huskies/server/src/chat/transport/matrix/bot/mentions.rs

//! Matrix mention detection — checks whether a message mentions the bot.
use matrix_sdk::ruma::events::room::message::{Relation, RoomMessageEventContentWithoutRelation};
use matrix_sdk::ruma::{OwnedEventId, OwnedUserId};
use std::collections::HashSet;
use tokio::sync::Mutex as TokioMutex;

/// Returns `true` if the message mentions the bot.
///
/// Checks both the plain-text `body` and an optional `formatted_body` (HTML).
/// Recognised forms:
/// - The bot's full Matrix user ID (e.g. `@timmy:homeserver.local`) in either body
/// - The localpart with `@` prefix (e.g. `@timmy`) with word-boundary check
/// - A `matrix.to` link containing the user ID (in `formatted_body`)
///
/// Short mentions are only counted when not immediately followed by an
/// alphanumeric character, hyphen, or underscore to avoid false positives.
pub fn mentions_bot(body: &str, formatted_body: Option<&str>, bot_user_id: &OwnedUserId) -> bool {
    let full_id = bot_user_id.as_str();
    let localpart = bot_user_id.localpart();

    // Check formatted_body for a matrix.to link containing the bot's user ID.
    if formatted_body.is_some_and(|html| html.contains(full_id)) {
        return true;
    }

    // Check plain body for the full ID.
    if body.contains(full_id) {
        return true;
    }

    // Check plain body for @localpart (e.g. "@timmy") with word boundaries.
    if contains_word(body, &format!("@{localpart}")) {
        return true;
    }

    false
}

/// Returns `true` if `haystack` contains `needle` at a word boundary.
#[allow(clippy::string_slice)] // all indices from find() or abs+needle.len() → always char boundaries
pub(super) fn contains_word(haystack: &str, needle: &str) -> bool {
    let mut start = 0;
    while let Some(rel) = haystack[start..].find(needle) {
        let abs = start + rel;
        let after = abs + needle.len();
        let next = haystack[after..].chars().next();
        let is_word_end = next.is_none_or(|c| !c.is_alphanumeric() && c != '-' && c != '_');
        if is_word_end {
            return true;
        }
        start = abs + 1;
    }
    false
}

/// Returns `true` if the message's `relates_to` field references an event that
/// the bot previously sent (i.e. the message is a reply or thread-reply to a
/// bot message).
pub(super) async fn is_reply_to_bot(
    relates_to: Option<&Relation<RoomMessageEventContentWithoutRelation>>,
    bot_sent_event_ids: &TokioMutex<HashSet<OwnedEventId>>,
) -> bool {
    let candidate_ids: Vec<&OwnedEventId> = match relates_to {
        Some(Relation::Reply { in_reply_to }) => vec![&in_reply_to.event_id],
        Some(Relation::Thread(thread)) => {
            let mut ids = vec![&thread.event_id];
            if let Some(irti) = &thread.in_reply_to {
                ids.push(&irti.event_id);
            }
            ids
        }
        _ => return false,
    };
    let guard = bot_sent_event_ids.lock().await;
    candidate_ids.iter().any(|id| guard.contains(*id))
}

/// Returns `true` when the message body appears to be explicitly addressed to
/// someone **other** than this bot.
///
/// Recognised address patterns at the start of the body:
/// - `"name: rest"` — display-name style (e.g. `"sally: do X"`)
/// - `"@name rest"` — @ mention style (e.g. `"@sally do X"`)
///
/// A message is only considered addressed to another party when the name does
/// **not** match either the bot's `bot_name` (case-insensitive) or the
/// localpart of its `bot_user_id`.
///
/// Used in ambient mode to suppress responses when a message is clearly
/// directed at a different participant (e.g. another bot in the same room).
#[allow(clippy::string_slice)] // word_end and colon_pos from find() → always char boundaries
pub fn is_addressed_to_other(body: &str, bot_user_id: &OwnedUserId, bot_name: &str) -> bool {
    let trimmed = body.trim_start();
    let lower = trimmed.to_lowercase();
    let bot_name_lower = bot_name.to_lowercase();
    let bot_localpart = bot_user_id.localpart().to_lowercase();

    // Pattern A: "@name …" at the start of the message.
    // Handles both "@localpart" and "@localpart:homeserver" forms.
    if let Some(rest) = lower.strip_prefix('@') {
        // Extract everything up to the first whitespace character.
        let word_end = rest.find(|c: char| c.is_whitespace()).unwrap_or(rest.len());
        let mention = &rest[..word_end]; // e.g. "sally" or "sally:example.com"

        // Strip the homeserver part to get just the localpart.
        let localpart = mention.split(':').next().unwrap_or(mention);

        if localpart.is_empty() {
            return false; // bare "@" — not an address
        }
        if localpart == bot_localpart {
            return false; // addressed to us
        }
        return true; // addressed to someone else
    }

    // Pattern B: "name: rest" — display-name style.
    // Only the text before the *first* colon is inspected.  We require that
    // the prefix contains no spaces so that ordinary sentences such as
    // "Here is a question: …" are not misread as bot addresses.
    if let Some(colon_pos) = lower.find(':') {
        let prefix = &lower[..colon_pos];

        // Single-word prefix (no spaces).
        if !prefix.contains(' ') && !prefix.is_empty() {
            if prefix == bot_name_lower || prefix == bot_localpart {
                return false; // addressed to us
            }
            return true; // addressed to someone else
        }

        // Multi-word prefix: only treat as an address if it is an exact
        // case-insensitive match for our display name.
        if prefix == bot_name_lower {
            return false; // addressed to us
        }
        // Otherwise the colon is part of a regular sentence — not an address.
    }

    false
}

// ---------------------------------------------------------------------------
// Tests
// ---------------------------------------------------------------------------

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

    fn make_user_id(s: &str) -> OwnedUserId {
        s.parse().unwrap()
    }

    // -- mentions_bot -------------------------------------------------------

    #[test]
    fn mentions_bot_by_full_id() {
        let uid = make_user_id("@timmy:homeserver.local");
        assert!(mentions_bot(
            "hello @timmy:homeserver.local can you help?",
            None,
            &uid
        ));
    }

    #[test]
    fn mentions_bot_by_localpart_at_start() {
        let uid = make_user_id("@timmy:homeserver.local");
        assert!(mentions_bot("@timmy please list open stories", None, &uid));
    }

    #[test]
    fn mentions_bot_by_localpart_mid_sentence() {
        let uid = make_user_id("@timmy:homeserver.local");
        assert!(mentions_bot("hey @timmy what's the status?", None, &uid));
    }

    #[test]
    fn mentions_bot_not_mentioned() {
        let uid = make_user_id("@timmy:homeserver.local");
        assert!(!mentions_bot(
            "can someone help me with this PR?",
            None,
            &uid
        ));
    }

    #[test]
    fn mentions_bot_no_false_positive_longer_username() {
        // "@timmybot" must NOT match "@timmy"
        let uid = make_user_id("@timmy:homeserver.local");
        assert!(!mentions_bot("hey @timmybot can you help?", None, &uid));
    }

    #[test]
    fn mentions_bot_at_end_of_string() {
        let uid = make_user_id("@timmy:homeserver.local");
        assert!(mentions_bot("shoutout to @timmy", None, &uid));
    }

    #[test]
    fn mentions_bot_followed_by_comma() {
        let uid = make_user_id("@timmy:homeserver.local");
        assert!(mentions_bot("@timmy, can you help?", None, &uid));
    }

    // -- is_reply_to_bot ----------------------------------------------------

    #[tokio::test]
    async fn is_reply_to_bot_direct_reply_match() {
        let sent: Arc<TokioMutex<HashSet<OwnedEventId>>> =
            Arc::new(TokioMutex::new(HashSet::new()));
        let event_id: OwnedEventId = "$abc123:example.com".parse().unwrap();
        sent.lock().await.insert(event_id.clone());

        let in_reply_to = matrix_sdk::ruma::events::relation::InReplyTo::new(event_id);
        let relates_to: Option<Relation<RoomMessageEventContentWithoutRelation>> =
            Some(Relation::Reply { in_reply_to });

        assert!(is_reply_to_bot(relates_to.as_ref(), &sent).await);
    }

    #[tokio::test]
    async fn is_reply_to_bot_direct_reply_no_match() {
        let sent: Arc<TokioMutex<HashSet<OwnedEventId>>> =
            Arc::new(TokioMutex::new(HashSet::new()));
        // sent is empty — this event was not sent by the bot

        let in_reply_to = matrix_sdk::ruma::events::relation::InReplyTo::new(
            "$other:example.com".parse::<OwnedEventId>().unwrap(),
        );
        let relates_to: Option<Relation<RoomMessageEventContentWithoutRelation>> =
            Some(Relation::Reply { in_reply_to });

        assert!(!is_reply_to_bot(relates_to.as_ref(), &sent).await);
    }

    #[tokio::test]
    async fn is_reply_to_bot_no_relation() {
        let sent: Arc<TokioMutex<HashSet<OwnedEventId>>> =
            Arc::new(TokioMutex::new(HashSet::new()));
        let relates_to: Option<Relation<RoomMessageEventContentWithoutRelation>> = None;
        assert!(!is_reply_to_bot(relates_to.as_ref(), &sent).await);
    }

    #[tokio::test]
    async fn is_reply_to_bot_thread_root_match() {
        let sent: Arc<TokioMutex<HashSet<OwnedEventId>>> =
            Arc::new(TokioMutex::new(HashSet::new()));
        let root_id: OwnedEventId = "$root123:example.com".parse().unwrap();
        sent.lock().await.insert(root_id.clone());

        // Thread reply where the thread root is the bot's message
        let thread = matrix_sdk::ruma::events::relation::Thread::plain(
            root_id,
            "$latest:example.com".parse::<OwnedEventId>().unwrap(),
        );
        let relates_to: Option<Relation<RoomMessageEventContentWithoutRelation>> =
            Some(Relation::Thread(thread));

        assert!(is_reply_to_bot(relates_to.as_ref(), &sent).await);
    }

    // -- is_addressed_to_other ----------------------------------------------

    #[test]
    fn addressed_to_other_display_name_colon() {
        // "sally: do X" — addressed to sally, not our bot (stu)
        let uid = make_user_id("@stu:homeserver.local");
        assert!(is_addressed_to_other("sally: do X", &uid, "stu"));
    }

    #[test]
    fn addressed_to_other_at_mention() {
        // "@sally do X" — addressed to sally, not our bot (stu)
        let uid = make_user_id("@stu:homeserver.local");
        assert!(is_addressed_to_other("@sally do X", &uid, "stu"));
    }

    #[test]
    fn addressed_to_other_at_mention_full_id() {
        // "@sally:homeserver.local do X" — localpart is still "sally"
        let uid = make_user_id("@stu:homeserver.local");
        assert!(is_addressed_to_other(
            "@sally:homeserver.local do X",
            &uid,
            "stu"
        ));
    }

    #[test]
    fn not_addressed_to_other_self_display_name() {
        // "stu: do X" — addressed to us
        let uid = make_user_id("@stu:homeserver.local");
        assert!(!is_addressed_to_other("stu: do X", &uid, "stu"));
    }

    #[test]
    fn not_addressed_to_other_self_at_mention() {
        // "@stu do X" — addressed to us
        let uid = make_user_id("@stu:homeserver.local");
        assert!(!is_addressed_to_other("@stu do X", &uid, "stu"));
    }

    #[test]
    fn not_addressed_to_other_self_at_mention_full_id() {
        // "@stu:homeserver.local do X" — addressed to us
        let uid = make_user_id("@stu:homeserver.local");
        assert!(!is_addressed_to_other(
            "@stu:homeserver.local do X",
            &uid,
            "stu"
        ));
    }

    #[test]
    fn not_addressed_to_other_no_addressee() {
        // No explicit addressee — ambient message for everyone
        let uid = make_user_id("@stu:homeserver.local");
        assert!(!is_addressed_to_other(
            "what's the status of the pipeline?",
            &uid,
            "stu"
        ));
    }

    #[test]
    fn not_addressed_to_other_sentence_with_colon() {
        // Regular sentence with colon — not an address
        let uid = make_user_id("@stu:homeserver.local");
        assert!(!is_addressed_to_other(
            "here is the answer: it depends",
            &uid,
            "stu"
        ));
    }

    #[test]
    fn not_addressed_to_other_display_name_case_insensitive() {
        // "STU: do X" — case-insensitive match against our name "stu"
        let uid = make_user_id("@stu:homeserver.local");
        assert!(!is_addressed_to_other("STU: do X", &uid, "stu"));
    }

    #[test]
    fn addressed_to_other_case_insensitive_other_name() {
        // "SALLY: do X" — addressed to sally, not us
        let uid = make_user_id("@stu:homeserver.local");
        assert!(is_addressed_to_other("SALLY: do X", &uid, "stu"));
    }
}