sieve: use block_list to filter authors

nixos-module.nix

@@ -55,6 +55,7 @@ let
     redis_password_file = cfg.redis.passwordFile;
     in_topic = "relay-in";
     prometheus_port = 9102;
+    blocklist = blocklistPath;
   };
 
   sieveSettings = lib.recursiveUpdate sieveDefaultSettings cfg.sieve.settings;

sieve/src/config.rs

@@ -9,6 +9,7 @@ pub struct Config {
     pub in_topic: String,
     priv_key_file: String,
     pub prometheus_port: u16,
+    pub blocklist: String,
 }
 
 impl Config {

sieve/src/main.rs

@@ -10,6 +10,7 @@ use cave::{
         self,
         fetch::authorized_fetch,
     },
+    block_list::BlockList,
     config::LoadConfig,
     feed,
 };
@@ -43,6 +44,7 @@ async fn main() {
     cave::init::init_logger(5557);
 
     let config = config::Config::load();
+    let block_list = BlockList::new(&config.blocklist).await;
 
     PrometheusBuilder::new()
         .with_http_listener(([0; 8], config.prometheus_port))
@@ -86,6 +88,7 @@ async fn main() {
         }
 
         let posts_cache = posts_cache.clone();
+        let block_list = block_list.clone();
         let client = client.clone();
         let mut store = store.clone();
         let priv_key = priv_key.clone();
@@ -137,6 +140,25 @@ async fn main() {
             };
 
             let author: activitypub::Actor = if let Some(author_url) = &post.attributed_to {
+                if let Ok(url) = Url::parse(author_url) {
+                    let host = if let Some(host) = url.host_str() {
+                        host
+                    } else {
+                        tracing::error!("No host in author {author_url}");
+                        return;
+                    };
+                    if block_list.is_blocked(host).await {
+                        tracing::warn!("Ignore blocked author {author_url}");
+                        metrics::counter!("sieve_activity", 1, "type" => "blocked_author");
+                        return;
+                    }
+                } else {
+                    tracing::error!("Invalid author: {author_url}");
+                    metrics::counter!("sieve_activity", 1, "type" => "invalid_author");
+                    return;
+
+                }
+
                 match authorized_fetch(&client, author_url, KEY_ID, &priv_key).await {
                     Ok(author) => {
                         metrics::counter!("sieve_activity", 1, "type" => "fetch_author_ok");