How to use the matminer.featurizers.conversions.StructureToComposition function in matminer

if self.composition_col in df.columns:
                logger.info(
                    self._log_prefix + "composition column already exists, "
                    "overwriting with composition from structure."
                )
            else:
                logger.info(
                    self._log_prefix + "Adding compositions from structures."
                )

            df = self._tidy_column(df, self.structure_col)

            # above tidy column will add oxidation states, these oxidation
            # states will then be transferred to composition.
            struct2comp = StructureToComposition(
                reduce=True,
                target_col_id=self.composition_col,
                overwrite_data=overwrite,
            )
            struct2comp.set_n_jobs(self.n_jobs)
            df = struct2comp.featurize_dataframe(df, self.structure_col)
        return df

"hl_mass_x": "hole mass_x",
            "hl_mass_y": "hole mass_y",
            "hl_mass_z": "hole mass_z",
            "kv": "bulk modulus",
            "magmom": "mu_b",
            "mbj_gap": "gap tbmbj",
            "mepsx": "epsilon_x tbmbj",
            "mepsy": "epsilon_y tbmbj",
            "mepsz": "epsilon_z tbmbj",
            "op_gap": "gap tbmbj",
            "strt": "structure",
            }

    df = df.rename(columns=colmap)
    df = df.drop(["multi_elastic", "fin_enp"], axis=1)
    s = StructureToComposition()
    df = s.featurize_dataframe(df, "structure")
    return df