Adjust content indexing to skip more unneeded stuff

This commit is contained in:
Ekaterina Vaartis 2021-08-22 19:38:03 +03:00 committed by FloatingGhost
parent 14ef6ce80f
commit 117f525fd6

View file

@ -52,13 +52,6 @@ def run(["index"]) do
timeout: :infinity timeout: :infinity
) )
|> Stream.chunk_every(chunk_size) |> 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 -> |> Stream.map(fn objects ->
Enum.map(objects, fn object -> Enum.map(objects, fn object ->
data = object.data data = object.data
@ -70,15 +63,34 @@ def run(["index"]) do
end end
{:ok, published, _} = DateTime.from_iso8601(data["published"]) {:ok, published, _} = DateTime.from_iso8601(data["published"])
{:ok, content} = FastSanitize.strip_tags(content_str)
%{ content =
id: object.id, with {:ok, scrubbed} <- FastSanitize.strip_tags(content_str),
content: content, trimmed <- String.trim(scrubbed) do
ap: data["id"], trimmed
published: published |> DateTime.to_unix() end
}
# Only index if there is anything in the string. If there is a single symbol,
# it's probably a dot from mastodon posts with just the picture
if String.length(content) > 1 do
%{
id: object.id,
content: content,
ap: data["id"],
published: published |> DateTime.to_unix()
}
else
nil
end
end) end)
|> Enum.filter(fn o -> not is_nil(o) end)
end)
|> Stream.transform(0, fn objects, acc ->
new_acc = acc + Enum.count(objects)
IO.puts("Indexed #{new_acc} entries")
{[objects], new_acc}
end) end)
|> Stream.each(fn objects -> |> Stream.each(fn objects ->
{:ok, result} = {:ok, result} =
@ -102,6 +114,9 @@ def run(["clear"]) do
endpoint = Pleroma.Config.get([Pleroma.Search.Meilisearch, :url]) endpoint = Pleroma.Config.get([Pleroma.Search.Meilisearch, :url])
{:ok, _} = Pleroma.HTTP.request(:delete, "#{endpoint}/indexes/objects", "", [], []) {:ok, _} =
Pleroma.HTTP.request(:delete, "#{endpoint}/indexes/objects/documents", "", [],
timeout: :infinity
)
end end
end end