def test_licence_url_multiple_urls(self):
     # only the first URL is extracted
     assert_equal(GeminiHarvester._extract_first_licence_url(
         ['Reference and PSMA Only',
          'http://www.test.gov.uk/licenseurl',
          'http://www.test.gov.uk/2nd_licenseurl']),
                  'http://www.test.gov.uk/licenseurl')
 def test_licence_url_embedded_at_start(self):
     # URL is embedded at the start of the text field and the
     # whole field is returned. Noting this unusual behaviour
     assert_equal(GeminiHarvester._extract_first_licence_url(
         ['http://www.test.gov.uk/licenseurl Reference and PSMA Only']),
                  'http://www.test.gov.uk/licenseurl Reference and PSMA Only')
 def test_licence_url_embedded(self):
     # URL is embedded within the text field and not extracted
     assert_equal(GeminiHarvester._extract_first_licence_url(
         ['Reference and PSMA Only http://www.test.gov.uk/licenseurl']),
                  None)
 def test_licence_url_normal(self):
     assert_equal(GeminiHarvester._extract_first_licence_url(
         ['Reference and PSMA Only',
          'http://www.test.gov.uk/licenseurl']),
                  'http://www.test.gov.uk/licenseurl')