Make the indexing batch differently and more, show number indexed

This commit is contained in:
Ekaterina Vaartis 2021-08-16 22:30:56 +03:00 committed by FloatingGhost
parent e5ac2ffa07
commit abf82a63ec

View file

@ -28,33 +28,46 @@ def run(["index"]) do
])
)
Pleroma.Repo.chunk_stream(
from(Pleroma.Object,
# Only index public posts which are notes and have some text
where:
fragment("data->>'type' = 'Note'") and
fragment("LENGTH(data->>'source') > 0") and
fragment("data->'to' \\? ?", ^Pleroma.Constants.as_public())
),
200,
:batches
)
|> Stream.map(fn objects ->
Enum.map(objects, fn object ->
data = object.data
%{id: object.id, source: data["source"], ap: data["id"]}
end)
end)
|> Stream.each(fn objects ->
{:ok, _} =
Pleroma.HTTP.post(
"#{endpoint}/indexes/objects/documents",
Jason.encode!(objects)
)
chunk_size = 100_000
IO.puts("Indexed #{Enum.count(objects)} entries")
end)
|> Stream.run()
Pleroma.Repo.transaction(
fn ->
Pleroma.Repo.stream(
from(Pleroma.Object,
# Only index public posts which are notes and have some text
where:
fragment("data->>'type' = 'Note'") and
fragment("LENGTH(data->>'source') > 0") and
fragment("data->'to' \\? ?", ^Pleroma.Constants.as_public()),
order_by: fragment("data->'published' DESC")
),
timeout: :infinity
)
|> Stream.chunk_every(chunk_size)
|> Stream.transform(0, fn objects, acc ->
new_acc = acc + Enum.count(objects)
IO.puts("Indexed #{new_acc} entries")
{[objects], new_acc}
end)
|> Stream.map(fn objects ->
Enum.map(objects, fn object ->
data = object.data
%{id: object.id, source: data["source"], ap: data["id"]}
end)
end)
|> Stream.each(fn objects ->
{:ok, _} =
Pleroma.HTTP.post(
"#{endpoint}/indexes/objects/documents",
Jason.encode!(objects)
)
end)
|> Stream.run()
end,
timeout: :infinity
)
end
def run(["clear"]) do