1 //! Container format detection.
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.
10 //! use nihav_core::detect::detect_format;
11 //! use std::fs::File;
12 //! use nihav_core::io::byteio::*;
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);
23 use std::io::SeekFrom;
24 use nihav_core::io::byteio::ByteReader;
26 /// Format detection score.
27 #[derive(Debug,Clone,Copy,PartialEq)]
28 pub enum DetectionScore {
29 /// Format is not detected.
31 /// Format matched by file extension.
33 /// Format matches by markers inside the file.
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)
58 fn val(&self) -> u64 {
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 }
71 fn read_val(&self, src: &mut ByteReader) -> Option<u64> {
74 let res = src.peek_byte();
75 if res.is_err() { return None; }
76 Some(u64::from(res.unwrap()))
79 let res = src.peek_u16be();
80 if res.is_err() { return None; }
81 Some(u64::from(res.unwrap()))
84 let res = src.peek_u16le();
85 if res.is_err() { return None; }
86 Some(u64::from(res.unwrap()))
89 let res = src.peek_u24be();
90 if res.is_err() { return None; }
91 Some(u64::from(res.unwrap()))
94 let res = src.peek_u24le();
95 if res.is_err() { return None; }
96 Some(u64::from(res.unwrap()))
99 let res = src.peek_u32be();
100 if res.is_err() { return None; }
101 Some(u64::from(res.unwrap()))
104 let res = src.peek_u32le();
105 if res.is_err() { return None; }
106 Some(u64::from(res.unwrap()))
109 let res = src.peek_u64be();
110 if res.is_err() { return None; }
114 let res = src.peek_u64le();
115 if res.is_err() { return None; }
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() }
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() }
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() }
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() }
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() }
149 Or(&'a CC<'a>, &'a CC<'a>),
160 fn eval(&self, src: &mut ByteReader) -> bool {
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) },
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; }
179 struct CheckItem<'a> {
185 struct DetectConditions<'a> {
186 demux_name: &'static str,
187 extensions: &'static str,
188 conditions: &'a [CheckItem<'a>],
191 const DETECTORS: &[DetectConditions] = &[
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")) },
204 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"RIFF") },
205 CheckItem{offs: 8, cond: &CC::Str(b"WAVEfmt ") }
211 conditions: &[CheckItem{offs: 4, cond: &CC::Or(&CC::Or(&CC::Str(b"mdat"),
213 &CC::Str(b"ftyp")) }],
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"),
221 &CC::Str(b"ftyp")) }],
226 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"FCMP")}],
231 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"2TSF")}],
236 conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}],
239 demux_name: "fable-imax",
241 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"IMAX") },
242 CheckItem{offs: 10, cond: &CC::Eq(Arg::U16LE(0x102)) }],
245 demux_name: "realaudio",
246 extensions: ".ra,.ram",
247 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}],
250 demux_name: "realmedia",
251 extensions: ".rm,.rmvb,.rma,.ra,.ram",
252 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) },
253 CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}],
256 demux_name: "real_ivr",
258 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}],
262 extensions: ".bik,.bk2",
263 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x42494B62), // BIKb
264 Arg::U32BE(0x42494B7B)), // BIKz
265 &CC::In(Arg::U32BE(0x4B423261), // KB2a
266 Arg::U32BE(0x4B42327B)))}], // KB2z
269 demux_name: "smacker",
271 conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}],
276 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"MAC ") },
277 CheckItem{offs: 4, cond: &CC::In(Arg::U16LE(3800), Arg::U16LE(3990))}],
282 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"fLaC") }],
287 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"TTA1") }],
290 demux_name: "wavpack",
292 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"wvpk") },
293 CheckItem{offs: 8, cond: &CC::In(Arg::U16LE(0x402), Arg::U16LE(0x410))}],
298 conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
299 CheckItem{offs: 2, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
304 conditions: &[CheckItem{offs: 0, cond: &CC::In(Arg::U16BE(1), Arg::U16BE(0xFF))},
305 CheckItem{offs: 3, cond: &CC::Str(b"\x0D\x0AVersion:Vivo/")}],
315 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") },
316 CheckItem{offs: 32, cond: &CC::Str(b"DATA")}],
326 conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"VXDS") }],
330 /// Tries to detect container format.
332 /// This function tries to determine container format using both file extension and checking against container specific markers inside.
333 /// In case of success the function returns short container name and the detection score.
334 /// Result should have the highest detection score among tested.
335 pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> {
336 let mut result = None;
337 let lname = name.to_lowercase();
338 for detector in DETECTORS {
339 let mut score = DetectionScore::No;
340 if !name.is_empty() {
341 for ext in detector.extensions.split(',') {
342 if lname.ends_with(ext) {
343 score = DetectionScore::ExtensionMatches;
348 let mut passed = !detector.conditions.is_empty();
349 for ck in detector.conditions {
350 let ret = src.seek(SeekFrom::Start(u64::from(ck.offs)));
355 if !ck.cond.eval(src) {
361 score = DetectionScore::MagicMatches;
363 if score == DetectionScore::MagicMatches {
364 return Some((detector.demux_name, score));
366 if result.is_none() && score != DetectionScore::No {
367 result = Some((detector.demux_name, score));
368 } else if result.is_some() {
369 let (_, oldsc) = result.unwrap();
370 if oldsc.less(score) {
371 result = Some((detector.demux_name, score));
378 /// Tries to detect container format for provided file name.
379 pub fn detect_format_by_name(name: &str) -> Option<&'static str> {
383 let lname = name.to_lowercase();
384 for detector in DETECTORS {
385 for ext in detector.extensions.split(',') {
386 if lname.ends_with(ext) {
387 return Some(detector.demux_name);
398 use crate::io::byteio::*;
401 fn test_avi_detect() {
402 let name = "assets/Indeo/laser05.avi";
403 let mut file = File::open(name).unwrap();
404 let mut fr = FileReader::new_read(&mut file);
405 let mut br = ByteReader::new(&mut fr);
406 let (name, score) = detect_format(name, &mut br).unwrap();
407 assert_eq!(name, "avi");
408 assert_eq!(score, DetectionScore::MagicMatches);
412 fn test_gdv_detect() {
413 let name = "assets/Game/intro1.gdv";
414 let mut file = File::open(name).unwrap();
415 let mut fr = FileReader::new_read(&mut file);
416 let mut br = ByteReader::new(&mut fr);
417 let (name, score) = detect_format(name, &mut br).unwrap();
418 assert_eq!(name, "gdv");
419 assert_eq!(score, DetectionScore::MagicMatches);