def get_body(self): try: element_body = self.get_element_body() except NoneMainDiv: return None else: try_remove(self.parser, element_body, "div.mna-tools") try_remove(self.parser, element_body, "div.mna-comment-call") try: element_body.remove(self.get_element_author()) except NoAuthorElement: pass return self.parser.tostring(element_body)
def get_body(self): try : element_body = self.get_element_body() except NoneMainDiv: return None else: div_header_element = self.parser.select(element_body, "div.header", 1) element_detail = self.parser.select(element_body, "div.details", 1) div_content_element = self.parser.select(element_body, "div.content", 1) drop_comments(element_body) try_remove(self.parser, element_body, "div.sidebar") try_remove(self.parser, element_detail, "div.footer") try_remove_from_selector_list(self.parser, div_header_element, ["h1", "div.picture", "div.date", "div.news-single-img", "div.metas_img", "strong"]) try_remove_from_selector_list(self.parser, div_content_element, ["div.tw_button", "div.wpfblike"]) try : description_element = self.parser.select(div_header_element, "div.description", 1) except BrokenPageError: pass else: text_content = description_element.text_content() if len(text_content.strip()) == 0 : description_element.drop_tree() else: if len(description_element) == 1: description_element.drop_tag() if len(div_header_element.text_content().strip()) == 0: div_header_element.drop_tree() if len(div_header_element) == 1: div_header_element.drop_tag() if len(element_detail) == 1: element_detail.drop_tag() div_content_element.drop_tag() return self.parser.tostring(element_body)
def get_body(self): try: element_body = self.get_element_body() except NoneMainDiv: return None else: div_header_element = self.parser.select(element_body, "div.header", 1) element_detail = self.parser.select(element_body, "div.details", 1) div_content_element = self.parser.select(element_body, "div.content", 1) drop_comments(element_body) try_remove(self.parser, element_body, "div.sidebar") try_remove(self.parser, element_detail, "div.footer") try_remove_from_selector_list(self.parser, div_header_element, [ "h1", "div.picture", "div.date", "div.news-single-img", "div.metas_img", "strong" ]) try_remove_from_selector_list(self.parser, div_content_element, ["div.tw_button", "div.wpfblike"]) try: description_element = self.parser.select( div_header_element, "div.description", 1) except BrokenPageError: pass else: text_content = description_element.text_content() if len(text_content.strip()) == 0: description_element.drop_tree() else: if len(description_element) == 1: description_element.drop_tag() if len(div_header_element.text_content().strip()) == 0: div_header_element.drop_tree() if len(div_header_element) == 1: div_header_element.drop_tag() if len(element_detail) == 1: element_detail.drop_tag() div_content_element.drop_tag() return self.parser.tostring(element_body)