Timing YAML vs. JSON parsing of the ACL Anthology people database

yaml_vs_json_parsing.py

import msgspec
import timeit
import yaml
from yaml import CLoader

yaml_path = "../data/yaml/people.yaml"
json_path = "../data/yaml/people.json"


def load_yaml():
    with open(yaml_path, "r", encoding="utf-8") as f:
        data = yaml.load(f, Loader=CLoader)

    return data


def load_json():
    with open(json_path, "rb") as f:
        data = msgspec.json.decode(f.read())

    return data


# Make sure JSON file exists and is equivalent to the YAML file
with open(json_path, "wb") as f:
    f.write(msgspec.json.encode(load_yaml()))


yaml_time = timeit.timeit("load_yaml()", globals=globals(), number=20)
json_time = timeit.timeit("load_json()", globals=globals(), number=20)
print(f"YAML loading took {yaml_time:2.2f}s")
print(f"JSON loading took {json_time:2.2f}s")