X-Git-Url: http://these/git/?a=blobdiff_plain;f=crates%2Fapub%2Fsrc%2Flib.rs;h=d4e34f9e740fd2bb3b444b759ec7e985c9841c93;hb=70fae9d68d65b1e4d153e30d3c065cc315b75eaf;hp=cc9d5f4875ee97beb11d9768c7dda00976f4b6eb;hpb=9e0fa99c695df5d1c5e615d6f5de2b9d4bc91205;p=lemmy.git

diff --git a/crates/apub/src/lib.rs b/crates/apub/src/lib.rs
index cc9d5f48..d4e34f9e 100644
--- a/crates/apub/src/lib.rs
+++ b/crates/apub/src/lib.rs
@@ -1,26 +1,65 @@
+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;
-mod context;
+pub(crate) mod activity_lists;
+pub mod api;
+pub(crate) mod collections;
 pub mod fetcher;
 pub mod http;
-pub mod migrations;
+pub(crate) mod mentions;
 pub mod objects;
-
-use crate::fetcher::post_or_comment::PostOrComment;
-use anyhow::{anyhow, Context};
-use lemmy_api_common::blocking;
-use lemmy_apub_lib::{activity_queue::send_activity, traits::ActorType};
-use lemmy_db_schema::{
-  newtypes::{CommunityId, DbUrl},
-  source::{activity::Activity, person::Person},
-  DbPool,
-};
-use lemmy_db_views_actor::community_person_ban_view::CommunityPersonBanView;
-use lemmy_utils::{location_info, settings::structs::Settings, LemmyError};
-use lemmy_websocket::LemmyContext;
-use log::info;
-use serde::Serialize;
-use std::net::IpAddr;
-use url::{ParseError, Url};
+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).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.
 ///
@@ -29,238 +68,141 @@ use url::{ParseError, Url};
 /// - 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)
-///
-pub(crate) fn check_is_apub_id_valid(
-  apub_id: &Url,
-  use_strict_allowlist: bool,
-  settings: &Settings,
-) -> Result<(), LemmyError> {
-  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(),
-      )
-    };
-  }
-
-  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 apub_id.scheme() != settings.get_protocol_string() {
-    return Err(anyhow!("invalid apub id scheme {}: {}", apub_id.scheme(), apub_id).into());
+#[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();
+
+  if !local_site_data
+    .local_site
+    .as_ref()
+    .map(|l| l.federation_enabled)
+    .unwrap_or(true)
+  {
+    Err(LemmyErrorType::FederationDisabled)?;
   }
 
-  // TODO: might be good to put the part above in one method, and below in another
-  //       (which only gets called in apub::objects)
-  //        -> no that doesnt make sense, we still need the code below for blocklist and strict allowlist
-  if let Some(blocked) = settings.to_owned().federation.blocked_instances {
-    if blocked.contains(&domain) {
-      return Err(anyhow!("{} is in federation blocklist", domain).into());
-    }
+  if local_site_data
+    .blocked_instances
+    .iter()
+    .any(|i| domain.eq(&i.domain))
+  {
+    Err(LemmyErrorType::DomainBlocked(domain.clone()))?;
   }
 
-  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 use_strict_allowlist || 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(anyhow!("{} not in federation allowlist", domain).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))
+  {
+    Err(LemmyErrorType::DomainNotInAllowList(domain))?;
   }
 
   Ok(())
 }
 
-#[async_trait::async_trait(?Send)]
-pub trait CommunityType {
-  fn followers_url(&self) -> Url;
-  async fn get_follower_inboxes(
-    &self,
-    pool: &DbPool,
-    settings: &Settings,
-  ) -> Result<Vec<Url>, LemmyError>;
+#[derive(Clone)]
+pub(crate) struct LocalSiteData {
+  local_site: Option<LocalSite>,
+  allowed_instances: Vec<Instance>,
+  blocked_instances: Vec<Instance>,
 }
 
