def setUp(self):
"""Set up local variables"""
self.jsonl_file = os.path.join(os.path.dirname(__file__),
'sample.jsonl')
self.txt_file = io_utils.change_extension(self.jsonl_file, 'txt')
self.copy_txt = io_utils.change_extension(self.jsonl_file, 'copy.txt')
io_utils.check_file_readable(self.jsonl_file)
def setUp(self):
"""Set up local variables"""
self.arpa = os.path.join(os.path.dirname(__file__), 'sample-model.arpa')
self.bin = io_utils.change_extension(self.arpa, 'bin')
self.tmp = io_utils.change_extension(self.arpa, 'tmp.bin')
io_utils.check_file_readable(self.arpa)
io_utils.check_file_readable(self.bin)
def setUp(self):
"""Set up local variables"""
self.csv_file = os.path.join(os.path.dirname(__file__), 'sample.csv')
self.jsonl_file = io_utils.change_extension(self.csv_file, 'jsonl')
self.ft_model = '/usr/share/ccquery/models/fastText/lid.176.bin'
io_utils.check_file_readable(self.csv_file)
io_utils.check_file_readable(self.jsonl_file)
io_utils.check_file_readable(self.ft_model)
self.copy_csv = io_utils.change_extension(self.csv_file, 'copy.csv')
self.copy_jsonl = io_utils.change_extension(self.csv_file,
'copy.jsonl')
def test_bad_extension(self):
"""Test wrong file extension"""
fn = io_utils.change_extension(self.jsonl, 'png')
with self.assertRaises(Exception) as context:
QueryAnalysis(fn, token='word', field='noisy')
self.assertTrue('Unknown file extension' in str(context.exception))
def setUp(self):
"""Set up local variables"""
self.extractor = WikiExtraction()
self.sample = os.path.join(os.path.dirname(__file__), 'sample.bz2')
self.data = os.path.join(os.path.dirname(__file__),
'sample-corpus.txt')
self.vocab = os.path.join(os.path.dirname(__file__),
'sample-vocab.txt')
io_utils.check_file_readable(self.sample)
io_utils.check_file_readable(self.data)
io_utils.check_file_readable(self.vocab)
# temporary files
self.files = {
'dld': os.path.join(os.path.dirname(__file__), 'dld.xml'),
'xml': io_utils.change_extension(self.sample, 'xml'),
'jsonl': io_utils.change_extension(self.sample, 'jsonl'),
'txt': io_utils.change_extension(self.sample, 'txt'),
'wvoc': io_utils.change_extension(self.sample, 'wvoc.txt'),
'wplot': io_utils.change_extension(self.sample, 'wvoc.png'),
'cvoc': io_utils.change_extension(self.sample, 'cvoc.txt'),
'cplot': io_utils.change_extension(self.sample, 'cvoc.png'),
}
def test_file(self):
self.assertEqual('__init__.py', io_utils.filename(self.empty_file))
self.assertEqual('.py', io_utils.extension(self.empty_file))
self.assertEqual('__init__', io_utils.basename(self.empty_file))
self.assertEqual('/src/tests/utils',
io_utils.dirname('/src/tests/utils/__init__.py'))
self.assertEqual(
'/src/tests/utils/__init__',
io_utils.path_without_ext('/src/tests/utils/__init__.py'))
self.assertEqual('0.0Bytes', io_utils.filesize(self.empty_file))
self.assertEqual('11.9KB', io_utils.filesize(self.archive))
self.assertEqual(0, io_utils.count_lines(self.empty_file))
self.assertTrue(io_utils.has_extension(self.empty_file, '.py'))
self.assertFalse(io_utils.has_extension(self.empty_file, '.txt'))
self.assertEqual(
'/src/tests/utils/__init__.txt',
io_utils.change_extension('/src/tests/utils/__init__.py', 'txt'))
self.assertEqual('', io_utils.change_extension('', 'txt'))