archivist_core/

lemmy.rs

//! Lemmy/Reddthat community monitor.
//!
//! Watches EVERY post + comment in a community (e.g. `c/fanfiction` on
//! reddthat.com), not just mentions. Fanfiction URLs get a metadata reply from
//! the bot; text that reads like a request ("looking for a fic where...") gets
//! an `/ask` search reply. State (last seen post/comment ids) is kept in Redis
//! under `archivist:lemmy:*` so the monitor is idempotent across restarts.
//!
//! Works against any Lemmy instance's public JSON API (`/api/v3/...`); no
//! websocket needed for polling.
// Serde needs these fields present to deserialize the API payloads, even
// though the monitor only reads a few of them.
#![allow(dead_code)]

use serde::Deserialize;

use crate::cache::PageCache;
use crate::config::BotConfig;
use crate::core::PlatformMessage;
use crate::dispatch::{do_ask, do_metadata, CoreCtx};
use crate::error::Result;
use crate::util::{extract_fanfic_url, normalize_url};

#[derive(Debug, Clone, Deserialize)]
struct LoginResponse {
    jwt: String,
}

#[derive(Debug, Clone, Deserialize)]
struct CommunityResponse {
    community_view: CommunityView,
}

#[derive(Debug, Clone, Deserialize)]
struct CommunityView {
    community: Community,
}

#[derive(Debug, Clone, Deserialize)]
struct Community {
    id: i64,
    name: String,
}

#[derive(Debug, Clone, Deserialize)]
struct PostsResponse {
    posts: Vec<PostView>,
}

#[derive(Debug, Clone, Deserialize)]
struct PostView {
    post: Post,
    creator: Person,
}

#[derive(Debug, Clone, Deserialize)]
struct Post {
    id: i64,
    name: String,
    body: Option<String>,
    url: Option<String>,
    creator_id: i64,
    published: String,
    community_id: i64,
}

#[derive(Debug, Clone, Deserialize)]
struct Person {
    name: String,
    // present in post views
}

#[derive(Debug, Clone, Deserialize)]
struct CommentsResponse {
    comments: Vec<CommentView>,
}

#[derive(Debug, Clone, Deserialize)]
struct CommentView {
    comment: Comment,
    creator: CommentPerson,
}

#[derive(Debug, Clone, Deserialize)]
struct Comment {
    id: i64,
    content: String,
    creator_id: i64,
    published: String,
}

#[derive(Debug, Clone, Deserialize)]
struct CommentPerson {
    name: String,
}

/// Redis key prefix for monitor state.
const STATE_PREFIX: &str = "archivist:lemmy";

/// Run the monitor loop forever (spawned as a background task by `main`).
/// Polls the community's new posts + comments; replies to anything that has a
/// fanfiction URL or reads like a fic request. Best-effort: every failure is
/// logged and the loop continues.
pub async fn run_monitor(ctx: CoreCtx) {
    let config = ctx.config.clone();
    if !config.lemmy_enabled {
        tracing::info!("Lemmy monitor disabled (FANFIC_ARCHIVIST_LEMMY_ENABLED=0)");
        return;
    }
    if config.lemmy_username.is_empty() || config.lemmy_password.is_empty() {
        tracing::error!("Lemmy monitor enabled but LEMMY_USERNAME/LEMMY_PASSWORD are empty — monitor will not run");
        return;
    }
    tracing::info!(
        "Lemmy monitor starting: {}/c/{} (poll {}s)",
        config.lemmy_url,
        config.lemmy_community,
        config.lemmy_poll_secs
    );

    let http = reqwest::Client::new();
    let jwt = match login(&config, &http).await {
        Ok(jwt) => jwt,
        Err(e) => {
            tracing::error!("Lemmy login failed: {e}");
            return;
        }
    };
    tracing::info!("Lemmy login OK (jwt len {})", jwt.len());

    let community_id = match resolve_community(&config, &http, &jwt).await {
        Ok(id) => id,
        Err(e) => {
            tracing::error!("Lemmy community resolve failed: {e}");
            return;
        }
    };
    tracing::info!("Monitoring community id {community_id}");

    let mut last_post = last_seen(&ctx.cache, "post").await;
    let mut last_comment = last_seen(&ctx.cache, "comment").await;

    let mut interval = tokio::time::interval(std::time::Duration::from_secs(config.lemmy_poll_secs));
    interval.tick().await; // first tick immediate

    loop {
        interval.tick().await;
        // Posts.
        match fetch_posts(&config, &http, &jwt, community_id, 1).await {
            Ok(posts) => {
                for pv in posts {
                    if pv.post.id <= last_post {
                        continue;
                    }
                    last_post = pv.post.id;
                    let text = format!(
                        "{}\n{}",
                        pv.post.name,
                        pv.post.body.clone().unwrap_or_default()
                    );
                    if let Some(url) = extract_fanfic_url(&text) {
                        let url = normalize_url(&url);
                        let msg = match do_metadata(&ctx, &url).await {
                            Ok(m) => m,
                            Err(e) => {
                                tracing::warn!("post {} metadata failed: {e}", pv.post.id);
                                continue;
                            }
                        };
                        let reply = render_reply(&msg);
                        if let Err(e) = reply_to_post(&config, &http, &jwt, pv.post.id, &reply).await
                        {
                            tracing::warn!("reply to post {} failed: {e}", pv.post.id);
                        } else {
                            tracing::info!("replied to post {} (fic url)", pv.post.id);
                        }
                    } else if looks_like_request(&text) {
                        let query = text.trim().to_string();
                        let msg = match do_ask(&ctx, 0, &query, None).await {
                            Ok(m) => m,
                            Err(e) => {
                                tracing::warn!("post {} ask failed: {e}", pv.post.id);
                                continue;
                            }
                        };
                        let reply = render_reply(&msg);
                        if let Err(e) = reply_to_post(&config, &http, &jwt, pv.post.id, &reply).await
                        {
                            tracing::warn!("reply to post {} failed: {e}", pv.post.id);
                        } else {
                            tracing::info!("replied to post {} (request)", pv.post.id);
                        }
                    }
                }
                let _ = ctx
                    .cache
                    .cache_raw(&format!("{STATE_PREFIX}:post"), 60 * 60 * 24 * 7, &serde_json::json!(last_post))
                    .await;
            }
            Err(e) => tracing::warn!("fetch posts failed: {e}"),
        }
        // Comments.
        match fetch_comments(&config, &http, &jwt, community_id, 1).await {
            Ok(comments) => {
                for cv in comments {
                    if cv.comment.id <= last_comment {
                        continue;
                    }
                    last_comment = cv.comment.id;
                    let text = cv.comment.content.clone();
                    if let Some(url) = extract_fanfic_url(&text) {
                        let url = normalize_url(&url);
                        let msg = match do_metadata(&ctx, &url).await {
                            Ok(m) => m,
                            Err(e) => {
                                tracing::warn!("comment {} metadata failed: {e}", cv.comment.id);
                                continue;
                            }
                        };
                        let reply = render_reply(&msg);
                        if let Err(e) =
                            reply_to_comment(&config, &http, &jwt, cv.comment.id, &reply).await
                        {
                            tracing::warn!("reply to comment {} failed: {e}", cv.comment.id);
                        } else {
                            tracing::info!("replied to comment {} (fic url)", cv.comment.id);
                        }
                    } else if looks_like_request(&text) {
                        let query = text.trim().to_string();
                        let msg = match do_ask(&ctx, 0, &query, None).await {
                            Ok(m) => m,
                            Err(e) => {
                                tracing::warn!("comment {} ask failed: {e}", cv.comment.id);
                                continue;
                            }
                        };
                        let reply = render_reply(&msg);
                        if let Err(e) =
                            reply_to_comment(&config, &http, &jwt, cv.comment.id, &reply).await
                        {
                            tracing::warn!("reply to comment {} failed: {e}", cv.comment.id);
                        } else {
                            tracing::info!("replied to comment {} (request)", cv.comment.id);
                        }
                    }
                }
                let _ = ctx
                    .cache
                    .cache_raw(&format!("{STATE_PREFIX}:comment"), 60 * 60 * 24 * 7, &serde_json::json!(last_comment))
                    .await;
            }
            Err(e) => tracing::warn!("fetch comments failed: {e}"),
        }
    }
}

/// Render a `PlatformMessage` as plain text for a Lemmy reply (Lemmy supports
/// markdown, so rich items keep their text).
fn render_reply(msg: &PlatformMessage) -> String {
    match msg {
        PlatformMessage::Text(s) => s.clone(),
        PlatformMessage::Rich { header, items, .. } => {
            let mut out = String::new();
            if let Some(h) = header {
                out.push_str(h);
                out.push('\n');
            }
            for (i, item) in items.iter().enumerate() {
                if i > 0 {
                    out.push_str("\n---\n");
                }
                if let Some(t) = &item.title {
                    out.push_str(&format!("**{}**\n", t));
                }
                if let Some(u) = &item.url {
                    out.push_str(&format!("<{u}>\n"));
                }
                out.push_str(&item.body);
                out.push('\n');
                for (k, v) in &item.fields {
                    out.push_str(&format!("**{k}**: {v}\n"));
                }
                if let Some(f) = &item.footer {
                    out.push_str(&format!("*{f}*\n"));
                }
            }
            out
        }
        PlatformMessage::File { .. } => "*(file uploads are not supported on this platform)*".into(),
        PlatformMessage::Ephemeral(inner) => render_reply(inner),
    }
}

/// Very light heuristic: does this text read like a fic request?
fn looks_like_request(text: &str) -> bool {
    let lower = text.to_lowercase();
    let words = [
        "looking for", "find a fic", "fic where", "fic about", "recommend",
        "recommendations", "any fics", "suggest", "suggestions", "need a fic",
        "help me find", "trope", "request",
    ];
    // Must be reasonably long to avoid replying to random chatter.
    if text.trim().chars().count() < 25 {
        return false;
    }
    words.iter().any(|w| lower.contains(w))
}

async fn login(config: &BotConfig, http: &reqwest::Client) -> Result<String> {
    let url = format!("{}/api/v3/user/login", config.lemmy_url.trim_end_matches('/'));
    let resp: LoginResponse = http
        .post(&url)
        .json(&serde_json::json!({
            "username_or_email": config.lemmy_username,
            "password": config.lemmy_password,
        }))
        .send()
        .await?
        .error_for_status()?
        .json()
        .await?;
    Ok(resp.jwt)
}

async fn resolve_community(config: &BotConfig, http: &reqwest::Client, jwt: &str) -> Result<i64> {
    let url = format!(
        "{}/api/v3/community?name={}",
        config.lemmy_url.trim_end_matches('/'),
        config.lemmy_community
    );
    let resp: CommunityResponse = http
        .get(&url)
        .header("Authorization", format!("Bearer {jwt}"))
        .send()
        .await?
        .error_for_status()?
        .json()
        .await?;
    Ok(resp.community_view.community.id)
}

async fn fetch_posts(
    config: &BotConfig,
    http: &reqwest::Client,
    jwt: &str,
    community_id: i64,
    page: i64,
) -> Result<Vec<PostView>> {
    let url = format!(
        "{}/api/v3/post/list?community_id={}&page={}&limit=20&sort=New",
        config.lemmy_url.trim_end_matches('/'),
        community_id,
        page
    );
    let resp: PostsResponse = http
        .get(&url)
        .header("Authorization", format!("Bearer {jwt}"))
        .send()
        .await?
        .error_for_status()?
        .json()
        .await?;
    Ok(resp.posts)
}

async fn fetch_comments(
    config: &BotConfig,
    http: &reqwest::Client,
    jwt: &str,
    community_id: i64,
    page: i64,
) -> Result<Vec<CommentView>> {
    let url = format!(
        "{}/api/v3/comment/list?community_id={}&page={}&limit=20&sort=New&max_depth=1",
        config.lemmy_url.trim_end_matches('/'),
        community_id,
        page
    );
    let resp: CommentsResponse = http
        .get(&url)
        .header("Authorization", format!("Bearer {jwt}"))
        .send()
        .await?
        .error_for_status()?
        .json()
        .await?;
    Ok(resp.comments)
}

async fn reply_to_post(
    config: &BotConfig,
    http: &reqwest::Client,
    jwt: &str,
    post_id: i64,
    content: &str,
) -> Result<()> {
    let url = format!(
        "{}/api/v3/comment",
        config.lemmy_url.trim_end_matches('/')
    );
    http.post(&url)
        .header("Authorization", format!("Bearer {jwt}"))
        .json(&serde_json::json!({
            "content": content,
            "post_id": post_id,
            "form_id": None::<String>,
        }))
        .send()
        .await?
        .error_for_status()?;
    Ok(())
}

async fn reply_to_comment(
    config: &BotConfig,
    http: &reqwest::Client,
    jwt: &str,
    comment_id: i64,
    content: &str,
) -> Result<()> {
    let url = format!(
        "{}/api/v3/comment",
        config.lemmy_url.trim_end_matches('/')
    );
    http.post(&url)
        .header("Authorization", format!("Bearer {jwt}"))
        .json(&serde_json::json!({
            "content": content,
            "parent_id": comment_id,
            "form_id": None::<String>,
        }))
        .send()
        .await?
        .error_for_status()?;
    Ok(())
}

/// Last seen id for a kind ("post" | "comment"), from Redis (best-effort).
async fn last_seen(cache: &PageCache, kind: &str) -> i64 {
    match cache.cached_raw(&format!("{STATE_PREFIX}:{kind}")).await {
        Some(v) => v.as_i64().unwrap_or(0),
        None => 0,
    }
}

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

    #[test]
    fn looks_like_request_detects() {
        assert!(looks_like_request(
            "Looking for a fic where Harry is raised by wolves, complete, over 50k words"
        ));
        assert!(looks_like_request("Any recommendations for slow-burn enemies to lovers?"));
    }

    #[test]
    fn looks_like_request_ignores_chatter() {
        assert!(!looks_like_request("hi"));
        assert!(!looks_like_request("what do you guys think about the new chapter?"));
    }

    #[test]
    fn render_reply_rich() {
        let msg = PlatformMessage::Rich {
            header: Some("search: \"dark harry\"".into()),
            items: vec![crate::core::RichItem::new("body text")
                .title("A Fic")
                .url("https://x")
                .field("Words", "50k")],
            actions: vec![],
        };
        let out = render_reply(&msg);
        assert!(out.contains("**A Fic**"));
        assert!(out.contains("body text"));
        assert!(out.contains("**Words**: 50k"));
    }
}