]>
Commit | Line | Data |
---|---|---|
1 | //! Container format detection. | |
2 | //! | |
3 | //! Usually user does not know the container format of the opened file. | |
4 | //! That is why format detection functionality is needed. | |
5 | //! This module contains the set of rules to detect container not merely by file extension but also by its content if possible. | |
6 | //! | |
7 | //! # Examples | |
8 | //! | |
9 | //! ```no_run | |
10 | //! use nihav_registry::detect::detect_format; | |
11 | //! use std::fs::File; | |
12 | //! use nihav_core::io::byteio::*; | |
13 | //! | |
14 | //! let name = "mediafile.ogv"; | |
15 | //! let mut file = File::open(name).unwrap(); | |
16 | //! let mut filereader = FileReader::new_read(&mut file); | |
17 | //! let mut br = ByteReader::new(&mut filereader); | |
18 | //! let result = detect_format(name, &mut br); | |
19 | //! if let Some((name, score)) = result { | |
20 | //! println!("detected format {} with score {:?}", name, score); | |
21 | //! } | |
22 | //! ``` | |
23 | use std::io::SeekFrom; | |
24 | use nihav_core::io::byteio::ByteReader; | |
25 | ||
26 | /// Format detection score. | |
27 | #[derive(Debug,Clone,Copy,PartialEq)] | |
28 | pub enum DetectionScore { | |
29 | /// Format is not detected. | |
30 | No, | |
31 | /// Format matched by file extension. | |
32 | ExtensionMatches, | |
33 | /// Format matches by markers inside the file. | |
34 | MagicMatches, | |
35 | } | |
36 | ||
37 | impl DetectionScore { | |
38 | /// Checks whether current detection score is less than a value it is compared against. | |
39 | pub fn less(self, other: DetectionScore) -> bool { | |
40 | (self as i32) < (other as i32) | |
41 | } | |
42 | } | |
43 | ||
44 | #[allow(dead_code)] | |
45 | enum Arg { | |
46 | Byte(u8), | |
47 | U16BE(u16), | |
48 | U16LE(u16), | |
49 | U24BE(u32), | |
50 | U24LE(u32), | |
51 | U32BE(u32), | |
52 | U32LE(u32), | |
53 | U64BE(u64), | |
54 | U64LE(u64), | |
55 | } | |
56 | ||
57 | impl Arg { | |
58 | fn val(&self) -> u64 { | |
59 | match *self { | |
60 | Arg::Byte(b) => { u64::from(b) } | |
61 | Arg::U16BE(v) => { u64::from(v) } | |
62 | Arg::U16LE(v) => { u64::from(v) } | |
63 | Arg::U24BE(v) => { u64::from(v) } | |
64 | Arg::U24LE(v) => { u64::from(v) } | |
65 | Arg::U32BE(v) => { u64::from(v) } | |
66 | Arg::U32LE(v) => { u64::from(v) } | |
67 | Arg::U64BE(v) => { v } | |
68 | Arg::U64LE(v) => { v } | |
69 | } | |
70 | } | |
71 | fn read_val(&self, src: &mut ByteReader) -> Option<u64> { | |
72 | match *self { | |
73 | Arg::Byte(_) => { | |
74 | let res = src.peek_byte(); | |
75 | if res.is_err() { return None; } | |
76 | Some(u64::from(res.unwrap())) | |
77 | } | |
78 | Arg::U16BE(_) => { | |
79 | let res = src.peek_u16be(); | |
80 | if res.is_err() { return None; } | |
81 | Some(u64::from(res.unwrap())) | |
82 | } | |
83 | Arg::U16LE(_) => { | |
84 | let res = src.peek_u16le(); | |
85 | if res.is_err() { return None; } | |
86 | Some(u64::from(res.unwrap())) | |
87 | } | |
88 | Arg::U24BE(_) => { | |
89 | let res = src.peek_u24be(); | |
90 | if res.is_err() { return None; } | |
91 | Some(u64::from(res.unwrap())) | |
92 | } | |
93 | Arg::U24LE(_) => { | |
94 | let res = src.peek_u24le(); | |
95 | if res.is_err() { return None; } | |
96 | Some(u64::from(res.unwrap())) | |
97 | } | |
98 | Arg::U32BE(_) => { | |
99 | let res = src.peek_u32be(); | |
100 | if res.is_err() { return None; } | |
101 | Some(u64::from(res.unwrap())) | |
102 | } | |
103 | Arg::U32LE(_) => { | |
104 | let res = src.peek_u32le(); | |
105 | if res.is_err() { return None; } | |
106 | Some(u64::from(res.unwrap())) | |
107 | } | |
108 | Arg::U64BE(_) => { | |
109 | let res = src.peek_u64be(); | |
110 | if res.is_err() { return None; } | |
111 | Some(res.unwrap()) | |
112 | } | |
113 | Arg::U64LE(_) => { | |
114 | let res = src.peek_u64le(); | |
115 | if res.is_err() { return None; } | |
116 | Some(res.unwrap()) | |
117 | } | |
118 | } | |
119 | } | |
120 | fn eq(&self, src: &mut ByteReader) -> bool { | |
121 | if let Some(rval) = self.read_val(src) { | |
122 | rval == self.val() | |
123 | } else { | |
124 | false | |
125 | } | |
126 | } | |
127 | fn ge(&self, src: &mut ByteReader) -> bool { | |
128 | if let Some(rval) = self.read_val(src) { | |
129 | rval >= self.val() | |
130 | } else { | |
131 | false | |
132 | } | |
133 | } | |
134 | fn gt(&self, src: &mut ByteReader) -> bool { | |
135 | if let Some(rval) = self.read_val(src) { | |
136 | rval > self.val() | |
137 | } else { | |
138 | false | |
139 | } | |
140 | } | |
141 | fn le(&self, src: &mut ByteReader) -> bool { | |
142 | if let Some(rval) = self.read_val(src) { | |
143 | rval <= self.val() | |
144 | } else { | |
145 | false | |
146 | } | |
147 | } | |
148 | fn lt(&self, src: &mut ByteReader) -> bool { | |
149 | if let Some(rval) = self.read_val(src) { | |
150 | rval < self.val() | |
151 | } else { | |
152 | false | |
153 | } | |
154 | } | |
155 | } | |
156 | ||
157 | #[allow(dead_code)] | |
158 | enum CC<'a> { | |
159 | Or(&'a CC<'a>, &'a CC<'a>), | |
160 | Eq(Arg), | |
161 | Str(&'static [u8]), | |
162 | In(Arg, Arg), | |
163 | Lt(Arg), | |
164 | Le(Arg), | |
165 | Gt(Arg), | |
166 | Ge(Arg), | |
167 | } | |
168 | ||
169 | impl<'a> CC<'a> { | |
170 | fn eval(&self, src: &mut ByteReader) -> bool { | |
171 | match *self { | |
172 | CC::Or(a, b) => { a.eval(src) || b.eval(src) }, | |
173 | CC::Eq(ref arg) => { arg.eq(src) }, | |
174 | CC::In(ref a, ref b) => { a.ge(src) && b.le(src) }, | |
175 | CC::Lt(ref arg) => { arg.lt(src) }, | |
176 | CC::Le(ref arg) => { arg.le(src) }, | |
177 | CC::Gt(ref arg) => { arg.gt(src) }, | |
178 | CC::Ge(ref arg) => { arg.ge(src) }, | |
179 | CC::Str(strng) => { | |
180 | let mut val: Vec<u8> = vec![0; strng.len()]; | |
181 | let res = src.peek_buf(val.as_mut_slice()); | |
182 | if res.is_err() { return false; } | |
183 | val == strng | |
184 | } | |
185 | } | |
186 | } | |
187 | } | |
188 | ||
189 | struct CheckItem<'a> { | |
190 | offs: u32, | |
191 | cond: &'a CC<'a>, | |
192 | } | |
193 | ||
194 | #[allow(dead_code)] | |
195 | struct DetectConditions<'a> { | |
196 | demux_name: &'static str, | |
197 | extensions: &'static str, | |
198 | conditions: &'a [CheckItem<'a>], | |
199 | } | |
200 | ||
201 | const DETECTORS: &[DetectConditions] = &[ | |
202 | DetectConditions { | |
203 | demux_name: "avi", | |
204 | extensions: ".avi", | |
205 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"RIFF"), &CC::Str(b"ON2 ")) }, | |
206 | CheckItem{offs: 8, cond: &CC::Or(&CC::Or(&CC::Str(b"AVI LIST"), | |
207 | &CC::Str(b"AVIXLIST")), | |
208 | &CC::Str(b"ON2fLIST")) }, | |
209 | ] | |
210 | }, | |
211 | DetectConditions { | |
212 | demux_name: "wav", | |
213 | extensions: ".wav", | |
214 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"RIFF") }, | |
215 | CheckItem{offs: 8, cond: &CC::Str(b"WAVEfmt ") } | |
216 | ] | |
217 | }, | |
218 | DetectConditions { | |
219 | demux_name: "mov", | |
220 | extensions: ".mov", | |
221 | conditions: &[CheckItem{offs: 4, cond: &CC::Or(&CC::Or(&CC::Str(b"mdat"), | |
222 | &CC::Str(b"moov")), | |
223 | &CC::Str(b"ftyp")) }], | |
224 | }, | |
225 | DetectConditions { | |
226 | demux_name: "gif", | |
227 | extensions: ".gif", | |
228 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"GIF87a"), | |
229 | &CC::Str(b"GIF89a")) }], | |
230 | }, | |
231 | DetectConditions { | |
232 | demux_name: "mov", | |
233 | extensions: ".mov", | |
234 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"\x00\x00\x00\x08wide") }, | |
235 | CheckItem{offs: 12, cond: &CC::Or(&CC::Or(&CC::Str(b"mdat"), | |
236 | &CC::Str(b"moov")), | |
237 | &CC::Str(b"ftyp")) }], | |
238 | }, | |
239 | DetectConditions { | |
240 | demux_name: "mov-macbin", | |
241 | extensions: ".mov,.bin", | |
242 | conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::Byte(0))}, | |
243 | CheckItem{offs: 0x41, cond: &CC::Str(b"MooV")}, | |
244 | CheckItem{offs: 0x7A, cond: &CC::Eq(Arg::Byte(0x81))}, | |
245 | CheckItem{offs: 0x7B, cond: &CC::Eq(Arg::Byte(0x81))}, | |
246 | CheckItem{offs: 0x84, cond: &CC::Str(b"mdat")}], | |
247 | }, | |
248 | DetectConditions { | |
249 | demux_name: "yuv4mpeg", | |
250 | extensions: ".y4m", | |
251 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"YUV4MPEG2 ") }], | |
252 | }, | |
253 | DetectConditions { | |
254 | demux_name: "armovie", | |
255 | extensions: ".rpl", | |
256 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"ARMovie\n") }], | |
257 | }, | |
258 | DetectConditions { | |
259 | demux_name: "flv", | |
260 | extensions: ".flv", | |
261 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FLV") }, | |
262 | CheckItem{offs: 3, cond: &CC::Le(Arg::Byte(1)) }], | |
263 | }, | |
264 | DetectConditions { | |
265 | demux_name: "ivf", | |
266 | extensions: ".ivf", | |
267 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(&[0x50, 0xEF, 0x81, 0x19, 0xB3, 0xBD, 0xD0, 0x11, 0xA3, 0xE5, 0x00, 0xA0, 0xC9, 0x24, 0x44])}, | |
268 | CheckItem{offs: 15, cond: &CC::Or(&CC::Eq(Arg::Byte(0x36)), &CC::Eq(Arg::Byte(0x37)))}], | |
269 | }, | |
270 | DetectConditions { | |
271 | demux_name: "dkivf", | |
272 | extensions: ".ivf", | |
273 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"DKIF\x00\x00")}, | |
274 | CheckItem{offs: 6, cond: &CC::Ge(Arg::U16LE(32))}], | |
275 | }, | |
276 | DetectConditions { | |
277 | demux_name: "arxel-cnm", | |
278 | extensions: ".cnm", | |
279 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"CNM UNR\x00")}], | |
280 | }, | |
281 | DetectConditions { | |
282 | demux_name: "fcmp", | |
283 | extensions: ".cmp", | |
284 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FCMP")}], | |
285 | }, | |
286 | DetectConditions { | |
287 | demux_name: "fst", | |
288 | extensions: ".fst", | |
289 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"2TSF")}], | |
290 | }, | |
291 | DetectConditions { | |
292 | demux_name: "gdv", | |
293 | extensions: ".gdv", | |
294 | conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}], | |
295 | }, | |
296 | DetectConditions { | |
297 | demux_name: "fable-imax", | |
298 | extensions: ".imx", | |
299 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"IMAX") }, | |
300 | CheckItem{offs: 10, cond: &CC::Eq(Arg::U16LE(0x102)) }], | |
301 | }, | |
302 | DetectConditions { | |
303 | demux_name: "hl-fmv", | |
304 | extensions: ".fmv", | |
305 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FMV*") }, | |
306 | CheckItem{offs: 4, cond: &CC::Eq(Arg::U32LE(0)) }], | |
307 | }, | |
308 | DetectConditions { | |
309 | demux_name: "legend-q", | |
310 | extensions: ".q", | |
311 | conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U16LE(0x6839))}, | |
312 | CheckItem{offs: 2, cond: &CC::In(Arg::Byte(3), Arg::Byte(7))}], | |
313 | }, | |
314 | DetectConditions { | |
315 | demux_name: "siff", | |
316 | extensions: ".vb,.vbc,.fcp,.son", | |
317 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"SIFF")}, | |
318 | CheckItem{offs: 4, cond: &CC::Or( | |
319 | &CC::Or( | |
320 | &CC::Str(b"VBV1VBHD"), | |
321 | &CC::Str(b"SOUNSHDR")), | |
322 | &CC::Str(b"FCPKFCHD"))}], | |
323 | }, | |
324 | DetectConditions { | |
325 | demux_name: "smush", | |
326 | extensions: ".san", | |
327 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"ANIM")}, | |
328 | CheckItem{offs: 8, cond: &CC::Str(b"AHDR")}], | |
329 | }, | |
330 | DetectConditions { | |
331 | demux_name: "smush-mcmp", | |
332 | extensions: ".imc", | |
333 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MCMP")}, | |
334 | CheckItem{offs: 6, cond: &CC::Eq(Arg::Byte(0))}, | |
335 | CheckItem{offs: 7, cond: &CC::Eq(Arg::Byte(0))}], | |
336 | }, | |
337 | DetectConditions { | |
338 | demux_name: "smush", | |
339 | extensions: ".snm", | |
340 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"SANM")}, | |
341 | CheckItem{offs: 8, cond: &CC::Str(b"SHDR")}], | |
342 | }, | |
343 | DetectConditions { | |
344 | demux_name: "realaudio", | |
345 | extensions: ".ra,.ram", | |
346 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}], | |
347 | }, | |
348 | DetectConditions { | |
349 | demux_name: "realmedia", | |
350 | extensions: ".rm,.rmvb,.rma,.ra,.ram", | |
351 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) }, | |
352 | CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}], | |
353 | }, | |
354 | DetectConditions { | |
355 | demux_name: "real_ivr", | |
356 | extensions: ".ivr", | |
357 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}], | |
358 | }, | |
359 | DetectConditions { | |
360 | demux_name: "bink", | |
361 | extensions: ".bik,.bk2", | |
362 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x42494B62), // BIKb | |
363 | Arg::U32BE(0x42494B7B)), // BIKz | |
364 | &CC::In(Arg::U32BE(0x4B423261), // KB2a | |
365 | Arg::U32BE(0x4B42327B)))}], // KB2z | |
366 | }, | |
367 | DetectConditions { | |
368 | demux_name: "smacker", | |
369 | extensions: ".smk", | |
370 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}], | |
371 | }, | |
372 | DetectConditions { | |
373 | demux_name: "ape", | |
374 | extensions: ".ape", | |
375 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MAC ") }, | |
376 | CheckItem{offs: 4, cond: &CC::In(Arg::U16LE(3800), Arg::U16LE(3990))}], | |
377 | }, | |
378 | DetectConditions { | |
379 | demux_name: "flac", | |
380 | extensions: ".flac", | |
381 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"fLaC") }], | |
382 | }, | |
383 | DetectConditions { | |
384 | demux_name: "tta", | |
385 | extensions: ".tta", | |
386 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"TTA1") }], | |
387 | }, | |
388 | DetectConditions { | |
389 | demux_name: "wavpack", | |
390 | extensions: ".wv", | |
391 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"wvpk") }, | |
392 | CheckItem{offs: 8, cond: &CC::In(Arg::U16LE(0x402), Arg::U16LE(0x410))}], | |
393 | }, | |
394 | DetectConditions { | |
395 | demux_name: "vivo", | |
396 | extensions: ".viv", | |
397 | conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))}, | |
398 | CheckItem{offs: 2, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}], | |
399 | }, | |
400 | DetectConditions { | |
401 | demux_name: "vivo", | |
402 | extensions: ".viv", | |
403 | conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))}, | |
404 | CheckItem{offs: 3, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}], | |
405 | }, | |
406 | DetectConditions { | |
407 | demux_name: "bmv", | |
408 | extensions: ".bmv", | |
409 | conditions: &[], | |
410 | }, | |
411 | DetectConditions { | |
412 | demux_name: "bmv3", | |
413 | extensions: ".bmv", | |
414 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") }, | |
415 | CheckItem{offs: 32, cond: &CC::Str(b"DATA")}], | |
416 | }, | |
417 | DetectConditions { | |
418 | demux_name: "sga", | |
