"""tools/sanitize.py: scrub a harvest dump before it becomes a fixture."""

from __future__ import annotations

import importlib.util
from pathlib import Path

_spec = importlib.util.spec_from_file_location(
    "sanitize", Path(__file__).parent.parent / "tools" / "sanitize.py"
)
sanitize = importlib.util.module_from_spec(_spec)
_spec.loader.exec_module(sanitize)


def _dump():
    return {
        "about": {"instanceUuid": "35bb7c82-8526-5aa7-a886-cd7f8e6be786"},
        "licenses": [{"licenseKey": "AAAAA-BBBBB-CCCCC-DDDDD-EEEEE", "editionKey": "esxBasic"}],
        "objects": {
            "HostSystem": [
                {
                    "_moref": "host-21",
                    "name": "esx01.corp.example.com",
                    "hardware.systemInfo": {"serialNumber": "CZJ1234567"},
                    "config.network.dnsConfig": {
                        "hostName": "esx01",
                        "domainName": "corp.example.com",
                        "address": ["10.1.2.3"],
                    },
                    "config.network.pnic": [{"mac": "3c:ec:ef:01:02:03"}],
                    "summary.managementServerIp": "0.0.0.0",
                }
            ],
            "VirtualMachine": [
                {
                    "_moref": "vm-7",
                    "name": "payroll-db",
                    "config.instanceUuid": "5003a1b2-0000-1111-2222-333344445555",
                    "config.hardware.device": [
                        {"backing": {"fileName": "[ds1] payroll-db/payroll-db.vmdk"}},
                        {"macAddress": "00:50:56:aa:bb:cc"},
                    ],
                    "guest.net": [{"ipAddress": ["10.1.2.50", "fe80::1"]}],
                }
            ],
        },
    }


def _text(data):
    import json

    return json.dumps(data)


class TestSanitize:
    def test_sensitive_values_are_gone_everywhere(self):
        clean = _text(sanitize.sanitize(_dump()))
        for secret in (
            "35bb7c82",
            "AAAAA-BBBBB",
            "esx01",
            "corp.example.com",
            "CZJ1234567",
            "10.1.2.3",
            "3c:ec:ef",
            "payroll-db",
            "5003a1b2",
            "00:50:56:aa:bb:cc",
            "10.1.2.50",
        ):
            assert secret not in clean, secret

    def test_structure_and_harmless_values_survive(self):
        clean = sanitize.sanitize(_dump())
        vm = clean["objects"]["VirtualMachine"][0]
        assert vm["_moref"] == "vm-7"
        assert clean["licenses"][0]["editionKey"] == "esxBasic"
        assert clean["objects"]["HostSystem"][0]["summary.managementServerIp"] == "0.0.0.0"
        path = vm["config.hardware.device"][0]["backing"]["fileName"]
        assert path.startswith("[ds1] ") and path.endswith(".vmdk")

    def test_placeholders_are_stable(self):
        """The same original value maps to the same placeholder, so references
        between objects (a VM's name inside its disk path) still line up."""
        clean = sanitize.sanitize(_dump())
        vm = clean["objects"]["VirtualMachine"][0]
        path = vm["config.hardware.device"][0]["backing"]["fileName"]
        assert path == f"[ds1] {vm['name']}/{vm['name']}.vmdk"

    def test_valid_shapes(self):
        clean = sanitize.sanitize(_dump())
        host = clean["objects"]["HostSystem"][0]
        assert host["config.network.pnic"][0]["mac"].count(":") == 5
        ip = clean["objects"]["VirtualMachine"][0]["guest.net"][0]["ipAddress"]
        assert ip[0].startswith("192.0.2.")
        assert ip[1].startswith("2001:db8::")