def _parse_match(self, match):
"""
If there is a group dict, return the dict;
even if there's only one value in the dict, return a dictionary;
If there is a group in match, return the group;
if there is only one value in the group, return the value;
if there has no group, return the whole matched string;
if there are many groups, return a tuple;
:param match:
:return:
"""
if not match:
if self.default:
return self.default
else:
raise NothingMatchedError(
f"Extract `{self._re_select}` error, "
f"please check selector or set parameter named `default`"
)
else:
string = match.group()
groups = match.groups()
group_dict = match.groupdict()
if group_dict:
return group_dict
if groups:
return groups[0] if len(groups) == 1 else groups
return string
def extract(self, html_etree: etree._Element, is_source: bool = False):
elements = self._get_elements(html_etree=html_etree)
if is_source:
return elements if self.many else elements[0]
if elements:
results = [self._parse_element(element) for element in elements]
elif self.default is None:
raise NothingMatchedError(
f"Extract `{self.css_select or self.xpath_select}` error, "
f"please check selector or set parameter named `default`")
else:
results = [self.default]
return results if self.many else results[0]