Page MenuHomePhorge

No OneTemporary

Size
14 KB
Referenced Files
None
Subscribers
None
diff --git a/lib/pleroma/user/search.ex b/lib/pleroma/user/search.ex
index 4426df13c..7680c2afd 100644
--- a/lib/pleroma/user/search.ex
+++ b/lib/pleroma/user/search.ex
@@ -1,198 +1,212 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.User.Search do
alias Pleroma.Repo
alias Pleroma.User
import Ecto.Query
@similarity_threshold 0.25
@limit 20
def search(query_string, opts \\ []) do
resolve = Keyword.get(opts, :resolve, false)
following = Keyword.get(opts, :following, false)
result_limit = Keyword.get(opts, :limit, @limit)
offset = Keyword.get(opts, :offset, 0)
for_user = Keyword.get(opts, :for_user)
# Strip the beginning @ off if there is a query
query_string = String.trim_leading(query_string, "@")
maybe_resolve(resolve, for_user, query_string)
{:ok, results} =
Repo.transaction(fn ->
Ecto.Adapters.SQL.query(
Repo,
"select set_limit(#{@similarity_threshold})",
[]
)
query_string
|> search_query(for_user, following)
|> paginate(result_limit, offset)
|> Repo.all()
end)
results
- |> filter_block_users(for_user)
end
- defp filter_block_users(accounts, %User{} = user) do
- Enum.filter(accounts, &(!User.blocks?(user, &1)))
- end
-
- defp filter_block_users(accounts, _), do: accounts
-
defp search_query(query_string, for_user, following) do
for_user
|> base_query(following)
+ |> filter_blocked_user(for_user)
+ |> filter_blocked_domains(for_user)
|> search_subqueries(query_string)
|> union_subqueries
|> distinct_query()
|> boost_search_rank_query(for_user)
|> subquery()
|> order_by(desc: :search_rank)
|> maybe_restrict_local(for_user)
end
defp base_query(_user, false), do: User
defp base_query(user, true), do: User.get_followers_query(user)
+ defp filter_blocked_user(query, %User{info: %{blocks: blocks}})
+ when length(blocks) > 0 do
+ from(q in query, where: not (q.ap_id in ^blocks))
+ end
+
+ defp filter_blocked_user(query, _), do: query
+
+ defp filter_blocked_domains(query, %User{info: %{domain_blocks: domain_blocks}})
+ when length(domain_blocks) > 0 do
+ domains = Enum.join(domain_blocks, ",")
+
+ from(
+ q in query,
+ where: fragment("substring(ap_id from '.*://([^/]*)') NOT IN (?)", ^domains)
+ )
+ end
+
+ defp filter_blocked_domains(query, _), do: query
+
defp paginate(query, limit, offset) do
from(q in query, limit: ^limit, offset: ^offset)
end
defp union_subqueries({fts_subquery, trigram_subquery}) do
from(s in trigram_subquery, union_all: ^fts_subquery)
end
defp search_subqueries(base_query, query_string) do
{
fts_search_subquery(base_query, query_string),
trigram_search_subquery(base_query, query_string)
}
end
defp distinct_query(q) do
from(s in subquery(q), order_by: s.search_type, distinct: s.id)
end
defp maybe_resolve(true, user, query) do
case {limit(), user} do
{:all, _} -> :noop
{:unauthenticated, %User{}} -> User.get_or_fetch(query)
{:unauthenticated, _} -> :noop
{false, _} -> User.get_or_fetch(query)
end
end
defp maybe_resolve(_, _, _), do: :noop
defp maybe_restrict_local(q, user) do
case {limit(), user} do
{:all, _} -> restrict_local(q)
{:unauthenticated, %User{}} -> q
{:unauthenticated, _} -> restrict_local(q)
{false, _} -> q
end
end
defp limit, do: Pleroma.Config.get([:instance, :limit_to_local_content], :unauthenticated)
defp restrict_local(q), do: where(q, [u], u.local == true)
defp boost_search_rank_query(query, nil), do: query
defp boost_search_rank_query(query, for_user) do
friends_ids = User.get_friends_ids(for_user)
followers_ids = User.get_followers_ids(for_user)
from(u in subquery(query),
select_merge: %{
search_rank:
fragment(
"""
CASE WHEN (?) THEN 0.5 + (?) * 1.3
WHEN (?) THEN 0.5 + (?) * 1.2
WHEN (?) THEN (?) * 1.1
ELSE (?) END
""",
u.id in ^friends_ids and u.id in ^followers_ids,
u.search_rank,
u.id in ^friends_ids,
u.search_rank,
u.id in ^followers_ids,
u.search_rank,
u.search_rank
)
}
)
end
@spec fts_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
