haggis-rs/src/node.rs

359 lines
12 KiB
Rust

use {
crate::{filetype::Flag, nix, Algorithm, Checksum, Error, File, FileType, Special},
md5::{Digest, Md5},
sha1::Sha1,
sha2::Sha256,
std::{
collections::HashMap,
env, fs,
io::{BufReader, BufWriter, Read, Write},
os::unix::fs::{symlink, MetadataExt},
path::{Path, PathBuf},
sync::Mutex,
},
};
#[derive(Debug, PartialEq)]
enum Kind {
Normal,
Dir,
Char,
Block,
Pipe,
}
impl From<u32> for Kind {
fn from(value: u32) -> Self {
if value & 0o60000 != 0 {
Self::Block
} else if value & 0o20000 != 0 {
Self::Char
} else if value & 0o10000 != 0 {
Self::Pipe
} else if value & 0o40000 != 0 {
Self::Dir
} else if value & 0o100000 != 0 {
Self::Normal
} else {
panic!();
}
}
}
/// A representation of a file and it's associated metadata.
#[derive(Debug)]
pub struct Node {
/// The filesystem path to this file
pub name: String,
/// The user id of this file's owner
pub uid: u32,
/// The group id of this file's owner
pub gid: u32,
/// The most recent modification time of this file
pub mtime: u64,
/// The Unix permissions bits of this file
pub mode: u16,
/// The type of file this node represents
pub filetype: FileType,
}
impl Node {
/// Reads a `Node` from an archive file or stream of Nodes.
/// > Note: this function reads an already created node. To create a new node
/// > from a file, use the `from_path` method.
pub fn read<T: Read>(reader: &mut T) -> Result<Self, Error> {
let mut len = [0; 2];
reader.read_exact(&mut len)?;
let len = u16::from_le_bytes(len);
if len == 0 {
return Ok(Self {
name: "".to_string(),
mode: 0,
uid: 0,
gid: 0,
mtime: 0,
filetype: FileType::Eof,
});
}
let mut name = Vec::with_capacity(len.into());
let mut handle = reader.take(len.into());
handle.read_to_end(&mut name)?;
let mut buf = [0; 18];
reader.read_exact(&mut buf)?;
let uid: [u8; 4] = buf[0..4].try_into()?;
let gid: [u8; 4] = buf[4..8].try_into()?;
let mtime: [u8; 8] = buf[8..16].try_into()?;
let raw_mode: [u8; 2] = buf[16..18].try_into()?;
let raw_mode = u16::from_le_bytes(raw_mode);
let mask: u16 = 0b111 << 13;
let mode = raw_mode & mask;
let flag: u8 = ((raw_mode & !mask) >> 13).try_into()?;
let flag: Flag = flag.try_into()?;
let filetype = FileType::read(reader, flag)?;
Ok(Self {
name: String::from_utf8(name)?,
uid: u32::from_le_bytes(uid),
gid: u32::from_le_bytes(gid),
mtime: u64::from_le_bytes(mtime),
mode,
filetype,
})
}
/// Write a `Node` struct into it's on-disk archive representation.
/// > Note: this function saves the data to the archive format's on-disk
/// > representation. To extract the contents of a `Node` and write out the
/// > file it represents, use the `extract` method instead.
pub fn write<T: Write>(&self, writer: &mut T) -> Result<(), Error> {
let len: u16 = self.name.len().try_into()?;
writer.write_all(&len.to_le_bytes())?;
writer.write_all(self.name.as_bytes())?;
writer.write_all(&self.mode.to_le_bytes())?;
[self.uid, self.gid]
.iter()
.try_for_each(|f| writer.write_all(&f.to_le_bytes()))?;
writer.write_all(&self.mtime.to_le_bytes())?;
self.filetype.write(writer)?;
Ok(())
}
/// Creates a new node from a file which exists on the filesystem
/// ### Parameters
/// - path - the path to this file
/// - checksum - a zeroed out `Checksum` variant to be used if the inline
/// checksumming feature is to be used
/// - links - this should be passed to each invocation of `from_path` used
/// during the creation of a single archive, to identify hard links and to
/// avoid writing their data out more than once.
pub fn from_path(
path: &str,
algorithm: Algorithm,
links: &Mutex<HashMap<u64, String>>,
) -> Result<Self, Error> {
let name = String::from(path);
let fd = fs::File::open(path)?;
let meta = fd.metadata()?;
let mode = meta.mode();
let uid = meta.uid();
let gid = meta.gid();
let mtime = meta.mtime().try_into()?;
let mut reader = BufReader::new(fd);
let ft = meta.file_type();
let filetype = 'blk: {
if ft.is_dir() {
FileType::Directory
} else if ft.is_symlink() {
let target = fs::read_link(path)?;
let target = target
.to_str()
.ok_or(Error::Other("bad path".to_string()))?
.to_string();
FileType::SoftLink(target)
} else {
if meta.nlink() > 1 {
if let Ok(mut list) = links.lock() {
let inode = meta.ino();
if let Some(target) = list.get(&inode).cloned() {
break 'blk FileType::HardLink(target);
} else {
list.insert(inode, name.clone());
}
}
}
let kind = Kind::from(mode);
if kind == Kind::Char {
let device = Special::from_rdev(meta.rdev());
break 'blk FileType::Character(device);
} else if kind == Kind::Block {
let device = Special::from_rdev(meta.rdev());
break 'blk FileType::Block(device);
} else if kind == Kind::Pipe {
break 'blk FileType::Fifo;
} else if kind == Kind::Normal {
let mut len = meta.len();
let mut data = Vec::with_capacity(len.try_into()?);
len = reader.read_to_end(&mut data)?.try_into()?;
let checksum = match algorithm {
Algorithm::Md5 => {
let mut hasher = Md5::new();
hasher.update(&data);
let cs = hasher.finalize().into();
Checksum::Md5(cs)
}
Algorithm::Sha1 => {
let mut hasher = Sha1::new();
hasher.update(&data);
let cs = hasher.finalize().into();
Checksum::Sha1(cs)
}
Algorithm::Sha256 => {
let mut hasher = Sha256::new();
hasher.update(&data);
let cs = hasher.finalize().into();
Checksum::Sha256(cs)
}
Algorithm::Skip => Checksum::Skip,
};
break 'blk FileType::Normal(File {
len,
checksum,
data,
});
} else {
return Err(Error::UnknownFileType);
}
}
};
Ok(Self {
name,
mode,
uid,
gid,
mtime,
filetype,
})
}
/// Recreates the original file on disk which this node represents
pub fn extract(&self, prefix: Option<&str>) -> Result<(), Error> {
let euid = nix::geteuid();
let path = 'blk: {
if let Some(prefix) = prefix {
if self.name.starts_with('/') {
break 'blk format!("{prefix}{}", self.name);
} else {
break 'blk format!("{prefix}/{}", self.name);
}
}
self.name.clone()
};
let n_path = if path.starts_with('/') {
PathBuf::from(&path)
} else {
if let Ok(mut p) = env::current_dir() {
p.push(&path);
p
} else {
PathBuf::from(&format!("./{path}"))
}
};
if let Some(p) = n_path.parent() {
if !p.exists() {
self.mkdir(&p)?;
}
}
match self.filetype {
FileType::Eof => {}
FileType::Fifo => {
nix::mkfifo(&path, self.mode)?;
if euid == 0 {
nix::chown(&path, self.uid, self.gid)?;
}
}
FileType::Block(ref b) => {
if euid == 0 {
nix::mknod(&path, self.mode, b.major, b.minor)?;
}
}
FileType::Character(ref c) => {
if euid == 0 {
nix::mknod(&path, self.mode, c.major, c.minor)?;
}
}
FileType::Normal(ref n) => {
{
let fd = fs::OpenOptions::new()
.create(true)
.truncate(true)
.write(true)
.open(&n_path)?;
let mut writer = BufWriter::new(fd);
writer.write_all(&n.data)?;
}
if euid == 0 {
nix::chown(&path, self.uid, self.gid)?;
}
nix::chmod(&path, self.mode)?;
}
FileType::HardLink(ref t) => {
let target = if let Some(prefix) = prefix {
format!("{prefix}/{t}")
} else {
t.to_string()
};
fs::hard_link(&target, &path)?;
}
FileType::SoftLink(ref t) => {
symlink(&self.name, t)?;
}
FileType::Directory => {
self.mkdir(&PathBuf::from(&path))?;
}
}
Ok(())
}
fn mkdir(&self, dir: &Path) -> Result<(), Error> {
if !dir.exists() {
fs::create_dir_all(dir)?;
}
if nix::geteuid() == 0 {
nix::chown(dir.to_str().ok_or(Error::NulError)?, self.uid, self.gid)?;
}
nix::chmod(
dir.to_str().ok_or(Error::BadPath)?,
self.mode & 0o7777 | 0o100,
)?;
Ok(())
}
}
#[cfg(test)]
mod tests {
use super::*;
use std::fmt::Write;
static LI: &[u8] = include_bytes!("../test/li.txt");
#[test]
fn from_path() {
let links = Mutex::new(HashMap::new());
let node = Node::from_path("test/li.txt", Algorithm::Sha256, &links).unwrap();
let FileType::Normal(f) = node.filetype else { panic!() };
assert_eq!(f.len, 1005);
let Checksum::Sha256(sum) = f.checksum else { panic!() };
let mut s = String::new();
for c in &sum {
write!(s, "{c:02x}").unwrap();
}
assert_eq!(
s,
"5f1b6e6e31682fb6683db2e78db11e624527c897618f1a5b0a0b5256f557c22d"
);
}
#[test]
fn write() {
let fd = std::fs::File::create("test/li.node").unwrap();
let mut writer = std::io::BufWriter::new(fd);
let links = Mutex::new(HashMap::new());
let node = Node::from_path("test/li.txt", Algorithm::Sha1, &links).unwrap();
node.write(&mut writer).unwrap();
}
#[test]
fn read() {
let fd = std::fs::File::open("test/li.node").unwrap();
let mut reader = BufReader::new(fd);
let node = Node::read(&mut reader).unwrap();
let FileType::Normal(f) = node.filetype else { panic!() };
let Checksum::Sha1(sum) = f.checksum else { panic!() };
let mut s = String::new();
for c in &sum {
write!(s, "{c:02x}").unwrap();
}
assert_eq!(s, "9bf3e5b5efd22f932e100b86c83482787e82a682");
assert_eq!(LI, f.data);
}
}