tag (enc, CBOR_TAG_INDIRECTION);
548 encode_sv (enc, sv);
549 }
380} 550}
381 551
382static void 552static void
383encode_nv (enc_t *enc, SV *sv) 553encode_nv (enc_t *enc, SV *sv)
384{ 554{
385 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
386 556
387 need (enc, 9); 557 need (enc, 9);
388 558
389 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
390 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391 //TODO: maybe I32? 561 //TODO: maybe I32?
392 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
393 { 563 {
394 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
395 565
396 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
397 567
398 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
400 570
401 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
403 } 573 }
404 else 574 else
405 { 575 {
406 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
407 577
408 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
409 579
410 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
411 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
412 582
413 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
422 592
423 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
424 { 594 {
425 STRLEN len; 595 STRLEN len;
426 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
427 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
428 } 598 }
429 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
430 encode_nv (enc, sv); 600 encode_nv (enc, sv);
431 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
432 { 602 {
433 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
434 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
435 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
436 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
437 else 607 else
438 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
439 } 609 }
440 else if (SvROK (sv)) 610 else if (SvROK (sv))
441 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
442 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
443 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446 else 616 else
447 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 617 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
448 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449} 619}
450 620
451static SV * 621static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
453{ 623{
454 enc_t enc; 624 enc_t enc = { 0 };
455 625
456 enc.cbor = *cbor; 626 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 630
462 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
632
633 if (cbor->flags & F_PACK_STRINGS)
634 {
635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
638 }
639
463 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
464 641
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 643 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
467 644
481 U8 *end; // end of input string 658 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 659 const char *err; // parse error, if != 0
483 CBOR cbor; 660 CBOR cbor;
484 U32 depth; // recursion depth 661 U32 depth; // recursion depth
485 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
663 AV *shareable;
664 AV *stringref;
665 SV *decode_tagged;
486} dec_t; 666} dec_t;
487 667
488#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
489 669
490#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 670#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 673#define DEC_DEC_DEPTH --dec->depth
494 674
495static UV 675static UV
496decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
497{ 677{
498 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
499 { 679 ++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;
504 680
505 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
506 WANT (2); 685 WANT (1);
507 dec->cur += 2; 686 dec->cur += 1;
508 return dec->cur[-1]; 687 return dec->cur[-1];
509 688 }
510 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
511 WANT (3); 691 WANT (2);
512 dec->cur += 3; 692 dec->cur += 2;
513 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
514 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
