tests/test_validator_schema_consistency.py

"""Guard against CREATE TABLE / INSERT column mismatches — catch them in-memory,
at validate_bundle() time, instead of ~10 minutes later as a cryptic Snowflake
'invalid identifier' error deep inside an executemany (see the Gap e2e failure:
FACT_CAMPAIGN_PERFORMANCE's row loader referenced a CHECKOUT_INITIATED column
CREATE TABLE never declared).
"""

from __future__ import annotations

from demoprep_app.dataset.contracts import DatasetBundle, DatasetColumn, DatasetTable
from demoprep_app.dataset.validator import ValidationReport, _check_schema_consistency
from demoprep_app.scenario.contract import ScenarioContract


def _scenario() -> ScenarioContract:
    return ScenarioContract(
        company_name="Acme",
        company_url="https://example.com",
        use_case="Marketing Funnel",
        scenario_type="campaign_performance",
        fact_grain="campaign-day",
    )


def test_schema_consistency_passes_for_a_clean_bundle():
    bundle = DatasetBundle(
        scenario=_scenario(),
        tables=[
            DatasetTable(
                name="FACT_CAMPAIGN_PERFORMANCE",
                grain="campaign-day",
                columns=[
                    DatasetColumn("FACT_CAMPAIGN_PERFORMANCE_KEY", "fact_key", "NUMBER", nullable=False),
                    DatasetColumn("IMPRESSIONS", "impressions", "NUMBER", nullable=False),
                    DatasetColumn("CLICKS", "clicks", "NUMBER", nullable=False),
                ],
                rows=[{"FACT_CAMPAIGN_PERFORMANCE_KEY": 1, "IMPRESSIONS": 100, "CLICKS": 5}],
                is_fact=True,
            )
        ],
    )
    report = ValidationReport()

    _check_schema_consistency(bundle, report)

    assert report.integrity_errors == []


def test_schema_consistency_catches_a_row_key_with_no_declared_column():
    # Mirrors the Gap failure: the row loader would INSERT a column
    # (CHECKOUT_INITIATED) that CREATE TABLE never declared for this table.
    bundle = DatasetBundle(
        scenario=_scenario(),
        tables=[
            DatasetTable(
                name="FACT_CAMPAIGN_PERFORMANCE",
                grain="campaign-day",
                columns=[
                    DatasetColumn("FACT_CAMPAIGN_PERFORMANCE_KEY", "fact_key", "NUMBER", nullable=False),
                    DatasetColumn("ADD_TO_CART", "add_to_cart", "NUMBER", nullable=False),
                    DatasetColumn("CONVERSIONS", "conversions", "NUMBER", nullable=False),
                ],
                rows=[
                    {
                        "FACT_CAMPAIGN_PERFORMANCE_KEY": 1,
                        "ADD_TO_CART": 10,
                        "CONVERSIONS": 2,
                        "CHECKOUT_INITIATED": 6,  # not a declared column
                    }
                ],
                is_fact=True,
            )
        ],
    )
    report = ValidationReport()

    _check_schema_consistency(bundle, report)

    assert len(report.integrity_errors) == 1
    assert "FACT_CAMPAIGN_PERFORMANCE" in report.integrity_errors[0]
    assert "CHECKOUT_INITIATED" in report.integrity_errors[0]


def test_schema_consistency_catches_column_names_that_collide_after_normalization():
    # Two distinct Python-level names that normalize to the same SQL identifier
    # would make CREATE TABLE emit a duplicate column.
    bundle = DatasetBundle(
        scenario=_scenario(),
        tables=[
            DatasetTable(
                name="FACT_CAMPAIGN_PERFORMANCE",
                grain="campaign-day",
                columns=[
                    DatasetColumn("Checkout Initiated", "checkout_initiated", "NUMBER", nullable=False),
                    DatasetColumn("Checkout-Initiated", "checkout_initiated_2", "NUMBER", nullable=False),
                ],
                rows=[{"Checkout Initiated": 1, "Checkout-Initiated": 2}],
                is_fact=True,
            )
        ],
    )
    report = ValidationReport()

    _check_schema_consistency(bundle, report)

    assert len(report.integrity_errors) == 1
    assert "CHECKOUT_INITIATED" in report.integrity_errors[0]
    assert "duplicate column" in report.integrity_errors[0]


def test_schema_consistency_ignores_empty_tables():
    bundle = DatasetBundle(
        scenario=_scenario(),
        tables=[
            DatasetTable(
                name="EMPTY_DIM",
                grain="empty",
                columns=[DatasetColumn("EMPTY_DIM_KEY", "dimension_key", "NUMBER", nullable=False)],
                rows=[],
            )
        ],
    )
    report = ValidationReport()

    _check_schema_consistency(bundle, report)

    assert report.integrity_errors == []