fix(schema-compiler): Use isNotDistinctFrom() for ClickHouse Tesseract multi-fact joins (#10494)

ClickHouse does not support OR/IS NULL patterns in JOIN ON clauses,
which causes Tesseract multi-fact queries to fail. Add support for a
function-style is_not_distinct_from expression template and register
ClickHouse's isNotDistinctFrom() function as the implementation.

Fixes #10493

---------

Co-authored-by: Ryan Clapp <ryan.clapp@getgarner.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Dmitry Patsura <talk@dmtry.me>

Author: 4 people

packages/cubejs-schema-compiler/src/adapter/ClickHouseQuery.ts

@@ -277,6 +277,7 @@ export class ClickHouseQuery extends BaseQuery {
     // ClickHouse intervals have a distinct type for each granularity
     delete templates.types.interval;
     delete templates.types.binary;
+    templates.expressions.is_not_distinct_from = 'isNotDistinctFrom({{ left }}, {{ right }})';
     return templates;
   }
 }

rust/cubesqlplanner/cubesqlplanner/src/planner/sql_templates/plan.rs

@@ -462,6 +462,12 @@ impl PlanSqlTemplates {
 
                 return self.binary_expr(left_column, &is_not_distinct_from_op, right_column);
             }
+            if self.supports_is_not_distinct_from_expr() {
+                return self.render.render_template(
+                    "expressions/is_not_distinct_from",
+                    context! { left => left_column.as_str(), right => right_column.as_str() },
+                );
+            }
             format!(
                 " OR ({} AND {})",
                 self.is_null_expr(&left_column, false)?,
@@ -486,6 +492,11 @@ impl PlanSqlTemplates {
             .contains_template("operators/is_not_distinct_from")
     }
 
+    pub fn supports_is_not_distinct_from_expr(&self) -> bool {
+        self.render
+            .contains_template("expressions/is_not_distinct_from")
+    }
+
     pub fn supports_generated_time_series(
         &self,
         predifined_granularity: bool,
@@ -798,3 +809,112 @@ impl PlanSqlTemplates {
         )
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::test_fixtures::cube_bridge::{MockDriverTools, MockSqlTemplatesRender};
+    use std::collections::HashMap;
+
+    fn plan_templates_with(extra_templates: Vec<(&str, &str)>) -> PlanSqlTemplates {
+        if extra_templates.is_empty() {
+            let driver_tools = Rc::new(MockDriverTools::new());
+            return PlanSqlTemplates::try_new(driver_tools, false).unwrap();
+        }
+        // Build a minimal template set with only what join_by_dimension_conditions needs
+        let mut t: HashMap<String, String> = HashMap::new();
+        t.insert(
+            "expressions/binary".to_string(),
+            "({{ left }} {{ op }} {{ right }})".to_string(),
+        );
+        t.insert(
+            "expressions/is_null".to_string(),
+            "({{ expr }} IS {% if negate %}NOT {% endif %}NULL)".to_string(),
+        );
+        for (k, v) in extra_templates {
+            t.insert(k.to_string(), v.to_string());
+        }
+        let render = MockSqlTemplatesRender::try_new(t).unwrap();
+        let driver_tools = Rc::new(MockDriverTools::with_sql_templates(render));
+        PlanSqlTemplates::try_new(driver_tools, false).unwrap()
+    }
+
+    #[test]
+    fn test_join_condition_no_null_check() {
+        let templates = plan_templates_with(vec![]);
+        let left = "t1.col".to_string();
+        let right = "t2.col".to_string();
+
+        let result = templates
+            .join_by_dimension_conditions(&left, &right, false)
+            .unwrap();
+        assert_eq!(result, "(t1.col = t2.col)");
+    }
+
+    #[test]
+    fn test_join_condition_null_check_fallback_or_is_null() {
+        // No is_not_distinct_from templates → falls back to OR (IS NULL AND IS NULL)
+        let templates = plan_templates_with(vec![]);
+        let left = "t1.col".to_string();
+        let right = "t2.col".to_string();
+
+        let result = templates
+            .join_by_dimension_conditions(&left, &right, true)
+            .unwrap();
+        assert_eq!(
+            result,
+            "(t1.col = t2.col OR ((t1.col IS NULL) AND (t2.col IS NULL)))"
+        );
+    }
+
+    #[test]
+    fn test_join_condition_null_check_binary_operator() {
+        // Postgres/BigQuery/Snowflake style: binary operator IS NOT DISTINCT FROM
+        let templates = plan_templates_with(vec![(
+            "operators/is_not_distinct_from",
+            "IS NOT DISTINCT FROM",
+        )]);
+        let left = "t1.col".to_string();
+        let right = "t2.col".to_string();
+
+        let result = templates
+            .join_by_dimension_conditions(&left, &right, true)
+            .unwrap();
+        assert_eq!(result, "(t1.col IS NOT DISTINCT FROM t2.col)");
+    }
+
+    #[test]
+    fn test_join_condition_null_check_expression_template() {
+        // ClickHouse style: function-call isNotDistinctFrom(left, right)
+        let templates = plan_templates_with(vec![(
+            "expressions/is_not_distinct_from",
+            "isNotDistinctFrom({{ left }}, {{ right }})",
+        )]);
+        let left = "t1.col".to_string();
+        let right = "t2.col".to_string();
+
+        let result = templates
+            .join_by_dimension_conditions(&left, &right, true)
+            .unwrap();
+        assert_eq!(result, "isNotDistinctFrom(t1.col, t2.col)");
+    }
+
+    #[test]
+    fn test_join_condition_binary_operator_takes_precedence_over_expression() {
+        // When both templates exist, the binary operator should be used
+        let templates = plan_templates_with(vec![
+            ("operators/is_not_distinct_from", "IS NOT DISTINCT FROM"),
+            (
+                "expressions/is_not_distinct_from",
+                "isNotDistinctFrom({{ left }}, {{ right }})",
+            ),
+        ]);
+        let left = "t1.col".to_string();
+        let right = "t2.col".to_string();
+
+        let result = templates
+            .join_by_dimension_conditions(&left, &right, true)
+            .unwrap();
+        assert_eq!(result, "(t1.col IS NOT DISTINCT FROM t2.col)");
+    }
+}