defp fts_search_subquery(query, term) do
processed_query =
term
|> String.replace(~r/\W+/, " ")
|> String.trim()
|> String.split()
|> Enum.map(&(&1 <> ":*"))
|> Enum.join(" | ")
from(
u in query,
select_merge: %{
search_type: ^0,
search_rank:
fragment(
"""
ts_rank_cd(
setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B'),
to_tsquery('simple', ?),
32
)
""",
u.nickname,
u.name,
^processed_query
)
},
where:
fragment(
"""
(setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B')) @@ to_tsquery('simple', ?)
""",
u.nickname,
u.name,
^processed_query
)
)
|> User.restrict_deactivated()
end
@spec trigram_search_subquery(User.t() | Ecto.Query.t(), String.t()) :: Ecto.Query.t()
defp trigram_search_subquery(query, term) do
from(
u in query,
select_merge: %{
# ^1 gives 'Postgrex expected a binary, got 1' for some weird reason
search_type: fragment("?", 1),
search_rank:
fragment(
"similarity(?, trim(? || ' ' || coalesce(?, '')))",
^term,
u.nickname,
u.name
)
},
where: fragment("trim(? || ' ' || coalesce(?, '')) % ?", u.nickname, u.name, ^term)
)
|> User.restrict_deactivated()
end
end
diff --git a/test/user_search_test.exs b/test/user_search_test.exs
index 957142e67..8f8472aae 100644
--- a/test/user_search_test.exs
+++ b/test/user_search_test.exs
@@ -1,212 +1,221 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2018 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.UserSearchTest do
alias Pleroma.Repo
alias Pleroma.User
use Pleroma.DataCase
import Pleroma.Factory
setup_all do
Tesla.Mock.mock_global(fn env -> apply(HttpRequestMock, :request, [env]) end)
:ok
end
describe "User.search" do
test "accepts limit parameter" do
Enum.each(0..4, &insert(:user, %{nickname: "john#{&1}"}))
assert length(User.search("john", limit: 3)) == 3
assert length(User.search("john")) == 5
end
test "accepts offset parameter" do
Enum.each(0..4, &insert(:user, %{nickname: "john#{&1}"}))
assert length(User.search("john", limit: 3)) == 3
assert length(User.search("john", limit: 3, offset: 3)) == 2
end
test "finds a user by full or partial nickname" do
user = insert(:user, %{nickname: "john"})
Enum.each(["john", "jo", "j"], fn query ->
assert user ==
User.search(query)
|> List.first()
|> Map.put(:search_rank, nil)
|> Map.put(:search_type, nil)
end)
end
test "finds a user by full or partial name" do
user = insert(:user, %{name: "John Doe"})
Enum.each(["John Doe", "JOHN", "doe", "j d", "j", "d"], fn query ->
assert user ==
User.search(query)
|> List.first()
|> Map.put(:search_rank, nil)
|> Map.put(:search_type, nil)
end)
end
test "finds users, preferring nickname matches over name matches" do
u1 = insert(:user, %{name: "lain", nickname: "nick1"})
u2 = insert(:user, %{nickname: "lain", name: "nick1"})
assert [u2.id, u1.id] == Enum.map(User.search("lain"), & &1.id)
end
test "finds users, considering density of matched tokens" do
u1 = insert(:user, %{name: "Bar Bar plus Word Word"})
u2 = insert(:user, %{name: "Word Word Bar Bar Bar"})
assert [u2.id, u1.id] == Enum.map(User.search("bar word"), & &1.id)
end
test "finds users, ranking by similarity" do
u1 = insert(:user, %{name: "lain"})
_u2 = insert(:user, %{name: "ean"})
u3 = insert(:user, %{name: "ebn", nickname: "lain@mastodon.social"})
u4 = insert(:user, %{nickname: "lain@pleroma.soykaf.com"})
assert [u4.id, u3.id, u1.id] == Enum.map(User.search("lain@ple", for_user: u1), & &1.id)
end
test "finds users, handling misspelled requests" do
u1 = insert(:user, %{name: "lain"})
assert [u1.id] == Enum.map(User.search("laiin"), & &1.id)
end
test "finds users, boosting ranks of friends and followers" do
u1 = insert(:user)
u2 = insert(:user, %{name: "Doe"})
follower = insert(:user, %{name: "Doe"})
friend = insert(:user, %{name: "Doe"})
{:ok, follower} = User.follow(follower, u1)
{:ok, u1} = User.follow(u1, friend)
assert [friend.id, follower.id, u2.id] --
Enum.map(User.search("doe", resolve: false, for_user: u1), & &1.id) == []
end
