"""Test the backfill CLI."""

import json
import re

import pytest

from tests.integration.utils import run_command


def test_backfill() -> None:
    """Test sync one command works.

    The files from `test/integration/data` have been uploaded to S3.
    """
    stdout, _ = run_command(
        [
            "poetry",
            "run",
            "pdpbackfill",
            "backfill",
            "process",
            "--bucket-name=qa-pdp-backfill",
            "--manifest-file=integration-test/manifest.json",
        ],
        expected_returncode=0,
    )

    # "integration-test/01_test_detach.csv" has 3 detach lines
    attach_matches = re.findall("OPERATION_DETACH", stdout)
    assert len(attach_matches) == 3

    # "integration-test/02_test_attach.csv" has 3 attach lines
    attach_matches = re.findall("OPERATION_ATTACH", stdout)
    assert len(attach_matches) == 3

    # There are 2 files in the backfill job.
    # - "integration-test/01_test_detach.csv"
    # - "integration-test/02_test_attach.csv"
    #
    # Both contain 1 unique identity id. Each file should have 1
    # `success`ful PDP attach-detach request and 0 `fail`.
    summary_matches = re.findall(r'"summary": ({"success": 1, "fail": 0})', stdout)
    assert len(summary_matches) == 2
    for match in summary_matches:
        assert json.loads(match) == {"success": 1, "fail": 0}


@pytest.mark.parametrize(
    "filename",
    [
        pytest.param(
            "bom_manifest.json",
            id="manifest is utf-8 with bom encoded, referencing a csv that is utf-8 with bom-encoded",
        ),
        pytest.param(
            "utf8_manifest_ref_bom_csv.json",
            id="manifest is utf-8 encoded, referencing a csv that is utf-8 with bom-encoded",
        ),
    ],
)
def test_backfill_bom(filename: str) -> None:
    """Test backfill command works with utf-8 with bom-encoded files.

    The files from `test/integration/data` have been uploaded to S3.
    """
    stdout, _ = run_command(
        [
            "poetry",
            "run",
            "pdpbackfill",
            "backfill",
            "process",
            "--bucket-name=qa-pdp-backfill",
            f"--manifest-file=integration-test/{filename}",
        ],
        expected_returncode=0,
    )

    # "integration-test/bom_supported.csv" has 3 attach lines
    # "integration-test/neo4j_export_style.csv" has 3 attach lines
    attach_matches = re.findall("OPERATION_ATTACH", stdout)
    assert len(attach_matches) == 6

    # There are 2 files in the backfill job.
    # - "integration-test/bom_supported.csv"
    # - "integration-test/neo4j_export_style.csv"
    #
    # It contains 2 unique identity ids and should have 2
    # `success`ful PDP attach-detach request and 0 `fail`.
    summary_matches = re.findall(r'"summary": ({"success": 1, "fail": 0})', stdout)
    assert len(summary_matches) == 2
    for match in summary_matches:
        assert json.loads(match) == {"success": 1, "fail": 0}


@pytest.mark.parametrize(
    "additional_args,expected_error_message",
    [
        pytest.param(
            [],
            "Missing option '--bucket-name'.",
            id="Missing both args",
        ),
        pytest.param(
            ["--bucket-name=qa-pdp-backfill"],
            "Missing option '--manifest-file'.",
            id="--manifest-file is missing",
        ),
        pytest.param(
            ["--manifest-file=integration-test/manifest.json"],
            "Missing option '--bucket-name'.",
            id="--bucket-name is missing",
        ),
    ],
)
def test_process_arguments(
    additional_args: list[str],
    expected_error_message: str,
) -> None:
    """Test sync all command requires environment argument."""
    command_args = [
        "poetry",
        "run",
        "pdpbackfill",
        "backfill",
        "process",
    ]
    command_args.extend(additional_args)
    stdout, _ = run_command(
        command_args,
        expected_returncode=2,
        description=json.dumps(f"Additional arguments: {additional_args}"),
    )

    assert expected_error_message in stdout