commit 600d6be8d99062733138e51cfe2930a61c8676ad · pyrox.dev/nixpkgs

+23 -9
pkgs/development/python-modules/trafilatura/default.nix
···

       1
       1
        
       {

     

       2
       2
        
         lib,

     

       3
       3
        
         buildPythonPackage,

     

       4
       4
       +
         fetchFromGitHub,

     

       5
       5
       +
       

     

       6
       6
       +
         # build-system

     

       7
       7
       +
         setuptools,

     

       8
       8
       +
       

     

       9
       9
       +
         # dependencies

     

       4
       10
        
         certifi,

     

       5
       11
        
         charset-normalizer,

     

       6
       12
        
         courlan,

     

       7
       7
       -
         fetchPypi,

     

       8
       13
        
         htmldate,

     

       9
       14
        
         justext,

     

       10
       15
        
         lxml,

     

       11
       11
       -
         pytestCheckHook,

     

       12
       12
       -
         pythonOlder,

     

       13
       13
       -
         setuptools,

     

       14
       16
        
         urllib3,

     

       17
       17
       +
       

     

       18
       18
       +
         # tests

     

       19
       19
       +
         pytestCheckHook,

     

       15
       20
        
       }:

     

       16
       21
        
       

     

       17
       22
        
       buildPythonPackage rec {

     
···

       19
       24
        
         version = "2.0.0";

     

       20
       25
        
         pyproject = true;

     

       21
       26
        
       

     

       22
       22
       -
         disabled = pythonOlder "3.9";

     

       23
       23
       -
       

     

       24
       24
       -
         src = fetchPypi {

     

       25
       25
       -
           inherit pname version;

     

       26
       26
       -
           hash = "sha256-zrcJSm7Ml+cv6nPH26NnFMXFtXe2Rw5FINyok3BtYkc=";

     

       27
       27
       +
         src = fetchFromGitHub {

     

       28
       28
       +
           owner = "adbar";

     

       29
       29
       +
           repo = "trafilatura";

     

       30
       30
       +
           tag = "v${version}";

     

       31
       31
       +
           hash = "sha256-Cf1W3JEGSMkVmRZVTXYsXzZK/Nt/aDG890Sf0/0OZAA=";

     

       27
       32
        
         };

     

       28
       33
        
       

     

       29
       34
        
         postPatch = ''

     
···

       48
       53
        
         nativeCheckInputs = [ pytestCheckHook ];

     

       49
       54
        
       

     

       50
       55
        
         disabledTests = [

     

       56
       56
       +
           # TypeError: argument of type 'NoneType' is not iterable

     

       57
       57
       +
           # https://github.com/adbar/trafilatura/issues/805

     

       58
       58
       +
           "test_external"

     

       59
       59
       +
           "test_extract"

     

       60
       60
       +
       

     

       61
       61
       +
           # AttributeError: 'NoneType' object has no attribute 'find'

     

       62
       62
       +
           # https://github.com/adbar/trafilatura/issues/805

     

       63
       63
       +
           "test_table_processing"

     

       64
       64
       +
       

     

       51
       65
        
           # Disable tests that require an internet connection

     

       52
       66
        
           "test_cli_pipeline"

     

       53
       67
        
           "test_crawl_page"