Implement automatic pruning of empty profiles

CHANGELOG.md

@@ -14,7 +14,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 - Added `limits` parameter group to configuration.
 - Made file size limit adjustable with `limits.media.file_size_limit` configuration option.
 - Added `limits.posts.character_limit` configuration parameter (replaces `post_character_limit`).
-- Implemented automatic pruning of remote posts (disabled by default).
+- Implemented automatic pruning of remote posts and empty profiles (disabled by default).
 
 ### Changed
 

contrib/mitra_config.yaml

@@ -38,6 +38,7 @@ registration:
 # Data retention parameters
 #retention:
 #  extraneous_posts: 50
+#  empty_profiles: 150
 
 # List of blocked domains
 #blocked_instances: []

src/config/retention.rs

@@ -3,6 +3,7 @@ use serde::Deserialize;
 #[derive(Clone, Deserialize)]
 pub struct RetentionConfig {
     pub extraneous_posts: Option<u32>,
+    pub empty_profiles: Option<u32>,
 }
 
 #[allow(clippy::derivable_impls)]
@@ -10,6 +11,7 @@ impl Default for RetentionConfig {
     fn default() -> Self {
         Self {
             extraneous_posts: None,
+            empty_profiles: None,
         }
     }
 }

src/job_queue/periodic_tasks.rs

@@ -18,7 +18,14 @@ use crate::ethereum::subscriptions::{
     update_expired_subscriptions,
 };
 use crate::monero::subscriptions::check_monero_subscriptions;
-use crate::models::posts::queries::{delete_post, find_extraneous_posts};
+use crate::models::{
+    posts::queries::{delete_post, find_extraneous_posts},
+    profiles::queries::{
+        delete_profile,
+        find_empty_profiles,
+        get_profile_by_id,
+    },
+};
 use crate::utils::datetime::days_before_now;
 
 pub async fn nft_monitor(
@@ -135,3 +142,22 @@ pub async fn delete_extraneous_posts(
     };
     Ok(())
 }
+
+pub async fn delete_empty_profiles(
+    config: &Config,
+    db_pool: &DbPool,
+) -> Result<(), Error> {
+    let db_client = &mut **get_database_client(db_pool).await?;
+    let updated_before = match config.retention.empty_profiles {
+        Some(days) => days_before_now(days),
+        None => return Ok(()), // not configured
+    };
+    let profiles = find_empty_profiles(db_client, &updated_before).await?;
+    for profile_id in profiles {
+        let profile = get_profile_by_id(db_client, &profile_id).await?;
+        let deletion_queue = delete_profile(db_client, &profile.id).await?;
+        deletion_queue.process(config).await;
+        log::info!("deleted profile {}", profile.acct);
+    };
+    Ok(())
+}

src/job_queue/scheduler.rs

@@ -19,6 +19,7 @@ enum PeriodicTask {
     IncomingActivityQueueExecutor,
     OutgoingActivityQueueExecutor,
     DeleteExtraneousPosts,
+    DeleteEmptyProfiles,
 }
 
 impl PeriodicTask {
@@ -32,6 +33,7 @@ impl PeriodicTask {
             Self::IncomingActivityQueueExecutor => 5,
             Self::OutgoingActivityQueueExecutor => 5,
             Self::DeleteExtraneousPosts => 3600,
+            Self::DeleteEmptyProfiles => 3600,
         }
     }
 
@@ -63,6 +65,9 @@ pub fn run(
         if config.retention.extraneous_posts.is_some() {
             scheduler_state.insert(PeriodicTask::DeleteExtraneousPosts, None);
         };
+        if config.retention.empty_profiles.is_some() {
+            scheduler_state.insert(PeriodicTask::DeleteEmptyProfiles, None);
+        };
 
         let mut interval = tokio::time::interval(Duration::from_secs(5));
         let mut token_waitlist_map: HashMap<Uuid, DateTime<Utc>> = HashMap::new();
@@ -103,6 +108,9 @@ pub fn run(
                     PeriodicTask::DeleteExtraneousPosts => {
                         delete_extraneous_posts(&config, &db_pool).await
                     },
+                    PeriodicTask::DeleteEmptyProfiles => {
+                        delete_empty_profiles(&config, &db_pool).await
+                    },
                 };
                 task_result.unwrap_or_else(|err| {
                     log::error!("{:?}: {}", task, err);