SNOW-1660544: The overwrite parameter does not behave as documented in write_pandas()

[chrisglass-slam]

Python version

Python 3.11.0rc1 (main, Apr 21 2024, 22:33:16) [GCC 11.4.0]

Operating system and processor architecture

Linux-6.5.0-1025-azure-x86_64-with-glibc2.35

Installed packages

snowflake==0.12.1
snowflake-connector-python==3.12.0
snowflake-snowpark-python==1.20.0
snowflake._legacy==0.11.0
snowflake.core==0.12.1

What did you do?

Using the write_pandas() function, I am passing "overwrite=True", but with "auto_create_table=False".

Observation:

The function performs a "CREATE TABLE IF NOT EXISTS". The table does exist, and the schema is maintained externally. This means the user I am using needs to be granted "CREATE TABLE" privileges on the table's schema.

Possible solution:

Switching the conditional on line 356 of pandas_tools.py as follows would probably solve the issue:


From `if auto_create_table or overwrite:` to `if auto_create_table and overwrite:`

What did you expect to see?

Expectation:

The function does not attempt to create a table (do not perform a "CREATE TABLE", with or without "IF NOT EXISTS").
The function performs a TRUNCATE of the table before loading the data with COPY INTO.

Can you set logging to DEBUG and collect the logs?

import logging
import os

for logger_name in ('snowflake.connector',):
    logger = logging.getLogger(logger_name)
    logger.setLevel(logging.DEBUG)
    ch = logging.StreamHandler()
    ch.setLevel(logging.DEBUG)
    ch.setFormatter(logging.Formatter('%(asctime)s - %(threadName)s %(filename)s:%(lineno)d - %(funcName)s() - %(levelname)s - %(message)s'))
    logger.addHandler(ch)

[sfc-gh-sghosh]

Hello @chrisglass-slam ,

Thanks for raising the issue, we are looking into it, will update.

Regards,
Sujan