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: "mov-macbin",
 235         extensions: ".mov,.bin",
 236         conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::Byte(0))},
 237                       CheckItem{offs: 0x41, cond: &CC::Str(b"MooV")},
 238                       CheckItem{offs: 0x7A, cond: &CC::Eq(Arg::Byte(0x81))},
 239                       CheckItem{offs: 0x7B, cond: &CC::Eq(Arg::Byte(0x81))},
 240                       CheckItem{offs: 0x84, cond: &CC::Str(b"mdat")}],
 241     },
 242     DetectConditions {
 243         demux_name: "yuv4mpeg",
 244         extensions: ".y4m",
 245         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"YUV4MPEG2 ") }],
 246     },
 247     DetectConditions {
 248         demux_name: "flv",
 249         extensions: ".flv",
 250         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FLV") },
 251                       CheckItem{offs: 3, cond: &CC::Le(Arg::Byte(1)) }],
 252     },
 253     DetectConditions {
 254         demux_name: "ivf",
 255         extensions: ".ivf",
 256         conditions: &[CheckItem{offs: 0, cond: &CC::Str(&[0x50, 0xEF, 0x81, 0x19, 0xB3, 0xBD, 0xD0, 0x11, 0xA3, 0xE5, 0x00, 0xA0, 0xC9, 0x24, 0x44])},
 257                       CheckItem{offs: 15, cond: &CC::Or(&CC::Eq(Arg::Byte(0x36)), &CC::Eq(Arg::Byte(0x37)))}],
 258     },
 259     DetectConditions {
 260         demux_name: "dkivf",
 261         extensions: ".ivf",
 262         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"DKIF\x00\x00")},
 263                       CheckItem{offs: 6, cond: &CC::Ge(Arg::U16LE(32))}],
 264     },
 265     DetectConditions {
 266         demux_name: "arxel-cnm",
 267         extensions: ".cnm",
 268         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"CNM UNR\x00")}],
 269     },
 270     DetectConditions {
 271         demux_name: "fcmp",
 272         extensions: ".cmp",
 273         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FCMP")}],
 274     },
 275     DetectConditions {
 276         demux_name: "fst",
 277         extensions: ".fst",
 278         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"2TSF")}],
 279     },
 280     DetectConditions {
 281         demux_name: "gdv",
 282         extensions: ".gdv",
 283         conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}],
 284     },
 285     DetectConditions {
 286         demux_name: "fable-imax",
 287         extensions: ".imx",
 288         conditions: &[CheckItem{offs:  0, cond: &CC::Str(b"IMAX") },
 289                       CheckItem{offs: 10, cond: &CC::Eq(Arg::U16LE(0x102)) }],
 290     },
 291     DetectConditions {
 292         demux_name: "hl-fmv",
 293         extensions: ".fmv",
 294         conditions: &[CheckItem{offs:  0, cond: &CC::Str(b"FMV*") },
 295                       CheckItem{offs:  4, cond: &CC::Eq(Arg::U32LE(0)) }],
 296     },
 297     DetectConditions {
 298         demux_name: "legend-q",
 299         extensions: ".q",
 300         conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U16LE(0x6839))},
 301                       CheckItem{offs: 2, cond: &CC::In(Arg::Byte(3), Arg::Byte(7))}],
 302     },
 303     DetectConditions {
 304         demux_name: "siff",
 305         extensions: ".vb,.vbc,.fcp,.son",
 306         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"SIFF")},
 307                       CheckItem{offs: 4, cond: &CC::Or(
 308                                     &CC::Or(
 309                                         &CC::Str(b"VBV1VBHD"),
 310                                         &CC::Str(b"SOUNSHDR")),
 311                                     &CC::Str(b"FCPKFCHD"))}],
 312     },
 313     DetectConditions {
 314         demux_name: "smush",
 315         extensions: ".san",
 316         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"ANIM")},
 317                       CheckItem{offs: 8, cond: &CC::Str(b"AHDR")}],
 318     },
 319     DetectConditions {
 320         demux_name: "smush-mcmp",
 321         extensions: ".imc",
 322         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MCMP")},
 323                       CheckItem{offs: 6, cond: &CC::Eq(Arg::Byte(0))},
 324                       CheckItem{offs: 7, cond: &CC::Eq(Arg::Byte(0))}],
 325     },
 326     DetectConditions {
 327         demux_name: "smush",
 328         extensions: ".snm",
 329         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"SANM")},
 330                       CheckItem{offs: 8, cond: &CC::Str(b"SHDR")}],
 331     },
 332     DetectConditions {
 333         demux_name: "realaudio",
 334         extensions: ".ra,.ram",
 335         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}],
 336     },
 337     DetectConditions {
 338         demux_name: "realmedia",
 339         extensions: ".rm,.rmvb,.rma,.ra,.ram",
 340         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) },
 341                       CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}],
 342     },
 343     DetectConditions {
 344         demux_name: "real_ivr",
 345         extensions: ".ivr",
 346         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}],
 347     },
 348     DetectConditions {
 349         demux_name: "bink",
 350         extensions: ".bik,.bk2",
 351         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x42494B62),     // BIKb
 352                                                                Arg::U32BE(0x42494B7B)),    // BIKz
 353                                                        &CC::In(Arg::U32BE(0x4B423261),     // KB2a
 354                                                                Arg::U32BE(0x4B42327B)))}], // KB2z
 355     },
 356     DetectConditions {
 357         demux_name: "smacker",
 358         extensions: ".smk",
 359         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}],
 360     },
 361     DetectConditions {
 362         demux_name: "ape",
 363         extensions: ".ape",
 364         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MAC ") },
 365                       CheckItem{offs: 4, cond: &CC::In(Arg::U16LE(3800), Arg::U16LE(3990))}],
 366     },
 367     DetectConditions {
 368         demux_name: "flac",
 369         extensions: ".flac",
 370         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"fLaC") }],
 371     },
 372     DetectConditions {
 373         demux_name: "tta",
 374         extensions: ".tta",
 375         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"TTA1") }],
 376     },
 377     DetectConditions {
 378         demux_name: "wavpack",
 379         extensions: ".wv",
 380         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"wvpk") },
 381                       CheckItem{offs: 8, cond: &CC::In(Arg::U16LE(0x402), Arg::U16LE(0x410))}],
 382     },
 383     DetectConditions {
 384         demux_name: "vivo",
 385         extensions: ".viv",
 386         conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
 387                       CheckItem{offs: 2, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
 388     },
 389     DetectConditions {
 390         demux_name: "vivo",
 391         extensions: ".viv",
 392         conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
 393                       CheckItem{offs: 3, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
 394     },
 395     DetectConditions {
 396         demux_name: "bmv",
 397         extensions: ".bmv",
 398         conditions: &[],
 399     },
 400     DetectConditions {
 401         demux_name: "bmv3",
 402         extensions: ".bmv",
 403         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") },
 404                       CheckItem{offs: 32, cond: &CC::Str(b"DATA")}],
 405     },
 406     DetectConditions {
 407         demux_name: "vmd",
 408         extensions: ".vmd",
 409         conditions: &[],
 410     },
 411     DetectConditions {
 412         demux_name: "vx",
 413         extensions: ".vx",
 414         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"VXDS") }],
 415     },
 416 ];
 417
 418 /// Tries to detect container format.
 419 ///
 420 /// This function tries to determine container format using both file extension and checking against container specific markers inside.
 421 /// In case of success the function returns short container name and the detection score.
 422 /// Result should have the highest detection score among tested.
 423 pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> {
 424     let mut result = None;
 425     let lname = name.to_lowercase();
 426     for detector in DETECTORS {
 427         let mut score = DetectionScore::No;
 428         if !name.is_empty() {
 429             for ext in detector.extensions.split(',') {
 430                 if lname.ends_with(ext) {
 431                     score = DetectionScore::ExtensionMatches;
 432                     break;
 433                 }
 434             }
 435         }
 436         let mut passed = !detector.conditions.is_empty();
 437         for ck in detector.conditions {
 438             let ret = src.seek(SeekFrom::Start(u64::from(ck.offs)));
 439             if ret.is_err() {
 440                 passed = false;
 441                 break;
 442             }
 443             if !ck.cond.eval(src) {
 444                 passed = false;
 445                 break;
 446             }
 447         }
 448         if passed {
 449             score = DetectionScore::MagicMatches;
 450         }
 451         if score == DetectionScore::MagicMatches {
 452             return Some((detector.demux_name, score));
 453         }
 454         if result.is_none() && score != DetectionScore::No {
 455             result = Some((detector.demux_name, score));
 456         } else if result.is_some() {
 457             let (_, oldsc) = result.unwrap();
 458             if oldsc.less(score) {
 459                 result = Some((detector.demux_name, score));
 460             }
 461         }
 462     }
 463     result
 464 }
 465
 466 /// Tries to detect container format for provided file name.
 467 pub fn detect_format_by_name(name: &str) -> Option<&'static str> {
 468     if name.is_empty() {
 469         return None;
 470     }
 471     let lname = name.to_lowercase();
 472     for detector in DETECTORS {
 473         for ext in detector.extensions.split(',') {
 474             if lname.ends_with(ext) {
 475                 return Some(detector.demux_name);
 476             }
 477         }
 478     }
 479     None
 480 }
 481
 482 #[cfg(test)]
 483 mod test {
 484     use super::*;
 485     use std::fs::File;
 486     use nihav_core::io::byteio::*;
 487
 488     #[test]
 489     fn test_avi_detect() {
 490         let name = "assets/Indeo/laser05.avi";
 491         let mut file = File::open(name).unwrap();
 492         let mut fr = FileReader::new_read(&mut file);
 493         let mut br = ByteReader::new(&mut fr);
 494         let (name, score) = detect_format(name, &mut br).unwrap();
 495         assert_eq!(name, "avi");
 496         assert_eq!(score, DetectionScore::MagicMatches);
 497     }
 498
 499     #[test]
 500     fn test_gdv_detect() {
 501         let name = "assets/Game/intro1.gdv";
 502         let mut file = File::open(name).unwrap();
 503         let mut fr = FileReader::new_read(&mut file);
 504         let mut br = ByteReader::new(&mut fr);
 505         let (name, score) = detect_format(name, &mut br).unwrap();
 506         assert_eq!(name, "gdv");
 507         assert_eq!(score, DetectionScore::MagicMatches);
 508     }
 509 }