Open bowdens opened 6 years ago
Also one thing to note is that there is an API limit for twitter, so it would be good if the bot can make sure by itself that it won't exceed the API limits. But that shouldn't be a problem so long as it doesn't get too popular.
@bowdens I think I can do both of them, but which implementation would you prefer? parse_tweet
could return multiple targets and then we could just add a loop. Making a Markov chain of multiple users would also be easy, because we could just add all the words to one bag.
@MateuszOkulus I think the multiple tweet implementation probably makes the most sense for the user. You could even add some way that the user could choose between the different modes (maybe the the user types -combine or something in the tweet). it's totally up to you which way you implement it though!
Currently the bot will only make a tweet for a single \@mention. For instance, if I was to tweet "\@sim_tweet \@bo_wd_en \@realDonaldTrump" the bot would tweet back to me with an impersonation of \@bo_wd_en but not \@realDonaldTrump.
There are two ways to handle this. One is to simply make the bot create multiple reply tweets for each \@mention, or the other is to add each \@mentioned user to the markov chain to create a mixed impersonation. So if I was to tweet as before with \@bo_wd_en and \@realDonaldTrump, then it would create a single tweet in response that would be a mix between an impersonation of me and Donald Trump.