pkgs/tools/text/nltk-data/default.nix

   1 { lib, newScope, fetchFromGitHub, unzip, stdenvNoCC }:
   2 let
   3   base = {
   4     version = "0-unstable-2024-07-29";
   5     nativeBuildInputs = [ unzip ];
   6     dontBuild = true;
   7     meta = with lib; {
   8       description = "NLTK Data";
   9       homepage = "https://github.com/nltk/nltk_data";
  10       license = licenses.asl20;
  11       platforms = platforms.all;
  12       maintainers = with maintainers; [ happysalada ];
  13     };
  14   };
  15   makeNltkDataPackage = {pname, location, hash}:
  16     let
  17       src = fetchFromGitHub {
  18         owner = "nltk";
  19         repo = "nltk_data";
  20         rev = "cfe82914f3c2d24363687f1db3b05e8b9f687e2b";
  21         inherit hash;
  22         sparseCheckout = [ "packages/${location}/${pname}.zip" ];
  23       };
  24     in
  25     stdenvNoCC.mkDerivation (base // {
  26       inherit pname src;
  27       inherit (base) version;
  28       installPhase = ''
  29         runHook preInstall
  30
  31         mkdir -p $out
  32         unzip ${src}/packages/${location}/${pname}.zip
  33         mkdir -p $out/${location}
  34         cp -R ${pname}/ $out/${location}
  35
  36         runHook postInstall
  37       '';
  38     });
  39 in
  40 lib.makeScope newScope (self: {
  41   punkt = makeNltkDataPackage {
  42     pname = "punkt";
  43     location = "tokenizers";
  44     hash = "sha256-OzMkruoYbFKqzuimOXIpE5lhHz8tmSqOFoLT+fjdTVg=";
  45   };
  46   punkt_tab = makeNltkDataPackage {
  47     pname = "punkt_tab";
  48     location = "tokenizers";
  49     hash = "sha256-OzMkruoYbFKqzuimOXIpE5lhHz8tmSqOFoLT+fjdTVg=";
  50   };
  51   averaged_perceptron_tagger = makeNltkDataPackage {
  52     pname = "averaged_perceptron_tagger";
  53     location = "taggers";
  54     hash = "sha256-tl3Cn2okhBkUtTXvAmFRx72Brez6iTGRdmFTwFmpk3M=";
  55   };
  56   snowball_data = makeNltkDataPackage {
  57     pname = "snowball_data";
  58     location = "stemmers";
  59     hash = "sha256-mNefwOPVJGz9kXV3LV4DuV7FJpNir/Nwg4ujd0CogEk=";
  60   };
  61   stopwords = makeNltkDataPackage {
  62     pname = "stopwords";
  63     location = "corpora";
  64     hash = "sha256-8lMjW5YI8h6dHJ/83HVY2OYGDyKPpgkUAKPISiAKqqk=";
  65   };
  66   wordnet = makeNltkDataPackage {
  67     pname = "wordnet";
  68     location = "corpora";
  69     hash = "sha256-8lMjW5YI8h6dHJ/83HVY2OYGDyKPpgkUAKPISiAKqqk=";
  70   };
  71 })