Пример #1
0
            imdb_id = url.split("/")[-2]
            image.retrieve(image_url,
                           "../www/static/images/posters/%s.png" % imdb_id)
            cursor.execute(
                "INSERT IGNORE INTO movies (imdb_id, name, image_url,  plot, created_time)"
                "VALUES (%s, %s, %s, %s, %s)",
                (imdb_id,
                 name.encode("utf-8"), image_url, plot.encode("utf-8"),
                 datetime.now().strftime("%Y-%m-%d %H:%M:%S")))

            if cursor.lastrowid != 0:
                self.distributor.add_task(DetailScrapeTask, {
                    "url": url,
                    "imdb_id": imdb_id,
                    "id": cursor.lastrowid
                })

        yield None


if __name__ == "__main__":
    logging.basicConfig(
        level=logging.DEBUG,
        filename="logs/imdb_nowplaying.log",
        filemode="w",
        format="%(asctime)s\t%(levelname)s\t%(threadName)s\t%(message)s")
    distributor = ThreadDistributor(1)
    distributor.add_task(NowPlayingScrapeTask,
                         "http://www.imdb.com/nowplaying/")
    distributor.run()
Пример #2
0
db_connect = MySQLdb.connect(DB["host"], DB["user"], DB["passwd"], DB["name"])
cursor = db_connect.cursor()


class InitializeTask(Task):
    """ Read tweet from database for recent movie
    
    """
    
    def run(self):
        cursor.execute("""SELECT `id`, `name` FROM movies
                       WHERE DATEDIFF(NOW(), created_time) <= %s""", 
                       TIME_TH)
        movies = cursor.fetchall()
        print movies
        
        yield None


if __name__ == "__main__":
    logging.basicConfig(
        level = logging.DEBUG,
        filename = "logs/tweets_error.log",
        filemod = "w",
        format="%(asctime)s\t%(levelname)s\t(%(threadName)-10s)\t%(message)s"
    )
    distributor = ThreadDistributor(N_THREADS)
    distributor.add_task(InitializeTask)
    distributor.run()
    cursor.close()