Merge branch 'develop' into feature/digest-email
[akkoma] / lib / pleroma / user / search.ex
1 # Pleroma: A lightweight social networking server
2 # Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
3 # SPDX-License-Identifier: AGPL-3.0-only
4
5 defmodule Pleroma.User.Search do
6 alias Pleroma.Repo
7 alias Pleroma.User
8 import Ecto.Query
9
10 @similarity_threshold 0.25
11 @limit 20
12
13 def search(query_string, opts \\ []) do
14 resolve = Keyword.get(opts, :resolve, false)
15 following = Keyword.get(opts, :following, false)
16 result_limit = Keyword.get(opts, :limit, @limit)
17 offset = Keyword.get(opts, :offset, 0)
18
19 for_user = Keyword.get(opts, :for_user)
20
21 # Strip the beginning @ off if there is a query
22 query_string = String.trim_leading(query_string, "@")
23
24 maybe_resolve(resolve, for_user, query_string)
25
26 {:ok, results} =
27 Repo.transaction(fn ->
28 Ecto.Adapters.SQL.query(
29 Repo,
30 "select set_limit(#{@similarity_threshold})",
31 []
32 )
33
34 query_string
35 |> search_query(for_user, following)
36 |> paginate(result_limit, offset)
37 |> Repo.all()
38 end)
39
40 results
41 end
42
43 defp search_query(query_string, for_user, following) do
44 for_user
45 |> base_query(following)
46 |> search_subqueries(query_string)
47 |> union_subqueries
48 |> distinct_query()
49 |> boost_search_rank_query(for_user)
50 |> subquery()
51 |> order_by(desc: :search_rank)
52 |> maybe_restrict_local(for_user)
53 end
54
55 defp base_query(_user, false), do: User
56 defp base_query(user, true), do: User.get_followers_query(user)
57
58 defp paginate(query, limit, offset) do
59 from(q in query, limit: ^limit, offset: ^offset)
60 end
61
62 defp union_subqueries({fts_subquery, trigram_subquery}) do
63 from(s in trigram_subquery, union_all: ^fts_subquery)
64 end
65
66 defp search_subqueries(base_query, query_string) do
67 {
68 fts_search_subquery(base_query, query_string),
69 trigram_search_subquery(base_query, query_string)
70 }
71 end
72
73 defp distinct_query(q) do
74 from(s in subquery(q), order_by: s.search_type, distinct: s.id)
75 end
76
77 defp maybe_resolve(true, user, query) do
78 case {limit(), user} do
79 {:all, _} -> :noop
80 {:unauthenticated, %User{}} -> User.get_or_fetch(query)
81 {:unauthenticated, _} -> :noop
82 {false, _} -> User.get_or_fetch(query)
83 end
84 end
85
86 defp maybe_resolve(_, _, _), do: :noop
87
88 defp maybe_restrict_local(q, user) do
89 case {limit(), user} do
90 {:all, _} -> restrict_local(q)
91 {:unauthenticated, %User{}} -> q
92 {:unauthenticated, _} -> restrict_local(q)
93 {false, _} -> q
94 end
95 end
96
97 defp limit, do: Pleroma.Config.get([:instance, :limit_to_local_content], :unauthenticated)
98
99 defp restrict_local(q), do: where(q, [u], u.local == true)
100
101 defp boost_search_rank_query(query, nil), do: query
102
103 defp boost_search_rank_query(query, for_user) do
104 friends_ids = User.get_friends_ids(for_user)
105 followers_ids = User.get_followers_ids(for_user)
106
107 from(u in subquery(query),
108 select_merge: %{
109 search_rank:
110 fragment(
111 """
112 CASE WHEN (?) THEN 0.5 + (?) * 1.3
113 WHEN (?) THEN 0.5 + (?) * 1.2
114 WHEN (?) THEN (?) * 1.1
115 ELSE (?) END
116 """,
117 u.id in ^friends_ids and u.id in ^followers_ids,
118 u.search_rank,
119 u.id in ^friends_ids,
120 u.search_rank,
121 u.id in ^followers_ids,
122 u.search_rank,
123 u.search_rank
124 )
125 }
126 )
127 end
128
129 @spec fts_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
130 defp fts_search_subquery(query, term) do
131 processed_query =
132 term
133 |> String.replace(~r/\W+/, " ")
134 |> String.trim()
135 |> String.split()
136 |> Enum.map(&(&1 <> ":*"))
137 |> Enum.join(" | ")
138
139 from(
140 u in query,
141 select_merge: %{
142 search_type: ^0,
143 search_rank:
144 fragment(
145 """
146 ts_rank_cd(
147 setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
148 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B'),
149 to_tsquery('simple', ?),
150 32
151 )
152 """,
153 u.nickname,
154 u.name,
155 ^processed_query
156 )
157 },
158 where:
159 fragment(
160 """
161 (setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
162 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B')) @@ to_tsquery('simple', ?)
163 """,
164 u.nickname,
165 u.name,
166 ^processed_query
167 )
168 )
169 |> User.restrict_deactivated()
170 end
171
172 @spec trigram_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
173 defp trigram_search_subquery(query, term) do
174 from(
175 u in query,
176 select_merge: %{
177 # ^1 gives 'Postgrex expected a binary, got 1' for some weird reason
178 search_type: fragment("?", 1),
179 search_rank:
180 fragment(
181 "similarity(?, trim(? || ' ' || coalesce(?, '')))",
182 ^term,
183 u.nickname,
184 u.name
185 )
186 },
187 where: fragment("trim(? || ' ' || coalesce(?, '')) % ?", u.nickname, u.name, ^term)
188 )
189 |> User.restrict_deactivated()
190 end
191 end