WIP: Add support for all ClickHouse builtin functions, QUALIFY and WITH FILL FROM/TO/STEP

crates/lib-dialects/src/clickhouse.rs

@@ -450,6 +450,70 @@ pub fn dialect(config: Option<&Value>) -> Dialect {
         ),
     ]);
 
+    // ClickHouse allows tuple literals as expression arguments in function calls,
+    // e.g. addInterval((INTERVAL 1 DAY, INTERVAL 1 YEAR), INTERVAL 1 MONTH).
+    clickhouse_dialect.replace_grammar(
+        "BaseExpressionElementGrammar",
+        ansi_dialect.grammar("BaseExpressionElementGrammar").copy(
+            Some(vec![Ref::new("TupleSegment").to_matchable()]),
+            None,
+            Some(Ref::new("ExpressionSegment").to_matchable()),
+            None,
+            Vec::new(),
+            false,
+        ),
+    );
+
+    // Function arguments in ClickHouse may include tuple literals, e.g.
+    // addInterval((INTERVAL 1 DAY, INTERVAL 1 YEAR), INTERVAL 1 MONTH).
+    clickhouse_dialect.replace_grammar(
+        "FunctionContentsExpressionGrammar",
+        one_of(vec![
+            Ref::new("TupleSegment").to_matchable(),
+            Ref::new("ExpressionSegment").to_matchable(),
+        ])
+        .to_matchable(),
+    );
+
+    // Prioritize delimited argument parsing so single tuple arguments are parsed as
+    // function arguments instead of generic bracketed expressions.
+    clickhouse_dialect.replace_grammar(
+        "FunctionContentsGrammar",
+        ansi_dialect.grammar("FunctionContentsGrammar").copy(
+            Some(vec![
+                Sequence::new(vec![
+                    Ref::keyword("DISTINCT").optional().to_matchable(),
+                    one_of(vec![
+                        Ref::new("StarSegment").to_matchable(),
+                        Delimited::new(vec![
+                            Ref::new("FunctionContentsExpressionGrammar").to_matchable(),
+                        ])
+                        .to_matchable(),
+                    ])
+                    .to_matchable(),
+                ])
+                .to_matchable(),
+            ]),
+            Some(0),
+            None,
+            None,
+            Vec::new(),
+            false,
+        ),
+    );
+
+    // ClickHouse supports aggregate modifiers after the function call, e.g.
+    // any(x) RESPECT NULLS.
+    clickhouse_dialect.replace_grammar(
+        "PostFunctionGrammar",
+        one_of(vec![
+            Ref::new("OverClauseSegment").to_matchable(),
+            Ref::new("FilterClauseGrammar").to_matchable(),
+            Ref::new("IgnoreRespectNullsGrammar").to_matchable(),
+        ])
+        .to_matchable(),
+    );
+
     clickhouse_dialect.replace_grammar(
         "WildcardExpressionSegment",
         ansi::wildcard_expression_segment().copy(
@@ -768,6 +832,22 @@ pub fn dialect(config: Option<&Value>) -> Dialect {
         .into(),
     )]);
 
+    clickhouse_dialect.add([(
+        "QualifyClauseSegment".into(),
+        NodeMatcher::new(SyntaxKind::QualifyClause, |_| {
+            Sequence::new(vec![
+                Ref::keyword("QUALIFY").to_matchable(),
+                MetaSegment::implicit_indent().to_matchable(),
+                optionally_bracketed(vec![Ref::new("ExpressionSegment").to_matchable()])
+                    .to_matchable(),
+                MetaSegment::dedent().to_matchable(),
+            ])
+            .to_matchable()
+        })
+        .to_matchable()
+        .into(),
+    )]);
+
     // We need to replace the UnorderedSelectStatementSegment to include PREWHERE
     clickhouse_dialect.replace_grammar(
         "UnorderedSelectStatementSegment",
@@ -800,6 +880,16 @@ pub fn dialect(config: Option<&Value>) -> Dialect {
                 Vec::new(),
                 false,
             )
+            .copy(
+                Some(vec![
+                    Ref::new("QualifyClauseSegment").optional().to_matchable(),
+                ]),
+                None,
+                Some(Ref::new("OrderByClauseSegment").optional().to_matchable()),
+                None,
+                Vec::new(),
+                false,
+            )
             .copy(
                 Some(vec![
                     Ref::new("FormatClauseSegment").optional().to_matchable(),
@@ -835,11 +925,7 @@ pub fn dialect(config: Option<&Value>) -> Dialect {
             .to_matchable(),
             Sequence::new(vec![
                 Ref::keyword("STEP").to_matchable(),
-                one_of(vec![
-                    Ref::new("NumericLiteralSegment").to_matchable(),
-                    Ref::new("IntervalExpressionSegment").to_matchable(),
-                ])
-                .to_matchable(),
+                Ref::new("ExpressionSegment").to_matchable(),
             ])
             .config(|this| this.optional())
             .to_matchable(),

