Restarting fetch to different filtering needs

[gawry]

It would be good to have a easy way to change tracked words once you are getting the stream feed.

[knutole]

I've been thinking about this too, but it appears to me that Twitter does not support it - I can't find anything in their API to suggest it, at least. I'd love to be proved wrong though.

[gawry]

I did a successful implementation, but doesn't seems very good as I don't stop the current connection.

    stream = tweetstream.TweetStream(configuration,ioloop=main_io_loop)

    stream.fetch("/1.1/statuses/filter.json?track="+tornado.escape.url_escape(words), callback=callback)

    def check_words():
        global words, stream
        with open('words.txt') as file:
            #using a file for testing purposes, real implementation should use access a db
            newwords = file.read()
        if words != newwords:
            words = newwords
            print newwords
            try:
                #start another process (and twitter kills the last one), it would be good to gracefully kill the current connection
                stream = tweetstream.TweetStream(configuration,ioloop=main_io_loop)
                stream.fetch("/1.1/statuses/filter.json?track="+tornado.escape.url_escape(words), callback=callback)
                interval_ms = 1000*60
                scheduler = tornado.ioloop.PeriodicCallback(check_words,interval_ms,io_loop = main_io_loop)
                scheduler.start()
                main_io_loop.start()
            except:
                pass
            file.close()

    interval_ms = 1000*60
    scheduler = tornado.ioloop.PeriodicCallback(check_words,interval_ms,io_loop = main_io_loop)
    scheduler.start()
    main_io_loop.start()