From ff3659adbad53a7a7175d1aef1a340c9c8c1bb80 Mon Sep 17 00:00:00 2001
From: Max Justus Spransy <maxjustus@gmail.com>
Date: Mon, 11 Nov 2024 13:38:37 -0800
Subject: [PATCH] support ClickHouse/DuckDB join variants

https://clickhouse.com/docs/en/sql-reference/statements/select/join
https://duckdb.org/docs/sql/query_syntax/from.html
---
 src/lib.rs       | 28 +++++++++++++++++++++
 src/tokenizer.rs | 65 ++++++++++++++++++++++++++++++++++++++----------
 2 files changed, 80 insertions(+), 13 deletions(-)

diff --git a/src/lib.rs b/src/lib.rs
index 5d7d2da..76451a2 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -320,6 +320,34 @@ mod tests {
         assert_eq!(format(input, &QueryParams::None, &options), expected);
     }
 
+    #[test]
+    fn it_formats_select_query_with_non_standard_join() {
+        let input = indoc!(
+            "
+            SELECT customer_id.from, COUNT(order_id) AS total FROM customers
+            INNER ANY JOIN orders ON customers.customer_id = orders.customer_id
+            LEFT
+            SEMI JOIN foo ON foo.id = customers.id
+            PASTE
+            JOIN bar
+            ;"
+        );
+        let options = FormatOptions::default();
+        let expected = indoc!(
+            "
+            SELECT
+              customer_id.from,
+              COUNT(order_id) AS total
+            FROM
+              customers
+              INNER ANY JOIN orders ON customers.customer_id = orders.customer_id
+              LEFT SEMI JOIN foo ON foo.id = customers.id
+              PASTE JOIN bar;"
+        );
+
+        assert_eq!(format(input, &QueryParams::None, &options), expected);
+    }
+
     #[test]
     fn it_formats_select_query_with_different_comments() {
         let input = indoc!(
diff --git a/src/tokenizer.rs b/src/tokenizer.rs
index 0d2ef78..25d4a33 100644
--- a/src/tokenizer.rs
+++ b/src/tokenizer.rs
@@ -580,24 +580,63 @@ fn get_newline_reserved_token<'a>(
     last_reserved_token: Option<Token<'a>>,
 ) -> impl FnMut(&'a str) -> IResult<&'a str, Token<'a>> {
     move |input: &'a str| {
-        let uc_input = get_uc_words(input, 3);
-        let result: IResult<&str, &str> = alt((
-            terminated(tag("AND"), end_of_word),
-            terminated(tag("CROSS APPLY"), end_of_word),
-            terminated(tag("CROSS JOIN"), end_of_word),
-            terminated(tag("ELSE"), end_of_word),
-            terminated(tag("INNER JOIN"), end_of_word),
+        let uc_input: String = get_uc_words(input, 3);
+
+        // We have to break up the alternatives into multiple subsets
+        // to avoid exceeding the alt() 21 element limit.
+
+        // Standard SQL joins
+        let standard_joins = alt((
             terminated(tag("JOIN"), end_of_word),
+            terminated(tag("INNER JOIN"), end_of_word),
             terminated(tag("LEFT JOIN"), end_of_word),
-            terminated(tag("LEFT OUTER JOIN"), end_of_word),
-            terminated(tag("OR"), end_of_word),
-            terminated(tag("OUTER APPLY"), end_of_word),
-            terminated(tag("OUTER JOIN"), end_of_word),
             terminated(tag("RIGHT JOIN"), end_of_word),
+            terminated(tag("FULL JOIN"), end_of_word),
+            terminated(tag("CROSS JOIN"), end_of_word),
+            terminated(tag("LEFT OUTER JOIN"), end_of_word),
             terminated(tag("RIGHT OUTER JOIN"), end_of_word),
-            terminated(tag("WHEN"), end_of_word),
+            terminated(tag("FULL OUTER JOIN"), end_of_word),
+        ));
+
+        // Warehouse-specific ANY/SEMI/ANTI joins
+        let specific_joins = alt((
+            terminated(tag("INNER ANY JOIN"), end_of_word),
+            terminated(tag("LEFT ANY JOIN"), end_of_word),
+            terminated(tag("RIGHT ANY JOIN"), end_of_word),
+            terminated(tag("ANY JOIN"), end_of_word),
+            terminated(tag("SEMI JOIN"), end_of_word),
+            terminated(tag("LEFT SEMI JOIN"), end_of_word),
+            terminated(tag("RIGHT SEMI JOIN"), end_of_word),
+            terminated(tag("LEFT ANTI JOIN"), end_of_word),
+            terminated(tag("RIGHT ANTI JOIN"), end_of_word),
+        ));
+
+        // Special joins and GLOBAL variants
+        let special_joins = alt((
+            terminated(tag("ASOF JOIN"), end_of_word),
+            terminated(tag("LEFT ASOF JOIN"), end_of_word),
+            terminated(tag("PASTE JOIN"), end_of_word),
+            terminated(tag("GLOBAL INNER JOIN"), end_of_word),
+            terminated(tag("GLOBAL LEFT JOIN"), end_of_word),
+            terminated(tag("GLOBAL RIGHT JOIN"), end_of_word),
+            terminated(tag("GLOBAL FULL JOIN"), end_of_word),
+        ));
+
+        // Legacy and logical operators
+        let operators = alt((
+            terminated(tag("CROSS APPLY"), end_of_word),
+            terminated(tag("OUTER APPLY"), end_of_word),
+            terminated(tag("AND"), end_of_word),
+            terminated(tag("OR"), end_of_word),
             terminated(tag("XOR"), end_of_word),
-        ))(&uc_input);
+            terminated(tag("WHEN"), end_of_word),
+            terminated(tag("ELSE"), end_of_word),
+        ));
+
+        // Combine all parsers
+        let result: IResult<&str, &str> =
+            alt((standard_joins, specific_joins, special_joins, operators))(&uc_input);
+
         if let Ok((_, token)) = result {
             let final_word = token.split(' ').last().unwrap();
             let input_end_pos =