[bencehezso]

Summary

A developer tool renders a fixed matrix of fictional sample assessments (AWS, Azure and Terraform state; with and without egress; offline and uploaded; and edge cases) through the real report pipeline, offline and deterministically. The same scenario files run in CI against a test catalogue, so the existing guards cover every case.

Changes

Scenarios

• Twelve scenario files in tests/fixtures/samples/ and two fictional tfstate fixtures; no real names, accounts, tenants or subscriptions.
• A loader that names the file and field of any error.

Tool (python -m tests.samples)

• Runs the real stages with the collectors replaced by scenario data, under a frozen clock: report.pdf, index.html, egress.pdf, egress.html, a text dump of each, an overview page, and optional contact sheets (PyMuPDF, optional).
• --case, --out, --catalogue, --sheets; never downloads a catalogue; exits non-zero when a case fails and still renders the rest.
• The list of files a case holds is in one place (EXPECTED_FILES).

Tests

• 46 tests: loader, a missing or inactive type fails only its case, the command, and per case the expected files, credential guard, term/wording/retired-term guards, fonts, notices, upload states, case-specific checks and determinism.
• html_text moves to tests/support/html.py, with a per-heading dump used for the HTML text files.

Repository

• .gitignore: /samples/ and the sample tfstate fixtures.

Compatibility

Config, exit codes, report output, the dataset and the API v2 definition copy are unchanged. No product code changes; the only files outside tests/ are .gitignore.