X-Git-Url: http://these/git/?a=blobdiff_plain;f=crates%2Fapub%2Fsrc%2Flib.rs;h=d4e34f9e740fd2bb3b444b759ec7e985c9841c93;hb=70fae9d68d65b1e4d153e30d3c065cc315b75eaf;hp=e758e706db23e753e5731a3be98a5416db357729;hpb=a2a594b7635db2241602be56250f7d9bf992f7b9;p=lemmy.git

diff --git a/crates/apub/src/lib.rs b/crates/apub/src/lib.rs
index e758e706..d4e34f9e 100644
--- a/crates/apub/src/lib.rs
+++ b/crates/apub/src/lib.rs
@@ -1,20 +1,20 @@
 use crate::fetcher::post_or_comment::PostOrComment;
-use activitypub_federation::{
-  core::{inbox::ActorPublicKey, signatures::PublicKey},
-  InstanceSettingsBuilder,
-  LocalInstance,
+use activitypub_federation::config::{Data, UrlVerifier};
+use async_trait::async_trait;
+use lemmy_api_common::context::LemmyContext;
+use lemmy_db_schema::{
+  source::{activity::ReceivedActivity, instance::Instance, local_site::LocalSite},
+  utils::{ActualDbPool, DbPool},
 };
-use anyhow::Context;
-use lemmy_api_common::utils::blocking;
-use lemmy_db_schema::{newtypes::DbUrl, source::activity::Activity, utils::DbPool};
-use lemmy_utils::{error::LemmyError, location_info, settings::structs::Settings};
-use lemmy_websocket::LemmyContext;
-use once_cell::sync::{Lazy, OnceCell};
-use std::env;
-use url::{ParseError, Url};
+use lemmy_utils::error::{LemmyError, LemmyErrorType, LemmyResult};
+use moka::future::Cache;
+use once_cell::sync::Lazy;
+use std::{sync::Arc, time::Duration};
+use url::Url;
 
 pub mod activities;
 pub(crate) mod activity_lists;
+pub mod api;
 pub(crate) mod collections;
 pub mod fetcher;
 pub mod http;
@@ -22,27 +22,43 @@ pub(crate) mod mentions;
 pub mod objects;
 pub mod protocol;
 
+pub const FEDERATION_HTTP_FETCH_LIMIT: u32 = 50;
+/// All incoming and outgoing federation actions read the blocklist/allowlist and slur filters
+/// multiple times. This causes a huge number of database reads if we hit the db directly. So we
+/// cache these values for a short time, which will already make a huge difference and ensures that
+/// changes take effect quickly.
+const BLOCKLIST_CACHE_DURATION: Duration = Duration::from_secs(60);
+
 static CONTEXT: Lazy<Vec<serde_json::Value>> = Lazy::new(|| {
   serde_json::from_str(include_str!("../assets/lemmy/context.json")).expect("parse context")
 });
 
-// TODO: store this in context? but its only used in this crate, no need to expose it elsewhere
-fn local_instance(context: &LemmyContext) -> &'static LocalInstance {
-  static LOCAL_INSTANCE: OnceCell<LocalInstance> = OnceCell::new();
-  LOCAL_INSTANCE.get_or_init(|| {
-    let settings = InstanceSettingsBuilder::default()
-      .http_fetch_retry_limit(context.settings().http_fetch_retry_limit)
-      .worker_count(context.settings().federation.worker_count)
-      .testing_send_sync(env::var("APUB_TESTING_SEND_SYNC").is_ok())
-      .verify_url_function(|url| check_apub_id_valid(url, &Settings::get()))
-      .build()
-      .expect("configure federation");
-    LocalInstance::new(
-      context.settings().hostname,
-      context.client().clone(),
-      settings,
-    )
-  })
+#[derive(Clone)]
+pub struct VerifyUrlData(pub ActualDbPool);
+
+#[async_trait]
+impl UrlVerifier for VerifyUrlData {
+  async fn verify(&self, url: &Url) -> Result<(), &'static str> {
+    let local_site_data = local_site_data_cached(&mut (&self.0).into())
+      .await
+      .expect("read local site data");
+    check_apub_id_valid(url, &local_site_data).map_err(|err| match err {
+      LemmyError {
+        error_type: LemmyErrorType::FederationDisabled,
+        ..
+      } => "Federation disabled",
+      LemmyError {
+        error_type: LemmyErrorType::DomainBlocked(_),
+        ..
+      } => "Domain is blocked",
+      LemmyError {
+        error_type: LemmyErrorType::DomainNotInAllowList(_),
+        ..
+      } => "Domain is not in allowlist",
+      _ => "Failed validating apub id",
+    })?;
+    Ok(())
+  }
 }
 
