Git of dcc - anni/blob - lib/mix/tasks/pleroma/search/meilisearch.ex

   1 # Pleroma: A lightweight social networking server
   2 # Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
   3 # SPDX-License-Identifier: AGPL-3.0-only
   4
   5 defmodule Mix.Tasks.Pleroma.Search.Meilisearch do
   6   require Pleroma.Constants
   7
   8   import Mix.Pleroma
   9   import Ecto.Query
  10
  11   import Pleroma.Search.Meilisearch,
  12     only: [meili_post: 2, meili_put: 2, meili_get: 1, meili_delete: 1]
  13
  14   def run(["index"]) do
  15     start_pleroma()
  16     Pleroma.HTML.compile_scrubbers()
  17
  18     meili_version =
  19       (
  20         {:ok, result} = meili_get("/version")
  21
  22         result["pkgVersion"]
  23       )
  24
  25     # The ranking rule syntax was changed but nothing about that is mentioned in the changelog
  26     if not Version.match?(meili_version, ">= 0.25.0") do
  27       raise "Meilisearch <0.24.0 not supported"
  28     end
  29
  30     {:ok, _} =
  31       meili_post(
  32         "/indexes/objects/settings/ranking-rules",
  33         [
  34           "published:desc",
  35           "words",
  36           "exactness",
  37           "proximity",
  38           "typo",
  39           "attribute",
  40           "sort"
  41         ]
  42       )
  43
  44     {:ok, _} =
  45       meili_post(
  46         "/indexes/objects/settings/searchable-attributes",
  47         [
  48           "content"
  49         ]
  50       )
  51
  52     IO.puts("Created indices. Starting to insert posts.")
  53
  54     chunk_size = Pleroma.Config.get([Pleroma.Search.Meilisearch, :initial_indexing_chunk_size])
  55
  56     Pleroma.Repo.transaction(
  57       fn ->
  58         query =
  59           from(Pleroma.Object,
  60             # Only index public and unlisted posts which are notes and have some text
  61             where:
  62               fragment("data->>'type' = 'Note'") and
  63                 (fragment("data->'to' \\? ?", ^Pleroma.Constants.as_public()) or
  64                    fragment("data->'cc' \\? ?", ^Pleroma.Constants.as_public())),
  65             order_by: [desc: fragment("data->'published'")]
  66           )
  67
  68         count = query |> Pleroma.Repo.aggregate(:count, :data)
  69         IO.puts("Entries to index: #{count}")
  70
  71         Pleroma.Repo.stream(
  72           query,
  73           timeout: :infinity
  74         )
  75         |> Stream.map(&Pleroma.Search.Meilisearch.object_to_search_data/1)
  76         |> Stream.filter(fn o -> not is_nil(o) end)
  77         |> Stream.chunk_every(chunk_size)
  78         |> Stream.transform(0, fn objects, acc ->
  79           new_acc = acc + Enum.count(objects)
  80
  81           # Reset to the beginning of the line and rewrite it
  82           IO.write("\r")
  83           IO.write("Indexed #{new_acc} entries")
  84
  85           {[objects], new_acc}
  86         end)
  87         |> Stream.each(fn objects ->
  88           result =
  89             meili_put(
  90               "/indexes/objects/documents",
  91               objects
  92             )
  93
  94           with {:ok, res} <- result do
  95             if not Map.has_key?(res, "uid") do
  96               IO.puts("\nFailed to index: #{inspect(result)}")
  97             end
  98           else
  99             e -> IO.puts("\nFailed to index due to network error: #{inspect(e)}")
 100           end
 101         end)
 102         |> Stream.run()
 103       end,
 104       timeout: :infinity
 105     )
 106
 107     IO.write("\n")
 108   end
 109
 110   def run(["clear"]) do
 111     start_pleroma()
 112
 113     meili_delete("/indexes/objects/documents")
 114   end
 115
 116   def run(["show-keys", master_key]) do
 117     start_pleroma()
 118
 119     endpoint = Pleroma.Config.get([Pleroma.Search.Meilisearch, :url])
 120
 121     {:ok, result} =
 122       Pleroma.HTTP.get(
 123         Path.join(endpoint, "/keys"),
 124         [{"Authorization", "Bearer #{master_key}"}]
 125       )
 126
 127     decoded = Jason.decode!(result.body)
 128
 129     if decoded["results"] do
 130       Enum.each(decoded["results"], fn %{"description" => desc, "key" => key} ->
 131         IO.puts("#{desc}: #{key}")
 132       end)
 133     else
 134       IO.puts("Error fetching the keys, check the master key is correct: #{inspect(decoded)}")
 135     end
 136   end
 137
 138   def run(["stats"]) do
 139     start_pleroma()
 140
 141     {:ok, result} = meili_get("/indexes/objects/stats")
 142     IO.puts("Number of entries: #{result["numberOfDocuments"]}")
 143     IO.puts("Indexing? #{result["isIndexing"]}")
 144   end
 145 end