[nihav.git] / nihav-core / src / detect.rs

use std::io::SeekFrom;
use crate::io::byteio::ByteReader;

#[derive(Debug,Clone,Copy,PartialEq)]
pub enum DetectionScore {
    No,
    ExtensionMatches,
    MagicMatches,
}

impl DetectionScore {
    pub fn less(&self, other: DetectionScore) -> bool {
        (*self as i32) < (other as i32)
    }
}

#[allow(dead_code)]
enum Arg {
    Byte(u8),
    U16BE(u16),
    U16LE(u16),
    U24BE(u32),
    U24LE(u32),
    U32BE(u32),
    U32LE(u32),
    U64BE(u64),
    U64LE(u64),
}

impl Arg {
    fn val(&self) -> u64 {
        match *self {
            Arg::Byte(b) => { b as u64 }
            Arg::U16BE(v) => { v as u64 }
            Arg::U16LE(v) => { v as u64 }
            Arg::U24BE(v) => { v as u64 }
            Arg::U24LE(v) => { v as u64 }
            Arg::U32BE(v) => { v as u64 }
            Arg::U32LE(v) => { v as u64 }
            Arg::U64BE(v) => { v }
            Arg::U64LE(v) => { v }
        }
    }
    fn read_val(&self, src: &mut ByteReader) -> Option<u64> {
        match *self {
            Arg::Byte(_) => {
                let res = src.peek_byte();
                if let Err(_) = res { return None; }
                Some(res.unwrap() as u64)
            }
            Arg::U16BE(_) => {
                let res = src.peek_u16be();
                if let Err(_) = res { return None; }
                Some(res.unwrap() as u64)
            }
            Arg::U16LE(_) => {
                let res = src.peek_u16le();
                if let Err(_) = res { return None; }
                Some(res.unwrap() as u64)
            }
            Arg::U24BE(_) => {
                let res = src.peek_u24be();
                if let Err(_) = res { return None; }
                Some(res.unwrap() as u64)
            }
            Arg::U24LE(_) => {
                let res = src.peek_u24le();
                if let Err(_) = res { return None; }
                Some(res.unwrap() as u64)
            }
            Arg::U32BE(_) => {
                let res = src.peek_u32be();
                if let Err(_) = res { return None; }
                Some(res.unwrap() as u64)
            }
            Arg::U32LE(_) => {
                let res = src.peek_u32le();
                if let Err(_) = res { return None; }
                Some(res.unwrap() as u64)
            }
            Arg::U64BE(_) => {
                let res = src.peek_u64be();
                if let Err(_) = res { return None; }
                Some(res.unwrap())
            }
            Arg::U64LE(_) => {
                let res = src.peek_u64le();
                if let Err(_) = res { return None; }
                Some(res.unwrap())
            }
        }
    }
    fn eq(&self, src: &mut ByteReader) -> bool {
        let val = self.read_val(src);
        if let None = val { false }
        else { val.unwrap() == self.val() }
    }
    fn ge(&self, src: &mut ByteReader) -> bool {
        let val = self.read_val(src);
        if let None = val { false }
        else { val.unwrap() >= self.val() }
    }
    fn gt(&self, src: &mut ByteReader) -> bool {
        let val = self.read_val(src);
        if let None = val { false }
        else { val.unwrap() > self.val() }
    }
    fn le(&self, src: &mut ByteReader) -> bool {
        let val = self.read_val(src);
        if let None = val { false }
        else { val.unwrap() <= self.val() }
    }
    fn lt(&self, src: &mut ByteReader) -> bool {
        let val = self.read_val(src);
        if let None = val { false }
        else { val.unwrap() < self.val() }
    }
}

#[allow(dead_code)]
enum CC<'a> {
    Or(&'a CC<'a>, &'a CC<'a>),
    Eq(Arg),
    Str(&'static [u8]),
    In(Arg, Arg),
    Lt(Arg),
    Le(Arg),
    Gt(Arg),
    Ge(Arg),
}

impl<'a> CC<'a> {
    fn eval(&self, src: &mut ByteReader) -> bool {
        match *self {
            CC::Or (ref a, ref b) => { a.eval(src) || b.eval(src) },
            CC::Eq(ref arg)      => { arg.eq(src) },
            CC::In(ref a, ref b) => { a.ge(src) && b.le(src) },
            CC::Lt(ref arg)      => { arg.lt(src) },
            CC::Le(ref arg)      => { arg.le(src) },
            CC::Gt(ref arg)      => { arg.gt(src) },
            CC::Ge(ref arg)      => { arg.ge(src) },
            CC::Str(str) => {
                let mut val: Vec<u8> = Vec::with_capacity(str.len());
                val.resize(str.len(), 0);
                let res = src.peek_buf(val.as_mut_slice());
                if let Err(_) = res { return false; }
                val == str
            }
        }
    }
}

struct CheckItem<'a> {
    offs: u32,
    cond: &'a CC<'a>,
}

#[allow(dead_code)]
struct DetectConditions<'a> {
    demux_name: &'static str,
    extensions: &'static str,
    conditions: &'a [CheckItem<'a>],
}

const DETECTORS: &[DetectConditions] = &[
    DetectConditions {
        demux_name: "avi",
        extensions: ".avi",
        conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"RIFF"), &CC::Str(b"ON2 ")) },
                      CheckItem{offs: 8, cond: &CC::Or(&CC::Or(&CC::Str(b"AVI LIST"),
                                                               &CC::Str(b"AVIXLIST")),
                                                               &CC::Str(b"ON2fLIST")) },
                     ]
    },
    DetectConditions {
        demux_name: "gdv",
        extensions: ".gdv",
        conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}],
    },
    DetectConditions {
        demux_name: "realaudio",
        extensions: ".ra,.ram",
        conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}],
    },
    DetectConditions {
        demux_name: "realmedia",
        extensions: ".rm,.rmvb,.rma,.ra,.ram",
        conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) },
                      CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}],
    },
    DetectConditions {
        demux_name: "real_ivr",
        extensions: ".ivr",
        conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}],
    },
    DetectConditions {
        demux_name: "bink",
        extensions: ".bik,.bk2",
        conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x32494B62),     // BIKb
                                                               Arg::U32BE(0x32494B7B)),    // BIKz
                                                       &CC::In(Arg::U32BE(0x4B423261),     // KB2a
                                                               Arg::U32BE(0x4B42327B)))}], // KB2z
    },
    DetectConditions {
        demux_name: "smacker",
        extensions: ".smk",
        conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}],
    },
    DetectConditions {
        demux_name: "bmv",
        extensions: ".bmv",
        conditions: &[],
    },
    DetectConditions {
        demux_name: "bmv3",
        extensions: ".bmv",
        conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") },
                      CheckItem{offs: 32, cond: &CC::Str(b"DATA")}],
    },
    DetectConditions {
        demux_name: "vmd",
        extensions: ".vmd",
        conditions: &[],
    },
];

pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> {
    let mut result = None;
    let lname = name.to_lowercase();
    for detector in DETECTORS {
        let mut score = DetectionScore::No;
        if name.len() > 0 {
            for ext in detector.extensions.split(',') {
                if lname.ends_with(ext) {
                    score = DetectionScore::ExtensionMatches;
                    break;
                }
            }
        }
        let mut passed = detector.conditions.len() > 0;
        for ck in detector.conditions {
            let ret = src.seek(SeekFrom::Start(ck.offs as u64));
            if let Err(_) = ret {
                passed = false;
                break;
            }
            if !ck.cond.eval(src) {
                passed = false;
                break;
            }
        }
        if passed {
            score = DetectionScore::MagicMatches;
        }
        if score == DetectionScore::MagicMatches {
            return Some((detector.demux_name, score));
        }
        if result.is_none() && score != DetectionScore::No {
            result = Some((detector.demux_name, score));
        } else if result.is_some() {
            let (_, oldsc) = result.unwrap();
            if oldsc.less(score) {
                result = Some((detector.demux_name, score));
            }
        }
    }
    result
}

#[cfg(test)]
mod test {
    use super::*;
    use std::fs::File;
    use crate::io::byteio::*;

    #[test]
    fn test_avi_detect() {
        let name = "assets/Indeo/laser05.avi";
        let mut file = File::open(name).unwrap();
        let mut fr = FileReader::new_read(&mut file);
        let mut br = ByteReader::new(&mut fr);
        let (name, score) = detect_format(name, &mut br).unwrap();
        assert_eq!(name, "avi");
        assert_eq!(score, DetectionScore::MagicMatches);
    }

    #[test]
    fn test_gdv_detect() {
        let name = "assets/Game/intro1.gdv";
        let mut file = File::open(name).unwrap();
        let mut fr = FileReader::new_read(&mut file);
        let mut br = ByteReader::new(&mut fr);
        let (name, score) = detect_format(name, &mut br).unwrap();
        assert_eq!(name, "gdv");
        assert_eq!(score, DetectionScore::MagicMatches);
    }
}
Commit	Line	Data
d8ce0de0	1	use std::io::SeekFrom;
aca89041	2	use crate::io::byteio::ByteReader;
d8ce0de0 KS	3
	4	#[derive(Debug,Clone,Copy,PartialEq)]
	5	pub enum DetectionScore {
	6	No,
	7	ExtensionMatches,
	8	MagicMatches,
	9	}
	10
	11	impl DetectionScore {
	12	pub fn less(&self, other: DetectionScore) -> bool {
	13	(*self as i32) < (other as i32)
	14	}
	15	}
	16
	17	#[allow(dead_code)]
b5bd2ae4	18	enum Arg {
d8ce0de0 KS	19	Byte(u8),
	20	U16BE(u16),
	21	U16LE(u16),
	22	U24BE(u32),
	23	U24LE(u32),
	24	U32BE(u32),
	25	U32LE(u32),
	26	U64BE(u64),
	27	U64LE(u64),
	28	}
	29
b5bd2ae4	30	impl Arg {
d8ce0de0 KS	31	fn val(&self) -> u64 {
d8ce0de0 KS	32	match *self {
b5bd2ae4 KS	33	Arg::Byte(b) => { b as u64 }
	34	Arg::U16BE(v) => { v as u64 }
	35	Arg::U16LE(v) => { v as u64 }
	36	Arg::U24BE(v) => { v as u64 }
	37	Arg::U24LE(v) => { v as u64 }
	38	Arg::U32BE(v) => { v as u64 }
	39	Arg::U32LE(v) => { v as u64 }
	40	Arg::U64BE(v) => { v }
	41	Arg::U64LE(v) => { v }
d8ce0de0 KS	42	}
	43	}
	44	fn read_val(&self, src: &mut ByteReader) -> Option<u64> {
	45	match *self {
b5bd2ae4	46	Arg::Byte(_) => {
d8ce0de0 KS	47	let res = src.peek_byte();
	48	if let Err(_) = res { return None; }
	49	Some(res.unwrap() as u64)
	50	}
b5bd2ae4	51	Arg::U16BE(_) => {
d8ce0de0 KS	52	let res = src.peek_u16be();
	53	if let Err(_) = res { return None; }
	54	Some(res.unwrap() as u64)
	55	}
b5bd2ae4	56	Arg::U16LE(_) => {
d8ce0de0 KS	57	let res = src.peek_u16le();
	58	if let Err(_) = res { return None; }
	59	Some(res.unwrap() as u64)
	60	}
b5bd2ae4	61	Arg::U24BE(_) => {
d8ce0de0 KS	62	let res = src.peek_u24be();
	63	if let Err(_) = res { return None; }
	64	Some(res.unwrap() as u64)
	65	}
b5bd2ae4	66	Arg::U24LE(_) => {
d8ce0de0 KS	67	let res = src.peek_u24le();
	68	if let Err(_) = res { return None; }
	69	Some(res.unwrap() as u64)
	70	}
b5bd2ae4	71	Arg::U32BE(_) => {
d8ce0de0 KS	72	let res = src.peek_u32be();
	73	if let Err(_) = res { return None; }
	74	Some(res.unwrap() as u64)
	75	}
b5bd2ae4	76	Arg::U32LE(_) => {
d8ce0de0 KS	77	let res = src.peek_u32le();
	78	if let Err(_) = res { return None; }
	79	Some(res.unwrap() as u64)
	80	}
b5bd2ae4	81	Arg::U64BE(_) => {
d8ce0de0 KS	82	let res = src.peek_u64be();
	83	if let Err(_) = res { return None; }
	84	Some(res.unwrap())
	85	}
b5bd2ae4	86	Arg::U64LE(_) => {
d8ce0de0 KS	87	let res = src.peek_u64le();
	88	if let Err(_) = res { return None; }
	89	Some(res.unwrap())
	90	}
	91	}
	92	}
	93	fn eq(&self, src: &mut ByteReader) -> bool {
	94	let val = self.read_val(src);
	95	if let None = val { false }
	96	else { val.unwrap() == self.val() }
	97	}
	98	fn ge(&self, src: &mut ByteReader) -> bool {
	99	let val = self.read_val(src);
	100	if let None = val { false }
	101	else { val.unwrap() >= self.val() }
	102	}
	103	fn gt(&self, src: &mut ByteReader) -> bool {
	104	let val = self.read_val(src);
	105	if let None = val { false }
	106	else { val.unwrap() > self.val() }
	107	}
	108	fn le(&self, src: &mut ByteReader) -> bool {
	109	let val = self.read_val(src);
	110	if let None = val { false }
	111	else { val.unwrap() <= self.val() }
	112	}
	113	fn lt(&self, src: &mut ByteReader) -> bool {
	114	let val = self.read_val(src);
	115	if let None = val { false }
	116	else { val.unwrap() < self.val() }
	117	}
	118	}
	119
	120	#[allow(dead_code)]
b5bd2ae4 KS	121	enum CC<'a> {
	122	Or(&'a CC<'a>, &'a CC<'a>),
	123	Eq(Arg),
	124	Str(&'static [u8]),
	125	In(Arg, Arg),
	126	Lt(Arg),
	127	Le(Arg),
	128	Gt(Arg),
	129	Ge(Arg),
d8ce0de0 KS	130	}
d8ce0de0 KS	131
b5bd2ae4	132	impl<'a> CC<'a> {
d8ce0de0 KS	133	fn eval(&self, src: &mut ByteReader) -> bool {
d8ce0de0 KS	134	match *self {
b5bd2ae4 KS	135	CC::Or (ref a, ref b) => { a.eval(src) \|\| b.eval(src) },
b5bd2ae4 KS	136	CC::Eq(ref arg) => { arg.eq(src) },
4d477e23	137	CC::In(ref a, ref b) => { a.ge(src) && b.le(src) },
b5bd2ae4 KS	138	CC::Lt(ref arg) => { arg.lt(src) },
	139	CC::Le(ref arg) => { arg.le(src) },
	140	CC::Gt(ref arg) => { arg.gt(src) },
	141	CC::Ge(ref arg) => { arg.ge(src) },
	142	CC::Str(str) => {
d8ce0de0 KS	143	let mut val: Vec<u8> = Vec::with_capacity(str.len());
	144	val.resize(str.len(), 0);
	145	let res = src.peek_buf(val.as_mut_slice());
	146	if let Err(_) = res { return false; }
	147	val == str
	148	}
	149	}
	150	}
	151	}
	152
	153	struct CheckItem<'a> {
	154	offs: u32,
b5bd2ae4	155	cond: &'a CC<'a>,
d8ce0de0 KS	156	}
	157
	158	#[allow(dead_code)]
	159	struct DetectConditions<'a> {
	160	demux_name: &'static str,
	161	extensions: &'static str,
	162	conditions: &'a [CheckItem<'a>],
	163	}
	164
	165	const DETECTORS: &[DetectConditions] = &[
	166	DetectConditions {
	167	demux_name: "avi",
	168	extensions: ".avi",
b5bd2ae4 KS	169	conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"RIFF"), &CC::Str(b"ON2 ")) },
	170	CheckItem{offs: 8, cond: &CC::Or(&CC::Or(&CC::Str(b"AVI LIST"),
	171	&CC::Str(b"AVIXLIST")),
	172	&CC::Str(b"ON2fLIST")) },
	173	]
d8ce0de0 KS	174	},
	175	DetectConditions {
	176	demux_name: "gdv",
	177	extensions: ".gdv",
b5bd2ae4	178	conditions: &[CheckItem{offs: 0, cond: &CC::Eq(Arg::U32LE(0x29111994))}],
d8ce0de0	179	},
ce52b3b5 KS	180	DetectConditions {
	181	demux_name: "realaudio",
	182	extensions: ".ra,.ram",
	183	conditions: &[CheckItem{offs: 0, cond: &CC::Str(b".ra\xFD")}],
	184	},
	185	DetectConditions {
	186	demux_name: "realmedia",
	187	extensions: ".rm,.rmvb,.rma,.ra,.ram",
db5cc44b	188	conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".RMF"), &CC::Str(b".RMP")) },
ce52b3b5 KS	189	CheckItem{offs: 4, cond: &CC::Ge(Arg::U32BE(10))}],
	190	},
	191	DetectConditions {
	192	demux_name: "real_ivr",
	193	extensions: ".ivr",
	194	conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b".R1M"), &CC::Str(b".REC"))}],
	195	},
c6c21059 KS	196	DetectConditions {
c6c21059 KS	197	demux_name: "bink",
