group-actor/src/command.rs

use once_cell::sync::Lazy;
use regex::{Regex, RegexSetBuilder};

#[derive(Debug, Clone, PartialEq)]
pub enum StatusCommand {
    Boost,
    Ignore,
    BanUser(String),
    UnbanUser(String),
    BanServer(String),
    UnbanServer(String),
    AddMember(String),
    RemoveMember(String),
    GrantAdmin(String),
    RemoveAdmin(String),
    Announce(String),
    OpenGroup,
    CloseGroup,
    Help,
    ListMembers,
    Leave,
}

macro_rules! p_user {
    () => {
        r"(@?[a-zA-Z0-9_.-]+@[a-zA-Z0-9_.-]+\.[a-z0-9_-]+|@[a-zA-Z0-9_.-]+)"
    }
}
macro_rules! p_server {
    () => {
        r"([a-zA-Z0-9_.-]+\.[a-zA-Z0-9_-]+)"
    }
}

macro_rules! command {
    ($($val:expr),+) => {
        Regex::new(concat!(r"(?:^|\s|>|\n)[\\/]", $($val,)+ r"(?:$|[!,]|\W)")).unwrap()
    }
}

static RE_BOOST: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"b(?:oost)?")
});

static RE_IGNORE: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"i(?:g(?:n(?:ore)?)?)?")
});

static RE_BAN_USER: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"ban\s+", p_user!())
});

static RE_UNBAN_USER: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"unban\s+", p_user!())
});

static RE_BAN_SERVER: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"ban\s+", p_server!())
});

static RE_UNBAN_SERVER: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"unban\s+", p_server!())
});

static RE_ADD_MEMBER: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"(?:add)\s+", p_user!())
});

static RE_REMOVE_MEMBER: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"(?:kick|remove)\s+", p_user!())
});

static RE_GRANT_ADMIN: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"(?:op|admin)\s+", p_user!())
});

static RE_REVOKE_ADMIN: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"(?:deop|deadmin)\s+", p_user!())
});

static RE_OPEN_GROUP: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"opengroup")
});

static RE_CLOSE_GROUP: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"closegroup")
});

static RE_HELP: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"help")
});

static RE_MEMBERS: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"(?:members)")
});

static RE_LEAVE: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    command!(r"(?:leave)")
});

static RE_ANNOUNCE: once_cell::sync::Lazy<Regex> = Lazy::new(|| {
    Regex::new(concat!(r"(?:^|\s|>|\n)[\\/]announce\s+(.*)$")).unwrap()
});

pub fn parse_status(content: &str) -> Vec<StatusCommand> {
    debug!("Raw content: {}", content);

    let content = content.replace("<br/>", " ");
    // let content = content.replace("<br />", " ");
    // let content = content.replace("<BR/>", " ");
    // let content = content.replace("<BR />", " ");

    let content = voca_rs::strip::strip_tags(&content);
    debug!("Stripped tags: {}", content);

    // short-circuiting commands

    if RE_IGNORE.is_match(&content) {
        debug!("IGNORE");
        return vec![StatusCommand::Ignore];
    }

    if RE_HELP.is_match(&content) {
        debug!("HELP");
        return vec![StatusCommand::Help];
    }

    // additive commands

    let mut commands = vec![];

    // one-use commands

    if RE_BOOST.is_match(&content) {
        debug!("BOOST");
        commands.push(StatusCommand::Boost);
    }

    if RE_LEAVE.is_match(&content) {
        debug!("LEAVE");
        commands.push(StatusCommand::Leave);
    }

    if RE_MEMBERS.is_match(&content) {
        debug!("MEMBERS");
        commands.push(StatusCommand::ListMembers);
    }

    if RE_OPEN_GROUP.is_match(&content) {
        debug!("OPEN GROUP");
        commands.push(StatusCommand::OpenGroup);
    }

    if RE_CLOSE_GROUP.is_match(&content) {
        debug!("CLOSE GROUP");
        commands.push(StatusCommand::CloseGroup);
    }

    if let Some(c) = RE_ANNOUNCE.captures(&content) {
        if let Some(s) = c.get(1) {
            let s = s.as_str().trim();
            debug!("ANNOUNCE: «{}»", s);
            commands.push(StatusCommand::Announce(s.to_owned()));
        }
    }

    // multi-occurence commands

    for c in RE_BAN_USER.captures_iter(&content) {
        if let Some(s) = c.get(1) {
            let s = s.as_str();
            let s = s.trim_start_matches('@');
            debug!("BAN USER: {}", s);
            commands.push(StatusCommand::BanUser(s.to_owned()));
        }
    }

    for c in RE_UNBAN_USER.captures_iter(&content) {
        if let Some(s) = c.get(1) {
            let s = s.as_str();
            let s = s.trim_start_matches('@');
            debug!("UNBAN USER: {}", s);
            commands.push(StatusCommand::UnbanUser(s.to_owned()));
        }
    }

    for c in RE_BAN_SERVER.captures_iter(&content) {
        if let Some(s) = c.get(1) {
            debug!("BAN SERVER: {}", s.as_str());
            commands.push(StatusCommand::BanServer(s.as_str().to_owned()));
        }
    }

    for c in RE_UNBAN_SERVER.captures_iter(&content) {
        if let Some(s) = c.get(1) {
            debug!("UNBAN SERVER: {}", s.as_str());
            commands.push(StatusCommand::UnbanServer(s.as_str().to_owned()));
        }
    }

    for c in RE_ADD_MEMBER.captures_iter(&content) {
        if let Some(s) = c.get(1) {
            let s = s.as_str();
            let s = s.trim_start_matches('@');
            debug!("ADD MEMBER: {}", s);
            commands.push(StatusCommand::AddMember(s.to_owned()));
        }
    }

    for c in RE_REMOVE_MEMBER.captures_iter(&content) {
        if let Some(s) = c.get(1) {
            let s = s.as_str();
            let s = s.trim_start_matches('@');
            debug!("UNBAN USER: {}", s);
            commands.push(StatusCommand::RemoveMember(s.to_owned()));
        }
    }

    for c in RE_GRANT_ADMIN.captures_iter(&content) {
        if let Some(s) = c.get(1) {
            let s = s.as_str();
            let s = s.trim_start_matches('@');
            debug!("ADD ADMIN: {}", s);
            commands.push(StatusCommand::GrantAdmin(s.to_owned()));
        }
    }

    for c in RE_REVOKE_ADMIN.captures_iter(&content) {
        if let Some(s) = c.get(1) {
            let s = s.as_str();
            let s = s.trim_start_matches('@');
            debug!("REMOVE ADMIN: {}", s);
            commands.push(StatusCommand::RemoveAdmin(s.to_owned()));
        }
    }

    commands
}

#[cfg(test)]
mod test {
    use crate::command::{parse_status, StatusCommand};

    use super::{RE_ADD_MEMBER, RE_ANNOUNCE, RE_BAN_SERVER, RE_BAN_USER, RE_BOOST, RE_CLOSE_GROUP, RE_GRANT_ADMIN,
                RE_HELP, RE_IGNORE, RE_LEAVE, RE_MEMBERS, RE_OPEN_GROUP, RE_REMOVE_MEMBER, RE_REVOKE_ADMIN};

    #[test]
    fn test_boost() {
        assert!(RE_BOOST.is_match("/b"));
        assert!(RE_BOOST.is_match(">/b"));
        assert!(RE_BOOST.is_match("/b mm"));
        assert!(RE_BOOST.is_match("/b."));
        assert!(RE_BOOST.is_match("\\b"));
        assert!(!RE_BOOST.is_match("boo/b"));
        assert!(RE_BOOST.is_match("bla\n/b"));
        assert!(RE_BOOST.is_match("/boost"));
        assert!(RE_BOOST.is_match("/boost\n"));
        assert!(RE_BOOST.is_match("/boost dfdfg"));
        assert!(!RE_BOOST.is_match("/boosty"));
        assert!(RE_BOOST.is_match("/b\nxxx"));
        assert!(!RE_BOOST.is_match("/bleble\n"));
    }

    #[test]
    fn test_ignore() {
        assert!(RE_IGNORE.is_match("/i"));
        assert!(RE_IGNORE.is_match("/ig"));
        assert!(RE_IGNORE.is_match("/ign"));
        assert!(RE_IGNORE.is_match("/i mm"));
        assert!(RE_IGNORE.is_match("/i."));
        assert!(RE_IGNORE.is_match("\\i"));
        assert!(!RE_IGNORE.is_match("boo/i"));
        assert!(RE_IGNORE.is_match("bla\n/i"));
        assert!(RE_IGNORE.is_match("/ignore"));
        assert!(RE_IGNORE.is_match("/ignore x"));
        assert!(RE_IGNORE.is_match("/ignore\n"));
        assert!(RE_IGNORE.is_match("/ignore dfdfg"));
        assert!(!RE_IGNORE.is_match("/ignorey"));
        assert!(RE_IGNORE.is_match("/i\nxxx"));
        assert!(!RE_IGNORE.is_match("/ileble\n"));
        assert!(!RE_IGNORE.is_match("/ignx"));
    }

    #[test]
    fn test_ban_user() {
        assert!(RE_BAN_USER.is_match("/ban lain@pleroma.soykaf.com"));
        assert!(RE_BAN_USER.is_match("/ban lain@stupidname.uk"));
        assert!(RE_BAN_USER.is_match("bababababa /ban lain@stupidname.uk lala"));
        assert!(!RE_BAN_USER.is_match("/ban stupidname.uk"));

        assert!(RE_BAN_USER.is_match("/ban @lain"));
        assert!(RE_BAN_USER.is_match("/ban @lain aaa"));

        assert!(RE_BAN_USER.is_match("/ban        \t lain@pleroma.soykaf.com"));
        assert!(RE_BAN_USER.is_match("/ban @lain@pleroma.soykaf.com"));
        assert!(RE_BAN_USER.is_match("/ban @l-a_i.n9@xn--999pleroma-weirdname.com"));
        assert!(RE_BAN_USER.is_match("/ban @LAIN@PleromA.soykaf.com"));

        let c = RE_BAN_USER.captures("/ban lain@pleroma.soykaf.com");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "lain@pleroma.soykaf.com");

        let c = RE_BAN_USER.captures("/ban lain@pleroma.soykaf.com xx");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "lain@pleroma.soykaf.com");

        let c = RE_BAN_USER.captures("/ban @lain@pleroma.soykaf.com");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "@lain@pleroma.soykaf.com");

        let c = RE_BAN_USER.captures("/ban @lain");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "@lain");

        let c = RE_BAN_USER.captures("/ban @lain xx");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "@lain");
    }

    #[test]
    fn test_ban_server() {
        assert!(!RE_BAN_SERVER.is_match("/ban lain@pleroma.soykaf.com"));
        assert!(RE_BAN_SERVER.is_match("/ban pleroma.soykaf.com"));
        assert!(RE_BAN_SERVER.is_match("/ban xn--999pleroma-weirdname.com"));
        assert!(RE_BAN_SERVER.is_match("/ban           \t           xn--999pleroma-weirdname.com"));
        assert!(RE_BAN_SERVER.is_match("mamama /ban pleroma.soykaf.com momomo"));
        assert!(!RE_BAN_SERVER.is_match("/ban @pleroma.soykaf.com"));

        let c = RE_BAN_SERVER.captures("/ban pleroma.soykaf.com");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "pleroma.soykaf.com");

        let c = RE_BAN_SERVER.captures("/ban pleroma.soykaf.com xx");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "pleroma.soykaf.com");
    }

    #[test]
    fn test_add_member() {
        assert!(RE_ADD_MEMBER.is_match("/add lain@pleroma.soykaf.com"));
        assert!(RE_ADD_MEMBER.is_match("/add @lain@pleroma.soykaf.com"));
        assert!(RE_ADD_MEMBER.is_match("\\add @lain"));

        let c = RE_ADD_MEMBER.captures("/add @lain");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "@lain");
    }

    #[test]
    fn test_remove_member() {
        assert!(!RE_REMOVE_MEMBER.is_match("/admin lain@pleroma.soykaf.com"));

        assert!(RE_REMOVE_MEMBER.is_match("/remove lain@pleroma.soykaf.com"));
        assert!(RE_REMOVE_MEMBER.is_match("/remove @lain@pleroma.soykaf.com"));
        assert!(RE_REMOVE_MEMBER.is_match("\\remove @lain"));

        assert!(RE_REMOVE_MEMBER.is_match("/kick @lain"));
        assert!(RE_REMOVE_MEMBER.is_match("/remove @lain"));

        let c = RE_REMOVE_MEMBER.captures("/kick lain@pleroma.soykaf.com");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "lain@pleroma.soykaf.com");
    }

    #[test]
    fn test_add_admin() {
        assert!(!RE_GRANT_ADMIN.is_match("/expel lain@pleroma.soykaf.com"));

        assert!(RE_GRANT_ADMIN.is_match("/admin lain@pleroma.soykaf.com"));
        assert!(RE_GRANT_ADMIN.is_match("/op @lain@pleroma.soykaf.com"));
        assert!(RE_GRANT_ADMIN.is_match("\\op @lain"));

        let c = RE_GRANT_ADMIN.captures("/op @lain@pleroma.soykaf.com");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "@lain@pleroma.soykaf.com");
    }

    #[test]
    fn test_remove_admin() {
        assert!(!RE_REVOKE_ADMIN.is_match("/admin lain@pleroma.soykaf.com"));

        assert!(RE_REVOKE_ADMIN.is_match("/deop @lain"));
        assert!(RE_REVOKE_ADMIN.is_match("/deadmin @lain"));

        let c = RE_REVOKE_ADMIN.captures("/deadmin @lain");
        assert!(c.is_some());
        assert_eq!(c.unwrap().get(1).unwrap().as_str(), "@lain");
    }

    #[test]
    fn test_opengroup() {
        assert!(!RE_OPEN_GROUP.is_match("/admin lain@pleroma.soykaf.com"));
        assert!(RE_OPEN_GROUP.is_match("/opengroup"));
        assert!(RE_OPEN_GROUP.is_match("x /opengroup"));
        assert!(RE_OPEN_GROUP.is_match("/opengroup dfgdfg"));
        assert!(RE_OPEN_GROUP.is_match("\n\n/opengroup\n dfgdfg\n\n"));
    }

    #[test]
    fn test_closegroup() {
        assert!(!RE_CLOSE_GROUP.is_match("/admin lain@pleroma.soykaf.com"));
        assert!(RE_CLOSE_GROUP.is_match("/closegroup"));
        assert!(RE_CLOSE_GROUP.is_match("x /closegroup"));
        assert!(RE_CLOSE_GROUP.is_match("/closegroup dfgdfg"));
        assert!(RE_CLOSE_GROUP.is_match("\n\n/closegroup\n dfgdfg\n\n"));
    }

    #[test]
    fn test_help() {
        assert!(!RE_HELP.is_match("/admin lain@pleroma.soykaf.com"));
        assert!(RE_HELP.is_match("/help"));

        assert!(!RE_HELP.is_match("/helpx"));
        assert!(!RE_HELP.is_match("a/help"));
        assert!(!RE_HELP.is_match("help"));

        assert!(RE_HELP.is_match("x /help"));
        assert!(RE_HELP.is_match("/help dfgdfg"));
        assert!(RE_HELP.is_match("\n\n/help\n dfgdfg\n\n"));
    }

    #[test]
    fn test_members() {
        assert!(!RE_MEMBERS.is_match("/admin lain@pleroma.soykaf.com"));
        assert!(RE_MEMBERS.is_match("/members"));
    }

    #[test]
    fn test_leave() {
        assert!(!RE_LEAVE.is_match("/list"));
        assert!(RE_LEAVE.is_match("/leave"));
        assert!(RE_LEAVE.is_match("/leave"));
        assert!(RE_LEAVE.is_match("x /leave"));
        assert!(RE_LEAVE.is_match("/leave z"));
    }

    #[test]
    fn test_announce() {
        assert!(!RE_ANNOUNCE.is_match("/list"));
        assert!(RE_ANNOUNCE.is_match("sdfsdffsd /announce b"));
        assert!(RE_ANNOUNCE.is_match("/announce bla bla bla"));
        assert!(RE_ANNOUNCE.is_match("sdfsdffsd /announce bla bla bla"));
        assert_eq!("bla bla bla", RE_ANNOUNCE.captures("sdfsdffsd /announce bla bla bla").unwrap().get(1).unwrap().as_str());
    }

    #[test]
    fn test_real_post() {
        assert_eq!(Vec::<StatusCommand>::new(), parse_status("Hello there is nothing here /fake command"));
        assert_eq!(vec![StatusCommand::Help], parse_status("lets see some \\help and /ban @lain"));
        assert_eq!(vec![StatusCommand::Ignore], parse_status("lets see some /ignore and /ban @lain"));
        assert_eq!(vec![
            StatusCommand::BanUser("lain".to_string()),
            StatusCommand::BanUser("piggo@piggo.space".to_string()),
            StatusCommand::BanServer("soykaf.com".to_string())
        ],
                   parse_status("let's /ban @lain! /ban @piggo@piggo.space and also /ban soykaf.com"));
    }

    #[test]
    fn test_strip() {
        assert_eq!(vec![StatusCommand::BanUser("betty".to_string())],
                   parse_status(r#"Let's bad the naughty bot: /ban <span class="h-card"><a class="u-url mention" data-user="9nXpaGZL88fPAiP8xU" href="https://piggo.space/users/betty" rel="ugc">@<span>betty</span></a></span>"#));
        assert_eq!(vec![StatusCommand::BanUser("betty@abstarbauze.com".to_string())],
                   parse_status(r#"Let's bad the naughty bot: /ban <span class="h-card"><a class="u-url mention" data-user="9nXpaGZL88fPAiP8xU" href="https://piggo.space/users/betty" rel="ugc">@<span>betty@abstarbauze.com</span></a></span>"#));
    }
}