]> git.nihav.org Git - nihav.git/blame - nihav-registry/src/detect.rs
avimux: do not record palette change chunks in OpenDML index
[nihav.git] / nihav-registry / src / detect.rs
CommitLineData
1b5ebe1d
KS
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
01f55c6a 10//! use nihav_registry::detect::detect_format;
1b5ebe1d
KS
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//! ```
d8ce0de0 23use std::io::SeekFrom;
32f7cbe5 24use nihav_core::io::byteio::ByteReader;
d8ce0de0 25
1b5ebe1d 26/// Format detection score.
d8ce0de0
KS
27#[derive(Debug,Clone,Copy,PartialEq)]
28pub enum DetectionScore {
1b5ebe1d 29 /// Format is not detected.
d8ce0de0 30 No,
1b5ebe1d 31 /// Format matched by file extension.
d8ce0de0 32 ExtensionMatches,
1b5ebe1d 33 /// Format matches by markers inside the file.
d8ce0de0
KS
34 MagicMatches,
35}
36
37impl DetectionScore {
1b5ebe1d 38 /// Checks whether current detection score is less than a value it is compared against.
e243ceb4
KS
39 pub fn less(self, other: DetectionScore) -> bool {
40 (self as i32) < (other as i32)
d8ce0de0
KS
41 }
42}
43
44#[allow(dead_code)]
b5bd2ae4 45enum Arg {
d8ce0de0
KS
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
b5bd2ae4 57impl Arg {
d8ce0de0
KS
58 fn val(&self) -> u64 {
59 match *self {
e243ceb4
KS
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) }
b5bd2ae4
KS
67 Arg::U64BE(v) => { v }
68 Arg::U64LE(v) => { v }
d8ce0de0
KS
69 }
70 }
71 fn read_val(&self, src: &mut ByteReader) -> Option<u64> {
72 match *self {
b5bd2ae4 73 Arg::Byte(_) => {
d8ce0de0 74 let res = src.peek_byte();
e243ceb4
KS
75 if res.is_err() { return None; }
76 Some(u64::from(res.unwrap()))
d8ce0de0 77 }
b5bd2ae4 78 Arg::U16BE(_) => {
d8ce0de0 79 let res = src.peek_u16be();
e243ceb4
KS
80 if res.is_err() { return None; }
81 Some(u64::from(res.unwrap()))
d8ce0de0 82 }
b5bd2ae4 83 Arg::U16LE(_) => {
d8ce0de0 84 let res = src.peek_u16le();
e243ceb4
KS
85 if res.is_err() { return None; }
86 Some(u64::from(res.unwrap()))
d8ce0de0 87 }
b5bd2ae4 88 Arg::U24BE(_) => {
d8ce0de0 89 let res = src.peek_u24be();
e243ceb4
KS
90 if res.is_err() { return None; }
91 Some(u64::from(res.unwrap()))
d8ce0de0 92 }
b5bd2ae4 93 Arg::U24LE(_) => {
d8ce0de0 94 let res = src.peek_u24le();
e243ceb4
KS
95 if res.is_err() { return None; }
96 Some(u64::from(res.unwrap()))
d8ce0de0 97 }
b5bd2ae4 98 Arg::U32BE(_) => {
d8ce0de0 99 let res = src.peek_u32be();
e243ceb4
KS
100 if res.is_err() { return None; }
101 Some(u64::from(res.unwrap()))
d8ce0de0 102 }
b5bd2ae4 103 Arg::U32LE(_) => {
d8ce0de0 104 let res = src.peek_u32le();
e243ceb4
KS
105 if res.is_err() { return None; }
106 Some(u64::from(res.unwrap()))
d8ce0de0 107 }
b5bd2ae4 108 Arg::U64BE(_) => {
d8ce0de0 109 let res = src.peek_u64be();
e243ceb4 110 if res.is_err() { return None; }
d8ce0de0
KS
111 Some(res.unwrap())
112 }
b5bd2ae4 113 Arg::U64LE(_) => {
d8ce0de0 114 let res = src.peek_u64le();
e243ceb4 115 if res.is_err() { return None; }
d8ce0de0
KS
116 Some(res.unwrap())
117 }
118 }
119 }
120 fn eq(&self, src: &mut ByteReader) -> bool {
01f55c6a
KS
121 if let Some(rval) = self.read_val(src) {
122 rval == self.val()
123 } else {
124 false
125 }
d8ce0de0
KS
126 }
127 fn ge(&self, src: &mut ByteReader) -> bool {
01f55c6a
KS
128 if let Some(rval) = self.read_val(src) {
129 rval >= self.val()
130 } else {
131 false
132 }
d8ce0de0
KS
133 }
134 fn gt(&self, src: &mut ByteReader) -> bool {
01f55c6a
KS
135 if let Some(rval) = self.read_val(src) {
136 rval > self.val()
137 } else {
138 false
139 }
d8ce0de0
KS
140 }
141 fn le(&self, src: &mut ByteReader) -> bool {
01f55c6a
KS
142 if let Some(rval) = self.read_val(src) {
143 rval <= self.val()
144 } else {
145 false
146 }
d8ce0de0
KS
147 }
148 fn lt(&self, src: &mut ByteReader) -> bool {
01f55c6a
KS
149 if let Some(rval) = self.read_val(src) {
150 rval < self.val()
151 } else {
152 false
153 }
d8ce0de0
KS
154 }
155}
156
157#[allow(dead_code)]
b5bd2ae4
KS
158enum 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),
d8ce0de0
KS
167}
168
b5bd2ae4 169impl<'a> CC<'a> {
d8ce0de0
KS
170 fn eval(&self, src: &mut ByteReader) -> bool {
171 match *self {
6f263099 172 CC::Or(a, b) => { a.eval(src) || b.eval(src) },
b5bd2ae4 173 CC::Eq(ref arg) => { arg.eq(src) },
4d477e23 174 CC::In(ref a, ref b) => { a.ge(src) && b.le(src) },
b5bd2ae4
KS
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) },
405cec9e
KS
179 CC::Str(strng) => {
180 let mut val: Vec<u8> = vec![0; strng.len()];
d8ce0de0 181 let res = src.peek_buf(val.as_mut_slice());
e243ceb4 182 if res.is_err() { return false; }
405cec9e 183 val == strng
d8ce0de0
KS
184 }
185 }
186 }
187}
188
189struct CheckItem<'a> {
190 offs: u32,
b5bd2ae4 191 cond: &'a CC<'a>,
d8ce0de0
KS
192}
193
194#[allow(dead_code)]
195struct DetectConditions<'a> {
196 demux_name: &'static str,
197 extensions: &'static str,
198 conditions: &'a [CheckItem<'a>],
199}
200
201const DETECTORS: &[DetectConditions] = &[
202 DetectConditions {
203 demux_name: "avi",
204 extensions: ".avi",
b5bd2ae4
KS
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 ]
d8ce0de0 210 },
283abfa6
KS
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 },
bdf66deb
KS
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 },
fc39649d
KS
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 },
bdf66deb
KS
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 },
88fd1059
KS
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 },
12b18d48
KS
248 DetectConditions {
249 demux_name: "mov-resfork",
250 extensions: ".mov",
251 conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32BE(0x100))},
252 CheckItem{offs: 0x108, cond: &CC::Str(b"moov")}],
253 },
dbb2cbc9
KS
254 DetectConditions {
255 demux_name: "yuv4mpeg",
256 extensions: ".y4m",
257 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"YUV4MPEG2 ") }],
258 },
2d6ff332
KS
259 DetectConditions {
260 demux_name: "armovie",
261 extensions: ".rpl",
262 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"ARMovie\n") }],
263 },
9383afcb
KS
264 DetectConditions {
265 demux_name: "tca",
266 extensions: ".tca",
267 conditions: &[CheckItem{offs: 0x00, cond: &CC::Str(b"ACEF") },
268 CheckItem{offs: 0x18, cond: &CC::Eq(Arg::U32LE(64))}],
269 },
92d9fb69
KS
270 DetectConditions {
271 demux_name: "flv",
272 extensions: ".flv",
273 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FLV") },
274 CheckItem{offs: 3, cond: &CC::Le(Arg::Byte(1)) }],
275 },
3a09aef3
KS
276 DetectConditions {
277 demux_name: "dvi",
278 extensions: ".avs,.dvi",
279 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"IVDV")},
280 CheckItem{offs: 12, cond: &CC::Str(b"SSVA")}],
281 },
e01d4987
KS
282 DetectConditions {
283 demux_name: "ivf",
284 extensions: ".ivf",
285 conditions: &[CheckItem{offs: 0, cond: &CC::Str(&[0x50, 0xEF, 0x81, 0x19, 0xB3, 0xBD, 0xD0, 0x11, 0xA3, 0xE5, 0x00, 0xA0, 0xC9, 0x24, 0x44])},
286 CheckItem{offs: 15, cond: &CC::Or(&CC::Eq(Arg::Byte(0x36)), &CC::Eq(Arg::Byte(0x37)))}],
287 },
fd7e6906 288 DetectConditions {
b8e71e0a 289 demux_name: "dkivf",
fd7e6906
KS
290 extensions: ".ivf",
291 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"DKIF\x00\x00")},
292 CheckItem{offs: 6, cond: &CC::Ge(Arg::U16LE(32))}],
293 },
d8ce0de0
KS
294 DetectConditions {
295 demux_name: "gdv",
296 extensions: ".gdv",
b5bd2ae4 297 conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}],
d8ce0de0 298 },
c17769db
KS
299 DetectConditions {
300 demux_name: "smush",
301 extensions: ".san",
302 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"ANIM")},
303 CheckItem{offs: 8, cond: &CC::Str(b"AHDR")}],
304 },
9e08bfdd
KS
305 DetectConditions {
306 demux_name: "smush-mcmp",
307 extensions: ".imc",
308 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MCMP")},
309 CheckItem{offs: 6, cond: &CC::Eq(Arg::Byte(0))},
310 CheckItem{offs: 7, cond: &CC::Eq(Arg::Byte(0))}],
311 },
c17769db
KS
312 DetectConditions {
313 demux_name: "smush",
314 extensions: ".snm",
315 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"SANM")},
316 CheckItem{offs: 8, cond: &CC::Str(b"SHDR")}],
317 },
bac0d858
KS
318 DetectConditions {
319 demux_name: "qpeg",
320 extensions: ".dvc",
321 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"IDVCd")}],
322 },
2d2f0127
KS
323 DetectConditions {
324 demux_name: "tealmov",
325 extensions: ".pdb",
326 conditions: &[CheckItem{offs: 0x3C, cond: &CC::Str(b"MvieTlMv")}],
327 },
ce52b3b5
KS
328 DetectConditions {
329 demux_name: "realaudio",
330 extensions: ".ra,.ram",
331 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}],
332 },
333 DetectConditions {
334 demux_name: "realmedia",
335 extensions: ".rm,.rmvb,.rma,.ra,.ram",
db5cc44b 336 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) },
ce52b3b5
KS
337 CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}],
338 },
339 DetectConditions {
340 demux_name: "real_ivr",
341 extensions: ".ivr",
342 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}],
343 },
c6c21059
KS
344 DetectConditions {
345 demux_name: "bink",
4d477e23 346 extensions: ".bik,.bk2",
4998874b
KS
347 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x42494B62), // BIKb
348 Arg::U32BE(0x42494B7B)), // BIKz
c6c21059
KS
349 &CC::In(Arg::U32BE(0x4B423261), // KB2a
350 Arg::U32BE(0x4B42327B)))}], // KB2z
351 },
606c448e
KS
352 DetectConditions {
353 demux_name: "smacker",
354 extensions: ".smk",
355 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}],
356 },
87927c57
KS
357 DetectConditions {
358 demux_name: "ape",
359 extensions: ".ape",
360 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MAC ") },
361 CheckItem{offs: 4, cond: &CC::In(Arg::U16LE(3800), Arg::U16LE(3990))}],
362 },
363 DetectConditions {
364 demux_name: "flac",
365 extensions: ".flac",
366 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"fLaC") }],
367 },
368 DetectConditions {
369 demux_name: "tta",
370 extensions: ".tta",
371 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"TTA1") }],
372 },
373 DetectConditions {
374 demux_name: "wavpack",
375 extensions: ".wv",
376 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"wvpk") },
377 CheckItem{offs: 8, cond: &CC::In(Arg::U16LE(0x402), Arg::U16LE(0x410))}],
378 },
128253cc 379 DetectConditions {
31cf33ac
KS
380 demux_name: "vivo",
381 extensions: ".viv",
382 conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
383 CheckItem{offs: 2, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
384 },
385 DetectConditions {
386 demux_name: "vivo",
387 extensions: ".viv",
388 conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
389 CheckItem{offs: 3, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
390 },
391 DetectConditions {
128253cc
KS
392 demux_name: "bmv",
393 extensions: ".bmv",
394 conditions: &[],
395 },
ecda1cc1
KS
396 DetectConditions {
397 demux_name: "bmv3",
398 extensions: ".bmv",
399 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") },
400 CheckItem{offs: 32, cond: &CC::Str(b"DATA")}],
401 },
fead60e3
KS
402 DetectConditions {
403 demux_name: "sga",
404 extensions: ".dtv,.avc",
405 conditions: &[],
406 },
f9fc73be
KS
407 DetectConditions {
408 demux_name: "sierra-seq",
409 extensions: ".seq",
410 conditions: &[],
411 },
9895bd7b
KS
412 DetectConditions {
413 demux_name: "vmd",
414 extensions: ".vmd",
415 conditions: &[],
416 },
d8ce0de0
KS
417];
418
1b5ebe1d
KS
419/// Tries to detect container format.
