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 1.64 by root, Tue Jun 27 02:03:24 2017 UTC

6#include <string.h> 6#include <string.h>
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#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
69
70// known tags
71enum cbor_tag
72{
73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
81
82 // rfc7049
83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
85 CBOR_TAG_POS_BIGNUM = 2, // byte string
86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
88 CBOR_TAG_BIGFLOAT = 5, // array
89
90 CBOR_TAG_CONV_B64U = 21, // base64url, any
91 CBOR_TAG_CONV_B64 = 22, // base64, any
92 CBOR_TAG_CONV_HEX = 23, // base16, any
93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
94
95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
100
101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
102};
103
14#define F_SHRINK 0x00000200UL 104#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 105#define F_ALLOW_UNKNOWN 0x00000002UL
106#define F_ALLOW_SHARING 0x00000004UL
107#define F_ALLOW_CYCLES 0x00000008UL
108#define F_FORBID_OBJECTS 0x00000010UL
109#define F_PACK_STRINGS 0x00000020UL
110#define F_TEXT_KEYS 0x00000040UL
111#define F_TEXT_STRINGS 0x00000080UL
112#define F_VALIDATE_UTF8 0x00000100UL
16 113
17#define INIT_SIZE 32 // initial scalar size to be allocated 114#define INIT_SIZE 32 // initial scalar size to be allocated
18 115
19#define SB do { 116#define SB do {
20#define SE } while (0) 117#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)
32 118
33#define IN_RANGE_INC(type,val,beg,end) \ 119#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 120 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 121 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 122
42#else 128#else
43# define CBOR_SLOW 0 129# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 130# define CBOR_STASH cbor_stash
45#endif 131#endif
46 132
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 133static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 134static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 135
50typedef struct { 136typedef struct {
51 U32 flags; 137 U32 flags;
52 U32 max_depth; 138 U32 max_depth;
53 STRLEN max_size; 139 STRLEN max_size;
140 SV *filter;
54 141
55 SV *cb_object; 142 // for the incremental parser
56 HV *cb_sk_object; 143 STRLEN incr_pos; // the current offset into the text
144 STRLEN incr_need; // minimum bytes needed to decode
145 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
57} CBOR; 146} CBOR;
58 147
59INLINE void 148ecb_inline void
60cbor_init (CBOR *cbor) 149cbor_init (CBOR *cbor)
61{ 150{
62 Zero (cbor, 1, CBOR); 151 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 152 cbor->max_depth = 512;
64} 153}
65 154
155ecb_inline void
156cbor_free (CBOR *cbor)
157{
158 SvREFCNT_dec (cbor->filter);
159 SvREFCNT_dec (cbor->incr_count);
160}
161
66///////////////////////////////////////////////////////////////////////////// 162/////////////////////////////////////////////////////////////////////////////
67// utility functions 163// utility functions
68 164
69INLINE SV * 165ecb_inline SV *
70get_bool (const char *name) 166get_bool (const char *name)
71{ 167{
72 SV *sv = get_sv (name, 1); 168 SV *sv = get_sv (name, 1);
73 169
74 SvREADONLY_on (sv); 170 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 171 SvREADONLY_on (SvRV (sv));
76 172
77 return sv; 173 return sv;
78} 174}
79 175
80INLINE void 176ecb_inline void
81shrink (SV *sv) 177shrink (SV *sv)
82{ 178{
83 sv_utf8_downgrade (sv, 1); 179 sv_utf8_downgrade (sv, 1);
84 180
85 if (SvLEN (sv) > SvCUR (sv) + 1) 181 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 186 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 187#endif
92 } 188 }
93} 189}
94 190
95///////////////////////////////////////////////////////////////////////////// 191// minimum length of a string to be registered for stringref
96// fp hell 192ecb_inline int
97 193minimum_string_length (UV idx)
98//TODO 194{
195 return idx <= 23 ? 3
196 : idx <= 0xffU ? 4
197 : idx <= 0xffffU ? 5
198 : idx <= 0xffffffffU ? 7
199 : 11;
200}
99 201
100///////////////////////////////////////////////////////////////////////////// 202/////////////////////////////////////////////////////////////////////////////
101// encoder 203// encoder
102 204
103// structure used for encoding CBOR 205// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 208 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 209 char *end; // SvEND (sv)
108 SV *sv; // result scalar 210 SV *sv; // result scalar
109 CBOR cbor; 211 CBOR cbor;
110 U32 depth; // recursion level 212 U32 depth; // recursion level
213 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
214 UV stringref_idx;
215 HV *shareable; // ptr => index, or 0
216 UV shareable_idx;
111} enc_t; 217} enc_t;
112 218
113INLINE void 219ecb_inline void
114need (enc_t *enc, STRLEN len) 220need (enc_t *enc, STRLEN len)
115{ 221{
116 if (expect_false (enc->cur + len >= enc->end)) 222 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
117 { 223 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 224 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 225 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 226 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 227 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 228 }
123} 229}
124 230
125INLINE void 231ecb_inline void
126encode_ch (enc_t *enc, char ch) 232encode_ch (enc_t *enc, char ch)
127{ 233{
128 need (enc, 1); 234 need (enc, 1);
129 *enc->cur++ = ch; 235 *enc->cur++ = ch;
130} 236}
132static void 238static void
133encode_uint (enc_t *enc, int major, UV len) 239encode_uint (enc_t *enc, int major, UV len)
134{ 240{
135 need (enc, 9); 241 need (enc, 9);
136 242
137 if (len < 24) 243 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 244 *enc->cur++ = major | len;
139 else if (len < 0x100) 245 else if (ecb_expect_true (len <= 0xffU))
140 { 246 {
141 *enc->cur++ = major | 24; 247 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 248 *enc->cur++ = len;
143 } 249 }
144 else if (len < 0x10000) 250 else if (len <= 0xffffU)
145 { 251 {
146 *enc->cur++ = major | 25; 252 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 253 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 254 *enc->cur++ = len;
149 } 255 }
150 else if (len < 0x100000000) 256 else if (len <= 0xffffffffU)
151 { 257 {
152 *enc->cur++ = major | 26; 258 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 260 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 262 *enc->cur++ = len;
157 } 263 }
158 else if (len) 264 else
159 { 265 {