419 | extensions: ".dtv,.avc", | |
420 | conditions: &[], | |
421 | }, | |
422 | DetectConditions { | |
423 | demux_name: "vmd", | |
424 | extensions: ".vmd", | |
425 | conditions: &[], | |
426 | }, | |
427 | DetectConditions { | |
428 | demux_name: "vx", | |
429 | extensions: ".vx", | |
430 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"VXDS") }], | |
431 | }, | |
432 | ]; | |
433 | ||
434 | /// Tries to detect container format. | |
435 | /// | |
436 | /// This function tries to determine container format using both file extension and checking against container specific markers inside. | |
437 | /// In case of success the function returns short container name and the detection score. | |
438 | /// Result should have the highest detection score among tested. | |
439 | pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> { | |
440 | let mut result = None; | |
441 | let lname = name.to_lowercase(); | |
442 | for detector in DETECTORS { | |
443 | let mut score = DetectionScore::No; | |
444 | if !name.is_empty() { | |
445 | for ext in detector.extensions.split(',') { | |
446 | if lname.ends_with(ext) { | |
447 | score = DetectionScore::ExtensionMatches; | |
448 | break; | |
449 | } | |
450 | } | |
451 | } | |
452 | let mut passed = !detector.conditions.is_empty(); | |
453 | for ck in detector.conditions { | |
454 | let ret = src.seek(SeekFrom::Start(u64::from(ck.offs))); | |
455 | if ret.is_err() { | |
456 | passed = false; | |
457 | break; | |
458 | } | |
459 | if !ck.cond.eval(src) { | |
460 | passed = false; | |
461 | break; | |
462 | } | |
463 | } | |
464 | if passed { | |
465 | score = DetectionScore::MagicMatches; | |
466 | } | |
467 | if score == DetectionScore::MagicMatches { | |
468 | return Some((detector.demux_name, score)); | |
469 | } | |
470 | if result.is_none() && score != DetectionScore::No { | |
471 | result = Some((detector.demux_name, score)); | |
472 | } else if result.is_some() { | |
473 | let (_, oldsc) = result.unwrap(); | |
474 | if oldsc.less(score) { | |
475 | result = Some((detector.demux_name, score)); | |
476 | } | |
477 | } | |
478 | } | |
479 | result | |
480 | } | |
481 | ||
482 | /// Tries to detect container format for provided file name. | |
483 | pub fn detect_format_by_name(name: &str) -> Option<&'static str> { | |
484 | if name.is_empty() { | |
485 | return None; | |
486 | } | |
487 | let lname = name.to_lowercase(); | |
488 | for detector in DETECTORS { | |
489 | for ext in detector.extensions.split(',') { | |
490 | if lname.ends_with(ext) { | |
491 | return Some(detector.demux_name); | |
492 | } | |
493 | } | |
494 | } | |
495 | None | |
496 | } | |
497 | ||
498 | #[cfg(test)] | |
499 | mod test { | |
500 | use super::*; | |
501 | use std::fs::File; | |
502 | use nihav_core::io::byteio::*; | |
503 | ||
504 | #[test] | |
505 | fn test_avi_detect() { | |
506 | let name = "assets/Indeo/laser05.avi"; | |
507 | let mut file = File::open(name).unwrap(); | |
508 | let mut fr = FileReader::new_read(&mut file); | |
509 | let mut br = ByteReader::new(&mut fr); | |
510 | let (name, score) = detect_format(name, &mut br).unwrap(); | |
511 | assert_eq!(name, "avi"); | |
512 | assert_eq!(score, DetectionScore::MagicMatches); | |
513 | } | |
514 | ||
515 | #[test] | |
516 | fn test_gdv_detect() { | |
517 | let name = "assets/Game/intro1.gdv"; | |
518 | let mut file = File::open(name).unwrap(); | |
519 | let mut fr = FileReader::new_read(&mut file); | |
520 | let mut br = ByteReader::new(&mut fr); | |
521 | let (name, score) = detect_format(name, &mut br).unwrap(); | |
522 | assert_eq!(name, "gdv"); | |
523 | assert_eq!(score, DetectionScore::MagicMatches); | |
524 | } | |
525 | } |