Allow dashes in domain name search
[akkoma] / lib / pleroma / user / search.ex
1 # Pleroma: A lightweight social networking server
2 # Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
3 # SPDX-License-Identifier: AGPL-3.0-only
4
5 defmodule Pleroma.User.Search do
6 alias Pleroma.EctoType.ActivityPub.ObjectValidators.Uri, as: UriType
7 alias Pleroma.Pagination
8 alias Pleroma.User
9
10 import Ecto.Query
11
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 # If this returns anything, it should bounce to the top
25 maybe_resolved = maybe_resolve(resolve, for_user, query_string)
26
27 top_user_ids =
28 []
29 |> maybe_add_resolved(maybe_resolved)
30 |> maybe_add_ap_id_match(query_string)
31 |> maybe_add_uri_match(query_string)
32
33 results =
34 query_string
35 |> search_query(for_user, following, top_user_ids)
36 |> Pagination.fetch_paginated(%{"offset" => offset, "limit" => result_limit}, :offset)
37
38 results
39 end
40
41 defp maybe_add_resolved(list, {:ok, %User{} = user}) do
42 [user.id | list]
43 end
44
45 defp maybe_add_resolved(list, _), do: list
46
47 defp maybe_add_ap_id_match(list, query) do
48 if user = User.get_cached_by_ap_id(query) do
49 [user.id | list]
50 else
51 list
52 end
53 end
54
55 defp maybe_add_uri_match(list, query) do
56 with {:ok, query} <- UriType.cast(query),
57 q = from(u in User, where: u.uri == ^query, select: u.id),
58 users = Pleroma.Repo.all(q) do
59 users ++ list
60 else
61 _ -> list
62 end
63 end
64
65 def sanitise_domain(domain) do
66 domain
67 |> String.replace(~r/[!-\,|@|?|<|>|[-`|{-~|\/|:|\s]+/, "")
68 end
69
70 defp format_query(query_string) do
71 # Strip the beginning @ off if there is a query
72 query_string = String.trim_leading(query_string, "@")
73
74 with [name, domain] <- String.split(query_string, "@") do
75 encoded_domain =
76 domain
77 |> sanitise_domain()
78 |> String.to_charlist()
79 |> :idna.encode()
80 |> to_string()
81
82 name <> "@" <> encoded_domain
83 else
84 _ -> query_string
85 end
86 end
87
88 defp search_query(query_string, for_user, following, top_user_ids) do
89 for_user
90 |> base_query(following)
91 |> filter_blocked_user(for_user)
92 |> filter_invisible_users()
93 |> filter_internal_users()
94 |> filter_blocked_domains(for_user)
95 |> fts_search(query_string)
96 |> select_top_users(top_user_ids)
97 |> trigram_rank(query_string)
98 |> boost_search_rank(for_user, top_user_ids)
99 |> subquery()
100 |> order_by(desc: :search_rank)
101 |> maybe_restrict_local(for_user)
102 |> filter_deactivated_users()
103 end
104
105 defp select_top_users(query, top_user_ids) do
106 from(u in query,
107 or_where: u.id in ^top_user_ids
108 )
109 end
110
111 defp fts_search(query, query_string) do
112 query_string = to_tsquery(query_string)
113
114 from(
115 u in query,
116 where:
117 fragment(
118 # The fragment must _exactly_ match `users_fts_index`, otherwise the index won't work
119 """
120 (
121 setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
122 setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B')
123 ) @@ to_tsquery('simple', ?)
