···

       
1
       
 
       
{

     

       
2
       
 
       
  lib,

     

       
3
       
-
       
  stdenv,

     

       
4
       
 
       
  buildPythonPackage,

     

       
5
       
 
       
  fetchFromGitHub,

     

       
6
       
 
       
  rustPlatform,

     

       
7
       
-
       
  pytestCheckHook,

     

       
8
       
-
       
  libiconv,

     

       
9
       
-
       
  numpy,

     

       
10
       
-
       
  protobuf,

     

       
11
       
 
       
  protoc,

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
12
       
 
       
  pyarrow,

     

       
13
       
 
       
  typing-extensions,

     

       
14
       
-
       
  pythonOlder,

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
15
       
 
       
}:

     

       
16
       
 
       


     

       
17
       
-
       
let

     

       
18
       
-
       
  arrow-testing = fetchFromGitHub {

     

       
19
       
-
       
    name = "arrow-testing";

     

       
20
       
-
       
    owner = "apache";

     

       
21
       
-
       
    repo = "arrow-testing";

     

       
22
       
-
       
    rev = "4d209492d514c2d3cb2d392681b9aa00e6d8da1c";

     

       
23
       
-
       
    hash = "sha256-IkiCbuy0bWyClPZ4ZEdkEP7jFYLhM7RCuNLd6Lazd4o=";

     

       
24
       
-
       
  };

     

       
25
       
-
       


     

       
26
       
-
       
  parquet-testing = fetchFromGitHub {

     

       
27
       
-
       
    name = "parquet-testing";

     

       
28
       
-
       
    owner = "apache";

     

       
29
       
-
       
    repo = "parquet-testing";

     

       
30
       
-
       
    rev = "50af3d8ce206990d81014b1862e5ce7380dc3e08";

     

       
31
       
-
       
    hash = "sha256-edyv/r5olkj09aHtm8LHZY0b3jUtLNUcufwI41qKYaY=";

     

       
32
       
-
       
  };

     

       
33
       
-
       
in

     

       
34
       
-
       


     

       
35
       
 
       
buildPythonPackage rec {

     

       
36
       
 
       
  pname = "datafusion";

     

       
37
       
-
       
  version = "40.1.0";

     

       
38
       
 
       
  pyproject = true;

     

       
39
       
 
       


     

       
40
       
 
       
  src = fetchFromGitHub {

     
···

       
42
       
 
       
    owner = "apache";

     

       
43
       
 
       
    repo = "arrow-datafusion-python";

     

       
44
       
 
       
    tag = version;

     

       
45
       
-
       
    hash = "sha256-5WOSlx4XW9zO6oTY16lWQElShLv0ubflVPfSSEGrFgg=";

     

       
0
       
 
       

     

       
0
       
 
       

     

       
46
       
 
       
  };

     

       
47
       
 
       


     

       
48
       
 
       
  cargoDeps = rustPlatform.fetchCargoVendor {

     

       
49
       
-
       
    name = "datafusion-cargo-deps";

     

       
50
       
-
       
    inherit src;

     

       
51
       
-
       
    hash = "sha256-xUpchV4UFEX1HkCpClOwxnEfGLVlOIX4UmzYKiUth9U=";

     

       
52
       
 
       
  };

     

       
53
       
 
       


     

       
54
       
 
       
  nativeBuildInputs = with rustPlatform; [

     
···

       
59
       
 
       


     

       
60
       
 
       
  buildInputs = [

     

       
61
       
 
       
    protobuf

     

       
62
       
-
       
  ]

     

       
63
       
-
       
  ++ lib.optionals stdenv.hostPlatform.isDarwin [

     

       
64
       
-
       
    libiconv

     

       
65
       
 
       
  ];

     

       
66
       
 
       


     

       
67
       
 
       
  dependencies = [

     
···

       
70
       
 
       
  ];

     

       
71
       
 
       


     

       
72
       
 
       
  nativeCheckInputs = [

     

       
0
       
 
       

     

       
0
       
 
       

     

       
73
       
 
       
    pytestCheckHook

     

       
74
       
-
       
    numpy

     

       
75
       
 
       
  ];

     

       
76
       
 
       


     

       
77
       
-
       
  pythonImportsCheck = [ "datafusion" ];

     

       
78
       
-
       


     

       
79
       
-
       
  pytestFlags = [

     

       
80
       
-
       
    "--pyargs"

     

       
81
       
-
       
    pname

     

       
82
       
 
       
  ];

     

       
83
       
 
       


     

       
84
       
 
       
  preCheck = ''

     

       
85
       
-
       
    pushd $TMPDIR

     

       
86
       
-
       
    ln -s ${arrow-testing} ./testing

     

       
87
       
-
       
    ln -s ${parquet-testing} ./parquet

     

       
88
       
 
       
  '';

     

       
89
       
 
       


     

       
90
       
-
       
  postCheck = ''

     

       
91
       
-
       
    popd

     

       
92
       
-
       
  '';

     

       
0
       
 
       

     

       
93
       
 
       


     

       
94
       
-
       
  meta = with lib; {

     

       
95
       
 
       
    description = "Extensible query execution framework";

     

       
96
       
 
       
    longDescription = ''

     

       
97
       
 
       
      DataFusion is an extensible query execution framework, written in Rust,

     
···

       
99
       
 
       