test "finds followers of user by partial name" do
u1 = insert(:user)
u2 = insert(:user, %{name: "Jimi"})
follower_jimi = insert(:user, %{name: "Jimi Hendrix"})
follower_lizz = insert(:user, %{name: "Lizz Wright"})
friend = insert(:user, %{name: "Jimi"})
{:ok, follower_jimi} = User.follow(follower_jimi, u1)
{:ok, _follower_lizz} = User.follow(follower_lizz, u2)
{:ok, u1} = User.follow(u1, friend)
assert Enum.map(User.search("jimi", following: true, for_user: u1), & &1.id) == [
follower_jimi.id
]
assert User.search("lizz", following: true, for_user: u1) == []
end
test "find local and remote users for authenticated users" do
u1 = insert(:user, %{name: "lain"})
u2 = insert(:user, %{name: "ebn", nickname: "lain@mastodon.social", local: false})
u3 = insert(:user, %{nickname: "lain@pleroma.soykaf.com", local: false})
results =
"lain"
|> User.search(for_user: u1)
|> Enum.map(& &1.id)
|> Enum.sort()
assert [u1.id, u2.id, u3.id] == results
end
test "find only local users for unauthenticated users" do
%{id: id} = insert(:user, %{name: "lain"})
insert(:user, %{name: "ebn", nickname: "lain@mastodon.social", local: false})
insert(:user, %{nickname: "lain@pleroma.soykaf.com", local: false})
assert [%{id: ^id}] = User.search("lain")
end
test "find only local users for authenticated users when `limit_to_local_content` is `:all`" do
Pleroma.Config.put([:instance, :limit_to_local_content], :all)
%{id: id} = insert(:user, %{name: "lain"})
insert(:user, %{name: "ebn", nickname: "lain@mastodon.social", local: false})
insert(:user, %{nickname: "lain@pleroma.soykaf.com", local: false})
assert [%{id: ^id}] = User.search("lain")
Pleroma.Config.put([:instance, :limit_to_local_content], :unauthenticated)
end
test "find all users for unauthenticated users when `limit_to_local_content` is `false`" do
Pleroma.Config.put([:instance, :limit_to_local_content], false)
u1 = insert(:user, %{name: "lain"})
u2 = insert(:user, %{name: "ebn", nickname: "lain@mastodon.social", local: false})
u3 = insert(:user, %{nickname: "lain@pleroma.soykaf.com", local: false})
results =
"lain"
|> User.search()
|> Enum.map(& &1.id)
|> Enum.sort()
assert [u1.id, u2.id, u3.id] == results
Pleroma.Config.put([:instance, :limit_to_local_content], :unauthenticated)
end
test "finds a user whose name is nil" do
_user = insert(:user, %{name: "notamatch", nickname: "testuser@pleroma.amplifie.red"})
user_two = insert(:user, %{name: nil, nickname: "lain@pleroma.soykaf.com"})
assert user_two ==
User.search("lain@pleroma.soykaf.com")
|> List.first()
|> Map.put(:search_rank, nil)
|> Map.put(:search_type, nil)
end
test "does not yield false-positive matches" do
insert(:user, %{name: "John Doe"})
Enum.each(["mary", "a", ""], fn query ->
assert [] == User.search(query)
end)
end
test "works with URIs" do
user = insert(:user)
results =
User.search("http://mastodon.example.org/users/admin", resolve: true, for_user: user)
result = results |> List.first()
user = User.get_cached_by_ap_id("http://mastodon.example.org/users/admin")
assert length(results) == 1
assert user == result |> Map.put(:search_rank, nil) |> Map.put(:search_type, nil)
end
test "excludes a blocked users from search result" do
user = insert(:user, %{nickname: "Bill"})
[blocked_user | users] = Enum.map(0..3, &insert(:user, %{nickname: "john#{&1}"}))
+
+ blocked_user2 =
+ insert(
+ :user,
+ %{nickname: "john awful", ap_id: "https://awful-and-rude-instance.com/user/bully"}
+ )
+
+ User.block_domain(user, "awful-and-rude-instance.com")
User.block(user, blocked_user)
account_ids = User.search("john", for_user: refresh_record(user)) |> collect_ids
assert account_ids == collect_ids(users)
refute Enum.member?(account_ids, blocked_user.id)
+ refute Enum.member?(account_ids, blocked_user2.id)
assert length(account_ids) == 3
end
end
end

File Metadata

Mime Type
text/x-diff
Expires
Sat, Oct 10, 2:30 PM (1 d, 11 h)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
1785273
Default Alt Text
(14 KB)

Event Timeline