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 (ref a, ref 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: "ivf",
 240         extensions: ".ivf",
 241         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"DKIF\x00\x00")},
 242                       CheckItem{offs: 6, cond: &CC::Ge(Arg::U16LE(32))}],
 243     },
 244     DetectConditions {
 245         demux_name: "fcmp",
 246         extensions: ".cmp",
 247         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FCMP")}],
 248     },
 249     DetectConditions {
 250         demux_name: "fst",
 251         extensions: ".fst",
 252         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"2TSF")}],
 253     },
 254     DetectConditions {
 255         demux_name: "gdv",
 256         extensions: ".gdv",
 257         conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}],
 258     },
 259     DetectConditions {
 260         demux_name: "fable-imax",
 261         extensions: ".imx",
 262         conditions: &[CheckItem{offs:  0, cond: &CC::Str(b"IMAX") },
 263                       CheckItem{offs: 10, cond: &CC::Eq(Arg::U16LE(0x102)) }],
 264     },
 265     DetectConditions {
 266         demux_name: "legend-q",
 267         extensions: ".q",
 268         conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U16LE(0x6839))},
 269                       CheckItem{offs: 2, cond: &CC::In(Arg::Byte(3), Arg::Byte(5))}],
 270     },
 271     DetectConditions {
 272         demux_name: "realaudio",
 273         extensions: ".ra,.ram",
 274         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}],
 275     },
 276     DetectConditions {
 277         demux_name: "realmedia",
 278         extensions: ".rm,.rmvb,.rma,.ra,.ram",
 279         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) },
 280                       CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}],
 281     },
 282     DetectConditions {
 283         demux_name: "real_ivr",
 284         extensions: ".ivr",
 285         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}],
 286     },
 287     DetectConditions {
 288         demux_name: "bink",
 289         extensions: ".bik,.bk2",
 290         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x42494B62),     // BIKb
 291                                                                Arg::U32BE(0x42494B7B)),    // BIKz
 292                                                        &CC::In(Arg::U32BE(0x4B423261),     // KB2a
 293                                                                Arg::U32BE(0x4B42327B)))}], // KB2z
 294     },
 295     DetectConditions {
 296         demux_name: "smacker",
 297         extensions: ".smk",
 298         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}],
 299     },
 300     DetectConditions {
 301         demux_name: "ape",
 302         extensions: ".ape",
 303         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MAC ") },
 304                       CheckItem{offs: 4, cond: &CC::In(Arg::U16LE(3800), Arg::U16LE(3990))}],
 305     },
 306     DetectConditions {
 307         demux_name: "flac",
 308         extensions: ".flac",
 309         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"fLaC") }],
 310     },
 311     DetectConditions {
 312         demux_name: "tta",
 313         extensions: ".tta",
 314         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"TTA1") }],
 315     },
 316     DetectConditions {
 317         demux_name: "wavpack",
 318         extensions: ".wv",
 319         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"wvpk") },
 320                       CheckItem{offs: 8, cond: &CC::In(Arg::U16LE(0x402), Arg::U16LE(0x410))}],
 321     },
 322     DetectConditions {
 323         demux_name: "vivo",
 324         extensions: ".viv",
 325         conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
 326                       CheckItem{offs: 2, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
 327     },
 328     DetectConditions {
 329         demux_name: "vivo",
 330         extensions: ".viv",
 331         conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
 332                       CheckItem{offs: 3, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
 333     },
 334     DetectConditions {
 335         demux_name: "bmv",
 336         extensions: ".bmv",
 337         conditions: &[],
 338     },
 339     DetectConditions {
 340         demux_name: "bmv3",
 341         extensions: ".bmv",
 342         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") },
 343                       CheckItem{offs: 32, cond: &CC::Str(b"DATA")}],
 344     },
 345     DetectConditions {
 346         demux_name: "vmd",
 347         extensions: ".vmd",
 348         conditions: &[],
 349     },
 350     DetectConditions {
 351         demux_name: "vx",
 352         extensions: ".vx",
 353         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"VXDS") }],
 354     },
 355 ];
 356
 357 /// Tries to detect container format.
 358 ///
 359 /// This function tries to determine container format using both file extension and checking against container specific markers inside.
 360 /// In case of success the function returns short container name and the detection score.
 361 /// Result should have the highest detection score among tested.
 362 pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> {
 363     let mut result = None;
 364     let lname = name.to_lowercase();
 365     for detector in DETECTORS {
 366         let mut score = DetectionScore::No;
 367         if !name.is_empty() {
 368             for ext in detector.extensions.split(',') {
 369                 if lname.ends_with(ext) {
 370                     score = DetectionScore::ExtensionMatches;
 371                     break;
 372                 }
 373             }
 374         }
 375         let mut passed = !detector.conditions.is_empty();
 376         for ck in detector.conditions {
 377             let ret = src.seek(SeekFrom::Start(u64::from(ck.offs)));
 378             if ret.is_err() {
 379                 passed = false;
 380                 break;
 381             }
 382             if !ck.cond.eval(src) {
 383                 passed = false;
 384                 break;
 385             }
 386         }
 387         if passed {
 388             score = DetectionScore::MagicMatches;
 389         }
 390         if score == DetectionScore::MagicMatches {
 391             return Some((detector.demux_name, score));
 392         }
 393         if result.is_none() && score != DetectionScore::No {
 394             result = Some((detector.demux_name, score));
 395         } else if result.is_some() {
 396             let (_, oldsc) = result.unwrap();
 397             if oldsc.less(score) {
 398                 result = Some((detector.demux_name, score));
 399             }
 400         }
 401     }
 402     result
 403 }
 404
 405 /// Tries to detect container format for provided file name.
 406 pub fn detect_format_by_name(name: &str) -> Option<&'static str> {
 407     if name.is_empty() {
 408         return None;
 409     }
 410     let lname = name.to_lowercase();
 411     for detector in DETECTORS {
 412         for ext in detector.extensions.split(',') {
 413             if lname.ends_with(ext) {
 414                 return Some(detector.demux_name);
 415             }
 416         }
 417     }
 418     None
 419 }
 420
 421 #[cfg(test)]
 422 mod test {
 423     use super::*;
 424     use std::fs::File;
 425     use nihav_core::io::byteio::*;
 426
 427     #[test]
 428     fn test_avi_detect() {
 429         let name = "assets/Indeo/laser05.avi";
 430         let mut file = File::open(name).unwrap();
 431         let mut fr = FileReader::new_read(&mut file);
 432         let mut br = ByteReader::new(&mut fr);
 433         let (name, score) = detect_format(name, &mut br).unwrap();
 434         assert_eq!(name, "avi");
 435         assert_eq!(score, DetectionScore::MagicMatches);
 436     }
 437
 438     #[test]
 439     fn test_gdv_detect() {
 440         let name = "assets/Game/intro1.gdv";
 441         let mut file = File::open(name).unwrap();
 442         let mut fr = FileReader::new_read(&mut file);
 443         let mut br = ByteReader::new(&mut fr);
 444         let (name, score) = detect_format(name, &mut br).unwrap();
 445         assert_eq!(name, "gdv");
 446         assert_eq!(score, DetectionScore::MagicMatches);
 447     }
 448 }