124 """,
125 u.nickname,
126 u.name,
127 ^query_string
128 )
129 )
130 end
131
132 defp to_tsquery(query_string) do
133 String.trim_trailing(query_string, "@" <> local_domain())
134 |> String.replace(~r/[!-\/|@|[-`|{-~|:-?]+/, " ")
135 |> String.trim()
136 |> String.split()
137 |> Enum.map(&(&1 <> ":*"))
138 |> Enum.join(" | ")
139 end
140
141 # Considers nickname match, localized nickname match, name match; preferences nickname match
142 defp trigram_rank(query, query_string) do
143 from(
144 u in query,
145 select_merge: %{
146 search_rank:
147 fragment(
148 """
149 similarity(?, ?) +
150 similarity(?, regexp_replace(?, '@.+', '')) +
151 similarity(?, trim(coalesce(?, '')))
152 """,
153 ^query_string,
154 u.nickname,
155 ^query_string,
156 u.nickname,
157 ^query_string,
158 u.name
159 )
160 }
161 )
162 end
163
164 defp base_query(%User{} = user, true), do: User.get_friends_query(user)
165 defp base_query(_user, _following), do: User
166
167 defp filter_invisible_users(query) do
168 from(q in query, where: q.invisible == false)
169 end
170
171 defp filter_internal_users(query) do
172 from(q in query, where: q.actor_type != "Application")
173 end
174
175 defp filter_deactivated_users(query) do
176 from(q in query, where: q.is_active == true)
177 end
178
179 defp filter_blocked_user(query, %User{} = blocker) do
180 query
181 |> join(:left, [u], b in Pleroma.UserRelationship,
182 as: :blocks,
183 on: b.relationship_type == ^:block and b.source_id == ^blocker.id and u.id == b.target_id
184 )
185 |> where([blocks: b], is_nil(b.target_id))
186 end
187
188 defp filter_blocked_user(query, _), do: query
189
190 defp filter_blocked_domains(query, %User{domain_blocks: domain_blocks})
191 when length(domain_blocks) > 0 do
192 domains = Enum.join(domain_blocks, ",")
193
194 from(
195 q in query,
196 where: fragment("substring(ap_id from '.*://([^/]*)') NOT IN (?)", ^domains)
197 )
198 end
199
200 defp filter_blocked_domains(query, _), do: query
201
202 defp maybe_resolve(true, user, query) do
203 case {limit(), user} do
204 {:all, _} -> :noop
205 {:unauthenticated, %User{}} -> User.get_or_fetch(query)
206 {:unauthenticated, _} -> :noop
207 {false, _} -> User.get_or_fetch(query)
208 end
209 end
210
211 defp maybe_resolve(_, _, _), do: :noop
212
213 defp maybe_restrict_local(q, user) do
214 case {limit(), user} do
215 {:all, _} -> restrict_local(q)
216 {:unauthenticated, %User{}} -> q
217 {:unauthenticated, _} -> restrict_local(q)
218 {false, _} -> q
219 end
220 end
221
222 defp limit, do: Pleroma.Config.get([:instance, :limit_to_local_content], :unauthenticated)
223
224 defp restrict_local(q), do: where(q, [u], u.local == true)
225
226 defp local_domain, do: Pleroma.Config.get([Pleroma.Web.Endpoint, :url, :host])
227
228 defp boost_search_rank(query, %User{} = for_user, top_user_ids) do
229 friends_ids = User.get_friends_ids(for_user)
230 followers_ids = User.get_followers_ids(for_user)
231
232 from(u in subquery(query),
233 select_merge: %{
234 search_rank:
235 fragment(
236 """
237 CASE WHEN (?) THEN (?) * 1.5
238 WHEN (?) THEN (?) * 1.3
239 WHEN (?) THEN (?) * 1.1
240 WHEN (?) THEN 9001
241 ELSE (?) END
242 """,
243 u.id in ^friends_ids and u.id in ^followers_ids,
244 u.search_rank,
245 u.id in ^friends_ids,
246 u.search_rank,
247 u.id in ^followers_ids,
248 u.search_rank,
249 u.id in ^top_user_ids,
250 u.search_rank
251 )
252 }
253 )
254 end
255
256 defp boost_search_rank(query, _for_user, top_user_ids) do
257 from(u in subquery(query),
258 select_merge: %{
259 search_rank:
260 fragment(
261 """
262 CASE WHEN (?) THEN 9001
263 ELSE (?) END
264 """,
265 u.id in ^top_user_ids,
266 u.search_rank
267 )
268 }
269 )
270 end
271 end