From 367f0c31c3c15f75aed1d3ba66914e4197c19596 Mon Sep 17 00:00:00 2001 From: Ivan Tashkinov Date: Thu, 31 Dec 2020 09:36:26 +0300 Subject: [PATCH] [#3213] Added query options support for Repo.chunk_stream/4. Used infinite timeout in transfer_hashtags select query. --- lib/mix/tasks/pleroma/database.ex | 11 +++++------ lib/pleroma/repo.ex | 6 +++--- 2 files changed, 8 insertions(+), 9 deletions(-) diff --git a/lib/mix/tasks/pleroma/database.ex b/lib/mix/tasks/pleroma/database.ex index d44bd3478..f903cf75b 100644 --- a/lib/mix/tasks/pleroma/database.ex +++ b/lib/mix/tasks/pleroma/database.ex @@ -149,9 +149,9 @@ def run(["transfer_hashtags"]) do tag: fragment("(?)->>'tag'", object.data) } ) - |> Pleroma.Repo.chunk_stream(100, :batches) + |> Repo.chunk_stream(100, :batches, timeout: :infinity) |> Stream.each(fn objects -> - Logger.info("Processing #{length(objects)} objects...") + Logger.info("Processing #{length(objects)} objects starting from id #{hd(objects).id}...") Enum.map( objects, @@ -165,10 +165,9 @@ def run(["transfer_hashtags"]) do with {:ok, hashtag_records} <- Hashtag.get_or_create_by_names(hashtags) do for hashtag_record <- hashtag_records do with {:ok, _} <- - Ecto.Adapters.SQL.query( - Repo, - "insert into hashtags_objects(hashtag_id, object_id) values " <> - "(#{hashtag_record.id}, #{object.id});" + Repo.query( + "insert into hashtags_objects(hashtag_id, object_id) values ($1, $2);", + [hashtag_record.id, object.id] ) do :noop else diff --git a/lib/pleroma/repo.ex b/lib/pleroma/repo.ex index 4524bd5e2..78711e6ac 100644 --- a/lib/pleroma/repo.ex +++ b/lib/pleroma/repo.ex @@ -63,8 +63,8 @@ def get_assoc(resource, association) do iex> Pleroma.Repo.chunk_stream(Pleroma.Activity.Queries.by_actor(ap_id), 500, :batches) """ @spec chunk_stream(Ecto.Query.t(), integer(), atom()) :: Enumerable.t() - def chunk_stream(query, chunk_size, returns_as \\ :one) do - # We don't actually need start and end funcitons of resource streaming, + def chunk_stream(query, chunk_size, returns_as \\ :one, query_options \\ []) do + # We don't actually need start and end functions of resource streaming, # but it seems to be the only way to not fetch records one-by-one and # have individual records be the elements of the stream, instead of # lists of records @@ -76,7 +76,7 @@ def chunk_stream(query, chunk_size, returns_as \\ :one) do |> order_by(asc: :id) |> where([r], r.id > ^last_id) |> limit(^chunk_size) - |> all() + |> all(query_options) |> case do [] -> {:halt, last_id}