… | |
… | |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
9 | #include <limits.h> |
9 | #include <limits.h> |
10 | #include <float.h> |
10 | #include <float.h> |
11 | |
11 | |
|
|
12 | #define ECB_NO_THREADS 1 |
12 | #include "ecb.h" |
13 | #include "ecb.h" |
13 | |
14 | |
14 | // compatibility with perl <5.18 |
15 | // compatibility with perl <5.18 |
15 | #ifndef HvNAMELEN_get |
16 | #ifndef HvNAMELEN_get |
16 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
17 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
… | |
… | |
19 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
20 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
20 | #endif |
21 | #endif |
21 | #ifndef HvNAMEUTF8 |
22 | #ifndef HvNAMEUTF8 |
22 | # define HvNAMEUTF8(hv) 0 |
23 | # define HvNAMEUTF8(hv) 0 |
23 | #endif |
24 | #endif |
|
|
25 | #ifndef SvREFCNT_dec_NN |
|
|
26 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
|
|
27 | #endif |
|
|
28 | |
|
|
29 | // known major and minor types |
|
|
30 | enum cbor_type |
|
|
31 | { |
|
|
32 | MAJOR_SHIFT = 5, |
|
|
33 | MINOR_MASK = 0x1f, |
|
|
34 | |
|
|
35 | MAJOR_POS_INT = 0 << MAJOR_SHIFT, |
|
|
36 | MAJOR_NEG_INT = 1 << MAJOR_SHIFT, |
|
|
37 | MAJOR_BYTES = 2 << MAJOR_SHIFT, |
|
|
38 | MAJOR_TEXT = 3 << MAJOR_SHIFT, |
|
|
39 | MAJOR_ARRAY = 4 << MAJOR_SHIFT, |
|
|
40 | MAJOR_MAP = 5 << MAJOR_SHIFT, |
|
|
41 | MAJOR_TAG = 6 << MAJOR_SHIFT, |
|
|
42 | MAJOR_MISC = 7 << MAJOR_SHIFT, |
|
|
43 | |
|
|
44 | // INT/STRING/ARRAY/MAP subtypes |
|
|
45 | LENGTH_EXT1 = 24, |
|
|
46 | LENGTH_EXT2 = 25, |
|
|
47 | LENGTH_EXT4 = 26, |
|
|
48 | LENGTH_EXT8 = 27, |
|
|
49 | |
|
|
50 | // SIMPLE types (effectively MISC subtypes) |
|
|
51 | SIMPLE_FALSE = 20, |
|
|
52 | SIMPLE_TRUE = 21, |
|
|
53 | SIMPLE_NULL = 22, |
|
|
54 | SIMPLE_UNDEF = 23, |
|
|
55 | |
|
|
56 | // MISC subtype (unused) |
|
|
57 | MISC_EXT1 = 24, |
|
|
58 | MISC_FLOAT16 = 25, |
|
|
59 | MISC_FLOAT32 = 26, |
|
|
60 | MISC_FLOAT64 = 27, |
|
|
61 | |
|
|
62 | // BYTES/TEXT/ARRAY/MAP |
|
|
63 | MINOR_INDEF = 31, |
|
|
64 | }; |
24 | |
65 | |
25 | // known tags |
66 | // known tags |
26 | enum cbor_tag |
67 | enum cbor_tag |
27 | { |
68 | { |
28 | // inofficial extensions (pending iana registration) |
69 | // extensions |
|
|
70 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
29 | CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object |
71 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
30 | CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object |
72 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
31 | CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing |
73 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
32 | CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing |
74 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
33 | CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref |
75 | CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
34 | CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref |
|
|
35 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
76 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
36 | |
77 | |
37 | // rfc7049 |
78 | // rfc7049 |
38 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
79 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
39 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
80 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
40 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
81 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
41 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
82 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
42 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
83 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
43 | CBOR_TAG_BIGFLOAT = 5, // array |
84 | CBOR_TAG_BIGFLOAT = 5, // array |
44 | |
85 | |
45 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
86 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
46 | CBOR_TAG_CONV_B64 = 22, // base64, any |
87 | CBOR_TAG_CONV_B64 = 22, // base64, any |
47 | CBOR_TAG_CONV_HEX = 23, // base16, any |
88 | CBOR_TAG_CONV_HEX = 23, // base16, any |
48 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
89 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
49 | |
90 | |
50 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
91 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
51 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
92 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
52 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
93 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
53 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
94 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
54 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
95 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
55 | |
96 | |
56 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
97 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
57 | }; |
98 | }; |
58 | |
99 | |
59 | #define F_SHRINK 0x00000001UL |
100 | #define F_SHRINK 0x00000001UL |
60 | #define F_ALLOW_UNKNOWN 0x00000002UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
61 | #define F_ALLOW_SHARING 0x00000004UL //TODO |
102 | #define F_ALLOW_SHARING 0x00000004UL |
|
|
103 | #define F_ALLOW_CYCLES 0x00000008UL |
|
|
104 | #define F_FORBID_OBJECTS 0x00000010UL |
62 | #define F_DEDUP_STRINGS 0x00000008UL //TODO |
105 | #define F_PACK_STRINGS 0x00000020UL |
63 | #define F_DEDUP_KEYS 0x00000010UL //TODO |
106 | #define F_TEXT_KEYS 0x00000040UL |
|
|
107 | #define F_TEXT_STRINGS 0x00000080UL |
|
|
108 | #define F_VALIDATE_UTF8 0x00000100UL |
64 | |
109 | |
65 | #define INIT_SIZE 32 // initial scalar size to be allocated |
110 | #define INIT_SIZE 32 // initial scalar size to be allocated |
66 | |
111 | |
67 | #define SB do { |
112 | #define SB do { |
68 | #define SE } while (0) |
113 | #define SE } while (0) |
… | |
… | |
80 | # define CBOR_SLOW 0 |
125 | # define CBOR_SLOW 0 |
81 | # define CBOR_STASH cbor_stash |
126 | # define CBOR_STASH cbor_stash |
82 | #endif |
127 | #endif |
83 | |
128 | |
84 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
129 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
85 | static SV *types_true, *types_false, *types_error, *sv_cbor; |
130 | static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter; |
86 | |
131 | |
87 | typedef struct { |
132 | typedef struct { |
88 | U32 flags; |
133 | U32 flags; |
89 | U32 max_depth; |
134 | U32 max_depth; |
90 | STRLEN max_size; |
135 | STRLEN max_size; |
|
|
136 | SV *filter; |
|
|
137 | |
|
|
138 | // for the incremental parser |
|
|
139 | STRLEN incr_pos; // the current offset into the text |
|
|
140 | STRLEN incr_need; // minimum bytes needed to decode |
|
|
141 | AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef. |
91 | } CBOR; |
142 | } CBOR; |
92 | |
143 | |
93 | ecb_inline void |
144 | ecb_inline void |
94 | cbor_init (CBOR *cbor) |
145 | cbor_init (CBOR *cbor) |
95 | { |
146 | { |
96 | Zero (cbor, 1, CBOR); |
147 | Zero (cbor, 1, CBOR); |
97 | cbor->max_depth = 512; |
148 | cbor->max_depth = 512; |
|
|
149 | } |
|
|
150 | |
|
|
151 | ecb_inline void |
|
|
152 | cbor_free (CBOR *cbor) |
|
|
153 | { |
|
|
154 | SvREFCNT_dec (cbor->filter); |
|
|
155 | SvREFCNT_dec (cbor->incr_count); |
98 | } |
156 | } |
99 | |
157 | |
100 | ///////////////////////////////////////////////////////////////////////////// |
158 | ///////////////////////////////////////////////////////////////////////////// |
101 | // utility functions |
159 | // utility functions |
102 | |
160 | |
… | |
… | |
124 | SvPV_renew (sv, SvCUR (sv) + 1); |
182 | SvPV_renew (sv, SvCUR (sv) + 1); |
125 | #endif |
183 | #endif |
126 | } |
184 | } |
127 | } |
185 | } |
128 | |
186 | |
|
|
187 | // minimum length of a string to be registered for stringref |
|
|
188 | ecb_inline int |
|
|
189 | minimum_string_length (UV idx) |
|
|
190 | { |
|
|
191 | return idx > 23 |
|
|
192 | ? idx > 0xffU |
|
|
193 | ? idx > 0xffffU |
|
|
194 | ? idx > 0xffffffffU |
|
|
195 | ? 11 |
|
|
196 | : 7 |
|
|
197 | : 5 |
|
|
198 | : 4 |
|
|
199 | : 3; |
|
|
200 | } |
|
|
201 | |
129 | ///////////////////////////////////////////////////////////////////////////// |
202 | ///////////////////////////////////////////////////////////////////////////// |
130 | // encoder |
203 | // encoder |
131 | |
204 | |
132 | // structure used for encoding CBOR |
205 | // structure used for encoding CBOR |
133 | typedef struct |
206 | typedef struct |
… | |
… | |
135 | char *cur; // SvPVX (sv) + current output position |
208 | char *cur; // SvPVX (sv) + current output position |
136 | char *end; // SvEND (sv) |
209 | char *end; // SvEND (sv) |
137 | SV *sv; // result scalar |
210 | SV *sv; // result scalar |
138 | CBOR cbor; |
211 | CBOR cbor; |
139 | U32 depth; // recursion level |
212 | U32 depth; // recursion level |
140 | HV *stringref; // string => index, or 0 |
213 | HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8) |
|
|
214 | UV stringref_idx; |
141 | HV *shareable; // ptr => index, or 0 |
215 | HV *shareable; // ptr => index, or 0 |
142 | UV shareable_idx; |
216 | UV shareable_idx; |
143 | } enc_t; |
217 | } enc_t; |
144 | |
218 | |
145 | ecb_inline void |
219 | ecb_inline void |
… | |
… | |
164 | static void |
238 | static void |
165 | encode_uint (enc_t *enc, int major, UV len) |
239 | encode_uint (enc_t *enc, int major, UV len) |
166 | { |
240 | { |
167 | need (enc, 9); |
241 | need (enc, 9); |
168 | |
242 | |
169 | if (len < 24) |
243 | if (ecb_expect_true (len < LENGTH_EXT1)) |
170 | *enc->cur++ = major | len; |
244 | *enc->cur++ = major | len; |
171 | else if (len <= 0xff) |
245 | else if (ecb_expect_true (len <= 0xffU)) |
172 | { |
246 | { |
173 | *enc->cur++ = major | 24; |
247 | *enc->cur++ = major | LENGTH_EXT1; |
174 | *enc->cur++ = len; |
248 | *enc->cur++ = len; |
175 | } |
249 | } |
176 | else if (len <= 0xffff) |
250 | else if (len <= 0xffffU) |
177 | { |
251 | { |
178 | *enc->cur++ = major | 25; |
252 | *enc->cur++ = major | LENGTH_EXT2; |
179 | *enc->cur++ = len >> 8; |
253 | *enc->cur++ = len >> 8; |
180 | *enc->cur++ = len; |
254 | *enc->cur++ = len; |
181 | } |
255 | } |
182 | else if (len <= 0xffffffff) |
256 | else if (len <= 0xffffffffU) |
183 | { |
257 | { |
184 | *enc->cur++ = major | 26; |
258 | *enc->cur++ = major | LENGTH_EXT4; |
185 | *enc->cur++ = len >> 24; |
259 | *enc->cur++ = len >> 24; |
186 | *enc->cur++ = len >> 16; |
260 | *enc->cur++ = len >> 16; |
187 | *enc->cur++ = len >> 8; |
261 | *enc->cur++ = len >> 8; |
188 | *enc->cur++ = len; |
262 | *enc->cur++ = len; |
189 | } |
263 | } |
190 | else |
264 | else |
191 | { |
265 | { |
192 | *enc->cur++ = major | 27; |
266 | *enc->cur++ = major | LENGTH_EXT8; |
193 | *enc->cur++ = len >> 56; |
267 | *enc->cur++ = len >> 56; |
194 | *enc->cur++ = len >> 48; |
268 | *enc->cur++ = len >> 48; |
195 | *enc->cur++ = len >> 40; |
269 | *enc->cur++ = len >> 40; |
196 | *enc->cur++ = len >> 32; |
270 | *enc->cur++ = len >> 32; |
197 | *enc->cur++ = len >> 24; |
271 | *enc->cur++ = len >> 24; |
… | |
… | |
199 | *enc->cur++ = len >> 8; |
273 | *enc->cur++ = len >> 8; |
200 | *enc->cur++ = len; |
274 | *enc->cur++ = len; |
201 | } |
275 | } |
202 | } |
276 | } |
203 | |
277 | |
204 | static void |
278 | ecb_inline void |
|
|
279 | encode_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 |
|
|
285 | ecb_noinline static void |
205 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
286 | encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) |
206 | { |
287 | { |
207 | 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 | |
|
|
307 | ecb_inline void |
|
|
308 | encode_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); |
208 | need (enc, len); |
318 | need (enc, len); |
209 | memcpy (enc->cur, str, len); |
319 | memcpy (enc->cur, str, len); |
210 | enc->cur += len; |
320 | enc->cur += len; |
211 | } |
321 | } |
212 | |
322 | |
213 | ecb_inline void |
323 | ecb_inline void |
214 | encode_tag (enc_t *enc, UV tag) |
324 | encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
215 | { |
325 | { |
216 | encode_uint (enc, 0xc0, tag); |
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); |
217 | } |
346 | } |
218 | |
347 | |
219 | static void encode_sv (enc_t *enc, SV *sv); |
348 | static void encode_sv (enc_t *enc, SV *sv); |
220 | |
349 | |
221 | static void |
350 | static void |
… | |
… | |
226 | if (enc->depth >= enc->cbor.max_depth) |
355 | if (enc->depth >= enc->cbor.max_depth) |
227 | croak (ERR_NESTING_EXCEEDED); |
356 | croak (ERR_NESTING_EXCEEDED); |
228 | |
357 | |
229 | ++enc->depth; |
358 | ++enc->depth; |
230 | |
359 | |
231 | encode_uint (enc, 0x80, len + 1); |
360 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
232 | |
361 | |
|
|
362 | if (SvMAGICAL (av)) |
233 | for (i = 0; i <= len; ++i) |
363 | for (i = 0; i <= len; ++i) |
234 | { |
364 | { |
235 | SV **svp = av_fetch (av, i, 0); |
365 | SV **svp = av_fetch (av, i, 0); |
236 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
366 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
237 | } |
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 | } |
238 | |
374 | |
239 | --enc->depth; |
375 | --enc->depth; |
240 | } |
376 | } |
241 | |
377 | |
242 | static void |
378 | static void |
… | |
… | |
251 | |
387 | |
252 | int pairs = hv_iterinit (hv); |
388 | int pairs = hv_iterinit (hv); |
253 | int mg = SvMAGICAL (hv); |
389 | int mg = SvMAGICAL (hv); |
254 | |
390 | |
255 | if (mg) |
391 | if (mg) |
256 | encode_ch (enc, 0xa0 | 31); |
392 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
257 | else |
393 | else |
258 | encode_uint (enc, 0xa0, pairs); |
394 | encode_uint (enc, MAJOR_MAP, pairs); |
259 | |
395 | |
260 | while ((he = hv_iternext (hv))) |
396 | while ((he = hv_iternext (hv))) |
261 | { |
397 | { |
262 | if (HeKLEN (he) == HEf_SVKEY) |
398 | if (HeKLEN (he) == HEf_SVKEY) |
263 | encode_sv (enc, HeSVKEY (he)); |
399 | encode_sv (enc, HeSVKEY (he)); |
264 | else |
400 | else |
265 | 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)); |
266 | |
402 | |
267 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
403 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
268 | } |
404 | } |
269 | |
405 | |
270 | if (mg) |
406 | if (mg) |
271 | encode_ch (enc, 0xe0 | 31); |
407 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
272 | |
408 | |
273 | --enc->depth; |
409 | --enc->depth; |
274 | } |
410 | } |
275 | |
411 | |
276 | // encode objects, arrays and special \0=false and \1=true values. |
412 | // encode objects, arrays and special \0=false and \1=true values. |
277 | static void |
413 | static void |
278 | encode_rv (enc_t *enc, SV *sv) |
414 | encode_rv (enc_t *enc, SV *sv) |
279 | { |
415 | { |
280 | SvGETMAGIC (sv); |
416 | SvGETMAGIC (sv); |
281 | |
|
|
282 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING) |
|
|
283 | && ecb_expect_false (SvREFCNT (sv) > 1)) |
|
|
284 | { |
|
|
285 | if (!enc->shareable) |
|
|
286 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
287 | |
|
|
288 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
289 | |
|
|
290 | if (SvOK (*svp)) |
|
|
291 | { |
|
|
292 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
293 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
294 | return; |
|
|
295 | } |
|
|
296 | else |
|
|
297 | { |
|
|
298 | sv_setuv (*svp, enc->shareable_idx); |
|
|
299 | ++enc->shareable_idx; |
|
|
300 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
301 | } |
|
|
302 | } |
|
|
303 | |
417 | |
304 | svtype svt = SvTYPE (sv); |
418 | svtype svt = SvTYPE (sv); |
305 | |
419 | |
306 | if (ecb_expect_false (SvOBJECT (sv))) |
420 | if (ecb_expect_false (SvOBJECT (sv))) |
307 | { |
421 | { |
… | |
… | |
314 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
428 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
315 | ? cbor_tagged_stash |
429 | ? cbor_tagged_stash |
316 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
430 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
317 | |
431 | |
318 | HV *stash = SvSTASH (sv); |
432 | HV *stash = SvSTASH (sv); |
319 | GV *method; |
|
|
320 | |
433 | |
321 | if (stash == boolean_stash) |
434 | if (stash == boolean_stash) |
322 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
435 | { |
|
|
436 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
|
|
437 | return; |
|
|
438 | } |
323 | else if (stash == error_stash) |
439 | else if (stash == error_stash) |
324 | encode_ch (enc, 0xe0 | 23); |
440 | { |
|
|
441 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
|
|
442 | return; |
|
|
443 | } |
325 | else if (stash == tagged_stash) |
444 | else if (stash == tagged_stash) |
326 | { |
445 | { |
327 | if (svt != SVt_PVAV) |
446 | if (svt != SVt_PVAV) |
328 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
447 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
329 | |
448 | |
330 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
449 | encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
331 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 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 (!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)) |
332 | } |
465 | { |
|
|
466 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
467 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
|
|
468 | return; |
|
|
469 | } |
|
|
470 | else |
|
|
471 | { |
|
|
472 | sv_setuv (*svp, enc->shareable_idx); |
|
|
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)))); |
333 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
486 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
334 | { |
487 | { |
335 | dSP; |
488 | dSP; |
336 | |
489 | |
337 | ENTER; SAVETMPS; PUSHMARK (SP); |
490 | ENTER; SAVETMPS; |
|
|
491 | PUSHMARK (SP); |
338 | // 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 |
339 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
493 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
340 | |
494 | |
341 | PUTBACK; |
495 | PUTBACK; |
342 | // G_SCALAR ensures that return value is 1 |
496 | // G_SCALAR ensures that return value is 1 |
… | |
… | |
355 | } |
509 | } |
356 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
510 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
357 | { |
511 | { |
358 | dSP; |
512 | dSP; |
359 | |
513 | |
360 | ENTER; SAVETMPS; PUSHMARK (SP); |
514 | ENTER; SAVETMPS; |
|
|
515 | SAVESTACK_POS (); |
|
|
516 | PUSHMARK (SP); |
361 | EXTEND (SP, 2); |
517 | EXTEND (SP, 2); |
362 | // we re-bless the reference to get overload and other niceties right |
518 | // we re-bless the reference to get overload and other niceties right |
363 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
519 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
364 | PUSHs (sv_cbor); |
520 | PUSHs (sv_cbor); |
365 | |
521 | |
… | |
… | |
370 | // catch this surprisingly common error |
526 | // catch this surprisingly common error |
371 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
527 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
372 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
528 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
373 | |
529 | |
374 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
530 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
375 | encode_uint (enc, 0x80, count + 1); |
531 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
376 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
532 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
377 | |
533 | |
378 | while (count) |
534 | while (count) |
379 | encode_sv (enc, SP[1 - count--]); |
535 | encode_sv (enc, SP[1 - count--]); |
380 | |
536 | |
381 | PUTBACK; |
537 | PUTBACK; |
… | |
… | |
402 | { |
558 | { |
403 | double nv = SvNVX (sv); |
559 | double nv = SvNVX (sv); |
404 | |
560 | |
405 | need (enc, 9); |
561 | need (enc, 9); |
406 | |
562 | |
407 | if (ecb_expect_false (nv == (U32)nv)) |
563 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
408 | encode_uint (enc, 0x00, (U32)nv); |
564 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
409 | //TODO: maybe I32? |
565 | //TODO: maybe I32? |
410 | else if (ecb_expect_false (nv == (float)nv)) |
566 | else if (ecb_expect_false (nv == (float)nv)) |
411 | { |
567 | { |
412 | uint32_t fp = ecb_float_to_binary32 (nv); |
568 | uint32_t fp = ecb_float_to_binary32 (nv); |
413 | |
569 | |
414 | *enc->cur++ = 0xe0 | 26; |
570 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
415 | |
571 | |
416 | if (!ecb_big_endian ()) |
572 | if (!ecb_big_endian ()) |
417 | fp = ecb_bswap32 (fp); |
573 | fp = ecb_bswap32 (fp); |
418 | |
574 | |
419 | memcpy (enc->cur, &fp, 4); |
575 | memcpy (enc->cur, &fp, 4); |
… | |
… | |
421 | } |
577 | } |
422 | else |
578 | else |
423 | { |
579 | { |
424 | uint64_t fp = ecb_double_to_binary64 (nv); |
580 | uint64_t fp = ecb_double_to_binary64 (nv); |
425 | |
581 | |
426 | *enc->cur++ = 0xe0 | 27; |
582 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
427 | |
583 | |
428 | if (!ecb_big_endian ()) |
584 | if (!ecb_big_endian ()) |
429 | fp = ecb_bswap64 (fp); |
585 | fp = ecb_bswap64 (fp); |
430 | |
586 | |
431 | memcpy (enc->cur, &fp, 8); |
587 | memcpy (enc->cur, &fp, 8); |
… | |
… | |
440 | |
596 | |
441 | if (SvPOKp (sv)) |
597 | if (SvPOKp (sv)) |
442 | { |
598 | { |
443 | STRLEN len; |
599 | STRLEN len; |
444 | char *str = SvPV (sv, len); |
600 | char *str = SvPV (sv, len); |
445 | encode_str (enc, SvUTF8 (sv), str, len); |
601 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
446 | } |
602 | } |
447 | else if (SvNOKp (sv)) |
603 | else if (SvNOKp (sv)) |
448 | encode_nv (enc, sv); |
604 | encode_nv (enc, sv); |
449 | else if (SvIOKp (sv)) |
605 | else if (SvIOKp (sv)) |
450 | { |
606 | { |
451 | if (SvIsUV (sv)) |
607 | if (SvIsUV (sv)) |
452 | encode_uint (enc, 0x00, SvUVX (sv)); |
608 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
453 | else if (SvIVX (sv) >= 0) |
609 | else if (SvIVX (sv) >= 0) |
454 | encode_uint (enc, 0x00, SvIVX (sv)); |
610 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
455 | else |
611 | else |
456 | encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); |
612 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
457 | } |
613 | } |
458 | else if (SvROK (sv)) |
614 | else if (SvROK (sv)) |
459 | encode_rv (enc, SvRV (sv)); |
615 | encode_rv (enc, SvRV (sv)); |
460 | else if (!SvOK (sv)) |
616 | else if (!SvOK (sv)) |
461 | encode_ch (enc, 0xe0 | 22); |
617 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
462 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
618 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
463 | encode_ch (enc, 0xe0 | 23); |
619 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
464 | else |
620 | else |
465 | 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", |
466 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
622 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
467 | } |
623 | } |
468 | |
624 | |
469 | static SV * |
625 | static SV * |
470 | encode_cbor (SV *scalar, CBOR *cbor) |
626 | encode_cbor (SV *scalar, CBOR *cbor) |
471 | { |
627 | { |
472 | enc_t enc = { }; |
628 | enc_t enc = { 0 }; |
473 | |
629 | |
474 | enc.cbor = *cbor; |
630 | enc.cbor = *cbor; |
475 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
631 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
476 | enc.cur = SvPVX (enc.sv); |
632 | enc.cur = SvPVX (enc.sv); |
477 | enc.end = SvEND (enc.sv); |
633 | enc.end = SvEND (enc.sv); |
478 | |
634 | |
479 | 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 | |
480 | encode_sv (&enc, scalar); |
644 | encode_sv (&enc, scalar); |
481 | |
645 | |
482 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
646 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
483 | *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 |
484 | |
648 | |
… | |
… | |
499 | const char *err; // parse error, if != 0 |
663 | const char *err; // parse error, if != 0 |
500 | CBOR cbor; |
664 | CBOR cbor; |
501 | U32 depth; // recursion depth |
665 | U32 depth; // recursion depth |
502 | U32 maxdepth; // recursion depth limit |
666 | U32 maxdepth; // recursion depth limit |
503 | AV *shareable; |
667 | AV *shareable; |
|
|
668 | AV *stringref; |
|
|
669 | SV *decode_tagged; |
|
|
670 | SV *err_sv; // optional sv for error, needs to be freed |
504 | } dec_t; |
671 | } dec_t; |
505 | |
672 | |
506 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
673 | // set dec->err to ERRSV |
|
|
674 | ecb_cold static void |
|
|
675 | err_errsv (dec_t *dec) |
|
|
676 | { |
|
|
677 | if (!dec->err) |
|
|
678 | { |
|
|
679 | dec->err_sv = newSVsv (ERRSV); |
507 | |
680 | |
508 | #define WANT(len) if (ecb_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; |
509 | |
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 |
|
|
690 | ecb_cold static void |
|
|
691 | err_set (dec_t *dec, const char *reason) |
|
|
692 | { |
|
|
693 | if (!dec->err) |
|
|
694 | dec->err = reason; |
|
|
695 | } |
|
|
696 | |
|
|
697 | ecb_cold static void |
|
|
698 | err_unexpected_end (dec_t *dec) |
|
|
699 | { |
|
|
700 | err_set (dec, "unexpected end of CBOR data"); |
|
|
701 | } |
|
|
702 | |
|
|
703 | ecb_cold static void |
|
|
704 | err_nesting_exceeded (dec_t *dec) |
|
|
705 | { |
|
|
706 | err_set (dec, ERR_NESTING_EXCEEDED); |
|
|
707 | } |
|
|
708 | |
|
|
709 | #define ERR_DO(do) SB do; goto fail; SE |
|
|
710 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
|
|
711 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
|
|
712 | |
|
|
713 | #define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec)) |
|
|
714 | |
510 | #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
715 | #define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED) |
511 | #define DEC_DEC_DEPTH --dec->depth |
716 | #define DEC_DEC_DEPTH --dec->depth |
512 | |
717 | |
513 | static UV |
718 | static UV |
514 | decode_uint (dec_t *dec) |
719 | decode_uint (dec_t *dec) |
515 | { |
720 | { |
516 | switch (*dec->cur & 31) |
721 | U8 m = *dec->cur & MINOR_MASK; |
517 | { |
722 | ++dec->cur; |
518 | case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: |
|
|
519 | case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: |
|
|
520 | case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23: |
|
|
521 | return *dec->cur++ & 31; |
|
|
522 | |
723 | |
523 | case 24: |
724 | if (ecb_expect_true (m < LENGTH_EXT1)) |
|
|
725 | return m; |
|
|
726 | else if (ecb_expect_true (m == LENGTH_EXT1)) |
|
|
727 | { |
524 | WANT (2); |
728 | WANT (1); |
525 | dec->cur += 2; |
729 | dec->cur += 1; |
526 | return dec->cur[-1]; |
730 | return dec->cur[-1]; |
527 | |
731 | } |
528 | case 25: |
732 | else if (ecb_expect_true (m == LENGTH_EXT2)) |
|
|
733 | { |
529 | WANT (3); |
734 | WANT (2); |
530 | dec->cur += 3; |
735 | dec->cur += 2; |
531 | return (((UV)dec->cur[-2]) << 8) |
736 | return (((UV)dec->cur[-2]) << 8) |
532 | | ((UV)dec->cur[-1]); |
737 | | ((UV)dec->cur[-1]); |
533 | |
738 | } |
534 | case 26: |
739 | else if (ecb_expect_true (m == LENGTH_EXT4)) |
|
|
740 | { |
535 | WANT (5); |
741 | WANT (4); |
536 | dec->cur += 5; |
742 | dec->cur += 4; |
537 | return (((UV)dec->cur[-4]) << 24) |
743 | return (((UV)dec->cur[-4]) << 24) |
538 | | (((UV)dec->cur[-3]) << 16) |
744 | | (((UV)dec->cur[-3]) << 16) |
539 | | (((UV)dec->cur[-2]) << 8) |
745 | | (((UV)dec->cur[-2]) << 8) |
540 | | ((UV)dec->cur[-1]); |
746 | | ((UV)dec->cur[-1]); |
541 | |
747 | } |
542 | case 27: |
748 | else if (ecb_expect_true (m == LENGTH_EXT8)) |
|
|
749 | { |
543 | WANT (9); |
750 | WANT (8); |
544 | dec->cur += 9; |
751 | dec->cur += 8; |
|
|
752 | |
|
|
753 | return |
|
|
754 | #if UVSIZE < 8 |
|
|
755 | 0 |
|
|
756 | #else |
545 | return (((UV)dec->cur[-8]) << 56) |
757 | (((UV)dec->cur[-8]) << 56) |
546 | | (((UV)dec->cur[-7]) << 48) |
758 | | (((UV)dec->cur[-7]) << 48) |
547 | | (((UV)dec->cur[-6]) << 40) |
759 | | (((UV)dec->cur[-6]) << 40) |
548 | | (((UV)dec->cur[-5]) << 32) |
760 | | (((UV)dec->cur[-5]) << 32) |
|
|
761 | #endif |
549 | | (((UV)dec->cur[-4]) << 24) |
762 | | (((UV)dec->cur[-4]) << 24) |
550 | | (((UV)dec->cur[-3]) << 16) |
763 | | (((UV)dec->cur[-3]) << 16) |
551 | | (((UV)dec->cur[-2]) << 8) |
764 | | (((UV)dec->cur[-2]) << 8) |
552 | | ((UV)dec->cur[-1]); |
765 | | ((UV)dec->cur[-1]); |
553 | |
766 | } |
554 | default: |
767 | else |
555 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
768 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
556 | } |
|
|
557 | |
769 | |
558 | fail: |
770 | fail: |
559 | return 0; |
771 | return 0; |
560 | } |
772 | } |
561 | |
773 | |
… | |
… | |
566 | { |
778 | { |
567 | AV *av = newAV (); |
779 | AV *av = newAV (); |
568 | |
780 | |
569 | DEC_INC_DEPTH; |
781 | DEC_INC_DEPTH; |
570 | |
782 | |
571 | if ((*dec->cur & 31) == 31) |
783 | if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF)) |
572 | { |
784 | { |
573 | ++dec->cur; |
785 | ++dec->cur; |
574 | |
786 | |
575 | for (;;) |
787 | for (;;) |
576 | { |
788 | { |
577 | WANT (1); |
789 | WANT (1); |
578 | |
790 | |
579 | if (*dec->cur == (0xe0 | 31)) |
791 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
580 | { |
792 | { |
581 | ++dec->cur; |
793 | ++dec->cur; |
582 | break; |
794 | break; |
583 | } |
795 | } |
584 | |
796 | |
585 | av_push (av, decode_sv (dec)); |
797 | av_push (av, decode_sv (dec)); |
586 | } |
798 | } |
587 | } |
799 | } |
588 | else |
800 | else |
589 | { |
801 | { |
590 | int i, len = decode_uint (dec); |
802 | UV i, len = decode_uint (dec); |
591 | |
803 | |
|
|
804 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
592 | av_fill (av, len - 1); |
805 | av_fill (av, len - 1); |
593 | |
806 | |
594 | for (i = 0; i < len; ++i) |
807 | for (i = 0; i < len; ++i) |
595 | AvARRAY (av)[i] = decode_sv (dec); |
808 | AvARRAY (av)[i] = decode_sv (dec); |
596 | } |
809 | } |
… | |
… | |
606 | |
819 | |
607 | static void |
820 | static void |
608 | decode_he (dec_t *dec, HV *hv) |
821 | decode_he (dec_t *dec, HV *hv) |
609 | { |
822 | { |
610 | // for speed reasons, we specialcase single-string |
823 | // for speed reasons, we specialcase single-string |
611 | // byte or utf-8 strings as keys. |
824 | // byte or utf-8 strings as keys, but only when !stringref |
612 | |
825 | |
613 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
826 | if (ecb_expect_true (!dec->stringref)) |
|
|
827 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
614 | { |
828 | { |
615 | I32 len = decode_uint (dec); |
829 | STRLEN len = decode_uint (dec); |
616 | char *key = (char *)dec->cur; |
830 | char *key = (char *)dec->cur; |
617 | |
831 | |
|
|
832 | WANT (len); |
618 | dec->cur += len; |
833 | dec->cur += len; |
619 | |
834 | |
620 | hv_store (hv, key, len, decode_sv (dec), 0); |
835 | hv_store (hv, key, len, decode_sv (dec), 0); |
|
|
836 | |
|
|
837 | return; |
621 | } |
838 | } |
622 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
839 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
623 | { |
840 | { |
624 | I32 len = decode_uint (dec); |
841 | STRLEN len = decode_uint (dec); |
625 | char *key = (char *)dec->cur; |
842 | char *key = (char *)dec->cur; |
626 | |
843 | |
|
|
844 | WANT (len); |
627 | dec->cur += len; |
845 | dec->cur += len; |
628 | |
846 | |
|
|
847 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
|
|
848 | if (!is_utf8_string (key, len)) |
|
|
849 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
|
|
850 | |
629 | hv_store (hv, key, -len, decode_sv (dec), 0); |
851 | hv_store (hv, key, -len, decode_sv (dec), 0); |
|
|
852 | |
|
|
853 | return; |
630 | } |
854 | } |
631 | else |
855 | |
632 | { |
|
|
633 | SV *k = decode_sv (dec); |
856 | SV *k = decode_sv (dec); |
634 | SV *v = decode_sv (dec); |
857 | SV *v = decode_sv (dec); |
635 | |
858 | |
|
|
859 | // we leak memory if uncaught exceptions are thrown by random magical |
|
|
860 | // methods, and this is hopefully the only place where it can happen, |
|
|
861 | // so if there is a chance of an exception, take the very slow path. |
|
|
862 | // since catching exceptions is "undocumented/internal/forbidden" by |
|
|
863 | // the new p5p powers, we need to call out to a perl function :/ |
|
|
864 | if (ecb_expect_false (SvAMAGIC (k))) |
|
|
865 | { |
|
|
866 | dSP; |
|
|
867 | |
|
|
868 | ENTER; SAVETMPS; |
|
|
869 | PUSHMARK (SP); |
|
|
870 | EXTEND (SP, 3); |
|
|
871 | PUSHs (sv_2mortal (newRV_inc ((SV *)hv))); |
|
|
872 | PUSHs (sv_2mortal (k)); |
|
|
873 | PUSHs (sv_2mortal (v)); |
|
|
874 | |
|
|
875 | PUTBACK; |
|
|
876 | call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL); |
|
|
877 | SPAGAIN; |
|
|
878 | |
|
|
879 | FREETMPS; LEAVE; |
|
|
880 | |
|
|
881 | if (SvTRUE (ERRSV)) |
|
|
882 | ERR_ERRSV; |
|
|
883 | |
|
|
884 | return; |
|
|
885 | } |
|
|
886 | |
636 | hv_store_ent (hv, k, v, 0); |
887 | hv_store_ent (hv, k, v, 0); |
637 | SvREFCNT_dec (k); |
888 | SvREFCNT_dec (k); |
638 | } |
889 | |
|
|
890 | fail: |
|
|
891 | ; |
639 | } |
892 | } |
640 | |
893 | |
641 | static SV * |
894 | static SV * |
642 | decode_hv (dec_t *dec) |
895 | decode_hv (dec_t *dec) |
643 | { |
896 | { |
644 | HV *hv = newHV (); |
897 | HV *hv = newHV (); |
645 | |
898 | |
646 | DEC_INC_DEPTH; |
899 | DEC_INC_DEPTH; |
647 | |
900 | |
648 | if ((*dec->cur & 31) == 31) |
901 | if (*dec->cur == (MAJOR_MAP | MINOR_INDEF)) |
649 | { |
902 | { |
650 | ++dec->cur; |
903 | ++dec->cur; |
651 | |
904 | |
652 | for (;;) |
905 | for (;;) |
653 | { |
906 | { |
654 | WANT (1); |
907 | WANT (1); |
655 | |
908 | |
656 | if (*dec->cur == (0xe0 | 31)) |
909 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
657 | { |
910 | { |
658 | ++dec->cur; |
911 | ++dec->cur; |
659 | break; |
912 | break; |
660 | } |
913 | } |
661 | |
914 | |
662 | decode_he (dec, hv); |
915 | decode_he (dec, hv); |
663 | } |
916 | } |
664 | } |
917 | } |
665 | else |
918 | else |
666 | { |
919 | { |
667 | int pairs = decode_uint (dec); |
920 | UV pairs = decode_uint (dec); |
|
|
921 | |
|
|
922 | WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes |
668 | |
923 | |
669 | while (pairs--) |
924 | while (pairs--) |
670 | decode_he (dec, hv); |
925 | decode_he (dec, hv); |
671 | } |
926 | } |
672 | |
927 | |
… | |
… | |
682 | static SV * |
937 | static SV * |
683 | decode_str (dec_t *dec, int utf8) |
938 | decode_str (dec_t *dec, int utf8) |
684 | { |
939 | { |
685 | SV *sv = 0; |
940 | SV *sv = 0; |
686 | |
941 | |
687 | if ((*dec->cur & 31) == 31) |
942 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
688 | { |
943 | { |
|
|
944 | // indefinite length strings |
689 | ++dec->cur; |
945 | ++dec->cur; |
690 | |
946 | |
|
|
947 | U8 major = *dec->cur & MAJOR_MISC; |
|
|
948 | |
691 | sv = newSVpvn ("", 0); |
949 | sv = newSVpvn ("", 0); |
692 | |
950 | |
693 | // not very fast, and certainly not robust against illegal input |
|
|
694 | for (;;) |
951 | for (;;) |
695 | { |
952 | { |
696 | WANT (1); |
953 | WANT (1); |
697 | |
954 | |
698 | if (*dec->cur == (0xe0 | 31)) |
955 | if ((*dec->cur - major) > LENGTH_EXT8) |
|
|
956 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
699 | { |
957 | { |
700 | ++dec->cur; |
958 | ++dec->cur; |
701 | break; |
959 | break; |
702 | } |
960 | } |
|
|
961 | else |
|
|
962 | ERR ("corrupted CBOR data (invalid chunks in indefinite length string)"); |
703 | |
963 | |
704 | sv_catsv (sv, decode_sv (dec)); |
964 | STRLEN len = decode_uint (dec); |
|
|
965 | |
|
|
966 | WANT (len); |
|
|
967 | sv_catpvn (sv, dec->cur, len); |
|
|
968 | dec->cur += len; |
705 | } |
969 | } |
706 | } |
970 | } |
707 | else |
971 | else |
708 | { |
972 | { |
709 | STRLEN len = decode_uint (dec); |
973 | STRLEN len = decode_uint (dec); |
710 | |
974 | |
711 | WANT (len); |
975 | WANT (len); |
712 | sv = newSVpvn (dec->cur, len); |
976 | sv = newSVpvn (dec->cur, len); |
713 | dec->cur += len; |
977 | dec->cur += len; |
|
|
978 | |
|
|
979 | if (ecb_expect_false (dec->stringref) |
|
|
980 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
|
|
981 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
714 | } |
982 | } |
715 | |
983 | |
716 | if (utf8) |
984 | if (utf8) |
|
|
985 | { |
|
|
986 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
|
|
987 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
|
|
988 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
|
|
989 | |
717 | SvUTF8_on (sv); |
990 | SvUTF8_on (sv); |
|
|
991 | } |
718 | |
992 | |
719 | return sv; |
993 | return sv; |
720 | |
994 | |
721 | fail: |
995 | fail: |
722 | SvREFCNT_dec (sv); |
996 | SvREFCNT_dec (sv); |
… | |
… | |
732 | WANT (1); |
1006 | WANT (1); |
733 | |
1007 | |
734 | switch (tag) |
1008 | switch (tag) |
735 | { |
1009 | { |
736 | case CBOR_TAG_MAGIC: |
1010 | case CBOR_TAG_MAGIC: |
737 | return decode_sv (dec); |
1011 | sv = decode_sv (dec); |
|
|
1012 | break; |
738 | |
1013 | |
739 | case CBOR_TAG_INDIRECTION: |
1014 | case CBOR_TAG_INDIRECTION: |
740 | return newRV_noinc (decode_sv (dec)); |
1015 | sv = newRV_noinc (decode_sv (dec)); |
|
|
1016 | break; |
|
|
1017 | |
|
|
1018 | case CBOR_TAG_STRINGREF_NAMESPACE: |
|
|
1019 | { |
|
|
1020 | // do nmot use SAVETMPS/FREETMPS, as these will |
|
|
1021 | // erase mortalised caches, e.g. "shareable" |
|
|
1022 | ENTER; |
|
|
1023 | |
|
|
1024 | SAVESPTR (dec->stringref); |
|
|
1025 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
1026 | |
|
|
1027 | sv = decode_sv (dec); |
|
|
1028 | |
|
|
1029 | LEAVE; |
|
|
1030 | } |
|
|
1031 | break; |
|
|
1032 | |
|
|
1033 | case CBOR_TAG_STRINGREF: |
|
|
1034 | { |
|
|
1035 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
|
|
1036 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
|
|
1037 | |
|
|
1038 | UV idx = decode_uint (dec); |
|
|
1039 | |
|
|
1040 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
|
|
1041 | ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
|
|
1042 | |
|
|
1043 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
|
|
1044 | } |
|
|
1045 | break; |
741 | |
1046 | |
742 | case CBOR_TAG_VALUE_SHAREABLE: |
1047 | case CBOR_TAG_VALUE_SHAREABLE: |
743 | { |
1048 | { |
744 | if (ecb_expect_false (!dec->shareable)) |
1049 | if (ecb_expect_false (!dec->shareable)) |
745 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
1050 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
746 | |
1051 | |
|
|
1052 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
|
|
1053 | { |
747 | sv = newSV (0); |
1054 | sv = newSV (0); |
748 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
1055 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
749 | |
1056 | |
750 | SV *osv = decode_sv (dec); |
1057 | SV *osv = decode_sv (dec); |
751 | sv_setsv (sv, osv); |
1058 | sv_setsv (sv, osv); |
752 | SvREFCNT_dec_NN (osv); |
1059 | SvREFCNT_dec_NN (osv); |
753 | } |
1060 | } |
754 | |
1061 | else |
755 | return sv; |
1062 | { |
|
|
1063 | av_push (dec->shareable, &PL_sv_undef); |
|
|
1064 | int idx = AvFILLp (dec->shareable); |
|
|
1065 | sv = decode_sv (dec); |
|
|
1066 | av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
|
|
1067 | } |
|
|
1068 | } |
|
|
1069 | break; |
756 | |
1070 | |
757 | case CBOR_TAG_VALUE_SHAREDREF: |
1071 | case CBOR_TAG_VALUE_SHAREDREF: |
758 | { |
1072 | { |
759 | if ((*dec->cur >> 5) != 0) |
1073 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
760 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
1074 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
761 | |
1075 | |
762 | UV idx = decode_uint (dec); |
1076 | UV idx = decode_uint (dec); |
763 | |
1077 | |
764 | if (!dec->shareable || idx > AvFILLp (dec->shareable)) |
1078 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
765 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1079 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
766 | |
1080 | |
767 | return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
1081 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
1082 | |
|
|
1083 | if (sv == &PL_sv_undef) |
|
|
1084 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
768 | } |
1085 | } |
|
|
1086 | break; |
769 | |
1087 | |
770 | case CBOR_TAG_PERL_OBJECT: |
1088 | case CBOR_TAG_PERL_OBJECT: |
771 | { |
1089 | { |
|
|
1090 | if (dec->cbor.flags & F_FORBID_OBJECTS) |
|
|
1091 | goto filter; |
|
|
1092 | |
772 | sv = decode_sv (dec); |
1093 | sv = decode_sv (dec); |
773 | |
1094 | |
774 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
1095 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
775 | ERR ("corrupted CBOR data (non-array perl object)"); |
1096 | ERR ("corrupted CBOR data (non-array perl object)"); |
776 | |
1097 | |
… | |
… | |
786 | if (!method) |
1107 | if (!method) |
787 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
1108 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
788 | |
1109 | |
789 | dSP; |
1110 | dSP; |
790 | |
1111 | |
791 | ENTER; SAVETMPS; PUSHMARK (SP); |
1112 | ENTER; SAVETMPS; |
|
|
1113 | PUSHMARK (SP); |
792 | EXTEND (SP, len + 1); |
1114 | EXTEND (SP, len + 1); |
793 | // we re-bless the reference to get overload and other niceties right |
1115 | // we re-bless the reference to get overload and other niceties right |
794 | PUSHs (*av_fetch (av, 0, 1)); |
1116 | PUSHs (*av_fetch (av, 0, 1)); |
795 | PUSHs (sv_cbor); |
1117 | PUSHs (sv_cbor); |
796 | |
1118 | |
… | |
… | |
804 | SPAGAIN; |
1126 | SPAGAIN; |
805 | |
1127 | |
806 | if (SvTRUE (ERRSV)) |
1128 | if (SvTRUE (ERRSV)) |
807 | { |
1129 | { |
808 | FREETMPS; LEAVE; |
1130 | FREETMPS; LEAVE; |
809 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1131 | ERR_ERRSV; |
810 | } |
1132 | } |
811 | |
1133 | |
812 | SvREFCNT_dec (sv); |
1134 | SvREFCNT_dec (sv); |
813 | sv = SvREFCNT_inc (POPs); |
1135 | sv = SvREFCNT_inc (POPs); |
814 | |
1136 | |
815 | PUTBACK; |
1137 | PUTBACK; |
816 | |
1138 | |
817 | FREETMPS; LEAVE; |
1139 | FREETMPS; LEAVE; |
818 | |
|
|
819 | return sv; |
|
|
820 | } |
1140 | } |
|
|
1141 | break; |
821 | |
1142 | |
822 | default: |
1143 | default: |
|
|
1144 | filter: |
823 | { |
1145 | { |
|
|
1146 | SV *tag_sv = newSVuv (tag); |
|
|
1147 | |
824 | sv = decode_sv (dec); |
1148 | sv = decode_sv (dec); |
825 | |
1149 | |
|
|
1150 | dSP; |
|
|
1151 | ENTER; SAVETMPS; |
|
|
1152 | SAVESTACK_POS (); |
|
|
1153 | PUSHMARK (SP); |
|
|
1154 | EXTEND (SP, 2); |
|
|
1155 | PUSHs (tag_sv); |
|
|
1156 | PUSHs (sv); |
|
|
1157 | |
|
|
1158 | PUTBACK; |
|
|
1159 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
|
|
1160 | SPAGAIN; |
|
|
1161 | |
|
|
1162 | if (SvTRUE (ERRSV)) |
|
|
1163 | { |
|
|
1164 | SvREFCNT_dec (tag_sv); |
|
|
1165 | FREETMPS; LEAVE; |
|
|
1166 | ERR_ERRSV; |
|
|
1167 | } |
|
|
1168 | |
|
|
1169 | if (count) |
|
|
1170 | { |
|
|
1171 | SvREFCNT_dec (tag_sv); |
|
|
1172 | SvREFCNT_dec (sv); |
|
|
1173 | sv = SvREFCNT_inc (POPs); |
|
|
1174 | } |
|
|
1175 | else |
|
|
1176 | { |
826 | AV *av = newAV (); |
1177 | AV *av = newAV (); |
827 | av_push (av, newSVuv (tag)); |
1178 | av_push (av, tag_sv); |
828 | av_push (av, sv); |
1179 | av_push (av, sv); |
829 | |
1180 | |
830 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1181 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
831 | ? cbor_tagged_stash |
1182 | ? cbor_tagged_stash |
832 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
1183 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
833 | |
|
|
834 | return sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
1184 | sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
835 | } |
1185 | } |
|
|
1186 | |
|
|
1187 | PUTBACK; |
|
|
1188 | |
|
|
1189 | FREETMPS; LEAVE; |
|
|
1190 | } |
|
|
1191 | break; |
836 | } |
1192 | } |
|
|
1193 | |
|
|
1194 | return sv; |
837 | |
1195 | |
838 | fail: |
1196 | fail: |
839 | SvREFCNT_dec (sv); |
1197 | SvREFCNT_dec (sv); |
840 | return &PL_sv_undef; |
1198 | return &PL_sv_undef; |
841 | } |
1199 | } |
… | |
… | |
843 | static SV * |
1201 | static SV * |
844 | decode_sv (dec_t *dec) |
1202 | decode_sv (dec_t *dec) |
845 | { |
1203 | { |
846 | WANT (1); |
1204 | WANT (1); |
847 | |
1205 | |
848 | switch (*dec->cur >> 5) |
1206 | switch (*dec->cur >> MAJOR_SHIFT) |
849 | { |
1207 | { |
850 | case 0: // unsigned int |
1208 | case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec)); |
851 | return newSVuv (decode_uint (dec)); |
1209 | case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec)); |
852 | case 1: // negative int |
1210 | case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0); |
853 | return newSViv (-1 - (IV)decode_uint (dec)); |
1211 | case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1); |
854 | case 2: // octet string |
1212 | case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec); |
855 | return decode_str (dec, 0); |
1213 | case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec); |
856 | case 3: // utf-8 string |
1214 | case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec); |
857 | return decode_str (dec, 1); |
1215 | |
858 | case 4: // array |
1216 | case MAJOR_MISC >> MAJOR_SHIFT: |
859 | return decode_av (dec); |
|
|
860 | case 5: // map |
|
|
861 | return decode_hv (dec); |
|
|
862 | case 6: // tag |
|
|
863 | return decode_tagged (dec); |
|
|
864 | case 7: // misc |
|
|
865 | switch (*dec->cur++ & 31) |
1217 | switch (*dec->cur++ & MINOR_MASK) |
866 | { |
1218 | { |
867 | case 20: |
1219 | case SIMPLE_FALSE: |
868 | #if CBOR_SLOW |
1220 | #if CBOR_SLOW |
869 | types_false = get_bool ("Types::Serialiser::false"); |
1221 | types_false = get_bool ("Types::Serialiser::false"); |
870 | #endif |
1222 | #endif |
871 | return newSVsv (types_false); |
1223 | return newSVsv (types_false); |
872 | case 21: |
1224 | case SIMPLE_TRUE: |
873 | #if CBOR_SLOW |
1225 | #if CBOR_SLOW |
874 | types_true = get_bool ("Types::Serialiser::true"); |
1226 | types_true = get_bool ("Types::Serialiser::true"); |
875 | #endif |
1227 | #endif |
876 | return newSVsv (types_true); |
1228 | return newSVsv (types_true); |
877 | case 22: |
1229 | case SIMPLE_NULL: |
878 | return newSVsv (&PL_sv_undef); |
1230 | return newSVsv (&PL_sv_undef); |
879 | case 23: |
1231 | case SIMPLE_UNDEF: |
880 | #if CBOR_SLOW |
1232 | #if CBOR_SLOW |
881 | types_error = get_bool ("Types::Serialiser::error"); |
1233 | types_error = get_bool ("Types::Serialiser::error"); |
882 | #endif |
1234 | #endif |
883 | return newSVsv (types_error); |
1235 | return newSVsv (types_error); |
884 | |
1236 | |
885 | case 25: |
1237 | case MISC_FLOAT16: |
886 | { |
1238 | { |
887 | WANT (2); |
1239 | WANT (2); |
888 | |
1240 | |
889 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1241 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
890 | dec->cur += 2; |
1242 | dec->cur += 2; |
891 | |
1243 | |
892 | return newSVnv (ecb_binary16_to_float (fp)); |
1244 | return newSVnv (ecb_binary16_to_float (fp)); |
893 | } |
1245 | } |
894 | |
1246 | |
895 | case 26: |
1247 | case MISC_FLOAT32: |
896 | { |
1248 | { |
897 | uint32_t fp; |
1249 | uint32_t fp; |
898 | WANT (4); |
1250 | WANT (4); |
899 | memcpy (&fp, dec->cur, 4); |
1251 | memcpy (&fp, dec->cur, 4); |
900 | dec->cur += 4; |
1252 | dec->cur += 4; |
… | |
… | |
903 | fp = ecb_bswap32 (fp); |
1255 | fp = ecb_bswap32 (fp); |
904 | |
1256 | |
905 | return newSVnv (ecb_binary32_to_float (fp)); |
1257 | return newSVnv (ecb_binary32_to_float (fp)); |
906 | } |
1258 | } |
907 | |
1259 | |
908 | case 27: |
1260 | case MISC_FLOAT64: |
909 | { |
1261 | { |
910 | uint64_t fp; |
1262 | uint64_t fp; |
911 | WANT (8); |
1263 | WANT (8); |
912 | memcpy (&fp, dec->cur, 8); |
1264 | memcpy (&fp, dec->cur, 8); |
913 | dec->cur += 8; |
1265 | dec->cur += 8; |
… | |
… | |
916 | fp = ecb_bswap64 (fp); |
1268 | fp = ecb_bswap64 (fp); |
917 | |
1269 | |
918 | return newSVnv (ecb_binary64_to_double (fp)); |
1270 | return newSVnv (ecb_binary64_to_double (fp)); |
919 | } |
1271 | } |
920 | |
1272 | |
921 | // 0..19 unassigned |
1273 | // 0..19 unassigned simple |
922 | // 24 reserved + unassigned (reserved values are not encodable) |
1274 | // 24 reserved + unassigned simple (reserved values are not encodable) |
|
|
1275 | // 28-30 unassigned misc |
|
|
1276 | // 31 break code |
923 | default: |
1277 | default: |
924 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1278 | ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)"); |
925 | } |
1279 | } |
926 | |
1280 | |
927 | break; |
1281 | break; |
928 | } |
1282 | } |
929 | |
1283 | |
… | |
… | |
932 | } |
1286 | } |
933 | |
1287 | |
934 | static SV * |
1288 | static SV * |
935 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1289 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
936 | { |
1290 | { |
937 | dec_t dec = { }; |
1291 | dec_t dec = { 0 }; |
938 | SV *sv; |
1292 | SV *sv; |
939 | STRLEN len; |
1293 | STRLEN len; |
940 | char *data = SvPVbyte (string, len); |
1294 | char *data = SvPVbyte (string, len); |
941 | |
1295 | |
942 | if (len > cbor->max_size && cbor->max_size) |
1296 | if (len > cbor->max_size && cbor->max_size) |
… | |
… | |
956 | if (dec.cur != dec.end && !dec.err) |
1310 | if (dec.cur != dec.end && !dec.err) |
957 | dec.err = "garbage after CBOR object"; |
1311 | dec.err = "garbage after CBOR object"; |
958 | |
1312 | |
959 | if (dec.err) |
1313 | if (dec.err) |
960 | { |
1314 | { |
|
|
1315 | if (dec.shareable) |
|
|
1316 | { |
|
|
1317 | // need to break cyclic links, which would all be in shareable |
|
|
1318 | int i; |
|
|
1319 | SV **svp; |
|
|
1320 | |
|
|
1321 | for (i = av_len (dec.shareable) + 1; i--; ) |
|
|
1322 | if ((svp = av_fetch (dec.shareable, i, 0))) |
|
|
1323 | sv_setsv (*svp, &PL_sv_undef); |
|
|
1324 | } |
|
|
1325 | |
961 | SvREFCNT_dec (sv); |
1326 | SvREFCNT_dec (sv); |
|
|
1327 | |
|
|
1328 | if (dec.err_sv) |
|
|
1329 | sv_2mortal (dec.err_sv); |
|
|
1330 | |
962 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1331 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
963 | } |
1332 | } |
964 | |
1333 | |
965 | sv = sv_2mortal (sv); |
1334 | sv = sv_2mortal (sv); |
966 | |
1335 | |
967 | return sv; |
1336 | return sv; |
968 | } |
1337 | } |
969 | |
1338 | |
|
|
1339 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1340 | // incremental parser |
|
|
1341 | |
|
|
1342 | #define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0) |
|
|
1343 | |
|
|
1344 | // returns 0 for notyet, 1 for success or error |
|
|
1345 | static int |
|
|
1346 | incr_parse (CBOR *self, SV *cborstr) |
|
|
1347 | { |
|
|
1348 | STRLEN cur; |
|
|
1349 | SvPV (cborstr, cur); |
|
|
1350 | |
|
|
1351 | while (ecb_expect_true (self->incr_need <= cur)) |
|
|
1352 | { |
|
|
1353 | // table of integer count bytes |
|
|
1354 | static I8 incr_len[MINOR_MASK + 1] = { |
|
|
1355 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1356 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1357 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1358 | 1, 2, 4, 8,-1,-1,-1,-2 |
|
|
1359 | }; |
|
|
1360 | |
|
|
1361 | const U8 *p = SvPVX (cborstr) + self->incr_pos; |
|
|
1362 | U8 m = *p & MINOR_MASK; |
|
|
1363 | IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1364 | I8 ilen = incr_len[m]; |
|
|
1365 | |
|
|
1366 | self->incr_need = self->incr_pos + 1; |
|
|
1367 | |
|
|
1368 | if (ecb_expect_false (ilen < 0)) |
|
|
1369 | { |
|
|
1370 | if (m != MINOR_INDEF) |
|
|
1371 | return 1; // error |
|
|
1372 | |
|
|
1373 | if (*p == (MAJOR_MISC | MINOR_INDEF)) |
|
|
1374 | { |
|
|
1375 | if (count >= 0) |
|
|
1376 | return 1; // error |
|
|
1377 | |
|
|
1378 | count = 1; |
|
|
1379 | } |
|
|
1380 | else |
|
|
1381 | { |
|
|
1382 | av_push (self->incr_count, newSViv (-1)); //TODO: nest |
|
|
1383 | count = -1; |
|
|
1384 | } |
|
|
1385 | } |
|
|
1386 | else |
|
|
1387 | { |
|
|
1388 | self->incr_need += ilen; |
|
|
1389 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1390 | return 0; |
|
|
1391 | |
|
|
1392 | int major = *p >> MAJOR_SHIFT; |
|
|
1393 | |
|
|
1394 | switch (major) |
|
|
1395 | { |
|
|
1396 | case MAJOR_TAG >> MAJOR_SHIFT: |
|
|
1397 | ++count; // tags merely prefix another value |
|
|
1398 | break; |
|
|
1399 | |
|
|
1400 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1401 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1402 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1403 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1404 | { |
|
|
1405 | UV len; |
|
|
1406 | |
|
|
1407 | if (ecb_expect_false (ilen)) |
|
|
1408 | { |
|
|
1409 | len = 0; |
|
|
1410 | |
|
|
1411 | do { |
|
|
1412 | len = (len << 8) | *++p; |
|
|
1413 | } while (--ilen); |
|
|
1414 | } |
|
|
1415 | else |
|
|
1416 | len = m; |
|
|
1417 | |
|
|
1418 | switch (major) |
|
|
1419 | { |
|
|
1420 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1421 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1422 | self->incr_need += len; |
|
|
1423 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1424 | return 0; |
|
|
1425 | |
|
|
1426 | break; |
|
|
1427 | |
|
|
1428 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1429 | len <<= 1; |
|
|
1430 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1431 | if (len) |
|
|
1432 | { |
|
|
1433 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
|
|
1434 | count = len + 1; |
|
|
1435 | } |
|
|
1436 | break; |
|
|
1437 | } |
|
|
1438 | } |
|
|
1439 | } |
|
|
1440 | } |
|
|
1441 | |
|
|
1442 | self->incr_pos = self->incr_need; |
|
|
1443 | |
|
|
1444 | if (count > 0) |
|
|
1445 | { |
|
|
1446 | while (!--count) |
|
|
1447 | { |
|
|
1448 | if (!AvFILLp (self->incr_count)) |
|
|
1449 | return 1; // done |
|
|
1450 | |
|
|
1451 | SvREFCNT_dec_NN (av_pop (self->incr_count)); |
|
|
1452 | count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1453 | } |
|
|
1454 | |
|
|
1455 | SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count; |
|
|
1456 | } |
|
|
1457 | } |
|
|
1458 | |
|
|
1459 | return 0; |
|
|
1460 | } |
|
|
1461 | |
|
|
1462 | |
970 | ///////////////////////////////////////////////////////////////////////////// |
1463 | ///////////////////////////////////////////////////////////////////////////// |
971 | // XS interface functions |
1464 | // XS interface functions |
972 | |
1465 | |
973 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1466 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
974 | |
1467 | |
… | |
… | |
982 | |
1475 | |
983 | types_true = get_bool ("Types::Serialiser::true" ); |
1476 | types_true = get_bool ("Types::Serialiser::true" ); |
984 | types_false = get_bool ("Types::Serialiser::false"); |
1477 | types_false = get_bool ("Types::Serialiser::false"); |
985 | types_error = get_bool ("Types::Serialiser::error"); |
1478 | types_error = get_bool ("Types::Serialiser::error"); |
986 | |
1479 | |
|
|
1480 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
|
|
1481 | |
987 | sv_cbor = newSVpv ("CBOR", 0); |
1482 | sv_cbor = newSVpv ("CBOR", 0); |
988 | SvREADONLY_on (sv_cbor); |
1483 | SvREADONLY_on (sv_cbor); |
|
|
1484 | |
|
|
1485 | assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1)); |
989 | } |
1486 | } |
990 | |
1487 | |
991 | PROTOTYPES: DISABLE |
1488 | PROTOTYPES: DISABLE |
992 | |
1489 | |
993 | void CLONE (...) |
1490 | void CLONE (...) |
… | |
… | |
1012 | void shrink (CBOR *self, int enable = 1) |
1509 | void shrink (CBOR *self, int enable = 1) |
1013 | ALIAS: |
1510 | ALIAS: |
1014 | shrink = F_SHRINK |
1511 | shrink = F_SHRINK |
1015 | allow_unknown = F_ALLOW_UNKNOWN |
1512 | allow_unknown = F_ALLOW_UNKNOWN |
1016 | allow_sharing = F_ALLOW_SHARING |
1513 | allow_sharing = F_ALLOW_SHARING |
1017 | dedup_keys = F_DEDUP_KEYS |
1514 | allow_cycles = F_ALLOW_CYCLES |
|
|
1515 | forbid_objects = F_FORBID_OBJECTS |
1018 | dedup_strings = F_DEDUP_STRINGS |
1516 | pack_strings = F_PACK_STRINGS |
|
|
1517 | text_keys = F_TEXT_KEYS |
|
|
1518 | text_strings = F_TEXT_STRINGS |
|
|
1519 | validate_utf8 = F_VALIDATE_UTF8 |
1019 | PPCODE: |
1520 | PPCODE: |
1020 | { |
1521 | { |
1021 | if (enable) |
1522 | if (enable) |
1022 | self->flags |= ix; |
1523 | self->flags |= ix; |
1023 | else |
1524 | else |
… | |
… | |
1029 | void get_shrink (CBOR *self) |
1530 | void get_shrink (CBOR *self) |
1030 | ALIAS: |
1531 | ALIAS: |
1031 | get_shrink = F_SHRINK |
1532 | get_shrink = F_SHRINK |
1032 | get_allow_unknown = F_ALLOW_UNKNOWN |
1533 | get_allow_unknown = F_ALLOW_UNKNOWN |
1033 | get_allow_sharing = F_ALLOW_SHARING |
1534 | get_allow_sharing = F_ALLOW_SHARING |
1034 | get_dedup_keys = F_DEDUP_KEYS |
1535 | get_allow_cycles = F_ALLOW_CYCLES |
|
|
1536 | get_forbid_objects = F_FORBID_OBJECTS |
1035 | get_dedup_strings = F_DEDUP_STRINGS |
1537 | get_pack_strings = F_PACK_STRINGS |
|
|
1538 | get_text_keys = F_TEXT_KEYS |
|
|
1539 | get_text_strings = F_TEXT_STRINGS |
|
|
1540 | get_validate_utf8 = F_VALIDATE_UTF8 |
1036 | PPCODE: |
1541 | PPCODE: |
1037 | XPUSHs (boolSV (self->flags & ix)); |
1542 | XPUSHs (boolSV (self->flags & ix)); |
1038 | |
1543 | |
1039 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1544 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1040 | PPCODE: |
1545 | PPCODE: |
… | |
… | |
1053 | XPUSHs (ST (0)); |
1558 | XPUSHs (ST (0)); |
1054 | |
1559 | |
1055 | int get_max_size (CBOR *self) |
1560 | int get_max_size (CBOR *self) |
1056 | CODE: |
1561 | CODE: |
1057 | RETVAL = self->max_size; |
1562 | RETVAL = self->max_size; |
|
|
1563 | OUTPUT: |
|
|
1564 | RETVAL |
|
|
1565 | |
|
|
1566 | void filter (CBOR *self, SV *filter = 0) |
|
|
1567 | PPCODE: |
|
|
1568 | SvREFCNT_dec (self->filter); |
|
|
1569 | self->filter = filter ? newSVsv (filter) : filter; |
|
|
1570 | XPUSHs (ST (0)); |
|
|
1571 | |
|
|
1572 | SV *get_filter (CBOR *self) |
|
|
1573 | CODE: |
|
|
1574 | RETVAL = self->filter ? self->filter : NEWSV (0, 0); |
1058 | OUTPUT: |
1575 | OUTPUT: |
1059 | RETVAL |
1576 | RETVAL |
1060 | |
1577 | |
1061 | void encode (CBOR *self, SV *scalar) |
1578 | void encode (CBOR *self, SV *scalar) |
1062 | PPCODE: |
1579 | PPCODE: |
… | |
… | |
1077 | EXTEND (SP, 2); |
1594 | EXTEND (SP, 2); |
1078 | PUSHs (sv); |
1595 | PUSHs (sv); |
1079 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1596 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1080 | } |
1597 | } |
1081 | |
1598 | |
|
|
1599 | void incr_parse (CBOR *self, SV *cborstr) |
|
|
1600 | ALIAS: |
|
|
1601 | incr_parse_multiple = 1 |
|
|
1602 | PPCODE: |
|
|
1603 | { |
|
|
1604 | if (SvUTF8 (cborstr)) |
|
|
1605 | sv_utf8_downgrade (cborstr, 0); |
|
|
1606 | |
|
|
1607 | if (!self->incr_count) |
|
|
1608 | { |
|
|
1609 | self->incr_count = newAV (); |
|
|
1610 | self->incr_pos = 0; |
|
|
1611 | self->incr_need = 1; |
|
|
1612 | |
|
|
1613 | av_push (self->incr_count, newSViv (1)); |
|
|
1614 | } |
|
|
1615 | |
|
|
1616 | do |
|
|
1617 | { |
|
|
1618 | if (!incr_parse (self, cborstr)) |
|
|
1619 | { |
|
|
1620 | if (self->incr_need > self->max_size && self->max_size) |
|
|
1621 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
|
|
1622 | (unsigned long)self->incr_need, (unsigned long)self->max_size); |
|
|
1623 | |
|
|
1624 | break; |
|
|
1625 | } |
|
|
1626 | |
|
|
1627 | SV *sv; |
|
|
1628 | char *offset; |
|
|
1629 | |
|
|
1630 | PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
|
|
1631 | XPUSHs (sv); |
|
|
1632 | |
|
|
1633 | sv_chop (cborstr, offset); |
|
|
1634 | |
|
|
1635 | av_clear (self->incr_count); |
|
|
1636 | av_push (self->incr_count, newSViv (1)); |
|
|
1637 | |
|
|
1638 | self->incr_pos = 0; |
|
|
1639 | self->incr_need = self->incr_pos + 1; |
|
|
1640 | } |
|
|
1641 | while (ix); |
|
|
1642 | } |
|
|
1643 | |
|
|
1644 | void incr_reset (CBOR *self) |
|
|
1645 | CODE: |
|
|
1646 | { |
|
|
1647 | SvREFCNT_dec (self->incr_count); |
|
|
1648 | self->incr_count = 0; |
|
|
1649 | } |
|
|
1650 | |
|
|
1651 | void DESTROY (CBOR *self) |
|
|
1652 | PPCODE: |
|
|
1653 | cbor_free (self); |
|
|
1654 | |
1082 | PROTOTYPES: ENABLE |
1655 | PROTOTYPES: ENABLE |
1083 | |
1656 | |
1084 | void encode_cbor (SV *scalar) |
1657 | void encode_cbor (SV *scalar) |
|
|
1658 | ALIAS: |
|
|
1659 | encode_cbor = 0 |
|
|
1660 | encode_cbor_sharing = F_ALLOW_SHARING |
1085 | PPCODE: |
1661 | PPCODE: |
1086 | { |
1662 | { |
1087 | CBOR cbor; |
1663 | CBOR cbor; |
1088 | cbor_init (&cbor); |
1664 | cbor_init (&cbor); |
|
|
1665 | cbor.flags |= ix; |
1089 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1666 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1090 | XPUSHs (scalar); |
1667 | XPUSHs (scalar); |
1091 | } |
1668 | } |
1092 | |
1669 | |
1093 | void decode_cbor (SV *cborstr) |
1670 | void decode_cbor (SV *cborstr) |