dd095d436697cc6398466cfc21b0d8be448abe74
[nihav.git] / nihav-registry / src / detect.rs
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(str) => {
180 let mut val: Vec<u8> = vec![0; str.len()];
181 let res = src.peek_buf(val.as_mut_slice());
182 if res.is_err() { return false; }
183 val == str
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: "mov",
227 extensions: ".mov",
228 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"\x00\x00\x00\x08wide") },
229 CheckItem{offs: 12, cond: &CC::Or(&CC::Or(&CC::Str(b"mdat"),
230 &CC::Str(b"moov")),
231 &CC::Str(b"ftyp")) }],
232 },
233 DetectConditions {
234 demux_name: "yuv4mpeg",
235 extensions: ".y4m",
236 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"YUV4MPEG2 ") }],
237 },
238 DetectConditions {
239 demux_name: "flv",
240 extensions: ".flv",
241 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FLV") },
242 CheckItem{offs: 3, cond: &CC::Le(Arg::Byte(1)) }],
243 },
244 DetectConditions {
245 demux_name: "ivf",
246 extensions: ".ivf",
247 conditions: &[CheckItem{offs: 0, cond: &CC::Str(&[0x50, 0xEF, 0x81, 0x19, 0xB3, 0xBD, 0xD0, 0x11, 0xA3, 0xE5, 0x00, 0xA0, 0xC9, 0x24, 0x44])},
248 CheckItem{offs: 15, cond: &CC::Or(&CC::Eq(Arg::Byte(0x36)), &CC::Eq(Arg::Byte(0x37)))}],
249 },
250 DetectConditions {
251 demux_name: "dkivf",
252 extensions: ".ivf",
253 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"DKIF\x00\x00")},
254 CheckItem{offs: 6, cond: &CC::Ge(Arg::U16LE(32))}],
255 },
256 DetectConditions {
257 demux_name: "fcmp",
258 extensions: ".cmp",
259 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FCMP")}],
260 },
261 DetectConditions {
262 demux_name: "fst",
263 extensions: ".fst",
264 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"2TSF")}],
265 },
266 DetectConditions {
267 demux_name: "gdv",
268 extensions: ".gdv",
269 conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}],
270 },
271 DetectConditions {
272 demux_name: "fable-imax",
273 extensions: ".imx",
274 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"IMAX") },
275 CheckItem{offs: 10, cond: &CC::Eq(Arg::U16LE(0x102)) }],
276 },
277 DetectConditions {
278 demux_name: "hl-fmv",
279 extensions: ".fmv",
280 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FMV*") },
281 CheckItem{offs: 4, cond: &CC::Eq(Arg::U32LE(0)) }],
282 },
283 DetectConditions {
284 demux_name: "legend-q",
285 extensions: ".q",
286 conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U16LE(0x6839))},
287 CheckItem{offs: 2, cond: &CC::In(Arg::Byte(3), Arg::Byte(7))}],
288 },
289 DetectConditions {
290 demux_name: "siff",
291 extensions: ".vb,.vbc,.fcp,.son",
292 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"SIFF")},
293 CheckItem{offs: 4, cond: &CC::Or(
294 &CC::Or(
295 &CC::Str(b"VBV1VBHD"),
296 &CC::Str(b"SOUNSHDR")),
297 &CC::Str(b"FCPKFCHD"))}],
298 },
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 },
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 },
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 },
318 DetectConditions {
319 demux_name: "realaudio",
320 extensions: ".ra,.ram",
321 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}],
322 },
323 DetectConditions {
324 demux_name: "realmedia",
325 extensions: ".rm,.rmvb,.rma,.ra,.ram",
326 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) },
327 CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}],
328 },
329 DetectConditions {
330 demux_name: "real_ivr",
331 extensions: ".ivr",
332 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}],
333 },
334 DetectConditions {
335 demux_name: "bink",
336 extensions: ".bik,.bk2",
337 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x42494B62), // BIKb
338 Arg::U32BE(0x42494B7B)), // BIKz
339 &CC::In(Arg::U32BE(0x4B423261), // KB2a
340 Arg::U32BE(0x4B42327B)))}], // KB2z
341 },
342 DetectConditions {
343 demux_name: "smacker",
344 extensions: ".smk",
345 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}],
346 },
347 DetectConditions {
348 demux_name: "ape",
349 extensions: ".ape",
350 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MAC ") },
351 CheckItem{offs: 4, cond: &CC::In(Arg::U16LE(3800), Arg::U16LE(3990))}],
352 },
353 DetectConditions {