    '';

     

       
100
       
 
       
    homepage = "https://arrow.apache.org/datafusion/";

     

       
101
       
 
       
    changelog = "https://github.com/apache/arrow-datafusion-python/blob/${version}/CHANGELOG.md";

     

       
102
       
-
       
    license = with licenses; [ asl20 ];

     

       
103
       
-
       
    maintainers = with maintainers; [ cpcloud ];

     

       
104
       
 
       
  };

     

       
105
       
 
       
}

     

···

       
1
       
 
       
{

     

       
2
       
 
       
  lib,

     

       
0
       
 
       

     

       
3
       
 
       
  buildPythonPackage,

     

       
4
       
 
       
  fetchFromGitHub,

     

       
5
       
 
       
  rustPlatform,

     

       
6
       
+
       


     

       
7
       
+
       
  # nativeBuildInputs

     

       
0
       
 
       

     

       
0
       
 
       

     

       
8
       
 
       
  protoc,

     

       
9
       
+
       


     

       
10
       
+
       
  # buildInputs

     

       
11
       
+
       
  protobuf,

     

       
12
       
+
       


     

       
13
       
+
       
  # dependencies

     

       
14
       
 
       
  pyarrow,

     

       
15
       
 
       
  typing-extensions,

     

       
16
       
+
       


     

       
17
       
+
       
  # tests

     

       
18
       
+
       
  numpy,

     

       
19
       
+
       
  pytest-asyncio,

     

       
20
       
+
       
  pytestCheckHook,

     

       
21
       
 
       
}:

     

       
22
       
 
       


     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
23
       
 
       
buildPythonPackage rec {

     

       
24
       
 
       
  pname = "datafusion";

     

       
25
       
+
       
  version = "48.0.0";

     

       
26
       
 
       
  pyproject = true;

     

       
27
       
 
       


     

       
28
       
 
       
  src = fetchFromGitHub {

     
···

       
30
       
 
       
    owner = "apache";

     

       
31
       
 
       
    repo = "arrow-datafusion-python";

     

       
32
       
 
       
    tag = version;

     

       
33
       
+
       
    # Fetch arrow-testing and parquet-testing (tests assets)

     

       
34
       
+
       
    fetchSubmodules = true;

     

       
35
       
+
       
    hash = "sha256-9IOkb31f4nFo9mWTr+z5ZG8xSXIZSgW3vCBgLaGxpfI=";

     

       
36
       
 
       
  };

     

       
37
       
 
       


     

       
38
       
 
       
  cargoDeps = rustPlatform.fetchCargoVendor {

     

       
39
       
+
       
    inherit pname src version;

     

       
40
       
+
       
    hash = "sha256-P9NFvhHAGgYIi36CHEPZPr8hmMNp5zrCcmE7NHx51k4=";

     

       
0
       
 
       

     

       
41
       
 
       
  };

     

       
42
       
 
       


     

       
43
       
 
       
  nativeBuildInputs = with rustPlatform; [

     
···

       
48
       
 
       


     

       
49
       
 
       
  buildInputs = [

     

       
50
       
 
       
    protobuf

     

       
0
       
 
       

     

       
0
       
 
       

     

       
0
       
 
       

     

       
51
       
 
       
  ];

     

       
52
       
 
       


     

       
53
       
 
       
  dependencies = [

     
···

       
56
       
 
       
  ];

     

       
57
       
 
       


     

       
58
       
 
       
  nativeCheckInputs = [

     

       
59
       
+
       
    numpy

     

       
60
       
+
       
    pytest-asyncio

     

       
61
       
 
       
    pytestCheckHook

     

       
0
       
 
       

     

       
62
       
 
       
  ];

     

       
63
       
 
       


     

       
64
       
+
       
  pythonImportsCheck = [

     

       
65
       
+
       
    "datafusion"

     

       
66
       
+
       
    "datafusion._internal"

     

       
0
       
 
       

     

       
0
       
 
       

     

       
67
       
 
       
  ];

     

       
68
       
 
       


     

       
69
       
 
       
  preCheck = ''

     

       
70
       
+
       
    rm -rf python/datafusion

     

       
0
       
 
       

     

       
0
       
 
       

     

       
71
       
 
       
  '';

     

       
72
       
 
       


     

       
73
       
+
       
  disabledTests = [

     

       
74
       
+
       
    # Exception: DataFusion error (requires internet access)

     

       
75
       
+
       
    "test_register_http_csv"

     

       
76
       
+
       
  ];

     

       
77
       
 
       


     

       
78
       
+
       
  meta = {

     

       
79
       
 
       
    description = "Extensible query execution framework";

     

       
80
       
 
       
    longDescription = ''

     

       
81
       
 
       
      DataFusion is an extensible query execution framework, written in Rust,

     
···

       
83
       
 
       
    '';

     

       
84
       
 
       
    homepage = "https://arrow.apache.org/datafusion/";

     

       
85
       
 
       
    changelog = "https://github.com/apache/arrow-datafusion-python/blob/${version}/CHANGELOG.md";

     

       
86
       
+
       
    license = with lib.licenses; [ asl20 ];

     

       
87
       
+
       
    maintainers = with lib.maintainers; [ cpcloud ];

     

       
88
       
 
       
  };

     

       
89
       
 
       
}