tirith_core/

normalize.rs

// Per-component normalization: decode only unreserved characters (RFC 3986 §2.3).
// Unreserved: A-Z, a-z, 0-9, '-', '.', '_', '~'

/// Check if a byte value represents an unreserved character.
fn is_unreserved(byte: u8) -> bool {
    byte.is_ascii_alphanumeric() || matches!(byte, b'-' | b'.' | b'_' | b'~')
}

/// Decode a hex character to its value.
fn hex_val(b: u8) -> Option<u8> {
    match b {
        b'0'..=b'9' => Some(b - b'0'),
        b'a'..=b'f' => Some(b - b'a' + 10),
        b'A'..=b'F' => Some(b - b'A' + 10),
        _ => None,
    }
}

/// Decode only unreserved percent-encoded characters in a string.
/// Returns the normalized string and whether any unreserved chars were decoded.
/// Hex digits in percent-triplets are always normalized to uppercase.
fn decode_unreserved_once(input: &str) -> (String, bool) {
    let bytes = input.as_bytes();
    let mut result = Vec::with_capacity(bytes.len());
    let mut decoded_any = false;
    let mut i = 0;

    while i < bytes.len() {
        if bytes[i] == b'%' && i + 2 < bytes.len() {
            if let (Some(hi), Some(lo)) = (hex_val(bytes[i + 1]), hex_val(bytes[i + 2])) {
                let decoded_byte = (hi << 4) | lo;
                if is_unreserved(decoded_byte) {
                    result.push(decoded_byte);
                    decoded_any = true;
                    i += 3;
                    continue;
                } else {
                    // Normalize hex to uppercase but keep encoded
                    result.push(b'%');
                    result.push(bytes[i + 1].to_ascii_uppercase());
                    result.push(bytes[i + 2].to_ascii_uppercase());
                    i += 3;
                    continue;
                }
            }
            // Invalid percent-triplet, leave as-is
            result.push(bytes[i]);
            i += 1;
        } else {
            result.push(bytes[i]);
            i += 1;
        }
    }

    (String::from_utf8_lossy(&result).into_owned(), decoded_any)
}

/// Normalize a URL path component (decode unreserved chars, up to 3 rounds).
/// Returns (normalized, raw, detected_double_encoding).
pub fn normalize_path(raw: &str) -> NormalizedComponent {
    let mut current = raw.to_string();
    let mut rounds = 0;

    // Always run at least one pass (for hex case normalization),
    // then continue if unreserved chars were decoded (up to 3 rounds).
    loop {
        let (decoded, did_decode) = decode_unreserved_once(&current);
        current = decoded;
        rounds += 1;
        if !did_decode || rounds >= 3 {
            break;
        }
    }

    // Detect double-encoding: look for %25XX patterns in the final result
    // This indicates a percent-encoded percent sign that decoded to %XX
    let double_encoded = detect_double_encoding(&current);

    NormalizedComponent {
        raw: raw.to_string(),
        normalized: current,
        double_encoded,
        rounds,
    }
}

/// Normalize a query/fragment component (same treatment as path).
pub fn normalize_query(raw: &str) -> NormalizedComponent {
    normalize_path(raw)
}

/// Detect genuine double-encoding: %25XX patterns (percent-encoded percent sign).
fn detect_double_encoding(s: &str) -> bool {
    let bytes = s.as_bytes();
    if bytes.len() < 5 {
        return false;
    }
    let mut i = 0;
    while i + 4 < bytes.len() {
        if bytes[i] == b'%'
            && bytes[i + 1] == b'2'
            && bytes[i + 2] == b'5'
            && hex_val(bytes[i + 3]).is_some()
            && hex_val(bytes[i + 4]).is_some()
        {
            return true;
        }
        i += 1;
    }
    false
}

/// Result of normalization.
#[derive(Debug, Clone)]
pub struct NormalizedComponent {
    pub raw: String,
    pub normalized: String,
    pub double_encoded: bool,
    pub rounds: u32,
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_unreserved_decoded() {
        // %41 = 'A' (unreserved) -> should be decoded
        let result = normalize_path("%41");
        assert_eq!(result.normalized, "A");
    }

    #[test]
    fn test_reserved_preserved() {
        // %2F = '/' (reserved) -> should stay encoded
        let result = normalize_path("%2F");
        assert_eq!(result.normalized, "%2F");
    }

    #[test]
    fn test_reserved_at_preserved() {
        // %40 = '@' (reserved) -> should stay encoded
        let result = normalize_path("%40");
        assert_eq!(result.normalized, "%40");
    }

    #[test]
    fn test_reserved_colon_preserved() {
        // %3A = ':' (reserved) -> should stay encoded
        let result = normalize_path("%3A");
        assert_eq!(result.normalized, "%3A");
    }

    #[test]
    fn test_reserved_question_preserved() {
        // %3F = '?' (reserved) -> should stay encoded
        let result = normalize_path("%3F");
        assert_eq!(result.normalized, "%3F");
    }

    #[test]
    fn test_hex_case_normalized() {
        // %2f (lowercase) -> %2F (uppercase, still reserved)
        let result = normalize_path("%2f");
        assert_eq!(result.normalized, "%2F");
    }

    #[test]
    fn test_double_encoding_detected() {
        // %252F decodes to %2F after one round (unreserved part of %25 = '%')
        // Actually %25 is '%' which is NOT unreserved, so it stays as %25
        // %252F stays as %252F -> but we detect the %25 pattern
        let result = normalize_path("%252F");
        assert!(result.double_encoded);
    }

    #[test]
    fn test_single_level_not_double_encoded() {
        // %2F is normal, not double-encoded
        let result = normalize_path("%2F");
        assert!(!result.double_encoded);
    }

    #[test]
    fn test_mixed_encoding() {
        // %41%2F -> A%2F (A decoded, / preserved)
        let result = normalize_path("%41%2F");
        assert_eq!(result.normalized, "A%2F");
    }

    #[test]
    fn test_tilde_decoded() {
        // %7E = '~' (unreserved) -> decoded
        let result = normalize_path("%7E");
        assert_eq!(result.normalized, "~");
    }

    #[test]
    fn test_hyphen_decoded() {
        // %2D = '-' (unreserved) -> decoded
        let result = normalize_path("%2D");
        assert_eq!(result.normalized, "-");
    }

    #[test]
    fn test_dot_decoded() {
        // %2E = '.' (unreserved) -> decoded
        let result = normalize_path("%2E");
        assert_eq!(result.normalized, ".");
    }

    #[test]
    fn test_underscore_decoded() {
        // %5F = '_' (unreserved) -> decoded
        let result = normalize_path("%5F");
        assert_eq!(result.normalized, "_");
    }

    #[test]
    fn test_no_encoding() {
        let result = normalize_path("/path/to/file");
        assert_eq!(result.normalized, "/path/to/file");
        // One pass always runs (for hex case normalization), even with no encodings
        assert_eq!(result.rounds, 1);
    }

    #[test]
    fn test_invalid_percent_triplet() {
        // %GG is not valid hex -> left as-is
        let result = normalize_path("%GG");
        assert_eq!(result.normalized, "%GG");
    }

    #[test]
    fn test_multiple_rounds() {
        // %2541 -> round 1: %25 stays (not unreserved), 41 stays as part of %2541
        // Actually %2541: %25 = '%' (not unreserved, stays), then '4', '1'
        // So it stays %2541 but we detect double encoding
        let result = normalize_path("%2541");
        assert!(result.double_encoded);
    }
}