fastflowtransform.testing.registry

Runner

Bases: Protocol

Callable signature for a generic test runner.

Returns:

Name	Type	Description
ok	bool	Whether the test passed.
message	str | None	Optional human-friendly message (usually set on failure).
example_sql	str | None	Optional example SQL (shown in summary on failure).

Source code in src/fastflowtransform/testing/registry.py

class Runner(Protocol):
    """Callable signature for a generic test runner.

    Returns:
        ok (bool): Whether the test passed.
        message (str | None): Optional human-friendly message (usually set on failure).
        example_sql (str | None): Optional example SQL (shown in summary on failure).
    """

    __name__: str

    def __call__(
        self, con: Any, table: str, column: str | None, params: dict[str, Any]
    ) -> tuple[bool, str | None, str | None]: ...

DQParamsBase

Bases: BaseModel

Base for all dynamically created DQ params models. Forbids unknown keys by default.

Source code in src/fastflowtransform/testing/registry.py

class DQParamsBase(BaseModel):
    """
    Base for all dynamically created DQ params models.
    Forbids unknown keys by default.
    """

    model_config = ConfigDict(extra="forbid")

run_accepted_values

run_accepted_values(con, table, column, params)

Runner for testing.accepted_values.

Source code in src/fastflowtransform/testing/registry.py

def run_accepted_values(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.accepted_values."""
    values = params.get("values") or []
    where = params.get("where")

    if column is None:
        example = "-- accepted_values: column parameter is required"
        return False, "missing required parameter: column", example

    if not values:
        # No values configured -> we treat this as a no-op check.
        example = f"-- accepted_values: no values provided; check is skipped for {table}.{column}"
        return True, None, example

    in_list = testing.sql_list(values)
    example = (
        f"select distinct {column} from {table} "
        + f"where {column} is not null and {column} not in ({in_list})"
        + (f" and ({where})" if where else "")
        + " limit 5"
    )

    col = column
    try:
        testing.accepted_values(con, table, col, values=values, where=where)
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_greater_equal

run_greater_equal(con, table, column, params)

Runner for testing.greater_equal (column >= threshold).

Source code in src/fastflowtransform/testing/registry.py

def run_greater_equal(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.greater_equal (column >= threshold)."""
    threshold = float(params.get("threshold", 0.0))
    if column is None:
        example = f"select count(*) from {table} where <column> < {threshold}"
        return False, "missing required parameter: column", example

    example = f"select count(*) from {table} where {column} < {threshold}"
    col = column
    try:
        testing.greater_equal(con, table, col, threshold=threshold)
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_non_negative_sum

run_non_negative_sum(con, table, column, params)

Runner for testing.non_negative_sum.

Source code in src/fastflowtransform/testing/registry.py

def run_non_negative_sum(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.non_negative_sum."""
    if column is None:
        example = f"select coalesce(sum(<column>), 0) from {table}"
        return False, "missing required parameter: column", example

    example = f"select coalesce(sum({column}), 0) from {table}"
    col = column
    try:
        testing.non_negative_sum(con, table, col)
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_row_count_between

run_row_count_between(con, table, column, params)

Runner for testing.row_count_between.

Source code in src/fastflowtransform/testing/registry.py

def run_row_count_between(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.row_count_between."""
    min_rows = int(params.get("min_rows", 1))
    max_rows_param = params.get("max_rows")
    max_rows = int(max_rows_param) if max_rows_param is not None else None

    example = f"select count(*) from {table}"
    try:
        testing.row_count_between(con, table, min_rows=min_rows, max_rows=max_rows)
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_freshness

run_freshness(con, table, column, params)

Runner for testing.freshness (max timestamp delay in minutes).

Source code in src/fastflowtransform/testing/registry.py

def run_freshness(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.freshness (max timestamp delay in minutes)."""
    if column is None:
        example = (
            f"select date_part('epoch', now() - max(<ts_column>)) / 60.0 as delay_min from {table}"
        )
        return False, "missing required parameter: column (ts_col)", example

    max_delay_raw = params.get("max_delay_minutes")
    example = f"select date_part('epoch', now() - max({column})) / 60.0 as delay_min from {table}"

    if max_delay_raw is None:
        return False, "missing required parameter: max_delay_minutes", example

    try:
        max_delay_int = int(max_delay_raw)
    except (TypeError, ValueError):
        return (
            False,
            f"invalid max_delay_minutes (expected integer minutes, got {max_delay_raw!r})",
            example,
        )

    col = column
    try:
        testing.freshness(con, table, col, max_delay_minutes=max_delay_int)
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_reconcile_equal

run_reconcile_equal(con, table, column, params)

Runner for testing.reconcile_equal (left == right within tolerances).

