def __get_comma_line_on_each(self, attributes): """Encodes charactors and converts from list to string. Args: attributes: A list of a list of strings like above. (ex. [['tennis','soccer',...], ....]) Returns: A list of strings of many keywords like above. (ex. -> ["tennis,soccer", ...]) """ return map(lambda x: Tool.conv_encoding(Constants.SPLIT_CHAR.join(x)), attributes)
def analyze_HTML(self, path, studentID): if len(self.back_studentID) > 0 and self.back_studentID != studentID: self.__end_one_student() with open(path, 'r') as f: parser = StudentHTMLParser(path) content = f.read() parser.feed(Tool.conv_encoding(content)) self.content += parser.content.encode("utf-8") + " " self.page_titles.append(parser.page_title) self.page_paths.append(parser.page_path) self.page_size += parser.page_size parser.close() self.back_studentID = studentID