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 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.53 by root, Sun Apr 24 19:51:41 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"
14
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
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};
13 65
14// known tags 66// known tags
15enum cbor_tag 67enum cbor_tag
16{ 68{
17 // inofficial extensions (pending iana registration) 69 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 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
20 77
21 // rfc7049 78 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
28 85
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 90
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 96
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 98};
42 99
43#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
44#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_UTF8_KEYS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
45 108
46#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
47 110
48#define SB do { 111#define SB do {
49#define SE } while (0) 112#define SE } while (0)
61# define CBOR_SLOW 0 124# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 125# define CBOR_STASH cbor_stash
63#endif 126#endif
64 127
65static 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::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 129static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 130
68typedef struct { 131typedef struct {
69 U32 flags; 132 U32 flags;
70 U32 max_depth; 133 U32 max_depth;
71 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.
72} CBOR; 141} CBOR;
73 142
74ecb_inline void 143ecb_inline void
75cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
76{ 145{
77 Zero (cbor, 1, CBOR); 146 Zero (cbor, 1, CBOR);
78 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);
79} 155}
80 156
81///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
82// utility functions 158// utility functions
83 159
105 SvPV_renew (sv, SvCUR (sv) + 1); 181 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 182#endif
107 } 183 }
108} 184}
109 185
110///////////////////////////////////////////////////////////////////////////// 186// minimum length of a string to be registered for stringref
111// fp hell 187ecb_inline int
112 188minimum_string_length (UV idx)
113//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}
114 200
115///////////////////////////////////////////////////////////////////////////// 201/////////////////////////////////////////////////////////////////////////////
116// encoder 202// encoder
117 203
118// structure used for encoding CBOR 204// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 207 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 208 char *end; // SvEND (sv)
123 SV *sv; // result scalar 209 SV *sv; // result scalar
124 CBOR cbor; 210 CBOR cbor;
125 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;
126} enc_t; 216} enc_t;
127 217
128ecb_inline void 218ecb_inline void
129need (enc_t *enc, STRLEN len) 219need (enc_t *enc, STRLEN len)
130{ 220{
147static void 237static void
148encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
149{ 239{
150 need (enc, 9); 240 need (enc, 9);
151 241
152 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
154 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
155 { 245 {
156 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 247 *enc->cur++ = len;
158 } 248 }
159 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
160 { 250 {
161 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 253 *enc->cur++ = len;
164 } 254 }
165 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
166 { 256 {
167 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 261 *enc->cur++ = len;
172 } 262 }
173 else 263 else
174 { 264 {
175 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 272 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 273 *enc->cur++ = len;
184 } 274 }
185} 275}
186 276
187static 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
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 286{
190 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);
191 need (enc, len); 317 need (enc, len);
192 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
193 enc->cur += len; 319 enc->cur += len;
194} 320}
195 321
322ecb_inline
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
196static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
197 348
198static void 349static void
199encode_av (enc_t *enc, AV *av) 350encode_av (enc_t *enc, AV *av)
200{ 351{
203 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
205 356
206 ++enc->depth; 357 ++enc->depth;
207 358
208 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 360
361 if (SvMAGICAL (av))
210 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
211 { 363 {
212 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 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 }
215 373
216 --enc->depth; 374 --enc->depth;
217} 375}
218 376
219static void 377static void
228 386
229 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
231 389
232 if (mg) 390 if (mg)
233 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 392 else
235 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
236 394
237 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
238 { 396 {
239 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
241 else 399 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_UTF8_KEYS | F_UTF8_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 401
244 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));
245 } 403 }
246 404
247 if (mg) 405 if (mg)
248 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 407
250 --enc->depth; 408 --enc->depth;
251} 409}
252 410
253// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
254static void 412static void
255encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
256{ 414{
257 svtype svt;
258
259 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
416
260 svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
261 418
262 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
263 { 420 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 421 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 422 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 428 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 430
274 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
275 GV *method;
276 432
277 if (stash == boolean_stash) 433 if (stash == boolean_stash)
278 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 }
279 else if (stash == error_stash) 438 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
281 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
282 { 444 {
283 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 447
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 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))
288 } 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
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 483 {
291 dSP; 484 dSP;
292 485
293 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
294 // 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
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 490
297 PUTBACK; 491 PUTBACK;
298 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
311 } 505 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 507 {
314 dSP; 508 dSP;
315 509
316 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
317 EXTEND (SP, 2); 513 EXTEND (SP, 2);
318 // 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
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
321 517
325 521
326 // catch this surprisingly common error 522 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 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));
329 525
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 529
334 while (count) 530 while (count)
335 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
336 532
337 PUTBACK; 533 PUTBACK;
344 } 540 }
345 else if (svt == SVt_PVHV) 541 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 542 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 543 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 544 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 545 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 546 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 547 encode_tag (enc, CBOR_TAG_INDIRECTION);
548 encode_sv (enc, sv);
549 }
369} 550}
370 551
371static void 552static void
372encode_nv (enc_t *enc, SV *sv) 553encode_nv (enc_t *enc, SV *sv)
373{ 554{
374 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
375 556
376 need (enc, 9); 557 need (enc, 9);
377 558
378 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 561 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
382 { 563 {
383 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
384 565
385 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
386 567
387 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
389 570
390 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
392 } 573 }
393 else 574 else
394 { 575 {
395 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
396 577
397 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
398 579
399 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
401 582
402 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
411 592
412 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
413 { 594 {
414 STRLEN len; 595 STRLEN len;
415 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_UTF8_STRINGS, SvUTF8 (sv), str, len);
417 } 598 }
418 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 600 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
421 { 602 {
422 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
424 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
426 else 607 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
428 } 609 }
429 else if (SvROK (sv)) 610 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 616 else
436 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",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 619}
439 620
440static SV * 621static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
442{ 623{
443 enc_t enc; 624 enc_t enc = { 0 };
444 625
445 enc.cbor = *cbor; 626 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 630
451 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
452 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
453 641
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *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
456 644
470 U8 *end; // end of input string 658 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 659 const char *err; // parse error, if != 0
472 CBOR cbor; 660 CBOR cbor;
473 U32 depth; // recursion depth 661 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
663 AV *shareable;
664 AV *stringref;
665 SV *decode_tagged;
475} dec_t; 666} dec_t;
476 667
477#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
478 669
479#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")
482#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
483 674
484static UV 675static UV
485decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
486{ 677{
487 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
488 { 679 ++dec->cur;
489 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
