flat.rs 98.0%

100 of 102 lines run, 51 of 54 functions · run #303 on main · run missed

//! A flat binary layout for local indexes (oxygen#41): fixed-width
//! little-endian numbers and length-prefixed bytes, read in one pass
//! without building a generic value tree. Only for caches this repository
//! derives and can derive again: never exchanged, never trusted beyond the
//! key they are stored under, and a reader that finds anything amiss fails
//! so the caller rebuilds.

use crate::hash::ObjectRef;
use crate::objects::ObjectError;

#[derive(Default)]
pub struct Writer(pub Vec<u8>);

impl Writer {
    pub fn u8(&mut self, v: u8) {
        self.0.push(v);
    }
    pub fn u32(&mut self, v: u32) {
        self.0.extend_from_slice(&v.to_le_bytes());
    }
    pub fn u64(&mut self, v: u64) {
        self.0.extend_from_slice(&v.to_le_bytes());
    }
    /// A count, which must fit in 32 bits.
    pub fn count(&mut self, n: usize) {
        self.u32(u32::try_from(n).expect("an index count fits in 32 bits"));
    }
    pub fn fixed(&mut self, b: &[u8]) {
        self.0.extend_from_slice(b);
    }
    pub fn bytes(&mut self, b: &[u8]) {
        self.count(b.len());
        self.0.extend_from_slice(b);
    }
    pub fn object(&mut self, r: &ObjectRef) {
        self.fixed(&r.0);
    }
    pub fn opt_object(&mut self, r: &Option<ObjectRef>) {
        match r {
            None => self.u8(0),
            Some(r) => {
                self.u8(1);
                self.object(r);
            }
        }
    }
}

pub struct Reader<'a> {
    buf: &'a [u8],
    pos: usize,
}

fn short() -> ObjectError {
    ObjectError::Schema("index: truncated".into())
}

impl<'a> Reader<'a> {
    pub fn new(buf: &'a [u8]) -> Reader<'a> {
        Reader { buf, pos: 0 }
    }
    pub fn at_end(&self) -> bool {
        self.pos == self.buf.len()
    }
    fn take(&mut self, n: usize) -> Result<&'a [u8], ObjectError> {
        let end = self.pos.checked_add(n).ok_or_else(short)?;
        let s = self.buf.get(self.pos..end).ok_or_else(short)?;
        self.pos = end;
        Ok(s)
    }
    pub fn u8(&mut self) -> Result<u8, ObjectError> {
        Ok(self.take(1)?[0])
    }
    pub fn u32(&mut self) -> Result<u32, ObjectError> {
        Ok(u32::from_le_bytes(self.fixed::<4>()?))
    }
    pub fn u64(&mut self) -> Result<u64, ObjectError> {
        Ok(u64::from_le_bytes(self.fixed::<8>()?))
    }
    /// A count; checked against what is left, so a corrupt one cannot ask
    /// for more room than the index could hold.
    pub fn count(&mut self) -> Result<usize, ObjectError> {
        let n = self.u32()? as usize;
        if n > self.buf.len() - self.pos {
            return Err(short());
        }
        Ok(n)
    }
    pub fn fixed<const N: usize>(&mut self) -> Result<[u8; N], ObjectError> {
        Ok(self.take(N)?.try_into().expect("N bytes"))
    }
    pub fn bytes(&mut self) -> Result<&'a [u8], ObjectError> {
        let n = self.u32()? as usize;
        self.take(n)
    }
    pub fn text(&mut self) -> Result<String, ObjectError> {
        String::from_utf8(self.bytes()?.to_vec())
            .map_err(|_| ObjectError::Schema("index: text is not UTF-8".into()))
    }
    pub fn object(&mut self) -> Result<ObjectRef, ObjectError> {
        Ok(ObjectRef(self.fixed::<32>()?))
    }
    pub fn opt_object(&mut self) -> Result<Option<ObjectRef>, ObjectError> {
        Ok(match self.u8()? {
            0 => None,
            _ => Some(self.object()?),
        })
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn numbers_and_bytes_round_trip_and_truncation_fails() {
        let mut w = Writer::default();
        w.u8(7);
        w.u32(70_000);
        w.u64(1 << 40);
        w.bytes(b"path/name");
        w.opt_object(&Some(ObjectRef([3; 32])));
        w.opt_object(&None);
        let mut r = Reader::new(&w.0);
        assert_eq!(r.u8().unwrap(), 7);
        assert_eq!(r.u32().unwrap(), 70_000);
        assert_eq!(r.u64().unwrap(), 1 << 40);
        assert_eq!(r.text().unwrap(), "path/name");
        assert_eq!(r.opt_object().unwrap(), Some(ObjectRef([3; 32])));
        assert_eq!(r.opt_object().unwrap(), None);
        assert!(r.at_end());
        let mut r = Reader::new(&w.0[..w.0.len() - 3]);
        r.fixed::<13>().unwrap();
        assert!(r.text().is_ok());
        assert!(r.opt_object().is_err());
    }
}

