]>
Commit | Line | Data |
---|---|---|
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 | |
10 | //! use nihav_core::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 | //! ``` | |
d8ce0de0 | 23 | use std::io::SeekFrom; |
32f7cbe5 | 24 | use nihav_core::io::byteio::ByteReader; |
d8ce0de0 | 25 | |
1b5ebe1d | 26 | /// Format detection score. |
d8ce0de0 KS |
27 | #[derive(Debug,Clone,Copy,PartialEq)] |
28 | pub 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 | ||
37 | impl 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 | 45 | enum 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 | 57 | impl 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 { | |
121 | let val = self.read_val(src); | |
e243ceb4 | 122 | if val.is_none() { false } |
d8ce0de0 KS |
123 | else { val.unwrap() == self.val() } |
124 | } | |
125 | fn ge(&self, src: &mut ByteReader) -> bool { | |
126 | let val = self.read_val(src); | |
e243ceb4 | 127 | if val.is_none() { false } |
d8ce0de0 KS |
128 | else { val.unwrap() >= self.val() } |
129 | } | |
130 | fn gt(&self, src: &mut ByteReader) -> bool { | |
131 | let val = self.read_val(src); | |
e243ceb4 | 132 | if val.is_none() { false } |
d8ce0de0 KS |
133 | else { val.unwrap() > self.val() } |
134 | } | |
135 | fn le(&self, src: &mut ByteReader) -> bool { | |
136 | let val = self.read_val(src); | |
e243ceb4 | 137 | if val.is_none() { false } |
d8ce0de0 KS |
138 | else { val.unwrap() <= self.val() } |
139 | } | |
140 | fn lt(&self, src: &mut ByteReader) -> bool { | |
141 | let val = self.read_val(src); | |
e243ceb4 | 142 | if val.is_none() { false } |
d8ce0de0 KS |
143 | else { val.unwrap() < self.val() } |
144 | } | |
145 | } | |
146 | ||
147 | #[allow(dead_code)] | |
b5bd2ae4 KS |
148 | enum CC<'a> { |
149 | Or(&'a CC<'a>, &'a CC<'a>), | |
150 | Eq(Arg), | |
151 | Str(&'static [u8]), | |
152 | In(Arg, Arg), | |
153 | Lt(Arg), | |
154 | Le(Arg), | |
155 | Gt(Arg), | |
156 | Ge(Arg), | |
d8ce0de0 KS |
157 | } |
158 | ||
b5bd2ae4 | 159 | impl<'a> CC<'a> { |
d8ce0de0 KS |
160 | fn eval(&self, src: &mut ByteReader) -> bool { |
161 | match *self { | |
b5bd2ae4 KS |
162 | CC::Or (ref a, ref b) => { a.eval(src) || b.eval(src) }, |
163 | CC::Eq(ref arg) => { arg.eq(src) }, | |
4d477e23 | 164 | CC::In(ref a, ref b) => { a.ge(src) && b.le(src) }, |
b5bd2ae4 KS |
165 | CC::Lt(ref arg) => { arg.lt(src) }, |
166 | CC::Le(ref arg) => { arg.le(src) }, | |
167 | CC::Gt(ref arg) => { arg.gt(src) }, | |
168 | CC::Ge(ref arg) => { arg.ge(src) }, | |
169 | CC::Str(str) => { | |
e243ceb4 | 170 | let mut val: Vec<u8> = vec![0; str.len()]; |
d8ce0de0 | 171 | let res = src.peek_buf(val.as_mut_slice()); |
e243ceb4 | 172 | if res.is_err() { return false; } |
d8ce0de0 KS |
173 | val == str |
174 | } | |
175 | } | |
176 | } | |
177 | } | |
178 | ||
179 | struct CheckItem<'a> { | |
180 | offs: u32, | |
b5bd2ae4 | 181 | cond: &'a CC<'a>, |
d8ce0de0 KS |
182 | } |
183 | ||
184 | #[allow(dead_code)] | |
185 | struct DetectConditions<'a> { | |
186 | demux_name: &'static str, | |
187 | extensions: &'static str, | |
188 | conditions: &'a [CheckItem<'a>], | |
189 | } | |
190 | ||
191 | const DETECTORS: &[DetectConditions] = &[ | |
192 | DetectConditions { | |
193 | demux_name: "avi", | |
194 | extensions: ".avi", | |
b5bd2ae4 KS |
195 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"RIFF"), &CC::Str(b"ON2 ")) }, |
196 | CheckItem{offs: 8, cond: &CC::Or(&CC::Or(&CC::Str(b"AVI LIST"), | |
197 | &CC::Str(b"AVIXLIST")), | |
198 | &CC::Str(b"ON2fLIST")) }, | |
199 | ] | |
d8ce0de0 KS |
200 | }, |
201 | DetectConditions { | |
202 | demux_name: "gdv", | |
203 | extensions: ".gdv", | |
b5bd2ae4 | 204 | conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}], |
d8ce0de0 | 205 | }, |
ce52b3b5 KS |
206 | DetectConditions { |
207 | demux_name: "realaudio", | |
208 | extensions: ".ra,.ram", | |
209 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}], | |
210 | }, | |
211 | DetectConditions { | |
212 | demux_name: "realmedia", | |
213 | extensions: ".rm,.rmvb,.rma,.ra,.ram", | |
db5cc44b | 214 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) }, |
ce52b3b5 KS |
215 | CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}], |
216 | }, | |
217 | DetectConditions { | |
