From 2f690a9306791669e66e97c968ac8b92b5822481 Mon Sep 17 00:00:00 2001
From: Pavel Sobolev <paveloom@riseup.net>
Date: Tue, 26 Sep 2023 12:47:09 +0300
Subject: [PATCH] python3Packages.sentence-splitter: init at 1.4

---
 .../sentence-splitter/default.nix             | 41 +++++++++++++++++++
 pkgs/top-level/python-packages.nix            |  2 +
 2 files changed, 43 insertions(+)
 create mode 100644 pkgs/development/python-modules/sentence-splitter/default.nix

diff --git a/pkgs/development/python-modules/sentence-splitter/default.nix b/pkgs/development/python-modules/sentence-splitter/default.nix
new file mode 100644
index 000000000000..a544a006ce80
--- /dev/null
+++ b/pkgs/development/python-modules/sentence-splitter/default.nix
@@ -0,0 +1,41 @@
+{ lib
+, buildPythonPackage
+, pythonOlder
+, fetchFromGitHub
+
+, pytestCheckHook
+, regex
+}:
+
+buildPythonPackage rec {
+  pname = "sentence-splitter";
+  version = "1.4";
+
+  disabled = pythonOlder "3.5";
+
+  src = fetchFromGitHub {
+    owner = "mediacloud";
+    repo = "sentence-splitter";
+    rev = version;
+    hash = "sha256-FxRi8fhKB9++lCTFpCAug0fxjkSVTKChLY84vkshR34=";
+  };
+
+  propagatedBuildInputs = [
+    regex
+  ];
+
+  nativeCheckInputs = [
+    pytestCheckHook
+  ];
+
+  pythonImportsCheck = [
+    "sentence_splitter"
+  ];
+
+  meta = with lib; {
+    description = "Text to sentence splitter using heuristic algorithm by Philipp Koehn and Josh Schroeder";
+    homepage = "https://github.com/mediacloud/sentence-splitter";
+    license = licenses.lgpl3Plus;
+    maintainers = with maintainers; [ paveloom ];
+  };
+}
diff --git a/pkgs/top-level/python-packages.nix b/pkgs/top-level/python-packages.nix
index 94fb3ef5d862..a057296f0eb8 100644
--- a/pkgs/top-level/python-packages.nix
+++ b/pkgs/top-level/python-packages.nix
@@ -11793,6 +11793,8 @@ self: super: with self; {
     inherit (pkgs) sentencepiece;
   };
 
+  sentence-splitter = callPackage ../development/python-modules/sentence-splitter { };
+
   sentence-transformers = callPackage ../development/python-modules/sentence-transformers { };
 
   sentinel = callPackage ../development/python-modules/sentinel { };