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_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 //! ```
  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         let val = self.read_val(src);
 122         if val.is_none() { false }
 123         else { val.unwrap() == self.val() }
 124     }
 125     fn ge(&self, src: &mut ByteReader) -> bool {
 126         let val = self.read_val(src);
 127         if val.is_none() { false }
 128         else { val.unwrap() >= self.val() }
 129     }
 130     fn gt(&self, src: &mut ByteReader) -> bool {
 131         let val = self.read_val(src);
 132         if val.is_none() { false }
 133         else { val.unwrap() > self.val() }
 134     }
 135     fn le(&self, src: &mut ByteReader) -> bool {
 136         let val = self.read_val(src);
 137         if val.is_none() { false }
 138         else { val.unwrap() <= self.val() }
 139     }
 140     fn lt(&self, src: &mut ByteReader) -> bool {
 141         let val = self.read_val(src);
 142         if val.is_none() { false }
 143         else { val.unwrap() < self.val() }
 144     }
 145 }
 146
 147 #[allow(dead_code)]
 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),
 157 }
 158
 159 impl<'a> CC<'a> {
 160     fn eval(&self, src: &mut ByteReader) -> bool {
 161         match *self {
 162             CC::Or (ref a, ref b) => { a.eval(src) || b.eval(src) },
 163             CC::Eq(ref arg)      => { arg.eq(src) },
 164             CC::In(ref a, ref b) => { a.ge(src) && b.le(src) },
 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) => {
 170                 let mut val: Vec<u8> = vec![0; str.len()];
 171                 let res = src.peek_buf(val.as_mut_slice());
 172                 if res.is_err() { return false; }
 173                 val == str
 174             }
 175         }
 176     }
 177 }
 178
 179 struct CheckItem<'a> {
 180     offs: u32,
 181     cond: &'a CC<'a>,
 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",
 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                      ]
 200     },
 201     DetectConditions {
 202         demux_name: "wav",
 203         extensions: ".wav",
 204         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"RIFF") },
 205                       CheckItem{offs: 8, cond: &CC::Str(b"WAVEfmt ") }
 206                      ]
 207     },
 208     DetectConditions {
 209         demux_name: "mov",
 210         extensions: ".mov",
 211         conditions: &[CheckItem{offs: 4, cond: &CC::Or(&CC::Or(&CC::Str(b"mdat"),
 212                                                                &CC::Str(b"moov")),
 213                                                                &CC::Str(b"ftyp")) }],
 214     },
 215     DetectConditions {
 216         demux_name: "mov",
 217         extensions: ".mov",
 218         conditions: &[CheckItem{offs:  0, cond: &CC::Str(b"\x00\x00\x00\x08wide") },
 219                       CheckItem{offs: 12, cond: &CC::Or(&CC::Or(&CC::Str(b"mdat"),
 220                                                                 &CC::Str(b"moov")),
 221                                                                 &CC::Str(b"ftyp")) }],
 222     },
 223     DetectConditions {
 224         demux_name: "gdv",
 225         extensions: ".gdv",
 226         conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}],
 227     },
 228     DetectConditions {
 229         demux_name: "realaudio",
 230         extensions: ".ra,.ram",
 231         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}],
 232     },
 233     DetectConditions {
 234         demux_name: "realmedia",
 235         extensions: ".rm,.rmvb,.rma,.ra,.ram",
 236         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) },
 237                       CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}],
 238     },
 239     DetectConditions {
 240         demux_name: "real_ivr",
 241         extensions: ".ivr",
 242         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}],
 243     },
 244     DetectConditions {
 245         demux_name: "bink",
 246         extensions: ".bik,.bk2",
 247         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x42494B62),     // BIKb
 248                                                                Arg::U32BE(0x42494B7B)),    // BIKz
 249                                                        &CC::In(Arg::U32BE(0x4B423261),     // KB2a
 250                                                                Arg::U32BE(0x4B42327B)))}], // KB2z
 251     },
 252     DetectConditions {
 253         demux_name: "smacker",
 254         extensions: ".smk",
 255         conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}],
 256     },
 257     DetectConditions {
 258         demux_name: "ape",
 259         extensions: ".ape",
 260         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MAC ") },
 261                       CheckItem{offs: 4, cond: &CC::In(Arg::U16LE(3800), Arg::U16LE(3990))}],
 262     },
 263     DetectConditions {
 264         demux_name: "flac",
 265         extensions: ".flac",
 266         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"fLaC") }],
 267     },
 268     DetectConditions {
 269         demux_name: "tta",
 270         extensions: ".tta",
 271         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"TTA1") }],
 272     },
 273     DetectConditions {
 274         demux_name: "wavpack",
 275         extensions: ".wv",
 276         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"wvpk") },
 277                       CheckItem{offs: 8, cond: &CC::In(Arg::U16LE(0x402), Arg::U16LE(0x410))}],
 278     },
 279     DetectConditions {
 280         demux_name: "vivo",
 281         extensions: ".viv",
 282         conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
 283                       CheckItem{offs: 2, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
 284     },
 285     DetectConditions {
 286         demux_name: "vivo",
 287         extensions: ".viv",
 288         conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
 289                       CheckItem{offs: 3, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
 290     },
 291     DetectConditions {
 292         demux_name: "bmv",
 293         extensions: ".bmv",
 294         conditions: &[],
 295     },
 296     DetectConditions {
 297         demux_name: "bmv3",
 298         extensions: ".bmv",
 299         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") },
 300                       CheckItem{offs: 32, cond: &CC::Str(b"DATA")}],
 301     },
 302     DetectConditions {
 303         demux_name: "vmd",
 304         extensions: ".vmd",
 305         conditions: &[],
 306     },
 307     DetectConditions {
 308         demux_name: "vx",
 309         extensions: ".vx",
 310         conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"VXDS") }],
 311     },
 312 ];
 313
 314 /// Tries to detect container format.
 315 ///
 316 /// This function tries to determine container format using both file extension and checking against container specific markers inside.
 317 /// In case of success the function returns short container name and the detection score.
 318 /// Result should have the highest detection score among tested.
 319 pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> {
 320     let mut result = None;
 321     let lname = name.to_lowercase();
 322     for detector in DETECTORS {
 323         let mut score = DetectionScore::No;
 324         if !name.is_empty() {
 325             for ext in detector.extensions.split(',') {
 326                 if lname.ends_with(ext) {
 327                     score = DetectionScore::ExtensionMatches;
 328                     break;
 329                 }
 330             }
 331         }
 332         let mut passed = !detector.conditions.is_empty();
 333         for ck in detector.conditions {
 334             let ret = src.seek(SeekFrom::Start(u64::from(ck.offs)));
 335             if ret.is_err() {
 336                 passed = false;
 337                 break;
 338             }
 339             if !ck.cond.eval(src) {
 340                 passed = false;
 341                 break;
 342             }
 343         }
 344         if passed {
 345             score = DetectionScore::MagicMatches;
 346         }
 347         if score == DetectionScore::MagicMatches {
 348             return Some((detector.demux_name, score));
 349         }
 350         if result.is_none() && score != DetectionScore::No {
 351             result = Some((detector.demux_name, score));
 352         } else if result.is_some() {
 353             let (_, oldsc) = result.unwrap();
 354             if oldsc.less(score) {
 355                 result = Some((detector.demux_name, score));
 356             }
 357         }
 358     }
 359     result
 360 }
 361
 362 /// Tries to detect container format for provided file name.
 363 pub fn detect_format_by_name(name: &str) -> Option<(&'static str)> {
 364     if name.is_empty() {
 365         return None;
 366     }
 367     let lname = name.to_lowercase();
 368     for detector in DETECTORS {
 369         for ext in detector.extensions.split(',') {
 370             if lname.ends_with(ext) {
 371                 return Some(detector.demux_name);
 372             }
 373         }
 374     }
 375     None
 376 }
 377
 378 #[cfg(test)]
 379 mod test {
 380     use super::*;
 381     use std::fs::File;
 382     use crate::io::byteio::*;
 383
 384     #[test]
 385     fn test_avi_detect() {
 386         let name = "assets/Indeo/laser05.avi";
 387         let mut file = File::open(name).unwrap();
 388         let mut fr = FileReader::new_read(&mut file);
 389         let mut br = ByteReader::new(&mut fr);
 390         let (name, score) = detect_format(name, &mut br).unwrap();
 391         assert_eq!(name, "avi");
 392         assert_eq!(score, DetectionScore::MagicMatches);
 393     }
 394
 395     #[test]
 396     fn test_gdv_detect() {
 397         let name = "assets/Game/intro1.gdv";
 398         let mut file = File::open(name).unwrap();
 399         let mut fr = FileReader::new_read(&mut file);
 400         let mut br = ByteReader::new(&mut fr);
 401         let (name, score) = detect_format(name, &mut br).unwrap();
 402         assert_eq!(name, "gdv");
 403         assert_eq!(score, DetectionScore::MagicMatches);
 404     }
 405 }