… | |
… | |
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 |
29 | CBOR_TAG_PERL_OBJECT = 256, |
70 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
30 | CBOR_TAG_GENERIC_OBJECT = 257, |
71 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
|
|
72 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
|
|
73 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
|
|
74 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
|
|
75 | CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
|
|
76 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
31 | |
77 | |
32 | // rfc7049 |
78 | // rfc7049 |
33 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
79 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
34 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
80 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
35 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
81 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
36 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
82 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
37 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
83 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
38 | CBOR_TAG_BIGFLOAT = 5, // array |
84 | CBOR_TAG_BIGFLOAT = 5, // array |
39 | |
85 | |
40 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
86 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
41 | CBOR_TAG_CONV_B64 = 22, // base64, any |
87 | CBOR_TAG_CONV_B64 = 22, // base64, any |
42 | CBOR_TAG_CONV_HEX = 23, // base16, any |
88 | CBOR_TAG_CONV_HEX = 23, // base16, any |
43 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
89 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
44 | |
90 | |
45 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
91 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
46 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
92 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
47 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
93 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
48 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
94 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
49 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
95 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
50 | |
96 | |
51 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
97 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
52 | }; |
98 | }; |
53 | |
99 | |
54 | #define F_SHRINK 0x00000200UL |
100 | #define F_SHRINK 0x00000001UL |
55 | #define F_ALLOW_UNKNOWN 0x00002000UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
|
|
102 | #define F_ALLOW_SHARING 0x00000004UL |
|
|
103 | #define F_ALLOW_CYCLES 0x00000008UL |
|
|
104 | #define F_PACK_STRINGS 0x00000010UL |
|
|
105 | #define F_TEXT_KEYS 0x00000020UL |
|
|
106 | #define F_TEXT_STRINGS 0x00000040UL |
|
|
107 | #define F_VALIDATE_UTF8 0x00000080UL |
56 | |
108 | |
57 | #define INIT_SIZE 32 // initial scalar size to be allocated |
109 | #define INIT_SIZE 32 // initial scalar size to be allocated |
58 | |
110 | |
59 | #define SB do { |
111 | #define SB do { |
60 | #define SE } while (0) |
112 | #define SE } while (0) |
… | |
… | |
72 | # define CBOR_SLOW 0 |
124 | # define CBOR_SLOW 0 |
73 | # define CBOR_STASH cbor_stash |
125 | # define CBOR_STASH cbor_stash |
74 | #endif |
126 | #endif |
75 | |
127 | |
76 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
128 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
77 | static SV *types_true, *types_false, *types_error, *sv_cbor; |
129 | static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter; |
78 | |
130 | |
79 | typedef struct { |
131 | typedef struct { |
80 | U32 flags; |
132 | U32 flags; |
81 | U32 max_depth; |
133 | U32 max_depth; |
82 | STRLEN max_size; |
134 | STRLEN max_size; |
|
|
135 | SV *filter; |
|
|
136 | |
|
|
137 | // for the incremental parser |
|
|
138 | STRLEN incr_pos; // the current offset into the text |
|
|
139 | STRLEN incr_need; // minimum bytes needed to decode |
|
|
140 | AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef. |
83 | } CBOR; |
141 | } CBOR; |
84 | |
142 | |
85 | ecb_inline void |
143 | ecb_inline void |
86 | cbor_init (CBOR *cbor) |
144 | cbor_init (CBOR *cbor) |
87 | { |
145 | { |
88 | Zero (cbor, 1, CBOR); |
146 | Zero (cbor, 1, CBOR); |
89 | cbor->max_depth = 512; |
147 | cbor->max_depth = 512; |
|
|
148 | } |
|
|
149 | |
|
|
150 | ecb_inline void |
|
|
151 | cbor_free (CBOR *cbor) |
|
|
152 | { |
|
|
153 | SvREFCNT_dec (cbor->filter); |
|
|
154 | SvREFCNT_dec (cbor->incr_count); |
90 | } |
155 | } |
91 | |
156 | |
92 | ///////////////////////////////////////////////////////////////////////////// |
157 | ///////////////////////////////////////////////////////////////////////////// |
93 | // utility functions |
158 | // utility functions |
94 | |
159 | |
… | |
… | |
116 | SvPV_renew (sv, SvCUR (sv) + 1); |
181 | SvPV_renew (sv, SvCUR (sv) + 1); |
117 | #endif |
182 | #endif |
118 | } |
183 | } |
119 | } |
184 | } |
120 | |
185 | |
121 | ///////////////////////////////////////////////////////////////////////////// |
186 | // minimum length of a string to be registered for stringref |
122 | // fp hell |
187 | ecb_inline int |
123 | |
188 | minimum_string_length (UV idx) |
124 | //TODO |
189 | { |
|
|
190 | return idx > 23 |
|
|
191 | ? idx > 0xffU |
|
|
192 | ? idx > 0xffffU |
|
|
193 | ? idx > 0xffffffffU |
|
|
194 | ? 11 |
|
|
195 | : 7 |
|
|
196 | : 5 |
|
|
197 | : 4 |
|
|
198 | : 3; |
|
|
199 | } |
125 | |
200 | |
126 | ///////////////////////////////////////////////////////////////////////////// |
201 | ///////////////////////////////////////////////////////////////////////////// |
127 | // encoder |
202 | // encoder |
128 | |
203 | |
129 | // structure used for encoding CBOR |
204 | // structure used for encoding CBOR |
… | |
… | |
132 | char *cur; // SvPVX (sv) + current output position |
207 | char *cur; // SvPVX (sv) + current output position |
133 | char *end; // SvEND (sv) |
208 | char *end; // SvEND (sv) |
134 | SV *sv; // result scalar |
209 | SV *sv; // result scalar |
135 | CBOR cbor; |
210 | CBOR cbor; |
136 | U32 depth; // recursion level |
211 | U32 depth; // recursion level |
|
|
212 | HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8) |
|
|
213 | UV stringref_idx; |
|
|
214 | HV *shareable; // ptr => index, or 0 |
|
|
215 | UV shareable_idx; |
137 | } enc_t; |
216 | } enc_t; |
138 | |
217 | |
139 | ecb_inline void |
218 | ecb_inline void |
140 | need (enc_t *enc, STRLEN len) |
219 | need (enc_t *enc, STRLEN len) |
141 | { |
220 | { |
… | |
… | |
158 | static void |
237 | static void |
159 | encode_uint (enc_t *enc, int major, UV len) |
238 | encode_uint (enc_t *enc, int major, UV len) |
160 | { |
239 | { |
161 | need (enc, 9); |
240 | need (enc, 9); |
162 | |
241 | |
163 | if (len < 24) |
242 | if (ecb_expect_true (len < LENGTH_EXT1)) |
164 | *enc->cur++ = major | len; |
243 | *enc->cur++ = major | len; |
165 | else if (len <= 0xff) |
244 | else if (ecb_expect_true (len <= 0xffU)) |
166 | { |
245 | { |
167 | *enc->cur++ = major | 24; |
246 | *enc->cur++ = major | LENGTH_EXT1; |
168 | *enc->cur++ = len; |
247 | *enc->cur++ = len; |
169 | } |
248 | } |
170 | else if (len <= 0xffff) |
249 | else if (len <= 0xffffU) |
171 | { |
250 | { |
172 | *enc->cur++ = major | 25; |
251 | *enc->cur++ = major | LENGTH_EXT2; |
173 | *enc->cur++ = len >> 8; |
252 | *enc->cur++ = len >> 8; |
174 | *enc->cur++ = len; |
253 | *enc->cur++ = len; |
175 | } |
254 | } |
176 | else if (len <= 0xffffffff) |
255 | else if (len <= 0xffffffffU) |
177 | { |
256 | { |
178 | *enc->cur++ = major | 26; |
257 | *enc->cur++ = major | LENGTH_EXT4; |
179 | *enc->cur++ = len >> 24; |
258 | *enc->cur++ = len >> 24; |
180 | *enc->cur++ = len >> 16; |
259 | *enc->cur++ = len >> 16; |
181 | *enc->cur++ = len >> 8; |
260 | *enc->cur++ = len >> 8; |
182 | *enc->cur++ = len; |
261 | *enc->cur++ = len; |
183 | } |
262 | } |
184 | else |
263 | else |
185 | { |
264 | { |
186 | *enc->cur++ = major | 27; |
265 | *enc->cur++ = major | LENGTH_EXT8; |
187 | *enc->cur++ = len >> 56; |
266 | *enc->cur++ = len >> 56; |
188 | *enc->cur++ = len >> 48; |
267 | *enc->cur++ = len >> 48; |
189 | *enc->cur++ = len >> 40; |
268 | *enc->cur++ = len >> 40; |
190 | *enc->cur++ = len >> 32; |
269 | *enc->cur++ = len >> 32; |
191 | *enc->cur++ = len >> 24; |
270 | *enc->cur++ = len >> 24; |
… | |
… | |
193 | *enc->cur++ = len >> 8; |
272 | *enc->cur++ = len >> 8; |
194 | *enc->cur++ = len; |
273 | *enc->cur++ = len; |
195 | } |
274 | } |
196 | } |
275 | } |
197 | |
276 | |
198 | static void |
277 | ecb_inline void |
|
|
278 | encode_tag (enc_t *enc, UV tag) |
|
|
279 | { |
|
|
280 | encode_uint (enc, MAJOR_TAG, tag); |
|
|
281 | } |
|
|
282 | |
|
|
283 | // exceptional (hopefully) slow path for byte strings that need to be utf8-encoded |
|
|
284 | ecb_noinline static void |
199 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
285 | encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) |
200 | { |
286 | { |
201 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
287 | STRLEN ulen = len; |
|
|
288 | U8 *p, *pend = (U8 *)str + len; |
|
|
289 | |
|
|
290 | for (p = (U8 *)str; p < pend; ++p) |
|
|
291 | ulen += *p >> 7; // count set high bits |
|
|
292 | |
|
|
293 | encode_uint (enc, MAJOR_TEXT, ulen); |
|
|
294 | |
|
|
295 | need (enc, ulen); |
|
|
296 | for (p = (U8 *)str; p < pend; ++p) |
|
|
297 | if (*p < 0x80) |
|
|
298 | *enc->cur++ = *p; |
|
|
299 | else |
|
|
300 | { |
|
|
301 | *enc->cur++ = 0xc0 + (*p >> 6); |
|
|
302 | *enc->cur++ = 0x80 + (*p & 63); |
|
|
303 | } |
|
|
304 | } |
|
|
305 | |
|
|
306 | ecb_inline void |
|
|
307 | encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
|
|
308 | { |
|
|
309 | if (ecb_expect_false (upgrade_utf8)) |
|
|
310 | if (!utf8) |
|
|
311 | { |
|
|
312 | encode_str_utf8 (enc, utf8, str, len); |
|
|
313 | return; |
|
|
314 | } |
|
|
315 | |
|
|
316 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
202 | need (enc, len); |
317 | need (enc, len); |
203 | memcpy (enc->cur, str, len); |
318 | memcpy (enc->cur, str, len); |
204 | enc->cur += len; |
319 | enc->cur += len; |
205 | } |
320 | } |
206 | |
321 | |
|
|
322 | ecb_inline void |
|
|
323 | encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
|
|
324 | { |
|
|
325 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
|
|
326 | { |
|
|
327 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
|
|
328 | |
|
|
329 | if (SvOK (*svp)) |
|
|
330 | { |
|
|
331 | // already registered, use stringref |
|
|
332 | encode_tag (enc, CBOR_TAG_STRINGREF); |
|
|
333 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
|
|
334 | return; |
|
|
335 | } |
|
|
336 | else if (len >= minimum_string_length (enc->stringref_idx)) |
|
|
337 | { |
|
|
338 | // register only |
|
|
339 | sv_setuv (*svp, enc->stringref_idx); |
|
|
340 | ++enc->stringref_idx; |
|
|
341 | } |
|
|
342 | } |
|
|
343 | |
|
|
344 | encode_str (enc, upgrade_utf8, utf8, str, len); |
|
|
345 | } |
|
|
346 | |
207 | static void encode_sv (enc_t *enc, SV *sv); |
347 | static void encode_sv (enc_t *enc, SV *sv); |
208 | |
348 | |
209 | static void |
349 | static void |
210 | encode_av (enc_t *enc, AV *av) |
350 | encode_av (enc_t *enc, AV *av) |
211 | { |
351 | { |
… | |
… | |
214 | if (enc->depth >= enc->cbor.max_depth) |
354 | if (enc->depth >= enc->cbor.max_depth) |
215 | croak (ERR_NESTING_EXCEEDED); |
355 | croak (ERR_NESTING_EXCEEDED); |
216 | |
356 | |
217 | ++enc->depth; |
357 | ++enc->depth; |
218 | |
358 | |
219 | encode_uint (enc, 0x80, len + 1); |
359 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
220 | |
360 | |
|
|
361 | if (SvMAGICAL (av)) |
221 | for (i = 0; i <= len; ++i) |
362 | for (i = 0; i <= len; ++i) |
222 | { |
363 | { |
223 | SV **svp = av_fetch (av, i, 0); |
364 | SV **svp = av_fetch (av, i, 0); |
224 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
365 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
225 | } |
366 | } |
|
|
367 | else |
|
|
368 | for (i = 0; i <= len; ++i) |
|
|
369 | { |
|
|
370 | SV *sv = AvARRAY (av)[i]; |
|
|
371 | encode_sv (enc, sv ? sv : &PL_sv_undef); |
|
|
372 | } |
226 | |
373 | |
227 | --enc->depth; |
374 | --enc->depth; |
228 | } |
375 | } |
229 | |
376 | |
230 | static void |
377 | static void |
… | |
… | |
239 | |
386 | |
240 | int pairs = hv_iterinit (hv); |
387 | int pairs = hv_iterinit (hv); |
241 | int mg = SvMAGICAL (hv); |
388 | int mg = SvMAGICAL (hv); |
242 | |
389 | |
243 | if (mg) |
390 | if (mg) |
244 | encode_ch (enc, 0xa0 | 31); |
391 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
245 | else |
392 | else |
246 | encode_uint (enc, 0xa0, pairs); |
393 | encode_uint (enc, MAJOR_MAP, pairs); |
247 | |
394 | |
248 | while ((he = hv_iternext (hv))) |
395 | while ((he = hv_iternext (hv))) |
249 | { |
396 | { |
250 | if (HeKLEN (he) == HEf_SVKEY) |
397 | if (HeKLEN (he) == HEf_SVKEY) |
251 | encode_sv (enc, HeSVKEY (he)); |
398 | encode_sv (enc, HeSVKEY (he)); |
252 | else |
399 | else |
253 | encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
400 | encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
254 | |
401 | |
255 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
402 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
256 | } |
403 | } |
257 | |
404 | |
258 | if (mg) |
405 | if (mg) |
259 | encode_ch (enc, 0xe0 | 31); |
406 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
260 | |
407 | |
261 | --enc->depth; |
408 | --enc->depth; |
262 | } |
409 | } |
263 | |
410 | |
264 | // encode objects, arrays and special \0=false and \1=true values. |
411 | // encode objects, arrays and special \0=false and \1=true values. |
265 | static void |
412 | static void |
266 | encode_rv (enc_t *enc, SV *sv) |
413 | encode_rv (enc_t *enc, SV *sv) |
267 | { |
414 | { |
268 | svtype svt; |
|
|
269 | |
|
|
270 | SvGETMAGIC (sv); |
415 | SvGETMAGIC (sv); |
|
|
416 | |
271 | svt = SvTYPE (sv); |
417 | svtype svt = SvTYPE (sv); |
272 | |
418 | |
273 | if (ecb_expect_false (SvOBJECT (sv))) |
419 | if (ecb_expect_false (SvOBJECT (sv))) |
274 | { |
420 | { |
275 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
421 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
276 | ? types_boolean_stash |
422 | ? types_boolean_stash |
… | |
… | |
281 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
427 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
282 | ? cbor_tagged_stash |
428 | ? cbor_tagged_stash |
283 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
429 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
284 | |
430 | |
285 | HV *stash = SvSTASH (sv); |
431 | HV *stash = SvSTASH (sv); |
286 | GV *method; |
|
|
287 | |
432 | |
288 | if (stash == boolean_stash) |
433 | if (stash == boolean_stash) |
289 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
434 | { |
|
|
435 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
|
|
436 | return; |
|
|
437 | } |
290 | else if (stash == error_stash) |
438 | else if (stash == error_stash) |
291 | encode_ch (enc, 0xe0 | 23); |
439 | { |
|
|
440 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
|
|
441 | return; |
|
|
442 | } |
292 | else if (stash == tagged_stash) |
443 | else if (stash == tagged_stash) |
293 | { |
444 | { |
294 | if (svt != SVt_PVAV) |
445 | if (svt != SVt_PVAV) |
295 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
446 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
296 | |
447 | |
297 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
448 | encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
298 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
449 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
|
|
450 | |
|
|
451 | return; |
|
|
452 | } |
|
|
453 | } |
|
|
454 | |
|
|
455 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
|
|
456 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
|
|
457 | { |
|
|
458 | if (!enc->shareable) |
|
|
459 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
460 | |
|
|
461 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
462 | |
|
|
463 | if (SvOK (*svp)) |
299 | } |
464 | { |
|
|
465 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
466 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
|
|
467 | return; |
|
|
468 | } |
|
|
469 | else |
|
|
470 | { |
|
|
471 | sv_setuv (*svp, enc->shareable_idx); |
|
|
472 | ++enc->shareable_idx; |
|
|
473 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
474 | } |
|
|
475 | } |
|
|
476 | |
|
|
477 | if (ecb_expect_false (SvOBJECT (sv))) |
|
|
478 | { |
|
|
479 | HV *stash = SvSTASH (sv); |
|
|
480 | GV *method; |
|
|
481 | |
300 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
482 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
301 | { |
483 | { |
302 | dSP; |
484 | dSP; |
303 | |
485 | |
304 | ENTER; SAVETMPS; PUSHMARK (SP); |
486 | ENTER; SAVETMPS; |
|
|
487 | PUSHMARK (SP); |
305 | // we re-bless the reference to get overload and other niceties right |
488 | // we re-bless the reference to get overload and other niceties right |
306 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
489 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
307 | |
490 | |
308 | PUTBACK; |
491 | PUTBACK; |
309 | // G_SCALAR ensures that return value is 1 |
492 | // G_SCALAR ensures that return value is 1 |
… | |
… | |
322 | } |
505 | } |
323 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
506 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
324 | { |
507 | { |
325 | dSP; |
508 | dSP; |
326 | |
509 | |
327 | ENTER; SAVETMPS; PUSHMARK (SP); |
510 | ENTER; SAVETMPS; |
|
|
511 | SAVESTACK_POS (); |
|
|
512 | PUSHMARK (SP); |
328 | EXTEND (SP, 2); |
513 | EXTEND (SP, 2); |
329 | // we re-bless the reference to get overload and other niceties right |
514 | // we re-bless the reference to get overload and other niceties right |
330 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
515 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
331 | PUSHs (sv_cbor); |
516 | PUSHs (sv_cbor); |
332 | |
517 | |
… | |
… | |
336 | |
521 | |
337 | // catch this surprisingly common error |
522 | // catch this surprisingly common error |
338 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
523 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
339 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
524 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
340 | |
525 | |
341 | encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); |
526 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
342 | encode_uint (enc, 0x80, count + 1); |
527 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
343 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
528 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
344 | |
529 | |
345 | while (count) |
530 | while (count) |
346 | encode_sv (enc, SP[1 - count--]); |
531 | encode_sv (enc, SP[1 - count--]); |
347 | |
532 | |
348 | PUTBACK; |
533 | PUTBACK; |
… | |
… | |
355 | } |
540 | } |
356 | else if (svt == SVt_PVHV) |
541 | else if (svt == SVt_PVHV) |
357 | encode_hv (enc, (HV *)sv); |
542 | encode_hv (enc, (HV *)sv); |
358 | else if (svt == SVt_PVAV) |
543 | else if (svt == SVt_PVAV) |
359 | encode_av (enc, (AV *)sv); |
544 | encode_av (enc, (AV *)sv); |
360 | else if (svt < SVt_PVAV) |
|
|
361 | { |
|
|
362 | STRLEN len = 0; |
|
|
363 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
364 | |
|
|
365 | if (len == 1 && *pv == '1') |
|
|
366 | encode_ch (enc, 0xe0 | 21); |
|
|
367 | else if (len == 1 && *pv == '0') |
|
|
368 | encode_ch (enc, 0xe0 | 20); |
|
|
369 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
370 | encode_ch (enc, 0xe0 | 23); |
|
|
371 | else |
|
|
372 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
373 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
374 | } |
|
|
375 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
376 | encode_ch (enc, 0xe0 | 23); |
|
|
377 | else |
545 | else |
378 | croak ("encountered %s, but CBOR can only represent references to arrays or hashes", |
546 | { |
379 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
547 | encode_tag (enc, CBOR_TAG_INDIRECTION); |
|
|
548 | encode_sv (enc, sv); |
|
|
549 | } |
380 | } |
550 | } |
381 | |
551 | |
382 | static void |
552 | static void |
383 | encode_nv (enc_t *enc, SV *sv) |
553 | encode_nv (enc_t *enc, SV *sv) |
384 | { |
554 | { |
385 | double nv = SvNVX (sv); |
555 | double nv = SvNVX (sv); |
386 | |
556 | |
387 | need (enc, 9); |
557 | need (enc, 9); |
388 | |
558 | |
389 | if (ecb_expect_false (nv == (U32)nv)) |
559 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
390 | encode_uint (enc, 0x00, (U32)nv); |
560 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
391 | //TODO: maybe I32? |
561 | //TODO: maybe I32? |
392 | else if (ecb_expect_false (nv == (float)nv)) |
562 | else if (ecb_expect_false (nv == (float)nv)) |
393 | { |
563 | { |
394 | uint32_t fp = ecb_float_to_binary32 (nv); |
564 | uint32_t fp = ecb_float_to_binary32 (nv); |
395 | |
565 | |
396 | *enc->cur++ = 0xe0 | 26; |
566 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
397 | |
567 | |
398 | if (!ecb_big_endian ()) |
568 | if (!ecb_big_endian ()) |
399 | fp = ecb_bswap32 (fp); |
569 | fp = ecb_bswap32 (fp); |
400 | |
570 | |
401 | memcpy (enc->cur, &fp, 4); |
571 | memcpy (enc->cur, &fp, 4); |
… | |
… | |
403 | } |
573 | } |
404 | else |
574 | else |
405 | { |
575 | { |
406 | uint64_t fp = ecb_double_to_binary64 (nv); |
576 | uint64_t fp = ecb_double_to_binary64 (nv); |
407 | |
577 | |
408 | *enc->cur++ = 0xe0 | 27; |
578 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
409 | |
579 | |
410 | if (!ecb_big_endian ()) |
580 | if (!ecb_big_endian ()) |
411 | fp = ecb_bswap64 (fp); |
581 | fp = ecb_bswap64 (fp); |
412 | |
582 | |
413 | memcpy (enc->cur, &fp, 8); |
583 | memcpy (enc->cur, &fp, 8); |
… | |
… | |
422 | |
592 | |
423 | if (SvPOKp (sv)) |
593 | if (SvPOKp (sv)) |
424 | { |
594 | { |
425 | STRLEN len; |
595 | STRLEN len; |
426 | char *str = SvPV (sv, len); |
596 | char *str = SvPV (sv, len); |
427 | encode_str (enc, SvUTF8 (sv), str, len); |
597 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
428 | } |
598 | } |
429 | else if (SvNOKp (sv)) |
599 | else if (SvNOKp (sv)) |
430 | encode_nv (enc, sv); |
600 | encode_nv (enc, sv); |
431 | else if (SvIOKp (sv)) |
601 | else if (SvIOKp (sv)) |
432 | { |
602 | { |
433 | if (SvIsUV (sv)) |
603 | if (SvIsUV (sv)) |
434 | encode_uint (enc, 0x00, SvUVX (sv)); |
604 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
435 | else if (SvIVX (sv) >= 0) |
605 | else if (SvIVX (sv) >= 0) |
436 | encode_uint (enc, 0x00, SvIVX (sv)); |
606 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
437 | else |
607 | else |
438 | encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); |
608 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
439 | } |
609 | } |
440 | else if (SvROK (sv)) |
610 | else if (SvROK (sv)) |
441 | encode_rv (enc, SvRV (sv)); |
611 | encode_rv (enc, SvRV (sv)); |
442 | else if (!SvOK (sv)) |
612 | else if (!SvOK (sv)) |
443 | encode_ch (enc, 0xe0 | 22); |
613 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
444 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
614 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
445 | encode_ch (enc, 0xe0 | 23); |
615 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
446 | else |
616 | else |
