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 1.4 by root, Sat Oct 26 21:06:37 2013 UTC vs.
Revision 1.39 by root, Sun Dec 1 14:45:03 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
64
65// known tags
66enum cbor_tag
67{
68 // extensions
69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
76
77 // rfc7049
78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
80 CBOR_TAG_POS_BIGNUM = 2, // byte string
81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
83 CBOR_TAG_BIGFLOAT = 5, // array
84
85 CBOR_TAG_CONV_B64U = 21, // base64url, any
86 CBOR_TAG_CONV_B64 = 22, // base64, any
87 CBOR_TAG_CONV_HEX = 23, // base16, any
88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
89
90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97};
98
14#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
16 105
17#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
18 107
19#define SB do { 108#define SB do {
20#define SE } while (0) 109#define SE } while (0)
21
22#if __GNUC__ >= 3
23# define expect(expr,value) __builtin_expect ((expr), (value))
24# define INLINE static inline
25#else
26# define expect(expr,value) (expr)
27# define INLINE static
28#endif
29
30#define expect_false(expr) expect ((expr) != 0, 0)
31#define expect_true(expr) expect ((expr) != 0, 1)
32 110
33#define IN_RANGE_INC(type,val,beg,end) \ 111#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 112 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 113 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 114
42#else 120#else
43# define CBOR_SLOW 0 121# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
45#endif 123#endif
46 124
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 127
50typedef struct { 128typedef struct {
51 U32 flags; 129 U32 flags;
52 U32 max_depth; 130 U32 max_depth;
53 STRLEN max_size; 131 STRLEN max_size;
54 132 SV *filter;
55 SV *cb_object;
56 HV *cb_sk_object;
57} CBOR; 133} CBOR;
58 134
59INLINE void 135ecb_inline void
60cbor_init (CBOR *cbor) 136cbor_init (CBOR *cbor)
61{ 137{
62 Zero (cbor, 1, CBOR); 138 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 139 cbor->max_depth = 512;
64} 140}
65 141
142ecb_inline void
143cbor_free (CBOR *cbor)
144{
145 SvREFCNT_dec (cbor->filter);
146}
147
66///////////////////////////////////////////////////////////////////////////// 148/////////////////////////////////////////////////////////////////////////////
67// utility functions 149// utility functions
68 150
69INLINE SV * 151ecb_inline SV *
70get_bool (const char *name) 152get_bool (const char *name)
71{ 153{
72 SV *sv = get_sv (name, 1); 154 SV *sv = get_sv (name, 1);
73 155
74 SvREADONLY_on (sv); 156 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 157 SvREADONLY_on (SvRV (sv));
76 158
77 return sv; 159 return sv;
78} 160}
79 161
80INLINE void 162ecb_inline void
81shrink (SV *sv) 163shrink (SV *sv)
82{ 164{
83 sv_utf8_downgrade (sv, 1); 165 sv_utf8_downgrade (sv, 1);
84 166
85 if (SvLEN (sv) > SvCUR (sv) + 1) 167 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 172 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 173#endif
92 } 174 }
93} 175}
94 176
95///////////////////////////////////////////////////////////////////////////// 177// minimum length of a string to be registered for stringref
96// fp hell 178ecb_inline int
97 179minimum_string_length (UV idx)
98//TODO 180{
181 return idx > 23
182 ? idx > 0xffU
183 ? idx > 0xffffU
184 ? idx > 0xffffffffU
185 ? 11
186 : 7
187 : 5
188 : 4
189 : 3;
190}
99 191
100///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
101// encoder 193// encoder
102 194
103// structure used for encoding CBOR 195// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
108 SV *sv; // result scalar 200 SV *sv; // result scalar
109 CBOR cbor; 201 CBOR cbor;
110 U32 depth; // recursion level 202 U32 depth; // recursion level
203 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
204 UV stringref_idx;
205 HV *shareable; // ptr => index, or 0
206 UV shareable_idx;
111} enc_t; 207} enc_t;
112 208
113INLINE void 209ecb_inline void
114need (enc_t *enc, STRLEN len) 210need (enc_t *enc, STRLEN len)
115{ 211{
116 if (expect_false (enc->cur + len >= enc->end)) 212 if (ecb_expect_false (enc->cur + len >= enc->end))
117 { 213 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 214 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 215 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 216 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 217 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 218 }
123} 219}
124 220