515 695 }
516 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
517 WANT (5); 698 WANT (4);
518 dec->cur += 5; 699 dec->cur += 4;
519 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
520 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
521 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
522 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
523 704 }
524 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
525 WANT (9); 707 WANT (8);
526 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
527 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
528 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
529 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
530 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
531 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
532 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
533 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
534 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
535 723 }
536 default: 724 else
537 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538 }
539 726
540fail: 727fail:
541 return 0; 728 return 0;
542} 729}
543 730
548{ 735{
549 AV *av = newAV (); 736 AV *av = newAV ();
550 737
551 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
552 739
553 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554 { 741 {
555 ++dec->cur; 742 ++dec->cur;
556 743
557 for (;;) 744 for (;;)
558 { 745 {
559 WANT (1); 746 WANT (1);
560 747
561 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
562 { 749 {
563 ++dec->cur; 750 ++dec->cur;
564 break; 751 break;
565 } 752 }
566 753
569 } 756 }
570 else 757 else
571 { 758 {
572 int i, len = decode_uint (dec); 759 int i, len = decode_uint (dec);
573 760
761 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); 762 av_fill (av, len - 1);
575 763
576 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
577 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
578 } 766 }
584 SvREFCNT_dec (av); 772 SvREFCNT_dec (av);
585 DEC_DEC_DEPTH; 773 DEC_DEC_DEPTH;
586 return &PL_sv_undef; 774 return &PL_sv_undef;
587} 775}
588 776
777static void
778decode_he (dec_t *dec, HV *hv)
779{
780 // for speed reasons, we specialcase single-string
781 // byte or utf-8 strings as keys, but only when !stringref
782
783 if (ecb_expect_true (!dec->stringref))
784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
785 {
786 I32 len = decode_uint (dec);
787 char *key = (char *)dec->cur;
788
789 WANT (len);
790 dec->cur += len;
791
792 hv_store (hv, key, len, decode_sv (dec), 0);
793
794 return;
795 }
796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
797 {
798 I32 len = decode_uint (dec);
799 char *key = (char *)dec->cur;
800
801 WANT (len);
802 dec->cur += len;
803
804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
805 if (!is_utf8_string (key, len))
806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
807
808 hv_store (hv, key, -len, decode_sv (dec), 0);
809
810 return;
811 }
812
813 SV *k = decode_sv (dec);
814 SV *v = decode_sv (dec);
815
816 hv_store_ent (hv, k, v, 0);
817 SvREFCNT_dec (k);
818
819fail:
820 ;
821}
822
589static SV * 823static SV *
590decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
591{ 825{
592 HV *hv = newHV (); 826 HV *hv = newHV ();
593 827
594 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
595 829
596 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
597 { 831 {
598 ++dec->cur; 832 ++dec->cur;
599 833
600 for (;;) 834 for (;;)
601 { 835 {
602 WANT (1); 836 WANT (1);
603 837
604 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
605 { 839 {
606 ++dec->cur; 840 ++dec->cur;
607 break; 841 break;
608 } 842 }
609 843
610 SV *k = decode_sv (dec); 844 decode_he (dec, hv);
611 SV *v = decode_sv (dec);
612
613 hv_store_ent (hv, k, v, 0);
614 SvREFCNT_dec (k);
615 } 845 }
616 } 846 }
617 else 847 else
618 { 848 {
619 int len = decode_uint (dec); 849 int pairs = decode_uint (dec);
620 850
621 while (len--) 851 while (pairs--)
622 { 852 decode_he (dec, hv);
623 SV *k = decode_sv (dec);
624 SV *v = decode_sv (dec);
625
626 hv_store_ent (hv, k, v, 0);
627 SvREFCNT_dec (k);
628 }
629 } 853 }
630 854
631 DEC_DEC_DEPTH; 855 DEC_DEC_DEPTH;
632 return newRV_noinc ((SV *)hv); 856 return newRV_noinc ((SV *)hv);
633 857
640static SV * 864static SV *
641decode_str (dec_t *dec, int utf8) 865decode_str (dec_t *dec, int utf8)
642{ 866{
643 SV *sv = 0; 867 SV *sv = 0;
644 868
645 if ((*dec->cur & 31) == 31) 869 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
646 { 870 {
871 // indefinite length strings
647 ++dec->cur; 872 ++dec->cur;
648 873
874 U8 major = *dec->cur & MAJOR_MISC;
875
649 sv = newSVpvn ("", 0); 876 sv = newSVpvn ("", 0);
650 877
651 // not very fast, and certainly not robust against illegal input
652 for (;;) 878 for (;;)
653 { 879 {
654 WANT (1); 880 WANT (1);
655 881
656 if (*dec->cur == (0xe0 | 31)) 882 if ((*dec->cur - major) > LENGTH_EXT8)
883 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
657 { 884 {
658 ++dec->cur; 885 ++dec->cur;
659 break; 886 break;
660 } 887 }
888 else
889 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
661 890
662 sv_catsv (sv, decode_sv (dec)); 891 STRLEN len = decode_uint (dec);
892
893 WANT (len);
894 sv_catpvn (sv, dec->cur, len);
895 dec->cur += len;
663 } 896 }
664 } 897 }
665 else 898 else
666 { 899 {
667 STRLEN len = decode_uint (dec); 900 STRLEN len = decode_uint (dec);
668 901
669 WANT (len); 902 WANT (len);
670 sv = newSVpvn (dec->cur, len); 903 sv = newSVpvn (dec->cur, len);
671 dec->cur += len; 904 dec->cur += len;
905
906 if (ecb_expect_false (dec->stringref)
907 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
908 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
672 } 909 }
673 910
674 if (utf8) 911 if (utf8)
912 {
913 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
914 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
915 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
916
675 SvUTF8_on (sv); 917 SvUTF8_on (sv);
918 }
676 919
677 return sv; 920 return sv;
678 921
679fail: 922fail:
680 SvREFCNT_dec (sv); 923 SvREFCNT_dec (sv);
682} 925}
683 926
684static SV * 927static SV *
685decode_tagged (dec_t *dec) 928decode_tagged (dec_t *dec)
686{ 929{
930 SV *sv = 0;
687 UV tag = decode_uint (dec); 931 UV tag = decode_uint (dec);
932
933 WANT (1);
934
935 switch (tag)
936 {
937 case CBOR_TAG_MAGIC:
688 SV *sv = decode_sv (dec); 938 sv = decode_sv (dec);
939 break;
689 940
690 if (tag == CBOR_TAG_MAGIC) 941 case CBOR_TAG_INDIRECTION:
691 return sv; 942 sv = newRV_noinc (decode_sv (dec));
692 else if (tag == CBOR_TAG_PERL_OBJECT) 943 break;
693 { 944
945 case CBOR_TAG_STRINGREF_NAMESPACE:
946 {
947 ENTER; SAVETMPS;
948
949 SAVESPTR (dec->stringref);
950 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
951
952 sv = decode_sv (dec);
953
954 FREETMPS; LEAVE;
955 }
956 break;
957
958 case CBOR_TAG_STRINGREF:
959 {
960 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
961 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
962
963 UV idx = decode_uint (dec);
964
965 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
966 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
967
968 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
969 }
970 break;
971
972 case CBOR_TAG_VALUE_SHAREABLE:
973 {
974 if (ecb_expect_false (!dec->shareable))
975 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
976
977 if (dec->cbor.flags & F_ALLOW_CYCLES)
978 {
979 sv = newSV (0);
980 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
981
982 SV *osv = decode_sv (dec);
983 sv_setsv (sv, osv);
984 SvREFCNT_dec_NN (osv);
985 }
986 else
987 {
988 av_push (dec->shareable, &PL_sv_undef);
989 int idx = AvFILLp (dec->shareable);
990 sv = decode_sv (dec);
991 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
992 }
993 }
994 break;
995
996 case CBOR_TAG_VALUE_SHAREDREF:
997 {
998 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
999 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1000
1001 UV idx = decode_uint (dec);
1002
1003 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
1004 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1005
1006 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1007
1008 if (sv == &PL_sv_undef)
