Merge branch 'remove-swift' into 'develop'
[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 |> filter_blocked_user(for_user)
47 |> filter_blocked_domains(for_user)
48 |> search_subqueries(query_string)
49 |> union_subqueries
50 |> distinct_query()
51 |> boost_search_rank_query(for_user)
52 |> subquery()
53 |> order_by(desc: :search_rank)
54 |> maybe_restrict_local(for_user)
55 end
56
57 defp base_query(_user, false), do: User
58 defp base_query(user, true), do: User.get_followers_query(user)
59
60 defp filter_blocked_user(query, %User{info: %{blocks: blocks}})
61 when length(blocks) > 0 do
62 from(q in query, where: not (q.ap_id in ^blocks))
63 end
64
65 defp filter_blocked_user(query, _), do: query
66
67 defp filter_blocked_domains(query, %User{info: %{domain_blocks: domain_blocks}})
68 when length(domain_blocks) > 0 do
69 domains = Enum.join(domain_blocks, ",")
70
71 from(
72 q in query,
73 where: fragment("substring(ap_id from '.*://([^/]*)') NOT IN (?)", ^domains)
74 )
75 end
76
77 defp filter_blocked_domains(query, _), do: query
78
79 defp paginate(query, limit, offset) do
80 from(q in query, limit: ^limit, offset: ^offset)
81 end
82
83 defp union_subqueries({fts_subquery, trigram_subquery}) do
84 from(s in trigram_subquery, union_all: ^fts_subquery)
85 end
86
87 defp search_subqueries(base_query, query_string) do
88 {
89 fts_search_subquery(base_query, query_string),
90 trigram_search_subquery(base_query, query_string)
91 }
92 end
93
94 defp distinct_query(q) do
95 from(s in subquery(q), order_by: s.search_type, distinct: s.id)
96 end
97
98 defp maybe_resolve(true, user, query) do
99 case {limit(), user} do
100 {:all, _} -> :noop
101 {:unauthenticated, %User{}} -> User.get_or_fetch(query)
102 {:unauthenticated, _} -> :noop
103 {false, _} -> User.get_or_fetch(query)
104 end
105 end
106
107 defp maybe_resolve(_, _, _), do: :noop
108
109 defp maybe_restrict_local(q, user) do
110 case {limit(), user} do
111 {:all, _} -> restrict_local(q)
112 {:unauthenticated, %User{}} -> q
113 {:unauthenticated, _} -> restrict_local(q)
114 {false, _} -> q
115 end
116 end
117
118 defp limit, do: Pleroma.Config.get([:instance, :limit_to_local_content], :unauthenticated)
119
120 defp restrict_local(q), do: where(q, [u], u.local == true)
121
122 defp boost_search_rank_query(query, nil), do: query
123
124 defp boost_search_rank_query(query, for_user) do
125 friends_ids = User.get_friends_ids(for_user)
126 followers_ids = User.get_followers_ids(for_user)
127
128 from(u in subquery(query),
129 select_merge: %{
130 search_rank:
131 fragment(
132 """
133 CASE WHEN (?) THEN 0.5 + (?) * 1.3
134 WHEN (?) THEN 0.5 + (?) * 1.2
135 WHEN (?) THEN (?) * 1.1
136 ELSE (?) END
137 """,
138 u.id in ^friends_ids and u.id in ^followers_ids,
139 u.search_rank,
140 u.id in ^friends_ids,
141 u.search_rank,
142 u.id in ^followers_ids,
143 u.search_rank,
144 u.search_rank
145 )
146 }
147 )
148 end
149
150 @spec fts_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
151 defp fts_search_subquery(query, term) do
152 processed_query =
153 String.trim_trailing(term, "@" <> local_domain())
154 |> String.replace(~r/\W+/, " ")
155 |> String.trim()
156 |> String.split()
157 |> Enum.map(&(&1 <> ":*"))
158 |> Enum.join(" | ")
159
160 from(
161 u in query,
162 select_merge: %{
163 search_type: ^0,
164 search_rank:
165 fragment(
166 """
167 ts_rank_cd(
168 setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
169 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B'),
170 to_tsquery('simple', ?),
171 32
172 )
173 """,
174 u.nickname,
175 u.name,
176 ^processed_query
177 )
178 },
179 where:
180 fragment(
181 """
182 (setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
183 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B')) @@ to_tsquery('simple', ?)
184 """,
185 u.nickname,
186 u.name,
187 ^processed_query
188 )
189 )
190 |> User.restrict_deactivated()
191 end
192
193 @spec trigram_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
194 defp trigram_search_subquery(query, term) do
195 term = String.trim_trailing(term, "@" <> local_domain())
196
197 from(
198 u in query,
199 select_merge: %{
200 # ^1 gives 'Postgrex expected a binary, got 1' for some weird reason
201 search_type: fragment("?", 1),
202 search_rank:
203 fragment(
204 "similarity(?, trim(? || ' ' || coalesce(?, '')))",
205 ^term,
206 u.nickname,
207 u.name
208 )
209 },
210 where: fragment("trim(? || ' ' || coalesce(?, '')) % ?", u.nickname, u.name, ^term)
211 )
212 |> User.restrict_deactivated()
213 end
214
215 defp local_domain, do: Pleroma.Config.get([Pleroma.Web.Endpoint, :url, :host])
216 end