125INLINE void 221ecb_inline void
126encode_ch (enc_t *enc, char ch) 222encode_ch (enc_t *enc, char ch)
127{ 223{
128 need (enc, 1); 224 need (enc, 1);
129 *enc->cur++ = ch; 225 *enc->cur++ = ch;
130} 226}
132static void 228static void
133encode_uint (enc_t *enc, int major, UV len) 229encode_uint (enc_t *enc, int major, UV len)
134{ 230{
135 need (enc, 9); 231 need (enc, 9);
136 232
137 if (len < 24) 233 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
139 else if (len <= 0xff) 235 else if (ecb_expect_true (len <= 0xffU))
140 { 236 {
141 *enc->cur++ = major | 24; 237 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 238 *enc->cur++ = len;
143 } 239 }
144 else if (len <= 0xffff) 240 else if (len <= 0xffffU)
145 { 241 {
146 *enc->cur++ = major | 25; 242 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 243 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 244 *enc->cur++ = len;
149 } 245 }
150 else if (len <= 0xffffffff) 246 else if (len <= 0xffffffffU)
151 { 247 {
152 *enc->cur++ = major | 26; 248 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 249 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 250 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 252 *enc->cur++ = len;
157 } 253 }
158 else 254 else
159 { 255 {
160 *enc->cur++ = major | 27; 256 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 257 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 258 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 259 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 260 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 261 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 263 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 264 *enc->cur++ = len;
169 } 265 }
170} 266}
171 267
172static void 268ecb_inline void
269encode_tag (enc_t *enc, UV tag)
270{
271 encode_uint (enc, MAJOR_TAG, tag);
272}
273
274ecb_inline void
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 275encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 276{
175 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
176 need (enc, len); 278 need (enc, len);
177 memcpy (enc->cur, str, len); 279 memcpy (enc->cur, str, len);
178 enc->cur += len; 280 enc->cur += len;
179} 281}
180 282
283static void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
285{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289
290 if (SvOK (*svp))
291 {
292 // already registered, use stringref
293 encode_tag (enc, CBOR_TAG_STRINGREF);
294 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
295 return;
296 }
297 else if (len >= minimum_string_length (enc->stringref_idx))
298 {
299 // register only
300 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx;
302 }
303 }
304
305 encode_str (enc, utf8, str, len);
306}
307
181static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
182 309
183static void 310static void
184encode_av (enc_t *enc, AV *av) 311encode_av (enc_t *enc, AV *av)
185{ 312{
188 if (enc->depth >= enc->cbor.max_depth) 315 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 316 croak (ERR_NESTING_EXCEEDED);
190 317
191 ++enc->depth; 318 ++enc->depth;
192 319
193 encode_uint (enc, 0x80, len + 1); 320 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 321
195 for (i = 0; i <= len; ++i) 322 for (i = 0; i <= len; ++i)
196 { 323 {
197 SV **svp = av_fetch (av, i, 0); 324 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 325 encode_sv (enc, svp ? *svp : &PL_sv_undef);
213 340
214 int pairs = hv_iterinit (hv); 341 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 342 int mg = SvMAGICAL (hv);
216 343
217 if (mg) 344 if (mg)
218 encode_ch (enc, 0xa0 | 31); 345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 346 else
220 encode_uint (enc, 0xa0, pairs); 347 encode_uint (enc, MAJOR_MAP, pairs);
221 348
222 while ((he = hv_iternext (hv))) 349 while ((he = hv_iternext (hv)))
223 { 350 {
224 if (HeKLEN (he) == HEf_SVKEY) 351 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 352 encode_sv (enc, HeSVKEY (he));
226 else 353 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 354 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 355
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 356 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 357 }
231 358
232 if (mg) 359 if (mg)
233 encode_ch (enc, 0xe0 | 31); 360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 361
235 --enc->depth; 362 --enc->depth;
236} 363}
237 364
238// encode objects, arrays and special \0=false and \1=true values. 365// encode objects, arrays and special \0=false and \1=true values.