crates/lib-dialects/src/clickhouse_keywords.rs

@@ -135,6 +135,7 @@ pub(crate) const UNRESERVED_KEYWORDS: &[&str] = &[
     "PROFILE",
     "PROJECTION",
     "QUARTER",
+    "QUALIFY",
     "QUERY",
     "QUOTA",
     "QUEUES",

crates/lib-dialects/test/fixtures/dialects/clickhouse/sqlfluff/complex_table_definition.yml

@@ -161,14 +161,13 @@ file:
                 - expression:
                   - quoted_literal: '''key'''
                 - comma: ','
-                - expression:
+                - tuple:
                   - bracketed:
                     - start_bracket: (
                     - column_reference:
                       - naked_identifier: category
                     - comma: ','
-                    - column_reference:
-                      - quoted_identifier: '''CATEGORY'''
+                    - quoted_literal: '''CATEGORY'''
                     - end_bracket: )
                 - comma: ','
                 - expression:
@@ -661,14 +660,13 @@ file:
             - expression:
               - quoted_literal: '''price'''
             - comma: ','
-            - expression:
+            - tuple:
               - bracketed:
                 - start_bracket: (
                 - column_reference:
                   - naked_identifier: category
                 - comma: ','
-                - column_reference:
-                  - quoted_identifier: '''RESOURCE_TYPE'''
+                - quoted_literal: '''RESOURCE_TYPE'''
                 - end_bracket: )
             - comma: ','
             - expression:

crates/lib-dialects/test/fixtures/dialects/clickhouse/sqlfluff/create_table.yml

@@ -143,7 +143,7 @@ file:
             - column_reference:
               - naked_identifier: date
           - comma: ','
-          - expression:
+          - tuple:
             - bracketed:
               - start_bracket: (
               - column_reference:

crates/lib-dialects/test/fixtures/dialects/clickhouse/sqlfluff/order_by_with_fill.yml

@@ -81,7 +81,8 @@ file:
       - expression:
         - numeric_literal: '5.51'
       - keyword: STEP
-      - numeric_literal: '0.5'
+      - expression:
+        - numeric_literal: '0.5'
 - statement_terminator: ;
 - statement:
   - select_statement:
@@ -173,7 +174,8 @@ file:
       - keyword: WITH
       - keyword: FILL
       - keyword: STEP
-      - numeric_literal: '5'
+      - expression:
+        - numeric_literal: '5'
 - statement_terminator: ;
 - statement:
   - select_statement:
@@ -260,7 +262,8 @@ file:
       - keyword: WITH
       - keyword: FILL
       - keyword: STEP
-      - numeric_literal: '5'
+      - expression:
+        - numeric_literal: '5'
       - comma: ','
       - column_reference:
         - naked_identifier: d2
@@ -352,10 +355,11 @@ file:
       - keyword: WITH
       - keyword: FILL
       - keyword: STEP
-      - interval_expression:
-        - keyword: INTERVAL
-        - numeric_literal: '1'
-        - date_part: DAY
+      - expression:
+        - interval_expression:
+          - keyword: INTERVAL
+          - numeric_literal: '1'
+          - date_part: DAY
       - comma: ','
       - column_reference:
         - naked_identifier: d2

crates/lib-dialects/test/fixtures/dialects/clickhouse/sqruff/any_respect_nulls_variants.sql

@@ -0,0 +1,15 @@
+SELECT any_respect_nulls(x) FROM t;
+SELECT any(x) RESPECT NULLS FROM t;
+SELECT anyRespectNulls(x) FROM t;
+SELECT any_value_respect_nulls(x) FROM t;
+SELECT anyValueRespectNulls(x) FROM t;
+SELECT any_value(x) RESPECT NULLS FROM t;
+SELECT first_value_respect_nulls(x) FROM t;
+SELECT firstValueRespectNulls(x) FROM t;
+SELECT first_value(x) RESPECT NULLS FROM t;
+SELECT anyLast_respect_nulls(x) FROM t;
+SELECT anyLastRespectNulls(x) FROM t;
+SELECT anyLast(x) RESPECT NULLS FROM t;
+SELECT last_value_respect_nulls(x) FROM t;
+SELECT lastValueRespectNulls(x) FROM t;
+SELECT last_value(x) RESPECT NULLS FROM t;