import json
import sys
from hashlib import sha256
from pathlib import Path

import pytest

from pingcap_datapipelines.cli import main, parse_tables
from pingcap_datapipelines.config import PipelineConfig
from pingcap_datapipelines.models import Column, Table


def test_parse_tables_rejects_an_explicit_empty_selection() -> None:
    with pytest.raises(ValueError, match="at least one exact table"):
        parse_tables(", ,")


def test_inspect_compact_prints_one_complete_json_line(monkeypatch, capsys) -> None:
    config = PipelineConfig(
        path=Path("source.yaml"),
        data={
            "source": {"name": "nexuscrm", "type": "tidbcloud", "database": "nexuscrm"},
            "target": {
                "project": "gcp-digital-bi",
                "location": "us-west2",
                "datasets": {"raw": "raw", "final": "final", "control": "control"},
            },
            "gcs": {
                "bucket": "bucket",
                "snapshot_prefix": "snapshots",
                "manifest_prefix": "manifests",
                "runtime_log_prefix": "logs",
            },
        },
    )
    tables = [
        Table(
            database="nexuscrm",
            name="Account",
            rows_estimate=1,
            columns=[Column("Id", "varchar", "VARCHAR(18)", False, 1)],
            primary_key=["Id"],
        ),
        Table(
            database="nexuscrm",
            name="_System_OutboxEvent",
            rows_estimate=2,
            columns=[Column("Id", "varchar", "VARCHAR(18)", False, 1)],
            primary_key=["Id"],
        ),
    ]
    monkeypatch.setattr("pingcap_datapipelines.cli.load_config", lambda _: config)
    monkeypatch.setattr("pingcap_datapipelines.cli.inspect_tables", lambda _: tables)
    monkeypatch.setattr(
        sys,
        "argv",
        ["pingcap-pipeline", "inspect", "--source", "source.yaml", "--compact"],
    )

    main()

    output = capsys.readouterr().out
    assert output.count("\n") == 1
    assert json.loads(output) == {
        "source": "nexuscrm",
        "source_database": "nexuscrm",
        "table_count": 2,
        "selection": {
            "mode": "configured",
            "requested_tables": None,
            "resolved_tables": ["Account", "_System_OutboxEvent"],
            "qualified_patterns": ["nexuscrm.Account", "nexuscrm._System_OutboxEvent"],
            "qualified_patterns_sha256": sha256(
                b"nexuscrm.Account\nnexuscrm._System_OutboxEvent"
            ).hexdigest(),
            "canonicalization": "utf8_newline_joined_sorted_qualified_patterns",
        },
        "tables": [
            {"name": "Account", "rows_estimate": 1, "primary_key": ["Id"], "watermark_columns": []},
            {
                "name": "_System_OutboxEvent",
                "rows_estimate": 2,
                "primary_key": ["Id"],
                "watermark_columns": [],
            },
        ],
    }


def test_inspect_explicit_tidb_subset_is_validated_and_hashed(monkeypatch, capsys) -> None:
    config = PipelineConfig(
        path=Path("source.yaml"),
        data={
            "source": {"name": "nexuscrm", "type": "tidbcloud", "database": "nexuscrm"},
            "target": {
                "project": "gcp-digital-bi",
                "location": "us-west2",
                "datasets": {"raw": "raw", "final": "final", "control": "control"},
            },
            "gcs": {
                "bucket": "bucket",
                "snapshot_prefix": "snapshots",
                "manifest_prefix": "manifests",
                "runtime_log_prefix": "logs",
            },
        },
    )
    selected = [
        Table("nexuscrm", "Account", 3, [], ["Id"]),
        Table("nexuscrm", "Order", 5, [], ["Id"]),
    ]
    calls: list[list[str]] = []
    monkeypatch.setattr("pingcap_datapipelines.cli.load_config", lambda _: config)
    monkeypatch.setattr(
        "pingcap_datapipelines.cli.inspect_exact_tidb_tables",
        lambda _config, tables: calls.append(tables) or selected,
    )
    monkeypatch.setattr(
        sys,
        "argv",
        [
            "pingcap-pipeline",
            "inspect",
            "--source",
            "source.yaml",
            "--tables",
            "Order,Account",
            "--compact",
        ],
    )

    main()

    payload = json.loads(capsys.readouterr().out)
    assert calls == [["Order", "Account"]]
    assert payload["table_count"] == 2
    assert payload["selection"]["mode"] == "explicit_subset"
    assert payload["selection"]["requested_tables"] == ["Order", "Account"]
    assert payload["selection"]["resolved_tables"] == ["Account", "Order"]
    assert payload["selection"]["qualified_patterns"] == ["nexuscrm.Account", "nexuscrm.Order"]


def test_load_passes_operator_bound_expected_tables(monkeypatch, capsys) -> None:
    config = PipelineConfig(
        path=Path("source.yaml"),
        data={
            "source": {"name": "nexuscrm", "type": "tidbcloud", "database": "nexuscrm"},
            "target": {
                "project": "gcp-digital-bi",
                "location": "us-west2",
                "datasets": {"raw": "raw", "final": "final", "control": "control"},
            },
            "gcs": {
                "bucket": "bucket",
                "snapshot_prefix": "snapshots",
                "manifest_prefix": "manifests",
                "runtime_log_prefix": "logs",
            },
        },
    )
    calls: list[dict[str, object]] = []
    monkeypatch.setattr("pingcap_datapipelines.cli.load_config", lambda _: config)
    monkeypatch.setattr(
        "pingcap_datapipelines.cli.run_load",
        lambda _config, **kwargs: calls.append(kwargs) or [{"status": "loaded"}],
    )
    monkeypatch.setattr(
        sys,
        "argv",
        [
            "pingcap-pipeline",
            "load",
            "--source",
            "source.yaml",
            "--manifest",
            "gs://bucket/manifest.json",
            "--tables",
            "Account,Order",
        ],
    )

    main()

    assert json.loads(capsys.readouterr().out) == [{"status": "loaded"}]
    assert calls == [
        {
            "manifest_uri": "gs://bucket/manifest.json",
            "latest": False,
            "all_unloaded": False,
            "tables": ["Account", "Order"],
        }
    ]