1		//! A flat binary layout for local indexes (oxygen#41): fixed-width
2		//! little-endian numbers and length-prefixed bytes, read in one pass
3		//! without building a generic value tree. Only for caches this repository
4		//! derives and can derive again: never exchanged, never trusted beyond the
5		//! key they are stored under, and a reader that finds anything amiss fails
6		//! so the caller rebuilds.
7
8		use crate::hash::ObjectRef;
9		use crate::objects::ObjectError;
10
11		#[derive(Default)]
12		pub struct Writer(pub Vec<u8>);
13
14		impl Writer {
15	349561×	pub fn u8(&mut self, v: u8) {
16	349561×	self.0.push(v);
17	349561×	}
18	386327×	pub fn u32(&mut self, v: u32) {
19	386327×	self.0.extend_from_slice(&v.to_le_bytes());
20	386327×	}
21	70739×	pub fn u64(&mut self, v: u64) {
22	70739×	self.0.extend_from_slice(&v.to_le_bytes());
23	70739×	}
24		/// A count, which must fit in 32 bits.
25	182994×	pub fn count(&mut self, n: usize) {
26	182994×	self.u32(u32::try_from(n).expect("an index count fits in 32 bits"));
27	182994×	}
28	704581×	pub fn fixed(&mut self, b: &[u8]) {
29	704581×	self.0.extend_from_slice(b);
30	704581×	}
31	112484×	pub fn bytes(&mut self, b: &[u8]) {
32	112484×	self.count(b.len());
33	112484×	self.0.extend_from_slice(b);
34	112484×	}
35	248918×	pub fn object(&mut self, r: &ObjectRef) {
36	248918×	self.fixed(&r.0);
37	248918×	}
38	30603×	pub fn opt_object(&mut self, r: &Option<ObjectRef>) {
39	30603×	match r {
40	30363×	None => self.u8(0),
41	240×	Some(r) => {
42	240×	self.u8(1);
43	240×	self.object(r);
44	240×	}
45		}
46	30603×	}
47		}
48
49		pub struct Reader<'a> {
50		buf: &'a [u8],
51		pos: usize,
52		}
53
54	3×	fn short() -> ObjectError {
55	3×	ObjectError::Schema("index: truncated".into())
56	3×	}
57
58		impl<'a> Reader<'a> {
59	21737×	pub fn new(buf: &'a [u8]) -> Reader<'a> {
60	21737×	Reader { buf, pos: 0 }
61	21737×	}
62	43395×	pub fn at_end(&self) -> bool {
63	43395×	self.pos == self.buf.len()
64	43395×	}
65	1206391×	fn take(&mut self, n: usize) -> Result<&'a [u8], ObjectError> {
66	1206391×	let end = self.pos.checked_add(n).ok_or_else(short)?;
67	1206391×	let s = self.buf.get(self.pos..end).ok_or_else(short)?;
68	1206388×	self.pos = end;
69	1206388×	Ok(s)
70	1206391×	}
71	237769×	pub fn u8(&mut self) -> Result<u8, ObjectError> {
72	237769×	Ok(self.take(1)?[0])
73	237769×	}
74	297449×	pub fn u32(&mut self) -> Result<u32, ObjectError> {
75	297449×	Ok(u32::from_le_bytes(self.fixed::<4>()?))
76	297449×	}
77	54102×	pub fn u64(&mut self) -> Result<u64, ObjectError> {
78	54102×	Ok(u64::from_le_bytes(self.fixed::<8>()?))
79	54102×	}
80		/// A count; checked against what is left, so a corrupt one cannot ask
81		/// for more room than the index could hold.
82	56275×	pub fn count(&mut self) -> Result<usize, ObjectError> {
83	56275×	let n = self.u32()? as usize;
84	56275×	if n > self.buf.len() - self.pos {
85	0×	return Err(short());
86	56275×	}
87	56275×	Ok(n)
88	56275×	}
89	866399×	pub fn fixed<const N: usize>(&mut self) -> Result<[u8; N], ObjectError> {
90	866399×	Ok(self.take(N)?.try_into().expect("N bytes"))
91	866399×	}
92	102223×	pub fn bytes(&mut self) -> Result<&'a [u8], ObjectError> {
93	102223×	let n = self.u32()? as usize;
94	102223×	self.take(n)
95	102223×	}
96	61484×	pub fn text(&mut self) -> Result<String, ObjectError> {
97	61484×	String::from_utf8(self.bytes()?.to_vec())
98	61484×	.map_err(\|_\| ObjectError::Schema("index: text is not UTF-8".into()))
99	61484×	}
100	177975×	pub fn object(&mut self) -> Result<ObjectRef, ObjectError> {
101	177975×	Ok(ObjectRef(self.fixed::<32>()?))
102	177975×	}
103	25565×	pub fn opt_object(&mut self) -> Result<Option<ObjectRef>, ObjectError> {
104	25565×	Ok(match self.u8()? {
105	25372×	0 => None,
106	193×	_ => Some(self.object()?),
107		})
108	25565×	}
109		}
110
111		#[cfg(test)]
112		mod tests {
113		use super::*;
114
115		#[test]
116	1×	fn numbers_and_bytes_round_trip_and_truncation_fails() {
117	1×	let mut w = Writer::default();
118	1×	w.u8(7);
119	1×	w.u32(70_000);
120	1×	w.u64(1 << 40);
121	1×	w.bytes(b"path/name");
122	1×	w.opt_object(&Some(ObjectRef([3; 32])));
123	1×	w.opt_object(&None);
124	1×	let mut r = Reader::new(&w.0);
125	1×	assert_eq!(r.u8().unwrap(), 7);
126	1×	assert_eq!(r.u32().unwrap(), 70_000);
127	1×	assert_eq!(r.u64().unwrap(), 1 << 40);
128	1×	assert_eq!(r.text().unwrap(), "path/name");
129	1×	assert_eq!(r.opt_object().unwrap(), Some(ObjectRef([3; 32])));
130	1×	assert_eq!(r.opt_object().unwrap(), None);
131	1×	assert!(r.at_end());
132	1×	let mut r = Reader::new(&w.0[..w.0.len() - 3]);
133	1×	r.fixed::<13>().unwrap();
134	1×	assert!(r.text().is_ok());
135	1×	assert!(r.opt_object().is_err());
136	1×	}
137		}

`/`	Search
`g` `c`	Code
`g` `h`	History
`g` `b`	Bookmarks
`g` `t`	Trust
`y`	Permalink to this state (files and trees)
`?`	This sheet

Coverage / crates/oxy-core/src/flat.rs 98.0%

Keyboard