I’ve looked into this, and I don’t think there is. The only filtering operator for the “stream” API seems to be “filter_level” which is not what we are looking for.
I can tell you that, using the twitter API designed for Python, I was able to extract the information “entities/urls” and sometimes “entities/media” from a tweet. “entities/urls” is by default an empty list (in Python) and will only be non-empty if there are urls to put inside it. If the tweet contains a picture or other form of media, it will show up in “entities/media”.
Sorry this is so late and specific to Python, but I hope it gets people on the right track!