commit 4a2c6cc55d3f21dfe98ed50b171dcc3fadbec294 · pyrox.dev/nixpkgs

+4 -10

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

···

       5
       5
        
         python3,

     

       6
       6
        
         makeWrapper,

     

       7
       7
        
         nix-update-script,

     

       8
       8
       -
         symlinkJoin,

     

       9
       9
       -
         nltk-data,

     

       10
       8
        
       }:

     

       11
       9
        
       let

     

       12
       10
        
         pythonEnv = python3.withPackages (

     
···

       147
       145
        
           ++ unstructured.optional-dependencies.all-docs

     

       148
       146
        
         );

     

       149
       147
        
         version = "0.0.82";

     

       150
       150
       -
         unstructured_api_nltk_data = symlinkJoin {

     

       151
       151
       -
           name = "unstructured_api_nltk_data";

     

       152
       152
       -
       

     

       153
       153
       -
           paths = [

     

       154
       154
       -
             nltk-data.punkt

     

       155
       155
       -
             nltk-data.averaged-perceptron-tagger

     

       156
       156
       -
           ];

     

       157
       157
       -
         };

     

       148
       148
       +
         unstructured_api_nltk_data = python3.pkgs.nltk.dataDir (d: [

     

       149
       149
       +
           d.punkt

     

       150
       150
       +
           d.averaged-perceptron-tagger

     

       151
       151
       +
         ]);

     

       158
       152
        
       in

     

       159
       153
        
       stdenvNoCC.mkDerivation {

     

       160
       154
        
         pname = "unstructured-api";

+4 -9

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

···

       6
       6
        
         gitMinimal,

     

       7
       7
        
         portaudio,

     

       8
       8
        
         playwright-driver,

     

       9
       9
       -
         symlinkJoin,

     

       10
       10
       -
         nltk-data,

     

       11
       9
        
         pythonOlder,

     

       12
       10
        
         pythonAtLeast,

     

       13
       11
        
         setuptools-scm,

     
···

       122
       120
        
       }:

     

       123
       121
        
       

     

       124
       122
        
       let

     

       125
       125
       -
         aider-nltk-data = symlinkJoin {

     

       126
       126
       -
           name = "aider-nltk-data";

     

       127
       127
       -
           paths = [

     

       128
       128
       -
             nltk-data.punkt-tab

     

       129
       129
       -
             nltk-data.stopwords

     

       130
       130
       -
           ];

     

       131
       131
       -
         };

     

       123
       123
       +
         aider-nltk-data = nltk.dataDir (d: [

     

       124
       124
       +
           d.punkt-tab

     

       125
       125
       +
           d.stopwords

     

       126
       126
       +
         ]);

     

       132
       127
        
       

     

       133
       128
        
         version = "0.83.1";

     

       134
       129
        
         aider-chat = buildPythonPackage {

+15

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

···

       1
       1
       +
       {

     

       2
       2
       +
         lib,

     

       3
       3
       +
         pkgs,

     

       4
       4
       +
         python3Packages,

     

       5
       5
       +
       }:

     

       6
       6
       +
       lib.makeOverridable (

     

       7
       7
       +
         { ... }@nltkDataPkgs:

     

       8
       8
       +
         f:

     

       9
       9
       +
         pkgs.symlinkJoin {

     

       10
       10
       +
           inherit (python3Packages.nltk) meta;

     

       11
       11
       +
           name = "nltk-data-dir";

     

       12
       12
       +
       

     

       13
       13
       +
           paths = f nltkDataPkgs;

     

       14
       14
       +
         }

     

       15
       15
       +
       ) python3Packages.nltk.data

+70 -9

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

···

       1
       1
        
       {

     

       2
       2
        
         lib,

     

       3
       3
       +
         pkgs,

     

       3
       4
        
         fetchPypi,

     

       4
       5
        
         buildPythonPackage,

     

       5
       6
        
         pythonOlder,

     
···

       7
       8
        
         joblib,

     

       8
       9
        
         regex,

     

       9
       10
        
         tqdm,

     

       11
       11
       +
       

     

       12
       12
       +
         # preInstallCheck

     

       13
       13
       +
         nltk,

     

       14
       14
       +
       

     

       15
       15
       +
         # nativeCheckInputs

     

       16
       16
       +
         matplotlib,

     

       17
       17
       +
         numpy,

     

       18
       18
       +
         pyparsing,

     

       19
       19
       +
         pytestCheckHook,

     

       20
       20
       +
         pytest-mock,

     

       10
       21
        
       }:

     

       11
       22
        
       

     

       12
       23
        
       buildPythonPackage rec {

     
···

       21
       32
        
           hash = "sha256-h9EnvT3kvYmk+BJl5fpZyxsZmydEAXU3D3QX0rx66Gg=";

     

       22
       33
        
         };

     

       23
       34
        
       

     

       24
       24
       -
         propagatedBuildInputs = [

     

       35
       35
       +
         dependencies = [

     

       25
       36
        
           click

     

       26
       37
        
           joblib

     

       27
       38
        
           regex

     

       28
       39
        
           tqdm

     

       29
       40
        
         ];

     

       30
       41
        
       

     

       31
       31
       -
         # Tests require some data, the downloading of which is impure. It would

     

       32
       32
       -
         # probably make sense to make the data another derivation, but then feeding

     

       33
       33
       -
         # that into the tests (given that we need nltk itself to download the data,

     

       34
       34
       -
         # unless there's an easy way to download it without nltk's downloader) might

     

       35
       35
       -
         # be complicated. For now let's just disable the tests and hope for the

     

       36
       36
       -
         # best.

