Accepting request 1045543 from devel:languages:python

- Complete nltk_data.tar.xz for offline testing
- Fix failing tests (gh#nltk/nltk#2969) by adding patches:
  - port-2to3.patch
  - skip-networked-test.patch
- Clean up the SPEC to get rid of rpmlint warnings.

OBS-URL: https://build.opensuse.org/request/show/1045543
OBS-URL: https://build.opensuse.org/package/show/openSUSE:Factory/python-nltk?expand=0&rev=12

nltk_data.tar.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f41383a4774bf7227f4563f46543460ba07a6921f7bcc6185519e87ea9e4323f
+size 453871052

port-2to3.patch

@@ -0,0 +1,48 @@
+---
+ nltk_data/corpora/pl196x/splitter.py                 |    4 ++--
+ nltk_data/taggers/universal_tagset/universal_tags.py |    5 -----
+ tools/find_deprecated.py                             |    2 +-
+ 3 files changed, 3 insertions(+), 8 deletions(-)
+
+--- a/nltk_data/corpora/pl196x/splitter.py
++++ b/nltk_data/corpora/pl196x/splitter.py
+@@ -1,4 +1,4 @@
+-#!/usr/bin/python
++#!/usr/bin/python3
+ 
+ import sys, re
+ 
+@@ -7,7 +7,7 @@ TEXTID = re.compile(r'<text id="(.*)">')
+ 
+ if __name__ == '__main__':
+     if len(sys.argv) != 2:
+-        print 'One argument required: a pl196x corpus to split.'
++        print('One argument required: a pl196x corpus to split.')
+         sys.exit()
+ 
+     inputFileName = sys.argv[1]
+--- a/nltk_data/taggers/universal_tagset/universal_tags.py
++++ b/nltk_data/taggers/universal_tagset/universal_tags.py
+@@ -22,11 +22,6 @@ X - other: foreign words, typos, abbrevi
+ @author: Nathan Schneider (nschneid)
+ @since: 2011-05-06
+ '''
+-
+-# Strive towards Python 3 compatibility
+-from __future__ import print_function, unicode_literals, division
+-from future_builtins import map, filter
+-
+ import re, glob
+ from collections import defaultdict
+ 
+--- a/tools/find_deprecated.py
++++ b/tools/find_deprecated.py
+@@ -29,7 +29,7 @@ import textwrap
+ import tokenize
+ from doctest import DocTestParser, register_optionflag
+ 
+-from cStringIO import StringIO
++from io import StringIO
+ 
+ import nltk.corpus
+ from nltk import defaultdict

python-nltk.changes

@@ -1,3 +1,12 @@
+-------------------------------------------------------------------
+Mon Dec 26 10:41:22 UTC 2022 - Matej Cepl <mcepl@suse.com>
+
+- Complete nltk_data.tar.xz for offline testing
+- Fix failing tests (gh#nltk/nltk#2969) by adding patches:
+  - port-2to3.patch
+  - skip-networked-test.patch
+- Clean up the SPEC to get rid of rpmlint warnings.
+
 -------------------------------------------------------------------
 Tue Mar 22 07:48:14 UTC 2022 - Matej Cepl <mcepl@suse.com>
 

python-nltk.rpmlintrc

@@ -0,0 +1 @@
+addFilter("E: zero-length /usr/lib/python3\.\d/site-packages/nltk/tbl/api\.py")

python-nltk.spec

@@ -25,7 +25,18 @@ Release:        0
 Summary:        Natural Language Toolkit
 License:        Apache-2.0
 URL:            http://nltk.org/
-Source:         https://files.pythonhosted.org/packages/source/n/nltk/%{pyname}-%{version}.zip
+Source0:        https://files.pythonhosted.org/packages/source/n/nltk/%{pyname}-%{version}.zip
+# Downloaded NLTK data via python3 -m nltk.downloader,
+# then unzip downloaded zip archive.
+# see https://www.nltk.org/data.html for more details
+Source1:        nltk_data.tar.xz
+Source99:       python-nltk.rpmlintrc
+# PATCH-FIX-UPSTREAM skip-networked-test.patch gh#nltk/nltk#2969 mcepl@suse.com
+# skip tests requiring network connection
+Patch0:         skip-networked-test.patch
+# PATCH-FIX-UPSTREAM port-2to3.patch bsc#[0-9]+ mcepl@suse.com
+# port scripts in nltk_data to Python 3
+Patch1:         port-2to3.patch
 BuildRequires:  %{python_module regex}
 BuildRequires:  %{python_module setuptools}
 BuildRequires:  %{python_module six}
@@ -33,6 +44,27 @@ BuildRequires:  %{pythons}
 BuildRequires:  fdupes
 BuildRequires:  python-rpm-macros
 BuildRequires:  unzip
+# For testing
+BuildRequires:  %{python_module tk}
+BuildRequires:  %{python_module click}
+BuildRequires:  %{python_module pytest}
+# BuildRequires:  %%{python_module gensim}
+BuildRequires:  %{python_module joblib}
+BuildRequires:  %{python_module Jinja2}
+BuildRequires:  %{python_module matplotlib}
+BuildRequires:  %{python_module numpy}
+BuildRequires:  %{python_module pyparsing}
+BuildRequires:  %{python_module pytest-cov}
+BuildRequires:  %{python_module pytest-mock}
+BuildRequires:  %{python_module python-crfsuite}
+BuildRequires:  %{python_module regex}
+BuildRequires:  %{python_module requests}
+BuildRequires:  %{python_module scikit-learn}
+BuildRequires:  %{python_module scipy}
+BuildRequires:  %{python_module text-unidecode}
+BuildRequires:  %{python_module tqdm}
+BuildRequires:  %{python_module twython}
+#
 Requires:       python-regex
 Requires:       python-six
 Recommends:     python-gensim
@@ -49,19 +81,49 @@ Requires(postun):update-alternatives
 BuildArch:      noarch
 %python_subpackages
 
