timedeltas.py

# Copyright 2025 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from __future__ import annotations

import dataclasses
import functools
import typing

from bigframes import dtypes
from bigframes import operations as ops
from bigframes.core import agg_expressions as ex_types
from bigframes.core import expression as ex
from bigframes.core import nodes, schema, utils
from bigframes.operations import aggregations as aggs


@dataclasses.dataclass
class _TypedExpr:
    expr: ex.Expression
    dtype: dtypes.Dtype

    @classmethod
    def create_op_expr(
        cls, op: typing.Union[ops.ScalarOp, ops.RowOp], *inputs: _TypedExpr
    ) -> _TypedExpr:
        expr = op.as_expr(*tuple(x.expr for x in inputs))  # type: ignore
        dtype = op.output_type(*tuple(x.dtype for x in inputs))
        return cls(expr, dtype)


def rewrite_timedelta_expressions(root: nodes.BigFrameNode) -> nodes.BigFrameNode:
    """
    Rewrites expressions to properly handle timedelta values, because this type does not exist
    in the SQL world.
    """
    if isinstance(root, nodes.ProjectionNode):
        updated_assignments = tuple(
            (_rewrite_expressions(expr, root.schema).expr, column_id)
            for expr, column_id in root.assignments
        )
        return nodes.ProjectionNode(root.child, updated_assignments)

    if isinstance(root, nodes.FilterNode):
        return nodes.FilterNode(
            root.child, _rewrite_expressions(root.predicate, root.schema).expr
        )

    if isinstance(root, nodes.OrderByNode):
        by = tuple(_rewrite_ordering_expr(x, root.schema) for x in root.by)
        return nodes.OrderByNode(root.child, by)

    if isinstance(root, nodes.WindowOpNode):
        return nodes.WindowOpNode(
            root.child,
            _rewrite_aggregation(root.expression, root.schema),
            root.window_spec,
            root.output_name,
            root.skip_reproject_unsafe,
        )

    if isinstance(root, nodes.AggregateNode):
        updated_aggregations = tuple(
            (_rewrite_aggregation(agg, root.child.schema), col_id)
            for agg, col_id in root.aggregations
        )
        return nodes.AggregateNode(
            root.child,
            updated_aggregations,
            root.by_column_ids,
            root.order_by,
            root.dropna,
        )

    return root


def _rewrite_ordering_expr(
    expr: nodes.OrderingExpression, schema: schema.ArraySchema
) -> nodes.OrderingExpression:
    by = _rewrite_expressions(expr.scalar_expression, schema).expr
    return nodes.OrderingExpression(by, expr.direction, expr.na_last)


@functools.cache
def _rewrite_expressions(expr: ex.Expression, schema: schema.ArraySchema) -> _TypedExpr:
    if isinstance(expr, ex.DerefOp):
        return _TypedExpr(expr, schema.get_type(expr.id.sql))

    if isinstance(expr, ex.ScalarConstantExpression):
        return _rewrite_scalar_constant_expr(expr)

    if isinstance(expr, ex.OpExpression):
        updated_inputs = tuple(
            map(lambda x: _rewrite_expressions(x, schema), expr.inputs)
        )
        return _rewrite_op_expr(expr, updated_inputs)

    raise AssertionError(f"Unexpected expression type: {type(expr)}")


def _rewrite_scalar_constant_expr(expr: ex.ScalarConstantExpression) -> _TypedExpr:
    if expr.value is None:
        return _TypedExpr(ex.const(None, expr.dtype), expr.dtype)
    if expr.dtype == dtypes.TIMEDELTA_DTYPE:
        int_repr = utils.timedelta_to_micros(expr.value)  # type: ignore
        return _TypedExpr(ex.const(int_repr, expr.dtype), expr.dtype)

    return _TypedExpr(expr, expr.dtype)


def _rewrite_op_expr(
    expr: ex.OpExpression, inputs: typing.Tuple[_TypedExpr, ...]
) -> _TypedExpr:
    if isinstance(expr.op, ops.SubOp):
        return _rewrite_sub_op(inputs[0], inputs[1])

    if isinstance(expr.op, ops.AddOp):
        return _rewrite_add_op(inputs[0], inputs[1])

    if isinstance(expr.op, ops.MulOp):
        return _rewrite_mul_op(inputs[0], inputs[1])

    if isinstance(expr.op, ops.DivOp):
        return _rewrite_div_op(inputs[0], inputs[1])

    if isinstance(expr.op, ops.FloorDivOp):
        # We need to re-write floor div because for numerics: int // float => float
        # but for timedeltas: int(timedelta) // float => int(timedelta)
        return _rewrite_floordiv_op(inputs[0], inputs[1])

    if isinstance(expr.op, ops.ToTimedeltaOp):
        return _rewrite_to_timedelta_op(expr.op, inputs[0])

    return _TypedExpr.create_op_expr(expr.op, *inputs)


def _rewrite_sub_op(left: _TypedExpr, right: _TypedExpr) -> _TypedExpr:
    if dtypes.is_datetime_like(left.dtype) and dtypes.is_datetime_like(right.dtype):
        return _TypedExpr.create_op_expr(ops.timestamp_diff_op, left, right)

    if dtypes.is_datetime_like(left.dtype) and right.dtype == dtypes.TIMEDELTA_DTYPE:
        return _TypedExpr.create_op_expr(ops.timestamp_sub_op, left, right)

    if left.dtype == dtypes.DATE_DTYPE and right.dtype == dtypes.DATE_DTYPE:
        return _TypedExpr.create_op_expr(ops.date_diff_op, left, right)

    if left.dtype == dtypes.DATE_DTYPE and right.dtype == dtypes.TIMEDELTA_DTYPE:
        return _TypedExpr.create_op_expr(ops.date_sub_op, left, right)

    return _TypedExpr.create_op_expr(ops.sub_op, left, right)


def _rewrite_add_op(left: _TypedExpr, right: _TypedExpr) -> _TypedExpr:
    if dtypes.is_datetime_like(left.dtype) and right.dtype == dtypes.TIMEDELTA_DTYPE:
        return _TypedExpr.create_op_expr(ops.timestamp_add_op, left, right)

    if left.dtype == dtypes.TIMEDELTA_DTYPE and dtypes.is_datetime_like(right.dtype):
        # Re-arrange operands such that timestamp is always on the left and timedelta is
        # always on the right.
        return _TypedExpr.create_op_expr(ops.timestamp_add_op, right, left)

    if left.dtype == dtypes.DATE_DTYPE and right.dtype == dtypes.TIMEDELTA_DTYPE:
        return _TypedExpr.create_op_expr(ops.date_add_op, left, right)

    if left.dtype == dtypes.TIMEDELTA_DTYPE and right.dtype == dtypes.DATE_DTYPE:
        # Re-arrange operands such that date is always on the left and timedelta is
        # always on the right.
        return _TypedExpr.create_op_expr(ops.date_add_op, right, left)

    return _TypedExpr.create_op_expr(ops.add_op, left, right)


def _rewrite_mul_op(left: _TypedExpr, right: _TypedExpr) -> _TypedExpr:
    result = _TypedExpr.create_op_expr(ops.mul_op, left, right)

    if left.dtype == dtypes.TIMEDELTA_DTYPE and dtypes.is_numeric(right.dtype):
        return _TypedExpr.create_op_expr(ops.timedelta_floor_op, result)
    if dtypes.is_numeric(left.dtype) and right.dtype == dtypes.TIMEDELTA_DTYPE:
        return _TypedExpr.create_op_expr(ops.timedelta_floor_op, result)

    return result


def _rewrite_div_op(left: _TypedExpr, right: _TypedExpr) -> _TypedExpr:
    result = _TypedExpr.create_op_expr(ops.div_op, left, right)

    if left.dtype == dtypes.TIMEDELTA_DTYPE and dtypes.is_numeric(right.dtype):
        return _TypedExpr.create_op_expr(ops.timedelta_floor_op, result)

    return result


def _rewrite_floordiv_op(left: _TypedExpr, right: _TypedExpr) -> _TypedExpr:
    result = _TypedExpr.create_op_expr(ops.floordiv_op, left, right)

    if left.dtype == dtypes.TIMEDELTA_DTYPE and dtypes.is_numeric(right.dtype):
        return _TypedExpr.create_op_expr(ops.timedelta_floor_op, result)

    return result


def _rewrite_to_timedelta_op(op: ops.ToTimedeltaOp, arg: _TypedExpr):
    if arg.dtype == dtypes.TIMEDELTA_DTYPE:
        # Do nothing for values that are already timedeltas
        return arg

    return _TypedExpr.create_op_expr(op, arg)


@functools.cache
def _rewrite_aggregation(
    aggregation: ex_types.Aggregation, schema: schema.ArraySchema
) -> ex_types.Aggregation:
    if not isinstance(aggregation, ex_types.UnaryAggregation):
        return aggregation

    if isinstance(aggregation.arg, ex.DerefOp):
        input_type = schema.get_type(aggregation.arg.id.sql)
    else:
        input_type = aggregation.arg.output_type

    if isinstance(aggregation.op, aggs.DiffOp):
        if dtypes.is_datetime_like(input_type):
            return ex_types.UnaryAggregation(
                aggs.TimeSeriesDiffOp(aggregation.op.periods), aggregation.arg
            )
        elif input_type == dtypes.DATE_DTYPE:
            return ex_types.UnaryAggregation(
                aggs.DateSeriesDiffOp(aggregation.op.periods), aggregation.arg
            )

    if isinstance(aggregation.op, aggs.StdOp) and input_type == dtypes.TIMEDELTA_DTYPE:
        return ex_types.UnaryAggregation(
            aggs.StdOp(should_floor_result=True), aggregation.arg
        )

    if isinstance(aggregation.op, aggs.MeanOp) and input_type == dtypes.TIMEDELTA_DTYPE:
        return ex_types.UnaryAggregation(
            aggs.MeanOp(should_floor_result=True), aggregation.arg
        )

    if (
        isinstance(aggregation.op, aggs.QuantileOp)
        and input_type == dtypes.TIMEDELTA_DTYPE
    ):
        return ex_types.UnaryAggregation(
            aggs.QuantileOp(q=aggregation.op.q, should_floor_result=True),
            aggregation.arg,
        )

    return aggregation