fix: make SQLAlchemy import lazy and declare it under [sink] extra

[0x070907]

Summary

This PR makes the import lazy and declares sqlalchemy under a proper [sink] optional extra in pyproject.toml so dependency declarations match actual runtime expectations.

Why

The Flask integration imports _sink_mode from sqlalchemy.py at module scope, and sqlalchemy.py imports SQLAlchemy at the top level. Since sqlalchemy was only listed under [dev], importing AdiuvareMiddleware crashed on a plain install(pip install .) .

1. adiuvare/integrations/sqlalchemy.py - Removed top-level from sqlalchemy import event. Moved it inside attach_sink() as a lazy import pointing at pip install adiuvare[sink]
2. pyproject.toml - Added [sink] optional extra sqlalchemy so dependency declarations match runtime expectations.
3. tests/test_flask.py - Added test_flask_middleware_import_does_not_require_sqlalchemy which proves flask import works without sqlalchemy.
4. tests/test_sink.py - Added pytest.importorskip guard at top,so it gracefully skips when sqlalchemy isn't installed instead of crashing

Scope

• bug fix
• documentation
• tests
• TUI / CLI improvement
• framework integration
• signal change
• configuration or route-policy change

Verification

Commands run:

#1. verify plain install no longer crashes on import for the flask app to run
pip install .
python -m pytest tests/test_flask.py tests/test_sink.py  -v 
# test_sink.py SKIPPED (sqlalchemy not installed)

---

#2.  verify sink extra pulls 
pip install adiuvare[sink]
pytest  tests/test_sink.py  -v

---

Notes:

• I ran the relevant tests locally
• I updated docs if behavior changed
• I updated screenshots if the TUI changed materially

Signal-specific notes

If this PR adds or changes a signal or a meaningful detection pattern, fill this in.

Signal type:

• hard signal (trackA)
• soft signal (trackB)
• not applicable

Source, reproducer, or proof: N/A

Benign case checked: N/A

Performance note: N/A

TUI notes

If this PR changes the TUI, include before/after screenshots or a short capture.

• before/after screenshots attached
• not applicable

Local-only file check

• I did not accidentally stage local-only files such as adiuvare.yaml or local conftest.py changes

Related issue

Closes #67

[0-Shimanshu]

The issue itself is real, but this PR is now a bit out of date against main. The pyproject.toml package-data entry for state/schema.sql is already present there, so the remaining value here is really the MANIFEST.in addition rather than both changes together.

The verification note also needs to be tightened. Importing init_state_db does not actually exercise the schema file read, so it does not fully prove the packaging issue is fixed. A better check would be to call init_state_db(...) against a temporary database path after a normal install and confirm that it no longer raises FileNotFoundError.

[0x070907]

Hi @0-Shimanshu, I have added the comment noqa : E402 in test_sink.py so that module level import at the top is ignored for the 3 lines after pytest.importorskip, as it should be ignored if the app shouldn't crash when sqlalchemy isn't installed.
Is this test coverage sufficient, or would you like me to add anything else ?

[0-Shimanshu]

The main fix looks correct now. Making the SQLAlchemy import lazy inside attach_sink() solves the plain Flask import crash, and adding the [flask] extra is the right dependency story.

Before merge, please do one small cleanup pass:

• Since adiuvare.integrations.sqlalchemy no longer imports SQLAlchemy at module scope, _sink_mode can stay imported at module scope in flask.py. It does not need to be imported inside every request.
• Please clean up the # noqa: E402 comments in tests/test_sink.py so they use the normal format.

After that, this is good to merge.

[0-Shimanshu]

The lazy import direction is right, but I want to adjust the dependency boundary a bit before merging.

SQLAlchemy sink is not really a Flask-specific feature. Django, Flask, FastAPI, or even plain Python usage could all use the sink path. So I do not think SQLAlchemy should live under a [flask] extra.

Could you update this PR to use a dedicated sink extra instead?

Something like:

[project.optional-dependencies]
sink = [
  "sqlalchemy",
]

[0x070907]

@0-Shimanshu , sure I will make the necessary changes and update the branch and let you know

[0x070907]

@0-Shimanshu , I have made the necessary changes and the flask app doesn't crash anymore.

test_sink.py is skipped if sqlachemy doesn't exist.

---

After i run pip install .[sink],all the tests passed

Should I update the README,quickstart.md and installation.md with [sink] extra ?
I have resolved all the merge conflicts.Kindly check

[0-Shimanshu]

The sink dependency should be documented as a dedicated extra, not under Flask.

For local development, use:
pip install ".[sink]"

That works because . means the current Python project, and pip reads the package name from pyproject.toml.
For GitHub install, document it as:
pip install "adiuvare[sink] @ git+https://github.com/0-Shimanshu/ADIUVARE.git"

And later, once we publish to PyPI, users can use:
pip install "adiuvare[sink]"

Please update README / quickstart / installation docs with this wording.
Also do one small formatting cleanup before merge:
remove the extra trailing spaces / blank lines in flask.py, sqlalchemy.py, and test_flask.py.

[0x070907]

Sure, I'll update it

[0-Shimanshu]

@0x070907 update
??

[0x070907]

Hi @0-Shimanshu, apologies for the delayed response - My exams are going on.

I've now addressed all the feedback:

• Removed trailing spaces in tests/test_flask.py.
• Updated README.md, docs/installation.md, docs/quickstart.md, and docs/integrations/sqlalchemy.md to document the [sink] extra.

[0x070907]

Should I update django.md as well?

[0-Shimanshu]

Should I update django.md as well?

like what can u first tell me the changes.

[0x070907]

Should I update django.md as well?

like what can u first tell me the changes.

I will add a note about the [sink] extra to docs/integrations/django.md so Django users know how to install SQLAlchemy if they want to use attach_sink() alongside Django

[0-Shimanshu]

Should I update django.md as well?

like what can u first tell me the changes.

I will add a note about the [sink] extra to docs/integrations/django.md so Django users know how to install SQLAlchemy if they want to use attach_sink() alongside Django

i think that would be good go ahead

[0x070907]

Should I update django.md as well?

like what can u first tell me the changes.

I will add a note about the [sink] extra to docs/integrations/django.md so Django users know how to install SQLAlchemy if they want to use attach_sink() alongside Django

i think that would be good go ahead

Its done,please check

[0-Shimanshu]

@0x070907 dont force push it break stuff and u pushed i project.toml unecesseary file

[0x070907]

okay,I wont force push again.But pyproject.toml is necessary because sink extra is added

[0-Shimanshu]

okay,I wont force push again.But pyproject.toml is necessary because sink extra is added

oo ya right

[0x070907]

@0-Shimanshu I accidentally got a merge commit on this branch,and in order to remove the commit without affecting any files or the commit history,I had to force push and I confirm that this hasn't affected any file or the main branch