collect_entity.py

import json
import re
import os


def extract_movie_ids(text):
    res = []
    for m in re.finditer(r'@(\d+)', text):
        res.append(m.group(1))
    return res

def collect_entity_mappings():
    files = [
        "./test_data_dbpedia.jsonl",
        "./train_data_dbpedia.jsonl",
        "valid_data_dbpedia.jsonl"
        ]
    entityName2entity = {}

    with open('entity2id.json', 'r', encoding='utf-8') as f:
        entity2id = json.load(f)

    movieId2entity = {}
    for path in files:
        with open(path, 'r') as f:
            for line in f:
                dialog = json.loads(line)

                for message in dialog["messages"]:
                    movieIds = extract_movie_ids(message["text"])
                    for name, entity in zip(message['entity_name'], message['entity']):
                        entityName2entity[name] = entity
                    for movie_name, entity in zip(message['movie_name'], message['movie']):
                        entityName2entity[movie_name] = entity
                    for movieId, entity in zip(movieIds, message['movie']):
                        movieId2entity[movieId] = entity

    entityName2id = {k: entity2id[entityName2entity[k]] for k in entityName2entity}
    with open('entityName2id.json', 'w') as f:
        json.dump(entityName2id, f)

    movieId2id = {k: entity2id[movieId2entity[k]] for k in movieId2entity}
    with open('movieId2id.json', 'w') as f:
        json.dump(movieId2id, f)



if __name__ == "__main__":
    collect_entity_mappings()