shrupl/src/file/mod.rs

mod checked;
mod chunk;
mod uploading;

use std::{ffi::OsStr, fs, io::Read, path::Path};

use base64::prelude::{BASE64_STANDARD_NO_PAD as BASE64, Engine};
use blake2b_simd::Params as Blake2b;

pub use checked::Checked;
pub use chunk::Chunk;
use log::{debug, warn};
pub use uploading::Uploading;

/// how many bytes to hash at once (default: 4 MiB)
const HASH_CHUNK_SIZE: usize = 4 * 1024 * 1024;

/// compute hash for a file given its path.
/// Hash function: `BLAKE2b`, 512 bit
///
/// # Params
///
/// - `path` to the file to hash
/// - `size` of that file
/// - `on_progress` will be called for each processed chunk (max. `HASH_CHUNK_SIZE`)
///
/// # Errors
///
/// - from `fs::File::{open, read}`
/// - Mismatch if given `size` does not match the file's size
fn compute_hash(path: &Path, size: u64, mut on_progress: impl FnMut(u64)) -> crate::Result<String> {
    let mut file = fs::File::open(path)?;

    // Blake2b-512 hasher (64 * 8 bit)
    let mut hasher = Blake2b::new().hash_length(64).to_state();

    // buffer
    let mut buffer = vec![0; HASH_CHUNK_SIZE];
    let mut bytes_read = 0;

    loop {
        let n = file.read(&mut buffer)?;
        if n == 0 {
            break;
        }
        hasher.update(&buffer[..n]);

        // `buf` size must be < 2 EiB
        bytes_read += n as u64;
        on_progress(n as u64);
    }

    if bytes_read != size {
        return crate::Error::mismatch(size, bytes_read);
    }

    let result = BASE64.encode(hasher.finalize());
    debug!("hashed {:?}: {result:?}", path.display());
    Ok(result)
}

/// check hash for a file given its path, return Ok(()) on success
///
/// # Params
///
/// - everything from `compute_hash`
/// - optionally, known `hash`
///
/// # Errors
///
/// - from `file::compute_hash`
/// - Mismatch if `hash` is `None`
/// - Mismatch if given `hash` does not match the computed hash
fn check_hash(
    path: &Path,
    size: u64,
    hash: Option<&str>,
    on_progress: impl FnMut(u64),
) -> crate::Result<()> {
    // check if hash is None
    let Some(expected) = hash else {
        return crate::Error::mismatch("hash", path.display());
    };

    // compute and check new hash
    let actual = &compute_hash(path, size, on_progress)?;

    if expected == actual {
        debug!("hash matches {expected:?}");
        Ok(())
    } else {
        warn!("hash mismatch for file {:?}", path.display());
        crate::Error::mismatch(expected, actual)
    }
}

pub trait FileTrait {
    /// extract the filename part of a `Path` reference
    ///
    /// # Panics
    ///
    /// Expects `path::Path::file_name` and `ffi::OsStr::to_str` to succeed on the given path
    fn extract_file_name(p: &Path) -> &str {
        p.file_name()
            .and_then(OsStr::to_str)
            .expect("bad file name")
    }

    /// get a reference to the file's name
    ///
    /// Uses `file::FileTrait::extract_file_name`, which may **panic**!
    fn get_name(&self) -> &str;

    /// get the file's size
    fn get_size(&self) -> u64;

    /// check this file's hash, return Ok(()) on success
    ///
    /// # Errors
    ///
    /// - from `file::check_hash`
    fn check_hash(&self, on_progress: impl FnMut(u64)) -> crate::Result<()>;
}

#[cfg(test)]
mod tests {
    use crate::test_util::{
        create_file,
        data::{DATA_LENGTHS_BAD, HASHES_STD_BAD, HASHES_STD_GOOD, cases, cases_with},
    };

    use super::*;

    #[test]
    fn compute_hash_as_expected() {
        for (content, size, expected_hash) in cases_with(HASHES_STD_GOOD) {
            let file = create_file(content);

            // to capture progress updates from `compute_hash`
            let mut read_total = 0;
            let callback = |n| read_total += n;
            let hash = compute_hash(file.path(), size, callback).unwrap();

            assert_eq!(hash, expected_hash);
            assert_eq!(read_total, size);
        }
    }

    #[test]
    fn hash_size_mismatch() {
        for (content, good_size, bad_size) in cases_with(DATA_LENGTHS_BAD) {
            let file = create_file(content);

            {
                // `compute_hash` with bad size
                let err = compute_hash(file.path(), bad_size, drop).unwrap_err();
                assert!(err.is_mismatch(bad_size.to_string(), good_size.to_string()));
            }

            {
                // `check_hash` with bad size
                let err = check_hash(file.path(), bad_size, Some("foobar"), drop).unwrap_err();
                assert!(err.is_mismatch(bad_size.to_string(), good_size.to_string()));
            }
        }
    }

    #[test]
    fn hash_value_none() {
        for (content, size) in cases() {
            let file = create_file(content);

            // `check_hash` with no hash
            let err = check_hash(file.path(), size, None, drop).unwrap_err();
            assert!(err.is_mismatch("hash", file.path().display().to_string()));
        }
    }

    #[test]
    fn hash_value_mismatch() {
        for ((content, size, good_hash), bad_hash) in
            cases_with(HASHES_STD_GOOD).zip(HASHES_STD_BAD)
        {
            let file = create_file(content);

            // `check_hash` with bad hash
            let err = check_hash(file.path(), size, Some(bad_hash), drop).unwrap_err();
            assert!(err.is_mismatch(bad_hash, good_hash));
        }
    }
}