Harden artifact loading for downloads and cache validation

[mbsantiago]

Summary

• add retry, timeout, and exponential backoff support to load_artifact downloads to make network fetches more resilient
• add optional artifact integrity validation (expected_sha256, expected_size) for both newly downloaded files and cached files
• tighten URL detection to HTTP(S) URLs only and clean up partial or invalid files from cache automatically

Testing

• ruff check src/audioclass/utils.py tests/test_utils.py --select E,F,B,I
• pytest tests/test_utils.py (fails in this environment: missing optional dependency soundfile imported by tests/conftest.py)

[Copilot]

Pull request overview

This PR hardens audioclass.utils.load_artifact by improving URL detection, adding resilient HTTP download behavior (timeout + retry/backoff + streaming), and introducing optional integrity checks for downloaded and cached artifacts.

Changes:

• Restrict URL detection to valid HTTP(S) URLs and add a regression test for Windows-like paths.
• Add streaming downloads with configurable timeout, retries, and exponential backoff.
• Add optional cached/downloaded artifact validation via expected SHA256 and/or expected file size, and clean up invalid/partial cache files.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/audioclass/utils.py	Adds HTTP(S)-only URL detection, download retry/backoff + streaming, and artifact integrity validation with cache cleanup.
tests/test_utils.py	Adds coverage for URL detection, timeout/streaming usage, retry behavior, validation failures, and partial-file cleanup.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

_download_with_retry writes into the same fileobj across retry attempts, but on a requests.RequestException it does not rewind/truncate the file before retrying. If a network error happens mid-stream (e.g., ChunkedEncodingError from iter_content), the next attempt will append to the partial content, producing a corrupted artifact even if the later attempt succeeds. Reset/truncate the file between attempts (or download into a temp file per attempt and only move into place after a successful full download).

Suggested change

	fileobj.seek(0)
	fileobj.truncate()

[Copilot]

basename = Path(urlparse(path).path).name can be empty for valid HTTP(S) URLs that end with / (or otherwise have no filename). In that case new_path becomes the cache directory itself and new_path.open('wb') will fail with IsADirectoryError. Consider explicitly handling the empty-basename case (e.g., derive a filename from the URL hash or raise a clear ValueError).

[codecov]

Codecov Report

❌ Patch coverage is 79.03226% with 13 lines in your changes missing coverage. Please review.
✅ Project coverage is 89.60%. Comparing base (39abc8e) to head (60adefa).

Files with missing lines	Patch %	Lines
src/audioclass/utils.py	79.03%	4 Missing and 9 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #10      +/-   ##
==========================================
- Coverage   90.81%   89.60%   -1.22%     
==========================================
  Files          17       17              
  Lines         599      654      +55     
  Branches       61       74      +13     
==========================================
+ Hits          544      586      +42     
- Misses         29       33       +4     
- Partials       26       35       +9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.