def test_html_detect(self): self.assertTrue(HTMLReader.is_file( io.BytesIO('<html><body>hi</body></html>'.encode('utf-16le')) )) self.assertFalse(HTMLReader.is_file( io.BytesIO('hello world!'.encode('utf-16le')) )) self.assertTrue(HTMLReader.is_file( io.BytesIO(b'<title>hello</title>hi') )) self.assertTrue(HTMLReader.is_file( io.BytesIO(b'<html><body>hello') )) self.assertTrue(HTMLReader.is_file( io.BytesIO( b'The document has moved <a href="somewhere.html">here</a>' ) )) self.assertTrue( HTMLReader.is_url(URLInfo.parse('example.com/index.htm')) ) self.assertTrue( HTMLReader.is_url(URLInfo.parse('example.com/index.html')) ) self.assertTrue( HTMLReader.is_url(URLInfo.parse('example.com/index.dhtm')) ) self.assertTrue( HTMLReader.is_url(URLInfo.parse('example.com/index.xhtml')) ) self.assertTrue( HTMLReader.is_url(URLInfo.parse('example.com/index.xht')) ) self.assertFalse( HTMLReader.is_url(URLInfo.parse('example.com/image.jpg')) ) self.assertTrue( HTMLReader.is_request(Request.new('example.com/index.html')) ) self.assertFalse( HTMLReader.is_request(Request.new('example.com/image.jpg')) ) response = Response('HTTP/1.0', '200', 'OK') response.fields['Content-Type'] = 'text/html' self.assertTrue(HTMLReader.is_response(response)) response = Response('HTTP/1.0', '200', 'OK') response.fields['Content-Type'] = 'image/png' self.assertFalse(HTMLReader.is_response(response))