X-Git-Url: http://these/git/?a=blobdiff_plain;f=crates%2Fapub%2Fsrc%2Flib.rs;h=9a45284f22fbf609dfbbdf0196847e685b8cdbf0;hb=e9e76549a88cfbdab36f00d302cceabcaaa24f4c;hp=74d4cbef306bbda7833dd7de78c2c01d064f9b5d;hpb=3ca2453fc675459290397c0cad11663ea0a5fab9;p=lemmy.git

diff --git a/crates/apub/src/lib.rs b/crates/apub/src/lib.rs
index 74d4cbef..9a45284f 100644
--- a/crates/apub/src/lib.rs
+++ b/crates/apub/src/lib.rs
@@ -1,57 +1,51 @@
-#[macro_use]
-extern crate lazy_static;
+use crate::fetcher::post_or_comment::PostOrComment;
+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 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 mod activity_queue;
-pub mod extensions;
+pub(crate) mod activity_lists;
+pub mod api;
+pub(crate) mod collections;
 pub mod fetcher;
 pub mod http;
-pub mod inbox;
+pub(crate) mod mentions;
 pub mod objects;
-pub mod routes;
-
-use crate::extensions::{
-  group_extensions::GroupExtension,
-  page_extension::PageExtension,
-  signatures::{PublicKey, PublicKeyExtension},
-};
-use activitystreams::{
-  activity::Follow,
-  actor,
-  base::AnyBase,
-  object::{ApObject, Note, Page},
-};
-use activitystreams_ext::{Ext1, Ext2};
-use anyhow::{anyhow, Context};
-use diesel::NotFound;
-use lemmy_api_structs::blocking;
-use lemmy_db_queries::{source::activity::Activity_, ApubObject, DbPool};
-use lemmy_db_schema::{
-  source::{
-    activity::Activity,
-    comment::Comment,
-    community::Community,
-    person::{Person as DbPerson, Person},
-    post::Post,
-    private_message::PrivateMessage,
-  },
-  DbUrl,
-};
-use lemmy_utils::{location_info, settings::structs::Settings, LemmyError};
-use lemmy_websocket::LemmyContext;
-use serde::Serialize;
-use std::net::IpAddr;
-use url::{ParseError, Url};
-
-/// Activitystreams type for community
-type GroupExt = Ext2<actor::ApActor<ApObject<actor::Group>>, GroupExtension, PublicKeyExtension>;
-/// Activitystreams type for person
-type PersonExt = Ext1<actor::ApActor<ApObject<actor::Person>>, PublicKeyExtension>;
-/// Activitystreams type for post
-type PageExt = Ext1<ApObject<Page>, PageExtension>;
-type NoteExt = ApObject<Note>;
-
-pub static APUB_JSON_CONTENT_TYPE: &str = "application/activity+json";
+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")
+});
+
+#[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)?;
+    Ok(())
+  }
+}
 
 /// Checks if the ID is allowed for sending or receiving.
 ///
@@ -60,342 +54,146 @@ pub static APUB_JSON_CONTENT_TYPE: &str = "application/activity+json";
 /// - 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)
-///
-/// Note that only one of allowlist and blacklist can be enabled, not both.
-fn check_is_apub_id_valid(apub_id: &Url) -> Result<(), LemmyError> {
-  let settings = Settings::get();
-  let domain = apub_id.domain().context(location_info!())?.to_string();
-  let local_instance = settings.get_hostname_without_port()?;
-
-  if !settings.federation().enabled {
-    return if domain == local_instance {
-      Ok(())
-    } else {
-      Err(
-        anyhow!(
-          "Trying to connect with {}, but federation is disabled",
-          domain
-        )
-        .into(),
-      )
-    };
+#[tracing::instrument(skip(local_site_data))]
+fn check_apub_id_valid(apub_id: &Url, local_site_data: &LocalSiteData) -> Result<(), &'static str> {
+  let domain = apub_id.domain().expect("apud id has domain").to_string();
+
+  if !local_site_data
+    .local_site
+    .as_ref()
+    .map(|l| l.federation_enabled)
+    .unwrap_or(true)
+  {
+    return Err("Federation disabled");
   }
 
