chore: implement ReplaceStrOp and RegexReplaceStrOp

Author: chelsea-lin

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

@@ -222,6 +222,7 @@ def _(op: ops.StrContainsOp, expr: TypedExpr) -> sge.Expression:
 def _(op: ops.StrRepeatOp, expr: TypedExpr) -> sge.Expression:
     return sge.Repeat(this=expr.expr, times=sge.convert(op.repeats))
 
+
 @UNARY_OP_REGISTRATION.register(ops.date_op)
 def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return sge.Date(this=expr.expr)
@@ -560,6 +561,18 @@ def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return sge.Extract(this=sge.Identifier(this="QUARTER"), expression=expr.expr)
 
 
+@UNARY_OP_REGISTRATION.register(ops.ReplaceStrOp)
+def _(op: ops.ReplaceStrOp, expr: TypedExpr) -> sge.Expression:
+    return sge.func("REPLACE", expr.expr, sge.convert(op.pat), sge.convert(op.repl))
+
+
+@UNARY_OP_REGISTRATION.register(ops.RegexReplaceStrOp)
+def _(op: ops.RegexReplaceStrOp, expr: TypedExpr) -> sge.Expression:
+    return sge.func(
+        "REGEXP_REPLACE", expr.expr, sge.convert(op.pat), sge.convert(op.repl)
+    )
+
+
 @UNARY_OP_REGISTRATION.register(ops.reverse_op)
 def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return sge.func("REVERSE", expr.expr)

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_regex_replace_str/out.sql

@@ -0,0 +1,13 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `string_col` AS `bfcol_0`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    REGEXP_REPLACE(`bfcol_0`, 'e', 'a') AS `bfcol_1`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `string_col`
+FROM `bfcte_1`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_replace_str/out.sql

@@ -0,0 +1,13 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `string_col` AS `bfcol_0`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    REPLACE(`bfcol_0`, 'e', 'a') AS `bfcol_1`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `string_col`
+FROM `bfcte_1`

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

@@ -431,6 +431,18 @@ def test_quarter(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_replace_str(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.ReplaceStrOp("e", "a"), "string_col")
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_regex_replace_str(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.RegexReplaceStrOp(r"e", "a"), "string_col")
+    snapshot.assert_match(sql, "out.sql")
+
+
 def test_reverse(scalar_types_df: bpd.DataFrame, snapshot):
     bf_df = scalar_types_df[["string_col"]]
     sql = _apply_unary_op(bf_df, ops.reverse_op, "string_col")