160 *enc->cur++ = major | 27; 266 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 267 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 268 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 269 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 270 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 271 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 274 *enc->cur++ = len;
169 } 275 }
170} 276}
171 277
172static void 278ecb_inline void
279encode_tag (enc_t *enc, UV tag)
280{
281 encode_uint (enc, MAJOR_TAG, tag);
282}
283
284// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
285ecb_noinline static void
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 286encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 287{
175 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 288 STRLEN ulen = len;
289 U8 *p, *pend = (U8 *)str + len;
290
291 for (p = (U8 *)str; p < pend; ++p)
292 ulen += *p >> 7; // count set high bits
293
294 encode_uint (enc, MAJOR_TEXT, ulen);
295
296 need (enc, ulen);
297 for (p = (U8 *)str; p < pend; ++p)
298 if (*p < 0x80)
299 *enc->cur++ = *p;
300 else
301 {
302 *enc->cur++ = 0xc0 + (*p >> 6);
303 *enc->cur++ = 0x80 + (*p & 63);
304 }
305}
306
307ecb_inline void
308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
176 need (enc, len); 318 need (enc, len);
177 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
178 enc->cur += len; 320 enc->cur += len;
179} 321}
180 322
323ecb_inline void
324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
325{
326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
327 {
328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
329
330 if (SvOK (*svp))
331 {
332 // already registered, use stringref
333 encode_tag (enc, CBOR_TAG_STRINGREF);
334 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
335 return;
336 }
337 else if (len >= minimum_string_length (enc->stringref_idx))
338 {
339 // register only
340 sv_setuv (*svp, enc->stringref_idx);
341 ++enc->stringref_idx;
342 }
343 }
344
345 encode_str (enc, upgrade_utf8, utf8, str, len);
346}
347
181static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
182 349
183static void 350static void
184encode_av (enc_t *enc, AV *av) 351encode_av (enc_t *enc, AV *av)
185{ 352{
188 if (enc->depth >= enc->cbor.max_depth) 355 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 356 croak (ERR_NESTING_EXCEEDED);
190 357
191 ++enc->depth; 358 ++enc->depth;
192 359
193 encode_uint (enc, 0x80, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 361
362 if (ecb_expect_false (SvMAGICAL (av)))
195 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
196 { 364 {
197 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
199 } 367 }
368 else
369 for (i = 0; i <= len; ++i)
370 {
371 SV *sv = AvARRAY (av)[i];
372 encode_sv (enc, sv ? sv : &PL_sv_undef);
373 }
200 374
201 --enc->depth; 375 --enc->depth;
202} 376}
203 377
204static void 378static void
212 ++enc->depth; 386 ++enc->depth;
213 387
214 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
216 390
217 if (mg) 391 if (ecb_expect_false (mg))
218 encode_ch (enc, 0xa0 | 31); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 393 else
220 encode_uint (enc, 0xa0, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
221 395
222 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
223 { 397 {
224 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
226 else 400 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 401 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 402
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 403 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 404 }
231 405
232 if (mg) 406 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xe0 | 31); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 408
235 --enc->depth; 409 --enc->depth;
236} 410}
237 411
238// encode objects, arrays and special \0=false and \1=true values. 412// encode objects, arrays and special \0=false and \1=true values.
239static void 413static void
240encode_rv (enc_t *enc, SV *sv) 414encode_rv (enc_t *enc, SV *sv)
241{ 415{
242 svtype svt;
243
244 SvGETMAGIC (sv); 416 SvGETMAGIC (sv);
417
245 svt = SvTYPE (sv); 418 svtype svt = SvTYPE (sv);
246 419
247 if (expect_false (SvOBJECT (sv))) 420 if (ecb_expect_false (SvOBJECT (sv)))
248 { 421 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 422 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 423 ? types_boolean_stash
424 : gv_stashpv ("Types::Serialiser::Boolean", 1);
425 HV *error_stash = !CBOR_SLOW || types_error_stash
426 ? types_error_stash
427 : gv_stashpv ("Types::Serialiser::Error", 1);
428 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
429 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 430 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 431
253 if (SvSTASH (sv) == stash) 432 HV *stash = SvSTASH (sv);
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 433
434 if (stash == boolean_stash)
435 {
436 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
437 return;
438 }
439 else if (stash == error_stash)
440 {
441 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
442 return;
443 }
444 else if (stash == tagged_stash)
445 {
446 if (svt != SVt_PVAV)
447 croak ("encountered CBOR::XS::Tagged object that isn't an array");
448
449 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
450 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
451
452 return;
453 }
454 }
455
456 if (ecb_expect_false (SvREFCNT (sv) > 1)
457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
458 {
459 if (ecb_expect_false (!enc->shareable))
460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
461
462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
463
464 if (SvOK (*svp))
465 {
466 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
467 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
468 return;
469 }
255 else 470 else
256 { 471 {
257#if 0 //TODO 472 sv_setuv (*svp, enc->shareable_idx);
258 if (enc->cbor.flags & F_CONV_BLESSED) 473 ++enc->shareable_idx;
474 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
475 }
476 }
477
478 if (ecb_expect_false (SvOBJECT (sv)))
479 {
480 HV *stash = SvSTASH (sv);
481 GV *method;
482
483 if (enc->cbor.flags & F_FORBID_OBJECTS)
484 croak ("encountered object '%s', but forbid_objects is enabled",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259 { 487 {
488 dSP;
489
490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
260 // we re-bless the reference to get overload and other niceties right 492 // 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))); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 494
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 495 PUTBACK;
496 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 497 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 498 SPAGAIN;
274 499