239static void 366static void
240encode_rv (enc_t *enc, SV *sv) 367encode_rv (enc_t *enc, SV *sv)
241{ 368{
242 svtype svt;
243
244 SvGETMAGIC (sv); 369 SvGETMAGIC (sv);
370
245 svt = SvTYPE (sv); 371 svtype svt = SvTYPE (sv);
246 372
247 if (expect_false (SvOBJECT (sv))) 373 if (ecb_expect_false (SvOBJECT (sv)))
248 { 374 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 375 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 376 ? types_boolean_stash
377 : gv_stashpv ("Types::Serialiser::Boolean", 1);
378 HV *error_stash = !CBOR_SLOW || types_error_stash
379 ? types_error_stash
380 : gv_stashpv ("Types::Serialiser::Error", 1);
381 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
382 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 383 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 384
253 if (SvSTASH (sv) == stash) 385 HV *stash = SvSTASH (sv);
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 386
387 if (stash == boolean_stash)
388 {
389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
390 return;
391 }
392 else if (stash == error_stash)
393 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 return;
396 }
397 else if (stash == tagged_stash)
398 {
399 if (svt != SVt_PVAV)
400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
401
402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
404
405 return;
406 }
407 }
408
409 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 {
412 if (!enc->shareable)
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416
417 if (SvOK (*svp))
418 {
419 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
420 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
421 return;
422 }
255 else 423 else
256 { 424 {
257#if 0 //TODO 425 sv_setuv (*svp, enc->shareable_idx);
258 if (enc->cbor.flags & F_CONV_BLESSED) 426 ++enc->shareable_idx;
427 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
428 }
429 }
430
431 if (ecb_expect_false (SvOBJECT (sv)))
432 {
433 HV *stash = SvSTASH (sv);
434 GV *method;
435
436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259 { 437 {
438 dSP;
439
440 ENTER; SAVETMPS; PUSHMARK (SP);
260 // we re-bless the reference to get overload and other niceties right 441 // we re-bless the reference to get overload and other niceties right
261 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
262
263 if (to_cbor)
264 {
265 dSP;
266
267 ENTER; SAVETMPS; PUSHMARK (SP);
268 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 442 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 443
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 444 PUTBACK;
445 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 446 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 447 SPAGAIN;
274 448
275 // catch this surprisingly common error 449 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 450 if (SvROK (TOPs) && SvRV (TOPs) == sv)
277 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 451 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 452
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 453 encode_sv (enc, POPs);
283 454
455 PUTBACK;
456
284 FREETMPS; LEAVE; 457 FREETMPS; LEAVE;
285 }
286 else if (enc->cbor.flags & F_ALLOW_BLESSED)
287 encode_str (enc, "null", 4, 0);
288 else
289 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
290 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
291 } 458 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 459 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
293 encode_str (enc, "null", 4, 0); 460 {
461 dSP;
462
463 ENTER; SAVETMPS; PUSHMARK (SP);
464 EXTEND (SP, 2);
465 // we re-bless the reference to get overload and other niceties right
466 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
467 PUSHs (sv_cbor);
468
469 PUTBACK;
470 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
471 SPAGAIN;
472
473 // catch this surprisingly common error
474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
476
477 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
478 encode_uint (enc, MAJOR_ARRAY, count + 1);
479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
480
481 while (count)
482 encode_sv (enc, SP[1 - count--]);
483
484 PUTBACK;
485
486 FREETMPS; LEAVE;
487 }
294 else 488 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 489 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 490 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 491 }
300 else if (svt == SVt_PVHV) 492 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 493 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 494 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 495 encode_av (enc, (AV *)sv);
304 else if (svt < SVt_PVAV)
305 {
306 STRLEN len = 0;
307 char *pv = svt ? SvPV (sv, len) : 0;
308
309 if (len == 1 && *pv == '1')
310 encode_ch (enc, 0xe0 | 21);
311 else if (len == 1 && *pv == '0')
312 encode_ch (enc, 0xe0 | 20);
313 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
314 encode_ch (enc, 0xe0 | 23);
315 else
316 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
317 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
318 }
319 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
320 encode_ch (enc, 0xe0 | 23);
321 else 496 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 497 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 498 encode_tag (enc, CBOR_TAG_INDIRECTION);
499 encode_sv (enc, sv);
500 }
324} 501}
325 502
326static void 503static void
327encode_nv (enc_t *enc, SV *sv) 504encode_nv (enc_t *enc, SV *sv)
328{ 505{
329 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
330 507
331 need (enc, 9); 508 need (enc, 9);
332 509
333 if (expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 512 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
337 { 514 {
338 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
339 516
340 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
341 518
342 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
344 521
345 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
347 } 524 }
348 else 525 else
349 { 526 {
350 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
351 528
352 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
353 530
354 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
356 533
357 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
366 543
367 if (SvPOKp (sv)) 544 if (SvPOKp (sv))
368 { 545 {
369 STRLEN len; 546 STRLEN len;
370 char *str = SvPV (sv, len); 547 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 548 encode_strref (enc, SvUTF8 (sv), str, len);
372 } 549 }