-pub enum EndpointType {
-  Community,
-  Person,
-  Post,
-  Comment,
-  PrivateMessage,
-}
-
-/// Generates an apub endpoint for a given domain, IE xyz.tld
-fn generate_apub_endpoint_for_domain(
-  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())
-}
-
-/// Generates the ActivityPub ID for a given object type and ID.
-pub fn generate_apub_endpoint(
-  endpoint_type: EndpointType,
-  name: &str,
-  protocol_and_hostname: &str,
-) -> Result<DbUrl, ParseError> {
-  generate_apub_endpoint_for_domain(endpoint_type, name, protocol_and_hostname)
-}
-
-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: 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())
-}
-
-/// Takes in a shortname of the type dessalines@xyz.tld or dessalines (assumed to be local), and outputs the actor id.
-/// Used in the API for communities and users.
-pub fn build_actor_id_from_shortname(
-  endpoint_type: EndpointType,
-  short_name: &str,
-  settings: &Settings,
-) -> Result<DbUrl, ParseError> {
-  let split = short_name.split('@').collect::<Vec<&str>>();
-
-  let name = split[0];
-
-  // If there's no @, its local
-  let domain = if split.len() == 1 {
-    settings.get_protocol_and_hostname()
-  } else {
-    format!("{}://{}", settings.get_protocol_string(), split[1])
-  };
-
-  generate_apub_endpoint_for_domain(endpoint_type, name, &domain)
-}
-
-/// Store a sent or received activity in the database, for logging purposes. These records are not
-/// persistent.
-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) 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?,
+  )
 }
 
-async fn check_community_or_site_ban(
-  person: &Person,
-  community_id: CommunityId,
-  pool: &DbPool,
+pub(crate) async fn check_apub_id_valid_with_strictness(
+  apub_id: &Url,
+  is_strict: bool,
+  context: &LemmyContext,
 ) -> Result<(), LemmyError> {
-  if person.banned {
-    return Err(anyhow!("Person is banned from site").into());
-  }
-  let person_id = person.id;
-  let is_banned =
-    move |conn: &'_ _| CommunityPersonBanView::get(conn, person_id, community_id).is_ok();
-  if blocking(pool, is_banned).await? {
-    return Err(anyhow!("Person is banned from community").into());
+  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 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(crate) async fn send_lemmy_activity<T: Serialize>(
-  context: &LemmyContext,
-  activity: &T,
-  activity_id: &Url,
-  actor: &dyn ActorType,
-  inboxes: Vec<Url>,
-  sensitive: bool,
+/// 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> {
-  if !context.settings().federation.enabled || inboxes.is_empty() {
-    return Ok(());
-  }
-
-  info!("Sending activity {}", activity_id.to_string());
-
-  // Don't send anything to ourselves
-  // TODO: this should be a debug assert
-  let hostname = context.settings().get_hostname_without_port()?;
-  let inboxes: Vec<&Url> = inboxes
-    .iter()
-    .filter(|i| i.domain().expect("valid inbox url") != hostname)
-    .collect();
-
-  let serialised_activity = serde_json::to_string(&activity)?;
+  ReceivedActivity::create(&mut data.pool(), &ap_id.clone().into()).await?;
+  Ok(())
+}
 
-  insert_activity(
-    activity_id,
-    serialised_activity.clone(),
-    true,
-    sensitive,
-    context.pool(),
-  )
-  .await?;
+#[async_trait::async_trait]
+pub trait SendActivity: Sync {
+  type Response: Sync + Send + Clone;
 
-  send_activity(
-    serialised_activity,
-    actor,
-    inboxes,
-    context.client(),
-    context.activity_queue(),
-  )
-  .await
+  async fn send_activity(
+    _request: &Self,
+    _response: &Self::Response,
+    _context: &Data<LemmyContext>,
+  ) -> Result<(), LemmyError> {
+    Ok(())
+  }
 }