125 refactor logging to add log to stderr function

[leorudczenko]

This merge request adds the function log_to_console, which allows users to enable and modify the etlhelper logging handler, as specified by issue #125.

UPDATE:

To fix the issues described in the comments, 2 extra changes have been made:

• A default null handler has been assigned to etlhelper using logging.getLogger("etlhelper").handlers.clear()
• A new pytest fixture has been added to temporarily enable the logger for a test

Closes #125

[leorudczenko]

The GitHub workflow successfully ran the tests, however that does not include all of the tests. The test file test/integration/etl/test_etl_logging.py needs to be updated to reflect the log changes.

[leorudczenko]

The test file test/integration/etl/test_etl_logging.py has now been updated to reflect the log changes. I have run these tests locally and confirmed they work as expected.

[leorudczenko]

There is 1 more test that is currently failing in test/integration/etl/test_abort.py, test_abort_etlhelper_threads[do_executemany_etl-executemany]. Once this has been fixed then this will be ready to merge

[leorudczenko]

The error seems to come from the fact that many files are still initialising an etlhelper logger using logger = logging.getLogger("etlhelper"). But since the __init__ file no longer creates the logger by default, the calls to get the logger in other files are actually create a standard new logger.

When adding logging.getLogger("etlhelper").addHandler(logging.NullHandler()), to the etlhelper/__init__.py file, running the test test_abort_etlhelper_threads alone passes. However, when running all the tests together, test_abort_etlhelper_threads fails again.

[leorudczenko]

test_abort_etlhelper_threads was still failing because test/integration/etl/test_etl_logging.py enabled the logging for etlhelper, but test_abort_etlhelper_threads requires the logging to be disabled.

To fix this, I've created a pytest fixture which yields an enabled etlhelper logger, and then sets the handler back to NullHandler after completion.

With this change, all of the tests are now passing as expected.

[volcan01010]

Tested locally during pairing session - all good!