implement StartsWithOp

Author: chelsea-lin

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

@@ -14,6 +14,7 @@
 
 from __future__ import annotations
 
+import functools
 import typing
 
 import pandas as pd
@@ -633,6 +634,18 @@ def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     )
 
 
+@UNARY_OP_REGISTRATION.register(ops.StartsWithOp)
+def _(op: ops.StartsWithOp, expr: TypedExpr) -> sge.Expression:
+    if not op.pat:
+        return sge.false()
+
+    def to_startswith(pat: str) -> sge.Expression:
+        return sge.func("STARTS_WITH", expr.expr, sge.convert(pat))
+
+    conditions = [to_startswith(pat) for pat in op.pat]
+    return functools.reduce(lambda x, y: sge.Or(this=x, expression=y), conditions)
+
+
 @UNARY_OP_REGISTRATION.register(ops.StrStripOp)
 def _(op: ops.StrStripOp, expr: TypedExpr) -> sge.Expression:
     return sge.Trim(this=sge.convert(op.to_strip), expression=expr.expr)

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_startswith/multiple_patterns.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
+    *,
+    STARTS_WITH(`bfcol_0`, 'ab') OR STARTS_WITH(`bfcol_0`, 'cd') 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_startswith/no_pattern.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
+    *,
+    FALSE 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_startswith/single_pattern.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
+    *,
+    STARTS_WITH(`bfcol_0`, 'ab') 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

@@ -501,6 +501,18 @@ def test_sqrt(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_startswith(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.StartsWithOp(pat=("ab",)), "string_col")
+    snapshot.assert_match(sql, "single_pattern.sql")
+
+    sql = _apply_unary_op(bf_df, ops.StartsWithOp(pat=("ab", "cd")), "string_col")
+    snapshot.assert_match(sql, "multiple_patterns.sql")
+
+    sql = _apply_unary_op(bf_df, ops.StartsWithOp(pat=()), "string_col")
+    snapshot.assert_match(sql, "no_pattern.sql")
+
+
 def test_str_get(scalar_types_df: bpd.DataFrame, snapshot):
     bf_df = scalar_types_df[["string_col"]]
     sql = _apply_unary_op(bf_df, ops.StrGetOp(1), "string_col")