Merge remote-tracking branch 'origin/develop' into translate-posts
Signed-off-by: mkljczk <git@mkljczk.pl>
This commit is contained in:
commit
08de5f94e3
118 changed files with 3560 additions and 929 deletions
59
lib/pleroma/language/language_detector.ex
Normal file
59
lib/pleroma/language/language_detector.ex
Normal file
|
|
@ -0,0 +1,59 @@
|
|||
# Pleroma: A lightweight social networking server
|
||||
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
|
||||
# SPDX-License-Identifier: AGPL-3.0-only
|
||||
|
||||
defmodule Pleroma.Language.LanguageDetector do
|
||||
import Pleroma.EctoType.ActivityPub.ObjectValidators.LanguageCode,
|
||||
only: [good_locale_code?: 1]
|
||||
|
||||
@words_threshold 4
|
||||
@config_impl Application.compile_env(:pleroma, [__MODULE__, :config_impl], Pleroma.Config)
|
||||
|
||||
def configured? do
|
||||
provider = get_provider()
|
||||
|
||||
!!provider and provider.configured?
|
||||
end
|
||||
|
||||
def missing_dependencies do
|
||||
provider = get_provider()
|
||||
|
||||
if provider do
|
||||
provider.missing_dependencies()
|
||||
else
|
||||
[]
|
||||
end
|
||||
end
|
||||
|
||||
# Strip tags from text, etc.
|
||||
defp prepare_text(text) do
|
||||
text
|
||||
|> Floki.parse_fragment!()
|
||||
|> Floki.filter_out(
|
||||
".h-card, .mention, .hashtag, .u-url, .quote-inline, .recipients-inline, code, pre"
|
||||
)
|
||||
|> Floki.text()
|
||||
end
|
||||
|
||||
def detect(text) do
|
||||
provider = get_provider()
|
||||
|
||||
text = prepare_text(text)
|
||||
word_count = text |> String.split(~r/\s+/) |> Enum.count()
|
||||
|
||||
if word_count < @words_threshold or !provider or !provider.configured? do
|
||||
nil
|
||||
else
|
||||
with language <- provider.detect(text),
|
||||
true <- good_locale_code?(language) do
|
||||
language
|
||||
else
|
||||
_ -> nil
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
defp get_provider do
|
||||
@config_impl.get([__MODULE__, :provider])
|
||||
end
|
||||
end
|
||||
47
lib/pleroma/language/language_detector/fasttext.ex
Normal file
47
lib/pleroma/language/language_detector/fasttext.ex
Normal file
|
|
@ -0,0 +1,47 @@
|
|||
# Pleroma: A lightweight social networking server
|
||||
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
|
||||
# SPDX-License-Identifier: AGPL-3.0-only
|
||||
|
||||
defmodule Pleroma.Language.LanguageDetector.Fasttext do
|
||||
import Pleroma.Web.Utils.Guards, only: [not_empty_string: 1]
|
||||
|
||||
alias Pleroma.Language.LanguageDetector.Provider
|
||||
|
||||
@behaviour Provider
|
||||
|
||||
@impl Provider
|
||||
def missing_dependencies do
|
||||
if Pleroma.Utils.command_available?("fasttext") do
|
||||
[]
|
||||
else
|
||||
["fasttext"]
|
||||
end
|
||||
end
|
||||
|
||||
@impl Provider
|
||||
def configured?, do: not_empty_string(get_model())
|
||||
|
||||
@impl Provider
|
||||
def detect(text) do
|
||||
text_path = Path.join(System.tmp_dir!(), "fasttext-#{Ecto.UUID.generate()}")
|
||||
|
||||
File.write(text_path, text |> String.replace(~r/\s+/, " "))
|
||||
|
||||
detected_language =
|
||||
case System.cmd("fasttext", ["predict", get_model(), text_path]) do
|
||||
{"__label__" <> language, _} ->
|
||||
language |> String.trim()
|
||||
|
||||
_ ->
|
||||
nil
|
||||
end
|
||||
|
||||
File.rm(text_path)
|
||||
|
||||
detected_language
|
||||
end
|
||||
|
||||
defp get_model do
|
||||
Pleroma.Config.get([__MODULE__, :model])
|
||||
end
|
||||
end
|
||||
11
lib/pleroma/language/language_detector/provider.ex
Normal file
11
lib/pleroma/language/language_detector/provider.ex
Normal file
|
|
@ -0,0 +1,11 @@
|
|||
# Pleroma: A lightweight social networking server
|
||||
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
|
||||
# SPDX-License-Identifier: AGPL-3.0-only
|
||||
|
||||
defmodule Pleroma.Language.LanguageDetector.Provider do
|
||||
@callback missing_dependencies() :: [String.t()]
|
||||
|
||||
@callback configured?() :: boolean()
|
||||
|
||||
@callback detect(text :: String.t()) :: String.t() | nil
|
||||
end
|
||||
Loading…
Add table
Add a link
Reference in a new issue