fix(impala/pyspark): use regexp_replace to strip

ibis/backends/impala/tests/snapshots/test_string_builtins/test_string_builtins/lstrip/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  LTRIM(`t0`.`string_col`, ' \t\n\r\v\f') AS `LStrip(string_col)`
+  REGEXP_REPLACE(`t0`.`string_col`, '^\\s+', '') AS `LStrip(string_col)`
 FROM `functional_alltypes` AS `t0`

ibis/backends/impala/tests/snapshots/test_string_builtins/test_string_builtins/rstrip/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  RTRIM(`t0`.`string_col`, ' \t\n\r\v\f') AS `RStrip(string_col)`
+  REGEXP_REPLACE(`t0`.`string_col`, '\\s+$', '') AS `RStrip(string_col)`
 FROM `functional_alltypes` AS `t0`

ibis/backends/impala/tests/snapshots/test_string_builtins/test_string_builtins/strip/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  RTRIM(LTRIM(`t0`.`string_col`, ' \t\n\r\v\f'), ' \t\n\r\v\f') AS `Strip(string_col)`
+  REGEXP_REPLACE(`t0`.`string_col`, '^\\s+|\\s+$', '') AS `Strip(string_col)`
 FROM `functional_alltypes` AS `t0`

ibis/backends/sql/compilers/impala.py

@@ -1,7 +1,5 @@
 from __future__ import annotations
 
-from string import whitespace as WHITESPACE
-
 import sqlglot as sg
 import sqlglot.expressions as sge
 
@@ -45,6 +43,7 @@ class ImpalaCompiler(SQLGlotCompiler):
         ops.ArgMin,
         ops.Covariance,
         ops.ExtractDayOfYear,
+        ops.Kurtosis,
         ops.Levenshtein,
         ops.Map,
         ops.Median,
@@ -59,7 +58,6 @@ class ImpalaCompiler(SQLGlotCompiler):
         ops.TimestampBucket,
         ops.TimestampDelta,
         ops.Unnest,
-        ops.Kurtosis,
     )
 
     SIMPLE_OPS = {
@@ -78,8 +76,8 @@ class ImpalaCompiler(SQLGlotCompiler):
         ops.ExtractEpochSeconds: "unix_timestamp",
         ops.Hash: "fnv_hash",
         ops.Ln: "ln",
-        ops.TypeOf: "typeof",
         ops.RegexReplace: "regexp_replace",
+        ops.TypeOf: "typeof",
     }
 
     @staticmethod
@@ -325,16 +323,13 @@ def visit_DateDelta(self, op, *, left, right, part):
         return self.f.datediff(left, right)
 
     def visit_LStrip(self, op, *, arg):
-        return self.f.anon.ltrim(arg, WHITESPACE)
+        return self.f.regexp_replace(arg, r"^\s+", "")
 
     def visit_RStrip(self, op, *, arg):
-        return self.f.anon.rtrim(arg, WHITESPACE)
+        return self.f.regexp_replace(arg, r"\s+$", "")
 
     def visit_Strip(self, op, *, arg):
-        # Impala's `TRIM` doesn't allow specifying characters to trim off, unlike
-        # Impala's `RTRIM` and `LTRIM` which accept a set of characters to
-        # remove.
-        return self.f.anon.rtrim(self.f.anon.ltrim(arg, WHITESPACE), WHITESPACE)
+        return self.f.regexp_replace(arg, r"^\s+|\s+$", "")
 
 
 compiler = ImpalaCompiler()

ibis/backends/sql/compilers/pyspark.py

@@ -85,8 +85,6 @@ class PySparkCompiler(SQLGlotCompiler):
         ops.EndsWith: "endswith",
         ops.Hash: "hash",
         ops.Log10: "log10",
-        ops.LStrip: "ltrim",
-        ops.RStrip: "rtrim",
         ops.MapLength: "size",
         ops.MapContains: "map_contains_key",
         ops.MapMerge: "map_concat",
@@ -684,5 +682,14 @@ def visit_ArraySum(self, op, *, arg):
     def visit_ArrayMean(self, op, *, arg):
         return self._array_reduction(dtype=op.dtype, arg=arg, output=operator.truediv)
 
+    def visit_LStrip(self, op, *, arg):
+        return self.f.regexp_replace(arg, r"^\s+", "")
+
+    def visit_RStrip(self, op, *, arg):
+        return self.f.regexp_replace(arg, r"\s+$", "")
+
+    def visit_Strip(self, op, *, arg):
+        return self.f.regexp_replace(arg, r"^\s+|\s+$", "")
+
 
 compiler = PySparkCompiler()

ibis/backends/tests/test_string.py

@@ -1144,8 +1144,9 @@ def string_temp_table(backend, con):
                 "aBc",
                 "🐍",
                 "ÉéÈèêç",
+                "fluf\f",
             ],
-            "index_col": [0, 1, 2, 3, 4, 5, 6],
+            "index_col": [0, 1, 2, 3, 4, 5, 6, 7],
         }
     )
 
@@ -1277,7 +1278,7 @@ def string_temp_table(backend, con):
         ),
         param(
             lambda t: t.string_col.find_in_set(["aBc", "123"]),
-            lambda _: pd.Series([-1, -1, -1, 1, 0, -1, -1], name="tmp"),
+            lambda _: pd.Series([-1, -1, -1, 1, 0, -1, -1, -1], name="tmp"),
             id="find_in_set",
             marks=[
                 pytest.mark.notyet(
@@ -1306,7 +1307,7 @@ def string_temp_table(backend, con):
         ),
         param(
             lambda t: t.string_col.find_in_set(["abc, 123"]),
-            lambda _: pd.Series([-1, -1, -1, -1, -1, -1, -1], name="tmp"),
+            lambda _: pd.Series([-1, -1, -1, -1, -1, -1, -1, -1], name="tmp"),
             id="find_in_set_w_comma",
             marks=[
                 pytest.mark.notyet(
@@ -1346,25 +1347,11 @@ def string_temp_table(backend, con):
             lambda t: t.string_col.lstrip(),
             lambda t: t.str.lstrip(),
             id="lstrip",
-            marks=[
-                pytest.mark.notyet(
-                    ["pyspark", "databricks"],
-                    raises=AssertionError,
-                    reason="Spark SQL LTRIM doesn't accept characters to trim",
-                ),
-            ],
         ),
         param(
             lambda t: t.string_col.rstrip(),
             lambda t: t.str.rstrip(),
             id="rstrip",
-            marks=[
-                pytest.mark.notyet(
-                    ["pyspark", "databricks"],
-                    raises=AssertionError,
-                    reason="Spark SQL RTRIM doesn't accept characters to trim",
-                ),
-            ],
         ),
         param(
             lambda t: t.string_col.strip(),
@@ -1423,6 +1410,7 @@ def test_string_methods_accents_and_emoji(
     │ aBc        │
     │ 🐍         │
     │ ÉéÈèêç     │
+    │ fluf\f     │
     └────────────┘
     """
     t = string_temp_table