1 # Pleroma: A lightweight social networking server
2 # Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
3 # SPDX-License-Identifier: AGPL-3.0-only
5 defmodule Pleroma.User.Search do
6 alias Pleroma.Pagination
11 @similarity_threshold 0.25
14 def search(query_string, opts \\ []) do
15 resolve = Keyword.get(opts, :resolve, false)
16 following = Keyword.get(opts, :following, false)
17 result_limit = Keyword.get(opts, :limit, @limit)
18 offset = Keyword.get(opts, :offset, 0)
20 for_user = Keyword.get(opts, :for_user)
22 query_string = format_query(query_string)
24 maybe_resolve(resolve, for_user, query_string)
27 Repo.transaction(fn ->
28 Ecto.Adapters.SQL.query(
30 "select set_limit(#{@similarity_threshold})",
35 |> search_query(for_user, following)
36 |> Pagination.fetch_paginated(%{"offset" => offset, "limit" => result_limit}, :offset)
42 defp format_query(query_string) do
43 # Strip the beginning @ off if there is a query
44 query_string = String.trim_leading(query_string, "@")
46 with [name, domain] <- String.split(query_string, "@"),
47 formatted_domain <- String.replace(domain, ~r/[!-\-|@|[-`|{-~|\/|:]+/, "") do
48 name <> "@" <> to_string(:idna.encode(formatted_domain))
54 defp search_query(query_string, for_user, following) do
56 |> base_query(following)
57 |> filter_blocked_user(for_user)
58 |> filter_blocked_domains(for_user)
59 |> search_subqueries(query_string)
62 |> boost_search_rank_query(for_user)
64 |> order_by(desc: :search_rank)
65 |> maybe_restrict_local(for_user)
68 defp base_query(_user, false), do: User
69 defp base_query(user, true), do: User.get_followers_query(user)
71 defp filter_blocked_user(query, %User{info: %{blocks: blocks}})
72 when length(blocks) > 0 do
73 from(q in query, where: not (q.ap_id in ^blocks))
76 defp filter_blocked_user(query, _), do: query
78 defp filter_blocked_domains(query, %User{info: %{domain_blocks: domain_blocks}})
79 when length(domain_blocks) > 0 do
80 domains = Enum.join(domain_blocks, ",")
84 where: fragment("substring(ap_id from '.*://([^/]*)') NOT IN (?)", ^domains)
88 defp filter_blocked_domains(query, _), do: query
90 defp union_subqueries({fts_subquery, trigram_subquery}) do
91 from(s in trigram_subquery, union_all: ^fts_subquery)
94 defp search_subqueries(base_query, query_string) do
96 fts_search_subquery(base_query, query_string),
97 trigram_search_subquery(base_query, query_string)
101 defp distinct_query(q) do
102 from(s in subquery(q), order_by: s.search_type, distinct: s.id)
105 defp maybe_resolve(true, user, query) do
106 case {limit(), user} do
108 {:unauthenticated, %User{}} -> User.get_or_fetch(query)
109 {:unauthenticated, _} -> :noop
110 {false, _} -> User.get_or_fetch(query)
114 defp maybe_resolve(_, _, _), do: :noop
116 defp maybe_restrict_local(q, user) do
117 case {limit(), user} do
118 {:all, _} -> restrict_local(q)
119 {:unauthenticated, %User{}} -> q
120 {:unauthenticated, _} -> restrict_local(q)
125 defp limit, do: Pleroma.Config.get([:instance, :limit_to_local_content], :unauthenticated)
127 defp restrict_local(q), do: where(q, [u], u.local == true)
129 defp boost_search_rank_query(query, nil), do: query
131 defp boost_search_rank_query(query, for_user) do
132 friends_ids = User.get_friends_ids(for_user)
133 followers_ids = User.get_followers_ids(for_user)
135 from(u in subquery(query),
140 CASE WHEN (?) THEN 0.5 + (?) * 1.3
141 WHEN (?) THEN 0.5 + (?) * 1.2
142 WHEN (?) THEN (?) * 1.1
145 u.id in ^friends_ids and u.id in ^followers_ids,
147 u.id in ^friends_ids,
149 u.id in ^followers_ids,
157 @spec fts_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
158 defp fts_search_subquery(query, term) do
160 String.trim_trailing(term, "@" <> local_domain())
161 |> String.replace(~r/[!-\/|@|[-`|{-~|:-?]+/, " ")
164 |> Enum.map(&(&1 <> ":*"))
175 setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
176 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B'),
177 to_tsquery('simple', ?),
189 (setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
190 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B')) @@ to_tsquery('simple', ?)
197 |> User.restrict_deactivated()
200 @spec trigram_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
201 defp trigram_search_subquery(query, term) do
202 term = String.trim_trailing(term, "@" <> local_domain())
207 # ^1 gives 'Postgrex expected a binary, got 1' for some weird reason
208 search_type: fragment("?", 1),
211 "similarity(?, trim(? || ' ' || coalesce(?, '')))",
217 where: fragment("trim(? || ' ' || coalesce(?, '')) % ?", u.nickname, u.name, ^term)
219 |> User.restrict_deactivated()
222 defp local_domain, do: Pleroma.Config.get([Pleroma.Web.Endpoint, :url, :host])