tjh.dev
1{ lib, buildPythonPackage, fetchPypi, isPy27
2, cachetools
3, cytoolz
4, jellyfish
5, matplotlib
6, networkx
7, numpy
8, pyemd
9, pyphen
10, pytest
11, requests
12, scikit-learn
13, scipy
14, spacy
15, srsly
16}:
17
18buildPythonPackage rec {
19 pname = "textacy";
20 version = "0.10.1";
21 disabled = isPy27;
22
23 src = fetchPypi {
24 inherit pname version;
25 sha256 = "ff72adc6dbb85db6981324e226fff77830da57d7fe7e4adb2cafd9dc2a8bfa7d";
26 };
27
28 propagatedBuildInputs = [
29 cachetools
30 cytoolz
31 jellyfish
32 matplotlib
33 networkx
34 numpy
35 pyemd
36 pyphen
37 requests
38 scikit-learn
39 scipy
40 spacy
41 srsly
42 ];
43
44 checkInputs = [ pytest ];
45 # almost all tests have to deal with downloading a dataset, only test pure tests
46 checkPhase = ''
47 pytest tests/test_text_utils.py \
48 tests/test_utils.py \
49 tests/preprocessing \
50 tests/datasets/test_base_dataset.py
51 '';
52
53 meta = with lib; {
54 # scikit-learn in pythonPackages is too new for textacy
55 # remove as soon as textacy support scikit-learn >= 0.24
56 broken = true;
57 description = "Higher-level text processing, built on spaCy";
58 homepage = "https://textacy.readthedocs.io/";
59 license = licenses.asl20;
60 maintainers = with maintainers; [ rvl ];
61 };
62}