"""The harvest sanitiser (tools/sanitize.py).

Fixtures are committed forever, so a miss here puts a real NAS's serial numbers
and addresses into git history. Worth testing even though it is a dev tool.
"""

from __future__ import annotations

import importlib.util
import pathlib

import pytest

_SPEC = importlib.util.spec_from_file_location(
    "qnap_sanitize", pathlib.Path(__file__).parent.parent / "tools" / "sanitize.py"
)
assert _SPEC and _SPEC.loader
sanitize = importlib.util.module_from_spec(_SPEC)
_SPEC.loader.exec_module(sanitize)


@pytest.fixture()
def scrubber():
    return sanitize.Scrubber()


class TestSerials:
    def test_labelled_serial_is_replaced(self, scrubber):
        out = scrubber.scrub("Serial Number: WD-WCC4N7RTKZ9P")
        assert "WD-WCC4N7RTKZ9P" not in out
        assert "SERIAL001XXXX" in out

    def test_lsblk_serial_field_is_replaced(self, scrubber):
        out = scrubber.scrub('NAME="sda" MODEL="WD40EFRX" SERIAL="WD-WCC4N7RTKZ9P"')
        assert "WD-WCC4N7RTKZ9P" not in out
        assert 'MODEL="WD40EFRX"' in out, "model names are not identifying and must survive"

    def test_same_serial_maps_to_the_same_placeholder(self, scrubber):
        """A disk serial appears in several harvested files; the cross-reference
        has to survive or the fixtures stop describing one coherent device."""
        first = scrubber.scrub("serial: ABC123456")
        second = scrubber.scrub('SERIAL="ABC123456"')
        assert "SERIAL001XXXX" in first
        assert "SERIAL001XXXX" in second

    def test_distinct_serials_get_distinct_placeholders(self, scrubber):
        out = scrubber.scrub("sn: AAA111222\nsn: BBB333444")
        assert "SERIAL001XXXX" in out
        assert "SERIAL002XXXX" in out


class TestNetworkIdentifiers:
    def test_mac_is_replaced(self, scrubber):
        out = scrubber.scrub("link/ether 24:5e:be:11:22:33 brd ff:ff:ff:ff:ff:ff")
        assert "24:5e:be:11:22:33" not in out
        assert "00:11:22:33:44:" in out

    def test_ipv4_is_replaced(self, scrubber):
        out = scrubber.scrub("inet 10.7.224.12/24")
        assert "10.7.224.12" not in out
        assert "192.0.2.1" in out

    def test_loopback_and_wildcard_survive(self, scrubber):
        """Replacing these makes fixtures unreadable and reveals nothing."""
        out = scrubber.scrub("0.0.0.0:445 127.0.0.1:22")
        assert "0.0.0.0" in out
        assert "127.0.0.1" in out

    def test_same_ip_maps_consistently(self, scrubber):
        out = scrubber.scrub("gw 10.0.0.1\nvia 10.0.0.1")
        assert out.count("192.0.2.1") == 2


class TestHostnames:
    def test_extra_host_is_replaced_case_insensitively(self):
        s = sanitize.Scrubber(["MyNAS"])
        out = s.scrub("Server Name = mynas\nhost MYNAS ok")
        assert "mynas" not in out.lower().replace("testnas", "")
        assert out.count("testnas") == 2

    def test_empty_host_entry_is_ignored(self):
        """An unset --extra-host must not turn into a regex that matches everything."""
        s = sanitize.Scrubber(["", None])
        assert s.scrub("untouched text") == "untouched text"