from pathlib import Path

from scripts.scrape_fsbp import parse_fsbp_index


_FIXTURES = Path(__file__).parent / "fixtures"


def test_parses_fsbp_index_rows():
    html = (_FIXTURES / "fsbp_index_sample.html").read_text()
    rows = parse_fsbp_index(html, base_url="https://docs.aws.amazon.com")
    by_id = {r.control_id: r for r in rows}
    assert "S3.5" in by_id
    assert by_id["S3.5"].title == "S3 general purpose buckets should require requests to use SSL"
    assert by_id["S3.5"].severity == "medium"
    assert by_id["S3.5"].detail_url.endswith("s3-controls.html#s3-5")
    assert "IAM.5" in by_id
    assert "Account.1" in by_id


def test_resource_types_attached_via_prefix_map():
    html = (_FIXTURES / "fsbp_index_sample.html").read_text()
    rows = parse_fsbp_index(html, base_url="https://docs.aws.amazon.com")
    by_id = {r.control_id: r for r in rows}
    assert "aws_s3_bucket" in by_id["S3.5"].resource_types
    assert "aws_iam_user" in by_id["IAM.5"].resource_types
    assert "aws_account_alternate_contact" in by_id["Account.1"].resource_types


def test_dedups_repeat_control_ids():
    """A control linked multiple times on the page should appear once."""
    html = """\
<html><body>
<p><a href="./s3-controls.html#s3-1">[S3.1] First link</a></p>
<p><a href="./s3-controls.html#s3-1">[S3.1] Second link to same control</a></p>
</body></html>"""
    rows = parse_fsbp_index(html, base_url="https://docs.aws.amazon.com")
    assert len([r for r in rows if r.control_id == "S3.1"]) == 1