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 Agent.start_link(fn -> @init_state end, name: __MODULE__)
16 Agent.get(__MODULE__, & &1)
19 def put(key, value) do
20 Agent.update(__MODULE__, fn state ->
21 Map.put(state, key, value)
25 def increment(key, increment \\ 1) do
26 Agent.update(__MODULE__, fn state ->
27 updated_value = (state[key] || 0) + increment
28 Map.put(state, key, updated_value)
40 alias Pleroma.DataMigration
45 defdelegate state(), to: State, as: :get
46 defdelegate put_state(key, value), to: State, as: :put
47 defdelegate increment_state(key, increment), to: State, as: :increment
49 defdelegate data_migration(), to: DataMigration, as: :populate_hashtags_table
52 GenServer.start_link(__MODULE__, nil, name: __MODULE__)
57 {:ok, nil, {:continue, :init_state}}
61 def handle_continue(:init_state, _state) do
62 {:ok, _} = State.start_link(nil)
64 put_state(:status, :init)
69 Config.get(:env) == :test ->
70 put_state(:status, :noop)
73 put_state(:status, :halt)
74 put_state(:message, "Data migration does not exist.")
76 dm.state == :manual ->
77 put_state(:status, :noop)
78 put_state(:message, "Data migration is in manual execution state.")
80 dm.state == :complete ->
84 send(self(), :migrate_hashtags)
91 def handle_info(:migrate_hashtags, state) do
92 data_migration = data_migration()
94 {:ok, data_migration} = DataMigration.update_state(data_migration, :running)
95 put_state(:status, :running)
97 Logger.info("Starting transferring object embedded hashtags to `hashtags` table...")
99 max_processed_id = data_migration.data["max_processed_id"] || 0
101 # Note: most objects have Mention-type AS2 tags and no hashtags (but we can't filter them out)
104 left_join: hashtag in assoc(object, :hashtags),
105 where: object.id > ^max_processed_id,
106 where: is_nil(hashtag.id),
108 fragment("(?)->'tag' IS NOT NULL AND (?)->'tag' != '[]'::jsonb", object.data, object.data),
111 tag: fragment("(?)->'tag'", object.data)
114 |> Repo.chunk_stream(100, :batches, timeout: :infinity)
115 |> Stream.each(fn objects ->
116 object_ids = Enum.map(objects, & &1.id)
120 |> Enum.map(&transfer_object_hashtags(&1))
121 |> Enum.filter(&(elem(&1, 0) == :error))
122 |> Enum.map(&elem(&1, 1))
124 for failed_id <- failed_ids do
127 "INSERT INTO data_migration_failed_ids(data_migration_id, record_id) " <>
128 "VALUES ($1, $2) ON CONFLICT DO NOTHING;",
129 [data_migration.id, failed_id]
135 "DELETE FROM data_migration_failed_ids WHERE id = ANY($1)",
136 [object_ids -- failed_ids]
139 max_object_id = Enum.at(object_ids, -1)
140 _ = DataMigration.update(data_migration, %{data: %{"max_processed_id" => max_object_id}})
142 increment_state(:processed_count, length(object_ids))
143 increment_state(:failed_count, length(failed_ids))
145 # A quick and dirty approach to controlling the load this background migration imposes
146 sleep_interval = Config.get([:populate_hashtags_table, :sleep_interval_ms], 0)
147 Process.sleep(sleep_interval)
151 with {:ok, %{rows: [[0]]}} <-
153 "SELECT COUNT(record_id) FROM data_migration_failed_ids WHERE data_migration_id = $1;",
156 put_state(:status, :complete)
157 _ = DataMigration.update_state(data_migration, :complete)
162 put_state(:status, :failed)
163 put_state(:message, "Please check data_migration_failed_ids records.")
169 defp transfer_object_hashtags(object) do
170 hashtags = Object.object_data_hashtags(%{"tag" => object.tag})
172 Repo.transaction(fn ->
173 with {:ok, hashtag_records} <- Hashtag.get_or_create_by_names(hashtags) do
174 for hashtag_record <- hashtag_records do
177 "insert into hashtags_objects(hashtag_id, object_id) values ($1, $2);",
178 [hashtag_record.id, object.id]
184 "ERROR: could not link object #{object.id} and hashtag " <>
185 "#{hashtag_record.id}: #{inspect(e)}"
188 Repo.rollback(object.id)
195 error = "ERROR: could not create hashtags for object #{object.id}: #{inspect(e)}"
197 Repo.rollback(object.id)
202 defp handle_success do
203 put_state(:status, :complete)
205 unless Config.improved_hashtag_timeline() do
206 Config.put(Config.improved_hashtag_timeline_path(), true)