275 // catch this surprisingly common error 500 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 501 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))); 502 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 503
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 504 encode_sv (enc, POPs);
283 505
506 PUTBACK;
507
284 FREETMPS; LEAVE; 508 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 } 509 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
293 encode_str (enc, "null", 4, 0); 511 {
512 dSP;
513
514 ENTER; SAVETMPS;
515 SAVESTACK_POS ();
516 PUSHMARK (SP);
517 EXTEND (SP, 2);
518 // we re-bless the reference to get overload and other niceties right
519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
520 PUSHs (sv_cbor);
521
522 PUTBACK;
523 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
524 SPAGAIN;
525
526 // catch this surprisingly common error
527 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
528 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
529
530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
531 encode_uint (enc, MAJOR_ARRAY, count + 1);
532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
533
534 while (count)
535 encode_sv (enc, SP[1 - count--]);
536
537 PUTBACK;
538
539 FREETMPS; LEAVE;
540 }
294 else 541 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 542 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 543 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 544 }
300 else if (svt == SVt_PVHV) 545 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 546 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 547 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 548 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 549 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 550 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 551 encode_tag (enc, CBOR_TAG_INDIRECTION);
552 encode_sv (enc, sv);
553 }
324} 554}
325 555
326static void 556static void
327encode_nv (enc_t *enc, SV *sv) 557encode_nv (enc_t *enc, SV *sv)
328{ 558{
329 double nv = SvNVX (sv); 559 double nv = SvNVX (sv);
330 560
331 need (enc, 9); 561 need (enc, 9);
332 562
333 if (expect_false (nv == (U32)nv)) 563 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 564 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 565 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 566 else if (ecb_expect_false (nv == (float)nv))
337 { 567 {
568 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
569
338 uint32_t fp = ecb_float_to_binary32 (nv); 570 uint32_t fp = ecb_float_to_binary32 (nv);
339
340 *enc->cur++ = 0xe0 | 26;
341 571
342 if (!ecb_big_endian ()) 572 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp); 573 fp = ecb_bswap32 (fp);
344 574
345 memcpy (enc->cur, &fp, 4); 575 memcpy (enc->cur, &fp, 4);
346 enc->cur += 4; 576 enc->cur += 4;
347 } 577 }
348 else 578 else
349 { 579 {
580 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
581
350 uint64_t fp = ecb_double_to_binary64 (nv); 582 uint64_t fp = ecb_double_to_binary64 (nv);
351
352 *enc->cur++ = 0xe0 | 27;
353 583
354 if (!ecb_big_endian ()) 584 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp); 585 fp = ecb_bswap64 (fp);
356 586
357 memcpy (enc->cur, &fp, 8); 587 memcpy (enc->cur, &fp, 8);
366 596
367 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
368 { 598 {
369 STRLEN len; 599 STRLEN len;
370 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
372 } 602 }
373 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 604 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
376 { 606 {
377 if (SvIsUV (sv)) 607 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv)); 608 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
379 else if (SvIVX (sv) >= 0) 609 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv)); 610 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
381 else 611 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 612 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
383 } 613 }
384 else if (SvROK (sv)) 614 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 615 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 616 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 617 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 618 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 619 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 620 else
391 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 621 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 622 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 623}
394 624
395static SV * 625static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
397{ 627{
398 enc_t enc; 628 enc_t enc = { 0 };
399 629
400 enc.cbor = *cbor; 630 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 634
406 SvPOK_only (enc.sv); 635 SvPOK_only (enc.sv);
636
637 if (cbor->flags & F_PACK_STRINGS)
638 {
639 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
640 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
641 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
642 }
643
407 encode_sv (&enc, scalar); 644 encode_sv (&enc, scalar);
408 645
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 646 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 647 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411 648
425 U8 *end; // end of input string 662 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 663 const char *err; // parse error, if != 0
427 CBOR cbor; 664 CBOR cbor;
428 U32 depth; // recursion depth 665 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
667 AV *shareable;
668 AV *stringref;
669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
430} dec_t; 671} dec_t;
431 672
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 673// set dec->err to ERRSV
674ecb_cold static void
675err_errsv (dec_t *dec)
676{
677 if (!dec->err)
678 {
679 dec->err_sv = newSVsv (ERRSV);
433 680
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 681 // chop off the trailing \n
682 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
683 *SvEND (dec->err_sv) = 0;
435 684
685 dec->err = SvPVutf8_nolen (dec->err_sv);
686 }
687}
688
689// the following functions are used to reduce code size and help the compiler to optimise
690ecb_cold static void
691err_set (dec_t *dec, const char *reason)
692{
693 if (!dec->err)
694 dec->err = reason;
695}
696
697ecb_cold static void
698err_unexpected_end (dec_t *dec)
699{
700 err_set (dec, "unexpected end of CBOR data");
701}
702
703#define ERR_DO(do) SB do; goto fail; SE
704#define ERR(reason) ERR_DO (err_set (dec, reason))
705#define ERR_ERRSV ERR_DO (err_errsv (dec))
706
707#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
708
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 709#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 710#define DEC_DEC_DEPTH --dec->depth
438 711
439static UV 712static UV
440decode_uint (dec_t *dec) 713decode_uint (dec_t *dec)
441{ 714{
442 switch (*dec->cur & 31) 715 U8 m = *dec->cur & MINOR_MASK;
443 { 716 ++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;
448 717
449 case 24: 718 if (ecb_expect_true (m < LENGTH_EXT1))
719 return m;
720 else if (ecb_expect_true (m == LENGTH_EXT1))
721 {
450 WANT (2); 722 WANT (1);
451 dec->cur += 2; 723 dec->cur += 1;
452 return dec->cur[-1]; 724 return dec->cur[-1];
453 725 }
454 case 25: 726 else if (ecb_expect_true (m == LENGTH_EXT2))
727 {
455 WANT (3); 728 WANT (2);
456 dec->cur += 3; 729 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 730 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 731 | ((UV)dec->cur[-1]);
459 732 }
460 case 26: 733 else if (ecb_expect_true (m == LENGTH_EXT4))
734 {
461 WANT (5); 735 WANT (4);
462 dec->cur += 5; 736 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 737 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 738 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 739 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 740 | ((UV)dec->cur[-1]);
467 741 }
468 case 27: 742 else if (ecb_expect_true (m == LENGTH_EXT8))
743 {
469 WANT (9); 744 WANT (8);
470 dec->cur += 9; 745 dec->cur += 8;
746
747 return
748#if UVSIZE < 8
749 0
750#else
471 return (((UV)dec->cur[-8]) << 56) 751 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 752 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 753 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 754 | (((UV)dec->cur[-5]) << 32)
755#endif
475 | (((UV)dec->cur[-4]) << 24) 756 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 757 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 758 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 759 | ((UV)dec->cur[-1]);
479 760 }
480 default: 761 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 762 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 763
484fail: 764fail:
485 return 0; 765 return 0;
486} 766}
487 767
492{ 772{
493 AV *av = newAV (); 773 AV *av = newAV ();
494 774
495 DEC_INC_DEPTH; 775 DEC_INC_DEPTH;
496 776
497 if ((*dec->cur & 31) == 31) 777 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 778 {
499 ++dec->cur; 779 ++dec->cur;
500 780
501 for (;;) 781 for (;;)
502 { 782 {
503 WANT (1); 783 WANT (1);
504 784
505 if (*dec->cur == (0xe0 | 31)) 785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
506 { 786 {
507 ++dec->cur; 787 ++dec->cur;
508 break; 788 break;
509 } 789 }
510 790
511 av_push (av, decode_sv (dec)); 791 av_push (av, decode_sv (dec));
512 } 792 }
513 } 793 }
514 else 794 else
515 { 795 {
516 int i, len = decode_uint (dec); 796 UV i, len = decode_uint (dec);
517 797
798 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); 799 av_fill (av, len - 1);
519 800
520 for (i = 0; i < len; ++i) 801 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 802 AvARRAY (av)[i] = decode_sv (dec);
522 } 803 }
523 804
524 DEC_DEC_DEPTH; 805 DEC_DEC_DEPTH;
525 return newRV_noinc ((SV *)av); 806 return newRV_noinc ((SV *)av);
526 807
527fail: 808fail:
528 SvREFCNT_dec (av); 809 SvREFCNT_dec_NN (av);
529 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 811 return &PL_sv_undef;
531} 812}
532 813
814static void
815decode_he (dec_t *dec, HV *hv)
816{
817 // for speed reasons, we specialcase single-string
818 // byte or utf-8 strings as keys, but only when !stringref
819
820 if (ecb_expect_true (!dec->stringref))
821 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
822 {
823 STRLEN len = decode_uint (dec);
824 char *key = (char *)dec->cur;
825
826 WANT (len);
827 dec->cur += len;
828
829 hv_store (hv, key, len, decode_sv (dec), 0);
830
831 return;
832 }
833 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
834 {
835 STRLEN len = decode_uint (dec);
836 char *key = (char *)dec->cur;
837
838 WANT (len);
839 dec->cur += len;
840
841 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
842 if (!is_utf8_string (key, len))
843 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
844
845 hv_store (hv, key, -len, decode_sv (dec), 0);
846
847 return;
848 }
849
850 SV *k = decode_sv (dec);
851 SV *v = decode_sv (dec);
852
853 // we leak memory if uncaught exceptions are thrown by random magical
854 // methods, and this is hopefully the only place where it can happen,
855 // so if there is a chance of an exception, take the very slow path.
856 // since catching exceptions is "undocumented/internal/forbidden" by
857 // the new p5p powers, we need to call out to a perl function :/
858 if (ecb_expect_false (SvAMAGIC (k)))
859 {
860 dSP;
861
862 ENTER; SAVETMPS;
863 PUSHMARK (SP);
864 EXTEND (SP, 3);
865 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
866 PUSHs (sv_2mortal (k));
867 PUSHs (sv_2mortal (v));
868
869 PUTBACK;
870 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
871 SPAGAIN;
872
873 FREETMPS; LEAVE;
874
875 if (SvTRUE (ERRSV))
876 ERR_ERRSV;
877
878 return;
879 }
880
881 hv_store_ent (hv, k, v, 0);
882 SvREFCNT_dec_NN (k);
883
884fail:
885 ;
886}
887
533static SV * 888static SV *
534decode_hv (dec_t *dec) 889decode_hv (dec_t *dec)
535{ 890{
536 HV *hv = newHV (); 891 HV *hv = newHV ();
537 892
538 DEC_INC_DEPTH; 893 DEC_INC_DEPTH;
539 894
540 if ((*dec->cur & 31) == 31) 895 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 896 {
542 ++dec->cur; 897 ++dec->cur;
543 898
544 for (;;) 899 for (;;)
545 { 900 {
546 WANT (1); 901 WANT (1);
547 902
548 if (*dec->cur == (0xe0 | 31)) 903 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
549 { 904 {
550 ++dec->cur; 905 ++dec->cur;
551 break; 906 break;
552 } 907 }
553 908
554 SV *k = decode_sv (dec); 909 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 910 }
559 } 911 }
560 else 912 else
561 { 913 {
562 int len = decode_uint (dec); 914 UV pairs = decode_uint (dec);
563 915
916 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
917
564 while (len--) 918 while (pairs--)
565 { 919 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 } 920 }
572 921
573 DEC_DEC_DEPTH; 922 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 923 return newRV_noinc ((SV *)hv);
575 924
576#if 0 925fail:
577 SV *sv; 926 SvREFCNT_dec_NN (hv);
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH; 927 DEC_DEC_DEPTH;
581 decode_ws (dec); 928 return &PL_sv_undef;
929}
582 930
583 for (;;) 931static SV *
584 { 932decode_str (dec_t *dec, int utf8)
585 // heuristic: assume that 933{
586 // a) decode_str + hv_store_ent are abysmally slow. 934 SV *sv = 0;
587 // b) most hash keys are short, simple ascii text. 935
588 // => try to "fast-match" such strings to avoid 936 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
589 // the overhead of decode_str + hv_store_ent.
590 { 937 {
591 SV *value; 938 // indefinite length strings
592 char *p = dec->cur; 939 ++dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594 940
941 U8 major = *dec->cur & MAJOR_MISC;
942
943 sv = newSVpvn ("", 0);
944
595 for (;;) 945 for (;;)
596 { 946 {
597 // the >= 0x80 is false on most architectures 947 WANT (1);
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 948
949 if ((*dec->cur - major) > LENGTH_EXT8)
950 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 951 {
600 // slow path, back up and use decode_str 952 ++dec->cur;
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; 953 break;
619 } 954 }
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 { 955 else
685 dSP; 956 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
686 int count;
687 957
688 ENTER; SAVETMPS; PUSHMARK (SP); 958 STRLEN len = decode_uint (dec);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691 959
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
731fail:
732 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH;
734 return &PL_sv_undef;
735}
736
737static SV *
738decode_str (dec_t *dec, int utf8)
739{
740 SV *sv;
741
742 if ((*dec->cur & 31) == 31)
743 {
744 ++dec->cur;
745
746 sv = newSVpvn ("", 0);
747
748 // not very fast, and certainly not robust against illegal input
749 for (;;)
750 {
751 WANT (1); 960 WANT (len);
752 961 sv_catpvn (sv, dec->cur, len);
753 if (*dec->cur == (0xe0 | 31)) 962 dec->cur += len;
754 {
755 ++dec->cur;
756 break;
757 }
758
759 SV *sv2 = decode_sv (dec);
760 sv_catsv (sv, sv2);
761 } 963 }
762 } 964 }
763 else 965 else
764 { 966 {
765 STRLEN len = decode_uint (dec); 967 STRLEN len = decode_uint (dec);
766 968
767 WANT (len); 969 WANT (len);
768 sv = newSVpvn (dec->cur, len); 970 sv = newSVpvn (dec->cur, len);
769 dec->cur += len; 971 dec->cur += len;
972
973 if (ecb_expect_false (dec->stringref)
974 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
975 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 976 }
771 977
772 if (utf8) 978 if (utf8)
979 {
980 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
981 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
982 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
983
773 SvUTF8_on (sv); 984 SvUTF8_on (sv);
985 }
774 986
775 return sv; 987 return sv;
776 988
777fail: 989fail:
990 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 991 return &PL_sv_undef;
779} 992}
780 993
781static SV * 994static SV *
782decode_tagged (dec_t *dec) 995decode_tagged (dec_t *dec)
783{ 996{
997 SV *sv = 0;
784 UV tag = decode_uint (dec); 998 UV tag = decode_uint (dec);
999
1000 WANT (1);
1001
1002 switch (tag)
1003 {
1004 case CBOR_TAG_MAGIC:
785 SV *sv = decode_sv (dec); 1005 sv = decode_sv (dec);
1006 break;
786 1007
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 1008 case CBOR_TAG_INDIRECTION:
1009 sv = newRV_noinc (decode_sv (dec));
1010 break;
1011
1012 case CBOR_TAG_STRINGREF_NAMESPACE:
1013 {
1014 // do not use SAVETMPS/FREETMPS, as these will
1015 // erase mortalised caches, e.g. "shareable"
1016 ENTER;
1017
1018 SAVESPTR (dec->stringref);
1019 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1020
1021 sv = decode_sv (dec);
1022
1023 LEAVE;
1024 }
1025 break;
1026
1027 case CBOR_TAG_STRINGREF:
1028 {
1029 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1030 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1031
1032 UV idx = decode_uint (dec);
1033
1034 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1035 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1036
1037 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1038 }
1039 break;
1040
1041 case CBOR_TAG_VALUE_SHAREABLE:
1042 {
1043 if (ecb_expect_false (!dec->shareable))
1044 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1045
1046 if (dec->cbor.flags & F_ALLOW_CYCLES)
1047 {
1048 sv = newSV (0);
1049 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1050
1051 SV *osv = decode_sv (dec);
1052 sv_setsv (sv, osv);
1053 SvREFCNT_dec_NN (osv);
1054 }
1055 else
1056 {
1057 av_push (dec->shareable, &PL_sv_undef);
1058 int idx = AvFILLp (dec->shareable);
1059 sv = decode_sv (dec);
1060 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1061 }
1062 }
1063 break;
1064
1065 case CBOR_TAG_VALUE_SHAREDREF:
1066 {
1067 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1068 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1069
1070 UV idx = decode_uint (dec);
1071
1072 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1073 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1074
1075 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1076
1077 if (sv == &PL_sv_undef)
1078 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1079 }
1080 break;
1081
1082 case CBOR_TAG_PERL_OBJECT:
1083 {
1084 if (dec->cbor.flags & F_FORBID_OBJECTS)
1085 goto filter;
1086
1087 sv = decode_sv (dec);
1088
1089 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1090 ERR ("corrupted CBOR data (non-array perl object)");
1091
1092 AV *av = (AV *)SvRV (sv);
1093 int len = av_len (av) + 1;
1094 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1095
1096 if (!stash)
1097 ERR ("cannot decode perl-object (package does not exist)");
1098
1099 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1100
1101 if (!method)
1102 ERR ("cannot decode perl-object (package does not have a THAW method)");
1103
1104 dSP;
1105
1106 ENTER; SAVETMPS;
1107 PUSHMARK (SP);
1108 EXTEND (SP, len + 1);
1109 // we re-bless the reference to get overload and other niceties right
1110 PUSHs (*av_fetch (av, 0, 1));
1111 PUSHs (sv_cbor);
1112
1113 int i;
1114
1115 for (i = 1; i < len; ++i)
1116 PUSHs (*av_fetch (av, i, 1));
1117
1118 PUTBACK;
1119 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1120 SPAGAIN;
1121
1122 if (SvTRUE (ERRSV))
1123 {
1124 FREETMPS; LEAVE;
1125 ERR_ERRSV;
1126 }
1127
1128 SvREFCNT_dec_NN (sv);
1129 sv = SvREFCNT_inc (POPs);
1130
1131 PUTBACK;
1132
1133 FREETMPS; LEAVE;
1134 }
1135 break;
1136
1137 default:
1138 filter:
1139 {
1140 SV *tag_sv = newSVuv (tag);
1141
1142 sv = decode_sv (dec);
1143
1144 dSP;
1145 ENTER; SAVETMPS;
1146 SAVESTACK_POS ();
1147 PUSHMARK (SP);
1148 EXTEND (SP, 2);
1149 PUSHs (tag_sv);
1150 PUSHs (sv);
1151
1152 PUTBACK;
1153 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1154 SPAGAIN;
1155
1156 if (SvTRUE (ERRSV))
1157 {
1158 SvREFCNT_dec_NN (tag_sv);
1159 FREETMPS; LEAVE;
1160 ERR_ERRSV;
1161 }
1162
1163 if (count)
1164 {
1165 SvREFCNT_dec_NN (tag_sv);
1166 SvREFCNT_dec_NN (sv);
1167 sv = SvREFCNT_inc_NN (POPs);
1168 }
1169 else
1170 {
1171 AV *av = newAV ();
1172 av_push (av, tag_sv);
1173 av_push (av, sv);
1174
1175 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1176 ? cbor_tagged_stash
1177 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1178 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1179 }
1180
1181 PUTBACK;
1182
1183 FREETMPS; LEAVE;
1184 }
1185 break;
1186 }
1187
788 return sv; 1188 return sv;
789 1189
790 AV *av = newAV (); 1190fail:
791 av_push (av, newSVuv (tag)); 1191 SvREFCNT_dec (sv);
792 av_push (av, sv); 1192 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1193}
795 1194
796static SV * 1195static SV *
797decode_sv (dec_t *dec) 1196decode_sv (dec_t *dec)
798{ 1197{
799 WANT (1); 1198 WANT (1);
800 1199
801 switch (*dec->cur >> 5) 1200 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1201 {
803 case 0: // unsigned int 1202 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1203 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1204 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1205 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1206 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1207 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1208 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1209
811 return decode_str (dec, 1); 1210 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) 1211 switch (*dec->cur++ & MINOR_MASK)
820 { 1212 {
821 case 20: 1213 case SIMPLE_FALSE:
822#if CBOR_SLOW 1214#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1215 types_false = get_bool ("Types::Serialiser::false");
824#endif 1216#endif
825 return newSVsv (cbor_false); 1217 return newSVsv (types_false);
826 case 21: 1218 case SIMPLE_TRUE:
827#if CBOR_SLOW 1219#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1220 types_true = get_bool ("Types::Serialiser::true");
829#endif 1221#endif
830 return newSVsv (cbor_true); 1222 return newSVsv (types_true);
831 case 22: 1223 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1224 return newSVsv (&PL_sv_undef);
1225 case SIMPLE_UNDEF:
1226#if CBOR_SLOW
1227 types_error = get_bool ("Types::Serialiser::error");
1228#endif
1229 return newSVsv (types_error);
833 1230
834 case 25: 1231 case MISC_FLOAT16:
835 { 1232 {
836 WANT (2); 1233 WANT (2);
837 1234
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1235 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1236 dec->cur += 2;
840 1237
841 return newSVnv (ecb_binary16_to_float (fp)); 1238 return newSVnv (ecb_binary16_to_float (fp));
842 } 1239 }
843 1240
844 case 26: 1241 case MISC_FLOAT32:
845 { 1242 {
846 uint32_t fp; 1243 uint32_t fp;
847 WANT (4); 1244 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1245 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1246 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1249 fp = ecb_bswap32 (fp);
853 1250
854 return newSVnv (ecb_binary32_to_float (fp)); 1251 return newSVnv (ecb_binary32_to_float (fp));
855 } 1252 }
856 1253
857 case 27: 1254 case MISC_FLOAT64:
858 { 1255 {
859 uint64_t fp; 1256 uint64_t fp;
860 WANT (8); 1257 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1258 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1259 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1262 fp = ecb_bswap64 (fp);
866 1263
867 return newSVnv (ecb_binary64_to_double (fp)); 1264 return newSVnv (ecb_binary64_to_double (fp));
868 } 1265 }
869 1266
870 // 0..19 unassigned 1267 // 0..19 unassigned simple
871 // 24 reserved + unassigned (reserved values are not encodable) 1268 // 24 reserved + unassigned simple (reserved values are not encodable)
1269 // 28-30 unassigned misc
1270 // 31 break code
872 default: 1271 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1272 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874 } 1273 }
875 1274
876 break; 1275 break;
877 } 1276 }
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 1277
935fail: 1278fail:
936 return &PL_sv_undef; 1279 return &PL_sv_undef;
937} 1280}
938 1281
939static SV * 1282static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1283decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1284{
942 dec_t dec; 1285 dec_t dec = { 0 };
943 SV *sv; 1286 SV *sv;
1287 STRLEN len;
1288 char *data = SvPVbyte (string, len);
944 1289
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) 1290 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", 1291 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); 1292 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1293
980 dec.cbor = *cbor; 1294 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1295 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1296 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 1297
989 sv = decode_sv (&dec); 1298 sv = decode_sv (&dec);
990 1299
991 if (offset_return) 1300 if (offset_return)
992 *offset_return = dec.cur; 1301 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1304 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1305 dec.err = "garbage after CBOR object";
997 1306
998 if (dec.err) 1307 if (dec.err)
999 { 1308 {
1309 if (dec.shareable)
1310 {
1311 // need to break cyclic links, which would all be in shareable
1312 int i;
1313 SV **svp;
1314
1315 for (i = av_len (dec.shareable) + 1; i--; )
1316 if ((svp = av_fetch (dec.shareable, i, 0)))
1317 sv_setsv (*svp, &PL_sv_undef);
1318 }
1319
1000 SvREFCNT_dec (sv); 1320 SvREFCNT_dec_NN (sv);
1321
1322 if (dec.err_sv)
1323 sv_2mortal (dec.err_sv);
1324
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1325 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002 } 1326 }
1003 1327
1004 sv = sv_2mortal (sv); 1328 sv = sv_2mortal (sv);
1005 1329
1006 return sv; 1330 return sv;
1007} 1331}
1008 1332
1333/////////////////////////////////////////////////////////////////////////////
1334// incremental parser
1335
1336#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1337
1338// returns 0 for notyet, 1 for success or error
1339static int
1340incr_parse (CBOR *self, SV *cborstr)
1341{
1342 STRLEN cur;
1343 SvPV (cborstr, cur);
1344
1345 while (ecb_expect_true (self->incr_need <= cur))
1346 {
1347 // table of integer count bytes
1348 static I8 incr_len[MINOR_MASK + 1] = {
1349 0, 0, 0, 0, 0, 0, 0, 0,
1350 0, 0, 0, 0, 0, 0, 0, 0,
1351 0, 0, 0, 0, 0, 0, 0, 0,
1352 1, 2, 4, 8,-1,-1,-1,-2
1353 };
1354
1355 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1356 U8 m = *p & MINOR_MASK;
1357 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1358 I8 ilen = incr_len[m];
1359
1360 self->incr_need = self->incr_pos + 1;
1361
1362 if (ecb_expect_false (ilen < 0))
1363 {
1364 if (m != MINOR_INDEF)
1365 return 1; // error
1366
1367 if (*p == (MAJOR_MISC | MINOR_INDEF))
1368 {
1369 if (count >= 0)
1370 return 1; // error
1371
1372 count = 1;
1373 }
1374 else
1375 {
1376 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1377 count = -1;
1378 }
1379 }
1380 else
1381 {
1382 self->incr_need += ilen;
1383 if (ecb_expect_false (self->incr_need > cur))
1384 return 0;
1385
1386 int major = *p >> MAJOR_SHIFT;
1387
1388 switch (major)
1389 {
1390 case MAJOR_TAG >> MAJOR_SHIFT:
1391 ++count; // tags merely prefix another value
1392 break;
1393
1394 case MAJOR_BYTES >> MAJOR_SHIFT:
1395 case MAJOR_TEXT >> MAJOR_SHIFT:
1396 case MAJOR_ARRAY >> MAJOR_SHIFT:
1397 case MAJOR_MAP >> MAJOR_SHIFT:
1398 {
1399 UV len;
1400
1401 if (ecb_expect_false (ilen))
1402 {
1403 len = 0;
1404
1405 do {
1406 len = (len << 8) | *++p;
1407 } while (--ilen);
1408 }
1409 else
1410 len = m;
1411
1412 switch (major)
1413 {
1414 case MAJOR_BYTES >> MAJOR_SHIFT:
1415 case MAJOR_TEXT >> MAJOR_SHIFT:
1416 self->incr_need += len;
1417 if (ecb_expect_false (self->incr_need > cur))
1418 return 0;
1419
1420 break;
1421
1422 case MAJOR_MAP >> MAJOR_SHIFT:
1423 len <<= 1;
1424 case MAJOR_ARRAY >> MAJOR_SHIFT:
1425 if (len)
1426 {
1427 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1428 count = len + 1;
1429 }
1430 break;
1431 }
1432 }
1433 }
1434 }
1435
1436 self->incr_pos = self->incr_need;
1437
1438 if (count > 0)
1439 {
1440 while (!--count)
1441 {
1442 if (!AvFILLp (self->incr_count))
1443 return 1; // done
1444
1445 SvREFCNT_dec_NN (av_pop (self->incr_count));
1446 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1447 }
1448
1449 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1450 }
1451 }
1452
1453 return 0;
1454}
1455
1456
1009///////////////////////////////////////////////////////////////////////////// 1457/////////////////////////////////////////////////////////////////////////////
1010// XS interface functions 1458// XS interface functions
1011 1459
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1460MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1461
1014BOOT: 1462BOOT:
1015{ 1463{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1464 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1465 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1466
1019 cbor_true = get_bool ("CBOR::XS::true"); 1467 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1468 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1469
1470 types_true = get_bool ("Types::Serialiser::true" );
1471 types_false = get_bool ("Types::Serialiser::false");
1472 types_error = get_bool ("Types::Serialiser::error");
1473
1474 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1475
1476 sv_cbor = newSVpv ("CBOR", 0);
1477 SvREADONLY_on (sv_cbor);
1478
1479 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1021} 1480}
1022 1481
1023PROTOTYPES: DISABLE 1482PROTOTYPES: DISABLE
1024 1483
1025void CLONE (...) 1484void CLONE (...)
1026 CODE: 1485 CODE:
1027 cbor_stash = 0; 1486 cbor_stash = 0;
1487 cbor_tagged_stash = 0;
1488 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1489 types_boolean_stash = 0;
1029 1490
1030void new (char *klass) 1491void new (char *klass)
1031 PPCODE: 1492 PPCODE:
1032{ 1493{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1494 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1502
1042void shrink (CBOR *self, int enable = 1) 1503void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1504 ALIAS:
1044 shrink = F_SHRINK 1505 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1506 allow_unknown = F_ALLOW_UNKNOWN
1507 allow_sharing = F_ALLOW_SHARING
1508 allow_cycles = F_ALLOW_CYCLES
1509 forbid_objects = F_FORBID_OBJECTS
1510 pack_strings = F_PACK_STRINGS
1511 text_keys = F_TEXT_KEYS
1512 text_strings = F_TEXT_STRINGS
1513 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1514 PPCODE:
1047{ 1515{
1048 if (enable) 1516 if (enable)
1049 self->flags |= ix; 1517 self->flags |= ix;
1050 else 1518 else
1055 1523
1056void get_shrink (CBOR *self) 1524void get_shrink (CBOR *self)
1057 ALIAS: 1525 ALIAS:
1058 get_shrink = F_SHRINK 1526 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1527 get_allow_unknown = F_ALLOW_UNKNOWN
1528 get_allow_sharing = F_ALLOW_SHARING
1529 get_allow_cycles = F_ALLOW_CYCLES
1530 get_forbid_objects = F_FORBID_OBJECTS
1531 get_pack_strings = F_PACK_STRINGS
1532 get_text_keys = F_TEXT_KEYS
1533 get_text_strings = F_TEXT_STRINGS
1534 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1535 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1536 XPUSHs (boolSV (self->flags & ix));
1062 1537
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1538void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1539 PPCODE:
1080 CODE: 1555 CODE:
1081 RETVAL = self->max_size; 1556 RETVAL = self->max_size;
1082 OUTPUT: 1557 OUTPUT:
1083 RETVAL 1558 RETVAL
1084 1559
1085#if 0 //TODO 1560void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1561 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1562 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1563 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1564 XPUSHs (ST (0));
1094}
1095 1565
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1566SV *get_filter (CBOR *self)
1097 PPCODE: 1567 CODE:
1098{ 1568 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1569 OUTPUT:
1100 self->cb_sk_object = newHV (); 1570 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 1571
1120void encode (CBOR *self, SV *scalar) 1572void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1573 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1574 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1575 XPUSHs (scalar);
1136 EXTEND (SP, 2); 1588 EXTEND (SP, 2);
1137 PUSHs (sv); 1589 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1590 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1591}
1140 1592
1593void incr_parse (CBOR *self, SV *cborstr)
1594 ALIAS:
1595 incr_parse_multiple = 1
1596 PPCODE:
1597{
1598 if (SvUTF8 (cborstr))
1599 sv_utf8_downgrade (cborstr, 0);
1600
1601 if (!self->incr_count)
1602 {
1603 self->incr_count = newAV ();
1604 self->incr_pos = 0;
1605 self->incr_need = 1;
1606
1607 av_push (self->incr_count, newSViv (1));
1608 }
1609
1610 do
1611 {
1612 if (!incr_parse (self, cborstr))
1613 {
1614 if (self->incr_need > self->max_size && self->max_size)
1615 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1616 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1617
1618 break;
1619 }
1620
1621 SV *sv;
1622 char *offset;
1623
1624 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1625 XPUSHs (sv);
1626
1627 sv_chop (cborstr, offset);
1628
1629 av_clear (self->incr_count);
1630 av_push (self->incr_count, newSViv (1));
1631
1632 self->incr_pos = 0;
1633 self->incr_need = self->incr_pos + 1;
1634 }
1635 while (ix);
1636}
1637
1638void incr_reset (CBOR *self)
1639 CODE:
1640{
1641 SvREFCNT_dec (self->incr_count);
1642 self->incr_count = 0;
1643}
1644
1141void DESTROY (CBOR *self) 1645void DESTROY (CBOR *self)
1142 CODE: 1646 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1647 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1648
1146PROTOTYPES: ENABLE 1649PROTOTYPES: ENABLE
1147 1650
1148void encode_cbor (SV *scalar) 1651void encode_cbor (SV *scalar)
1652 ALIAS:
1653 encode_cbor = 0
1654 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1655 PPCODE:
1150{ 1656{
1151 CBOR cbor; 1657 CBOR cbor;
1152 cbor_init (&cbor); 1658 cbor_init (&cbor);
1659 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1660 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1661 XPUSHs (scalar);
1155} 1662}
1156 1663
1157void decode_cbor (SV *cborstr) 1664void decode_cbor (SV *cborstr)
1161 cbor_init (&cbor); 1668 cbor_init (&cbor);
1162 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1669 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1163 XPUSHs (cborstr); 1670 XPUSHs (cborstr);
1164} 1671}
1165 1672
1673#ifdef __AFL_COMPILER
1674
1675void
1676afl_init ()
1677 CODE:
1678 __AFL_INIT ();
1679
1680int
1681afl_loop (unsigned int count = 10000)
1682 CODE:
1683 RETVAL = __AFL_LOOP (count);
1684 OUTPUT:
1685 RETVAL
1686
1687#endif
1688

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines