filter: add content filter

Signed-off-by: William Casarin <jb55@jb55.com>

noteguard.toml

@@ -1,10 +1,13 @@
 
-pipeline = ["protected_events", "kinds", "whitelist", "ratelimit"]
+pipeline = ["protected_events", "kinds", "content", "whitelist", "ratelimit"]
 
 [filters.ratelimit]
 posts_per_minute = 8
 whitelist = ["127.0.0.1"]
 
+[filters.content]
+filters = ["https://cdn.nostr.build/i/some-spammy-or-abusive-image-image.png"]
+
 [filters.whitelist]
 pubkeys = ["16c21558762108afc34e4ff19e4ed51d9a48f79e0c34531efc423d21ab435e93"]
 ips = ["127.0.0.1"]

src/filters/content.rs

@@ -0,0 +1,23 @@
+use crate::{Action, InputMessage, NoteFilter, OutputMessage};
+use serde::Deserialize;
+
+#[derive(Deserialize, Default)]
+pub struct Content {
+    filters: Vec<String>,
+}
+
+impl NoteFilter for Content {
+    fn filter_note(&mut self, msg: &InputMessage) -> OutputMessage {
+        for filter in &self.filters {
+            if msg.event.content.contains(filter) {
+                return OutputMessage::new(msg.event.id.clone(), Action::ShadowReject, None);
+            }
+        }
+
+        OutputMessage::new(msg.event.id.clone(), Action::Accept, None)
+    }
+
+    fn name(&self) -> &'static str {
+        "content"
+    }
+}

src/filters/mod.rs

@@ -1,3 +1,4 @@
+mod content;
 mod kinds;
 mod protected_events;
 mod ratelimit;
@@ -6,6 +7,7 @@ mod whitelist;
 #[cfg(feature = "forwarder")]
 mod forwarder;
 
+pub use content::Content;
 pub use kinds::Kinds;
 pub use protected_events::ProtectedEvents;
 pub use ratelimit::RateLimit;

src/main.rs

@@ -1,14 +1,14 @@
-use noteguard::filters::{Kinds, ProtectedEvents, RateLimit, Whitelist};
+use noteguard::filters::{Content, Kinds, ProtectedEvents, RateLimit, Whitelist};
 
 #[cfg(feature = "forwarder")]
 use noteguard::filters::Forwarder;
 
+use log::info;
 use noteguard::{Action, InputMessage, NoteFilter, OutputMessage};
 use serde::de::DeserializeOwned;
 use serde::Deserialize;
 use std::collections::HashMap;
 use std::io::{self, Read};
-use log::info;
 
 #[derive(Deserialize)]
 struct Config {
@@ -49,6 +49,7 @@ impl Noteguard {
         self.register_filter::<Whitelist>();
         self.register_filter::<ProtectedEvents>();
         self.register_filter::<Kinds>();
+        self.register_filter::<Content>();
 
         #[cfg(feature = "forwarder")]
         self.register_filter::<Forwarder>();
@@ -117,7 +118,6 @@ fn serialize_output_message(msg: &OutputMessage) -> String {
     serde_json::to_string(msg).expect("OutputMessage should always serialize correctly")
 }
 
-
 fn noteguard() {
     env_logger::init();
     info!("running noteguard");
@@ -157,7 +157,11 @@ fn noteguard() {
         };
 
         if input_message.message_type != "new" {
-            let out = OutputMessage::new(input_message.event.id.clone(), Action::Reject, Some("invalid strfry write policy input".to_string()));
+            let out = OutputMessage::new(
+                input_message.event.id.clone(),
+                Action::Reject,
+                Some("invalid strfry write policy input".to_string()),
+            );
             println!("{}", serialize_output_message(&out));
             continue;
         }
@@ -169,7 +173,6 @@ fn noteguard() {
     }
 }
 
-
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -197,7 +200,11 @@ mod tests {
     }
 
     // Helper function to create a mock OutputMessage
-    fn create_mock_output_message(event_id: &str, action: Action, msg: Option<&str>) -> OutputMessage {
+    fn create_mock_output_message(
+        event_id: &str,
+        action: Action,
+        msg: Option<&str>,
+    ) -> OutputMessage {
         OutputMessage {
             id: event_id.to_string(),
             action,
@@ -210,7 +217,9 @@ mod tests {
         let noteguard = Noteguard::new();
         assert!(noteguard.registered_filters.contains_key("ratelimit"));
         assert!(noteguard.registered_filters.contains_key("whitelist"));
-        assert!(noteguard.registered_filters.contains_key("protected_events"));
+        assert!(noteguard
+            .registered_filters
+            .contains_key("protected_events"));
         assert!(noteguard.registered_filters.contains_key("kinds"));
     }
 
@@ -219,12 +228,15 @@ mod tests {
         let mut noteguard = Noteguard::new();
 
         // Create a mock config with one filter (RateLimit)
-        let config: Config = toml::from_str(r#"
+        let config: Config = toml::from_str(
+            r#"
             pipeline = ["ratelimit"]
 
             [filters.ratelimit]
             posts_per_minute = 3
-        "#).expect("Failed to parse config");
+        "#,
+        )
+        .expect("Failed to parse config");
 
         assert!(noteguard.load_config(&config).is_ok());
         assert_eq!(noteguard.loaded_filters.len(), 1);
@@ -235,14 +247,19 @@ mod tests {
         let mut noteguard = Noteguard::new();
 
         // Create a mock config with one filter (RateLimit)
-        let config: Config = toml::from_str(r#"
+        let config: Config = toml::from_str(
+            r#"
             pipeline = ["ratelimit"]
 
             [filters.ratelimit]
             posts_per_minute = 3
-        "#).expect("Failed to parse config");
+        "#,
+        )
+        .expect("Failed to parse config");
 
-        noteguard.load_config(&config).expect("Failed to load config");
+        noteguard
+            .load_config(&config)
+            .expect("Failed to load config");
 
         let input_message = create_mock_input_message("test_event_1", "new");
         let output_message = noteguard.run(input_message);
@@ -255,13 +272,18 @@ mod tests {
         let mut noteguard = Noteguard::new();
 
         // Create a mock config with one filter (ProtectedEvents) which will shadow reject the input
-        let config: Config = toml::from_str(r#"
+        let config: Config = toml::from_str(
+            r#"
             pipeline = ["protected_events"]
 
             [filters.protected_events]
-        "#).expect("Failed to parse config");
+        "#,
+        )
+        .expect("Failed to parse config");
 
-        noteguard.load_config(&config).expect("Failed to load config");
+        noteguard
+            .load_config(&config)
+            .expect("Failed to load config");
 
         let input_message = create_mock_input_message("test_event_3", "new");
         let output_message = noteguard.run(input_message);
@@ -274,13 +296,18 @@ mod tests {
         let mut noteguard = Noteguard::new();
 
         // Create a mock config with one filter (Whitelist) which will reject the input
-        let config: Config = toml::from_str(r#"
+        let config: Config = toml::from_str(
+            r#"
             pipeline = ["whitelist"]
             [filters.whitelist]
             pubkeys = ["something"]
-        "#).expect("Failed to parse config");
+        "#,
+        )
+        .expect("Failed to parse config");
 
-        noteguard.load_config(&config).expect("Failed to load config");
+        noteguard
+            .load_config(&config)
+            .expect("Failed to load config");
 
         let input_message = create_mock_input_message("test_event_2", "new");
         let output_message = noteguard.run(input_message);
@@ -308,7 +335,8 @@ mod tests {
         }
         "#;
 
-        let input_message: InputMessage = serde_json::from_str(input_json).expect("Failed to deserialize input message");
+        let input_message: InputMessage =
+            serde_json::from_str(input_json).expect("Failed to deserialize input message");
         assert_eq!(input_message.event.id, "test_event_5");
         assert_eq!(input_message.message_type, "new");
     }

test/inputs

@@ -1,3 +1,4 @@
+{"type": "new","receivedAt":12345,"sourceType":"IP4","sourceInfo": "127.0.0.3","event":{"id":"eabb9491951fc64d85b6cba28a7d53c333c3f1d59779d63abccf9a8f61a0bdef","pubkey":"5f54041530509de28550475bfe73db709609a4ee1e59281527ba81692923418f","created_at":1722305142,"kind":1,"tags":[["t","lightning"],["t","zapping"],["t","introductions"],["t","freefrom"],["t","help"],["t","nostr"],["t","plebchain"],["t","zapathon"],["t","damus"],["t","bitcoin"],["t","anime"],["t","grownostr"],["t","music"],["t","programming"],["t","cat"],["t","dog"],["t","nsfw"],["t","asknostr"],["t","foodstr"],["t","sports"],["t","acg"],["t","game"],["t","movie"],["t","book"],["t","art"],["t","tech"],["t","science"]],"content":"Freefrom is running out of funds and is about to close down. Please scan the QR code to donate.\n        Freefrom 资金不足即将倒闭，请扫描二维码捐款\n        https://ice.frostsky.com/2024/07/29/9e447d397cac0c27993f0cbcb6efb235.png\n        https://cdn.nostr.build/i/270de92fa4e5118a28352720b7f26c360aa2dbfdb1c12517f3e3117a2f9de47a.png\n        ","sig":"6fa13bedffb2250e2437af6898ab3b11444eb3aef5fd4b484b25658632d283774186f98a024c225581cd9bbaccb0966fa1360d13fd5671f97c7f7bf515f94166"}}
 {"type": "new","receivedAt":12345,"sourceType":"IP4","sourceInfo": "127.0.0.3","event":{"id": "70651d96a2b6b3431cc06b7543249ccd22ab5c203c6aa590b7688f916f252f8f","pubkey": "879d67486027539073d6531d271e3791b15c3e48becbfe4c3727e93355330cc8","created_at": 1720545068,"kind": 1,"tags": [["-"]],"content": "hello there","sig": "21a901e3663bac846493df588ad2185751a5a2826a64c26afb9edce8f9d9344cf00c1ea43016e7faca69da661eadd2731b457a0c31b207ab6ed509a047bf7845"}}
 {"type": "new","receivedAt":12345,"sourceType":"IP4","sourceInfo": "127.0.0.3","event":{"id": "68421a122cef086512b2c5bd29ca6285ced8bd8e302e347e3c5d90466c860a76","pubkey": "16c21558762108afc34e4ff19e4ed51d9a48f79e0c34531efc423d21ab435e93","created_at": 1720408658,"kind": 1,"tags": [],"content": "hi","sig": "7b76471744ded2b720ca832cdc89e670f6093ce38aeef55a5c6a4e077883d7d80dda1e9051032fb1faa1c3c212c517e93ee42b3ceac8e8e9b04bad46a361de90"}}
 {"type": "new","receivedAt":12345,"sourceType":"IP4","sourceInfo": "127.0.0.3","event":{"id": "68421a122cef086512b2c5bd29ca6285ced8bd8e302e347e3c5d90466c860a76","pubkey": "16c21558762108afc34e4ff19e4ed51d9a48f79e0c34531efc423d21ab435e93","created_at": 1720408658,"kind": 1,"tags": [],"content": "hi","sig": "7b76471744ded2b720ca832cdc89e670f6093ce38aeef55a5c6a4e077883d7d80dda1e9051032fb1faa1c3c212c517e93ee42b3ceac8e8e9b04bad46a361de90"}}