def _prepare_post_text()

in sourcecode/scoring/topic_model.py [0:0]

• 12 lines of code
• 4 McCabe index (conditional complexity)

  def _prepare_post_text(self, notes: pd.DataFrame) -> pd.DataFrame:
    """Concatenate all notes within each post into a single row associated with the post.

    Args:
      notes: dataframe containing raw note text

    Returns:
      DataFrame with one post per row containing note text
    """
    postNoteText = (
      notes[[c.tweetIdKey, c.summaryKey]]
      .fillna({c.summaryKey: ""})
      .groupby(c.tweetIdKey)[c.summaryKey]
      .apply(lambda postNotes: " ".join(postNotes))
      .reset_index(drop=False)
    )
    # Default tokenization for CountVectorizer will not split on underscore, which
    # results in very long tokens containing many words inside of URLs.  Removing
    # underscores allows us to keep default splitting while fixing that problem.
    postNoteText[c.summaryKey] = [
      text.replace("_", " ") for text in postNoteText[c.summaryKey].values
    ]
    return postNoteText