pyrox.dev
1{ lib
2, buildPythonPackage
3, fetchPypi
4, pythonOlder
5, poetry-core
6, numpy
7, pandas
8, pydateinfer
9, python-dateutil
10, scipy
11, type-infer
12, dataclasses-json
13, colorlog
14, pydantic
15, nltk-data
16, symlinkJoin
17}:
18let
19 testNltkData = symlinkJoin {
20 name = "nltk-test-data";
21 paths = [ nltk-data.punkt nltk-data.stopwords ];
22 };
23in
24buildPythonPackage rec {
25 pname = "dataprep-ml";
26 version = "0.0.18";
27 pyproject = true;
28
29 disabled = pythonOlder "3.8";
30
31 # using PyPI as github repo does not contain tags or release branches
32 src = fetchPypi {
33 pname = "dataprep_ml";
34 inherit version;
35 hash = "sha256-nIqyRwv62j8x5Fy7ILMLWxw6yJmkkNRE1zyUlfvRYTI=";
36 };
37
38 nativeBuildInputs = [
39 poetry-core
40 ];
41
42 propagatedBuildInputs = [
43 numpy
44 pandas
45 pydateinfer
46 python-dateutil
47 scipy
48 type-infer
49 dataclasses-json
50 colorlog
51 pydantic
52 ];
53
54 # PyPI tarball has no tests
55 doCheck = false;
56
57 # Package import requires NLTK data to be downloaded
58 # It is the only way to set NLTK_DATA environment variable,
59 # so that it is available in pythonImportsCheck
60 env.NLTK_DATA = testNltkData;
61 pythonImportsCheck = [
62 "dataprep_ml"
63 "dataprep_ml.cleaners"
64 "dataprep_ml.helpers"
65 "dataprep_ml.imputers"
66 "dataprep_ml.insights"
67 "dataprep_ml.recommenders"
68 "dataprep_ml.splitters"
69 ];
70
71 meta = with lib; {
72 description = "Data utilities for Machine Learning pipelines";
73 homepage = "https://github.com/mindsdb/dataprep_ml";
74 license = licenses.gpl3Only;
75 maintainers = with maintainers; [ mbalatsko ];
76 };
77}