Exemple #1
0
    def parse(self, response):
        self.driver.get(response.url)
        self.scroll_down_n_times(1)

        questions = self.get_questions()
        time.sleep(self.SLEEP_TIME)

        # best_answers = self.get_best_answers()
        # time.sleep(self.SLEEP_TIME)

        dates = self.get_question_dates()
        time.sleep(self.SLEEP_TIME)

        # assert len(questions) == len(best_answers) == len(dates)

        if not self.CATEGORY:
            categories = self.get_categories()
            time.sleep(self.SLEEP_TIME)
            assert len(questions) == len(categories)

        for i in range(len(questions)):
            date     = DbManager.convert_date_string_to_datetime(dates[i])
            category = self.CATEGORY if self.CATEGORY else categories[i]
            question = { 'content': questions[i], 'upvotes': -1, 'category': category, 'timestamp': date }
            DbManager.add_question(question)

            # answer = { 'content': best_answers[i], 'upvotes': -1, 'is_best_answer': True, 'timestamp': datetime.now()}
            # DbManager.add_answer(answer)

        self.driver.close()