def initial_registry(self, paths):
"""
Build initial report registry from given paths.
To fit to validator behaviors, if a path is an existing local file path
will be resolved to its absolute path. If it does not exists or an URL
it is left unchanged.
Arguments:
paths (list): List of page path(s) which have been required for
checking.
Returns:
list: Registry of required path with initial value ``None``,
except for unexisting file paths which will contain a critical
error log.
"""
registry = []
for path in paths:
path_key = path
if is_local_ressource(path):
if os.path.exists(path):
path_key = os.path.abspath(path)
registry.append((path_key, None))
return registry
def get_urls(self, path):
"""
Parse given sitemap to get urls.
Arguments:
path (string): Sitemap path.
Returns:
list: List of urls.
"""
if is_local_ressource(path):
ressource = self.get_file_ressource(path)
else:
ressource = self.get_url_ressource(path)
contenttype = self.contenttype(path)
if contenttype == "json":
return self.parse_sitemap_json(ressource)
elif contenttype == "xml":
return self.parse_sitemap_xml(ressource)
# Should never occurs
else:
msg = ("Unable to parse ressource from given path, unknowed "
"content type: {}".format(contenttype))
raise SitemapInvalidError(msg)
def validate_sitemap_path(logger, path):
"""
Validate sitemap file path.
Invalid sitemap file path is a critical error which should stop program
execution.
Arguments:
logger (logging.logger): Logging object to output error messages.
paths (list): List of path to validate, only filepaths are checked.
Returns:
bool: ``True`` if given filepath is a valid local path or an url, else
it returns ``False``.
"""
if is_local_ressource(path):
if not os.path.exists(path):
msg = "Given sitemap path does not exists: {}".format(path)
logger.critical(msg)
return False
elif os.path.isdir(path):
msg = "Directory sitemap path are not supported: {}".format(path)
logger.critical(msg)
return False
return True
def check_local_filepath(self, path):
"""
Check local file path exist and is not a directory.
Arguments:
logger (logging.logger): Logging object to output error messages.
paths (list): List of path to validate, only filepaths are checked.
"""
if is_local_ressource(path):
if not os.path.exists(path):
msg = "Given path does not exists: {}".format(path)
return msg
elif os.path.isdir(path):
msg = "Directory path are not supported: {}".format(path)
return msg
return False
def test_is_local_ressource(path, expected):
"""
Should detect if given path is a filepath.
"""
assert expected == is_local_ressource(path)