haggis-rs/src/node.rs

use {
    crate::{Checksum, Error, File, FileType, Special},
    md5::{Digest, Md5},
    sha1::Sha1,
    sha2::Sha256,
    std::{
        collections::HashMap,
        fs,
        io::{BufReader, Read, Write},
        os::unix::fs::MetadataExt,
        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 Unix permissions bits of this file
    pub mode: u32,
    /// 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 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; 8];
        reader.read_exact(&mut len)?;
        let len = u64::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.try_into()?);
        let mut handle = reader.take(len);
        handle.read_exact(&mut name)?;
        let mut buf = [0; 20];
        reader.read_exact(&mut buf)?;
        let mode: [u8; 4] = buf[..4].try_into()?;
        let uid: [u8; 4] = buf[4..8].try_into()?;
        let gid: [u8; 4] = buf[8..12].try_into()?;
        let mtime: [u8; 8] = buf[12..].try_into()?;
        let filetype = FileType::read(reader)?;
        Ok(Self {
            name: String::from_utf8(name)?,
            mode: u32::from_le_bytes(mode),
            uid: u32::from_le_bytes(uid),
            gid: u32::from_le_bytes(gid),
            mtime: u64::from_le_bytes(mtime),
            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 = self.name.len() as u64;
        writer.write_all(&len.to_le_bytes())?;
        writer.write_all(self.name.as_bytes())?;
        [self.mode, 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,
        checksum: Checksum,
        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().try_into()?);
                    break 'blk FileType::Character(device);
                } else if kind == Kind::Block {
                    let device = Special::from_rdev(meta.rdev().try_into()?);
                    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 checksum {
                        Checksum::Md5(mut cs) => {
                            let mut hasher = Md5::new();
                            hasher.update(&data);
                            cs = hasher.finalize().into();
                            Checksum::Md5(cs)
                        }
                        Checksum::Sha1(mut cs) => {
                            let mut hasher = Sha1::new();
                            hasher.update(&data);
                            cs = hasher.finalize().into();
                            Checksum::Sha1(cs)
                        }
                        Checksum::Sha256(mut cs) => {
                            let mut hasher = Sha256::new();
                            hasher.update(&data);
                            cs = hasher.finalize().into();
                            Checksum::Sha256(cs)
                        }
                        Checksum::Skip => checksum,
                    };
                    break 'blk FileType::Normal(File {
                        len,
                        checksum,
                        data,
                    });
                } else {
                    return Err(Error::UnknownFileType);
                }
            }
        };
        Ok(Self {
            name,
            mode,
            uid,
            gid,
            mtime,
            filetype,
        })
    }

    pub fn extract(&self, prefix: Option<&str>) -> Result<(), Error> {
        todo!()
    }
}