-  let host = apub_id.host_str().context(location_info!())?;
-  let host_as_ip = host.parse::<IpAddr>();
-  if host == "localhost" || host_as_ip.is_ok() {
-    return Err(anyhow!("invalid hostname {}: {}", host, apub_id).into());
+  if local_site_data
+    .blocked_instances
+    .iter()
+    .any(|i| domain.eq(&i.domain))
+  {
+    return Err("Domain is blocked");
   }
 
-  if apub_id.scheme() != Settings::get().get_protocol_string() {
-    return Err(anyhow!("invalid apub id scheme {}: {}", apub_id.scheme(), apub_id).into());
+  // 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))
+  {
+    return Err("Domain is not in allowlist");
   }
 
-  let allowed_instances = Settings::get().get_allowed_instances();
-  let blocked_instances = Settings::get().get_blocked_instances();
-
-  if allowed_instances.is_none() && blocked_instances.is_none() {
-    Ok(())
-  } else if let Some(mut allowed) = allowed_instances {
-    // need to allow this explicitly because apub receive might contain objects from our local
-    // instance. split is needed to remove the port in our federation test setup.
-    allowed.push(local_instance);
-
-    if allowed.contains(&domain) {
-      Ok(())
-    } else {
-      Err(anyhow!("{} not in federation allowlist", domain).into())
-    }
-  } else if let Some(blocked) = blocked_instances {
-    if blocked.contains(&domain) {
-      Err(anyhow!("{} is in federation blocklist", domain).into())
-    } else {
-      Ok(())
-    }
-  } else {
-    panic!("Invalid config, both allowed_instances and blocked_instances are specified");
-  }
-}
-
-/// Common functions for ActivityPub objects, which are implemented by most (but not all) objects
-/// and actors in Lemmy.
-#[async_trait::async_trait(?Send)]
-pub trait ApubObjectType {
-  async fn send_create(&self, creator: &DbPerson, context: &LemmyContext)
-    -> Result<(), LemmyError>;
-  async fn send_update(&self, creator: &DbPerson, context: &LemmyContext)
-    -> Result<(), LemmyError>;
-  async fn send_delete(&self, creator: &DbPerson, context: &LemmyContext)
-    -> Result<(), LemmyError>;
-  async fn send_undo_delete(
-    &self,
-    creator: &DbPerson,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-  async fn send_remove(&self, mod_: &DbPerson, context: &LemmyContext) -> Result<(), LemmyError>;
-  async fn send_undo_remove(
-    &self,
-    mod_: &DbPerson,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-}
-
-#[async_trait::async_trait(?Send)]
-pub trait ApubLikeableType {
-  async fn send_like(&self, creator: &DbPerson, context: &LemmyContext) -> Result<(), LemmyError>;
-  async fn send_dislike(
-    &self,
-    creator: &DbPerson,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-  async fn send_undo_like(
-    &self,
-    creator: &DbPerson,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-}
-
-/// Common methods provided by ActivityPub actors (community and person). Not all methods are
-/// implemented by all actors.
-#[async_trait::async_trait(?Send)]
-pub trait ActorType {
-  fn is_local(&self) -> bool;
-  fn actor_id(&self) -> Url;
-
-  // TODO: every actor should have a public key, so this shouldnt be an option (needs to be fixed in db)
-  fn public_key(&self) -> Option<String>;
-  fn private_key(&self) -> Option<String>;
-
-  fn get_shared_inbox_or_inbox_url(&self) -> Url;
-
-  /// Outbox URL is not generally used by Lemmy, so it can be generated on the fly (but only for
-  /// local actors).
-  fn get_outbox_url(&self) -> Result<Url, LemmyError> {
-    if !self.is_local() {
-      return Err(anyhow!("get_outbox_url() called for remote actor").into());
-    }
-    Ok(Url::parse(&format!("{}/outbox", &self.actor_id()))?)
-  }
-
-  fn get_public_key_ext(&self) -> Result<PublicKeyExtension, LemmyError> {
-    Ok(
-      PublicKey {
-        id: format!("{}#main-key", self.actor_id()),
-        owner: self.actor_id(),
-        public_key_pem: self.public_key().context(location_info!())?,
-      }
-      .to_ext(),
-    )
-  }
-}
-
-#[async_trait::async_trait(?Send)]
-pub trait CommunityType {
-  async fn get_follower_inboxes(&self, pool: &DbPool) -> Result<Vec<Url>, LemmyError>;
-  async fn send_accept_follow(
-    &self,
-    follow: Follow,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-
-  async fn send_delete(&self, context: &LemmyContext) -> Result<(), LemmyError>;
-  async fn send_undo_delete(&self, context: &LemmyContext) -> Result<(), LemmyError>;
-
-  async fn send_remove(&self, context: &LemmyContext) -> Result<(), LemmyError>;
-  async fn send_undo_remove(&self, context: &LemmyContext) -> Result<(), LemmyError>;
-
-  async fn send_announce(
-    &self,
-    activity: AnyBase,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-
-  async fn send_add_mod(
-    &self,
-    actor: &Person,
-    added_mod: Person,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-  async fn send_remove_mod(
-    &self,
-    actor: &Person,
-    removed_mod: Person,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-}
-
-#[async_trait::async_trait(?Send)]
-pub trait UserType {
-  async fn send_follow(
-    &self,
-    follow_actor_id: &Url,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
-  async fn send_unfollow(
-    &self,
-    follow_actor_id: &Url,
-    context: &LemmyContext,
-  ) -> Result<(), LemmyError>;
+  Ok(())
 }
 
