refactor: support agg_ops.DenseRankOp and RankOp for sqlglot compiler

Author: chelsea-lin

bigframes/core/compile/sqlglot/aggregations/unary_compiler.py

@@ -47,6 +47,15 @@ def _(
     return apply_window_if_present(sge.func("COUNT", column.expr), window)
 
 
+@UNARY_OP_REGISTRATION.register(agg_ops.DenseRankOp)
+def _(
+    op: agg_ops.DenseRankOp,
+    column: typed_expr.TypedExpr,
+    window: typing.Optional[window_spec.WindowSpec] = None,
+) -> sge.Expression:
+    return apply_window_if_present(sge.func("DENSE_RANK"), window)
+
+
 @UNARY_OP_REGISTRATION.register(agg_ops.MaxOp)
 def _(
     op: agg_ops.MaxOp,

bigframes/operations/aggregations.py

@@ -519,6 +519,8 @@ def implicitly_inherits_order(self):
 
 @dataclasses.dataclass(frozen=True)
 class DenseRankOp(UnaryWindowOp):
+    name: ClassVar[str] = "dense_rank"
+
     @property
     def skips_nulls(self):
         return False

tests/unit/core/compile/sqlglot/aggregations/snapshots/test_unary_compiler/test_dense_rank/out.sql

@@ -0,0 +1,19 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `int64_col` AS `bfcol_0`,
+    `rowindex` AS `bfcol_1`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    DENSE_RANK() OVER (
+      ORDER BY `bfcol_0` IS NULL ASC NULLS LAST, `bfcol_0` ASC NULLS LAST
+      RANGE BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING
+    ) AS `bfcol_4`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `bfuid_col_1`,
+  `bfcol_0` AS `int64_col`,
+  `bfcol_4` AS `agg_int64`
+FROM `bfcte_1`

tests/unit/core/compile/sqlglot/aggregations/test_unary_compiler.py

@@ -17,7 +17,14 @@
 import pytest
 
 from bigframes.core import agg_expressions as agg_exprs
-from bigframes.core import array_value, identifiers, nodes
+from bigframes.core import (
+    array_value,
+    expression,
+    identifiers,
+    nodes,
+    ordering,
+    window_spec,
+)
 from bigframes.operations import aggregations as agg_ops
 import bigframes.pandas as bpd
 
@@ -38,6 +45,24 @@ def _apply_unary_agg_ops(
     return sql
 
 
+def _apply_unary_window_op(
+    obj: bpd.DataFrame,
+    op: agg_exprs.UnaryAggregation,
+    window_spec: window_spec.WindowSpec,
+    new_name: str,
+) -> str:
+    win_node = nodes.WindowOpNode(
+        obj._block.expr.node,
+        expression=op,
+        window_spec=window_spec,
+        output_name=identifiers.ColumnId(new_name),
+    )
+    result = array_value.ArrayValue(win_node)
+
+    sql = result.session._executor.to_sql(result, enable_cache=False)
+    return sql
+
+
 def test_count(scalar_types_df: bpd.DataFrame, snapshot):
     col_name = "int64_col"
     bf_df = scalar_types_df[[col_name]]
@@ -47,6 +72,18 @@ def test_count(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_dense_rank(scalar_types_df: bpd.DataFrame, snapshot):
+    col_name = "int64_col"
+    bf_df = scalar_types_df[[col_name]]
+    agg_expr = agg_exprs.UnaryAggregation(
+        agg_ops.DenseRankOp(), expression.deref("int64_col")
+    )
+    window = window_spec.WindowSpec(ordering=(ordering.ascending_over(col_name),))
+    sql = _apply_unary_window_op(bf_df, agg_expr, window, "agg_int64")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
 def test_max(scalar_types_df: bpd.DataFrame, snapshot):
     col_name = "int64_col"
     bf_df = scalar_types_df[[col_name]]