def clean(self):
cleaned = super(CheckedFileForm, self).clean()
if self._errors:
return cleaned
# Check the URL is a valid URL and exists
exists, fmt, size, error_msg = url_exists(cleaned['url'])
if not exists:
self._errors['url'] = \
[error_msg]
# TODO: Consider uncommenting this
# if fmt == 'HTML':
# self._errors['url'] = \
# [_("This appears to be a web page and not a data file")]
cleaned['is_broken'] = False
cleaned['last_check'] = datetime.datetime.now()
cleaned['format'] = fmt
cleaned['size'] = size
return cleaned
def test_url_exists_ok(self):
exists, fmt, size, error = url_exists('https://google.com')
assert exists
assert fmt == 'HTML', fmt
assert not error
def test_url_missing_proto(self):
exists, fmt, size, error = url_exists('data.gov.uk')
assert not exists
assert fmt == ''
assert 'https' in error
def test_url_does_not_exist(self):
exists, fmt, size, error = url_exists('https://12345.12345.12345.org')
assert not exists
assert fmt == ''
assert 'Failed to connect' in error
def test_url_exists_csv(self):
exists, fmt, size, error = url_exists(
'https://data.gov.uk/data/site-usage/data_all.csv')
assert exists
assert fmt == 'CSV', fmt
assert not error