fix(duckdb): generate SIMILAR TO instead of tilde to workaround sqlglot issue

Author: cpcloud

ibis/backends/duckdb/registry.py

@@ -313,6 +313,7 @@ def _array_filter(t, op):
         ops.RegexReplace: fixed_arity(
             lambda *args: sa.func.regexp_replace(*args, sa.text("'g'")), 3
         ),
+        ops.RegexSearch: fixed_arity(lambda x, y: x.op("SIMILAR TO")(y), 2),
         ops.StringContains: fixed_arity(sa.func.contains, 2),
         ops.ApproxMedian: reduction(
             # without inline text, duckdb fails with

ibis/backends/tests/snapshots/test_string/test_rlike/bigquery/out.sql

@@ -0,0 +1,5 @@
+SELECT
+  t0.*
+FROM `ibis-gbq.ibis_gbq_testing.functional_alltypes` AS t0
+WHERE
+  REGEXP_CONTAINS(t0.`string_col`, '0')

ibis/backends/tests/snapshots/test_string/test_rlike/clickhouse/out.sql

@@ -0,0 +1,5 @@
+SELECT
+  *
+FROM functional_alltypes AS t0
+WHERE
+  match(t0.string_col, '0')

ibis/backends/tests/snapshots/test_string/test_rlike/duckdb/out.sql

@@ -0,0 +1,19 @@
+SELECT
+  t0.index,
+  t0."Unnamed: 0",
+  t0.id,
+  t0.bool_col,
+  t0.tinyint_col,
+  t0.smallint_col,
+  t0.int_col,
+  t0.bigint_col,
+  t0.float_col,
+  t0.double_col,
+  t0.date_string_col,
+  t0.string_col,
+  t0.timestamp_col,
+  t0.year,
+  t0.month
+FROM functional_alltypes AS t0
+WHERE
+  t0.string_col SIMILAR TO '0'

ibis/backends/tests/snapshots/test_string/test_rlike/impala/out.sql

@@ -0,0 +1,5 @@
+SELECT
+  t0.*
+FROM ibis_testing.`functional_alltypes` AS t0
+WHERE
+  t0.`string_col` RLIKE '0'

ibis/backends/tests/snapshots/test_string/test_rlike/mysql/out.sql

@@ -0,0 +1,19 @@
+SELECT
+  t0.`index`,
+  t0.`Unnamed: 0`,
+  t0.id,
+  t0.bool_col = 1 AS bool_col,
+  t0.tinyint_col,
+  t0.smallint_col,
+  t0.int_col,
+  t0.bigint_col,
+  t0.float_col,
+  t0.double_col,
+  t0.date_string_col,
+  t0.string_col,
+  t0.timestamp_col,
+  t0.year,
+  t0.month
+FROM functional_alltypes AS t0
+WHERE
+  REGEXP_LIKE(t0.string_col, '0')

ibis/backends/tests/snapshots/test_string/test_rlike/postgres/out.sql

@@ -0,0 +1,19 @@
+SELECT
+  t0.index,
+  t0."Unnamed: 0",
+  t0.id,
+  t0.bool_col,
+  t0.tinyint_col,
+  t0.smallint_col,
+  t0.int_col,
+  t0.bigint_col,
+  t0.float_col,
+  t0.double_col,
+  t0.date_string_col,
+  t0.string_col,
+  t0.timestamp_col,
+  t0.year,
+  t0.month
+FROM functional_alltypes AS t0
+WHERE
+  t0.string_col ~ '0'

ibis/backends/tests/snapshots/test_string/test_rlike/snowflake/out.sql

@@ -0,0 +1,19 @@
+SELECT
+  t0."index",
+  t0."Unnamed: 0",
+  t0."id",
+  t0."bool_col",
+  t0."tinyint_col",
+  t0."smallint_col",
+  t0."int_col",
+  t0."bigint_col",
+  t0."float_col",
+  t0."double_col",
+  t0."date_string_col",
+  t0."string_col",
+  t0."timestamp_col",
+  t0."year",
+  t0."month"
+FROM ibis_testing.voltrondataphillip."FUNCTIONAL_ALLTYPES" AS t0
+WHERE
+  REGEXP_LIKE(t0."string_col", '0')

ibis/backends/tests/snapshots/test_string/test_rlike/sqlite/out.sql

@@ -0,0 +1,19 @@
+SELECT
+  t0."index",
+  t0."Unnamed: 0",
+  t0.id,
+  t0.bool_col,
+  t0.tinyint_col,
+  t0.smallint_col,
+  t0.int_col,
+  t0.bigint_col,
+  t0.float_col,
+  t0.double_col,
+  t0.date_string_col,
+  t0.string_col,
+  STRFTIME('%Y-%m-%d %H:%M:%f', t0.timestamp_col) AS timestamp_col,
+  t0.year,
+  t0.month
+FROM main.functional_alltypes AS t0
+WHERE
+  _IBIS_SQLITE_REGEX_SEARCH(t0.string_col, '0')

ibis/backends/tests/snapshots/test_string/test_rlike/trino/out.sql

@@ -0,0 +1,19 @@
+SELECT
+  t0.index,
+  t0."unnamed: 0",
+  t0.id,
+  t0.bool_col,
+  t0.tinyint_col,
+  t0.smallint_col,
+  t0.int_col,
+  t0.bigint_col,
+  t0.float_col,
+  t0.double_col,
+  t0.date_string_col,
+  t0.string_col,
+  t0.timestamp_col,
+  t0.year,
+  t0.month
+FROM functional_alltypes AS t0
+WHERE
+  REGEXP_POSITION(t0.string_col, '0') <> -1

ibis/backends/tests/test_string.py

@@ -861,3 +861,23 @@ def test_multiple_subs(con):
     expr = ibis.literal("foo").substitute(m)
     result = con.execute(expr)
     assert result == "FOO"
+
+
+@pytest.mark.never(
+    ["dask", "datafusion", "pandas", "polars", "pyspark"],
+    raises=NotImplementedError,
+    reason="not a SQL backend",
+)
+@pytest.mark.notyet(
+    ["mssql"],
+    raises=OperationNotDefinedError,
+    reason="Doesn't support regular expressions",
+)
+@pytest.mark.notyet(
+    ["druid"],
+    raises=ValueError,
+    reason="sqlglot doesn't support a druid dialect",
+)
+def test_rlike(snapshot, alltypes):
+    expr = alltypes[alltypes.string_col.rlike('0')]
+    snapshot.assert_match(str(ibis.to_sql(expr)), "out.sql")