KeyError in organize_speaker_convo_history()

[jming]

Summary of the issue

---

KeyError                                  Traceback (most recent call last) Cell In [21], line 1 ----> 1 true_corpus.organize_speaker_convo_history()

File /usr/local/lib/python3.10/site-packages/convokit/model/corpus.py:1401, in Corpus.organize_speaker_convo_history(self, utterance_filter)    1399 for convo, utts in convo_utts.items():    1400     sorted_utts = sorted(utts, key=lambda x: (x[1], x[0])) -> 1401     self.set_speaker_convo_info(    1402         speaker, convo, "utterance_ids", [x[0] for x in sorted_utts]    1403     )    1404     self.set_speaker_convo_info(speaker, convo, "start_time", sorted_utts[0][1])    1405     self.set_speaker_convo_info(speaker, convo, "n_utterances", len(sorted_utts))

File /usr/local/lib/python3.10/site-packages/convokit/model/corpus.py:1351, in Corpus.set_speaker_convo_info(self, speaker_id, convo_id, key, value)    1349 if convo_id not in speaker.meta["conversations"]:    1350     speaker.meta["conversations"][convo_id] = {} -> 1351 speaker.meta["conversations"][convo_id][key] = value

KeyError: '1dhwyu'

Steps to reproduce

true_corpus = Corpus(filename=download('subreddit-TrueChristian'), storage_type="db") true_corpus.organize_speaker_convo_history()

Additional information

ConvoKit version: pip3 install git+https://github.com/CornellNLP/Cornell-Conversational-Analysis-Toolkit.git

[jpwchang]

I've pushed a small patch to fix this issue - if you uninstall and then reinstall ConvoKit from the git repo, it should work now!

[cristiandnm]

Let us know if this fixes the issue, so we can close this here.