1009 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1010 }
1011 break;
1012
1013 case CBOR_TAG_PERL_OBJECT:
1014 {
1015 sv = decode_sv (dec);
1016
694 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1017 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
695 ERR ("corrupted CBOR data (non-array perl object)"); 1018 ERR ("corrupted CBOR data (non-array perl object)");
696 1019
697 AV *av = (AV *)SvRV (sv); 1020 AV *av = (AV *)SvRV (sv);
698 int len = av_len (av) + 1; 1021 int len = av_len (av) + 1;
699 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1022 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
700 1023
701 if (!stash) 1024 if (!stash)
702 ERR ("cannot decode perl-object (package does not exist)"); 1025 ERR ("cannot decode perl-object (package does not exist)");
703 1026
704 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1027 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
705 1028
706 if (!method) 1029 if (!method)
707 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1030 ERR ("cannot decode perl-object (package does not have a THAW method)");
708 1031
709 dSP; 1032 dSP;
710 1033
711 ENTER; SAVETMPS; PUSHMARK (SP); 1034 ENTER; SAVETMPS;
1035 PUSHMARK (SP);
712 EXTEND (SP, len + 1); 1036 EXTEND (SP, len + 1);
713 // we re-bless the reference to get overload and other niceties right 1037 // we re-bless the reference to get overload and other niceties right
714 PUSHs (*av_fetch (av, 0, 1)); 1038 PUSHs (*av_fetch (av, 0, 1));
715 PUSHs (sv_cbor); 1039 PUSHs (sv_cbor);
716 1040
717 int i; 1041 int i;
718 1042
719 for (i = 1; i < len; ++i) 1043 for (i = 1; i < len; ++i)
720 PUSHs (*av_fetch (av, i, 1)); 1044 PUSHs (*av_fetch (av, i, 1));
721 1045
722 PUTBACK; 1046 PUTBACK;
723 call_sv ((SV *)GvCV (method), G_SCALAR); 1047 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
724 SPAGAIN; 1048 SPAGAIN;
725 1049
1050 if (SvTRUE (ERRSV))
1051 {
1052 FREETMPS; LEAVE;
1053 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1054 }
1055
726 SvREFCNT_dec (sv); 1056 SvREFCNT_dec (sv);
727 sv = SvREFCNT_inc (POPs); 1057 sv = SvREFCNT_inc (POPs);
728 1058
729 PUTBACK; 1059 PUTBACK;
730 1060
731 FREETMPS; LEAVE; 1061 FREETMPS; LEAVE;
1062 }
1063 break;
732 1064
733 return sv; 1065 default:
734 } 1066 {
735 else 1067 sv = decode_sv (dec);
736 { 1068
1069 dSP;
1070 ENTER; SAVETMPS;
1071 SAVESTACK_POS ();
1072 PUSHMARK (SP);
1073 EXTEND (SP, 2);
1074 PUSHs (newSVuv (tag));
1075 PUSHs (sv);
1076
1077 PUTBACK;
1078 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1079 SPAGAIN;
1080
1081 if (SvTRUE (ERRSV))
1082 {
1083 FREETMPS; LEAVE;
1084 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1085 }
1086
1087 if (count)
1088 {
1089 SvREFCNT_dec (sv);
1090 sv = SvREFCNT_inc (POPs);
1091 }
1092 else
1093 {
737 AV *av = newAV (); 1094 AV *av = newAV ();
738 av_push (av, newSVuv (tag)); 1095 av_push (av, newSVuv (tag));
739 av_push (av, sv); 1096 av_push (av, sv);
740 1097
741 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1098 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
742 ? cbor_tagged_stash 1099 ? cbor_tagged_stash
743 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1100 : gv_stashpv ("CBOR::XS::Tagged" , 1);
744
745 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1101 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1102 }
1103
1104 PUTBACK;
1105
1106 FREETMPS; LEAVE;
1107 }
1108 break;
746 } 1109 }
1110
1111 return sv;
747 1112
748fail: 1113fail:
749 SvREFCNT_dec (sv); 1114 SvREFCNT_dec (sv);
750 return &PL_sv_undef; 1115 return &PL_sv_undef;
751} 1116}
753static SV * 1118static SV *
754decode_sv (dec_t *dec) 1119decode_sv (dec_t *dec)
755{ 1120{
756 WANT (1); 1121 WANT (1);
757 1122
758 switch (*dec->cur >> 5) 1123 switch (*dec->cur >> MAJOR_SHIFT)
759 { 1124 {
760 case 0: // unsigned int 1125 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
761 return newSVuv (decode_uint (dec)); 1126 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
762 case 1: // negative int 1127 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
763 return newSViv (-1 - (IV)decode_uint (dec)); 1128 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
764 case 2: // octet string 1129 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
765 return decode_str (dec, 0); 1130 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
766 case 3: // utf-8 string 1131 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
767 return decode_str (dec, 1); 1132
768 case 4: // array 1133 case MAJOR_MISC >> MAJOR_SHIFT:
769 return decode_av (dec);
770 case 5: // map
771 return decode_hv (dec);
772 case 6: // tag
773 return decode_tagged (dec);
774 case 7: // misc
775 switch (*dec->cur++ & 31) 1134 switch (*dec->cur++ & MINOR_MASK)
776 { 1135 {
777 case 20: 1136 case SIMPLE_FALSE:
778#if CBOR_SLOW 1137#if CBOR_SLOW
779 types_false = get_bool ("Types::Serialiser::false"); 1138 types_false = get_bool ("Types::Serialiser::false");
780#endif 1139#endif
781 return newSVsv (types_false); 1140 return newSVsv (types_false);
782 case 21: 1141 case SIMPLE_TRUE:
783#if CBOR_SLOW 1142#if CBOR_SLOW
784 types_true = get_bool ("Types::Serialiser::true"); 1143 types_true = get_bool ("Types::Serialiser::true");
785#endif 1144#endif
786 return newSVsv (types_true); 1145 return newSVsv (types_true);
787 case 22: 1146 case SIMPLE_NULL:
788 return newSVsv (&PL_sv_undef); 1147 return newSVsv (&PL_sv_undef);
789 case 23: 1148 case SIMPLE_UNDEF:
790#if CBOR_SLOW 1149#if CBOR_SLOW
791 types_error = get_bool ("Types::Serialiser::error"); 1150 types_error = get_bool ("Types::Serialiser::error");
792#endif 1151#endif
793 return newSVsv (types_error); 1152 return newSVsv (types_error);
794 1153
795 case 25: 1154 case MISC_FLOAT16:
796 { 1155 {
797 WANT (2); 1156 WANT (2);
798 1157
799 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1158 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
800 dec->cur += 2; 1159 dec->cur += 2;
801 1160
802 return newSVnv (ecb_binary16_to_float (fp)); 1161 return newSVnv (ecb_binary16_to_float (fp));
803 } 1162 }
804 1163
805 case 26: 1164 case MISC_FLOAT32:
806 { 1165 {
807 uint32_t fp; 1166 uint32_t fp;
808 WANT (4); 1167 WANT (4);
809 memcpy (&fp, dec->cur, 4); 1168 memcpy (&fp, dec->cur, 4);
810 dec->cur += 4; 1169 dec->cur += 4;
813 fp = ecb_bswap32 (fp); 1172 fp = ecb_bswap32 (fp);
814 1173
815 return newSVnv (ecb_binary32_to_float (fp)); 1174 return newSVnv (ecb_binary32_to_float (fp));
816 } 1175 }
817 1176
818 case 27: 1177 case MISC_FLOAT64:
819 { 1178 {
820 uint64_t fp; 1179 uint64_t fp;
821 WANT (8); 1180 WANT (8);
822 memcpy (&fp, dec->cur, 8); 1181 memcpy (&fp, dec->cur, 8);
823 dec->cur += 8; 1182 dec->cur += 8;
826 fp = ecb_bswap64 (fp); 1185 fp = ecb_bswap64 (fp);
827 1186
828 return newSVnv (ecb_binary64_to_double (fp)); 1187 return newSVnv (ecb_binary64_to_double (fp));
829 } 1188 }
830 1189
831 // 0..19 unassigned 1190 // 0..19 unassigned simple
832 // 24 reserved + unassigned (reserved values are not encodable) 1191 // 24 reserved + unassigned simple (reserved values are not encodable)
1192 // 28-30 unassigned misc
1193 // 31 break code
833 default: 1194 default:
834 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1195 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
835 } 1196 }
836 1197
837 break; 1198 break;
838 } 1199 }
839 1200
842} 1203}
843 1204
844static SV * 1205static SV *
845decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1206decode_cbor (SV *string, CBOR *cbor, char **offset_return)
846{ 1207{
847 dec_t dec; 1208 dec_t dec = { 0 };
848 SV *sv; 1209 SV *sv;
1210 STRLEN len;
1211 char *data = SvPVbyte (string, len);
849 1212
850 /* work around bugs in 5.10 where manipulating magic values
851 * makes perl ignore the magic in subsequent accesses.
852 * also make a copy of non-PV values, to get them into a clean
853 * state (SvPV should do that, but it's buggy, see below).
854 */
855 /*SvGETMAGIC (string);*/
856 if (SvMAGICAL (string) || !SvPOK (string))
857 string = sv_2mortal (newSVsv (string));
858
859 SvUPGRADE (string, SVt_PV);
860
861 /* work around a bug in perl 5.10, which causes SvCUR to fail an
862 * assertion with -DDEBUGGING, although SvCUR is documented to
863 * return the xpv_cur field which certainly exists after upgrading.
864 * according to nicholas clark, calling SvPOK fixes this.
865 * But it doesn't fix it, so try another workaround, call SvPV_nolen
866 * and hope for the best.
867 * Damnit, SvPV_nolen still trips over yet another assertion. This
868 * assertion business is seriously broken, try yet another workaround
869 * for the broken -DDEBUGGING.
870 */
871 {
872#ifdef DEBUGGING
873 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
874#else
875 STRLEN offset = SvCUR (string);
876#endif
877
878 if (offset > cbor->max_size && cbor->max_size) 1213 if (len > cbor->max_size && cbor->max_size)
879 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1214 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
880 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1215 (unsigned long)len, (unsigned long)cbor->max_size);
881 }
882
883 sv_utf8_downgrade (string, 0);
884 1216
885 dec.cbor = *cbor; 1217 dec.cbor = *cbor;
886 dec.cur = (U8 *)SvPVX (string); 1218 dec.cur = (U8 *)data;
887 dec.end = (U8 *)SvEND (string); 1219 dec.end = (U8 *)data + len;
888 dec.err = 0;
889 dec.depth = 0;
890 1220
891 sv = decode_sv (&dec); 1221 sv = decode_sv (&dec);
892 1222
893 if (offset_return) 1223 if (offset_return)
894 *offset_return = dec.cur; 1224 *offset_return = dec.cur;
897 if (dec.cur != dec.end && !dec.err) 1227 if (dec.cur != dec.end && !dec.err)
898 dec.err = "garbage after CBOR object"; 1228 dec.err = "garbage after CBOR object";
899 1229
900 if (dec.err) 1230 if (dec.err)
901 { 1231 {
1232 if (dec.shareable)
1233 {
1234 // need to break cyclic links, which whould all be in shareable
1235 int i;
1236 SV **svp;
1237
1238 for (i = av_len (dec.shareable) + 1; i--; )
1239 if ((svp = av_fetch (dec.shareable, i, 0)))
1240 sv_setsv (*svp, &PL_sv_undef);
1241 }
1242
902 SvREFCNT_dec (sv); 1243 SvREFCNT_dec (sv);
903 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1244 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
904 } 1245 }
905 1246
906 sv = sv_2mortal (sv); 1247 sv = sv_2mortal (sv);
907 1248
908 return sv; 1249 return sv;
909} 1250}
910 1251
1252/////////////////////////////////////////////////////////////////////////////
1253// incremental parser
1254
1255#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1256
1257// returns 0 for notyet, 1 for success or error
1258static int
1259incr_parse (CBOR *self, SV *cborstr)
1260{
1261 STRLEN cur;
1262 SvPV (cborstr, cur);
1263
1264 while (ecb_expect_true (self->incr_need <= cur))
1265 {
1266 // table of integer count bytes
1267 static I8 incr_len[MINOR_MASK + 1] = {
1268 0, 0, 0, 0, 0, 0, 0, 0,
1269 0, 0, 0, 0, 0, 0, 0, 0,
1270 0, 0, 0, 0, 0, 0, 0, 0,
1271 1, 2, 4, 8,-1,-1,-1,-2
1272 };
1273
1274 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1275 U8 m = *p & MINOR_MASK;
1276 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1277 I8 ilen = incr_len[m];
1278
1279 self->incr_need = self->incr_pos + 1;
1280
1281 if (ecb_expect_false (ilen < 0))
1282 {
1283 if (m != MINOR_INDEF)
1284 return 1; // error
1285
1286 if (*p == (MAJOR_MISC | MINOR_INDEF))
1287 {
1288 if (count >= 0)
1289 return 1; // error
1290
1291 count = 1;
1292 }
1293 else
1294 {
1295 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1296 count = -1;
1297 }
1298 }
1299 else
1300 {
1301 self->incr_need += ilen;
1302 if (ecb_expect_false (self->incr_need > cur))
1303 return 0;
1304
1305 int major = *p >> MAJOR_SHIFT;
1306
1307 switch (major)
1308 {
1309 case MAJOR_TAG >> MAJOR_SHIFT:
1310 ++count; // tags merely prefix another value
1311 break;
1312
1313 case MAJOR_BYTES >> MAJOR_SHIFT:
1314 case MAJOR_TEXT >> MAJOR_SHIFT:
1315 case MAJOR_ARRAY >> MAJOR_SHIFT:
1316 case MAJOR_MAP >> MAJOR_SHIFT:
1317 {
1318 UV len;
1319
1320 if (ecb_expect_false (ilen))
1321 {
1322 len = 0;
1323
1324 do {
1325 len = (len << 8) | *++p;
1326 } while (--ilen);
1327 }
1328 else
1329 len = m;
1330
1331 switch (major)
1332 {
1333 case MAJOR_BYTES >> MAJOR_SHIFT:
1334 case MAJOR_TEXT >> MAJOR_SHIFT:
1335 self->incr_need += len;
1336 if (ecb_expect_false (self->incr_need > cur))
1337 return 0;
1338
1339 break;
1340
1341 case MAJOR_MAP >> MAJOR_SHIFT:
1342 len <<= 1;
1343 case MAJOR_ARRAY >> MAJOR_SHIFT:
1344 if (len)
1345 {
1346 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1347 count = len + 1;
1348 }
1349 break;
1350 }
1351 }
1352 }
1353 }
1354
1355 self->incr_pos = self->incr_need;
1356
1357 if (count > 0)
1358 {
1359 while (!--count)
1360 {
1361 if (!AvFILLp (self->incr_count))
1362 return 1; // done
1363
1364 SvREFCNT_dec_NN (av_pop (self->incr_count));
1365 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1366 }
1367
1368 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1369 }
1370 }
1371
1372 return 0;
1373}
1374
1375
911///////////////////////////////////////////////////////////////////////////// 1376/////////////////////////////////////////////////////////////////////////////
912// XS interface functions 1377// XS interface functions
913 1378
914MODULE = CBOR::XS PACKAGE = CBOR::XS 1379MODULE = CBOR::XS PACKAGE = CBOR::XS
915 1380
922 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1387 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
923 1388
924 types_true = get_bool ("Types::Serialiser::true" ); 1389 types_true = get_bool ("Types::Serialiser::true" );
925 types_false = get_bool ("Types::Serialiser::false"); 1390 types_false = get_bool ("Types::Serialiser::false");
926 types_error = get_bool ("Types::Serialiser::error"); 1391 types_error = get_bool ("Types::Serialiser::error");
1392
1393 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
927 1394
928 sv_cbor = newSVpv ("CBOR", 0); 1395 sv_cbor = newSVpv ("CBOR", 0);
929 SvREADONLY_on (sv_cbor); 1396 SvREADONLY_on (sv_cbor);
930} 1397}
931 1398
952 1419
953void shrink (CBOR *self, int enable = 1) 1420void shrink (CBOR *self, int enable = 1)
954 ALIAS: 1421 ALIAS:
955 shrink = F_SHRINK 1422 shrink = F_SHRINK
956 allow_unknown = F_ALLOW_UNKNOWN 1423 allow_unknown = F_ALLOW_UNKNOWN
1424 allow_sharing = F_ALLOW_SHARING
1425 allow_cycles = F_ALLOW_CYCLES
1426 pack_strings = F_PACK_STRINGS
1427 text_keys = F_TEXT_KEYS
1428 text_strings = F_TEXT_STRINGS
1429 validate_utf8 = F_VALIDATE_UTF8
957 PPCODE: 1430 PPCODE:
958{ 1431{
959 if (enable) 1432 if (enable)
960 self->flags |= ix; 1433 self->flags |= ix;
961 else 1434 else
966 1439
967void get_shrink (CBOR *self) 1440void get_shrink (CBOR *self)
968 ALIAS: 1441 ALIAS:
969 get_shrink = F_SHRINK 1442 get_shrink = F_SHRINK
970 get_allow_unknown = F_ALLOW_UNKNOWN 1443 get_allow_unknown = F_ALLOW_UNKNOWN
1444 get_allow_sharing = F_ALLOW_SHARING
1445 get_allow_cycles = F_ALLOW_CYCLES
1446 get_pack_strings = F_PACK_STRINGS
1447 get_text_keys = F_TEXT_KEYS
1448 get_text_strings = F_TEXT_STRINGS
1449 get_validate_utf8 = F_VALIDATE_UTF8
971 PPCODE: 1450 PPCODE:
972 XPUSHs (boolSV (self->flags & ix)); 1451 XPUSHs (boolSV (self->flags & ix));
973 1452
974void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1453void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
975 PPCODE: 1454 PPCODE:
988 XPUSHs (ST (0)); 1467 XPUSHs (ST (0));
989 1468
990int get_max_size (CBOR *self) 1469int get_max_size (CBOR *self)
991 CODE: 1470 CODE:
992 RETVAL = self->max_size; 1471 RETVAL = self->max_size;
1472 OUTPUT:
1473 RETVAL
1474
1475void filter (CBOR *self, SV *filter = 0)
1476 PPCODE:
1477 SvREFCNT_dec (self->filter);
1478 self->filter = filter ? newSVsv (filter) : filter;
1479 XPUSHs (ST (0));
1480
1481SV *get_filter (CBOR *self)
1482 CODE:
1483 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
993 OUTPUT: 1484 OUTPUT:
994 RETVAL 1485 RETVAL
995 1486
996void encode (CBOR *self, SV *scalar) 1487void encode (CBOR *self, SV *scalar)
997 PPCODE: 1488 PPCODE:
1012 EXTEND (SP, 2); 1503 EXTEND (SP, 2);
1013 PUSHs (sv); 1504 PUSHs (sv);
1014 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1505 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1015} 1506}
1016 1507
1508void incr_parse (CBOR *self, SV *cborstr)
1509 ALIAS:
1510 incr_parse_multiple = 1
1511 PPCODE:
1512{
1513 if (SvUTF8 (cborstr))
1514 sv_utf8_downgrade (cborstr, 0);
1515
1516 if (!self->incr_count)
1517 {
1518 self->incr_count = newAV ();
1519 self->incr_pos = 0;
1520 self->incr_need = 1;
1521
1522 av_push (self->incr_count, newSViv (1));
1523 }
1524
1525 do
1526 {
1527 if (!incr_parse (self, cborstr))
1528 {
1529 if (self->incr_need > self->max_size && self->max_size)
1530 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1531 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1532
1533 break;
1534 }
1535
1536 SV *sv;
1537 char *offset;
1538
1539 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1540 XPUSHs (sv);
1541
1542 sv_chop (cborstr, offset);
1543
1544 av_clear (self->incr_count);
1545 av_push (self->incr_count, newSViv (1));
1546
1547 self->incr_pos = 0;
1548 self->incr_need = self->incr_pos + 1;
1549 }
1550 while (ix);
1551}
1552
1553void incr_reset (CBOR *self)
1554 CODE:
1555{
1556 SvREFCNT_dec (self->incr_count);
1557 self->incr_count = 0;
1558}
1559
1560void DESTROY (CBOR *self)
1561 PPCODE:
1562 cbor_free (self);
1563
1017PROTOTYPES: ENABLE 1564PROTOTYPES: ENABLE
1018 1565
1019void encode_cbor (SV *scalar) 1566void encode_cbor (SV *scalar)
1567 ALIAS:
1568 encode_cbor = 0
1569 encode_cbor_sharing = F_ALLOW_SHARING
1020 PPCODE: 1570 PPCODE:
1021{ 1571{
1022 CBOR cbor; 1572 CBOR cbor;
1023 cbor_init (&cbor); 1573 cbor_init (&cbor);
1574 cbor.flags |= ix;
1024 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1575 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1025 XPUSHs (scalar); 1576 XPUSHs (scalar);
1026} 1577}
1027 1578
1028void decode_cbor (SV *cborstr) 1579void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines