python3Packages.nltk.dataDir: init

Add helper function to install NLTK data packages like `wordnet` and `punkt-tab`.
Should be used in conjunction with `makeWrapperArgs`,
e.g. `makeWrapperArgs = [ "--set" "NLTK_DATA" "${nltk.dataDir(d: [ d.wordnet d.punkt-tab ])}" ];`

pkgs/development/python-modules/nltk/data-dir.nix

@@ -0,0 +1,15 @@
+{
+  lib,
+  pkgs,
+  python3Packages,
+}:
+lib.makeOverridable (
+  { ... }@nltkDataPkgs:
+  f:
+  pkgs.symlinkJoin {
+    inherit (python3Packages.nltk) meta;
+    name = "nltk-data-dir";
+
+    paths = f nltkDataPkgs;
+  }
+) python3Packages.nltk.data

pkgs/development/python-modules/nltk/default.nix

@@ -41,6 +41,7 @@ buildPythonPackage rec {
 
   passthru = {
     data = pkgs.nltk-data;
+    dataDir = pkgs.callPackage ./data-dir.nix { };
   };
 
   meta = with lib; {