search_relevance_6.py

def search_tfidf_df(tfidf_df, text_df, query_string: str):
    """
    - Parameters: tfidf_df (Pandas DataFrame) representing a tf-idf
      matrix, text_df (Pandas DataFrame) with a "text" column and rows
      that correspond to the tfidf_df, and query_string (string).
    - Returns: A new dataframe that only contains rows from text_df where
      the corresponding tf-idf value was greater than zero for each of
      the terms in query_string. Additional columns are added to show the
      tf-idf value for each term and the sum of the tf-idf values. 
    """
    terms = query_string.lower().split(" ")
    filters = [tfidf_df[term] > 0 for term in terms]
    filtered_tfidf_df = tfidf_df[np.all(filters, axis=0)][terms]
    filtered_tfidf_df["tfidf_sum"] = filtered_tfidf_df.agg(sum, axis=1)
    full_df = text_df.merge(filtered_tfidf_df, 
                            left_index=True, right_index=True)

    return full_df.sort_values("tfidf_sum", ascending=False)