refactor evaluation data

def make_dataframe(annotations, opentapioca):

    data = []

    print("opentapioca keys", opentapioca.keys())

    for annotation in annotations.values():

        text, true_labels, pred_labels = annotation

        filename = text.get("document")

        labels_tapioca = sorted(opentapioca.get('choices'))

        if filename not in opentapioca.keys():

            print("Skipping", filename, "not found in OpenTapioca predictions")

            continue

        labels_tapioca = (opentapioca.get(filename, {}).get('choices'))

        options = dict([(v.get('value'), v.get('html').split('-')[0]) for v in text.get("options")])

    #    print('options', options)

        labels_true = [v.get("value", {}).get("choices") for v in true_labels if "choices" in v.get("value",{})]

        if labels_true:

            labels_true = sorted(labels_true[0])

            labels_true = labels_true[0]

        labels_pred = [v.get("value", {}).get("choices") for v in pred_labels if "choices" in v.get("value",{})]

        labels_pred = sorted(labels_pred[0])

        print(filename)

        if labels_pred:

            labels_pred = labels_pred[0]

#        print(filename)

#        print("labels_tapi", sorted(labels_tapioca))

#        print("labels_true", sorted(labels_true))

#        print("labels_pred", sorted(labels_pred))

        for k, v in options.items():

            data.append({"filename": filename, "qname": k, "entity": v, "annotation": k in labels_true, "prediction": k in labels_pred, "opentapioca": k in labels_tapioca, "correct_my": (k in labels_pred) == (k in labels_true), "correct_tapioca": (k in labels_tapioca) == (k in labels_true)})

    print(df['correct_tapioca'].groupby(df['correct_tapioca']).count())

    return df

def main(argv):

def main():

    parser = argparse.ArgumentParser(description="Evaluate NEL annotations")

    parser.add_argument("database_file", help="Path to the SQLite database file")

    parser.add_argument("project_id", help="Project ID to evaluate")

    parser.add_argument("opentapioca_file", help="OpenTapioca predictions in JSON")

    parser.add_argument("opentapioca_files", help="Directory with OpenTapioca predictions in JSON")

    parser.add_argument("output_dirname", help="Directory to save the output")

    args = parser.parse_args()

    database_file = args.database_file

    project_id = args.project_id

    opentapioca_file = args.opentapioca_file

    opentapioca_files = args.opentapioca_files

    output_dirname = args.output_dirname

    if not os.path.exists(database_file):

        print(f"Database file {database_file} does not exist.")

        sys.exit(1)

    if not os.path.exists(opentapioca_file):

        print(f"OpenTapioca file {opentapioca_file} does not exist.")

    if not os.path.exists(opentapioca_files):

        print(f"Directory with OpenTapioca files {opentapioca_files} does not exist.")

        sys.exit(1)

    opentapioca = None

    with open(opentapioca_file, 'r') as f:

    opentapioca = {}

    for opentapioca_file in os.listdir(opentapioca_files):

        with open(os.path.join(opentapioca_files, opentapioca_file), 'r', encoding='utf-8') as f:

            opentapioca_json = json.load(f)

            filename = os.path.basename(opentapioca_file).split('.')[0]

            try:

            opentapioca = opentapioca_json.get('predictions')[0]["result"][-1].get('value')

                opentapioca[filename] = opentapioca_json.get('predictions')[0]["result"][-1].get('value')

            except:

            print("OpenTapioca JSON does not contain predictions or is not well formatted")

            sys.exit(1)

                pass

    if not opentapioca:

    if not opentapioca.keys():

        print("OpenTapioca predictions not found")

        sys.exit(1)

    print("OpenTapioca predictions", opentapioca)

    annotations = read_database(database_file, project_id)

    if not os.path.exists(output_dirname):

        df.to_csv(f, sep='\t')

if __name__ == "__main__":

    main(sys.argv)

 No newline at end of file

    main()

 No newline at end of file