[AIRFLOW-4760] Fix package DAGs disappearing from DagBag when reloaded

[danmactough]

Make sure you have checked all steps below.

Jira

• My PR addresses the following Airflow Jira issues and references them in the PR title. For example, "[AIRFLOW-XXX] My Airflow PR"
  • https://issues.apache.org/jira/browse/AIRFLOW-4760

Description

• Here are some details about my PR, including screenshots of any UI changes:

AIRFLOW-2900 introduced a change that broke Airflow whenever it tries to reprocess a packaged DAG. Since that change, the fileloc column in the database for packaged DAGs isn't an actual filepath. It looks something like: /usr/local/airflow/dags/package.zip/my_dag.py -- notice that the path to the DAG inside the zip is appended to the zip file's path. Then, when the DagBag#process_file method tries to load that filepath, it bails because that filepath is invalid.

This PR fixes that behavior by updating the DagBag#get_dag method such that when it fetches the DAG from the database and proceeds to reprocess the file, instead of using the orm_dag.fileloc property directly, we reuse the correct_maybe_zipped helper function to ensure that the passed value is an actual filepath.

Before

After

Tests

• My PR adds the following unit tests OR does not need testing for this extremely good reason:

  • add a sanity check test to ensure that the fileloc property does not change for packaged DAGs
  • add a test to verify that packaged DAGs can be refreshed
  • add a test that plain .py DAGs can also be refreshed (seems silly to have test coverage for only packaged DAGs 😄)

Commits

• My commits all reference Jira issues in their subject lines, and I have squashed multiple commits if they address the same issue. In addition, my commits follow the guidelines from "How to write a good git commit message":
  1. Subject is separated from body by a blank line
  2. Subject is limited to 50 characters (not including Jira issue reference)
  3. Subject does not end with a period
  4. Subject uses the imperative mood ("add", not "adding")
  5. Body wraps at 72 characters
  6. Body explains "what" and "why", not "how"

Documentation

• In case of new functionality, my PR adds documentation that describes how to use it.
  • All the public functions and the classes in the PR contain docstrings that explain what it does
  • If you implement backwards incompatible changes, please leave a note in the Updating.md so we can assign it to a appropriate release

Code Quality

• Passes flake8

[ashb]

Code looks good - though there is already a text/dags/test_zip.zip that we should use instead of creating a new example dag (those show up in every new install which we don't want here)

[danmactough]

Thanks for the lightning-fast review @ashb. ⚡️Updated and waiting for Travis...

[codecov-io]

Codecov Report

Merging #5404 into master will decrease coverage by 68.16%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #5404       +/-   ##
==========================================
- Coverage   78.97%   10.8%   -68.17%     
==========================================
  Files         483     483               
  Lines       30254   30254               
==========================================
- Hits        23893    3270    -20623     
- Misses       6361   26984    +20623

Impacted Files	Coverage Δ
airflow/models/dagbag.py	17.39% <ø> (-74.4%)	⬇️
airflow/models/dag.py	25.15% <ø> (-66.62%)	⬇️
...low/contrib/operators/wasb_delete_blob_operator.py	0% <0%> (-100%)	⬇️
airflow/example_dags/subdags/subdag.py	0% <0%> (-100%)	⬇️
airflow/contrib/sensors/emr_base_sensor.py	0% <0%> (-100%)	⬇️
airflow/contrib/operators/gcs_list_operator.py	0% <0%> (-100%)	⬇️
airflow/example_dags/example_subdag_operator.py	0% <0%> (-100%)	⬇️
airflow/contrib/operators/file_to_gcs.py	0% <0%> (-100%)	⬇️
airflow/contrib/sensors/cassandra_record_sensor.py	0% <0%> (-100%)	⬇️
airflow/www/blueprints.py	0% <0%> (-100%)	⬇️
... and 394 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f710a0d...8658ea6. Read the comment docs.

[danmactough]

Phew! 😓 Wow, @ashb, that change caused a totally unexpected (to me) failure in Travis. Very painful to track down, but it's ✅ now!