def sanitize_url(obj):
"""Takes a string representing a URL and sanitizes it.
Args:
obj: a string representing a URL.
Returns:
An empty string if the URL does not start with http:// or https://
except when the string is empty. Otherwise, returns the original
URL.
Raises:
AssertionError: The string is non-empty and does not start with
http:// or https://
"""
if obj == '':
return obj
url_components = python_utils.url_split(obj)
quoted_url_components = (python_utils.url_quote(component)
for component in url_components)
raw = python_utils.url_unsplit(quoted_url_components)
acceptable = html_cleaner.filter_a('a', 'href', obj)
assert acceptable, ('Invalid URL: Sanitized URL should start with '
'\'http://\' or \'https://\'; received %s' % raw)
return raw
def set_url_query_parameter(url, param_name, param_value):
"""Set or replace a query parameter, and return the modified URL.
Args:
url: str. URL string which contains the query parameter.
param_name: str. Parameter name to be removed.
param_value: str. Set the parameter value, if it exists.
Returns:
str. Formated URL that has query parameter set or replaced.
Raises:
Exception. If the query parameter sent is not of string type,
them this exception is raised.
"""
if not isinstance(param_name, python_utils.BASESTRING):
raise Exception(
'URL query parameter name must be a string, received %s'
% param_name)
scheme, netloc, path, query_string, fragment = python_utils.url_split(url)
query_params = python_utils.parse_query_string(query_string)
query_params[param_name] = [param_value]
new_query_string = python_utils.url_encode(query_params, doseq=True)
return python_utils.url_unsplit(
(scheme, netloc, path, new_query_string, fragment))
def test_url_unsplit(self):
response = python_utils.url_split('http://www.google.com')
self.assertEqual(python_utils.url_unsplit(response),
'http://www.google.com')
