feat(datasets): support Google Sheets URLs in dataset loader

[ykd007]

Closes #86

What

Adds transparent Google Sheets URL normalization to fetch_csv_content.

When a public Google Sheets share/edit link is passed as a dataset URL, it is automatically rewritten to the /export?format=csv form before fetching — no change required from callers.

How

• _normalize_google_sheets_url(url) — pure regex transform, handles /edit#gid=N, query-param gid, and passes through URLs that are already in export format
• fetch_csv_content calls the normalizer before httpx.get, with follow_redirects=True added for robustness
• import re moved to module level

Tests

5 unit tests added to test_data.py covering: passthrough (non-Sheets URL), edit+gid conversion, edit-no-gid conversion, already-export passthrough, pub-output-csv passthrough.

[Copilot]

Pull request overview

Adds transparent Google Sheets URL normalization to fetch_csv_content, so public Sheets share/edit links are automatically rewritten to the /export?format=csv form before fetching. This resolves issue #86 by letting datasets configured with Sheets URLs be loaded without manual conversion.

Changes:

• New _normalize_google_sheets_url helper that regex-matches Sheets URLs, preserves already-exported forms, and appends gid when present.
• fetch_csv_content now normalizes the URL and uses follow_redirects=True for robustness.
• 5 unit tests covering passthrough and conversion cases.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
agentic_security/probe_data/data.py	Adds _normalize_google_sheets_url and integrates it into fetch_csv_content; module-level re import.
agentic_security/probe_data/test_data.py	Adds TestNormalizeGoogleSheetsUrl covering passthrough, edit→export with/without gid, and export/pub passthrough.

---

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

[ykd007]

The 6 test failures in CI are pre-existing ModuleNotFoundError issues for anthropic and openai — unrelated to this PR. They appear on main as well (the maintainer has a fix(ci): commit in flight right now). Our TestNormalizeGoogleSheetsUrl tests all pass within the 346 passing tests. Pre-Commit checks are now green ✅

[msoedov]

@ykd007 thank you for the patch!