let encoding = pijul::change::get_encoding(&data);

        let encoding = detect_encoding(&data);

// Detection logic borrowed from <https://github.com/Wilfred/difftastic/blob/1436c8eac39dcea07f8c24a8128284a25b416e8d/src/files.rs#L141>.
/// Binary data returns `None`
fn detect_encoding(data: &[u8]) -> Option<pijul::Encoding> {
    if let Ok(_) = std::str::from_utf8(data) {
        return Some(pijul::Encoding(encoding_rs::UTF_8));
    }
    // Only consider the first 1,000 bytes, as tree_magic_mini
    // considers the entire file, which is very slow on large files.
    let mut magic_bytes = data;
    if magic_bytes.len() > 1000 {
        magic_bytes = &magic_bytes[..1000];
    }
    let mime = tree_magic_mini::from_u8(magic_bytes);

    // Use MIME type detection to guess whether a file is binary. This
    // has false positives and false negatives, so only check the MIME
    // type after allowing perfect text files (see issue #433).
    match mime {
        // Treat pdf as binary.
        "application/pdf" => return None,
        // application/* is a mix of stuff, application/json is fine
        // but application/zip is binary that often decodes as valid
        // UTF-16.
        //
        // See
        // <https://developer.mozilla.org/en-US/docs/Web/HTTP/Guides/MIME_types/Common_types>
        // for a list of MIME types.
        "application/x-archive" => return None,
        "application/x-bzip" => return None,
        "application/x-bzip2" => return None,
        "application/x-7zip-compressed" => return None,
        "application/gzip" => return None,
        "application/zip" => return None,
        "application/zstd" => return None,
        // Treat all image content as binary.
        v if v.starts_with("image/") => return None,
        // Treat all audio content as binary.
        v if v.starts_with("audio/") => return None,
        // Treat all video content as binary.
        v if v.starts_with("video/") => return None,
        // Treat all font content as binary.
        v if v.starts_with("font/") => return None,
        _ => {}
    }
    // If the input bytes are *almost* valid UTF-8, treat them as
    // UTF-8. This is helpful when the user has written a small number
    // of bad bytes to a file. Users would still like to be able to
    // diff these files.
    let utf8_string = String::from_utf8_lossy(data).to_string();
    let num_utf8_invalid = utf8_string
        .chars()
        .take(50000)
        .filter(|c| *c == std::char::REPLACEMENT_CHARACTER || *c == '\0')
        .count();
    if num_utf8_invalid <= 2 {
        return Some(pijul::Encoding(encoding_rs::UTF_8));
    }
    // Fallback to pijul encoding detection
    pijul::change::get_encoding(&data)
}

workspace = true
[dependencies.encoding_rs]

workspace = true
[dependencies.tree_magic_mini]

[workspace.dependencies.encoding_rs]
version = "0.8"

[workspace.dependencies.tree_magic_mini]
version = "3"

[[package]]
name = "fixedbitset"
version = "0.5.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1d674e81391d1e1ab681a28d99df07927c6d4aa5b027d7da16ba32d1d21ecd99"

 "encoding_rs",

 "tree_magic_mini",

[[package]]
name = "petgraph"
version = "0.8.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8701b58ea97060d5e5b155d383a69952a60943f0e6dfe30b04c287beb0b27455"
dependencies = [
 "fixedbitset",
 "hashbrown 0.15.4",
 "indexmap",
]

name = "tree_magic_mini"
version = "3.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b8765b90061cba6c22b5831f675da109ae5561588290f9fa2317adab2714d5a6"
dependencies = [
 "memchr",
 "nom 8.0.0",
 "petgraph",
]
[[package]]