218 | demux_name: "real_ivr", | |
219 | extensions: ".ivr", | |
220 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}], | |
221 | }, | |
c6c21059 KS |
222 | DetectConditions { |
223 | demux_name: "bink", | |
4d477e23 | 224 | extensions: ".bik,.bk2", |
4998874b KS |
225 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x42494B62), // BIKb |
226 | Arg::U32BE(0x42494B7B)), // BIKz | |
c6c21059 KS |
227 | &CC::In(Arg::U32BE(0x4B423261), // KB2a |
228 | Arg::U32BE(0x4B42327B)))}], // KB2z | |
229 | }, | |
606c448e KS |
230 | DetectConditions { |
231 | demux_name: "smacker", | |
232 | extensions: ".smk", | |
233 | conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}], | |
234 | }, | |
128253cc KS |
235 | DetectConditions { |
236 | demux_name: "bmv", | |
237 | extensions: ".bmv", | |
238 | conditions: &[], | |
239 | }, | |
ecda1cc1 KS |
240 | DetectConditions { |
241 | demux_name: "bmv3", | |
242 | extensions: ".bmv", | |
243 | conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") }, | |
244 | CheckItem{offs: 32, cond: &CC::Str(b"DATA")}], | |
245 | }, | |
9895bd7b KS |
246 | DetectConditions { |
247 | demux_name: "vmd", | |
248 | extensions: ".vmd", | |
249 | conditions: &[], | |
250 | }, | |
d8ce0de0 KS |
251 | ]; |
252 | ||
1b5ebe1d KS |
253 | /// Tries to detect container format. |
254 | /// | |
255 | /// This function tries to determine container format using both file extension and checking against container specific markers inside. | |
256 | /// In case of success the function returns short container name and the detection score. | |
257 | /// Result should have the highest detection score among tested. | |
d8ce0de0 KS |
258 | pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> { |
259 | let mut result = None; | |
260 | let lname = name.to_lowercase(); | |
261 | for detector in DETECTORS { | |
262 | let mut score = DetectionScore::No; | |
e243ceb4 | 263 | if !name.is_empty() { |
d8ce0de0 KS |
264 | for ext in detector.extensions.split(',') { |
265 | if lname.ends_with(ext) { | |
266 | score = DetectionScore::ExtensionMatches; | |
267 | break; | |
268 | } | |
269 | } | |
270 | } | |
e243ceb4 | 271 | let mut passed = !detector.conditions.is_empty(); |
d8ce0de0 | 272 | for ck in detector.conditions { |
e243ceb4 KS |
273 | let ret = src.seek(SeekFrom::Start(u64::from(ck.offs))); |
274 | if ret.is_err() { | |
d8ce0de0 KS |
275 | passed = false; |
276 | break; | |
277 | } | |
278 | if !ck.cond.eval(src) { | |
279 | passed = false; | |
280 | break; | |
281 | } | |
282 | } | |
283 | if passed { | |
284 | score = DetectionScore::MagicMatches; | |
285 | } | |
286 | if score == DetectionScore::MagicMatches { | |
287 | return Some((detector.demux_name, score)); | |
288 | } | |
4d477e23 | 289 | if result.is_none() && score != DetectionScore::No { |
d8ce0de0 | 290 | result = Some((detector.demux_name, score)); |
4d477e23 | 291 | } else if result.is_some() { |
d8ce0de0 KS |
292 | let (_, oldsc) = result.unwrap(); |
293 | if oldsc.less(score) { | |
294 | result = Some((detector.demux_name, score)); | |
295 | } | |
296 | } | |
297 | } | |
298 | result | |
299 | } | |
300 | ||
301 | #[cfg(test)] | |
302 | mod test { | |
303 | use super::*; | |
304 | use std::fs::File; | |
aca89041 | 305 | use crate::io::byteio::*; |
d8ce0de0 KS |
306 | |
307 | #[test] | |
308 | fn test_avi_detect() { | |
250c49f6 | 309 | let name = "assets/Indeo/laser05.avi"; |
d8ce0de0 KS |
310 | let mut file = File::open(name).unwrap(); |
311 | let mut fr = FileReader::new_read(&mut file); | |
312 | let mut br = ByteReader::new(&mut fr); | |
313 | let (name, score) = detect_format(name, &mut br).unwrap(); | |
314 | assert_eq!(name, "avi"); | |
315 | assert_eq!(score, DetectionScore::MagicMatches); | |
316 | } | |
317 | ||
318 | #[test] | |
319 | fn test_gdv_detect() { | |
250c49f6 | 320 | let name = "assets/Game/intro1.gdv"; |
d8ce0de0 KS |
321 | let mut file = File::open(name).unwrap(); |
322 | let mut fr = FileReader::new_read(&mut file); | |
323 | let mut br = ByteReader::new(&mut fr); | |
324 | let (name, score) = detect_format(name, &mut br).unwrap(); | |
325 | assert_eq!(name, "gdv"); | |
326 | assert_eq!(score, DetectionScore::MagicMatches); | |
327 | } | |
b5bd2ae4 | 328 | } |