 /// Checks if the ID is allowed for sending or receiving.
@@ -52,172 +68,141 @@ fn local_instance(context: &LemmyContext) -> &'static LocalInstance {
 /// - the correct scheme (either http or https)
 /// - URL being in the allowlist (if it is active)
 /// - URL not being in the blocklist (if it is active)
-///
-/// `use_strict_allowlist` should be true only when parsing a remote community, or when parsing a
-/// post/comment in a local community.
-#[tracing::instrument(skip(settings))]
-fn check_apub_id_valid(apub_id: &Url, settings: &Settings) -> Result<(), &'static str> {
+#[tracing::instrument(skip(local_site_data))]
+fn check_apub_id_valid(apub_id: &Url, local_site_data: &LocalSiteData) -> Result<(), LemmyError> {
   let domain = apub_id.domain().expect("apud id has domain").to_string();
-  let local_instance = settings
-    .get_hostname_without_port()
-    .expect("local hostname is valid");
-  if domain == local_instance {
-    return Ok(());
-  }
 
-  if !settings.federation.enabled {
-    return Err("Federation disabled");
+  if !local_site_data
+    .local_site
+    .as_ref()
+    .map(|l| l.federation_enabled)
+    .unwrap_or(true)
+  {
+    Err(LemmyErrorType::FederationDisabled)?;
   }
 
-  if apub_id.scheme() != settings.get_protocol_string() {
-    return Err("Invalid protocol scheme");
+  if local_site_data
+    .blocked_instances
+    .iter()
+    .any(|i| domain.eq(&i.domain))
+  {
+    Err(LemmyErrorType::DomainBlocked(domain.clone()))?;
   }
 
-  if let Some(blocked) = settings.to_owned().federation.blocked_instances {
-    if blocked.contains(&domain) {
-      return Err("Domain is blocked");
-    }
-  }
-
-  if let Some(allowed) = settings.to_owned().federation.allowed_instances {
-    if !allowed.contains(&domain) {
-      return Err("Domain is not in allowlist");
-    }
+  // Only check this if there are instances in the allowlist
+  if !local_site_data.allowed_instances.is_empty()
+    && !local_site_data
+      .allowed_instances
+      .iter()
+      .any(|i| domain.eq(&i.domain))
+  {
+    Err(LemmyErrorType::DomainNotInAllowList(domain))?;
   }
 
   Ok(())
 }
 
-#[tracing::instrument(skip(settings))]
-pub(crate) fn check_apub_id_valid_with_strictness(
+#[derive(Clone)]
+pub(crate) struct LocalSiteData {
+  local_site: Option<LocalSite>,
+  allowed_instances: Vec<Instance>,
+  blocked_instances: Vec<Instance>,
+}
+
+pub(crate) async fn local_site_data_cached(
+  pool: &mut DbPool<'_>,
+) -> LemmyResult<Arc<LocalSiteData>> {
+  static CACHE: Lazy<Cache<(), Arc<LocalSiteData>>> = Lazy::new(|| {
+    Cache::builder()
+      .max_capacity(1)
+      .time_to_live(BLOCKLIST_CACHE_DURATION)
+      .build()
+  });
+  Ok(
+    CACHE
+      .try_get_with((), async {
+        let (local_site, allowed_instances, blocked_instances) =
+          lemmy_db_schema::try_join_with_pool!(pool => (
+            // LocalSite may be missing
+            |pool| async {
+              Ok(LocalSite::read(pool).await.ok())
+            },
+            Instance::allowlist,
+            Instance::blocklist
+          ))?;
+
+        Ok::<_, diesel::result::Error>(Arc::new(LocalSiteData {
+          local_site,
+          allowed_instances,
+          blocked_instances,
+        }))
+      })
+      .await?,
+  )
+}
+
+pub(crate) async fn check_apub_id_valid_with_strictness(
   apub_id: &Url,
   is_strict: bool,
-  settings: &Settings,
+  context: &LemmyContext,
 ) -> Result<(), LemmyError> {
-  check_apub_id_valid(apub_id, settings).map_err(LemmyError::from_message)?;
   let domain = apub_id.domain().expect("apud id has domain").to_string();
-  let local_instance = settings
+  let local_instance = context
+    .settings()
     .get_hostname_without_port()
     .expect("local hostname is valid");
   if domain == local_instance {
     return Ok(());
   }
 
-  if let Some(mut allowed) = settings.to_owned().federation.allowed_instances {
-    // Only check allowlist if this is a community, or strict allowlist is enabled.
-    let strict_allowlist = settings.to_owned().federation.strict_allowlist;
-    if is_strict || strict_allowlist {
-      // need to allow this explicitly because apub receive might contain objects from our local
-      // instance.
-      allowed.push(local_instance);
-
-      if !allowed.contains(&domain) {
-        return Err(LemmyError::from_message(
-          "Federation forbidden by strict allowlist",
-        ));
-      }
+  let local_site_data = local_site_data_cached(&mut context.pool()).await?;
+  check_apub_id_valid(apub_id, &local_site_data)?;
+
+  // Only check allowlist if this is a community, and there are instances in the allowlist
+  if is_strict && !local_site_data.allowed_instances.is_empty() {
+    // need to allow this explicitly because apub receive might contain objects from our local
+    // instance.
+    let mut allowed_and_local = local_site_data
+      .allowed_instances
+      .iter()
+      .map(|i| i.domain.clone())
+      .collect::<Vec<String>>();
+    let local_instance = context
+      .settings()
+      .get_hostname_without_port()
+      .expect("local hostname is valid");
+    allowed_and_local.push(local_instance);
+
+    let domain = apub_id.domain().expect("apud id has domain").to_string();
+    if !allowed_and_local.contains(&domain) {
+      return Err(LemmyErrorType::FederationDisabledByStrictAllowList)?;
     }
   }
   Ok(())
 }
 
-pub enum EndpointType {
-  Community,
-  Person,
-  Post,
-  Comment,
-  PrivateMessage,
-}
-
-/// Generates an apub endpoint for a given domain, IE xyz.tld
-pub fn generate_local_apub_endpoint(
-  endpoint_type: EndpointType,
-  name: &str,
-  domain: &str,
-) -> Result<DbUrl, ParseError> {
-  let point = match endpoint_type {
-    EndpointType::Community => "c",
-    EndpointType::Person => "u",
-    EndpointType::Post => "post",
-    EndpointType::Comment => "comment",
-    EndpointType::PrivateMessage => "private_message",
-  };
-
-  Ok(Url::parse(&format!("{}/{}/{}", domain, point, name))?.into())
-}
-
-pub fn generate_followers_url(actor_id: &DbUrl) -> Result<DbUrl, ParseError> {
-  Ok(Url::parse(&format!("{}/followers", actor_id))?.into())
-}
-
-pub fn generate_inbox_url(actor_id: &DbUrl) -> Result<DbUrl, ParseError> {
-  Ok(Url::parse(&format!("{}/inbox", actor_id))?.into())
-}
-
-pub fn generate_site_inbox_url(actor_id: &DbUrl) -> Result<DbUrl, ParseError> {
-  let mut actor_id: Url = actor_id.clone().into();
-  actor_id.set_path("site_inbox");
-  Ok(actor_id.into())
-}
-
-pub fn generate_shared_inbox_url(actor_id: &DbUrl) -> Result<DbUrl, LemmyError> {
-  let actor_id: Url = actor_id.clone().into();
-  let url = format!(
-    "{}://{}{}/inbox",
-    &actor_id.scheme(),
-    &actor_id.host_str().context(location_info!())?,
-    if let Some(port) = actor_id.port() {
-      format!(":{}", port)
-    } else {
-      "".to_string()
-    },
-  );
-  Ok(Url::parse(&url)?.into())
-}
-
-pub fn generate_outbox_url(actor_id: &DbUrl) -> Result<DbUrl, ParseError> {
-  Ok(Url::parse(&format!("{}/outbox", actor_id))?.into())
-}
-
-fn generate_moderators_url(community_id: &DbUrl) -> Result<DbUrl, LemmyError> {
-  Ok(Url::parse(&format!("{}/moderators", community_id))?.into())
-}
-
-/// Store a sent or received activity in the database, for logging purposes. These records are not
-/// persistent.
-#[tracing::instrument(skip(pool))]
-async fn insert_activity(
+/// Store received activities in the database.
+///
+/// This ensures that the same activity doesnt get received and processed more than once, which
+/// would be a waste of resources.
+#[tracing::instrument(skip(data))]
+async fn insert_received_activity(
   ap_id: &Url,
-  activity: serde_json::Value,
-  local: bool,
-  sensitive: bool,
-  pool: &DbPool,
-) -> Result<bool, LemmyError> {
-  let ap_id = ap_id.to_owned().into();
-  Ok(
-    blocking(pool, move |conn| {
-      Activity::insert(conn, ap_id, activity, local, sensitive)
-    })
-    .await??,
-  )
+  data: &Data<LemmyContext>,
+) -> Result<(), LemmyError> {
+  ReceivedActivity::create(&mut data.pool(), &ap_id.clone().into()).await?;
+  Ok(())
 }
 
-/// Common methods provided by ActivityPub actors (community and person). Not all methods are
-/// implemented by all actors.
-pub trait ActorType: ActorPublicKey {
-  fn actor_id(&self) -> Url;
-
-  fn private_key(&self) -> Option<String>;
-
-  fn inbox_url(&self) -> Url;
-
-  fn shared_inbox_url(&self) -> Option<Url>;
-
-  fn shared_inbox_or_inbox_url(&self) -> Url {
-    self.shared_inbox_url().unwrap_or_else(|| self.inbox_url())
-  }
+#[async_trait::async_trait]
+pub trait SendActivity: Sync {
+  type Response: Sync + Send + Clone;
 
-  fn get_public_key(&self) -> PublicKey {
-    PublicKey::new_main_key(self.actor_id(), self.public_key().to_string())
+  async fn send_activity(
+    _request: &Self,
+    _response: &Self::Response,
+    _context: &Data<LemmyContext>,
+  ) -> Result<(), LemmyError> {
+    Ok(())
   }
 }