"""Unit tests for Setlive workflow tasks."""

from datetime import datetime
import re
from unittest.mock import MagicMock
from unittest.mock import patch

import pytest

from feed_ingestion.flows.setlive import config
from feed_ingestion.flows.setlive import tasks

_date = '2024-01-01'
_report_name = 'opt_ins'


class TestBootstrap(object):
    """Test bootstrap."""

    @pytest.fixture
    def expected_bootstrap_response(self):
        """Response for a bootstrap task."""
        _date_dt = datetime.strptime(_date, '%Y-%m-%d').date()
        filename = config.reports[_report_name]['filename_template'].format(
            date=_date_dt
        )
        file_pattern = rf'.*\/{filename}'
        return {
            'feed_name': '_'.join([config.feed_name, _report_name]),
            'date': _date,
            'report_name': _report_name,
            'secrets_path': config.secrets_path,
            'drop_bucket': config.drop_bucket,
            'drop_path': config.drop_path,
            's3_dir_path': f's3://{config.drop_bucket}/{config.drop_path}',
            'file_pattern': file_pattern,
            'snowflake_pattern': filename,
            'staging_raw_table': config.reports[_report_name][
                'staging_raw_table'],
            'staging_raw_table_errors': config.reports[_report_name][
                'staging_raw_table_errors'],
            'pii_columns': config.reports[_report_name]['pii_columns'],
        }

    @patch('feed_ingestion.flows.setlive.tasks.garcon_feed_status')
    def test_bootstrap(self, feed_status_mock, expected_bootstrap_response):
        """Test bootstrap task."""
        context = {
            'activity': MagicMock(),
            'date': _date,
            'report': _report_name,
            'reload': None,
        }

        result = tasks.bootstrap(**context)

        assert result == expected_bootstrap_response


def test_filename_template_matches_current_vendor_format():
    """The configured pattern should match the observed Setlive filename."""
    report_date = datetime.strptime('2026-05-05', '%Y-%m-%d').date()
    pattern = config.reports['opt_ins']['filename_template'].format(
        date=report_date
    )

    assert re.match(
        pattern,
        'artistGroupId-2026-05-05T12-00-00-000Z.csv',
    )
    assert not re.match(
        pattern,
        'artistGroupId-2026-05-06T12-00-00-000Z.csv',
    )


@patch('feed_ingestion.flows.setlive.tasks.SetliveSnowflakeExecutor')
def test_load_staging_raw_table_reports_typical(mock_executor):
    """Test typical staging raw load sequence."""
    activity = MagicMock()
    feed_name = 'setlive_opt_ins'
    date = '2024-01-01'
    report_name = 'opt_ins'
    s3_dir_path = 's3://bucket/path'
    staging_raw_table_name = 'staging_table'
    staging_raw_errors_table_name = 'staging_table_errors'
    file_names = {'files': ['file1.csv', 'file2.csv']}
    snowflake_pattern = '.*-2024-01-01T[^/]*\\.csv$'
    pii_columns = [
        'EMAIL',
        'FIRST_NAME',
        'LAST_NAME',
        'POSTAL_CODE',
        'PHONE_NUMBER',
    ]

    mock_exec = mock_executor.return_value.__enter__.return_value

    tasks.load_staging_raw_table_reports(
        activity,
        feed_name,
        date,
        report_name,
        s3_dir_path,
        staging_raw_table_name,
        staging_raw_errors_table_name,
        file_names,
        snowflake_pattern,
        pii_columns,
    )

    mock_exec.create_temp_staging_raw_table.assert_called_once_with(
        report_name=report_name,
        temp_staging_raw_table='temp_staging_table_20240101',
    )
    mock_exec.assign_masking_on_temp_staging_raw_table.assert_called_once_with(
        temp_staging_raw_table='temp_staging_table_20240101',
        pii_columns=pii_columns,
    )
    mock_exec.create_stage.assert_called_once_with(
        s3_dir_path=s3_dir_path,
        stage_name='stage_setlive_20240101',
    )
    mock_exec.load_temp_staging_raw_table.assert_called_once_with(
        temp_staging_raw_table='temp_staging_table_20240101',
        file_name='.*(file1\\.csv|file2\\.csv)$',
        stage_name='stage_setlive_20240101',
    )
    mock_exec.clean_staging_raw_table.assert_called_once_with(
        staging_raw_table=staging_raw_table_name,
        staging_raw_errors_table=staging_raw_errors_table_name,
        file_name='^(file1\\.csv|file2\\.csv)$',
        date=date,
    )
    mock_exec.load_staging_raw_table.assert_called_once_with(
        staging_raw_table=staging_raw_table_name,
        staging_raw_errors_table=staging_raw_errors_table_name,
        temp_staging_raw_table='temp_staging_table_20240101',
        report_name=report_name,
        date=date,
        file_name='^(file1\\.csv|file2\\.csv)$',
    )
    mock_exec.drop_table.assert_called_once_with(
        table='temp_staging_table_20240101',
    )
    mock_exec.drop_stage.assert_called_once_with(
        stage_name='stage_setlive_20240101',
    )


@patch('feed_ingestion.flows.setlive.tasks.task_status')
@patch('feed_ingestion.flows.setlive.tasks.common_check_files_on_s3')
def test_mark_ingested_files(mock_check_files_on_s3, mock_task_status):
    """Test that matched files are persisted in task status."""
    mock_check_files_on_s3.return_value = {
        'source_files_dict': {
            'files': ['artistGroupId-2026-05-05T12-00-00-000Z.csv']
        }
    }

    tasks.mark_ingested_files(
        activity=MagicMock(),
        feed_name='setlive_opt_ins',
        date='2026-05-05',
        s3_download_path='s3://prod-orcdbucket/feed-drop/SetLive/',
        file_pattern=r'.*\/.*-2026-05-05T[^/]*\.csv$',
    )

    mock_task_status.set_values.assert_any_call(
        'setlive_opt_ins',
        '2026-05-05',
        'ingested_files_status',
        ['artistGroupId-2026-05-05T12-00-00-000Z.csv'],
    )
    mock_task_status.set_values.assert_any_call(
        'setlive_opt_ins',
        '2026-05-05',
        'ingested_files',
        ['artistGroupId-2026-05-05T12-00-00-000Z.csv'],
    )


@patch('feed_ingestion.flows.setlive.tasks.common_check_files_on_s3')
def test_check_files_on_s3_returns_stop_payload(mock_check_files_on_s3):
    """Test stop responses are passed through unchanged."""
    mock_check_files_on_s3.return_value = {'stop': True}

    result = tasks.check_files_on_s3(
        activity=MagicMock(),
        feed_name='setlive_opt_ins',
        date='2026-05-05',
        s3_download_path='s3://prod-orcdbucket/feed-drop/SetLive/',
        file_pattern=r'.*\/.*-2026-05-05T[^/]*\.csv$',
        list_output=True,
    )

    assert result == {'stop': True}


@patch('feed_ingestion.flows.setlive.tasks.task_status')
@patch('feed_ingestion.flows.setlive.tasks.common_check_files_on_s3')
def test_check_files_on_s3_returns_only_new_files(
    mock_check_files_on_s3,
    mock_task_status,
):
    """Only newly discovered files should be passed to the load step."""
    mock_check_files_on_s3.return_value = {
        'source_files_dict': {
            'files': [
                'existing-2026-05-09T19-37-34-434Z.csv',
                'new-2026-05-09T19-37-34-434Z.csv',
            ]
        }
    }
    mock_task_status.get_values.return_value = [
        'existing-2026-05-09T19-37-34-434Z.csv'
    ]

    result = tasks.check_files_on_s3(
        activity=MagicMock(),
        feed_name='setlive_opt_ins',
        date='2026-05-09',
        s3_download_path='s3://prod-orcdbucket/feed-drop/SetLive/',
        file_pattern=r'.*\/.*-2026-05-09T[^/]*\.csv$',
        list_output=True,
    )

    assert result == {
        'source_files_dict': {
            'files': ['new-2026-05-09T19-37-34-434Z.csv']
        }
    }