def intents_to_df_cosine_prep()

in src/dfcx_scrapi/core/intents.py [0:0]

• 33 lines of code
• 8 McCabe index (conditional complexity)

    def intents_to_df_cosine_prep(
        self,
        agent_id: str = None) -> Tuple[pd.DataFrame, Dict[str,str]]:
        """Exports a dataframe and defaultdict of Intents for use with Cosine
        Similarity tools.

        Args:
          agent_id: agent to pull list of intents from

        Returns:
          A tuple containing two values:
            A Pandas Dataframe of Intents and TPs
            A Defaultdict(List) that is prepped to feed to the
              Cosine similarity tool (offline)
        """
        if not agent_id:
            agent_id = self.agent_id

        intent_dict = defaultdict(list)
        intents = self.list_intents(agent_id)

        for intent in intents:  # pylint: disable=R1702
            if intent.display_name == "Default Negative Intent":
                pass
            else:
                if "training_phrases" in intent:
                    for training_phrase in intent.training_phrases:
                        text_list = []
                        if len(training_phrase.parts) > 1:
                            for item in training_phrase.parts:
                                text_list.append(item.text)
                            intent_dict[intent.display_name].append(
                                "".join(text_list))
                        else:
                            intent_dict[intent.display_name].append(
                                training_phrase.parts[0].text
                            )
                else:
                    intent_dict[intent.display_name].append("")

        dataframe = pd.DataFrame.from_dict(
            intent_dict, orient="index").transpose()
        dataframe = dataframe.stack().to_frame().reset_index(level=1)
        dataframe = dataframe.rename(
            columns={"level_1": "display_name",
            0: "training_phrase"}).reset_index(drop=True)
        dataframe = dataframe.sort_values(["display_name", "training_phrase"])

        return dataframe, intent_dict