1 # Pleroma: A lightweight social networking server
2 # Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
3 # SPDX-License-Identifier: AGPL-3.0-only
5 defmodule Pleroma.Migrators.HashtagsTableMigrator do
12 alias __MODULE__.State
14 alias Pleroma.DataMigration
19 defdelegate state(), to: State, as: :get
20 defdelegate put_stat(key, value), to: State, as: :put
21 defdelegate increment_stat(key, increment), to: State, as: :increment
23 defdelegate data_migration(), to: DataMigration, as: :populate_hashtags_table
25 @reg_name {:global, __MODULE__}
27 def whereis, do: GenServer.whereis(@reg_name)
32 GenServer.start_link(__MODULE__, nil, name: @reg_name)
41 {:ok, nil, {:continue, :init_state}}
45 def handle_continue(:init_state, _state) do
46 {:ok, _} = State.start_link(nil)
48 put_stat(:status, :init)
51 manual_migrations = Config.get([:instance, :manual_data_migrations], [])
54 Config.get(:env) == :test ->
55 put_stat(:status, :noop)
58 put_stat(:status, :halt)
59 put_stat(:message, "Data migration does not exist.")
61 dm.state == :manual or dm.name in manual_migrations ->
62 put_stat(:status, :noop)
63 put_stat(:message, "Data migration is in manual execution state.")
65 dm.state == :complete ->
69 send(self(), :migrate_hashtags)
76 def handle_info(:migrate_hashtags, state) do
77 data_migration = data_migration()
79 persistent_data = Map.take(data_migration.data, ["max_processed_id"])
81 {:ok, data_migration} =
82 DataMigration.update(data_migration, %{state: :running, data: persistent_data})
84 put_stat(:status, :running)
86 Logger.info("Starting transferring object embedded hashtags to `hashtags` table...")
88 max_processed_id = data_migration.data["max_processed_id"] || 0
90 # Note: most objects have Mention-type AS2 tags and no hashtags (but we can't filter them out)
93 left_join: hashtag in assoc(object, :hashtags),
94 where: object.id > ^max_processed_id,
95 where: is_nil(hashtag.id),
97 fragment("(?)->'tag' IS NOT NULL AND (?)->'tag' != '[]'::jsonb", object.data, object.data),
100 tag: fragment("(?)->'tag'", object.data)
103 |> Repo.chunk_stream(100, :batches, timeout: :infinity)
104 |> Stream.each(fn objects ->
105 object_ids = Enum.map(objects, & &1.id)
109 |> Enum.map(&transfer_object_hashtags(&1))
110 |> Enum.filter(&(elem(&1, 0) == :error))
111 |> Enum.map(&elem(&1, 1))
113 for failed_id <- failed_ids do
116 "INSERT INTO data_migration_failed_ids(data_migration_id, record_id) " <>
117 "VALUES ($1, $2) ON CONFLICT DO NOTHING;",
118 [data_migration.id, failed_id]
124 "DELETE FROM data_migration_failed_ids WHERE id = ANY($1)",
125 [object_ids -- failed_ids]
128 max_object_id = Enum.at(object_ids, -1)
130 put_stat(:max_processed_id, max_object_id)
131 increment_stat(:processed_count, length(object_ids))
132 increment_stat(:failed_count, length(failed_ids))
134 persist_stats(data_migration)
136 # A quick and dirty approach to controlling the load this background migration imposes
137 sleep_interval = Config.get([:populate_hashtags_table, :sleep_interval_ms], 0)
138 Process.sleep(sleep_interval)
142 with {:ok, %{rows: [[0]]}} <-
144 "SELECT COUNT(record_id) FROM data_migration_failed_ids WHERE data_migration_id = $1;",
147 _ = DataMigration.update_state(data_migration, :complete)
152 _ = DataMigration.update_state(data_migration, :failed)
154 put_stat(:status, :failed)
155 put_stat(:message, "Please check data_migration_failed_ids records.")
161 defp transfer_object_hashtags(object) do
162 hashtags = Object.object_data_hashtags(%{"tag" => object.tag})
164 Repo.transaction(fn ->
165 with {:ok, hashtag_records} <- Hashtag.get_or_create_by_names(hashtags) do
166 for hashtag_record <- hashtag_records do
169 "insert into hashtags_objects(hashtag_id, object_id) values ($1, $2);",
170 [hashtag_record.id, object.id]
176 "ERROR: could not link object #{object.id} and hashtag " <>
177 "#{hashtag_record.id}: #{inspect(e)}"
180 Repo.rollback(object.id)
187 error = "ERROR: could not create hashtags for object #{object.id}: #{inspect(e)}"
189 Repo.rollback(object.id)
194 defp persist_stats(data_migration) do
195 runner_state = Map.drop(state(), [:status])
196 _ = DataMigration.update(data_migration, %{data: runner_state})
199 defp handle_success do
200 put_stat(:status, :complete)
202 unless Config.improved_hashtag_timeline() do
203 Config.put(Config.improved_hashtag_timeline_path(), true)
209 def force_continue do
210 send(whereis(), :migrate_hashtags)
214 {:ok, _} = DataMigration.update(data_migration(), %{state: :pending, data: %{}})