490 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
491 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
492 return *dec->cur++ & 31;
493 680
494 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
495 WANT (2); 685 WANT (1);
496 dec->cur += 2; 686 dec->cur += 1;
497 return dec->cur[-1]; 687 return dec->cur[-1];
498 688 }
499 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
500 WANT (3); 691 WANT (2);
501 dec->cur += 3; 692 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
504 695 }
505 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
506 WANT (5); 698 WANT (4);
507 dec->cur += 5; 699 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
512 704 }
513 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
514 WANT (9); 707 WANT (8);
515 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
516 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
520 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
524 723 }
525 default: 724 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 726
529fail: 727fail:
530 return 0; 728 return 0;
531} 729}
532 730
537{ 735{
538 AV *av = newAV (); 736 AV *av = newAV ();
539 737
540 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
541 739
542 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 741 {
544 ++dec->cur; 742 ++dec->cur;
545 743
546 for (;;) 744 for (;;)
547 { 745 {
548 WANT (1); 746 WANT (1);
549 747
550 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
551 { 749 {
552 ++dec->cur; 750 ++dec->cur;
553 break; 751 break;
554 } 752 }
555 753
558 } 756 }
559 else 757 else
560 { 758 {
561 int i, len = decode_uint (dec); 759 int i, len = decode_uint (dec);
562 760
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
563 av_fill (av, len - 1); 762 av_fill (av, len - 1);
564 763
565 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
567 } 766 }
573 SvREFCNT_dec (av); 772 SvREFCNT_dec (av);
574 DEC_DEC_DEPTH; 773 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 774 return &PL_sv_undef;
576} 775}
577 776
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
578static SV * 823static SV *
579decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
580{ 825{
581 HV *hv = newHV (); 826 HV *hv = newHV ();
582 827
583 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
584 829
585 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 831 {
587 ++dec->cur; 832 ++dec->cur;
588 833
589 for (;;) 834 for (;;)
590 { 835 {
591 WANT (1); 836 WANT (1);
592 837
593 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
594 { 839 {
595 ++dec->cur; 840 ++dec->cur;
596 break; 841 break;
597 } 842 }
598 843
599 SV *k = decode_sv (dec); 844 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 845 }
605 } 846 }
606 else 847 else
607 { 848 {
608 int len = decode_uint (dec); 849 int pairs = decode_uint (dec);
609 850
610 while (len--) 851 while (pairs--)
611 { 852 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 853 }
619 854
620 DEC_DEC_DEPTH; 855 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 856 return newRV_noinc ((SV *)hv);
622 857
629static SV * 864static SV *
630decode_str (dec_t *dec, int utf8) 865decode_str (dec_t *dec, int utf8)
631{ 866{
632 SV *sv = 0; 867 SV *sv = 0;
633 868
634 if ((*dec->cur & 31) == 31) 869 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
635 { 870 {
871 // indefinite length strings
636 ++dec->cur; 872 ++dec->cur;
637 873
874 U8 major = *dec->cur & MAJOR_MISC;
875
638 sv = newSVpvn ("", 0); 876 sv = newSVpvn ("", 0);
639 877
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 878 for (;;)
642 { 879 {
643 WANT (1); 880 WANT (1);
644 881
645 if (*dec->cur == (0xe0 | 31)) 882 if ((*dec->cur - major) > LENGTH_EXT8)
883 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 884 {
647 ++dec->cur; 885 ++dec->cur;
648 break; 886 break;
649 } 887 }
888 else
889 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 890
651 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;
652 } 896 }
653 } 897 }
654 else 898 else
655 { 899 {
656 STRLEN len = decode_uint (dec); 900 STRLEN len = decode_uint (dec);
657 901
658 WANT (len); 902 WANT (len);
659 sv = newSVpvn (dec->cur, len); 903 sv = newSVpvn (dec->cur, len);
660 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));
661 } 909 }
662 910
663 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
664 SvUTF8_on (sv); 917 SvUTF8_on (sv);
918 }
665 919
666 return sv; 920 return sv;
667 921
668fail: 922fail:
669 SvREFCNT_dec (sv); 923 SvREFCNT_dec (sv);
671} 925}
672 926
673static SV * 927static SV *
674decode_tagged (dec_t *dec) 928decode_tagged (dec_t *dec)
675{ 929{
930 SV *sv = 0;
676 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:
677 SV *sv = decode_sv (dec); 938 sv = decode_sv (dec);
939 break;
678 940
679 if (tag == CBOR_TAG_MAGIC) 941 case CBOR_TAG_INDIRECTION:
680 return sv; 942 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 943 break;
682 { 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
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1017 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 1018 ERR ("corrupted CBOR data (non-array perl object)");
685 1019
686 AV *av = (AV *)SvRV (sv); 1020 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 1021 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1022 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 1023
690 if (!stash) 1024 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 1025 ERR ("cannot decode perl-object (package does not exist)");
692 1026
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1027 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 1028
695 if (!method) 1029 if (!method)
696 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)");