447 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
617 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
448 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
618 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
449 | } |
619 | } |
450 | |
620 | |
451 | static SV * |
621 | static SV * |
452 | encode_cbor (SV *scalar, CBOR *cbor) |
622 | encode_cbor (SV *scalar, CBOR *cbor) |
453 | { |
623 | { |
454 | enc_t enc; |
624 | enc_t enc = { 0 }; |
455 | |
625 | |
456 | enc.cbor = *cbor; |
626 | enc.cbor = *cbor; |
457 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
627 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
458 | enc.cur = SvPVX (enc.sv); |
628 | enc.cur = SvPVX (enc.sv); |
459 | enc.end = SvEND (enc.sv); |
629 | enc.end = SvEND (enc.sv); |
460 | enc.depth = 0; |
|
|
461 | |
630 | |
462 | SvPOK_only (enc.sv); |
631 | SvPOK_only (enc.sv); |
|
|
632 | |
|
|
633 | if (cbor->flags & F_PACK_STRINGS) |
|
|
634 | { |
|
|
635 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
|
|
636 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
637 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
638 | } |
|
|
639 | |
463 | encode_sv (&enc, scalar); |
640 | encode_sv (&enc, scalar); |
464 | |
641 | |
465 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
642 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
466 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
643 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
467 | |
644 | |
… | |
… | |
481 | U8 *end; // end of input string |
658 | U8 *end; // end of input string |
482 | const char *err; // parse error, if != 0 |
659 | const char *err; // parse error, if != 0 |
483 | CBOR cbor; |
660 | CBOR cbor; |
484 | U32 depth; // recursion depth |
661 | U32 depth; // recursion depth |
485 | U32 maxdepth; // recursion depth limit |
662 | U32 maxdepth; // recursion depth limit |
|
|
663 | AV *shareable; |
|
|
664 | AV *stringref; |
|
|
665 | SV *decode_tagged; |
|
|
666 | SV *err_sv; // optional sv for error, needs to be freed |
486 | } dec_t; |
667 | } dec_t; |
487 | |
668 | |
488 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
669 | // set dec->err to ERRSV |
|
|
670 | ecb_cold static void |
|
|
671 | err_errsv (dec_t *dec) |
|
|
672 | { |
|
|
673 | if (!dec->err) |
|
|
674 | { |
|
|
675 | dec->err_sv = newSVsv (ERRSV); |
489 | |
676 | |
490 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
677 | // chop off the trailing \n |
|
|
678 | SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1); |
|
|
679 | *SvEND (dec->err_sv) = 0; |
491 | |
680 | |
|
|
681 | dec->err = SvPVutf8_nolen (dec->err_sv); |
|
|
682 | } |
|
|
683 | } |
|
|
684 | |
|
|
685 | // the following functions are used to reduce code size and help the compiler to optimise |
|
|
686 | ecb_cold static void |
|
|
687 | err_set (dec_t *dec, const char *reason) |
|
|
688 | { |
|
|
689 | if (!dec->err) |
|
|
690 | dec->err = reason; |
|
|
691 | } |
|
|
692 | |
|
|
693 | ecb_cold static void |
|
|
694 | err_unexpected_end (dec_t *dec) |
|
|
695 | { |
|
|
696 | err_set (dec, "unexpected end of CBOR data"); |
|
|
697 | } |
|
|
698 | |
|
|
699 | ecb_cold static void |
|
|
700 | err_nesting_exceeded (dec_t *dec) |
|
|
701 | { |
|
|
702 | err_set (dec, ERR_NESTING_EXCEEDED); |
|
|
703 | } |
|
|
704 | |
|
|
705 | #define ERR_DO(do) SB do; goto fail; SE |
|
|
706 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
|
|
707 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
|
|
708 | |
|
|
709 | #define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec)) |
|
|
710 | |
492 | #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
711 | #define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED) |
493 | #define DEC_DEC_DEPTH --dec->depth |
712 | #define DEC_DEC_DEPTH --dec->depth |
494 | |
713 | |
495 | static UV |
714 | static UV |
496 | decode_uint (dec_t *dec) |
715 | decode_uint (dec_t *dec) |
497 | { |
716 | { |
498 | switch (*dec->cur & 31) |
717 | U8 m = *dec->cur & MINOR_MASK; |
499 | { |
718 | ++dec->cur; |
500 | case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: |
|
|
501 | case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: |
|
|
502 | case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23: |
|
|
503 | return *dec->cur++ & 31; |
|
|
504 | |
719 | |
505 | case 24: |
720 | if (ecb_expect_true (m < LENGTH_EXT1)) |
|
|
721 | return m; |
|
|
722 | else if (ecb_expect_true (m == LENGTH_EXT1)) |
|
|
723 | { |
506 | WANT (2); |
724 | WANT (1); |
507 | dec->cur += 2; |
725 | dec->cur += 1; |
508 | return dec->cur[-1]; |
726 | return dec->cur[-1]; |
509 | |
727 | } |
510 | case 25: |
728 | else if (ecb_expect_true (m == LENGTH_EXT2)) |
|
|
729 | { |
511 | WANT (3); |
730 | WANT (2); |
512 | dec->cur += 3; |
731 | dec->cur += 2; |
513 | return (((UV)dec->cur[-2]) << 8) |
732 | return (((UV)dec->cur[-2]) << 8) |
514 | | ((UV)dec->cur[-1]); |
733 | | ((UV)dec->cur[-1]); |
515 | |
734 | } |
516 | case 26: |
735 | else if (ecb_expect_true (m == LENGTH_EXT4)) |
|
|
736 | { |
517 | WANT (5); |
737 | WANT (4); |
518 | dec->cur += 5; |
738 | dec->cur += 4; |
519 | return (((UV)dec->cur[-4]) << 24) |
739 | return (((UV)dec->cur[-4]) << 24) |
520 | | (((UV)dec->cur[-3]) << 16) |
740 | | (((UV)dec->cur[-3]) << 16) |
521 | | (((UV)dec->cur[-2]) << 8) |
741 | | (((UV)dec->cur[-2]) << 8) |
522 | | ((UV)dec->cur[-1]); |
742 | | ((UV)dec->cur[-1]); |
523 | |
743 | } |
524 | case 27: |
744 | else if (ecb_expect_true (m == LENGTH_EXT8)) |
|
|
745 | { |
525 | WANT (9); |
746 | WANT (8); |
526 | dec->cur += 9; |
747 | dec->cur += 8; |
|
|
748 | |
|
|
749 | return |
|
|
750 | #if UVSIZE < 8 |
|
|
751 | 0 |
|
|
752 | #else |
527 | return (((UV)dec->cur[-8]) << 56) |
753 | (((UV)dec->cur[-8]) << 56) |
528 | | (((UV)dec->cur[-7]) << 48) |
754 | | (((UV)dec->cur[-7]) << 48) |
529 | | (((UV)dec->cur[-6]) << 40) |
755 | | (((UV)dec->cur[-6]) << 40) |
530 | | (((UV)dec->cur[-5]) << 32) |
756 | | (((UV)dec->cur[-5]) << 32) |
|
|
757 | #endif |
531 | | (((UV)dec->cur[-4]) << 24) |
758 | | (((UV)dec->cur[-4]) << 24) |
532 | | (((UV)dec->cur[-3]) << 16) |
759 | | (((UV)dec->cur[-3]) << 16) |
533 | | (((UV)dec->cur[-2]) << 8) |
760 | | (((UV)dec->cur[-2]) << 8) |
534 | | ((UV)dec->cur[-1]); |
761 | | ((UV)dec->cur[-1]); |
535 | |
762 | } |
536 | default: |
763 | else |
537 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
764 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
538 | } |
|
|
539 | |
765 | |
540 | fail: |
766 | fail: |
541 | return 0; |
767 | return 0; |
542 | } |
768 | } |
543 | |
769 | |
… | |
… | |
548 | { |
774 | { |
549 | AV *av = newAV (); |
775 | AV *av = newAV (); |
550 | |
776 | |
551 | DEC_INC_DEPTH; |
777 | DEC_INC_DEPTH; |
552 | |
778 | |
553 | if ((*dec->cur & 31) == 31) |
779 | if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF)) |
554 | { |
780 | { |
555 | ++dec->cur; |
781 | ++dec->cur; |
556 | |
782 | |
557 | for (;;) |
783 | for (;;) |
558 | { |
784 | { |
559 | WANT (1); |
785 | WANT (1); |
560 | |
786 | |
561 | if (*dec->cur == (0xe0 | 31)) |
787 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
562 | { |
788 | { |
563 | ++dec->cur; |
789 | ++dec->cur; |
564 | break; |
790 | break; |
565 | } |
791 | } |
566 | |
792 | |
567 | av_push (av, decode_sv (dec)); |
793 | av_push (av, decode_sv (dec)); |
568 | } |
794 | } |
569 | } |
795 | } |
570 | else |
796 | else |
571 | { |
797 | { |
572 | int i, len = decode_uint (dec); |
798 | UV i, len = decode_uint (dec); |
573 | |
799 | |
|
|
800 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
574 | av_fill (av, len - 1); |
801 | av_fill (av, len - 1); |
575 | |
802 | |
576 | for (i = 0; i < len; ++i) |
803 | for (i = 0; i < len; ++i) |
577 | AvARRAY (av)[i] = decode_sv (dec); |
804 | AvARRAY (av)[i] = decode_sv (dec); |
578 | } |
805 | } |
… | |
… | |
588 | |
815 | |
589 | static void |
816 | static void |
590 | decode_he (dec_t *dec, HV *hv) |
817 | decode_he (dec_t *dec, HV *hv) |
591 | { |
818 | { |
592 | // for speed reasons, we specialcase single-string |
819 | // for speed reasons, we specialcase single-string |
593 | // byte or utf-8 strings as keys. |
820 | // byte or utf-8 strings as keys, but only when !stringref |
594 | |
821 | |
595 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
822 | if (ecb_expect_true (!dec->stringref)) |
|
|
823 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
596 | { |
824 | { |
597 | I32 len = decode_uint (dec); |
825 | STRLEN len = decode_uint (dec); |
598 | char *key = (char *)dec->cur; |
826 | char *key = (char *)dec->cur; |
599 | |
827 | |
|
|
828 | WANT (len); |
600 | dec->cur += len; |
829 | dec->cur += len; |
601 | |
830 | |
602 | hv_store (hv, key, len, decode_sv (dec), 0); |
831 | hv_store (hv, key, len, decode_sv (dec), 0); |
|
|
832 | |
|
|
833 | return; |
603 | } |
834 | } |
604 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
835 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
605 | { |
836 | { |
606 | I32 len = decode_uint (dec); |
837 | STRLEN len = decode_uint (dec); |
607 | char *key = (char *)dec->cur; |
838 | char *key = (char *)dec->cur; |
608 | |
839 | |
|
|
840 | WANT (len); |
609 | dec->cur += len; |
841 | dec->cur += len; |
610 | |
842 | |
|
|
843 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
|
|
844 | if (!is_utf8_string (key, len)) |
|
|
845 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
|
|
846 | |
611 | hv_store (hv, key, -len, decode_sv (dec), 0); |
847 | hv_store (hv, key, -len, decode_sv (dec), 0); |
|
|
848 | |
|
|
849 | return; |
612 | } |
850 | } |
613 | else |
851 | |
614 | { |
|
|
615 | SV *k = decode_sv (dec); |
852 | SV *k = decode_sv (dec); |
616 | SV *v = decode_sv (dec); |
853 | SV *v = decode_sv (dec); |
617 | |
854 | |
|
|
855 | // we leak memory if uncaught exceptions are thrown by random magical |
|
|
856 | // methods, and this is hopefully the only place where it can happen, |
|
|
857 | // so if there is a chance of an exception, take the very slow path. |
|
|
858 | // since catching exceptions is "undocumented/internal/forbidden" by |
|
|
859 | // the new p5p powers, we need to call out to a perl function :/ |
|
|
860 | if (ecb_expect_false (SvAMAGIC (k))) |
|
|
861 | { |
|
|
862 | dSP; |
|
|
863 | |
|
|
864 | ENTER; SAVETMPS; |
|
|
865 | PUSHMARK (SP); |
|
|
866 | EXTEND (SP, 3); |
|
|
867 | PUSHs (sv_2mortal (newRV_inc ((SV *)hv))); |
|
|
868 | PUSHs (sv_2mortal (k)); |
|
|
869 | PUSHs (sv_2mortal (v)); |
|
|
870 | |
|
|
871 | PUTBACK; |
|
|
872 | call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL); |
|
|
873 | SPAGAIN; |
|
|
874 | |
|
|
875 | FREETMPS; LEAVE; |
|
|
876 | |
|
|
877 | if (SvTRUE (ERRSV)) |
|
|
878 | ERR_ERRSV; |
|
|
879 | |
|
|
880 | return; |
|
|
881 | } |
|
|
882 | |
618 | hv_store_ent (hv, k, v, 0); |
883 | hv_store_ent (hv, k, v, 0); |
619 | SvREFCNT_dec (k); |
884 | SvREFCNT_dec (k); |
620 | } |
885 | |
|
|
886 | fail: |
|
|
887 | ; |
621 | } |
888 | } |
622 | |
889 | |
623 | static SV * |
890 | static SV * |
624 | decode_hv (dec_t *dec) |
891 | decode_hv (dec_t *dec) |
625 | { |
892 | { |
626 | HV *hv = newHV (); |
893 | HV *hv = newHV (); |
627 | |
894 | |
628 | DEC_INC_DEPTH; |
895 | DEC_INC_DEPTH; |
629 | |
896 | |
630 | if ((*dec->cur & 31) == 31) |
897 | if (*dec->cur == (MAJOR_MAP | MINOR_INDEF)) |
631 | { |
898 | { |
632 | ++dec->cur; |
899 | ++dec->cur; |
633 | |
900 | |
634 | for (;;) |
901 | for (;;) |
635 | { |
902 | { |
636 | WANT (1); |
903 | WANT (1); |
637 | |
904 | |
638 | if (*dec->cur == (0xe0 | 31)) |
905 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
639 | { |
906 | { |
640 | ++dec->cur; |
907 | ++dec->cur; |
641 | break; |
908 | break; |
642 | } |
909 | } |
643 | |
910 | |
644 | decode_he (dec, hv); |
911 | decode_he (dec, hv); |
645 | } |
912 | } |
646 | } |
913 | } |
647 | else |
914 | else |
648 | { |
915 | { |
649 | int pairs = decode_uint (dec); |
916 | UV pairs = decode_uint (dec); |
|
|
917 | |
|
|
918 | WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes |
650 | |
919 | |
651 | while (pairs--) |
920 | while (pairs--) |
652 | decode_he (dec, hv); |
921 | decode_he (dec, hv); |
653 | } |
922 | } |
654 | |
923 | |
… | |
… | |
664 | static SV * |
933 | static SV * |
665 | decode_str (dec_t *dec, int utf8) |
934 | decode_str (dec_t *dec, int utf8) |
666 | { |
935 | { |
667 | SV *sv = 0; |
936 | SV *sv = 0; |
668 | |
937 | |
669 | if ((*dec->cur & 31) == 31) |
938 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
670 | { |
939 | { |
|
|
940 | // indefinite length strings |
671 | ++dec->cur; |
941 | ++dec->cur; |
672 | |
942 | |
|
|
943 | U8 major = *dec->cur & MAJOR_MISC; |
|
|
944 | |
673 | sv = newSVpvn ("", 0); |
945 | sv = newSVpvn ("", 0); |
674 | |
946 | |
675 | // not very fast, and certainly not robust against illegal input |
|
|
676 | for (;;) |
947 | for (;;) |
677 | { |
948 | { |
678 | WANT (1); |
949 | WANT (1); |
679 | |
950 | |
680 | if (*dec->cur == (0xe0 | 31)) |
951 | if ((*dec->cur - major) > LENGTH_EXT8) |
|
|
952 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
681 | { |
953 | { |
682 | ++dec->cur; |
954 | ++dec->cur; |
683 | break; |
955 | break; |
684 | } |
956 | } |
|
|
957 | else |
|
|
958 | ERR ("corrupted CBOR data (invalid chunks in indefinite length string)"); |
685 | |
959 | |
686 | sv_catsv (sv, decode_sv (dec)); |
960 | STRLEN len = decode_uint (dec); |
|
|
961 | |
|
|
962 | WANT (len); |
|
|
963 | sv_catpvn (sv, dec->cur, len); |
|
|
964 | dec->cur += len; |
687 | } |
965 | } |
688 | } |
966 | } |
689 | else |
967 | else |
690 | { |
968 | { |
691 | STRLEN len = decode_uint (dec); |
969 | STRLEN len = decode_uint (dec); |
692 | |
970 | |
693 | WANT (len); |
971 | WANT (len); |
694 | sv = newSVpvn (dec->cur, len); |
972 | sv = newSVpvn (dec->cur, len); |
695 | dec->cur += len; |
973 | dec->cur += len; |
|
|
974 | |
|
|
975 | if (ecb_expect_false (dec->stringref) |
|
|
976 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
|
|
977 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
696 | } |
978 | } |
697 | |
979 | |
698 | if (utf8) |
980 | if (utf8) |
|
|
981 | { |
|
|
982 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
|
|
983 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
|
|
984 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
|
|
985 | |
699 | SvUTF8_on (sv); |
986 | SvUTF8_on (sv); |
|
|
987 | } |
700 | |
988 | |
701 | return sv; |
989 | return sv; |
702 | |
990 | |
703 | fail: |
991 | fail: |
704 | SvREFCNT_dec (sv); |
992 | SvREFCNT_dec (sv); |
… | |
… | |
706 | } |
994 | } |
707 | |
995 | |
708 | static SV * |
996 | static SV * |
709 | decode_tagged (dec_t *dec) |
997 | decode_tagged (dec_t *dec) |
710 | { |
998 | { |
|
|
999 | SV *sv = 0; |
711 | UV tag = decode_uint (dec); |
1000 | UV tag = decode_uint (dec); |
|
|
1001 | |
|
|
1002 | WANT (1); |
|
|
1003 | |
|
|
1004 | switch (tag) |
|
|
1005 | { |
|
|
1006 | case CBOR_TAG_MAGIC: |
712 | SV *sv = decode_sv (dec); |
1007 | sv = decode_sv (dec); |
|
|
1008 | break; |
713 | |
1009 | |
714 | if (tag == CBOR_TAG_MAGIC) |
1010 | case CBOR_TAG_INDIRECTION: |
715 | return sv; |
1011 | sv = newRV_noinc (decode_sv (dec)); |
716 | else if (tag == CBOR_TAG_PERL_OBJECT) |
1012 | break; |
717 | { |
1013 | |
|
|
1014 | case CBOR_TAG_STRINGREF_NAMESPACE: |
|
|
1015 | { |
|
|
1016 | // do nmot use SAVETMPS/FREETMPS, as these will |
|
|
1017 | // erase mortalised caches, e.g. "shareable" |
|
|
1018 | ENTER; |
|
|
1019 | |
|
|
1020 | SAVESPTR (dec->stringref); |
|
|
1021 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
1022 | |
|
|
1023 | sv = decode_sv (dec); |
|
|
1024 | |
|
|
1025 | LEAVE; |
|
|
1026 | } |
|
|
1027 | break; |
|
|
1028 | |
|
|
1029 | case CBOR_TAG_STRINGREF: |
|
|
1030 | { |
|
|
1031 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
|
|
1032 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
|
|
1033 | |
|
|
1034 | UV idx = decode_uint (dec); |
|
|
1035 | |
|
|
1036 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
|
|
1037 | ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
|
|
1038 | |
|
|
1039 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
|
|
1040 | } |
|
|
1041 | break; |
|
|
1042 | |
|
|
1043 | case CBOR_TAG_VALUE_SHAREABLE: |
|
|
1044 | { |
|
|
1045 | if (ecb_expect_false (!dec->shareable)) |
|
|
1046 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
1047 | |
|
|
1048 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
|
|
1049 | { |
|
|
1050 | sv = newSV (0); |
|
|
1051 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
|
|
1052 | |
|
|
1053 | SV *osv = decode_sv (dec); |
|
|
1054 | sv_setsv (sv, osv); |
|
|
1055 | SvREFCNT_dec_NN (osv); |
|
|
1056 | } |
|
|
1057 | else |
|
|
1058 | { |
|
|
1059 | av_push (dec->shareable, &PL_sv_undef); |
|
|
1060 | int idx = AvFILLp (dec->shareable); |
|
|
1061 | sv = decode_sv (dec); |
|
|
1062 | av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
|
|
1063 | } |
|
|
1064 | } |
|
|
1065 | break; |
|
|
1066 | |
|
|
1067 | case CBOR_TAG_VALUE_SHAREDREF: |
|
|
1068 | { |
|
|
1069 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
|
|
1070 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
|
|
1071 | |
|
|
1072 | UV idx = decode_uint (dec); |
|
|
1073 | |
|
|
1074 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
|
|
1075 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
|
|
1076 | |
|
|
1077 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
1078 | |
|
|
1079 | if (sv == &PL_sv_undef) |
|
|
1080 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
|
|
1081 | } |
|
|
1082 | break; |
|
|
1083 | |
|
|
1084 | case CBOR_TAG_PERL_OBJECT: |
|
|
1085 | { |
|
|
1086 | sv = decode_sv (dec); |
|
|
1087 | |
718 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
1088 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
719 | ERR ("corrupted CBOR data (non-array perl object)"); |
1089 | ERR ("corrupted CBOR data (non-array perl object)"); |
720 | |
1090 | |
721 | AV *av = (AV *)SvRV (sv); |
1091 | AV *av = (AV *)SvRV (sv); |
722 | int len = av_len (av) + 1; |
1092 | int len = av_len (av) + 1; |
723 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
1093 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
724 | |
1094 | |
725 | if (!stash) |
1095 | if (!stash) |
726 | ERR ("cannot decode perl-object (package does not exist)"); |
1096 | ERR ("cannot decode perl-object (package does not exist)"); |
727 | |
1097 | |
728 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
1098 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
729 | |
1099 | |
730 | if (!method) |
1100 | if (!method) |
731 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
1101 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
732 | |
1102 | |
733 | dSP; |
1103 | dSP; |
734 | |
1104 | |
735 | ENTER; SAVETMPS; PUSHMARK (SP); |
1105 | ENTER; SAVETMPS; |
|
|
1106 | PUSHMARK (SP); |
736 | EXTEND (SP, len + 1); |
1107 | EXTEND (SP, len + 1); |
737 | // we re-bless the reference to get overload and other niceties right |
1108 | // we re-bless the reference to get overload and other niceties right |
738 | PUSHs (*av_fetch (av, 0, 1)); |
1109 | PUSHs (*av_fetch (av, 0, 1)); |
739 | PUSHs (sv_cbor); |
1110 | PUSHs (sv_cbor); |
740 | |
1111 | |
741 | int i; |
1112 | int i; |
742 | |
1113 | |
743 | for (i = 1; i < len; ++i) |
1114 | for (i = 1; i < len; ++i) |
744 | PUSHs (*av_fetch (av, i, 1)); |
1115 | PUSHs (*av_fetch (av, i, 1)); |
745 | |
1116 | |
746 | PUTBACK; |
1117 | PUTBACK; |
747 | call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
1118 | call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
748 | SPAGAIN; |
1119 | SPAGAIN; |
749 | |
1120 | |
750 | if (SvTRUE (ERRSV)) |
1121 | if (SvTRUE (ERRSV)) |
751 | { |
1122 | { |
|
|
1123 | FREETMPS; LEAVE; |
|
|
1124 | ERR_ERRSV; |
|
|
1125 | } |
|
|
1126 | |
|
|
1127 | SvREFCNT_dec (sv); |
|
|
1128 | sv = SvREFCNT_inc (POPs); |
|
|
1129 | |
|
|
1130 | PUTBACK; |
|
|
1131 | |
752 | FREETMPS; LEAVE; |
1132 | FREETMPS; LEAVE; |
753 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1133 | } |
|
|
1134 | break; |
|
|
1135 | |
|
|
1136 | default: |
754 | } |
1137 | { |
|
|
1138 | SV *tag_sv = newSVuv (tag); |
755 | |
1139 | |
|
|
1140 | sv = decode_sv (dec); |
|
|
1141 | |
|
|
1142 | dSP; |
|
|
1143 | ENTER; SAVETMPS; |
|
|
1144 | SAVESTACK_POS (); |
|
|
1145 | PUSHMARK (SP); |
|
|
1146 | EXTEND (SP, 2); |
|
|
1147 | PUSHs (tag_sv); |
|
|
1148 | PUSHs (sv); |
|
|
1149 | |
|
|
1150 | PUTBACK; |
|
|
1151 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
|
|
1152 | SPAGAIN; |
|
|
1153 | |
|
|
1154 | if (SvTRUE (ERRSV)) |
|
|
1155 | { |
|
|
1156 | SvREFCNT_dec (tag_sv); |
|
|
1157 | FREETMPS; LEAVE; |
|
|
1158 | ERR_ERRSV; |
|
|
1159 | } |
|
|
1160 | |
|
|
1161 | if (count) |
|
|
1162 | { |
|
|
1163 | SvREFCNT_dec (tag_sv); |
756 | SvREFCNT_dec (sv); |
1164 | SvREFCNT_dec (sv); |
757 | sv = SvREFCNT_inc (POPs); |
1165 | sv = SvREFCNT_inc (POPs); |
758 | |
1166 | } |
759 | PUTBACK; |
1167 | else |
760 | |
1168 | { |
761 | FREETMPS; LEAVE; |
|
|
762 | |
|
|
763 | return sv; |
|
|
764 | } |
|
|
765 | else |
|
|
766 | { |
|
|
767 | AV *av = newAV (); |
1169 | AV *av = newAV (); |
768 | av_push (av, newSVuv (tag)); |
1170 | av_push (av, tag_sv); |
769 | av_push (av, sv); |
1171 | av_push (av, sv); |
770 | |
1172 | |
771 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1173 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
772 | ? cbor_tagged_stash |
1174 | ? cbor_tagged_stash |
773 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
1175 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
774 | |
|
|
775 | return sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
1176 | sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
|
|
1177 | } |
|
|
1178 | |
|
|
1179 | PUTBACK; |
|
|
1180 | |
|
|
1181 | FREETMPS; LEAVE; |
|
|
1182 | } |
|
|
1183 | break; |
776 | } |
1184 | } |
|
|
1185 | |
|
|
1186 | return sv; |
777 | |
1187 | |
778 | fail: |
1188 | fail: |
779 | SvREFCNT_dec (sv); |
1189 | SvREFCNT_dec (sv); |
780 | return &PL_sv_undef; |
1190 | return &PL_sv_undef; |
781 | } |
1191 | } |
… | |
… | |
783 | static SV * |
1193 | static SV * |
784 | decode_sv (dec_t *dec) |
1194 | decode_sv (dec_t *dec) |
785 | { |
1195 | { |
786 | WANT (1); |
1196 | WANT (1); |
787 | |
1197 | |
788 | switch (*dec->cur >> 5) |
1198 | switch (*dec->cur >> MAJOR_SHIFT) |
789 | { |
1199 | { |
790 | case 0: // unsigned int |
1200 | case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec)); |
791 | return newSVuv (decode_uint (dec)); |
1201 | case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec)); |
792 | case 1: // negative int |
1202 | case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0); |
793 | return newSViv (-1 - (IV)decode_uint (dec)); |
1203 | case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1); |
794 | case 2: // octet string |
1204 | case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec); |
795 | return decode_str (dec, 0); |
1205 | case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec); |
796 | case 3: // utf-8 string |
1206 | case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec); |
797 | return decode_str (dec, 1); |
1207 | |
798 | case 4: // array |
1208 | case MAJOR_MISC >> MAJOR_SHIFT: |
799 | return decode_av (dec); |
|
|
800 | case 5: // map |
|
|
801 | return decode_hv (dec); |
|
|
802 | case 6: // tag |
|
|
803 | return decode_tagged (dec); |
|
|
804 | case 7: // misc |
|
|
805 | switch (*dec->cur++ & 31) |
1209 | switch (*dec->cur++ & MINOR_MASK) |
806 | { |
1210 | { |
807 | case 20: |
1211 | case SIMPLE_FALSE: |
808 | #if CBOR_SLOW |
1212 | #if CBOR_SLOW |
809 | types_false = get_bool ("Types::Serialiser::false"); |
1213 | types_false = get_bool ("Types::Serialiser::false"); |
810 | #endif |
1214 | #endif |
811 | return newSVsv (types_false); |
1215 | return newSVsv (types_false); |
812 | case 21: |
1216 | case SIMPLE_TRUE: |
813 | #if CBOR_SLOW |
1217 | #if CBOR_SLOW |
814 | types_true = get_bool ("Types::Serialiser::true"); |
1218 | types_true = get_bool ("Types::Serialiser::true"); |
815 | #endif |
1219 | #endif |
816 | return newSVsv (types_true); |
1220 | return newSVsv (types_true); |
817 | case 22: |
1221 | case SIMPLE_NULL: |
818 | return newSVsv (&PL_sv_undef); |
1222 | return newSVsv (&PL_sv_undef); |
819 | case 23: |
1223 | case SIMPLE_UNDEF: |
820 | #if CBOR_SLOW |
1224 | #if CBOR_SLOW |
821 | types_error = get_bool ("Types::Serialiser::error"); |
1225 | types_error = get_bool ("Types::Serialiser::error"); |
822 | #endif |
1226 | #endif |
823 | return newSVsv (types_error); |
1227 | return newSVsv (types_error); |
824 | |
1228 | |
825 | case 25: |
1229 | case MISC_FLOAT16: |
826 | { |
1230 | { |
827 | WANT (2); |
1231 | WANT (2); |
828 | |
1232 | |
829 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1233 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
830 | dec->cur += 2; |
1234 | dec->cur += 2; |
831 | |
1235 | |
832 | return newSVnv (ecb_binary16_to_float (fp)); |
1236 | return newSVnv (ecb_binary16_to_float (fp)); |
833 | } |
1237 | } |
834 | |
1238 | |
835 | case 26: |
1239 | case MISC_FLOAT32: |
836 | { |
1240 | { |
837 | uint32_t fp; |
1241 | uint32_t fp; |
838 | WANT (4); |
1242 | WANT (4); |
839 | memcpy (&fp, dec->cur, 4); |
1243 | memcpy (&fp, dec->cur, 4); |
840 | dec->cur += 4; |
1244 | dec->cur += 4; |
… | |
… | |
843 | fp = ecb_bswap32 (fp); |
1247 | fp = ecb_bswap32 (fp); |
844 | |
1248 | |
845 | return newSVnv (ecb_binary32_to_float (fp)); |
1249 | return newSVnv (ecb_binary32_to_float (fp)); |
846 | } |
1250 | } |
847 | |
1251 | |
848 | case 27: |
1252 | case MISC_FLOAT64: |
849 | { |
1253 | { |
850 | uint64_t fp; |
1254 | uint64_t fp; |
851 | WANT (8); |
1255 | WANT (8); |
852 | memcpy (&fp, dec->cur, 8); |
1256 | memcpy (&fp, dec->cur, 8); |
853 | dec->cur += 8; |
1257 | dec->cur += 8; |
… | |
… | |
856 | fp = ecb_bswap64 (fp); |
1260 | fp = ecb_bswap64 (fp); |
857 | |
1261 | |
858 | return newSVnv (ecb_binary64_to_double (fp)); |
1262 | return newSVnv (ecb_binary64_to_double (fp)); |
859 | } |
1263 | } |
860 | |
1264 | |
861 | // 0..19 unassigned |
1265 | // 0..19 unassigned simple |
862 | // 24 reserved + unassigned (reserved values are not encodable) |
1266 | // 24 reserved + unassigned simple (reserved values are not encodable) |
|
|
1267 | // 28-30 unassigned misc |
|
|
1268 | // 31 break code |
863 | default: |
1269 | default: |
864 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1270 | ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)"); |
865 | } |
1271 | } |
866 | |
1272 | |
867 | break; |
1273 | break; |
868 | } |
1274 | } |
869 | |
1275 | |
… | |
… | |
872 | } |
1278 | } |
873 | |
1279 | |
874 | static SV * |
1280 | static SV * |
875 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1281 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
876 | { |
1282 | { |
877 | dec_t dec; |
1283 | dec_t dec = { 0 }; |
878 | SV *sv; |
1284 | SV *sv; |
879 | STRLEN len; |
1285 | STRLEN len; |
880 | char *data = SvPVbyte (string, len); |
1286 | char *data = SvPVbyte (string, len); |
881 | |
1287 | |
882 | if (len > cbor->max_size && cbor->max_size) |
1288 | if (len > cbor->max_size && cbor->max_size) |
… | |
… | |
884 | (unsigned long)len, (unsigned long)cbor->max_size); |
1290 | (unsigned long)len, (unsigned long)cbor->max_size); |
885 | |
1291 | |
886 | dec.cbor = *cbor; |
1292 | dec.cbor = *cbor; |
887 | dec.cur = (U8 *)data; |
1293 | dec.cur = (U8 *)data; |
888 | dec.end = (U8 *)data + len; |
1294 | dec.end = (U8 *)data + len; |
889 | dec.err = 0; |
|
|
890 | dec.depth = 0; |
|
|
891 | |
1295 | |
892 | sv = decode_sv (&dec); |
1296 | sv = decode_sv (&dec); |
893 | |
1297 | |
894 | if (offset_return) |
1298 | if (offset_return) |
895 | *offset_return = dec.cur; |
1299 | *offset_return = dec.cur; |
… | |
… | |
898 | if (dec.cur != dec.end && !dec.err) |
1302 | if (dec.cur != dec.end && !dec.err) |
899 | dec.err = "garbage after CBOR object"; |
1303 | dec.err = "garbage after CBOR object"; |
900 | |
1304 | |
901 | if (dec.err) |
1305 | if (dec.err) |
902 | { |
1306 | { |
|
|
1307 | if (dec.shareable) |
|
|
1308 | { |
|
|
1309 | // need to break cyclic links, which would all be in shareable |
|
|
1310 | int i; |
|
|
1311 | SV **svp; |
|
|
1312 | |
|
|
1313 | for (i = av_len (dec.shareable) + 1; i--; ) |
|
|
1314 | if ((svp = av_fetch (dec.shareable, i, 0))) |
|
|
1315 | sv_setsv (*svp, &PL_sv_undef); |
|
|
1316 | } |
|
|
1317 | |
903 | SvREFCNT_dec (sv); |
1318 | SvREFCNT_dec (sv); |
|
|
1319 | |
|
|
1320 | if (dec.err_sv) |
|
|
1321 | sv_2mortal (dec.err_sv); |
|
|
1322 | |
904 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1323 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
905 | } |
1324 | } |
906 | |
1325 | |
907 | sv = sv_2mortal (sv); |
1326 | sv = sv_2mortal (sv); |
908 | |
1327 | |
909 | return sv; |
1328 | return sv; |
910 | } |
1329 | } |
911 | |
1330 | |
|
|
1331 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1332 | // incremental parser |
|
|
1333 | |
|
|
1334 | #define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0) |
|
|
1335 | |
|
|
1336 | // returns 0 for notyet, 1 for success or error |
|
|
1337 | static int |
|
|
1338 | incr_parse (CBOR *self, SV *cborstr) |
|
|
1339 | { |
|
|
1340 | STRLEN cur; |
|
|
1341 | SvPV (cborstr, cur); |
|
|
1342 | |
|
|
1343 | while (ecb_expect_true (self->incr_need <= cur)) |
|
|
1344 | { |
|
|
1345 | // table of integer count bytes |
|
|
1346 | static I8 incr_len[MINOR_MASK + 1] = { |
|
|
1347 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1348 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1349 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1350 | 1, 2, 4, 8,-1,-1,-1,-2 |
|
|
1351 | }; |
|
|
1352 | |
|
|
1353 | const U8 *p = SvPVX (cborstr) + self->incr_pos; |
|
|
1354 | U8 m = *p & MINOR_MASK; |
|
|
1355 | IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1356 | I8 ilen = incr_len[m]; |
|
|
1357 | |
|
|
1358 | self->incr_need = self->incr_pos + 1; |
|
|
1359 | |
|
|
1360 | if (ecb_expect_false (ilen < 0)) |
|
|
1361 | { |
|
|
1362 | if (m != MINOR_INDEF) |
|
|
1363 | return 1; // error |
|
|
1364 | |
|
|
1365 | if (*p == (MAJOR_MISC | MINOR_INDEF)) |
|
|
1366 | { |
|
|
1367 | if (count >= 0) |
|
|
1368 | return 1; // error |
|
|
1369 | |
|
|
1370 | count = 1; |
|
|
1371 | } |
|
|
1372 | else |
|
|
1373 | { |
|
|
1374 | av_push (self->incr_count, newSViv (-1)); //TODO: nest |
|
|
1375 | count = -1; |
|
|
1376 | } |
|
|
1377 | } |
|
|
1378 | else |
|
|
1379 | { |
|
|
1380 | self->incr_need += ilen; |
|
|
1381 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1382 | return 0; |
|
|
1383 | |
|
|
1384 | int major = *p >> MAJOR_SHIFT; |
|
|
1385 | |
|
|
1386 | switch (major) |
|
|
1387 | { |
|
|
1388 | case MAJOR_TAG >> MAJOR_SHIFT: |
|
|
1389 | ++count; // tags merely prefix another value |
|
|
1390 | break; |
|
|
1391 | |
|
|
1392 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1393 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1394 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1395 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1396 | { |
|
|
1397 | UV len; |
|
|
1398 | |
|
|
1399 | if (ecb_expect_false (ilen)) |
|
|
1400 | { |
|
|
