aria2-rust-pro/crates/aria2-rust-pro-compat/src/config.rs

//! Config parsing and normalization helpers for aria2-style option files.

use crate::options::{OptionScope, option_spec, reserved_option_names};

/// Where a compat config document originated.
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
pub enum ConfigLocationKind {
    /// The document was loaded from a file on disk.
    File,
    /// The document was provided inline as raw text.
    Inline,
    /// The document originated from an RPC payload.
    Rpc,
    /// The document originated from environment variables.
    Env,
    /// The document originated from CLI arguments.
    Cli,
}
/// Scope inferred for a config document or directive set.
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
pub enum ConfigScope {
    /// Only global options are expected.
    Global,
    /// Only per-download options are expected.
    PerDownload,
    /// The document can contain both global and per-download options.
    Mixed,
}
/// Source category for parsed config data.
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
pub enum ConfigSource {
    /// User-authored config file such as `aria2.conf`.
    UserConfig,
    /// Persisted session file content.
    SessionFile,
    /// Input-file content that expands downloads.
    InputFile,
    /// Runtime overrides originating from transient inputs.
    RuntimeOverride,
    /// Named profile content.
    Profile,
}

/// A single `name=value` config directive.
#[derive(Clone, Debug, Eq, PartialEq)]
pub struct ConfigDirective {
    /// Directive name after canonicalization.
    pub name: String,
    /// Optional directive value.
    pub value: Option<String>,
}
/// Parsed config directives with source metadata.
#[derive(Clone, Debug, Eq, PartialEq)]
pub struct ConfigAst {
    /// Ordered directives found in the source input.
    pub directives: Vec<ConfigDirective>,
    /// Origin category for the parsed directives.
    pub source: ConfigSource,
}
/// Parsed config document with location metadata.
#[derive(Clone, Debug, Eq, PartialEq)]
pub struct ConfigDocument {
    /// Ordered directives found in the document.
    pub directives: Vec<ConfigDirective>,
    /// Where the document was loaded from.
    pub location: ConfigLocationKind,
    /// Scope classification for the document.
    pub scope: ConfigScope,
}
/// Session save/load compatibility metadata.
#[derive(Clone, Debug, Eq, PartialEq)]
pub struct SessionFileModel {
    /// Path where the session file is stored.
    pub session_path: String,
    /// Optional input file associated with the session.
    pub input_path: Option<String>,
    /// Autosave interval for session persistence.
    pub autosave_interval_secs: u64,
}
/// Named profile loaded through compat config handling.
#[derive(Clone, Debug, Eq, PartialEq)]
pub struct ConfigProfile {
    /// Profile name.
    pub name: String,
    /// Source category for the profile.
    pub source: ConfigSource,
    /// Parsed profile document.
    pub document: ConfigDocument,
}

/// Errors produced while parsing compat config input.
#[derive(Clone, Debug, Eq, PartialEq)]
pub enum ConfigParseError {
    /// A line could not be parsed as a valid directive.
    InvalidDirective(String),
    /// A directive that requires a value omitted one.
    MissingValue(String),
    /// A directive named an unknown option.
    UnknownOption(String),
    /// A directive used an option name reserved for internal use.
    ReservedOption(String),
}
impl std::fmt::Display for ConfigParseError {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            Self::InvalidDirective(line) => write!(f, "invalid config directive: {line}"),
            Self::MissingValue(name) => write!(f, "missing value for option: {name}"),
            Self::UnknownOption(name) => write!(f, "unknown option: {name}"),
            Self::ReservedOption(name) => write!(f, "reserved option name: {name}"),
        }
    }
}
impl std::error::Error for ConfigParseError {}

/// Canonicalizes a config option name through the option registry.
fn canonicalize_option_name(name: &str) -> String {
    let name = normalize_option_token(name);
    option_spec(name).map_or_else(|| name.to_owned(), |spec| spec.name.to_owned())
}

/// Removes a UTF-8 byte-order mark when one prefixes a text fragment.
fn strip_utf8_bom(text: &str) -> &str {
    text.strip_prefix('\u{feff}').unwrap_or(text)
}

/// Normalizes a CLI- or config-style option token to its raw name form.
fn normalize_option_token(name: &str) -> &str {
    let name = strip_utf8_bom(name.trim());
    name.strip_prefix("--").unwrap_or_else(|| {
        name.strip_prefix('-')
            .filter(|value| !value.is_empty())
            .map_or(name, |name| name)
    })
}

/// Removes trailing comments that are safely separated from a value.
fn strip_safe_trailing_comment(value: &str) -> &str {
    for (index, ch) in value.char_indices() {
        let Some(prefix) = value.get(..index) else {
            continue;
        };
        if (ch == '#' || ch == ';') && prefix.chars().next_back().is_some_and(char::is_whitespace) {
            return prefix.trim_end();
        }
    }
    value.trim()
}

/// Parses a single aria2-style config line, ignoring blank lines and comments.
///
/// # Errors
///
/// Returns [`ConfigParseError`] when the line is malformed or omits a required value.
pub fn parse_config_line(line: &str) -> Result<Option<ConfigDirective>, ConfigParseError> {
    let trimmed = strip_utf8_bom(line).trim();
    if trimmed.is_empty() || trimmed.starts_with('#') || trimmed.starts_with(';') {
        return Ok(None);
    }
    let (name, value) = trimmed
        .split_once('=')
        .ok_or_else(|| ConfigParseError::InvalidDirective(trimmed.to_owned()))?;
    let name = normalize_option_token(name);
    if name.is_empty() {
        return Err(ConfigParseError::InvalidDirective(trimmed.to_owned()));
    }
    let value = strip_safe_trailing_comment(value);
    if value.is_empty() {
        return Err(ConfigParseError::MissingValue(name.to_owned()));
    }
    Ok(Some(ConfigDirective {
        name: canonicalize_option_name(name),
        value: Some(value.to_owned()),
    }))
}

/// Parses a single config line and enforces the known-option / non-reserved subset.
///
/// # Errors
///
/// Returns [`ConfigParseError`] when the line is malformed, reserved, or names an unknown option.
pub fn parse_config_line_strict(line: &str) -> Result<Option<ConfigDirective>, ConfigParseError> {
    let directive = parse_config_line(line)?;
    if let Some(ref d) = directive {
        if reserved_option_names().iter().any(|r| r.name == d.name) {
            return Err(ConfigParseError::ReservedOption(d.name.clone()));
        }
        if option_spec(d.name.as_str()).is_none() {
            return Err(ConfigParseError::UnknownOption(d.name.clone()));
        }
    }
    Ok(directive)
}

/// Parses a full config document using strict option validation.
///
/// # Errors
///
/// Returns [`ConfigParseError`] when any non-comment line is malformed, reserved, or unknown.
pub fn parse_config(text: &str) -> Result<Vec<ConfigDirective>, ConfigParseError> {
    let mut directives = Vec::new();
    for line in text.lines() {
        if let Some(d) = parse_config_line_strict(line)? {
            directives.push(d);
        }
    }
    Ok(directives)
}
/// Parses a full config document while tolerating unknown options.
///
/// # Errors
///
/// Returns [`ConfigParseError`] when any non-comment line is malformed or omits a required value.
pub fn parse_config_lenient(text: &str) -> Result<Vec<ConfigDirective>, ConfigParseError> {
    let mut directives = Vec::new();
    for line in text.lines() {
        if let Some(d) = parse_config_line(line)? {
            directives.push(d);
        }
    }
    Ok(directives)
}

/// Infers config scope for a named option when the option is known.
#[must_use]
pub fn infer_scope(name: &str) -> Option<ConfigScope> {
    option_spec(name).map(|s| match s.metadata.scope {
        OptionScope::Global => ConfigScope::Global,
        OptionScope::PerDownload => ConfigScope::PerDownload,
        OptionScope::Both => ConfigScope::Mixed,
    })
}

/// Returns a document whose directive names have been canonicalized.
#[must_use]
pub fn normalize_document(mut document: ConfigDocument) -> ConfigDocument {
    for directive in &mut document.directives {
        directive.name = canonicalize_option_name(&directive.name);
    }
    document
}

/// Loads an inline named profile using lenient parsing semantics.
///
/// # Errors
///
/// Returns [`ConfigParseError`] when the profile text contains malformed directives.
pub fn load_profile(name: &str, text: &str) -> Result<ConfigProfile, ConfigParseError> {
    let directives = parse_config_lenient(text)?;
    Ok(ConfigProfile {
        name: name.to_owned(),
        source: ConfigSource::Profile,
        document: ConfigDocument {
            directives,
            location: ConfigLocationKind::Inline,
            scope: ConfigScope::Mixed,
        },
    })
}

#[cfg(test)]
mod tests {
    use super::{
        ConfigDirective, ConfigDocument, ConfigLocationKind, ConfigParseError, ConfigScope,
        ConfigSource, load_profile, normalize_document, parse_config_line,
        parse_config_line_strict,
    };

    #[test]
    fn parse_config_line_strict_canonicalizes_known_alias_names() {
        let directive = parse_config_line_strict("http-want-digest=true")
            .expect("strict alias parse should succeed")
            .expect("directive should be present");

        assert_eq!(directive.name, "no-want-digest-header");
        assert_eq!(directive.value.as_deref(), Some("true"));
    }

    #[test]
    fn parse_config_line_lenient_canonicalizes_known_alias_names() {
        let directive = parse_config_line("http-want-digest=true")
            .expect("lenient alias parse should succeed")
            .expect("directive should be present");

        assert_eq!(directive.name, "no-want-digest-header");
        assert_eq!(directive.value.as_deref(), Some("true"));
    }

    #[test]
    fn load_profile_keeps_profile_metadata_while_canonicalizing_known_aliases() {
        let profile = load_profile(
            "demo",
            "# comment\nhttp-want-digest=true\nrpc-listen-all=true\n",
        )
        .expect("profile should load");

        assert_eq!(profile.name, "demo");
        assert_eq!(profile.source, ConfigSource::Profile);
        assert_eq!(profile.document.location, ConfigLocationKind::Inline);
        assert_eq!(profile.document.scope, ConfigScope::Mixed);
        let [first, second] = profile.document.directives.as_slice() else {
            panic!("profile should contain exactly two directives");
        };
        assert_eq!(first.name, "no-want-digest-header");
        assert_eq!(second.name, "rpc-listen-all");
    }

    #[test]
    fn parse_config_line_strict_still_rejects_unknown_options() {
        let error = parse_config_line_strict("not-a-real-option=true")
            .expect_err("unknown option should still be rejected");

        assert_eq!(
            error,
            ConfigParseError::UnknownOption("not-a-real-option".to_owned())
        );
    }

    #[test]
    fn parse_config_line_accepts_bom_and_cli_style_option_spellings() {
        let parameterized = parse_config_line_strict("\u{feff}--parameterized-uri=true")
            .expect("long CLI spelling should parse")
            .expect("directive should be present");
        assert_eq!(parameterized.name, "parameterized-uri");
        assert_eq!(parameterized.value.as_deref(), Some("true"));

        let remote_time = parse_config_line_strict("-R=true")
            .expect("short CLI alias should parse")
            .expect("directive should be present");
        assert_eq!(remote_time.name, "remote-time");
        assert_eq!(remote_time.value.as_deref(), Some("true"));
    }

    #[test]
    fn parse_config_line_strips_safe_trailing_comments_but_preserves_uri_fragments() {
        let referer = parse_config_line_strict(
            "referer=http://example.invalid/download#frag   # copied note",
        )
        .expect("referer with fragment should parse")
        .expect("directive should be present");
        assert_eq!(referer.name, "referer");
        assert_eq!(
            referer.value.as_deref(),
            Some("http://example.invalid/download#frag")
        );

        let tracker = parse_config_line("bt-tracker=udp://tracker.invalid:80/announce ; mirror")
            .expect("tracker line should parse")
            .expect("directive should be present");
        assert_eq!(tracker.name, "bt-tracker");
        assert_eq!(
            tracker.value.as_deref(),
            Some("udp://tracker.invalid:80/announce")
        );
    }

    #[test]
    fn load_profile_canonicalizes_extended_cli_spellings_and_comments() {
        let profile = load_profile(
            "compat",
            "\u{feff}--select-file=1-3,5 # keep files\n-R=true\n",
        )
        .expect("profile should load");

        let [first, second] = profile.document.directives.as_slice() else {
            panic!("profile should contain exactly two directives");
        };
        assert_eq!(first.name, "select-file");
        assert_eq!(first.value.as_deref(), Some("1-3,5"));
        assert_eq!(second.name, "remote-time");
        assert_eq!(second.value.as_deref(), Some("true"));
    }

    #[test]
    fn normalize_document_canonicalizes_cli_style_and_bom_prefixed_names() {
        let normalized = normalize_document(ConfigDocument {
            directives: vec![
                ConfigDirective {
                    name: "\u{feff}--parameterized-uri".to_owned(),
                    value: Some("true".to_owned()),
                },
                ConfigDirective {
                    name: "-R".to_owned(),
                    value: Some("true".to_owned()),
                },
            ],
            location: ConfigLocationKind::Inline,
            scope: ConfigScope::Mixed,
        });

        let [first, second] = normalized.directives.as_slice() else {
            panic!("normalized document should contain exactly two directives");
        };
        assert_eq!(first.name, "parameterized-uri");
        assert_eq!(second.name, "remote-time");
    }
}