def test_link_extractor(self): self.assertEqual(utils.clean_links('http://www.catb.org/jargon/html/S/schroedinbug.html\n'), ['http://www.catb.org/jargon/html/S/schroedinbug.html']) self.assertEqual(utils.clean_links('catb.org/jargon/html/S/schroedinbug.html'), []) self.assertEqual(utils.clean_links('http:// catb.org/jargon/html/S/schroedinbug.html'), []) self.assertEqual(utils.clean_links('http://www.catb.org/jargon/html/S/schroedinbug.html\nhttp://www.catb.org'), ['http://www.catb.org/jargon/html/S/schroedinbug.html', 'http://www.catb.org']) self.assertEqual(utils.clean_links('http://www.catb.org/jargon/html/S/schroedinbug.html http://www.catb.org'), ['http://www.catb.org/jargon/html/S/schroedinbug.html', 'http://www.catb.org']) self.assertEqual(utils.clean_links('http://www.catb.org/jargon/html/S/schroedinbug.html-http://www.catb.org'), ['http://www.catb.org/jargon/html/S/schroedinbug.html-http://www.catb.org']) self.assertEqual(utils.clean_links('https://www.catb.org/jargon/html/S/schroedinbug.html\nhttps://www.catb.org'), ['https://www.catb.org/jargon/html/S/schroedinbug.html', 'https://www.catb.org']) self.assertEqual(utils.clean_links('ftp://www.catb.org/jargon/html/S/schroedinbug.html\nftp://www.catb.org'), ['ftp://www.catb.org/jargon/html/S/schroedinbug.html', 'ftp://www.catb.org'])
def test_link_extractor(self): self.assertEqual( utils.clean_links( 'http://www.catb.org/jargon/html/S/schroedinbug.html\n'), ['http://www.catb.org/jargon/html/S/schroedinbug.html']) self.assertEqual( utils.clean_links('catb.org/jargon/html/S/schroedinbug.html'), []) self.assertEqual( utils.clean_links( 'http:// catb.org/jargon/html/S/schroedinbug.html'), []) self.assertEqual( utils.clean_links( 'http://www.catb.org/jargon/html/S/schroedinbug.html\nhttp://www.catb.org' ), [ 'http://www.catb.org/jargon/html/S/schroedinbug.html', 'http://www.catb.org' ]) self.assertEqual( utils.clean_links( 'http://www.catb.org/jargon/html/S/schroedinbug.html http://www.catb.org' ), [ 'http://www.catb.org/jargon/html/S/schroedinbug.html', 'http://www.catb.org' ]) self.assertEqual( utils.clean_links( 'http://www.catb.org/jargon/html/S/schroedinbug.html-http://www.catb.org' ), [ 'http://www.catb.org/jargon/html/S/schroedinbug.html-http://www.catb.org' ]) self.assertEqual( utils.clean_links( 'https://www.catb.org/jargon/html/S/schroedinbug.html\nhttps://www.catb.org' ), [ 'https://www.catb.org/jargon/html/S/schroedinbug.html', 'https://www.catb.org' ]) self.assertEqual( utils.clean_links( 'ftp://www.catb.org/jargon/html/S/schroedinbug.html\nftp://www.catb.org' ), [ 'ftp://www.catb.org/jargon/html/S/schroedinbug.html', 'ftp://www.catb.org' ])
def add_links(self, inp): """Returns true (evaluating expression) on nonemtpy and false on empty input """ if len(inp) == 0: return False parsed_input = utils.clean_links(inp) if self.type == "links": for link in parsed_input: self.links.append(link) else: # == dlc for link in parsed_input: dlc_links = dlc_to_links(link) if dlc_links != None: self.links.extend(dlc_links) return len(parsed_input)