Simplify feed generation internals by splitting response building and shared feed flow

[Abhijeet17o]

Hi, I want to propose a backend refactor for feeds that keeps behavior the same but makes the code much easier to maintain.

Right now, too much feed logic is concentrated in one place and endpoint setup is repeated across views:

1. Large multi-format response logic lives in utils.py.
2. Similar request setup is repeated in feeds.py.
3. Parameter handling, queryset construction, and output formatting are tightly coupled in utils.py.

What I propose to implement:

1. Keep the existing feeds_response entrypoint, but split format-specific paths into internal handlers:
2. json handler
3. csv handler
4. txt handler
5. stix21 handler
6. Extract shared feed setup flow used by feed endpoints so common logic is centralized (request params normalization, defaults/prioritization, queryset prep).
7. Preserve current API contract and endpoint behavior, including pagination/auth/throttling behavior.

Target files:

1. utils.py
2. feeds.py

Validation:

1. Ensure existing tests still pass, especially test_feeds_view.py and test_feeds_advanced_view.py.
2. Add focused regression tests for format parity and dispatcher behavior if needed.

This is not a feature rewrite and not an architecture change, just a behavior-preserving internal simplification to reduce future risk and improve maintainability

[Abhijeet17o]

Hey @regulartim, let me know what you think about this. If it aligns then can I open a PR and start implementation on it!

[regulartim]

Hey @Abhijeet17o !

request setup is repeated in feeds.py, feeds.py, and feeds.py.
tightly coupled in utils.py, utils.py, and utils.py

I am sorry, but your text screams AI slop. Is this a hallucination artifact where a model was told to list files but just repeated the same name three times?

I already thought about an object oriented approach to simplify the feed generation, but it is a low-value and high-risk change that directly impacts public-facing code paths. I don't think I want to assign that task to a new contributor.

[Abhijeet17o]

Hi @regulartim,
I did use an AI tool to help format the issue description, and I must have missed the repeated filenames while posting. I should have checked it more carefully.
Regarding the analysis, I did indeed derived that directly by reading the code for the generation of the feed. The AI was simply assisting with the formatting of the issue description itself.
That said, I understand your concern about the risk of refactoring the code by a new contributor. But if you think it's worth giving me a shot then I konw I can deliver. But after all it's your decision, I'd be happy either way.