From: operator and username changes

MJo · 2021-08-19T09:46:03.677Z

The from: operator can be used as part of a query to search for tweets by a specific user, specified by their username (the handle).

I am wondering how this query deals with changes of a handle over time. For example, if user 123 has at some point given up the handle “threedigits”, which was then picked up by user 456.

would a query from:threedigits return only tweets by user 456, or also some historical tweets by user 123?

If so, is there any reason to query by username rather than the user’s id?

Thanks!

IgorBrigadir · 2021-08-19T12:13:49.201Z

Do you have an example of a user like this? I’d love to test it out too. I was looking for this for twarc too.

MJo · 2021-08-19T13:42:59.118Z

Sadly not yet, I am just curious.

Probably these cases are rather rare, right?
I have not noticed one in the timelines I have looked at.

I could probably create a test case using my own account - I just hope that nobody will steal my handle from under my nose

IgorBrigadir · 2021-08-19T13:52:28.737Z

That raises a different question - what if handles “swap” between two accounts?

I did end up finding an example of a username that’s all numbers that could be mistaken for an id, like https://twitter.com/92536

So there’s the issue of all digit usernames and swapped usernames. Right now we’re handling it like this:

github.com

DocNow/twarc/blob/6483e834819b6bc519176822d35cf30ca4cafe1d/twarc/client2.py#L888-L904


      
          def _ensure_user_id(self, user):
              """
              Always return a valid user id, look up if not numeric.
              """
              user = str(user)
              is_numeric = re.match(r"^\d+$", user)
          
          
    if len(user) > 15 or (is_numeric and self._id_exists(user)):
                  return user
              else:
                  results = next(self.user_lookup([user], usernames=True))
                  if "data" in results and len(results["data"]) > 0:
                      return results["data"][0]["id"]
                  elif is_numeric:
                      return user
                  else:
                      raise ValueError(f"No such user {user}")

MJo · 2021-08-21T14:12:38.912Z

So I have found a couple of situations where I believe a handle has been moved from one user to another.
I can only find these in a data from the old sampled stream, however.

Using the V2API endpoints (and indeed, when looking up the tweets themselves in other apps) it appears that when a user changes their handle, that change is retroactively applied to all their tweets. Or maybe the current handle is simply linked to the id anytime a request is made?

In any case, I believe that answers my initial question. Looking up tweets from:threedigits will only return tweets from the user who bears that handle currently. And it will also return any tweets from before he adopted that handle, if a change has taken place.

Is that plausible?

As far as the issue with all digit usernames is concerned, that is also an interesting consideration. I will see if i find cases like that.