bc05639b53b5fb7ffac420fb4bdd8f727fd7852a
[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.Pagination
7 alias Pleroma.Repo
8 alias Pleroma.User
9 import Ecto.Query
10
11 @similarity_threshold 0.25
12 @limit 20
13
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)
19
20 for_user = Keyword.get(opts, :for_user)
21
22 query_string = format_query(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 |> Pagination.fetch_paginated(%{"offset" => offset, "limit" => result_limit}, :offset)
37 |> Map.get(:items)
38 end)
39
40 results
41 end
42
43 defp format_query(query_string) do
44 # Strip the beginning @ off if there is a query
45 query_string = String.trim_leading(query_string, "@")
46
47 with [name, domain] <- String.split(query_string, "@"),
48 formatted_domain <- String.replace(domain, ~r/[!-\-|@|[-`|{-~|\/|:|\s]+/, "") do
49 name <> "@" <> to_string(:idna.encode(formatted_domain))
50 else
51 _ -> query_string
52 end
53 end
54
55 defp search_query(query_string, for_user, following) do
56 for_user
57 |> base_query(following)
58 |> filter_blocked_user(for_user)
59 |> filter_blocked_domains(for_user)
60 |> search_subqueries(query_string)
61 |> union_subqueries
62 |> distinct_query()
63 |> boost_search_rank_query(for_user)
64 |> subquery()
65 |> order_by(desc: :search_rank)
66 |> maybe_restrict_local(for_user)
67 end
68
69 defp base_query(_user, false), do: User
70 defp base_query(user, true), do: User.get_followers_query(user)
71
72 defp filter_blocked_user(query, %User{info: %{blocks: blocks}})
73 when length(blocks) > 0 do
74 from(q in query, where: not (q.ap_id in ^blocks))
75 end
76
77 defp filter_blocked_user(query, _), do: query
78
79 defp filter_blocked_domains(query, %User{info: %{domain_blocks: domain_blocks}})
80 when length(domain_blocks) > 0 do
81 domains = Enum.join(domain_blocks, ",")
82
83 from(
84 q in query,
85 where: fragment("substring(ap_id from '.*://([^/]*)') NOT IN (?)", ^domains)
86 )
87 end
88
89 defp filter_blocked_domains(query, _), do: query
90
91 defp union_subqueries({fts_subquery, trigram_subquery}) do
92 from(s in trigram_subquery, union_all: ^fts_subquery)
93 end
94
95 defp search_subqueries(base_query, query_string) do
96 {
97 fts_search_subquery(base_query, query_string),
98 trigram_search_subquery(base_query, query_string)
99 }
100 end
101
102 defp distinct_query(q) do
103 from(s in subquery(q), order_by: s.search_type, distinct: s.id)
104 end
105
106 defp maybe_resolve(true, user, query) do
107 case {limit(), user} do
108 {:all, _} -> :noop
109 {:unauthenticated, %User{}} -> User.get_or_fetch(query)
110 {:unauthenticated, _} -> :noop
111 {false, _} -> User.get_or_fetch(query)
112 end
113 end
114
115 defp maybe_resolve(_, _, _), do: :noop
116
117 defp maybe_restrict_local(q, user) do
118 case {limit(), user} do
119 {:all, _} -> restrict_local(q)
120 {:unauthenticated, %User{}} -> q
121 {:unauthenticated, _} -> restrict_local(q)
122 {false, _} -> q
123 end
124 end
125
126 defp limit, do: Pleroma.Config.get([:instance, :limit_to_local_content], :unauthenticated)
127
128 defp restrict_local(q), do: where(q, [u], u.local == true)
129
130 defp boost_search_rank_query(query, nil), do: query
131
132 defp boost_search_rank_query(query, for_user) do
133 friends_ids = User.get_friends_ids(for_user)
134 followers_ids = User.get_followers_ids(for_user)
135
136 from(u in subquery(query),
137 select_merge: %{
138 search_rank:
139 fragment(
140 """
141 CASE WHEN (?) THEN 0.5 + (?) * 1.3
142 WHEN (?) THEN 0.5 + (?) * 1.2
143 WHEN (?) THEN (?) * 1.1
144 ELSE (?) END
145 """,
146 u.id in ^friends_ids and u.id in ^followers_ids,
147 u.search_rank,
148 u.id in ^friends_ids,
149 u.search_rank,
150 u.id in ^followers_ids,
151 u.search_rank,
152 u.search_rank
153 )
154 }
155 )
156 end
157
158 @spec fts_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
159 defp fts_search_subquery(query, term) do
160 processed_query =
161 String.trim_trailing(term, "@" <> local_domain())
162 |> String.replace(~r/[!-\/|@|[-`|{-~|:-?]+/, " ")
163 |> String.trim()
164 |> String.split()
165 |> Enum.map(&(&1 <> ":*"))
166 |> Enum.join(" | ")
167
168 from(
169 u in query,
170 select_merge: %{
171 search_type: ^0,
172 search_rank:
173 fragment(
174 """
175 ts_rank_cd(
176 setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
177 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B'),
178 to_tsquery('simple', ?),
179 32
180 )
181 """,
182 u.nickname,
183 u.name,
184 ^processed_query
185 )
186 },
187 where:
188 fragment(
189 """
190 (setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
191 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B')) @@ to_tsquery('simple', ?)
192 """,
193 u.nickname,
194 u.name,
195 ^processed_query
196 )
197 )
198 |> User.restrict_deactivated()
199 end
200
201 @spec trigram_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
202 defp trigram_search_subquery(query, term) do
203 term = String.trim_trailing(term, "@" <> local_domain())
204
205 from(
206 u in query,
207 select_merge: %{
208 # ^1 gives 'Postgrex expected a binary, got 1' for some weird reason
209 search_type: fragment("?", 1),
210 search_rank:
211 fragment(
212 "similarity(?, trim(? || ' ' || coalesce(?, '')))",
213 ^term,
214 u.nickname,
215 u.name
216 )
217 },
218 where: fragment("trim(? || ' ' || coalesce(?, '')) % ?", u.nickname, u.name, ^term)
219 )
220 |> User.restrict_deactivated()
221 end
222
223 defp local_domain, do: Pleroma.Config.get([Pleroma.Web.Endpoint, :url, :host])
224 end