4d477e23	198	extensions: ".bik,.bk2",
c6c21059 KS	199	conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::In(Arg::U32BE(0x32494B62), // BIKb
	200	Arg::U32BE(0x32494B7B)), // BIKz
	201	&CC::In(Arg::U32BE(0x4B423261), // KB2a
	202	Arg::U32BE(0x4B42327B)))}], // KB2z
	203	},
606c448e KS	204	DetectConditions {
	205	demux_name: "smacker",
	206	extensions: ".smk",
	207	conditions: &[CheckItem{offs: 0, cond: &CC::Or(&CC::Str(b"SMK2"), &CC::Str(b"SMK4"))}],
	208	},
128253cc KS	209	DetectConditions {
	210	demux_name: "bmv",
	211	extensions: ".bmv",
	212	conditions: &[],
	213	},
ecda1cc1 KS	214	DetectConditions {
	215	demux_name: "bmv3",
	216	extensions: ".bmv",
	217	conditions: &[CheckItem{offs: 0, cond: &CC::Str(b"BMVi") },
	218	CheckItem{offs: 32, cond: &CC::Str(b"DATA")}],
	219	},
9895bd7b KS	220	DetectConditions {
	221	demux_name: "vmd",
	222	extensions: ".vmd",
	223	conditions: &[],
	224	},
d8ce0de0 KS	225	];
	226
	227	pub fn detect_format(name: &str, src: &mut ByteReader) -> Option<(&'static str, DetectionScore)> {
	228	let mut result = None;
	229	let lname = name.to_lowercase();
	230	for detector in DETECTORS {
	231	let mut score = DetectionScore::No;
	232	if name.len() > 0 {
	233	for ext in detector.extensions.split(',') {
	234	if lname.ends_with(ext) {
	235	score = DetectionScore::ExtensionMatches;
	236	break;
	237	}
	238	}
	239	}
	240	let mut passed = detector.conditions.len() > 0;
	241	for ck in detector.conditions {
	242	let ret = src.seek(SeekFrom::Start(ck.offs as u64));
	243	if let Err(_) = ret {
	244	passed = false;
	245	break;
	246	}
	247	if !ck.cond.eval(src) {
	248	passed = false;
	249	break;
	250	}
	251	}
	252	if passed {
	253	score = DetectionScore::MagicMatches;
	254	}
	255	if score == DetectionScore::MagicMatches {
	256	return Some((detector.demux_name, score));
	257	}
4d477e23	258	if result.is_none() && score != DetectionScore::No {
d8ce0de0	259	result = Some((detector.demux_name, score));
4d477e23	260	} else if result.is_some() {
d8ce0de0 KS	261	let (_, oldsc) = result.unwrap();
	262	if oldsc.less(score) {
	263	result = Some((detector.demux_name, score));
	264	}
	265	}
	266	}
	267	result
	268	}
	269
	270	#[cfg(test)]
	271	mod test {
	272	use super::*;
	273	use std::fs::File;
aca89041	274	use crate::io::byteio::*;
d8ce0de0 KS	275
	276	#[test]
	277	fn test_avi_detect() {
250c49f6	278	let name = "assets/Indeo/laser05.avi";
d8ce0de0 KS	279	let mut file = File::open(name).unwrap();
	280	let mut fr = FileReader::new_read(&mut file);
	281	let mut br = ByteReader::new(&mut fr);
	282	let (name, score) = detect_format(name, &mut br).unwrap();
	283	assert_eq!(name, "avi");
	284	assert_eq!(score, DetectionScore::MagicMatches);
	285	}
	286
	287	#[test]
	288	fn test_gdv_detect() {
250c49f6	289	let name = "assets/Game/intro1.gdv";
d8ce0de0 KS	290	let mut file = File::open(name).unwrap();
	291	let mut fr = FileReader::new_read(&mut file);
	292	let mut br = ByteReader::new(&mut fr);
	293	let (name, score) = detect_format(name, &mut br).unwrap();
	294	assert_eq!(name, "gdv");
	295	assert_eq!(score, DetectionScore::MagicMatches);
	296	}
b5bd2ae4	297	}