1401 | len = 0; |
|
|
1402 | |
|
|
1403 | do { |
|
|
1404 | len = (len << 8) | *++p; |
|
|
1405 | } while (--ilen); |
|
|
1406 | } |
|
|
1407 | else |
|
|
1408 | len = m; |
|
|
1409 | |
|
|
1410 | switch (major) |
|
|
1411 | { |
|
|
1412 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1413 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1414 | self->incr_need += len; |
|
|
1415 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1416 | return 0; |
|
|
1417 | |
|
|
1418 | break; |
|
|
1419 | |
|
|
1420 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1421 | len <<= 1; |
|
|
1422 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1423 | if (len) |
|
|
1424 | { |
|
|
1425 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
|
|
1426 | count = len + 1; |
|
|
1427 | } |
|
|
1428 | break; |
|
|
1429 | } |
|
|
1430 | } |
|
|
1431 | } |
|
|
1432 | } |
|
|
1433 | |
|
|
1434 | self->incr_pos = self->incr_need; |
|
|
1435 | |
|
|
1436 | if (count > 0) |
|
|
1437 | { |
|
|
1438 | while (!--count) |
|
|
1439 | { |
|
|
1440 | if (!AvFILLp (self->incr_count)) |
|
|
1441 | return 1; // done |
|
|
1442 | |
|
|
1443 | SvREFCNT_dec_NN (av_pop (self->incr_count)); |
|
|
1444 | count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1445 | } |
|
|
1446 | |
|
|
1447 | SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count; |
|
|
1448 | } |
|
|
1449 | } |
|
|
1450 | |
|
|
1451 | return 0; |
|
|
1452 | } |
|
|
1453 | |
|
|
1454 | |
912 | ///////////////////////////////////////////////////////////////////////////// |
1455 | ///////////////////////////////////////////////////////////////////////////// |
913 | // XS interface functions |
1456 | // XS interface functions |
914 | |
1457 | |
915 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1458 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
916 | |
1459 | |
… | |
… | |
924 | |
1467 | |
925 | types_true = get_bool ("Types::Serialiser::true" ); |
1468 | types_true = get_bool ("Types::Serialiser::true" ); |
926 | types_false = get_bool ("Types::Serialiser::false"); |
1469 | types_false = get_bool ("Types::Serialiser::false"); |
927 | types_error = get_bool ("Types::Serialiser::error"); |
1470 | types_error = get_bool ("Types::Serialiser::error"); |
928 | |
1471 | |
|
|
1472 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
|
|
1473 | |
929 | sv_cbor = newSVpv ("CBOR", 0); |
1474 | sv_cbor = newSVpv ("CBOR", 0); |
930 | SvREADONLY_on (sv_cbor); |
1475 | SvREADONLY_on (sv_cbor); |
|
|
1476 | |
|
|
1477 | assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1)); |
931 | } |
1478 | } |
932 | |
1479 | |
933 | PROTOTYPES: DISABLE |
1480 | PROTOTYPES: DISABLE |
934 | |
1481 | |
935 | void CLONE (...) |
1482 | void CLONE (...) |
… | |
… | |
953 | |
1500 | |
954 | void shrink (CBOR *self, int enable = 1) |
1501 | void shrink (CBOR *self, int enable = 1) |
955 | ALIAS: |
1502 | ALIAS: |
956 | shrink = F_SHRINK |
1503 | shrink = F_SHRINK |
957 | allow_unknown = F_ALLOW_UNKNOWN |
1504 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
1505 | allow_sharing = F_ALLOW_SHARING |
|
|
1506 | allow_cycles = F_ALLOW_CYCLES |
|
|
1507 | pack_strings = F_PACK_STRINGS |
|
|
1508 | text_keys = F_TEXT_KEYS |
|
|
1509 | text_strings = F_TEXT_STRINGS |
|
|
1510 | validate_utf8 = F_VALIDATE_UTF8 |
958 | PPCODE: |
1511 | PPCODE: |
959 | { |
1512 | { |
960 | if (enable) |
1513 | if (enable) |
961 | self->flags |= ix; |
1514 | self->flags |= ix; |
962 | else |
1515 | else |
… | |
… | |
967 | |
1520 | |
968 | void get_shrink (CBOR *self) |
1521 | void get_shrink (CBOR *self) |
969 | ALIAS: |
1522 | ALIAS: |
970 | get_shrink = F_SHRINK |
1523 | get_shrink = F_SHRINK |
971 | get_allow_unknown = F_ALLOW_UNKNOWN |
1524 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
1525 | get_allow_sharing = F_ALLOW_SHARING |
|
|
1526 | get_allow_cycles = F_ALLOW_CYCLES |
|
|
1527 | get_pack_strings = F_PACK_STRINGS |
|
|
1528 | get_text_keys = F_TEXT_KEYS |
|
|
1529 | get_text_strings = F_TEXT_STRINGS |
|
|
1530 | get_validate_utf8 = F_VALIDATE_UTF8 |
972 | PPCODE: |
1531 | PPCODE: |
973 | XPUSHs (boolSV (self->flags & ix)); |
1532 | XPUSHs (boolSV (self->flags & ix)); |
974 | |
1533 | |
975 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1534 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
976 | PPCODE: |
1535 | PPCODE: |
… | |
… | |
989 | XPUSHs (ST (0)); |
1548 | XPUSHs (ST (0)); |
990 | |
1549 | |
991 | int get_max_size (CBOR *self) |
1550 | int get_max_size (CBOR *self) |
992 | CODE: |
1551 | CODE: |
993 | RETVAL = self->max_size; |
1552 | RETVAL = self->max_size; |
|
|
1553 | OUTPUT: |
|
|
1554 | RETVAL |
|
|
1555 | |
|
|
1556 | void filter (CBOR *self, SV *filter = 0) |
|
|
1557 | PPCODE: |
|
|
1558 | SvREFCNT_dec (self->filter); |
|
|
1559 | self->filter = filter ? newSVsv (filter) : filter; |
|
|
1560 | XPUSHs (ST (0)); |
|
|
1561 | |
|
|
1562 | SV *get_filter (CBOR *self) |
|
|
1563 | CODE: |
|
|
1564 | RETVAL = self->filter ? self->filter : NEWSV (0, 0); |
994 | OUTPUT: |
1565 | OUTPUT: |
995 | RETVAL |
1566 | RETVAL |
996 | |
1567 | |
997 | void encode (CBOR *self, SV *scalar) |
1568 | void encode (CBOR *self, SV *scalar) |
998 | PPCODE: |
1569 | PPCODE: |
… | |
… | |
1013 | EXTEND (SP, 2); |
1584 | EXTEND (SP, 2); |
1014 | PUSHs (sv); |
1585 | PUSHs (sv); |
1015 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1586 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1016 | } |
1587 | } |
1017 | |
1588 | |
|
|
1589 | void incr_parse (CBOR *self, SV *cborstr) |
|
|
1590 | ALIAS: |
|
|
1591 | incr_parse_multiple = 1 |
|
|
1592 | PPCODE: |
|
|
1593 | { |
|
|
1594 | if (SvUTF8 (cborstr)) |
|
|
1595 | sv_utf8_downgrade (cborstr, 0); |
|
|
1596 | |
|
|
1597 | if (!self->incr_count) |
|
|
1598 | { |
|
|
1599 | self->incr_count = newAV (); |
|
|
1600 | self->incr_pos = 0; |
|
|
1601 | self->incr_need = 1; |
|
|
1602 | |
|
|
1603 | av_push (self->incr_count, newSViv (1)); |
|
|
1604 | } |
|
|
1605 | |
|
|
1606 | do |
|
|
1607 | { |
|
|
1608 | if (!incr_parse (self, cborstr)) |
|
|
1609 | { |
|
|
1610 | if (self->incr_need > self->max_size && self->max_size) |
|
|
1611 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
|
|
1612 | (unsigned long)self->incr_need, (unsigned long)self->max_size); |
|
|
1613 | |
|
|
1614 | break; |
|
|
1615 | } |
|
|
1616 | |
|
|
1617 | SV *sv; |
|
|
1618 | char *offset; |
|
|
1619 | |
|
|
1620 | PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
|
|
1621 | XPUSHs (sv); |
|
|
1622 | |
|
|
1623 | sv_chop (cborstr, offset); |
|
|
1624 | |
|
|
1625 | av_clear (self->incr_count); |
|
|
1626 | av_push (self->incr_count, newSViv (1)); |
|
|
1627 | |
|
|
1628 | self->incr_pos = 0; |
|
|
1629 | self->incr_need = self->incr_pos + 1; |
|
|
1630 | } |
|
|
1631 | while (ix); |
|
|
1632 | } |
|
|
1633 | |
|
|
1634 | void incr_reset (CBOR *self) |
|
|
1635 | CODE: |
|
|
1636 | { |
|
|
1637 | SvREFCNT_dec (self->incr_count); |
|
|
1638 | self->incr_count = 0; |
|
|
1639 | } |
|
|
1640 | |
|
|
1641 | void DESTROY (CBOR *self) |
|
|
1642 | PPCODE: |
|
|
1643 | cbor_free (self); |
|
|
1644 | |
1018 | PROTOTYPES: ENABLE |
1645 | PROTOTYPES: ENABLE |
1019 | |
1646 | |
1020 | void encode_cbor (SV *scalar) |
1647 | void encode_cbor (SV *scalar) |
|
|
1648 | ALIAS: |
|
|
1649 | encode_cbor = 0 |
|
|
1650 | encode_cbor_sharing = F_ALLOW_SHARING |
1021 | PPCODE: |
1651 | PPCODE: |
1022 | { |
1652 | { |
1023 | CBOR cbor; |
1653 | CBOR cbor; |
1024 | cbor_init (&cbor); |
1654 | cbor_init (&cbor); |
|
|
1655 | cbor.flags |= ix; |
1025 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1656 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1026 | XPUSHs (scalar); |
1657 | XPUSHs (scalar); |
1027 | } |
1658 | } |
1028 | |
1659 | |
1029 | void decode_cbor (SV *cborstr) |
1660 | void decode_cbor (SV *cborstr) |