BigQueryのテーブルのサイズ一覧を出すスクリプト

import re

from google.cloud import bigquery
from google.cloud.bigquery.table import Table

MATCH_ALL = r'.*'
MATCH_NONE = r'^$'

client = bigquery.Client(project="your-project")


def list_table_id(dataset_id, include_pattern=MATCH_ALL, exclude_pattern=MATCH_NONE):
    dataset_ref = client.dataset(dataset_id)
    tables = list(client.list_tables(dataset_ref))
    ret = []
    for table in tables:
        if re.search(include_pattern, table.table_id) and not re.search(exclude_pattern, table.table_id):
            ret.append(table.table_id)
    return ret


if __name__ == "__main__":
    datasets = list(client.list_datasets())
    for dataset in datasets:
        dataset_id = dataset.dataset_id
        try:
            for table_id in list_table_id(dataset_id):
                try:
                    dataset_ref = client.dataset(dataset_id)
                    table_ref = dataset_ref.table(table_id)
                    table: Table = client.get_table(table_ref)
                    print(f"{dataset_id}\t{table_id}\t{table.num_bytes}")
                except Exception as e:
                    print(f"{dataset_id}\t{table_id}\terror\t{e}")
        except Exception as e:
            print(f"{dataset_id}\terror\t{e}")