use crate::fetcher::post_or_comment::PostOrComment;
-use anyhow::{anyhow, Context};
-use lemmy_api_common::blocking;
-use lemmy_db_schema::{newtypes::DbUrl, source::activity::Activity, DbPool};
-use lemmy_utils::{location_info, settings::structs::Settings, LemmyError};
-use serde::{Deserialize, Deserializer};
-use std::net::IpAddr;
-use url::{ParseError, Url};
+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(crate) mod activity_lists;
+pub mod api;
pub(crate) mod collections;
-mod context;
pub mod fetcher;
pub mod http;
pub(crate) mod mentions;
-pub mod migrations;
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")
+});
+
+#[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.
///
/// In particular, it checks for:
/// - 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))]
-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 {
- let err = anyhow!(
- "Trying to connect with {}, but federation is disabled",
- domain
- );
- Err(LemmyError::from_error_message(err, "federation_disabled"))
- };
+#[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() {
- let err = anyhow!("invalid hostname {}: {}", host, apub_id);
- return Err(LemmyError::from_error_message(err, "invalid_hostname"));
- }
-
- if apub_id.scheme() != settings.get_protocol_string() {
- let err = anyhow!("invalid apub id scheme {}: {}", apub_id.scheme(), apub_id);
- return Err(LemmyError::from_error_message(err, "invalid_scheme"));
- }
-
- // 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) {
- let err = anyhow!("{} is in federation blocklist", domain);
- return Err(LemmyError::from_error_message(err, "federation_blocked"));
- }
+ if local_site_data
+ .blocked_instances
+ .iter()
+ .any(|i| domain.eq(&i.domain))
+ {
+ return Err("Domain is blocked");
}
- 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) {
- let err = anyhow!("{} not in federation allowlist", domain);
- return Err(LemmyError::from_error_message(
- err,
- "federation_not_allowed",
- ));
- }
- }
+ // 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");
}
Ok(())
}
-pub(crate) fn deserialize_one_or_many<'de, T, D>(deserializer: D) -> Result<Vec<T>, D::Error>
-where
- T: Deserialize<'de>,
- D: Deserializer<'de>,
-{
- #[derive(Deserialize)]
- #[serde(untagged)]
- enum OneOrMany<T> {
- One(T),
- Many(Vec<T>),
- }
-
- let result: OneOrMany<T> = Deserialize::deserialize(deserializer)?;
- Ok(match result {
- OneOrMany::Many(list) => list,
- OneOrMany::One(value) => vec![value],
- })
-}
-
-pub(crate) fn deserialize_one<'de, T, D>(deserializer: D) -> Result<[T; 1], D::Error>
-where
- T: Deserialize<'de>,
- D: Deserializer<'de>,
-{
- #[derive(Deserialize)]
- #[serde(untagged)]
- enum MaybeArray<T> {
- Simple(T),
- Array([T; 1]),
- }
-
- let result: MaybeArray<T> = Deserialize::deserialize(deserializer)?;
- Ok(match result {
- MaybeArray::Simple(value) => [value],
- MaybeArray::Array(value) => value,
- })
-}
-
-pub(crate) fn deserialize_skip_error<'de, T, D>(deserializer: D) -> Result<T, D::Error>
-where
- T: Deserialize<'de> + Default,
- D: Deserializer<'de>,
-{
- let result = Deserialize::deserialize(deserializer);
- Ok(match result {
- Ok(o) => o,
- Err(_) => Default::default(),
- })
-}
-
-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 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(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 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(crate) async fn check_apub_id_valid_with_strictness(
+ apub_id: &Url,
+ is_strict: bool,
+ context: &LemmyContext,
+) -> 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(());
+ }
-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())
+ 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!"),
+ })?;
+
+ // 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 fn generate_outbox_url(actor_id: &DbUrl) -> Result<DbUrl, ParseError> {
- Ok(Url::parse(&format!("{}/outbox", actor_id))?.into())
+/// 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(())
}
-fn generate_moderators_url(community_id: &DbUrl) -> Result<DbUrl, LemmyError> {
- Ok(Url::parse(&format!("{}/moderators", community_id))?.into())
-}
+#[async_trait::async_trait]
+pub trait SendActivity: Sync {
+ type Response: Sync + Send + Clone;
-/// 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(
- 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??,
- )
+ async fn send_activity(
+ _request: &Self,
+ _response: &Self::Response,
+ _context: &Data<LemmyContext>,
+ ) -> Result<(), LemmyError> {
+ Ok(())
+ }
}
projects / lemmy.git / blobdiff