def get_clean_text(fileid):
try:
raw = acquire.load_etext(fileid)
tx = strip_headers.strip_headers(raw).strip()
raw_title = tx.split('\n', 1)[0]
tidy_title = raw_title.translate(
str.maketrans('', '', string.punctuation))
return ({"title": tidy_title, "text": tx})
except Exception as e:
print(e)
return ({})
def test_unreachable_mirror(self):
self.request_head_response(ok=False)
with self.assertRaises(UnknownDownloadUriException):
text.load_etext(1)
def test_load_etext(self):
etext = text.load_etext(2701)
self.assertIsInstance(etext, str)
self.assertGreater(len(etext), 1000)
def test_invalid_etext(self):
with self.assertRaises(UnknownDownloadUriException):
text.load_etext(1, mirror='http://example.com')
def test_unreachable_mirror(self):
self.request_head_response(ok=False)
with self.assertRaises(UnknownDownloadUriException):
text.load_etext(1)
def test_load_etext(self):
etext = text.load_etext(2701)
self.assertIsInstance(etext, str)
self.assertGreater(len(etext), 1000)
def test_invalid_etext(self):
with self.assertRaises(UnknownDownloadUriException):
text.load_etext(1, mirror='http://example.com')