     

       37
       37
       -
         doCheck = false;

     

       42
       42
       +
         # Use new passthru function to pass dependencies required for testing

     

       43
       43
       +
         preInstallCheck = ''

     

       44
       44
       +
           export NLTK_DATA=${

     

       45
       45
       +
             nltk.dataDir (

     

       46
       46
       +
               d: with d; [

     

       47
       47
       +
                 averaged-perceptron-tagger-eng

     

       48
       48
       +
                 averaged-perceptron-tagger-rus

     

       49
       49
       +
                 brown

     

       50
       50
       +
                 cess-cat

     

       51
       51
       +
                 cess-esp

     

       52
       52
       +
                 conll2007

     

       53
       53
       +
                 floresta

     

       54
       54
       +
                 gutenberg

     

       55
       55
       +
                 inaugural

     

       56
       56
       +
                 indian

     

       57
       57
       +
                 large-grammars

     

       58
       58
       +
                 nombank-1-0

     

       59
       59
       +
                 omw-1-4

     

       60
       60
       +
                 pl196x

     

       61
       61
       +
                 porter-test

     

       62
       62
       +
                 ptb

     

       63
       63
       +
                 punkt-tab

     

       64
       64
       +
                 rte

     

       65
       65
       +
                 sinica-treebank

     

       66
       66
       +
                 stopwords

     

       67
       67
       +
                 tagsets-json

     

       68
       68
       +
                 treebank

     

       69
       69
       +
                 twitter-samples

     

       70
       70
       +
                 udhr

     

       71
       71
       +
                 universal-tagset

     

       72
       72
       +
                 wmt15-eval

     

       73
       73
       +
                 wordnet

     

       74
       74
       +
                 wordnet-ic

     

       75
       75
       +
                 words

     

       76
       76
       +
               ]

     

       77
       77
       +
             )

     

       78
       78
       +
           }

     

       79
       79
       +
         '';

     

       80
       80
       +
       

     

       81
       81
       +
         nativeCheckInputs = [

     

       82
       82
       +
           pytestCheckHook

     

       83
       83
       +
           matplotlib

     

       84
       84
       +
           numpy

     

       85
       85
       +
           pyparsing

     

       86
       86
       +
           pytest-mock

     

       87
       87
       +
       

     

       88
       88
       +
           pkgs.which

     

       89
       89
       +
         ];

     

       90
       90
       +
       

     

       91
       91
       +
         disabledTestPaths = [

     

       92
       92
       +
           "nltk/test/unit/test_downloader.py" # Touches network

     

       93
       93
       +
         ];

     

       38
       94
        
       

     

       39
       95
        
         pythonImportsCheck = [ "nltk" ];

     

       40
       96
        
       

     

       97
       97
       +
         passthru = {

     

       98
       98
       +
           data = pkgs.nltk-data;

     

       99
       99
       +
           dataDir = pkgs.callPackage ./data-dir.nix { };

     

       100
       100
       +
         };

     

       101
       101
       +
       

     

       41
       102
        
         meta = with lib; {

     

       42
       103
        
           description = "Natural Language Processing ToolKit";

     

       43
       104
        
           mainProgram = "nltk";

     

       44
       105
        
           homepage = "http://nltk.org/";

     

       45
       106
        
           license = licenses.asl20;

     

       46
       46
       -
           maintainers = [ ];

     

       107
       107
       +
           maintainers = [ lib.maintainers.bengsparks ];

     

       47
       108
        
         };

     

       48
       109
        
       }

+5 -10

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

···

       16
       16
        
         python-dateutil,

     

       17
       17
        
         scipy,

     

       18
       18
        
         toml,

     

       19
       19
       -
         nltk-data,

     

       20
       20
       -
         symlinkJoin,

     

       21
       19
        
       }:

     

       22
       20
        
       let

     

       23
       23
       -
         testNltkData = symlinkJoin {

     

       24
       24
       -
           name = "nltk-test-data";

     

       25
       25
       -
           paths = [

     

       26
       26
       -
             nltk-data.punkt

     

       27
       27
       -
             nltk-data.punkt-tab

     

       28
       28
       -
             nltk-data.stopwords

     

       29
       29
       -
           ];

     

       30
       30
       -
         };

     

       21
       21
       +
         testNltkData = nltk.dataDir (d: [

     

       22
       22
       +
           d.punkt

     

       23
       23
       +
           d.punkt-tab

     

       24
       24
       +
           d.stopwords

     

       25
       25
       +
         ]);

     

       31
       26
        
       

     

       32
       27
        
         version = "0.0.21";

     

       33
       28
        
         tag = "v${version}";