-pub enum EndpointType {
-  Community,
-  Person,
-  Post,
-  Comment,
-  PrivateMessage,
+#[derive(Clone)]
+pub(crate) struct LocalSiteData {
+  local_site: Option<LocalSite>,
+  allowed_instances: Vec<Instance>,
+  blocked_instances: Vec<Instance>,
 }
 
-/// Generates the ActivityPub ID for a given object type and ID.
-pub fn generate_apub_endpoint(
-  endpoint_type: EndpointType,
-  name: &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",
-  };
-
+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(
-    Url::parse(&format!(
-      "{}/{}/{}",
-      Settings::get().get_protocol_and_hostname(),
-      point,
-      name
-    ))?
-    .into(),
+    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 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_shared_inbox_url(actor_id: &DbUrl) -> Result<DbUrl, LemmyError> {
-  let actor_id = actor_id.clone().into_inner();
-  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(crate) 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.
-pub(crate) async fn insert_activity<T>(
-  ap_id: &Url,
-  activity: T,
-  local: bool,
-  sensitive: bool,
-  pool: &DbPool,
-) -> Result<(), LemmyError>
-where
-  T: Serialize + std::fmt::Debug + Send + 'static,
-{
-  let ap_id = ap_id.to_owned().into();
-  blocking(pool, move |conn| {
-    Activity::insert(conn, ap_id, &activity, local, sensitive)
-  })
-  .await??;
-  Ok(())
-}
-
-pub(crate) enum PostOrComment {
-  Comment(Box<Comment>),
-  Post(Box<Post>),
-}
-
-/// Tries to find a post or comment in the local database, without any network requests.
-/// This is used to handle deletions and removals, because in case we dont have the object, we can
-/// simply ignore the activity.
-pub(crate) async fn find_post_or_comment_by_id(
+pub(crate) async fn check_apub_id_valid_with_strictness(
+  apub_id: &Url,
+  is_strict: bool,
   context: &LemmyContext,
-  apub_id: Url,
-) -> Result<PostOrComment, LemmyError> {
-  let ap_id = apub_id.clone();
-  let post = blocking(context.pool(), move |conn| {
-    Post::read_from_apub_id(conn, &ap_id.into())
-  })
-  .await?;
-  if let Ok(p) = post {
-    return Ok(PostOrComment::Post(Box::new(p)));
+) -> Result<(), LemmyError> {
+  let domain = apub_id.domain().expect("apud id has domain").to_string();
+  let local_instance = context
+    .settings()
+    .get_hostname_without_port()
+    .expect("local hostname is valid");
+  if domain == local_instance {
+    return Ok(());
   }
 
-  let ap_id = apub_id.clone();
-  let comment = blocking(context.pool(), move |conn| {
-    Comment::read_from_apub_id(conn, &ap_id.into())
-  })
-  .await?;
-  if let Ok(c) = comment {
-    return Ok(PostOrComment::Comment(Box::new(c)));
-  }
+  let local_site_data = local_site_data_cached(&mut context.pool()).await?;
+  check_apub_id_valid(apub_id, &local_site_data).map_err(|err| match err {
+    "Federation disabled" => LemmyErrorType::FederationDisabled,
+    "Domain is blocked" => LemmyErrorType::DomainBlocked,
+    "Domain is not in allowlist" => LemmyErrorType::DomainNotInAllowList,
+    _ => panic!("Could not handle apub error!"),
+  })?;
 
-  Err(NotFound.into())
+  // 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(())
 }
 
-#[derive(Debug)]
-pub(crate) enum Object {
-  Comment(Box<Comment>),
-  Post(Box<Post>),
-  Community(Box<Community>),
-  Person(Box<DbPerson>),
-  PrivateMessage(Box<PrivateMessage>),
+/// 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,
+  data: &Data<LemmyContext>,
+) -> Result<(), LemmyError> {
+  ReceivedActivity::create(&mut data.pool(), &ap_id.clone().into()).await?;
+  Ok(())
 }
 
-pub(crate) async fn find_object_by_id(
-  context: &LemmyContext,
-  apub_id: Url,
-) -> Result<Object, LemmyError> {
-  let ap_id = apub_id.clone();
-  if let Ok(pc) = find_post_or_comment_by_id(context, ap_id.to_owned()).await {
-    return Ok(match pc {
-      PostOrComment::Post(p) => Object::Post(Box::new(*p)),
-      PostOrComment::Comment(c) => Object::Comment(Box::new(*c)),
-    });
-  }
-
-  let ap_id = apub_id.clone();
-  let person = blocking(context.pool(), move |conn| {
-    DbPerson::read_from_apub_id(conn, &ap_id.into())
-  })
-  .await?;
-  if let Ok(u) = person {
-    return Ok(Object::Person(Box::new(u)));
-  }
+#[async_trait::async_trait]
+pub trait SendActivity: Sync {
+  type Response: Sync + Send + Clone;
 
-  let ap_id = apub_id.clone();
-  let community = blocking(context.pool(), move |conn| {
-    Community::read_from_apub_id(conn, &ap_id.into())
-  })
-  .await?;
-  if let Ok(c) = community {
-    return Ok(Object::Community(Box::new(c)));
-  }
-
-  let private_message = blocking(context.pool(), move |conn| {
-    PrivateMessage::read_from_apub_id(conn, &apub_id.into())
-  })
-  .await?;
-  if let Ok(pm) = private_message {
-    return Ok(Object::PrivateMessage(Box::new(pm)));
+  async fn send_activity(
+    _request: &Self,
+    _response: &Self::Response,
+    _context: &Data<LemmyContext>,
+  ) -> Result<(), LemmyError> {
+    Ok(())
   }
-
-  Err(NotFound.into())
 }