354 demux_name: "flac",
355 extensions: ".flac",
356 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"fLaC") }],
357 },
358 DetectConditions {
359 demux_name: "tta",
360 extensions: ".tta",
361 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"TTA1") }],
362 },
363 DetectConditions {
364 demux_name: "wavpack",
365 extensions: ".wv",
366 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"wvpk") },
367 CheckItem{offs: 8, cond: &CC::In(Arg::U16LE(0x402), Arg::U16LE(0x410))}],
368 },
369 DetectConditions {
370 demux_name: "vivo",
371 extensions: ".viv",
372 conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
373 CheckItem{offs: 2, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
374 },
375 DetectConditions {
376 demux_name: "vivo",
377 extensions: ".viv",
378 conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
379 CheckItem{offs: 3, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
380 },
381 DetectConditions {
382 demux_name: "bmv",
383 extensions: ".bmv",
384 conditions: &[],
385 },
386 DetectConditions {
387 demux_name: "bmv3",
388 extensions: ".bmv",
389 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") },
390 CheckItem{offs: 32, cond: &CC::Str(b"DATA")}],
391 },
392 DetectConditions {
393 demux_name: "vmd",
394 extensions: ".vmd",
395 conditions: &[],
396 },
397 DetectConditions {
398 demux_name: "vx",
399 extensions: ".vx",
400 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"VXDS") }],
401 },
402 ];
403
404 /// Tries to detect container format.
405 ///
406 /// This function tries to determine container format using both file extension and checking against container specific markers inside.
407 /// In case of success the function returns short container name and the detection score.
408 /// Result should have the highest detection score among tested.
409 pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> {
410 let mut result = None;
411 let lname = name.to_lowercase();
412 for detector in DETECTORS {
413 let mut score = DetectionScore::No;
414 if !name.is_empty() {
415 for ext in detector.extensions.split(',') {
416 if lname.ends_with(ext) {
417 score = DetectionScore::ExtensionMatches;
418 break;
419 }
420 }
421 }
422 let mut passed = !detector.conditions.is_empty();
423 for ck in detector.conditions {
424 let ret = src.seek(SeekFrom::Start(u64::from(ck.offs)));
425 if ret.is_err() {
426 passed = false;
427 break;
428 }
429 if !ck.cond.eval(src) {
430 passed = false;
431 break;
432 }
433 }
434 if passed {
435 score = DetectionScore::MagicMatches;
436 }
437 if score == DetectionScore::MagicMatches {
438 return Some((detector.demux_name, score));
439 }
440 if result.is_none() && score != DetectionScore::No {
441 result = Some((detector.demux_name, score));
442 } else if result.is_some() {
443 let (_, oldsc) = result.unwrap();
444 if oldsc.less(score) {
445 result = Some((detector.demux_name, score));
446 }
447 }
448 }
449 result
450 }
451
452 /// Tries to detect container format for provided file name.
453 pub fn detect_format_by_name(name: &str) -> Option<&'static str> {
454 if name.is_empty() {
455 return None;
456 }
457 let lname = name.to_lowercase();
458 for detector in DETECTORS {
459 for ext in detector.extensions.split(',') {
460 if lname.ends_with(ext) {
461 return Some(detector.demux_name);
462 }
463 }
464 }
465 None
466 }
467
468 #[cfg(test)]
469 mod test {
470 use super::*;
471 use std::fs::File;
472 use nihav_core::io::byteio::*;
473
474 #[test]
475 fn test_avi_detect() {
476 let name = "assets/Indeo/laser05.avi";
477 let mut file = File::open(name).unwrap();
478 let mut fr = FileReader::new_read(&mut file);
479 let mut br = ByteReader::new(&mut fr);
480 let (name, score) = detect_format(name, &mut br).unwrap();
481 assert_eq!(name, "avi");
482 assert_eq!(score, DetectionScore::MagicMatches);
483 }
484
485 #[test]
486 fn test_gdv_detect() {
487 let name = "assets/Game/intro1.gdv";
488 let mut file = File::open(name).unwrap();
489 let mut fr = FileReader::new_read(&mut file);
490 let mut br = ByteReader::new(&mut fr);
491 let (name, score) = detect_format(name, &mut br).unwrap();
492 assert_eq!(name, "gdv");
493 assert_eq!(score, DetectionScore::MagicMatches);
494 }
495 }