[Bug] BigQuery.upsert fails depending on table's column order

[Jason94]

GoogleBigQuery.upsert() fails if the columns are ordered differently in the database's table than in the Parsons table.

If the mismatched columns are the same data type, it will silently insert the data into the wrong columns. If the mismatched columns are different data types, then the upsert will fail with an error.

To Reproduce

This script will throw an error when it attempts the upsert because columns a and b have different types:

from parsons import Table
from parsons.google.google_bigquery import GoogleBigQuery

def main():
    bq = GoogleBigQuery()
    table_name = "test.upsert_test_data"

    data = Table([{"a": 1, "b": "abc"}, {"a": 3, "b": "def"}])

    bq.copy(data, table_name, if_exists="drop")

    data2 = Table(
        [
            {"b": "xyz", "a": 1},
            {"b": "zyx", "a": 5},
        ]
    )

    bq.upsert(data2, table_name, "a")

    print(bq.query(f"SELECT * FROM {table_name}"))

if __name__ == "__main__":
    main()

Your Environment

• Version of Parsons used (if you know): main branch as of 3/21/2024
• Operating System and version (desktop or mobile): Windows 10