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,
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
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
|> order_by(asc: :id)
|> where([r], r.id > ^last_id)
|> limit(^chunk_size)
- |> all()
+ |> all(query_options)
|> case do
[] ->
{:halt, last_id}