path: root/server/src/apub/fetcher.rs

use crate::{
  api::site::SearchResponse,
  apub::{is_apub_id_valid, FromApub, GroupExt, PageExt, PersonExt, APUB_JSON_CONTENT_TYPE},
  blocking,
  request::{retry, RecvError},
  routes::nodeinfo::{NodeInfo, NodeInfoWellKnown},
  DbPool, LemmyError,
};
use activitystreams::object::Note;
use activitystreams_new::{base::BaseExt, prelude::*, primitives::XsdAnyUri};
use actix_web::client::Client;
use chrono::NaiveDateTime;
use diesel::{result::Error::NotFound, PgConnection};
use lemmy_db::{
  comment::{Comment, CommentForm},
  comment_view::CommentView,
  community::{Community, CommunityForm, CommunityModerator, CommunityModeratorForm},
  community_view::CommunityView,
  naive_now,
  post::{Post, PostForm},
  post_view::PostView,
  user::{UserForm, User_},
  user_view::UserView,
  Crud, Joinable, SearchType,
};
use lemmy_utils::get_apub_protocol_string;
use log::debug;
use serde::Deserialize;
use std::{fmt::Debug, time::Duration};
use url::Url;

static ACTOR_REFETCH_INTERVAL_SECONDS: i64 = 24 * 60 * 60;

// Fetch nodeinfo metadata from a remote instance.
async fn _fetch_node_info(client: &Client, domain: &str) -> Result<NodeInfo, LemmyError> {
  let well_known_uri = Url::parse(&format!(
    "{}://{}/.well-known/nodeinfo",
    get_apub_protocol_string(),
    domain
  ))?;

  let well_known = fetch_remote_object::<NodeInfoWellKnown>(client, &well_known_uri).await?;
  let nodeinfo = fetch_remote_object::<NodeInfo>(client, &well_known.links.href).await?;

  Ok(nodeinfo)
}

/// Fetch any type of ActivityPub object, handling things like HTTP headers, deserialisation,
/// timeouts etc.
pub async fn fetch_remote_object<Response>(
  client: &Client,
  url: &Url,
) -> Result<Response, LemmyError>
where
  Response: for<'de> Deserialize<'de>,
{
  if !is_apub_id_valid(&url) {
    return Err(format_err!("Activitypub uri invalid or blocked: {}", url).into());
  }

  let timeout = Duration::from_secs(60);

  let json = retry(|| {
    client
      .get(url.as_str())
      .header("Accept", APUB_JSON_CONTENT_TYPE)
      .timeout(timeout)
      .send()
  })
  .await?
  .json()
  .await
  .map_err(|e| {
    debug!("Receive error, {}", e);
    RecvError(e.to_string())
  })?;

  Ok(json)
}

/// The types of ActivityPub objects that can be fetched directly by searching for their ID.
#[serde(untagged)]
#[derive(serde::Deserialize, Debug)]
pub enum SearchAcceptedObjects {
  Person(Box<PersonExt>),
  Group(Box<GroupExt>),
  Page(Box<PageExt>),
  Comment(Box<Note>),
}

/// Attempt to parse the query as URL, and fetch an ActivityPub object from it.
///
/// Some working examples for use with the docker/federation/ setup:
/// http://lemmy_alpha:8540/c/main, or !main@lemmy_alpha:8540
/// http://lemmy_alpha:8540/u/lemmy_alpha, or @lemmy_alpha@lemmy_alpha:8540
/// http://lemmy_alpha:8540/post/3
/// http://lemmy_alpha:8540/comment/2
pub async fn search_by_apub_id(
  query: &str,
  client: &Client,
  pool: &DbPool,
) -> Result<SearchResponse, LemmyError> {
  // Parse the shorthand query url
  let query_url = if query.contains('@') {
    debug!("{}", query);
    let split = query.split('@').collect::<Vec<&str>>();

    // User type will look like ['', username, instance]
    // Community will look like [!community, instance]
    let (name, instance) = if split.len() == 3 {
      (format!("/u/{}", split[1]), split[2])
    } else if split.len() == 2 {
      if split[0].contains('!') {
        let split2 = split[0].split('!').collect::<Vec<&str>>();
        (format!("/c/{}", split2[1]), split[1])
      } else {
        return Err(format_err!("Invalid search query: {}", query).into());
      }
    } else {
      return Err(format_err!("Invalid search query: {}", query).into());
    };

    let url = format!("{}://{}{}", get_apub_protocol_string(), instance, name);
    Url::parse(&url)?
  } else {
    Url::parse(&query)?
  };

  let mut response = SearchResponse {
    type_: SearchType::All.to_string(),
    comments: vec![],
    posts: vec![],
    communities: vec![],
    users: vec![],
  };

  let response = match fetch_remote_object::<SearchAcceptedObjects>(client, &query_url).await? {
    SearchAcceptedObjects::Person(p) => {
      let user_uri = p.inner.id().unwrap().to_string();

      let user = get_or_fetch_and_upsert_remote_user(&user_uri, client, pool).await?;

      response.users = vec![blocking(pool, move |conn| UserView::read(conn, user.id)).await??];

      response
    }
    SearchAcceptedObjects::Group(g) => {
      let community_uri = g.inner.id().unwrap().to_string();

      let community =
        get_or_fetch_and_upsert_remote_community(&community_uri, client, pool).await?;

      // TODO Maybe at some point in the future, fetch all the history of a community
      // fetch_community_outbox(&c, conn)?;
      response.communities = vec![
        blocking(pool, move |conn| {
          CommunityView::read(conn, community.id, None)
        })
        .await??,
      ];

      response
    }
    SearchAcceptedObjects::Page(p) => {
      let post_form = PostForm::from_apub(&p, client, pool).await?;

      let p = blocking(pool, move |conn| upsert_post(&post_form, conn)).await??;
      response.posts = vec![blocking(pool, move |conn| PostView::read(conn, p.id, None)).await??];

      response
    }
    SearchAcceptedObjects::Comment(c) => {
      let post_url = c
        .object_props
        .get_many_in_reply_to_xsd_any_uris()
        .unwrap()
        .next()
        .unwrap()
        .to_string();

      // TODO: also fetch parent comments if any
      let post = fetch_remote_object(client, &Url::parse(&post_url)?).await?;
      let post_form = PostForm::from_apub(&post, client, pool).await?;
      let comment_form = CommentForm::from_apub(&c, client, pool).await?;

      blocking(pool, move |conn| upsert_post(&post_form, conn)).await??;
      let c = blocking(pool, move |conn| upsert_comment(&comment_form, conn)).await??;
      response.comments =
        vec![blocking(pool, move |conn| CommentView::read(conn, c.id, None)).await??];

      response
    }
  };

  Ok(response)
}

/// Check if a remote user exists, create if not found, if its too old update it.Fetch a user, insert/update it in the database and return the user.
pub async fn get_or_fetch_and_upsert_remote_user(
  apub_id: &str,
  client: &Client,
  pool: &DbPool,
) -> Result<User_, LemmyError> {
  let apub_id_owned = apub_id.to_owned();
  let user = blocking(pool, move |conn| {
    User_::read_from_actor_id(conn, &apub_id_owned)
  })
  .await?;

  match user {
    // If its older than a day, re-fetch it
    Ok(u) if !u.local && should_refetch_actor(u.last_refreshed_at) => {
      debug!("Fetching and updating from remote user: {}", apub_id);
      let person = fetch_remote_object::<PersonExt>(client, &Url::parse(apub_id)?).await?;

      let mut uf =