373 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 551 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
376 { 553 {
377 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
379 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
381 else 558 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
383 } 560 }
384 else if (SvROK (sv)) 561 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 567 else
391 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 568 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 570}
394 571
395static SV * 572static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 573encode_cbor (SV *scalar, CBOR *cbor)
397{ 574{
398 enc_t enc; 575 enc_t enc = { };
399 576
400 enc.cbor = *cbor; 577 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 579 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 580 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 581
406 SvPOK_only (enc.sv); 582 SvPOK_only (enc.sv);
583
584 if (cbor->flags & F_PACK_STRINGS)
585 {
586 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
587 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
588 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
589 }
590
407 encode_sv (&enc, scalar); 591 encode_sv (&enc, scalar);
408 592
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 593 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 594 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411 595
425 U8 *end; // end of input string 609 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 610 const char *err; // parse error, if != 0
427 CBOR cbor; 611 CBOR cbor;
428 U32 depth; // recursion depth 612 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 613 U32 maxdepth; // recursion depth limit
614 AV *shareable;
615 AV *stringref;
616 SV *decode_tagged;
430} dec_t; 617} dec_t;
431 618
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 619#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
433 620
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 621#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
435 622
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 623#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
438 625
439static UV 626static UV
440decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
441{ 628{
442 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
443 { 630 ++dec->cur;
444 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
445 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
446 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
447 return *dec->cur++ & 31;
448 631
449 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
450 WANT (2); 636 WANT (1);
451 dec->cur += 2; 637 dec->cur += 1;
452 return dec->cur[-1]; 638 return dec->cur[-1];
453 639 }
454 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
455 WANT (3); 642 WANT (2);
456 dec->cur += 3; 643 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
459 646 }
460 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
461 WANT (5); 649 WANT (4);
462 dec->cur += 5; 650 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
467 655 }
468 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
469 WANT (9); 658 WANT (8);
470 dec->cur += 9; 659 dec->cur += 8;
660
661 return
662#if UVSIZE < 8
663 0
664#else
471 return (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
669#endif
475 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
479 674 }
480 default: 675 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 677
484fail: 678fail:
485 return 0; 679 return 0;
486} 680}
487 681
492{ 686{
493 AV *av = newAV (); 687 AV *av = newAV ();
494 688
495 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
496 690
497 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 692 {
499 ++dec->cur; 693 ++dec->cur;
500 694
501 for (;;) 695 for (;;)
502 { 696 {
503 WANT (1); 697 WANT (1);
504 698
505 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
506 { 700 {
507 ++dec->cur; 701 ++dec->cur;
508 break; 702 break;
509 } 703 }
510 704
513 } 707 }
514 else 708 else
515 { 709 {
516 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
517 711
712 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
518 av_fill (av, len - 1); 713 av_fill (av, len - 1);
519 714
520 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
522 } 717 }
528 SvREFCNT_dec (av); 723 SvREFCNT_dec (av);
529 DEC_DEC_DEPTH; 724 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 725 return &PL_sv_undef;
531} 726}
532 727
728static void
729decode_he (dec_t *dec, HV *hv)
730{
731 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref
733
734 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 {
737 I32 len = decode_uint (dec);
738 char *key = (char *)dec->cur;
739
740 dec->cur += len;
741
742 hv_store (hv, key, len, decode_sv (dec), 0);
743
744 return;
745 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756
757 hv_store (hv, key, -len, decode_sv (dec), 0);
758
759 return;
760 }
761
762 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec);
764
765 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k);
767
768fail:
769 ;
770}
771
533static SV * 772static SV *
534decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
535{ 774{
536 HV *hv = newHV (); 775 HV *hv = newHV ();
537 776
538 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
539 778
540 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 780 {
542 ++dec->cur; 781 ++dec->cur;
543 782
544 for (;;) 783 for (;;)
545 { 784 {
546 WANT (1); 785 WANT (1);
547 786
548 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
549 { 788 {
550 ++dec->cur; 789 ++dec->cur;
551 break; 790 break;
552 } 791 }
553 792
554 SV *k = decode_sv (dec); 793 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 794 }
559 } 795 }
560 else 796 else
561 { 797 {
562 int len = decode_uint (dec); 798 int pairs = decode_uint (dec);
563 799
564 while (len--) 800 while (pairs--)
565 { 801 decode_he (dec, hv);
566 SV *k = decode_sv (dec);
567 SV *v = decode_sv (dec);
568
569 hv_store_ent (hv, k, v, 0);
570 }
571 } 802 }
572 803
573 DEC_DEC_DEPTH; 804 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 805 return newRV_noinc ((SV *)hv);
575
576#if 0
577 SV *sv;
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH;
581 decode_ws (dec);
582
583 for (;;)
584 {
585 // heuristic: assume that
586 // a) decode_str + hv_store_ent are abysmally slow.
587 // b) most hash keys are short, simple ascii text.
588 // => try to "fast-match" such strings to avoid
589 // the overhead of decode_str + hv_store_ent.
590 {
591 SV *value;
592 char *p = dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594
595 for (;;)
596 {
597 // the >= 0x80 is false on most architectures
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
599 {
600 // slow path, back up and use decode_str
601 SV *key = decode_str (dec);
602 if (!key)
603 goto fail;
604
605 decode_ws (dec); EXPECT_CH (':');
606
607 decode_ws (dec);
608 value = decode_sv (dec);
609 if (!value)
610 {
611 SvREFCNT_dec (key);
612 goto fail;
613 }
614
615 hv_store_ent (hv, key, value, 0);
616 SvREFCNT_dec (key);
617
618 break;
619 }
620 else if (*p == '"')
621 {
622 // fast path, got a simple key
623 char *key = dec->cur;
624 int len = p - key;
625 dec->cur = p + 1;
626
627 decode_ws (dec); EXPECT_CH (':');
628
629 decode_ws (dec);
630 value = decode_sv (dec);
631 if (!value)
632 goto fail;
633
634 hv_store (hv, key, len, value, 0);
635
636 break;
637 }
638
639 ++p;
640 }
641 }
642
643 decode_ws (dec);
644
645 if (*dec->cur == '}')
646 {
647 ++dec->cur;
648 break;
649 }
650
651 if (*dec->cur != ',')
652 ERR (", or } expected while parsing object/hash");
653
654 ++dec->cur;
655
656 decode_ws (dec);
657
658 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
659 {
660 ++dec->cur;
661 break;
662 }
663 }
664
665 DEC_DEC_DEPTH;
666 sv = newRV_noinc ((SV *)hv);
667
668 // check filter callbacks
669 if (dec->cbor.flags & F_HOOK)
670 {
671 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
672 {
673 HE *cb, *he;
674
675 hv_iterinit (hv);
676 he = hv_iternext (hv);
677 hv_iterinit (hv);
678
679 // the next line creates a mortal sv each time its called.
680 // might want to optimise this for common cases.
681 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
682
683 if (cb)
684 {
685 dSP;
686 int count;
687
688 ENTER; SAVETMPS; PUSHMARK (SP);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691
692 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
693
694 if (count == 1)
695 {
696 sv = newSVsv (POPs);
697 FREETMPS; LEAVE;
698 return sv;
699 }
700
701 SvREFCNT_inc (sv);
702 FREETMPS; LEAVE;
703 }
704 }
705
706 if (dec->cbor.cb_object)
707 {
708 dSP;
709 int count;
710
711 ENTER; SAVETMPS; PUSHMARK (SP);
712 XPUSHs (sv_2mortal (sv));
713
714 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
715
716 if (count == 1)
717 {
718 sv = newSVsv (POPs);
719 FREETMPS; LEAVE;
720 return sv;
721 }
722
723 SvREFCNT_inc (sv);
724 FREETMPS; LEAVE;
725 }
726 }
727
728 return sv;
729#endif
730 806
731fail: 807fail:
732 SvREFCNT_dec (hv); 808 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH; 809 DEC_DEC_DEPTH;
734 return &PL_sv_undef; 810 return &PL_sv_undef;
735} 811}
736 812
737static SV * 813static SV *
738decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
739{ 815{
740 SV *sv; 816 SV *sv = 0;
741 817
742 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
743 { 819 {
820 // indefinite length strings
744 ++dec->cur; 821 ++dec->cur;
745 822
823 U8 major = *dec->cur & MAJOR_MISC;
824
746 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
747 826
748 // not very fast, and certainly not robust against illegal input
749 for (;;) 827 for (;;)
750 { 828 {
751 WANT (1); 829 WANT (1);
752 830
753 if (*dec->cur == (0xe0 | 31)) 831 if ((*dec->cur - major) > LENGTH_EXT8)
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
754 { 833 {
755 ++dec->cur; 834 ++dec->cur;
756 break; 835 break;
757 } 836 }
837 else
838 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
758 839
759 SV *sv2 = decode_sv (dec); 840 STRLEN len = decode_uint (dec);
760 sv_catsv (sv, sv2); 841
842 WANT (len);
843 sv_catpvn (sv, dec->cur, len);
844 dec->cur += len;
761 } 845 }
762 } 846 }
763 else 847 else
764 { 848 {
765 STRLEN len = decode_uint (dec); 849 STRLEN len = decode_uint (dec);
766 850
767 WANT (len); 851 WANT (len);
768 sv = newSVpvn (dec->cur, len); 852 sv = newSVpvn (dec->cur, len);
769 dec->cur += len; 853 dec->cur += len;
854
855 if (ecb_expect_false (dec->stringref)
856 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
857 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 858 }
771 859
772 if (utf8) 860 if (utf8)
861 {
862 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
863 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
864 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
865
773 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
774 868
775 return sv; 869 return sv;
776 870
777fail: 871fail:
872 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 873 return &PL_sv_undef;
779} 874}
780 875
781static SV * 876static SV *
782decode_tagged (dec_t *dec) 877decode_tagged (dec_t *dec)
783{ 878{
879 SV *sv = 0;
784 UV tag = decode_uint (dec); 880 UV tag = decode_uint (dec);
881
882 WANT (1);
883
884 switch (tag)
885 {
886 case CBOR_TAG_MAGIC:
785 SV *sv = decode_sv (dec); 887 sv = decode_sv (dec);
888 break;
786 889
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 890 case CBOR_TAG_INDIRECTION:
891 sv = newRV_noinc (decode_sv (dec));
892 break;
893
894 case CBOR_TAG_STRINGREF_NAMESPACE:
895 {
896 ENTER; SAVETMPS;
897
898 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900
901 sv = decode_sv (dec);
902
903 FREETMPS; LEAVE;
904 }
905 break;
906
907 case CBOR_TAG_STRINGREF:
908 {
909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
911
912 UV idx = decode_uint (dec);
913
914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
915 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
916
917 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
918 }
919 break;
920
921 case CBOR_TAG_VALUE_SHAREABLE:
922 {
923 if (ecb_expect_false (!dec->shareable))
924 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
925
926 if (dec->cbor.flags & F_ALLOW_CYCLES)
927 {
928 sv = newSV (0);
929 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
930
931 SV *osv = decode_sv (dec);
932 sv_setsv (sv, osv);
933 SvREFCNT_dec_NN (osv);
934 }
935 else
936 {
937 av_push (dec->shareable, &PL_sv_undef);
938 int idx = AvFILLp (dec->shareable);
939 sv = decode_sv (dec);
940 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
941 }
942 }
943 break;
944
945 case CBOR_TAG_VALUE_SHAREDREF:
946 {
947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
949
950 UV idx = decode_uint (dec);
951
952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
953 ERR ("corrupted CBOR data (sharedref index out of bounds)");
954
955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956
957 if (sv == &PL_sv_undef)
958 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
959 }
960 break;
961
962 case CBOR_TAG_PERL_OBJECT:
963 {
964 sv = decode_sv (dec);
965
966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
967 ERR ("corrupted CBOR data (non-array perl object)");
968
969 AV *av = (AV *)SvRV (sv);
970 int len = av_len (av) + 1;
971 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
972
973 if (!stash)
974 ERR ("cannot decode perl-object (package does not exist)");
975
976 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
977
978 if (!method)
979 ERR ("cannot decode perl-object (package does not have a THAW method)");
980
981 dSP;
982
983 ENTER; SAVETMPS; PUSHMARK (SP);
984 EXTEND (SP, len + 1);
985 // we re-bless the reference to get overload and other niceties right
986 PUSHs (*av_fetch (av, 0, 1));
987 PUSHs (sv_cbor);
988
989 int i;
990
991 for (i = 1; i < len; ++i)
992 PUSHs (*av_fetch (av, i, 1));
993
994 PUTBACK;
995 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
996 SPAGAIN;
997
998 if (SvTRUE (ERRSV))
999 {
1000 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1002 }
1003
1004 SvREFCNT_dec (sv);
1005 sv = SvREFCNT_inc (POPs);
1006
1007 PUTBACK;
1008
1009 FREETMPS; LEAVE;
1010 }
1011 break;
1012
1013 default:
1014 {
1015 sv = decode_sv (dec);
1016
1017 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP);
1019 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag));
1021 PUSHs (sv);
1022
1023 PUTBACK;
1024 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1025 SPAGAIN;
1026
1027 if (SvTRUE (ERRSV))
1028 {
1029 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1031 }
1032
1033 if (count)
1034 {
1035 SvREFCNT_dec (sv);
1036 sv = SvREFCNT_inc (POPs);
1037 }
1038 else
1039 {
1040 AV *av = newAV ();
1041 av_push (av, newSVuv (tag));
1042 av_push (av, sv);
1043
1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1045 ? cbor_tagged_stash
1046 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1047 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1048 }
1049
1050 PUTBACK;
1051
1052 FREETMPS; LEAVE;
1053 }
1054 break;
1055 }
1056
788 return sv; 1057 return sv;
789 1058
790 AV *av = newAV (); 1059fail:
791 av_push (av, newSVuv (tag)); 1060 SvREFCNT_dec (sv);
792 av_push (av, sv); 1061 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1062}
795 1063
796static SV * 1064static SV *
797decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
798{ 1066{
799 WANT (1); 1067 WANT (1);
800 1068
801 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1070 {
803 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1078
811 return decode_str (dec, 1); 1079 case MAJOR_MISC >> MAJOR_SHIFT:
812 case 4: // array
813 return decode_av (dec);
814 case 5: // map
815 return decode_hv (dec);
816 case 6: // tag
817 return decode_tagged (dec);
818 case 7: // misc
819 switch (*dec->cur++ & 31) 1080 switch (*dec->cur++ & MINOR_MASK)
820 { 1081 {
821 case 20: 1082 case SIMPLE_FALSE:
822#if CBOR_SLOW 1083#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
824#endif 1085#endif
825 return newSVsv (cbor_false); 1086 return newSVsv (types_false);
826 case 21: 1087 case SIMPLE_TRUE:
827#if CBOR_SLOW 1088#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
829#endif 1090#endif
830 return newSVsv (cbor_true); 1091 return newSVsv (types_true);
831 case 22: 1092 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
1094 case SIMPLE_UNDEF:
1095#if CBOR_SLOW
1096 types_error = get_bool ("Types::Serialiser::error");
1097#endif
1098 return newSVsv (types_error);
833 1099
834 case 25: 1100 case MISC_FLOAT16:
835 { 1101 {
836 WANT (2); 1102 WANT (2);
837 1103
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1105 dec->cur += 2;
840 1106
841 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
842 } 1108 }
843 1109
844 case 26: 1110 case MISC_FLOAT32:
845 { 1111 {
846 uint32_t fp; 1112 uint32_t fp;
847 WANT (4); 1113 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1115 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
853 1119
854 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
855 } 1121 }
856 1122
857 case 27: 1123 case MISC_FLOAT64:
858 { 1124 {
859 uint64_t fp; 1125 uint64_t fp;
860 WANT (8); 1126 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1128 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
866 1132
867 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
868 } 1134 }
869 1135
870 // 0..19 unassigned 1136 // 0..19 unassigned simple
871 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
872 default: 1138 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
874 } 1140 }
875 1141
876 break; 1142 break;
877 } 1143 }
878#if 0
879 switch (*dec->cur)
880 {
881 //case '"': ++dec->cur; return decode_str (dec);
882 case '[': ++dec->cur; return decode_av (dec);
883 case '{': ++dec->cur; return decode_hv (dec);
884
885 case '-':
886 case '0': case '1': case '2': case '3': case '4':
887 case '5': case '6': case '7': case '8': case '9':
888 //TODO return decode_num (dec);
889
890 case 't':
891 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
892 {
893 dec->cur += 4;
894#if CBOR_SLOW
895 cbor_true = get_bool ("CBOR::XS::true");
896#endif
897 return newSVsv (cbor_true);
898 }
899 else
900 ERR ("'true' expected");
901
902 break;
903
904 case 'f':
905 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
906 {
907 dec->cur += 5;
908#if CBOR_SLOW
909 cbor_false = get_bool ("CBOR::XS::false");
910#endif
911 return newSVsv (cbor_false);
912 }
913 else
914 ERR ("'false' expected");
915
916 break;
917
918 case 'n':
919 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
920 {
921 dec->cur += 4;
922 return newSVsv (&PL_sv_undef);
923 }
924 else
925 ERR ("'null' expected");
926
927 break;
928
929 default:
930 ERR ("malformed CBOR string, neither array, object, number, string or atom");
931 break;
932 }
933#endif
934 1144
935fail: 1145fail:
936 return &PL_sv_undef; 1146 return &PL_sv_undef;
937} 1147}
938 1148
939static SV * 1149static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1150decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1151{
942 dec_t dec; 1152 dec_t dec = { };
943 SV *sv; 1153 SV *sv;
1154 STRLEN len;
1155 char *data = SvPVbyte (string, len);
944 1156
945 /* work around bugs in 5.10 where manipulating magic values
946 * makes perl ignore the magic in subsequent accesses.
947 * also make a copy of non-PV values, to get them into a clean
948 * state (SvPV should do that, but it's buggy, see below).
949 */
950 /*SvGETMAGIC (string);*/
951 if (SvMAGICAL (string) || !SvPOK (string))
952 string = sv_2mortal (newSVsv (string));
953
954 SvUPGRADE (string, SVt_PV);
955
956 /* work around a bug in perl 5.10, which causes SvCUR to fail an
957 * assertion with -DDEBUGGING, although SvCUR is documented to
958 * return the xpv_cur field which certainly exists after upgrading.
959 * according to nicholas clark, calling SvPOK fixes this.
960 * But it doesn't fix it, so try another workaround, call SvPV_nolen
961 * and hope for the best.
962 * Damnit, SvPV_nolen still trips over yet another assertion. This
963 * assertion business is seriously broken, try yet another workaround
964 * for the broken -DDEBUGGING.
965 */
966 {
967#ifdef DEBUGGING
968 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
969#else
970 STRLEN offset = SvCUR (string);
971#endif
972
973 if (offset > cbor->max_size && cbor->max_size) 1157 if (len > cbor->max_size && cbor->max_size)
974 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1158 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
975 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1159 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1160
980 dec.cbor = *cbor; 1161 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1162 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1163 dec.end = (U8 *)data + len;
983 dec.err = 0;
984 dec.depth = 0;
985
986 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
987 ;//TODO dec.cbor.flags |= F_HOOK;
988 1164
989 sv = decode_sv (&dec); 1165 sv = decode_sv (&dec);
990 1166
991 if (offset_return) 1167 if (offset_return)
992 *offset_return = dec.cur; 1168 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1171 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
997 1173
998 if (dec.err) 1174 if (dec.err)
999 { 1175 {
1176 if (dec.shareable)
1177 {
1178 // need to break cyclic links, which whould all be in shareable
1179 int i;
1180 SV **svp;
1181
1182 for (i = av_len (dec.shareable) + 1; i--; )
1183 if ((svp = av_fetch (dec.shareable, i, 0)))
1184 sv_setsv (*svp, &PL_sv_undef);
1185 }
1186
1000 SvREFCNT_dec (sv); 1187 SvREFCNT_dec (sv);
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1188 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002 } 1189 }
1003 1190
1004 sv = sv_2mortal (sv); 1191 sv = sv_2mortal (sv);
1005 1192
1006 return sv; 1193 return sv;
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1199MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1200
1014BOOT: 1201BOOT:
1015{ 1202{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1203 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1204 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1205
1019 cbor_true = get_bool ("CBOR::XS::true"); 1206 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1207 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1208
1209 types_true = get_bool ("Types::Serialiser::true" );
1210 types_false = get_bool ("Types::Serialiser::false");
1211 types_error = get_bool ("Types::Serialiser::error");
1212
1213 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1214
1215 sv_cbor = newSVpv ("CBOR", 0);
1216 SvREADONLY_on (sv_cbor);
1021} 1217}
1022 1218
1023PROTOTYPES: DISABLE 1219PROTOTYPES: DISABLE
1024 1220
1025void CLONE (...) 1221void CLONE (...)
1026 CODE: 1222 CODE:
1027 cbor_stash = 0; 1223 cbor_stash = 0;
1224 cbor_tagged_stash = 0;
1225 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1226 types_boolean_stash = 0;
1029 1227
1030void new (char *klass) 1228void new (char *klass)
1031 PPCODE: 1229 PPCODE:
1032{ 1230{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1231 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1239
1042void shrink (CBOR *self, int enable = 1) 1240void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1241 ALIAS:
1044 shrink = F_SHRINK 1242 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1243 allow_unknown = F_ALLOW_UNKNOWN
1244 allow_sharing = F_ALLOW_SHARING
1245 allow_cycles = F_ALLOW_CYCLES
1246 pack_strings = F_PACK_STRINGS
1247 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1248 PPCODE:
1047{ 1249{
1048 if (enable) 1250 if (enable)
1049 self->flags |= ix; 1251 self->flags |= ix;
1050 else 1252 else
1055 1257
1056void get_shrink (CBOR *self) 1258void get_shrink (CBOR *self)
1057 ALIAS: 1259 ALIAS:
1058 get_shrink = F_SHRINK 1260 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1261 get_allow_unknown = F_ALLOW_UNKNOWN
1262 get_allow_sharing = F_ALLOW_SHARING
1263 get_allow_cycles = F_ALLOW_CYCLES
1264 get_pack_strings = F_PACK_STRINGS
1265 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1266 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1267 XPUSHs (boolSV (self->flags & ix));
1062 1268
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1269void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1270 PPCODE:
1080 CODE: 1286 CODE:
1081 RETVAL = self->max_size; 1287 RETVAL = self->max_size;
1082 OUTPUT: 1288 OUTPUT:
1083 RETVAL 1289 RETVAL
1084 1290
1085#if 0 //TODO 1291void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1292 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1293 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1294 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1295 XPUSHs (ST (0));
1094}
1095 1296
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1297SV *get_filter (CBOR *self)
1097 PPCODE: 1298 CODE:
1098{ 1299 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1300 OUTPUT:
1100 self->cb_sk_object = newHV (); 1301 RETVAL
1101
1102 if (SvOK (cb))
1103 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1104 else
1105 {
1106 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1107
1108 if (!HvKEYS (self->cb_sk_object))
1109 {
1110 SvREFCNT_dec (self->cb_sk_object);
1111 self->cb_sk_object = 0;
1112 }
1113 }
1114
1115 XPUSHs (ST (0));
1116}
1117
1118#endif
1119 1302
1120void encode (CBOR *self, SV *scalar) 1303void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1304 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1305 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1306 XPUSHs (scalar);
1137 PUSHs (sv); 1320 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1321 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1322}
1140 1323
1141void DESTROY (CBOR *self) 1324void DESTROY (CBOR *self)
1142 CODE: 1325 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1326 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1327
1146PROTOTYPES: ENABLE 1328PROTOTYPES: ENABLE
1147 1329
1148void encode_cbor (SV *scalar) 1330void encode_cbor (SV *scalar)
1331 ALIAS:
1332 encode_cbor = 0
1333 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1334 PPCODE:
1150{ 1335{
1151 CBOR cbor; 1336 CBOR cbor;
1152 cbor_init (&cbor); 1337 cbor_init (&cbor);
1338 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1339 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1340 XPUSHs (scalar);
1155} 1341}
1156 1342
1157void decode_cbor (SV *cborstr) 1343void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines