forgery/src/workers.rs

use crate::classifier::Classifier;
use crate::data::UserId;
use crate::db::{Db, IsSpam};
use crate::email;
use crate::scrape;
use anyhow::anyhow;
use forgejo_api::Forgejo;
use std::collections::HashMap;
use std::path::Path;
use std::sync::{Arc, Mutex};

use crate::FORGEJO_POLL_DELAY;
use crate::GRACE_PERIOD;
use crate::{ActuallyBan, Config};
use crate::{GUESS_LEGIT_THRESHOLD, GUESS_SPAM_THRESHOLD};

// Worker to refresh user data by periodically polling Forgejo

async fn try_refresh_user_data(
    forge: &Forgejo,
    db: Arc<Mutex<Db>>,
    classifier: Arc<Mutex<Classifier>>,
) -> anyhow::Result<()> {
    {
        let db = &db.lock().unwrap();
        let d = db.last_scrape.elapsed()?;
        if d < FORGEJO_POLL_DELAY {
            return Ok(());
        }
    }

    eprintln!("Fetching user data");
    let users = scrape::get_user_data(forge).await?;

    let db: &mut Db = &mut db.lock().unwrap();
    let classifier = &classifier.lock().unwrap();

    // NB: Some user accounts may have been deleted since last fetch (hopefully
    // they were spammers).
    // Such users will appear in the current [db] but not in the new [users].
    // We don't want to keep them in the database, so we rebuild a fresh [db]
    // containing only data for users who still exist.

    let mut newdb = Db::from_users(users, HashMap::new(), classifier);

    // Import spam classification from the previous Db
    for (&user_id, user_data) in &newdb.users {
        let &score = newdb.score.get(&user_id).unwrap();
        if let Some(&user_was_spam) = db.is_spam.get(&user_id) {
            if (user_was_spam.as_bool() && score < GUESS_SPAM_THRESHOLD)
                || (!user_was_spam.as_bool() && score > GUESS_LEGIT_THRESHOLD)
            {
                eprintln!(
                    "Score for user {} changed past threshold; discarding our current classification",
                    user_data.login
                );
            } else {
                newdb.is_spam.insert(user_id, user_was_spam);
            }
        }
    }

    // switch to [newdb]
    let _ = std::mem::replace(db, newdb);

    db.store_to_path(Path::new("db.json")).unwrap(); // FIXME

    Ok(())
}

pub async fn refresh_user_data(
    forge: Arc<Forgejo>,
    db: Arc<Mutex<Db>>,
    classifier: Arc<Mutex<Classifier>>,
) {
    loop {
        if let Err(e) = try_refresh_user_data(&forge, db.clone(), classifier.clone()).await {
            eprintln!("Error refreshing user data: {:?}", e);
        }
        tokio::time::sleep(FORGEJO_POLL_DELAY.mul_f32(0.1)).await;
    }
}

// Worker to delete spam accounts after their grace period expired

async fn try_purge_account(config: &Config, forge: &Forgejo, login: &str) -> anyhow::Result<()> {
    if let ActuallyBan::No = config.actually_ban {
        eprintln!("[Simulating: delete account of user {login}]");
        return Ok(());
    }

    eprintln!("Deleting account of user {login}");
    forge
        .admin_delete_user(
            login,
            forgejo_api::structs::AdminDeleteUserQuery { purge: Some(true) },
        )
        .await?;
    eprintln!("Success");
    Ok(())
}

pub async fn purge_spammer_accounts(config: Arc<Config>, forge: Arc<Forgejo>, db: Arc<Mutex<Db>>) {
    loop {
        let mut classified_users = Vec::new();
        {
            let db = &db.lock().unwrap();
            for (id, user, is_spam) in db.classified_users() {
                classified_users.push((id, user.login.clone(), is_spam));
            }
        }

        for (user_id, login, is_spam) in classified_users {
            if let IsSpam::Spam {
                classified_at,
                locked,
                notified,
            } = is_spam
            {
                match classified_at.elapsed() {
                    Ok(duration) if duration > GRACE_PERIOD => {
                        if !locked {
                            // NOTE: this is a minimum sanity check, but can this
                            // realistically happen? could we do better than printing a
                            // message in this case?
                            eprintln!("WARN: grace period for {login} expired but account is not locked! Skip user.");
                            continue;
                        }
                        if !notified {
                            // NOTE: we delete accounts even if we failed to notify them by email.
                            // Could we do better? (But if we do not delete in this case,
                            // spammers could perhaps prevent being deleted by providing
                            // non-working emails…)
                            eprintln!(
                                "WARN: grace period for {login} expired but we failed to send a \
                                       notification email. Deleting anyway..."
                            );
                        }

                        if let Err(e) = try_purge_account(&config, &forge, &login).await {
                            eprintln!("Error while deleting spammer account {login}: {:?}", e)
                        } else {
                            eprintln!("Deleted spammer account {login}");
                            let db = &mut db.lock().unwrap();
                            db.users.remove(&user_id);
                            db.is_spam.remove(&user_id);
                            db.score.remove(&user_id);
                            db.tokens.remove(&user_id);
                            db.store_to_path(Path::new("db.json")).unwrap(); // FIXME
                        }
                    }
                    _ => (),
                }
            }
        }

        tokio::time::sleep(std::time::Duration::from_secs(3600)).await;
    }
}

// Lock a user account and send a notification email.
// Since this can fail, we put it into a worker that periodically retries on any
// user marked as spam and not already locked/notified.

async fn lock_user_account(forge: &Forgejo, username: &str) -> anyhow::Result<()> {
    let opts = forgejo_api::structs::EditUserOption {
        // boilerplate: we do not change these settings
        active: None,
        admin: None,
        allow_create_organization: None,
        allow_git_hook: None,
        allow_import_local: None,
        description: None,
        email: None,
        full_name: None,
        location: None,
        login_name: None,
        max_repo_creation: None,
        must_change_password: None,
        password: None,
        pronouns: None,
        restricted: None,
        source_id: None,
        website: None,
        // lock the account and set its visibility to private: the user's
        // description and info will not be publicly visible
        prohibit_login: Some(true),
        visibility: Some("private".to_string()),
    };
    forge.admin_edit_user(username, opts).await?;
    Ok(())
}

pub async fn try_lock_and_notify_user(
    config: &Config,
    forge: &Forgejo,
    db: Arc<Mutex<Db>>,
    user_id: UserId,
) -> anyhow::Result<()> {
    let (login, email, is_spam) = {
        let db = &db.lock().unwrap();
        let user = db.users.get(&user_id).unwrap();
        let is_spam = match db.is_spam.get(&user_id) {
            Some(IsSpam::Spam {
                classified_at,
                locked,
                notified,
            }) => Some((*classified_at, *locked, *notified)),
            _ => None,
        };
        (user.login.clone(), user.email.clone(), is_spam)
    };

    if let Some((classified_at, locked, notified)) = is_spam {
        if !locked {
            match &config.actually_ban {
                ActuallyBan::Yes { .. } => {
                    eprintln!("Locking account of user {login}");
                    lock_user_account(forge, &login).await?;
                    eprintln!("Success");
                }
                ActuallyBan::No => eprintln!("[Simulating: lock account of user {login}]"),
            }

            let db = &mut db.lock().unwrap();
            db.is_spam.insert(
                user_id,
                IsSpam::Spam {
                    classified_at,
                    locked: true,
                    notified,
                },
            );
            db.store_to_path(Path::new("db.json")).unwrap(); // FIXME
        }

        if !notified {
            match &config.actually_ban {
                ActuallyBan::Yes { smtp } => {
                    eprintln!("Sending notification email to user {login}");
                    email::send_locked_account_notice(config, smtp, &login, &email).await?;
                    eprintln!("Success");
                }
                ActuallyBan::No => {
                    eprintln!("[Simulating: send notification email to user {login}]")
                }
            }
            let db = &mut db.lock().unwrap();
            db.is_spam.insert(
                user_id,
                IsSpam::Spam {
                    classified_at,
                    locked: true,
                    notified: true,
                },
            );
            db.store_to_path(Path::new("db.json")).unwrap(); // FIXME
        }

        Ok(())
    } else {
        Err(anyhow!(
            "Tried to lock user {} who was not classified as spam",
            login
        ))
    }
}

pub async fn lock_and_notify_users(config: Arc<Config>, forge: Arc<Forgejo>, db: Arc<Mutex<Db>>) {
    let mut spammers = Vec::new();
    {
        let db = &db.lock().unwrap();
        for (id, user, is_spam) in db.classified_users() {
            if is_spam.as_bool() {
                spammers.push((id, user.login.clone()))
            }
        }
    }

    for (user_id, login) in spammers {
        try_lock_and_notify_user(&config, &forge, db.clone(), user_id)
            .await
            .unwrap_or_else(|err| eprintln!("Failed to lock or notify user {login}: {err}"));
    }

    tokio::time::sleep(std::time::Duration::from_secs(3600)).await;
}