feat(clickhouse): implement ops.RandomScalar

Author: krzysztof-kwitt

ibis/backends/clickhouse/compiler/values.py

@@ -998,6 +998,7 @@ def formatter(op, **kw):
     ops.Tan: "tan",
     ops.Pi: "pi",
     ops.E: "e",
+    ops.RandomScalar: "randCanonical",
     # Unary aggregates
     ops.ApproxMedian: "median",
     # TODO: there is also a `uniq` function which is the

ibis/backends/tests/test_generic.py

@@ -504,10 +504,6 @@ def test_order_by(backend, alltypes, df, key, df_kwargs):
 
 
 @pytest.mark.notimpl(["dask", "datafusion", "impala", "pandas", "polars", "mssql"])
-@pytest.mark.notyet(
-    ["clickhouse"],
-    reason="clickhouse doesn't have a [0.0, 1.0) random implementation",
-)
 def test_order_by_random(alltypes):
     expr = alltypes.filter(_.id < 100).order_by(ibis.random()).limit(5)
     r1 = expr.execute()

ibis/backends/tests/test_numeric.py

@@ -543,10 +543,6 @@ def test_sa_default_numeric_precision_and_scale(
 
 
 @pytest.mark.notimpl(["dask", "datafusion", "impala", "pandas", "sqlite", "polars"])
-@pytest.mark.notyet(
-    ["clickhouse"],
-    reason="backend doesn't implement a [0.0, 1.0) or [0.0, 1.0] RANDOM() function",
-)
 def test_random(con):
     expr = ibis.random()
     result = con.execute(expr)