697 1031
698 dSP; 1032 dSP;
699 1033
700 ENTER; SAVETMPS; PUSHMARK (SP); 1034 ENTER; SAVETMPS;
1035 PUSHMARK (SP);
701 EXTEND (SP, len + 1); 1036 EXTEND (SP, len + 1);
702 // 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
703 PUSHs (*av_fetch (av, 0, 1)); 1038 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 1039 PUSHs (sv_cbor);
705 1040
706 int i; 1041 int i;
707 1042
708 for (i = 1; i < len; ++i) 1043 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 1044 PUSHs (*av_fetch (av, i, 1));
710 1045
711 PUTBACK; 1046 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1047 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1048 SPAGAIN;
714 1049
1050 if (SvTRUE (ERRSV))
1051 {
1052 FREETMPS; LEAVE;
1053 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1054 }
1055
1056 SvREFCNT_dec (sv);
715 sv = SvREFCNT_inc (POPs); 1057 sv = SvREFCNT_inc (POPs);
716 1058
717 PUTBACK; 1059 PUTBACK;
718 1060
719 FREETMPS; LEAVE; 1061 FREETMPS; LEAVE;
1062 }
1063 break;
720 1064
721 return sv; 1065 default:
722 } 1066 {
723 else 1067 sv = decode_sv (dec);
724 { 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 {
725 AV *av = newAV (); 1094 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1095 av_push (av, newSVuv (tag));
727 av_push (av, sv); 1096 av_push (av, sv);
728 1097
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1098 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1099 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1100 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 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;
734 } 1109 }
1110
1111 return sv;
735 1112
736fail: 1113fail:
737 SvREFCNT_dec (sv); 1114 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1115 return &PL_sv_undef;
739} 1116}
741static SV * 1118static SV *
742decode_sv (dec_t *dec) 1119decode_sv (dec_t *dec)
743{ 1120{
744 WANT (1); 1121 WANT (1);
745 1122
746 switch (*dec->cur >> 5) 1123 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1124 {
748 case 0: // unsigned int 1125 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1126 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1127 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1128 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1129 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1130 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1131 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1132
756 case 4: // array 1133 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1134 switch (*dec->cur++ & MINOR_MASK)
764 { 1135 {
765 case 20: 1136 case SIMPLE_FALSE:
766#if CBOR_SLOW 1137#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1138 types_false = get_bool ("Types::Serialiser::false");
768#endif 1139#endif
769 return newSVsv (types_false); 1140 return newSVsv (types_false);
770 case 21: 1141 case SIMPLE_TRUE:
771#if CBOR_SLOW 1142#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1143 types_true = get_bool ("Types::Serialiser::true");
773#endif 1144#endif
774 return newSVsv (types_true); 1145 return newSVsv (types_true);
775 case 22: 1146 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1147 return newSVsv (&PL_sv_undef);
777 case 23: 1148 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1149#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1150 types_error = get_bool ("Types::Serialiser::error");
780#endif 1151#endif
781 return newSVsv (types_error); 1152 return newSVsv (types_error);
782 1153
783 case 25: 1154 case MISC_FLOAT16:
784 { 1155 {
785 WANT (2); 1156 WANT (2);
786 1157
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1158 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1159 dec->cur += 2;
789 1160
790 return newSVnv (ecb_binary16_to_float (fp)); 1161 return newSVnv (ecb_binary16_to_float (fp));
791 } 1162 }
792 1163
793 case 26: 1164 case MISC_FLOAT32:
794 { 1165 {
795 uint32_t fp; 1166 uint32_t fp;
796 WANT (4); 1167 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1168 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1169 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1172 fp = ecb_bswap32 (fp);
802 1173
803 return newSVnv (ecb_binary32_to_float (fp)); 1174 return newSVnv (ecb_binary32_to_float (fp));
804 } 1175 }
805 1176
806 case 27: 1177 case MISC_FLOAT64:
807 { 1178 {
808 uint64_t fp; 1179 uint64_t fp;
809 WANT (8); 1180 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1181 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1182 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1185 fp = ecb_bswap64 (fp);
815 1186
816 return newSVnv (ecb_binary64_to_double (fp)); 1187 return newSVnv (ecb_binary64_to_double (fp));
817 } 1188 }
818 1189
819 // 0..19 unassigned 1190 // 0..19 unassigned simple
820 // 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
821 default: 1194 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1195 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
823 } 1196 }
824 1197
825 break; 1198 break;
826 } 1199 }
827 1200
830} 1203}
831 1204
832static SV * 1205static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1206decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1207{
835 dec_t dec; 1208 dec_t dec = { 0 };
836 SV *sv; 1209 SV *sv;
1210 STRLEN len;
1211 char *data = SvPVbyte (string, len);
837 1212
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1213 if (len > cbor->max_size && cbor->max_size)
867 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",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1215 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1216
873 dec.cbor = *cbor; 1217 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1218 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1219 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1220