+# changedir = nltk/test
+
 %description
 NLTK -- the Natural Language Toolkit -- is a suite of
 Python modules, data sets and tutorials supporting research and
 development in Natural Language Processing.
 
 %prep
-%autosetup -p1 -n %{pyname}-%{version}
+%autosetup -p1 -a1 -n %{pyname}-%{version}
 
-# sed -i "1,4{/\/usr\/bin\/env/d}" nltk/corpus/reader/knbc.py
+# Remove obsolete scripts
-# sed -i "1,4{/\/usr\/bin\/env/d}" nltk/test/runtests.py
+rm tools/nltk_term_index.py tools/run_doctests.py nltk_data/corpora/semcor/semcor.py
-# sed -i "1,4{/\/usr\/bin\/env/d}" nltk/test/unit/test_tgrep.py
+
-# sed -i "1,4{/\/usr\/bin\/env/d}" nltk/tgrep.py
+# Fix EOL
-# sed -i "1,4{/\/usr\/bin\/env/d}" nltk/tokenize/stanford_segmenter.py
+sed -i 's/\r/\n/g; s/\n$//' \
+    README.md \
+    nltk/corpus/reader/knbc.py \
+    nltk/test/unit/test_tgrep.py \
+    nltk/tgrep.py \
+    nltk/tokenize/stanford_segmenter.py \
+    nltk/corpus/reader/knbc.py \
+    nltk/test/unit/test_tgrep.py \
+    nltk/tgrep.py \
+    nltk/tokenize/stanford_segmenter.py \
+    nltk/corpus/reader/knbc.py \
+    nltk/test/unit/test_tgrep.py \
+    nltk/tgrep.py \
+    nltk/tokenize/stanford_segmenter.py
+
+# Remove unrequired shebangs
+sed -E -i "/#![[:space:]]*\/usr\/bin\/env python/d" \
+    nltk/tgrep.py \
+    nltk/tokenize/stanford_segmenter.py \
+    nltk/test/unit/test_tgrep.py \
+    nltk/corpus/reader/knbc.py
+
+# Switch shebangs to the standard Python interpreter
+sed -E -i "s|#![[:space:]]*%{_bindir}/env python|#!%{_bindir}/python3|" \
+    setup.py \
+    tools/global_replace.py \
+    nltk_data/corpora/pl196x/splitter.py \
+    tools/find_deprecated.py \
+    tools/svnmime.py
 
 %build
 %python_build
@@ -75,8 +137,9 @@ chmod -x %{buildroot}%{$python_sitelib}/nltk/test/dependency.doctest
 }
 
 %check
-# FOLLOWING http://www.nltk.org/install.html
+export NLTK_DATA=$(readlink -f ./nltk_data/)
-%python_exec -c "import nltk" || exit 1
+# export PYTEST_ADDOPTS="--doctest-modules"
+%pytest -k 'not network'
 
 %post
 %python_install_alternative nltk

skip-networked-test.patch

@@ -0,0 +1,35 @@
+---
+ nltk/test/unit/test_downloader.py |    4 ++++
+ setup.cfg                         |    4 ++++
+ 2 files changed, 8 insertions(+)
+
+--- a/nltk/test/unit/test_downloader.py
++++ b/nltk/test/unit/test_downloader.py
+@@ -1,6 +1,9 @@
+ from nltk import download
+ 
++import pytest
+ 
++
++@pytest.mark.network
+ def test_downloader_using_existing_parent_download_dir(tmp_path):
+     """Test that download works properly when the parent folder of the download_dir exists"""
+ 
+@@ -9,6 +12,7 @@ def test_downloader_using_existing_paren
+     assert download_status is True
+ 
+ 
++@pytest.mark.network
+ def test_downloader_using_non_existing_parent_download_dir(tmp_path):
+     """Test that download works properly when the parent folder of the download_dir does not exist"""
+ 
+--- a/setup.cfg
++++ b/setup.cfg
+@@ -1,3 +1,7 @@
++[tool:pytest]
++markers =
++   network: test case requires network connection
++
+ [metadata]
+ license_files = 
+ 	LICENSE.txt