420///
421/// This function tries to determine container format using both file extension and checking against container specific markers inside.
422/// In case of success the function returns short container name and the detection score.
423/// Result should have the highest detection score among tested.
d8ce0de0
KS
424pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> {
425 let mut result = None;
426 let lname = name.to_lowercase();
427 for detector in DETECTORS {
428 let mut score = DetectionScore::No;
e243ceb4 429 if !name.is_empty() {
d8ce0de0
KS
430 for ext in detector.extensions.split(',') {
431 if lname.ends_with(ext) {
432 score = DetectionScore::ExtensionMatches;
433 break;
434 }
435 }
436 }
e243ceb4 437 let mut passed = !detector.conditions.is_empty();
d8ce0de0 438 for ck in detector.conditions {
e243ceb4
KS
439 let ret = src.seek(SeekFrom::Start(u64::from(ck.offs)));
440 if ret.is_err() {
d8ce0de0
KS
441 passed = false;
442 break;
443 }
444 if !ck.cond.eval(src) {
445 passed = false;
446 break;
447 }
448 }
449 if passed {
450 score = DetectionScore::MagicMatches;
451 }
452 if score == DetectionScore::MagicMatches {
453 return Some((detector.demux_name, score));
454 }
4d477e23 455 if result.is_none() && score != DetectionScore::No {
d8ce0de0 456 result = Some((detector.demux_name, score));
4d477e23 457 } else if result.is_some() {
d8ce0de0
KS
458 let (_, oldsc) = result.unwrap();
459 if oldsc.less(score) {
460 result = Some((detector.demux_name, score));
461 }
462 }
463 }
464 result
465}
466
4b56e8c9 467/// Tries to detect container format for provided file name.
ac818eac 468pub fn detect_format_by_name(name: &str) -> Option<&'static str> {
4b56e8c9
KS
469 if name.is_empty() {
470 return None;
471 }
472 let lname = name.to_lowercase();
473 for detector in DETECTORS {
474 for ext in detector.extensions.split(',') {
475 if lname.ends_with(ext) {
476 return Some(detector.demux_name);
477 }
478 }
479 }
480 None
481}
482
d8ce0de0
KS
483#[cfg(test)]
484mod test {
485 use super::*;
486 use std::fs::File;
01f55c6a 487 use nihav_core::io::byteio::*;
d8ce0de0
KS
488
489 #[test]
490 fn test_avi_detect() {
250c49f6 491 let name = "assets/Indeo/laser05.avi";
d8ce0de0
KS
492 let mut file = File::open(name).unwrap();
493 let mut fr = FileReader::new_read(&mut file);
494 let mut br = ByteReader::new(&mut fr);
495 let (name, score) = detect_format(name, &mut br).unwrap();
496 assert_eq!(name, "avi");
497 assert_eq!(score, DetectionScore::MagicMatches);
498 }
499
500 #[test]
501 fn test_gdv_detect() {
250c49f6 502 let name = "assets/Game/intro1.gdv";
d8ce0de0
KS
503 let mut file = File::open(name).unwrap();
504 let mut fr = FileReader::new_read(&mut file);
505 let mut br = ByteReader::new(&mut fr);
506 let (name, score) = detect_format(name, &mut br).unwrap();
507 assert_eq!(name, "gdv");
508 assert_eq!(score, DetectionScore::MagicMatches);
509 }
b5bd2ae4 510}