879 sv = decode_sv (&dec); 1221 sv = decode_sv (&dec);
880 1222
881 if (offset_return) 1223 if (offset_return)
882 *offset_return = dec.cur; 1224 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1227 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1228 dec.err = "garbage after CBOR object";
887 1229
888 if (dec.err) 1230 if (dec.err)
889 { 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
890 SvREFCNT_dec (sv); 1243 SvREFCNT_dec (sv);
891 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);
892 } 1245 }
893 1246
894 sv = sv_2mortal (sv); 1247 sv = sv_2mortal (sv);
895 1248
896 return sv; 1249 return sv;
897} 1250}
898 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
899///////////////////////////////////////////////////////////////////////////// 1376/////////////////////////////////////////////////////////////////////////////
900// XS interface functions 1377// XS interface functions
901 1378
902MODULE = CBOR::XS PACKAGE = CBOR::XS 1379MODULE = CBOR::XS PACKAGE = CBOR::XS
903 1380
910 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1387 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
911 1388
912 types_true = get_bool ("Types::Serialiser::true" ); 1389 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1390 types_false = get_bool ("Types::Serialiser::false");
914 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);
915 1394
916 sv_cbor = newSVpv ("CBOR", 0); 1395 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1396 SvREADONLY_on (sv_cbor);
918} 1397}
919 1398
940 1419
941void shrink (CBOR *self, int enable = 1) 1420void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1421 ALIAS:
943 shrink = F_SHRINK 1422 shrink = F_SHRINK
944 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 utf8_strings = F_UTF8_STRINGS
1428 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1429 PPCODE:
946{ 1430{
947 if (enable) 1431 if (enable)
948 self->flags |= ix; 1432 self->flags |= ix;
949 else 1433 else
954 1438
955void get_shrink (CBOR *self) 1439void get_shrink (CBOR *self)
956 ALIAS: 1440 ALIAS:
957 get_shrink = F_SHRINK 1441 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1442 get_allow_unknown = F_ALLOW_UNKNOWN
1443 get_allow_sharing = F_ALLOW_SHARING
1444 get_allow_cycles = F_ALLOW_CYCLES
1445 get_pack_strings = F_PACK_STRINGS
1446 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1447 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1448 XPUSHs (boolSV (self->flags & ix));
961 1449
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1450void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1451 PPCODE:
979 CODE: 1467 CODE:
980 RETVAL = self->max_size; 1468 RETVAL = self->max_size;
981 OUTPUT: 1469 OUTPUT:
982 RETVAL 1470 RETVAL
983 1471
984#if 0 //TODO 1472void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1473 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1474 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1475 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1476 XPUSHs (ST (0));
993}
994 1477
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1478SV *get_filter (CBOR *self)
996 PPCODE: 1479 CODE:
997{ 1480 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1481 OUTPUT:
999 self->cb_sk_object = newHV (); 1482 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1483
1019void encode (CBOR *self, SV *scalar) 1484void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1485 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1486 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1487 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1500 EXTEND (SP, 2);
1036 PUSHs (sv); 1501 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1502 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1503}
1039 1504
1040#if 0 1505void incr_parse (CBOR *self, SV *cborstr)
1506 ALIAS:
1507 incr_parse_multiple = 1
1508 PPCODE:
1509{
1510 if (SvUTF8 (cborstr))
1511 sv_utf8_downgrade (cborstr, 0);
1512
1513 if (!self->incr_count)
1514 {
1515 self->incr_count = newAV ();
1516 self->incr_pos = 0;
1517 self->incr_need = 1;
1518
1519 av_push (self->incr_count, newSViv (1));
1520 }
1521
1522 do
1523 {
1524 if (!incr_parse (self, cborstr))
1525 {
1526 if (self->incr_need > self->max_size && self->max_size)
1527 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1528 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1529
1530 break;
1531 }
1532
1533 SV *sv;
1534 char *offset;
1535
1536 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1537 XPUSHs (sv);
1538
1539 sv_chop (cborstr, offset);
1540
1541 av_clear (self->incr_count);
1542 av_push (self->incr_count, newSViv (1));
1543
1544 self->incr_pos = 0;
1545 self->incr_need = self->incr_pos + 1;
1546 }
1547 while (ix);
1548}
1549
1550void incr_reset (CBOR *self)
1551 CODE:
1552{
1553 SvREFCNT_dec (self->incr_count);
1554 self->incr_count = 0;
1555}
1041 1556
1042void DESTROY (CBOR *self) 1557void DESTROY (CBOR *self)
1043 CODE: 1558 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1559 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1560
1049PROTOTYPES: ENABLE 1561PROTOTYPES: ENABLE
1050 1562
1051void encode_cbor (SV *scalar) 1563void encode_cbor (SV *scalar)
1564 ALIAS:
1565 encode_cbor = 0
1566 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1567 PPCODE:
1053{ 1568{
1054 CBOR cbor; 1569 CBOR cbor;
1055 cbor_init (&cbor); 1570 cbor_init (&cbor);
1571 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1572 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1573 XPUSHs (scalar);
1058} 1574}
1059 1575
1060void decode_cbor (SV *cborstr) 1576void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines