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 1.3 by root, Sat Oct 26 11:08:34 2013 UTC vs.
Revision 1.54 by root, Sun Apr 24 20:08:41 2016 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
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};
65
66// known tags
67enum cbor_tag
68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
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
77
78 // rfc7049
79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
81 CBOR_TAG_POS_BIGNUM = 2, // byte string
82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
84 CBOR_TAG_BIGFLOAT = 5, // array
85
86 CBOR_TAG_CONV_B64U = 21, // base64url, any
87 CBOR_TAG_CONV_B64 = 22, // base64, any
88 CBOR_TAG_CONV_HEX = 23, // base16, any
89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
90
91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98};
99
14#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
16 108
17#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
18 110
19#define SB do { 111#define SB do {
20#define SE } while (0) 112#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)
32 113
33#define IN_RANGE_INC(type,val,beg,end) \ 114#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 115 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 116 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 117
42#else 123#else
43# define CBOR_SLOW 0 124# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 125# define CBOR_STASH cbor_stash
45#endif 126#endif
46 127
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 128static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 129static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 130
50typedef struct { 131typedef struct {
51 U32 flags; 132 U32 flags;
52 U32 max_depth; 133 U32 max_depth;
53 STRLEN max_size; 134 STRLEN max_size;
135 SV *filter;
54 136
55 SV *cb_object; 137 // for the incremental parser
56 HV *cb_sk_object; 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.
57} CBOR; 141} CBOR;
58 142
59INLINE void 143ecb_inline void
60cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
61{ 145{
62 Zero (cbor, 1, CBOR); 146 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 147 cbor->max_depth = 512;
64} 148}
65 149
150ecb_inline void
151cbor_free (CBOR *cbor)
152{
153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
155}
156
66///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
67// utility functions 158// utility functions
68 159
69INLINE SV * 160ecb_inline SV *
70get_bool (const char *name) 161get_bool (const char *name)
71{ 162{
72 SV *sv = get_sv (name, 1); 163 SV *sv = get_sv (name, 1);
73 164
74 SvREADONLY_on (sv); 165 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 166 SvREADONLY_on (SvRV (sv));
76 167
77 return sv; 168 return sv;
78} 169}
79 170
80INLINE void 171ecb_inline void
81shrink (SV *sv) 172shrink (SV *sv)
82{ 173{
83 sv_utf8_downgrade (sv, 1); 174 sv_utf8_downgrade (sv, 1);
84 175
85 if (SvLEN (sv) > SvCUR (sv) + 1) 176 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 181 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 182#endif
92 } 183 }
93} 184}
94 185
95///////////////////////////////////////////////////////////////////////////// 186// minimum length of a string to be registered for stringref
96// fp hell 187ecb_inline int
97 188minimum_string_length (UV idx)
98//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}
99 200
100///////////////////////////////////////////////////////////////////////////// 201/////////////////////////////////////////////////////////////////////////////
101// encoder 202// encoder
102 203
103// structure used for encoding CBOR 204// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 207 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 208 char *end; // SvEND (sv)
108 SV *sv; // result scalar 209 SV *sv; // result scalar
109 CBOR cbor; 210 CBOR cbor;
110 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;
111} enc_t; 216} enc_t;
112 217
113INLINE void 218ecb_inline void
114need (enc_t *enc, STRLEN len) 219need (enc_t *enc, STRLEN len)
115{ 220{
116 if (expect_false (enc->cur + len >= enc->end)) 221 if (ecb_expect_false (enc->cur + len >= enc->end))
117 { 222 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 223 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 224 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 225 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 226 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 227 }
123} 228}
124 229
125INLINE void 230ecb_inline void
126encode_ch (enc_t *enc, char ch) 231encode_ch (enc_t *enc, char ch)
127{ 232{
128 need (enc, 1); 233 need (enc, 1);
129 *enc->cur++ = ch; 234 *enc->cur++ = ch;
130} 235}
132static void 237static void
133encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
134{ 239{
135 need (enc, 9); 240 need (enc, 9);
136 241
137 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
139 else if (len < 0x100) 244 else if (ecb_expect_true (len <= 0xffU))
140 { 245 {
141 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 247 *enc->cur++ = len;
143 } 248 }
144 else if (len < 0x10000) 249 else if (len <= 0xffffU)
145 { 250 {
146 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 253 *enc->cur++ = len;
149 } 254 }
150 else if (len < 0x100000000) 255 else if (len <= 0xffffffffU)
151 { 256 {
152 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 261 *enc->cur++ = len;
157 } 262 }
158 else if (len) 263 else
159 { 264 {
160 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 272 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 273 *enc->cur++ = len;
169 } 274 }
170} 275}
171 276
172static 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
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 286{
175 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);
176 need (enc, len); 317 need (enc, len);
177 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
178 enc->cur += len; 319 enc->cur += len;
179} 320}
180 321
322ecb_inline void
323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
324{
325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
326 {
327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
328
329 if (SvOK (*svp))
330 {
331 // already registered, use stringref
332 encode_tag (enc, CBOR_TAG_STRINGREF);
333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
334 return;
335 }
336 else if (len >= minimum_string_length (enc->stringref_idx))
337 {
338 // register only
339 sv_setuv (*svp, enc->stringref_idx);
340 ++enc->stringref_idx;
341 }
342 }
343
344 encode_str (enc, upgrade_utf8, utf8, str, len);
345}
346
181static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
182 348
183static void 349static void
184encode_av (enc_t *enc, AV *av) 350encode_av (enc_t *enc, AV *av)
185{ 351{
188 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
190 356
191 ++enc->depth; 357 ++enc->depth;
192 358
193 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 360
361 if (SvMAGICAL (av))
195 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
196 { 363 {
197 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
199 } 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 }
200 373
201 --enc->depth; 374 --enc->depth;
202} 375}
203 376
204static void 377static void
213 386
214 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
216 389
217 if (mg) 390 if (mg)
218 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 392 else
220 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
221 394
222 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
223 { 396 {
224 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
226 else 399 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 401
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 403 }
231 404
232 if (mg) 405 if (mg)
233 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 407
235 --enc->depth; 408 --enc->depth;
236} 409}
237 410
238// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
239static void 412static void
240encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
241{ 414{
242 svtype svt;
243
244 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
416
245 svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
246 418
247 if (expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
248 { 420 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 421 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 422 ? types_boolean_stash
423 : gv_stashpv ("Types::Serialiser::Boolean", 1);
424 HV *error_stash = !CBOR_SLOW || types_error_stash
425 ? types_error_stash
426 : gv_stashpv ("Types::Serialiser::Error", 1);
427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
428 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 430
253 if (SvSTASH (sv) == stash) 431 HV *stash = SvSTASH (sv);
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 432
433 if (stash == boolean_stash)
434 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
436 return;
437 }
438 else if (stash == error_stash)
439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
443 else if (stash == tagged_stash)
444 {
445 if (svt != SVt_PVAV)
446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
447
448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 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))
464 {
465 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
466 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
467 return;
468 }
255 else 469 else
256 { 470 {
257#if 0 //TODO 471 sv_setuv (*svp, enc->shareable_idx);
258 if (enc->cbor.flags & F_CONV_BLESSED) 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
482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259 { 483 {
484 dSP;
485
486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
260 // 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
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))); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 490
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 491 PUTBACK;
492 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 493 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 494 SPAGAIN;
274 495
275 // catch this surprisingly common error 496 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 497 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))); 498 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 499
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 500 encode_sv (enc, POPs);
283 501
502 PUTBACK;
503
284 FREETMPS; LEAVE; 504 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 } 505 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
293 encode_str (enc, "null", 4, 0); 507 {
508 dSP;
509
510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
513 EXTEND (SP, 2);
514 // we re-bless the reference to get overload and other niceties right
515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
516 PUSHs (sv_cbor);
517
518 PUTBACK;
519 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
520 SPAGAIN;
521
522 // catch this surprisingly common error
523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
525
526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
527 encode_uint (enc, MAJOR_ARRAY, count + 1);
528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
529
530 while (count)
531 encode_sv (enc, SP[1 - count--]);
532
533 PUTBACK;
534
535 FREETMPS; LEAVE;
536 }
294 else 537 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 538 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 540 }
300 else if (svt == SVt_PVHV) 541 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 542 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 543 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 544 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 545 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 546 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 547 encode_tag (enc, CBOR_TAG_INDIRECTION);
548 encode_sv (enc, sv);
549 }
324} 550}
325 551
326static void 552static void
327encode_nv (enc_t *enc, SV *sv) 553encode_nv (enc_t *enc, SV *sv)
328{ 554{
329 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
330 556
331 need (enc, 9); 557 need (enc, 9);
332 558
333 if (expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 561 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
337 { 563 {
338 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
339 565
340 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
341 567
342 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
344 570
345 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
347 } 573 }
348 else 574 else
349 { 575 {
350 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
351 577
352 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
353 579
354 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
356 582
357 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
366 592
367 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
368 { 594 {
369 STRLEN len; 595 STRLEN len;
370 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
372 } 598 }
373 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 600 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
376 { 602 {
377 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
379 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
381 else 607 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
383 } 609 }
384 else if (SvROK (sv)) 610 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 616 else
391 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",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 619}
394 620
395static SV * 621static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
397{ 623{
398 enc_t enc; 624 enc_t enc = { 0 };
399 625
400 enc.cbor = *cbor; 626 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 630
406 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
407 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
408 641
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *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
411 644
425 U8 *end; // end of input string 658 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 659 const char *err; // parse error, if != 0
427 CBOR cbor; 660 CBOR cbor;
428 U32 depth; // recursion depth 661 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
663 AV *shareable;
664 AV *stringref;
665 SV *decode_tagged;
430} dec_t; 666} dec_t;
431 667
432#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
433 669
434#define WANT(len) if (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")
435 671
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 672#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
438 674
439static UV 675static UV
440decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
441{ 677{
442 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
443 { 679 ++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;
448 680
449 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
450 WANT (2); 685 WANT (1);
451 dec->cur += 2; 686 dec->cur += 1;
452 return dec->cur[-1]; 687 return dec->cur[-1];
453 688 }
454 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
455 WANT (3); 691 WANT (2);
456 dec->cur += 3; 692 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
459 695 }
460 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
461 WANT (5); 698 WANT (4);
462 dec->cur += 5; 699 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
467 704 }
468 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
469 WANT (9); 707 WANT (8);
470 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
471 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
475 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
479 723 }
480 default: 724 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 726
484fail: 727fail:
485 return 0; 728 return 0;
486} 729}
487 730
492{ 735{
493 AV *av = newAV (); 736 AV *av = newAV ();
494 737
495 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
496 739
497 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 741 {
499 ++dec->cur; 742 ++dec->cur;
500 743
501 for (;;) 744 for (;;)
502 { 745 {
503 WANT (1); 746 WANT (1);
504 747
505 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
506 { 749 {
507 ++dec->cur; 750 ++dec->cur;
508 break; 751 break;
509 } 752 }
510 753
513 } 756 }
514 else 757 else
515 { 758 {
516 int i, len = decode_uint (dec); 759 int i, len = decode_uint (dec);
517 760
761 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); 762 av_fill (av, len - 1);
519 763
520 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
522 } 766 }
528 SvREFCNT_dec (av); 772 SvREFCNT_dec (av);
529 DEC_DEC_DEPTH; 773 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 774 return &PL_sv_undef;
531} 775}
532 776
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
533static SV * 823static SV *
534decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
535{ 825{
536 HV *hv = newHV (); 826 HV *hv = newHV ();
537 827
538 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
539 829
540 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 831 {
542 ++dec->cur; 832 ++dec->cur;
543 833
544 for (;;) 834 for (;;)
545 { 835 {
546 WANT (1); 836 WANT (1);
547 837
548 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
549 { 839 {
550 ++dec->cur; 840 ++dec->cur;
551 break; 841 break;
552 } 842 }
553 843
554 SV *k = decode_sv (dec); 844 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 845 }
559 } 846 }
560 else 847 else
561 { 848 {
562 int len = decode_uint (dec); 849 int pairs = decode_uint (dec);
563 850
564 while (len--) 851 while (pairs--)
565 { 852 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 } 853 }
572 854
573 DEC_DEC_DEPTH; 855 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 856 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
730 857
731fail: 858fail:
732 SvREFCNT_dec (hv); 859 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH; 860 DEC_DEC_DEPTH;
734 return &PL_sv_undef; 861 return &PL_sv_undef;
735} 862}
736 863
737static SV * 864static SV *
738decode_str (dec_t *dec, int utf8) 865decode_str (dec_t *dec, int utf8)
739{ 866{
740 SV *sv; 867 SV *sv = 0;
741 868
742 if ((*dec->cur & 31) == 31) 869 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
743 { 870 {
871 // indefinite length strings
744 ++dec->cur; 872 ++dec->cur;
745 873
874 U8 major = *dec->cur & MAJOR_MISC;
875
746 sv = newSVpvn ("", 0); 876 sv = newSVpvn ("", 0);
747 877
748 // not very fast, and certainly not robust against illegal input
749 for (;;) 878 for (;;)
750 { 879 {
751 WANT (1); 880 WANT (1);
752 881
753 if (*dec->cur == (0xe0 | 31)) 882 if ((*dec->cur - major) > LENGTH_EXT8)
883 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
754 { 884 {
755 ++dec->cur; 885 ++dec->cur;
756 break; 886 break;
757 } 887 }
888 else
889 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
758 890
759 SV *sv2 = decode_sv (dec); 891 STRLEN len = decode_uint (dec);
760 sv_catsv (sv, sv2); 892
893 WANT (len);
894 sv_catpvn (sv, dec->cur, len);
895 dec->cur += len;
761 } 896 }
762 } 897 }
763 else 898 else
764 { 899 {
765 STRLEN len = decode_uint (dec); 900 STRLEN len = decode_uint (dec);
766 901
767 WANT (len); 902 WANT (len);
768 sv = newSVpvn (dec->cur, len); 903 sv = newSVpvn (dec->cur, len);
769 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));
770 } 909 }
771 910
772 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
773 SvUTF8_on (sv); 917 SvUTF8_on (sv);
918 }
774 919
775 return sv; 920 return sv;
776 921
777fail: 922fail:
923 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 924 return &PL_sv_undef;
779} 925}
780 926
781static SV * 927static SV *
782decode_tagged (dec_t *dec) 928decode_tagged (dec_t *dec)
783{ 929{
930 SV *sv = 0;
784 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:
785 SV *sv = decode_sv (dec); 938 sv = decode_sv (dec);
939 break;
786 940
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 941 case CBOR_TAG_INDIRECTION:
942 sv = newRV_noinc (decode_sv (dec));
943 break;
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
1017 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1018 ERR ("corrupted CBOR data (non-array perl object)");
1019
1020 AV *av = (AV *)SvRV (sv);
1021 int len = av_len (av) + 1;
1022 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1023
1024 if (!stash)
1025 ERR ("cannot decode perl-object (package does not exist)");
1026
1027 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1028
1029 if (!method)
1030 ERR ("cannot decode perl-object (package does not have a THAW method)");
1031
1032 dSP;
1033
1034 ENTER; SAVETMPS;
1035 PUSHMARK (SP);
1036 EXTEND (SP, len + 1);
1037 // we re-bless the reference to get overload and other niceties right
1038 PUSHs (*av_fetch (av, 0, 1));
1039 PUSHs (sv_cbor);
1040
1041 int i;
1042
1043 for (i = 1; i < len; ++i)
1044 PUSHs (*av_fetch (av, i, 1));
1045
1046 PUTBACK;
1047 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1048 SPAGAIN;
1049
1050 if (SvTRUE (ERRSV))
1051 {
1052 FREETMPS; LEAVE;
1053 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1054 }
1055
1056 SvREFCNT_dec (sv);
1057 sv = SvREFCNT_inc (POPs);
1058
1059 PUTBACK;
1060
1061 FREETMPS; LEAVE;
1062 }
1063 break;
1064
1065 default:
1066 {
1067 sv = decode_sv (dec);
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 {
1094 AV *av = newAV ();
1095 av_push (av, newSVuv (tag));
1096 av_push (av, sv);
1097
1098 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1099 ? cbor_tagged_stash
1100 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1101 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1102 }
1103
1104 PUTBACK;
1105
1106 FREETMPS; LEAVE;
1107 }
1108 break;
1109 }
1110
788 return sv; 1111 return sv;
789 1112
790 AV *av = newAV (); 1113fail:
791 av_push (av, newSVuv (tag)); 1114 SvREFCNT_dec (sv);
792 av_push (av, sv); 1115 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1116}
795 1117
796static SV * 1118static SV *
797decode_sv (dec_t *dec) 1119decode_sv (dec_t *dec)
798{ 1120{
799 WANT (1); 1121 WANT (1);
800 1122
801 switch (*dec->cur >> 5) 1123 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1124 {
803 case 0: // unsigned int 1125 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1126 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1127 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1128 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1129 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1130 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1131 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1132
811 return decode_str (dec, 1); 1133 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) 1134 switch (*dec->cur++ & MINOR_MASK)
820 { 1135 {
821 case 20: 1136 case SIMPLE_FALSE:
822#if CBOR_SLOW 1137#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1138 types_false = get_bool ("Types::Serialiser::false");
824#endif 1139#endif
825 return newSVsv (cbor_false); 1140 return newSVsv (types_false);
826 case 21: 1141 case SIMPLE_TRUE:
827#if CBOR_SLOW 1142#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1143 types_true = get_bool ("Types::Serialiser::true");
829#endif 1144#endif
830 return newSVsv (cbor_true); 1145 return newSVsv (types_true);
831 case 22: 1146 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1147 return newSVsv (&PL_sv_undef);
1148 case SIMPLE_UNDEF:
1149#if CBOR_SLOW
1150 types_error = get_bool ("Types::Serialiser::error");
1151#endif
1152 return newSVsv (types_error);
833 1153
834 case 25: 1154 case MISC_FLOAT16:
835 { 1155 {
836 WANT (2); 1156 WANT (2);
837 1157
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1158 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1159 dec->cur += 2;
840 1160
841 return newSVnv (ecb_binary16_to_float (fp)); 1161 return newSVnv (ecb_binary16_to_float (fp));
842 } 1162 }
843 1163
844 case 26: 1164 case MISC_FLOAT32:
845 { 1165 {
846 uint32_t fp; 1166 uint32_t fp;
847 WANT (4); 1167 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1168 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1169 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1172 fp = ecb_bswap32 (fp);
853 1173
854 return newSVnv (ecb_binary32_to_float (fp)); 1174 return newSVnv (ecb_binary32_to_float (fp));
855 } 1175 }
856 1176
857 case 27: 1177 case MISC_FLOAT64:
858 { 1178 {
859 uint64_t fp; 1179 uint64_t fp;
860 WANT (8); 1180 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1181 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1182 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1185 fp = ecb_bswap64 (fp);
866 1186
867 return newSVnv (ecb_binary64_to_double (fp)); 1187 return newSVnv (ecb_binary64_to_double (fp));
868 } 1188 }
869 1189
870 // 0..19 unassigned 1190 // 0..19 unassigned simple
871 // 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
872 default: 1194 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1195 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874 } 1196 }
875 1197
876 break; 1198 break;
877 } 1199 }
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 1200
935fail: 1201fail:
936 return &PL_sv_undef; 1202 return &PL_sv_undef;
937} 1203}
938 1204
939static SV * 1205static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1206decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1207{
942 dec_t dec; 1208 dec_t dec = { 0 };
943 SV *sv; 1209 SV *sv;
1210 STRLEN len;
1211 char *data = SvPVbyte (string, len);
944 1212
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) 1213 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", 1214 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); 1215 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1216
980 dec.cbor = *cbor; 1217 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1218 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1219 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 1220
989 sv = decode_sv (&dec); 1221 sv = decode_sv (&dec);
990 1222
991 if (offset_return) 1223 if (offset_return)
992 *offset_return = dec.cur; 1224 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1227 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1228 dec.err = "garbage after CBOR object";
997 1229
998 if (dec.err) 1230 if (dec.err)
999 { 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
1000 SvREFCNT_dec (sv); 1243 SvREFCNT_dec (sv);
1001 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);
1002 } 1245 }
1003 1246
1004 sv = sv_2mortal (sv); 1247 sv = sv_2mortal (sv);
1005 1248
1006 return sv; 1249 return sv;
1007} 1250}
1008 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
1009///////////////////////////////////////////////////////////////////////////// 1376/////////////////////////////////////////////////////////////////////////////
1010// XS interface functions 1377// XS interface functions
1011 1378
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1379MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1380
1014BOOT: 1381BOOT:
1015{ 1382{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1383 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1384 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1385
1019 cbor_true = get_bool ("CBOR::XS::true"); 1386 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1387 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1388
1389 types_true = get_bool ("Types::Serialiser::true" );
1390 types_false = get_bool ("Types::Serialiser::false");
1391 types_error = get_bool ("Types::Serialiser::error");
1392
1393 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1394
1395 sv_cbor = newSVpv ("CBOR", 0);
1396 SvREADONLY_on (sv_cbor);
1021} 1397}
1022 1398
1023PROTOTYPES: DISABLE 1399PROTOTYPES: DISABLE
1024 1400
1025void CLONE (...) 1401void CLONE (...)
1026 CODE: 1402 CODE:
1027 cbor_stash = 0; 1403 cbor_stash = 0;
1404 cbor_tagged_stash = 0;
1405 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1406 types_boolean_stash = 0;
1029 1407
1030void new (char *klass) 1408void new (char *klass)
1031 PPCODE: 1409 PPCODE:
1032{ 1410{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1411 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1419
1042void shrink (CBOR *self, int enable = 1) 1420void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1421 ALIAS:
1044 shrink = F_SHRINK 1422 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1423 allow_unknown = F_ALLOW_UNKNOWN
1424 allow_sharing = F_ALLOW_SHARING
1425 allow_cycles = F_ALLOW_CYCLES
1426 pack_strings = F_PACK_STRINGS
1427 text_keys = F_TEXT_KEYS
1428 text_strings = F_TEXT_STRINGS
1429 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1430 PPCODE:
1047{ 1431{
1048 if (enable) 1432 if (enable)
1049 self->flags |= ix; 1433 self->flags |= ix;
1050 else 1434 else
1055 1439
1056void get_shrink (CBOR *self) 1440void get_shrink (CBOR *self)
1057 ALIAS: 1441 ALIAS:
1058 get_shrink = F_SHRINK 1442 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1443 get_allow_unknown = F_ALLOW_UNKNOWN
1444 get_allow_sharing = F_ALLOW_SHARING
1445 get_allow_cycles = F_ALLOW_CYCLES
1446 get_pack_strings = F_PACK_STRINGS
1447 get_text_keys = F_TEXT_KEYS
1448 get_text_strings = F_TEXT_STRINGS
1449 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1450 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1451 XPUSHs (boolSV (self->flags & ix));
1062 1452
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1453void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1454 PPCODE:
1080 CODE: 1470 CODE:
1081 RETVAL = self->max_size; 1471 RETVAL = self->max_size;
1082 OUTPUT: 1472 OUTPUT:
1083 RETVAL 1473 RETVAL
1084 1474
1085#if 0 //TODO 1475void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1476 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1477 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1478 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1479 XPUSHs (ST (0));
1094}
1095 1480
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1481SV *get_filter (CBOR *self)
1097 PPCODE: 1482 CODE:
1098{ 1483 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1484 OUTPUT:
1100 self->cb_sk_object = newHV (); 1485 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 1486
1120void encode (CBOR *self, SV *scalar) 1487void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1488 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1489 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1490 XPUSHs (scalar);
1136 EXTEND (SP, 2); 1503 EXTEND (SP, 2);
1137 PUSHs (sv); 1504 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1505 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1506}
1140 1507
1508void incr_parse (CBOR *self, SV *cborstr)
1509 ALIAS:
1510 incr_parse_multiple = 1
1511 PPCODE:
1512{
1513 if (SvUTF8 (cborstr))
1514 sv_utf8_downgrade (cborstr, 0);
1515
1516 if (!self->incr_count)
1517 {
1518 self->incr_count = newAV ();
1519 self->incr_pos = 0;
1520 self->incr_need = 1;
1521
1522 av_push (self->incr_count, newSViv (1));
1523 }
1524
1525 do
1526 {
1527 if (!incr_parse (self, cborstr))
1528 {
1529 if (self->incr_need > self->max_size && self->max_size)
1530 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1531 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1532
1533 break;
1534 }
1535
1536 SV *sv;
1537 char *offset;
1538
1539 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1540 XPUSHs (sv);
1541
1542 sv_chop (cborstr, offset);
1543
1544 av_clear (self->incr_count);
1545 av_push (self->incr_count, newSViv (1));
1546
1547 self->incr_pos = 0;
1548 self->incr_need = self->incr_pos + 1;
1549 }
1550 while (ix);
1551}
1552
1553void incr_reset (CBOR *self)
1554 CODE:
1555{
1556 SvREFCNT_dec (self->incr_count);
1557 self->incr_count = 0;
1558}
1559
1141void DESTROY (CBOR *self) 1560void DESTROY (CBOR *self)
1142 CODE: 1561 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1562 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1563
1146PROTOTYPES: ENABLE 1564PROTOTYPES: ENABLE
1147 1565
1148void encode_cbor (SV *scalar) 1566void encode_cbor (SV *scalar)
1567 ALIAS:
1568 encode_cbor = 0
1569 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1570 PPCODE:
1150{ 1571{
1151 CBOR cbor; 1572 CBOR cbor;
1152 cbor_init (&cbor); 1573 cbor_init (&cbor);
1574 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1575 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1576 XPUSHs (scalar);
1155} 1577}
1156 1578
1157void decode_cbor (SV *cborstr) 1579void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines