binhex-rs 0.1.0

#![doc = include_str!("../README.md")]
mod fork_reader;
mod lookup_table;
mod seek_over;
mod six_bit_decoder;
mod six_bit_rle_reader;
mod util;

use std::fs;
use std::io;
use std::io::{Read as _, Seek as _};
use std::path;

use binrw::{binread, BinReaderExt};
use crc::CRC_16_XMODEM;
use fourcc_rs::FourCC;
use macintosh_utils::FinderFlags;

pub use fork_reader::ForkReader;
use seek_over::SeekOver;
use six_bit_rle_reader::SixBitRleReader;
use util::ReadByte;

use crate::six_bit_decoder::SixBitDecoder;

/// General error used by the crate
#[derive(Debug, thiserror::Error)]
pub enum Error {
    /// The BinHex 4.0 header could not be located
    #[error("The BinHex 4.0 header could not be located")]
    HeaderNotFound,

    /// A read or seek operation on the underlying reader failed
    #[error(transparent)]
    Io(#[from] io::Error),

    /// Some data could not be decoded from the underlying reader
    #[error(transparent)]
    BinRw(#[from] binrw::Error),

    /// The binhex encoded stream contains unknown characters
    #[error("An unexpected character appeared within encoded data")]
    InvalidCharacter,

    /// The underlying stream did not provide enough data for decompression
    #[error("Unexpected EOF")]
    UnexpectedEof,

    /// The seek operation is not supported
    #[error("Seeking backwards or from the end is not supported at the moment")]
    UnsupportedSeek,
}

impl From<Error> for io::Error {
    fn from(val: Error) -> Self {
        match val {
            Error::Io(io) => io,
            other => io::Error::other(Box::new(other)),
        }
    }
}

const VERIFICATION_CHUNK_SIZE: usize = 1 << 20 /* bytes => 1 Mb */;

#[derive(Debug)]
/// Identifies the location where checksum verification failed
pub enum Checksum {
    /// The archive header is broken
    Header,
    /// The data fork is invalid
    DataFork,
    /// The resource fork is invalid
    ResourceFork,
}

#[derive(Debug, thiserror::Error)]
/// Error used to signal invalid checksums
pub enum VerificationError {
    #[error("A checksum did not match")]
    /// A crc check has failed
    ChecksumMismatch(Checksum),

    #[error(transparent)]
    /// The underlying stream returned an io::Error during seek or read
    Io(#[from] io::Error),

    #[error(transparent)]
    /// BinRw was unable to decode some data
    BinRw(#[from] binrw::Error),
}

#[binread]
#[derive(Debug, Clone)]
#[br(big)]
/// On-disk structure describing contents of the BinHex archive
pub struct ArchiveHeader {
    /// Original name of the embedded file
    #[br(map(macintosh_utils::string))]
    pub name: String,
    #[br(temp)]
    _name_terminator: u8,
    /// Four-byte type code for the embedded file
    pub file_code: FourCC,
    /// Four-byte identifier for the program that created the file
    pub creator_code: FourCC,
    /// Finder flags as observed at the time of compression
    pub finder_flags: FinderFlags,
    /// Number of bytes in the uncompressed data fork
    pub data_len: u32,
    /// Number of bytes in the uncompressed resource fork
    pub resource_len: u32,
    /// CRC-16 checksum of the header bytes
    ///
    /// The CRC-16 configuration is `CRC-16-XMODEM` the CRC of ASCII "123456789" is 0x31C3.
    pub header_checksum: u16,
}

impl ArchiveHeader {
    /// Fixed size of the header on disk
    pub const FIXED_SIZE: usize = 22;
}

/// Reader for BinHex archives
pub struct Archive<R> {
    header: ArchiveHeader,
    reader: SixBitRleReader<R>,
}

impl<R> Archive<R> {
    /// Original name of the embedded file
    pub fn name(&self) -> &str {
        self.header.name.as_str()
    }

    /// The file's four-byte type identifier
    pub fn file_code(&self) -> FourCC {
        self.header.file_code
    }

    /// The file's four-byte creator code
    pub fn creator_code(&self) -> FourCC {
        self.header.creator_code
    }

    /// Finder flags at the time of compression
    pub fn finder_flags(&self) -> FinderFlags {
        self.header.finder_flags
    }

    /// Number of bytes in the uncompressed data fork
    pub fn data_len(&self) -> usize {
        self.header.data_len as usize
    }

    /// Number of bytes in the uncompressed resource fork
    pub fn resource_len(&self) -> usize {
        self.header.resource_len as usize
    }

    /// CRC-16 checksum of the header bytes
    ///
    /// The CRC-16 configuration is `CRC-16-XMODEM` the CRC of ASCII "123456789" is 0x31C3.
    pub fn header_checksum(&self) -> u16 {
        self.header.header_checksum
    }
}

impl<R: io::Read + io::Seek> Archive<R> {
    /// Fixed header that introduces the start of encoded data
    pub const HEADER: &[u8; 45] = b"(This file must be converted with BinHex 4.0)";

    /// Try to read an archive from the reader.
    ///
    /// It searches for the start marker and reads the archive header after that. Returns an error
    /// if either fails.
    pub fn try_from(mut inner: R) -> Result<Self, Error> {
        inner
            .seek_over_string(Self::HEADER)
            .map_err(|_| Error::HeaderNotFound)?;

        let data_start = loop {
            if inner.read_byte()? == b':' {
                break inner.stream_position()?;
            }
        };

        let mut inner = SixBitRleReader::new(inner, data_start);
        let header: ArchiveHeader = inner.read_be()?;

        Ok(Self {
            reader: inner,
            header,
        })
    }

    /// Open the data fork for reading
    ///
    /// Since reading backwards through RLE encoded data is expensive, this should be called before
    /// accessing the resource fork, if the application wants to read both
    pub fn data_fork(&mut self) -> Result<ForkReader<&mut SixBitRleReader<R>>, Error> {
        self.reset()?;

        let position = self.reader.position;
        let length = self.data_len() as u64;

        Ok(ForkReader::new(&mut self.reader, position, length))
    }

    /// Open the resource fork for reading
    ///
    /// Since reading backwards through RLE encoded data is expensive, this should be called after
    /// accessing the data fork, if the application wants to read both
    pub fn resource_fork(&mut self) -> Result<ForkReader<&mut SixBitRleReader<R>>, Error> {
        self.reset()?;

        self.reader.seek(io::SeekFrom::Current(
            self.header.data_len as i64 + /* skip crc as well */ 2,
        ))?;

        let position = self.reader.position;
        let length = self.resource_len() as u64;

        Ok(ForkReader::new(&mut self.reader, position, length))
    }

    /// Verify checksums of the whole archive, this resets the stream and reads the whole file to
    /// calculate the checksums.
    pub fn verify(&mut self) -> Result<(), VerificationError> {
        self.reset()?;

        let crc = crc::Crc::<u16>::new(&CRC_16_XMODEM);

        // Verify header checksum
        let mut buf = vec![0u8; ArchiveHeader::FIXED_SIZE];
        self.reader.read_exact(&mut buf)?;

        let name_length = buf[0] as usize;
        buf.append(&mut vec![0u8; name_length]);
        self.reader.read_exact(
            &mut buf[(ArchiveHeader::FIXED_SIZE)..(ArchiveHeader::FIXED_SIZE + name_length)],
        )?;

        let mut digest = crc.digest();
        digest.update(&buf);
        if digest.finalize() != 0 {
            return Err(VerificationError::ChecksumMismatch(Checksum::Header));
        }

        // Verify data fork checksum
        let mut chunk = [0u8; VERIFICATION_CHUNK_SIZE];
        let mut digest = crc.digest();
        let data_len_offset = name_length + 12;
        let data_len = u32::from_be_bytes([
            buf[data_len_offset],
            buf[data_len_offset + 1],
            buf[data_len_offset + 2],
            buf[data_len_offset + 3],
        ]) as usize
            + 2;

        for _ in 0..(data_len / VERIFICATION_CHUNK_SIZE) {
            self.reader.read_exact(&mut chunk)?;
            digest.update(&chunk);
        }

        let rest = data_len % VERIFICATION_CHUNK_SIZE;
        self.reader.read_exact(&mut chunk[0..rest])?;
        digest.update(&chunk[0..rest]);
        if digest.finalize() != 0 {
            return Err(VerificationError::ChecksumMismatch(Checksum::DataFork));
        }

