modernize and standardize: uv, Python 3.14, pathlib, pytest, pre-commit, canonical CI

[soustruh]

Summary

Brings component-onedrive up to current CF team standards. No runtime/behaviour changes — only tooling, build, and code-style modernization. Six focused commits, each independently reviewable.

Commits (in order)

#	Commit	What it does
1	apply ruff formatting baseline	Run ruff format and ruff check --fix (with extend-select=I,UP and line-length=120) once across the codebase. Establishes a clean baseline so subsequent commits show only their real intent.
2	migrate to uv with Python 3.14 📦	Replace pip + requirements.txt with uv + pyproject.toml on Python 3.14. Multi-stage Dockerfile (lean base/production, test adds dev deps + runs ruff check + pytest). Drop legacy/Windows/flake8 cruft (flake8.cfg, deploy.sh, docs/imgs/, Windows .ps1/.bat scripts, tests/__init__.py). Populate empty URL stub files. No build-essential — all deps are pure Python or ship wheels for python:3.14-slim.
3	convert os.path to pathlib	Mechanical: os.path.join → Path() /, os.path.exists → .exists(), os.path.basename → .name. os.sep usage in _split_path_mask deliberately untouched (would risk behaviour drift in cross-platform splitting logic).
4	migrate unittest to pytest	Drop unittest.TestCase and unittest.main() in favour of plain pytest functions. Keep from unittest import mock (still the standard library mock).
5	add pre-commit hooks	.pre-commit-config.yaml with ruff-format-check, ruff-check, ty-check, pytest as local hooks. ty is advisory here (; true so it never blocks) — appropriate for an existing component. Verified end-to-end with pre-commit run --all-files.
6	adopt canonical CI workflow	Replace ~230 lines of handcrafted push.yml with a 17-line call to the reusable keboola/component-ci/.github/workflows/component-pipeline.yml@master.

Testing

• docker build --target test + docker run → pytest passes locally
• pre-commit run --all-files → ruff-format-check, ruff-check, pytest all green; ty-check shows 4 advisory diagnostics (non-blocking, follow-up)
• Re-tested end-to-end in our testing project and at client's stack on the new image

[matyas-jirat-keboola]

@keboola-pr-reviewer review

[matyas-jirat-keboola]

@keboola-pr-reviewer review

[keboola-pr-reviewer]

Verdict: needs_human (risk 2/5) · profile component-factory

The vast majority of this PR is mechanical and safe: ruff formatting, uv/pyproject.toml migration, flake8/deploy.sh deletions, pathlib conversion, pytest migration, pre-commit setup, and canonical CI adoption all fall squarely within the "a…

Concerns:

• src/client/client.py: _get_request_tokens() and Authorization header construction are credential-handling paths (OAuth token refresh) — policy requires human review regardless of change size.
• src/component.py: _save_refresh_token_state / _save_to_state persist refresh tokens — same credential-handling escalation rule applies.

Suggested reviewers: @keboola/component-factory

[matyas-jirat-keboola]

Hmm škoda že to neschválil botík... To by chtělo poladit

[soustruh]

The vast majority of this PR is mechanical and safe: ruff formatting, uv/pyproject.toml migration, flake8/deploy.sh deletions, pathlib conversion, pytest migration, pre-commit setup, and canonical CI adoption all fall squarely within the "a…

Hmm škoda že to neschválil botík... To by chtělo poladit

Too bad the bot didn't even finish the sentence… 😏 😁