feat(clickhouse): implement trim, pad and string predicates

Author: cpcloud

ibis/backends/clickhouse/registry.py

@@ -566,6 +566,46 @@ def _string_like(translator, expr):
     )
 
 
+def _string_ilike(translator, expr):
+    op = expr.op()
+    return 'lower({}) LIKE lower({})'.format(
+        translator.translate(op.arg),
+        translator.translate(op.pattern),
+    )
+
+
+def _startswith(translator, expr):
+    op = expr.op()
+    arg = op.arg
+    start = op.start
+    tr_arg = translator.translate(arg)
+    tr_start = translator.translate(start)
+    return f"startsWith({tr_arg}, {tr_start})"
+
+
+def _endswith(translator, expr):
+    op = expr.op()
+    arg = translator.translate(op.arg)
+    end = translator.translate(op.end)
+    return f"endsWith({arg}, {end})"
+
+
+def _lpad(translator, expr):
+    op = expr.op()
+    arg = translator.translate(op.arg)
+    length = translator.translate(op.length)
+    pad = translator.translate(op.pad)
+    return f"leftPad({arg}, {length}, {pad})"
+
+
+def _rpad(translator, expr):
+    op = expr.op()
+    arg = translator.translate(op.arg)
+    length = translator.translate(op.length)
+    pad = translator.translate(op.pad)
+    return f"rightPad({arg}, {length}, {pad})"
+
+
 def _group_concat(translator, expr):
     arg, sep, where = expr.op().args
     if where is not None:
@@ -646,6 +686,14 @@ def _group_concat(translator, expr):
     ops.StringJoin: _string_join,
     ops.StringSplit: _string_split,
     ops.StringSQLLike: _string_like,
+    ops.StringSQLILike: _string_ilike,
+    ops.StartsWith: _startswith,
+    ops.EndsWith: _endswith,
+    ops.LPad: _lpad,
+    ops.RPad: _rpad,
+    ops.LStrip: _unary('trimLeft'),
+    ops.RStrip: _unary('trimRight'),
+    ops.Strip: _unary('trimBoth'),
     ops.Repeat: _string_repeat,
     ops.RegexSearch: _fixed_arity('match', 2),
     # TODO: extractAll(haystack, pattern)[index + 1]

ibis/backends/tests/test_string.py

@@ -48,14 +48,7 @@ def test_string_col_is_unicode(backend, alltypes, df):
             lambda t: t.string_col.ilike('6%'),
             lambda t: t.string_col.str.contains('6.*'),
             id='ilike',
-            marks=pytest.mark.notimpl(
-                [
-                    "clickhouse",
-                    "datafusion",
-                    "impala",
-                    "pyspark",
-                ]
-            ),
+            marks=pytest.mark.notimpl(["datafusion", "impala", "pyspark"]),
         ),
         param(
             lambda t: t.string_col.re_search(r'[[:digit:]]+'),
@@ -124,13 +117,11 @@ def test_string_col_is_unicode(backend, alltypes, df):
             lambda t: t.string_col.lpad(10, 'a'),
             lambda t: t.string_col.str.pad(10, fillchar='a', side='left'),
             id='lpad',
-            marks=pytest.mark.notimpl(["clickhouse"]),
         ),
         param(
             lambda t: t.string_col.rpad(10, 'a'),
             lambda t: t.string_col.str.pad(10, fillchar='a', side='right'),
             id='rpad',
-            marks=pytest.mark.notimpl(["clickhouse"]),
         ),
         param(
             lambda t: t.string_col.find_in_set(['1']),
@@ -179,43 +170,28 @@ def test_string_col_is_unicode(backend, alltypes, df):
             lambda t: t.string_col.startswith('foo'),
             lambda t: t.string_col.str.startswith('foo'),
             id='startswith',
-            marks=pytest.mark.notimpl(
-                ["clickhouse", "dask", "datafusion", "pandas"]
-            ),
+            marks=pytest.mark.notimpl(["dask", "datafusion", "pandas"]),
         ),
         param(
             lambda t: t.string_col.endswith('foo'),
             lambda t: t.string_col.str.endswith('foo'),
             id='endswith',
-            marks=pytest.mark.notimpl(
-                ["clickhouse", "dask", "datafusion", "pandas"]
-            ),
+            marks=pytest.mark.notimpl(["dask", "datafusion", "pandas"]),
         ),
         param(
             lambda t: t.string_col.strip(),
             lambda t: t.string_col.str.strip(),
             id='strip',
-            marks=pytest.mark.notimpl(["clickhouse"]),
         ),
         param(
             lambda t: t.string_col.lstrip(),
             lambda t: t.string_col.str.lstrip(),
             id='lstrip',
-            marks=pytest.mark.notimpl(
-                [
-                    "clickhouse",
-                ]
-            ),
         ),
         param(
             lambda t: t.string_col.rstrip(),
             lambda t: t.string_col.str.rstrip(),
             id='rstrip',
-            marks=pytest.mark.notimpl(
-                [
-                    "clickhouse",
-                ]
-            ),
         ),
         param(
             lambda t: t.string_col.capitalize(),