fIx failing masking tests for python < 3.10

[potiuk]

Seems that the number of times user is printed in stack trace depend on Python version. The fix in #27335 seems to only have worked for Python 3.10 with the 1.0.0 of exceptiongroup the stack trace has less stack levels.

---

^ Add meaningful description above

Read the Pull Request Guidelines for more information.
In case of fundamental code changes, an Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in a newsfragment file, named {pr_number}.significant.rst or {issue_number}.significant.rst, in newsfragments.

[potiuk]

Running for "full tests" and with change in setup.py to get latest version of exceptiongroup - apparently the test results sligthly differ for different Python version

[potiuk]

just the trino case (which means BTW it is not question of resources - just race)

[Taragolis]

Finally I could reproduce this error locally.

I do not know about actual nature of this error but what I found

• Error not happen in case of exceptiongroup==1.0.0rc9
• exceptiongroup monkey patch traceback.TracebackException on import as result tests not failed if no import exceptiongroup. However this test run in the end of all Test: Core
• In PyCharm Debugger with Python 3.9.9 caplog.text returns expected value. I thought Debugger also might monkey patch traceback.TracebackException

ERROR Err
Traceback (most recent call last):
  File ".../test_secrets_masker.py", line 165, in test_masking_in_implicit_context_exceptions
    raise RuntimeError(f"Cannot connect to user:{p}")
RuntimeError: Cannot connect to user:***

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File ".../test_secrets_masker.py", line 167, in test_masking_in_implicit_context_exceptions
    raise RuntimeError(f"Exception: {ex1}")
RuntimeError: Exception: Cannot connect to user:***

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File ".../test_secrets_masker.py", line 169, in test_masking_in_implicit_context_exceptions
    raise RuntimeError(f"Exception: {ex2}")
RuntimeError: Exception: Exception: Cannot connect to user:***

• In the other hand pytest get this value.

ERROR Err
Traceback (most recent call last):
  File ".../test_secrets_masker.py", line 167, in test_masking_in_implicit_context_exceptions
    raise RuntimeError(f"Exception: {ex1}")
RuntimeError: Exception: Cannot connect to user:***

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File ".../test_secrets_masker.py", line 169, in test_masking_in_implicit_context_exceptions
    raise RuntimeError(f"Exception: {ex2}")
RuntimeError: Exception: Exception: Cannot connect to user:***

• Set env var EXCEPTIONGROUP_NO_PATCH to 1 not help

• Add parameter --tb=native to pytest run not help

• exceptiongroup is dependency of cattrs

root@4b131dd73f72:/opt/airflow# pipdeptree --reverse --packages exceptiongroup --python `which python`
exceptiongroup==1.0.0rc9
  - cattrs==22.2.0 [requires: exceptiongroup]
    - apache-airflow==2.5.0.dev0 [requires: cattrs>=22.1.0]
    - looker-sdk==22.18.0 [requires: cattrs>=1.3]

[potiuk]

Yeah. it's weird but eventually i think the approach i used works fine. We really do not care how many details and levels are in the stack trace we just care that password is not leaked. And checking that there is no plain password and that there are some 'masked' passwords (and more thab one in this case) is 'good enough' I think

[potiuk]

I think what really happens here is that sometimes we have one less exception in the chain because some of the monkey patching is done earlier or later (and it does not really matter) - but this is not something we should care about in this test - we just care about masking implicit passwords in stack trace

[potiuk]

BTW. We got the main "green" again :) 🎉