        // Verify resource fork checksum
        let mut digest = crc.digest();
        let resource_len_offset = name_length + 16;
        let resource_len = u32::from_be_bytes([
            buf[resource_len_offset],
            buf[resource_len_offset + 1],
            buf[resource_len_offset + 2],
            buf[resource_len_offset + 3],
        ]) as usize
            + 2;
        for _ in 0..(resource_len / VERIFICATION_CHUNK_SIZE) {
            self.reader.read_exact(&mut chunk)?;
            digest.update(&chunk);
        }

        let rest = resource_len % VERIFICATION_CHUNK_SIZE;
        self.reader.read_exact(&mut chunk[0..rest])?;
        digest.update(&chunk[0..rest]);
        if digest.finalize() != 0 {
            return Err(VerificationError::ChecksumMismatch(Checksum::ResourceFork));
        }

        Ok(())
    }

    /// Returns the underlying reader that was used initially to open the archive
    /// Note that the read position has probably changed
    pub fn into_inner(self) -> R {
        self.reader.into_inner()
    }

    fn reset(&mut self) -> io::Result<()> {
        self.reader.reset()
    }
}

impl Archive<fs::File> {
    /// Try to open the archive specified by `path`
    pub fn open<P: AsRef<path::Path>>(path: P) -> Result<Self, Error> {
        Self::try_from(fs::File::open(path)?)
    }

    /// Try to clone the archive which would allow you to read resource and data forks
    /// simultaneously
    pub fn try_clone(&self) -> io::Result<Self> {
        Ok(Self {
            header: self.header.clone(),
            reader: self.reader.try_clone()?,
        })
    }
}

impl<R: Clone> Clone for Archive<R> {
    /// Clone the archive, allowing you to access resource and data forks at the same time
    fn clone(&self) -> Self {
        Self {
            reader: self.reader.clone(),
            header: self.header.clone(),
        }
    }
}

pub fn probe<R: io::Read + io::Seek>(reader: R) -> bool {
    Archive::try_from(reader).is_ok()
}

#[cfg(test)]
mod test {
    use std::fs;
    use std::io;
    use std::io::Read as _;

    use fourcc_rs::fourcc;

    use super::Archive;
    use crate::{Checksum, VerificationError};

    #[test]
    fn decode() {
        let file = fs::File::open("./sample-file.hqx").unwrap();
        let archive = Archive::try_from(file).unwrap();
        assert_eq!(archive.name(), "binhex.test.sit");
        assert_eq!(archive.file_code(), fourcc!("SITD"));
        assert_eq!(archive.creator_code(), fourcc!("SIT!"));
        assert_eq!(archive.data_len(), 380);
        assert_eq!(archive.resource_len(), 0);
        assert_eq!(archive.header_checksum(), 0x6e3c);
    }

    #[test]
    fn successfully_verify() {
        let mut reader = Archive::open("sample-file.hqx").unwrap();
        assert!(reader.verify().is_ok());
    }

    #[test]
    fn failed_header_verification() {
        let mut file = fs::File::open("./sample-file.hqx").unwrap();
        let mut buffer = Vec::new();
        file.read_to_end(&mut buffer).unwrap();

        // Corrupt a bit in the archive header
        // this will change the decoded filename from "binhex.test.sit" to "biéhex.test.sit"
        buffer[137 + 3] |= 1 << 3;

        let reader = io::Cursor::new(buffer);
        let mut reader = Archive::try_from(reader).unwrap();

        let result = reader.verify();
        assert!(matches!(
            result,
            Err(VerificationError::ChecksumMismatch(Checksum::Header))
        ));
    }

    #[test]
    fn failed_data_verification() {
        let mut file = fs::File::open("./sample-file.hqx").unwrap();
        let mut buffer = Vec::new();
        file.read_to_end(&mut buffer).unwrap();

        // corrupt a bit in the archive data
        buffer[240 + 3] |= 1 << 3;

        let reader = io::Cursor::new(buffer);
        let mut reader = Archive::try_from(reader).unwrap();

        let result = reader.verify();
        assert!(matches!(
            result,
            Err(VerificationError::ChecksumMismatch(Checksum::DataFork))
        ));
    }
}