Function decode_utf8_chars

alpm-mtree/src/path_decoder.rs:30–72

Decodes UTF-8 characters from a string using MTREE-specific escape sequences. MTREE uses various decodings. 1. the VIS_CSTYLE encoding of `strsvis(3)`, which encodes a specific set of characters. Of these, only the following control characters are allowed in filenames: - \s Space - \t Tab - \r Carriage Return - \n Line Feed 2. `#` is encoded as `\#` to differentiate between comments. 3. For all o

(input: &mut &str)

/// have convenient backtracking and error messages in case we encounter invalid escape
/// sequences or malformed escaped UTF-8.
pub fn decode_utf8_chars(input: &mut &str) -> ModalResult<String> {
    // This is the string we'll accumulated the decoded path into.
    let mut path = String::new();

    loop {
        // Parse the string until we hit a `\`
        let part = take_while(0.., |c| c != '\\').parse_next(input)?;
        path.push_str(part);

        if input.is_empty() {
            break;
        }

        // We hit a `\`. See if it's an expected escape sequence.
        // If none of the expected sequences are encountered, fail and throw an error.
        let escaped = alt((
            "\\s".map(|s: &str| s.to_string()),
            "\\t".map(|s: &str| s.to_string()),
            "\\r".map(|s: &str| s.to_string()),
            "\\n".map(|s: &str| s.to_string()),
            "\\#".map(|s: &str| s.to_string()),
            unicode_char,
            fail.context(StrContext::Label("escape sequence"))
                .context(StrContext::Expected(StrContextValue::Description(
                    "VIS_CSTYLE encoding or encoded octal triplets for unicode chars.",
                ))),
        ))
        .parse_next(input)?;

        let unescaped = match escaped.as_str() {
            "\\s" => " ".to_string(),
            "\\t" => "\t".to_string(),
            "\\r" => "\r".to_string(),
            "\\n" => "\n".to_string(),
            "\\#" => "#".to_string(),
            _ => escaped,
        };

        path.push_str(&unescaped);
    }

    Ok(path)
}

/// Parse and convert a single octal triplet string into a byte.
///