python3Packages.nltk: add data(Dir) passthru, run tests (#409680)

pkgs/by-name/un/unstructured-api/package.nix

@@ -5,8 +5,6 @@
   python3,
   makeWrapper,
   nix-update-script,
-  symlinkJoin,
-  nltk-data,
 }:
 let
   pythonEnv = python3.withPackages (
@@ -147,14 +145,10 @@ let
     ++ unstructured.optional-dependencies.all-docs
   );
   version = "0.0.82";
-  unstructured_api_nltk_data = symlinkJoin {
-    name = "unstructured_api_nltk_data";
-
-    paths = [
-      nltk-data.punkt
-      nltk-data.averaged-perceptron-tagger
-    ];
-  };
+  unstructured_api_nltk_data = python3.pkgs.nltk.dataDir (d: [
+    d.punkt
+    d.averaged-perceptron-tagger
+  ]);
 in
 stdenvNoCC.mkDerivation {
   pname = "unstructured-api";

pkgs/development/python-modules/aider-chat/default.nix

@@ -6,8 +6,6 @@
   gitMinimal,
   portaudio,
   playwright-driver,
-  symlinkJoin,
-  nltk-data,
   pythonOlder,
   pythonAtLeast,
   setuptools-scm,
@@ -122,13 +120,10 @@
 }:
 
 let
-  aider-nltk-data = symlinkJoin {
-    name = "aider-nltk-data";
-    paths = [
-      nltk-data.punkt-tab
-      nltk-data.stopwords
-    ];
-  };
+  aider-nltk-data = nltk.dataDir (d: [
+    d.punkt-tab
+    d.stopwords
+  ]);
 
   version = "0.83.1";
   aider-chat = buildPythonPackage {

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

@@ -1,5 +1,6 @@
 {
   lib,
+  pkgs,
   fetchPypi,
   buildPythonPackage,
   pythonOlder,
@@ -7,6 +8,16 @@
   joblib,
   regex,
   tqdm,
+
+  # preInstallCheck
+  nltk,
+
+  # nativeCheckInputs
+  matplotlib,
+  numpy,
+  pyparsing,
+  pytestCheckHook,
+  pytest-mock,
 }:
 
 buildPythonPackage rec {
@@ -21,28 +32,78 @@ buildPythonPackage rec {
     hash = "sha256-h9EnvT3kvYmk+BJl5fpZyxsZmydEAXU3D3QX0rx66Gg=";
   };
 
-  propagatedBuildInputs = [
+  dependencies = [
     click
     joblib
     regex
     tqdm
   ];
 
-  # Tests require some data, the downloading of which is impure. It would
-  # probably make sense to make the data another derivation, but then feeding
-  # that into the tests (given that we need nltk itself to download the data,
-  # unless there's an easy way to download it without nltk's downloader) might
-  # be complicated. For now let's just disable the tests and hope for the
-  # best.
-  doCheck = false;
+  # Use new passthru function to pass dependencies required for testing
+  preInstallCheck = ''
+    export NLTK_DATA=${
+      nltk.dataDir (
+        d: with d; [
+          averaged-perceptron-tagger-eng
+          averaged-perceptron-tagger-rus
+          brown
+          cess-cat
+          cess-esp
+          conll2007
+          floresta
+          gutenberg
+          inaugural
+          indian
+          large-grammars
+          nombank-1-0
+          omw-1-4
+          pl196x
+          porter-test
+          ptb
+          punkt-tab
+          rte
+          sinica-treebank
+          stopwords
+          tagsets-json
+          treebank
+          twitter-samples
+          udhr
+          universal-tagset
+          wmt15-eval
+          wordnet
+          wordnet-ic
+          words
+        ]
+      )
+    }
+  '';
+
+  nativeCheckInputs = [
+    pytestCheckHook
+    matplotlib
+    numpy
+    pyparsing
+    pytest-mock
+
+    pkgs.which
+  ];
+
+  disabledTestPaths = [
+    "nltk/test/unit/test_downloader.py" # Touches network
+  ];
 
   pythonImportsCheck = [ "nltk" ];
 
+  passthru = {
+    data = pkgs.nltk-data;
+    dataDir = pkgs.callPackage ./data-dir.nix { };
+  };
+
   meta = with lib; {
     description = "Natural Language Processing ToolKit";
     mainProgram = "nltk";
     homepage = "http://nltk.org/";
     license = licenses.asl20;
-    maintainers = [ ];
+    maintainers = [ lib.maintainers.bengsparks ];
   };
 }

pkgs/development/python-modules/type-infer/default.nix

@@ -16,18 +16,13 @@
   python-dateutil,
   scipy,
   toml,
-  nltk-data,
-  symlinkJoin,
 }:
 let
-  testNltkData = symlinkJoin {
-    name = "nltk-test-data";
-    paths = [
-      nltk-data.punkt
-      nltk-data.punkt-tab
-      nltk-data.stopwords
-    ];
-  };
+  testNltkData = nltk.dataDir (d: [
+    d.punkt
+    d.punkt-tab
+    d.stopwords
+  ]);
 
   version = "0.0.21";
   tag = "v${version}";