def test_read_write_file_lines_gzip(self):
expected = [sent.text for sent in self.spacy_doc.sents]
filename = os.path.join(self.tempdir,
'test_read_write_file_lines.txt.gzip')
fileio.write_file_lines(expected, filename)
observed = [line.strip() for line in fileio.read_file_lines(filename)]
self.assertEqual(observed, expected)
def test_read_write_file_lines_bz2(self):
expected = [sent.text for sent in self.spacy_doc.sents]
filename = os.path.join(self.tempdir, 'test_read_write_file_lines.txt.bz2')
fileio.write_file_lines(expected, filename,
mode='wb' if PY2 else 'wt')
observed = [line.strip() for line
in fileio.read_file_lines(filename, mode='r' if PY2 else 'rt')]
self.assertEqual(observed, expected)
def test_read_write_file_lines_bz2(self):
expected = [sent.text for sent in self.spacy_doc.sents]
tempdir = tempfile.mkdtemp()
filename = os.path.join(tempdir, 'test_read_write_file_lines.txt.bz2')
fileio.write_file_lines(expected, filename)
observed = [line.strip() for line in fileio.read_file_lines(filename)]
os.remove(filename)
os.rmdir(tempdir)
self.assertEqual(observed, expected)
def test_read_write_file_lines_bytes(self):
expected = [unicode_to_bytes(sent.text) for sent in self.spacy_doc.sents]
for ext in ('.txt', '.gz', '.bz2', '.xz'):
filename = os.path.join(
self.tempdir, 'test_read_write_file_lines_bytes' + ext)
fileio.write_file_lines(expected, filename, mode='wb',
auto_make_dirs=True)
observed = [line.strip() for line
in fileio.read_file_lines(filename, mode='rb')]
self.assertEqual(observed, expected)
def test_read_write_file_lines_bz2(self):
expected = [sent.text for sent in self.spacy_doc.sents]
filename = os.path.join(self.tempdir,
'test_read_write_file_lines.txt.bz2')
fileio.write_file_lines(expected, filename, mode='wb' if PY2 else 'wt')
observed = [
line.strip()
for line in fileio.read_file_lines(filename,
mode='r' if PY2 else 'rt')
]
self.assertEqual(observed, expected)
def test_read_write_file_lines_unicode(self):
expected = [sent.text for sent in self.spacy_doc.sents]
for ext in ('.txt', '.gz', '.bz2', '.xz'):
filename = os.path.join(
self.tempdir, 'test_read_write_file_lines_unicode' + ext)
if PY2 is True and ext != '.txt':
self.assertRaises(
ValueError, fileio.open_sesame,
filename, 'wt', None, True)
else:
fileio.write_file_lines(expected, filename, mode='wt',
auto_make_dirs=True)
observed = [line.strip() for line
in fileio.read_file_lines(filename, mode='rt')]
self.assertEqual(observed, expected)
def test_read_write_file_lines_unicode(self):
expected = [sent.text for sent in self.spacy_doc.sents]
for ext in ('.txt', '.gz', '.bz2', '.xz'):
filename = os.path.join(
self.tempdir, 'test_read_write_file_lines_unicode' + ext)
if PY2 is True and ext != '.txt':
self.assertRaises(
ValueError, fileio.open_sesame,
filename, 'wt', None, True)
else:
fileio.write_file_lines(expected, filename, mode='wt',
auto_make_dirs=True)
observed = [line.strip() for line
in fileio.read_file_lines(filename, mode='rt')]
self.assertEqual(observed, expected)
def test_read_write_file_lines_bytes(self):
expected = [
unicode_to_bytes(sent.text) for sent in self.spacy_doc.sents
]
for ext in ('.txt', '.gz', '.bz2', '.xz'):
filename = os.path.join(self.tempdir,
'test_read_write_file_lines_bytes' + ext)
if is_python2 is True and ext == '.xz':
self.assertRaises(ValueError, fileio.open_sesame, filename,
'wb', 'utf-8', True)
else:
fileio.write_file_lines(expected,
filename,
mode='wb',
auto_make_dirs=True)
observed = [
line.strip()
for line in fileio.read_file_lines(filename, mode='rb')
]
self.assertEqual(observed, expected)
def test_read_write_file_lines_gzip(self):
expected = [sent.text for sent in self.spacy_doc.sents]
filename = os.path.join(self.tempdir, 'test_read_write_file_lines.txt.gzip')
fileio.write_file_lines(expected, filename)
observed = [line.strip() for line in fileio.read_file_lines(filename)]
self.assertEqual(observed, expected)