Source code in src/fastflowtransform/testing/registry.py

def run_reconcile_equal(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.reconcile_equal (left == right within tolerances)."""
    left = params.get("left")
    right = params.get("right")
    abs_tol = params.get("abs_tolerance")
    rel_tol = params.get("rel_tolerance_pct")

    if not isinstance(left, dict) or not isinstance(right, dict):
        example = "-- reconcile_equal requires 'left' and 'right' dict parameters"
        return False, "missing or invalid 'left'/'right' parameters", example

    example = _example_scalar_side(left) + ";\n" + _example_scalar_side(right)

    try:
        testing.reconcile_equal(
            con,
            left=left,
            right=right,
            abs_tolerance=abs_tol,
            rel_tolerance_pct=rel_tol,
        )
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_reconcile_ratio_within

run_reconcile_ratio_within(con, table, column, params)

Runner for testing.reconcile_ratio_within (min_ratio <= L/R <= max_ratio).

Source code in src/fastflowtransform/testing/registry.py

def run_reconcile_ratio_within(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.reconcile_ratio_within (min_ratio <= L/R <= max_ratio)."""
    left = params.get("left")
    right = params.get("right")
    min_ratio = params.get("min_ratio")
    max_ratio = params.get("max_ratio")

    if not isinstance(left, dict) or not isinstance(right, dict):
        example = "-- reconcile_ratio_within requires 'left' and 'right' dict parameters"
        return False, "missing or invalid 'left'/'right' parameters", example

    if min_ratio is None or max_ratio is None:
        example = _example_scalar_side(left) + ";\n" + _example_scalar_side(right)
        return False, "missing required parameters: min_ratio / max_ratio", example

    example = _example_scalar_side(left) + ";\n" + _example_scalar_side(right)

    try:
        testing.reconcile_ratio_within(
            con,
            left=left,
            right=right,
            min_ratio=float(min_ratio),
            max_ratio=float(max_ratio),
        )
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_reconcile_diff_within

run_reconcile_diff_within(con, table, column, params)

Runner for testing.reconcile_diff_within (|L - R| <= max_abs_diff).

Source code in src/fastflowtransform/testing/registry.py

def run_reconcile_diff_within(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.reconcile_diff_within (|L - R| <= max_abs_diff)."""
    left = params.get("left")
    right = params.get("right")
    max_abs_diff = params.get("max_abs_diff")

    if not isinstance(left, dict) or not isinstance(right, dict):
        example = "-- reconcile_diff_within requires 'left' and 'right' dict parameters"
        return False, "missing or invalid 'left'/'right' parameters", example

    if max_abs_diff is None:
        example = _example_scalar_side(left) + ";\n" + _example_scalar_side(right)
        return False, "missing required parameter: max_abs_diff", example

    example = _example_scalar_side(left) + ";\n" + _example_scalar_side(right)

    try:
        testing.reconcile_diff_within(
            con,
            left=left,
            right=right,
            max_abs_diff=float(max_abs_diff),
        )
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_reconcile_coverage

run_reconcile_coverage(con, table, column, params)

Runner for testing.reconcile_coverage (anti-join count == 0).

Source code in src/fastflowtransform/testing/registry.py

def run_reconcile_coverage(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.reconcile_coverage (anti-join count == 0)."""
    source = params.get("source")
    target = params.get("target")
    source_where = params.get("source_where")
    target_where = params.get("target_where")

    if not isinstance(source, dict) or not isinstance(target, dict):
        example = "-- reconcile_coverage requires 'source' and 'target' dict parameters"
        return False, "missing or invalid 'source'/'target' parameters", example

    example = _example_coverage_sql(source, target, source_where, target_where)

    try:
        testing.reconcile_coverage(
            con,
            source=source,
            target=target,
            source_where=source_where,
            target_where=target_where,
        )
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

run_relationships

run_relationships(con, table, column, params)

Runner for testing.relationships (FK-style anti join).

Source code in src/fastflowtransform/testing/registry.py

def run_relationships(
    con: Any, table: str, column: str | None, params: dict[str, Any]
) -> tuple[bool, str | None, str | None]:
    """Runner for testing.relationships (FK-style anti join)."""
    field = params.get("field") or column
    to_table = params.get("_to_relation") or params.get("to")
    to_field = params.get("to_field") or "id"
    where = params.get("where")
    to_where = params.get("to_where")

    example = _example_relationship_sql(
        table, field or "<field>", to_table, to_field, where, to_where
    )

    if not field:
        return False, "missing required parameter: field (or column)", example
    if not to_table:
        return False, "missing required parameter: to", example

    try:
        testing.relationships(
            con,
            table=table,
            field=field,
            to_table=to_table,
            to_field=to_field,
            where=where,
            to_where=to_where,
        )
        return True, None, example
    except testing.TestFailure as e:
        return False, str(e), example

register_python_test

register_python_test(kind, runner, *, params_model=None, origin=None, overwrite=False)

Register a custom Python test.

Parameters:

Name	Type	Description	Default
kind	str	logical test type (e.g. "no_future_orders").	required
runner	Runner	callable implementing Runner.	required
params_model	type[BaseModel] | None	optional Pydantic model for the params dict.	None
origin	str | None	string used in error messages (e.g. module path).	None
overwrite	bool	if True, replace an existing test with the same kind.	False

Source code in src/fastflowtransform/testing/registry.py

def register_python_test(
    kind: str,
    runner: Runner,
    *,
    params_model: type[BaseModel] | None = None,
    origin: str | None = None,
    overwrite: bool = False,
) -> None:
    """
    Register a custom Python test.

    Args:
        kind: logical test type (e.g. "no_future_orders").
        runner: callable implementing Runner.
        params_model: optional Pydantic model for the params dict.
        origin: string used in error messages (e.g. module path).
        overwrite: if True, replace an existing test with the same kind.
    """
    if kind in TESTS and not overwrite:
        raise ValueError(
            f"Test type {kind!r} is already registered "
            f"(origin={TEST_ORIGINS.get(kind, '<builtin>')!r})"
        )

    if kind in TESTS and overwrite:
        logger.warning(
            "Overwriting DQ test %r (previous origin=%r, new origin=%r)",
            kind,
            TEST_ORIGINS.get(kind),
            origin,
        )

    TESTS[kind] = runner

    if params_model is not None:
        TEST_PARAM_MODELS[kind] = params_model
    # If overwriting and no new params_model is given, keep the old one if present.
    elif overwrite and kind in TEST_PARAM_MODELS:
        pass
    else:
        # Default to a generic params model if you have one, or leave it unset
        TEST_PARAM_MODELS.pop(kind, None)

    if origin is not None:
        TEST_ORIGINS[kind] = origin
    elif overwrite:
        # if overwriting without explicit origin, don't change existing origin
        pass

register_sql_test

register_sql_test(kind, path, *, params_model=None, overwrite=False)

Register a custom SQL-based test from a *.ff.sql file.

kind: logical test type (e.g. "no_future_orders"). path: filesystem path to the template. params_model: optional Pydantic model for params. overwrite: if True, allow overriding an existing test of the same kind.

Source code in src/fastflowtransform/testing/registry.py

def register_sql_test(
    kind: str,
    path: Path,
    *,
    params_model: type[BaseModel] | None = None,
    overwrite: bool = False,
) -> None:
    """
    Register a custom SQL-based test from a *.ff.sql file.

    kind: logical test type (e.g. "no_future_orders").
    path: filesystem path to the template.
    params_model: optional Pydantic model for params.
    overwrite: if True, allow overriding an existing test of the same kind.
    """
    origin = str(path)
    META_KEYS = {"type", "table", "column", "severity", "tags", "name"}

    def _runner(
        con: Any, table: str, column: str | None, params: dict[str, Any]
    ) -> tuple[bool, str | None, str | None]:
        # 1) Strip generic test metadata and validate params if a schema is provided
        raw_params: dict[str, Any] = dict(params or {})
        core_params: dict[str, Any] = {k: v for k, v in raw_params.items() if k not in META_KEYS}

        if params_model is not None:
            try:
                cfg = params_model.model_validate(core_params)
            except ValidationError as exc:
                err_msg = _format_param_validation_error(kind, origin, exc)
                raise testing.TestFailure(err_msg) from exc
            # Use normalized params (e.g. converted types, defaults)
            params_validated = cfg.model_dump(exclude_none=True)
        else:
            params_validated = core_params

        # 2) Render the SQL template with a stable context
        env = REGISTRY.get_env()
        raw = path.read_text(encoding="utf-8")
        tmpl = env.from_string(raw)

        ctx: dict[str, Any] = {
            "kind": kind,
            "table": table,
            "column": column,
            "params": params_validated,
            # always present, so templates can safely do `{% if where %}`
            "where": params_validated.get("where"),
        }

        try:
            sql = tmpl.render(**ctx)
        except Exception as exc:
            raise testing.TestFailure(
                f"[{kind}] Failed to render SQL template for {origin}: {exc}"
            ) from exc

        # 3) Execute the SQL: convention here is "fail if count(*) > 0"
        n = _scalar(con, sql)
        ok = int(n or 0) == 0
        msg: str | None = None if ok else f"{kind} failed: {n} offending row(s)"
        example_sql = sql
        return ok, msg, example_sql

    register_python_test(
        kind,
        _runner,
        params_model=params_model,
        origin=origin,
        overwrite=overwrite,
    )