X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fuser%2Fsearch.ex;h=03f2c552f2f29b82cf20909d2d830ea284f1dc85;hb=05b5241314182c5aab2907e27d4c5f46d7617f56;hp=6b55df483b76944ba6276f25033ea30f76841628;hpb=6cb31edd76fd42a0e33bc365d982cd02e3578d6c;p=akkoma diff --git a/lib/pleroma/user/search.ex b/lib/pleroma/user/search.ex index 6b55df483..03f2c552f 100644 --- a/lib/pleroma/user/search.ex +++ b/lib/pleroma/user/search.ex @@ -1,10 +1,12 @@ # Pleroma: A lightweight social networking server -# Copyright © 2017-2019 Pleroma Authors +# Copyright © 2017-2020 Pleroma Authors # SPDX-License-Identifier: AGPL-3.0-only defmodule Pleroma.User.Search do + alias Pleroma.EctoType.ActivityPub.ObjectValidators.Uri, as: UriType alias Pleroma.Pagination alias Pleroma.User + import Ecto.Query @limit 20 @@ -19,42 +21,87 @@ defmodule Pleroma.User.Search do query_string = format_query(query_string) - maybe_resolve(resolve, for_user, query_string) + # If this returns anything, it should bounce to the top + maybe_resolved = maybe_resolve(resolve, for_user, query_string) + + top_user_ids = + [] + |> maybe_add_resolved(maybe_resolved) + |> maybe_add_ap_id_match(query_string) + |> maybe_add_uri_match(query_string) results = query_string - |> search_query(for_user, following) + |> search_query(for_user, following, top_user_ids) |> Pagination.fetch_paginated(%{"offset" => offset, "limit" => result_limit}, :offset) results end + defp maybe_add_resolved(list, {:ok, %User{} = user}) do + [user.id | list] + end + + defp maybe_add_resolved(list, _), do: list + + defp maybe_add_ap_id_match(list, query) do + if user = User.get_cached_by_ap_id(query) do + [user.id | list] + else + list + end + end + + defp maybe_add_uri_match(list, query) do + with {:ok, query} <- UriType.cast(query), + %User{} = user <- Pleroma.Repo.get_by(User, uri: query) do + [user.id | list] + else + _ -> list + end + end + defp format_query(query_string) do # Strip the beginning @ off if there is a query query_string = String.trim_leading(query_string, "@") - with [name, domain] <- String.split(query_string, "@"), - formatted_domain <- String.replace(domain, ~r/[!-\-|@|[-`|{-~|\/|:|\s]+/, "") do - name <> "@" <> to_string(:idna.encode(formatted_domain)) + with [name, domain] <- String.split(query_string, "@") do + encoded_domain = + domain + |> String.replace(~r/[!-\-|@|[-`|{-~|\/|:|\s]+/, "") + |> String.to_charlist() + |> :idna.encode() + |> to_string() + + name <> "@" <> encoded_domain else _ -> query_string end end - defp search_query(query_string, for_user, following) do + defp search_query(query_string, for_user, following, top_user_ids) do for_user |> base_query(following) |> filter_blocked_user(for_user) |> filter_invisible_users() + |> filter_discoverable_users() + |> filter_internal_users() |> filter_blocked_domains(for_user) |> fts_search(query_string) + |> select_top_users(top_user_ids) |> trigram_rank(query_string) - |> boost_search_rank(for_user) + |> boost_search_rank(for_user, top_user_ids) |> subquery() |> order_by(desc: :search_rank) |> maybe_restrict_local(for_user) end + defp select_top_users(query, top_user_ids) do + from(u in query, + or_where: u.id in ^top_user_ids + ) + end + defp fts_search(query, query_string) do query_string = to_tsquery(query_string) @@ -62,11 +109,15 @@ defmodule Pleroma.User.Search do u in query, where: fragment( + # The fragment must _exactly_ match `users_fts_index`, otherwise the index won't work """ - (to_tsvector('simple', ?) || to_tsvector('simple', ?)) @@ to_tsquery('simple', ?) + ( + setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') || + setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B') + ) @@ to_tsquery('simple', ?) """, - u.name, u.nickname, + u.name, ^query_string ) ) @@ -81,28 +132,44 @@ defmodule Pleroma.User.Search do |> Enum.join(" | ") end + # Considers nickname match, localized nickname match, name match; preferences nickname match defp trigram_rank(query, query_string) do from( u in query, select_merge: %{ search_rank: fragment( - "similarity(?, trim(? || ' ' || coalesce(?, '')))", + """ + similarity(?, ?) + + similarity(?, regexp_replace(?, '@.+', '')) + + similarity(?, trim(coalesce(?, ''))) + """, + ^query_string, + u.nickname, ^query_string, u.nickname, + ^query_string, u.name ) } ) end - defp base_query(_user, false), do: User - defp base_query(user, true), do: User.get_followers_query(user) + defp base_query(%User{} = user, true), do: User.get_friends_query(user) + defp base_query(_user, _following), do: User defp filter_invisible_users(query) do from(q in query, where: q.invisible == false) end + defp filter_discoverable_users(query) do + from(q in query, where: q.discoverable == true) + end + + defp filter_internal_users(query) do + from(q in query, where: q.actor_type != "Application") + end + defp filter_blocked_user(query, %User{} = blocker) do query |> join(:left, [u], b in Pleroma.UserRelationship, @@ -152,7 +219,7 @@ defmodule Pleroma.User.Search do defp local_domain, do: Pleroma.Config.get([Pleroma.Web.Endpoint, :url, :host]) - defp boost_search_rank(query, %User{} = for_user) do + defp boost_search_rank(query, %User{} = for_user, top_user_ids) do friends_ids = User.get_friends_ids(for_user) followers_ids = User.get_followers_ids(for_user) @@ -164,6 +231,7 @@ defmodule Pleroma.User.Search do CASE WHEN (?) THEN (?) * 1.5 WHEN (?) THEN (?) * 1.3 WHEN (?) THEN (?) * 1.1 + WHEN (?) THEN 9001 ELSE (?) END """, u.id in ^friends_ids and u.id in ^followers_ids, @@ -172,11 +240,26 @@ defmodule Pleroma.User.Search do u.search_rank, u.id in ^followers_ids, u.search_rank, + u.id in ^top_user_ids, u.search_rank ) } ) end - defp boost_search_rank(query, _for_user), do: query + defp boost_search_rank(query, _for_user, top_user_ids) do + from(u in subquery(query), + select_merge: %{ + search_rank: + fragment( + """ + CASE WHEN (?) THEN 9001 + ELSE (?) END + """, + u.id in ^top_user_ids, + u.search_rank + ) + } + ) + end end