enable smooth environment variable naming transition by MoralCode · Pull Request #311 · chaoss/CollectOSS

MoralCode · 2026-05-08T20:52:34Z

Description
This PR introduces a new SystemEnv class that lives alongside SystemConfig and aims to centralize/unify all environment variable access.

The primary purpose of this is to slightly decouple the requested environment variable name from the name where the value is actually pulled from. For example, even if all in-code references use COLLECTOSS_DB for the database connection string, this layer will allow a values to be read from AUGUR_DB as a fallback if COLLECTOSS_DB (preferred) is not set.

This class is also set up to print a deprecation warning/notice whenever the AUGUR variant of an environment variable is used as a fallback. The message makes it clear that this fallback behavior is only temporary and may not be around forever (although I don't see this code being removed anytime in the foreseeable future)

This PR includes a unit tests for this new functionality and works as intended, including edge cases such as when the key the code requests doesn't use a known prefix (os.getenv("CONFIG_DATADIR"))

It also includes a special case function get_bool() because there were ~3 different places in the code where a small python one-liner (that i wrote) had been copied around for convenience, and this was a good opportunity to refactor that to prevent those three versions from ever drifting in implementation from each other. There are certainly more places this logic can be reused as well as we unify the way all env vars are processed.

Environment variables handled via Click (the CLI library) are addressed too. the click automatic prefix has been updated and the ~5 lines in 1 file where we were explicitly setting the variables are now updated to include both variants.

This also addresses the init, config, and entrypoint bash scripts uses at startup by replacing them with python equivalents (and enabling them to utilize the new environment variable switcher as well). This effectively means the earliest phases of the startup sequence in CollectOSS have been cleaned up somewhat

This PR resolves one of the large items that needs to be done to complete #279 for the next release.

Notes for Reviewers
This change renames all code references to use the new COLLECTOSS_ prefix, with the SystemEnv adapting these accesses to be able to read from AUGUR_ variables if COLLECTOSS_ ones aren't present.

This PR works in the sense that CollectOSS is able to start up with this change present with the default env

Signed commits

Yes, I signed my commits.

github-actions · 2026-05-08T20:53:14Z

@@ -130,7 +131,7 @@ def start(ctx, disable_collection, development, pidfile, port):
    processes = start_celery_worker_processes((core_worker_count, secondary_worker_count, facade_worker_count), disable_collection)


[pylint] _{reported by reviewdog 🐶}
W0621: Redefining name 'processes' from outer scope (line 396) (redefined-outer-name)

github-actions · 2026-05-08T20:53:14Z

@@ -15,6 +15,7 @@
 import requests
 from redis.exceptions import ConnectionError as RedisConnectionError


[pylint] _{reported by reviewdog 🐶}
W0611: Unused ConnectionError imported from redis.exceptions as RedisConnectionError (unused-import)

github-actions · 2026-05-08T20:53:14Z

@@ -31,7 +32,7 @@

 from keyman.KeyClient import KeyClient, KeyPublisher


[pylint] _{reported by reviewdog 🐶}
W0611: Unused KeyClient imported from keyman.KeyClient (unused-import)

shlokgilda

Added some inline comments