Merge pull request #25589 from TszKitLo40/support-distinct-on

src/Parsers/ParserSelectQuery.cpp

@@ -1,4 +1,5 @@
 #include <memory>
+#include <Parsers/ASTLiteral.h>
 #include <Parsers/ASTSelectQuery.h>
 #include <Parsers/IParserBase.h>
 #include <Parsers/CommonParsers.h>
@@ -16,11 +17,12 @@ namespace DB
 
 namespace ErrorCodes
 {
-    extern const int TOP_AND_LIMIT_TOGETHER;
-    extern const int WITH_TIES_WITHOUT_ORDER_BY;
+    extern const int FIRST_AND_NEXT_TOGETHER;
     extern const int LIMIT_BY_WITH_TIES_IS_NOT_SUPPORTED;
     extern const int ROW_AND_ROWS_TOGETHER;
-    extern const int FIRST_AND_NEXT_TOGETHER;
+    extern const int SYNTAX_ERROR;
+    extern const int TOP_AND_LIMIT_TOGETHER;
+    extern const int WITH_TIES_WITHOUT_ORDER_BY;
 }
 
 
@@ -32,6 +34,7 @@ bool ParserSelectQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     ParserKeyword s_select("SELECT");
     ParserKeyword s_all("ALL");
     ParserKeyword s_distinct("DISTINCT");
+    ParserKeyword s_distinct_on("DISTINCT ON");
     ParserKeyword s_from("FROM");
     ParserKeyword s_prewhere("PREWHERE");
     ParserKeyword s_where("WHERE");
@@ -77,12 +80,13 @@ bool ParserSelectQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     ASTPtr limit_by_length;
     ASTPtr limit_by_offset;
     ASTPtr limit_by_expression_list;
+    ASTPtr distinct_on_expression_list;
     ASTPtr limit_offset;
     ASTPtr limit_length;
     ASTPtr top_length;
     ASTPtr settings;
 
-    /// WITH expr list
+    /// WITH expr_list
     {
         if (s_with.ignore(pos, expected))
         {
@@ -94,7 +98,7 @@ bool ParserSelectQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
         }
     }
 
-    /// SELECT [ALL/DISTINCT] [TOP N [WITH TIES]] expr list
+    /// SELECT [ALL/DISTINCT [ON (expr_list)]] [TOP N [WITH TIES]] expr_list
     {
         bool has_all = false;
         if (!s_select.ignore(pos, expected))
@@ -103,13 +107,27 @@ bool ParserSelectQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
         if (s_all.ignore(pos, expected))
             has_all = true;
 
-        if (s_distinct.ignore(pos, expected))
+        if (s_distinct_on.ignore(pos, expected))
+        {
+            if (open_bracket.ignore(pos, expected))
+            {
+                if (!exp_list.parse(pos, distinct_on_expression_list, expected))
+                    return false;
+                if (!close_bracket.ignore(pos, expected))
+                    return false;
+            }
+            else
+                return false;
+        }
+        else if (s_distinct.ignore(pos, expected))
+        {
             select_query->distinct = true;
+        }
 
         if (!has_all && s_all.ignore(pos, expected))
             has_all = true;
 
-        if (has_all && select_query->distinct)
+        if (has_all && (select_query->distinct || distinct_on_expression_list))
             return false;
 
         if (s_top.ignore(pos, expected))
@@ -256,13 +274,19 @@ bool ParserSelectQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
             select_query->limit_with_ties = true;
         }
 
+        if (limit_with_ties_occured && distinct_on_expression_list)
+            throw Exception("Can not use WITH TIES alongside LIMIT BY/DISTINCT ON", ErrorCodes::LIMIT_BY_WITH_TIES_IS_NOT_SUPPORTED);
+
         if (s_by.ignore(pos, expected))
         {
             /// WITH TIES was used alongside LIMIT BY
             /// But there are other kind of queries like LIMIT n BY smth LIMIT m WITH TIES which are allowed.
             /// So we have to ignore WITH TIES exactly in LIMIT BY state.
             if (limit_with_ties_occured)
-                throw Exception("Can not use WITH TIES alongside LIMIT BY", ErrorCodes::LIMIT_BY_WITH_TIES_IS_NOT_SUPPORTED);
+                throw Exception("Can not use WITH TIES alongside LIMIT BY/DISTINCT ON", ErrorCodes::LIMIT_BY_WITH_TIES_IS_NOT_SUPPORTED);
+
+            if (distinct_on_expression_list)
+                throw Exception("Can not use DISTINCT ON alongside LIMIT BY", ErrorCodes::SYNTAX_ERROR);
 
             limit_by_length = limit_length;
             limit_by_offset = limit_offset;
@@ -335,6 +359,17 @@ bool ParserSelectQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
         }
     }
 
+    if (distinct_on_expression_list)
+    {
+        /// DISTINCT ON and LIMIT BY are mutually exclusive, checked before
+        assert (limit_by_expression_list == nullptr);
+
+        /// Transform `DISTINCT ON expr` to `LIMIT 1 BY expr`
+        limit_by_expression_list = distinct_on_expression_list;
+        limit_by_length = std::make_shared<ASTLiteral>(Field{UInt8(1)});
+        distinct_on_expression_list = nullptr;
+    }
+
     /// Because TOP n in totally equals LIMIT n
     if (top_length)
         limit_length = top_length;

tests/queries/0_stateless/01917_distinct_on.reference

@@ -0,0 +1,8 @@
+1	1	1
+2	2	2
+1	2	2
+1	1	1
+2	2	2
+1	2	2
+1	1	1
+2	2	2

tests/queries/0_stateless/01917_distinct_on.sql

@@ -0,0 +1,23 @@
+DROP TABLE IF EXISTS t1;
+
+CREATE TABLE t1 (`a` UInt32, `b` UInt32, `c` UInt32 ) ENGINE = Memory;
+INSERT INTO t1 VALUES (1, 1, 1), (1, 1, 2), (2, 2, 2), (1, 2, 2);
+
+SELECT DISTINCT ON (a, b) a, b, c FROM t1;
+SELECT DISTINCT ON (a, b) * FROM t1;
+SELECT DISTINCT ON (a) * FROM t1;
+
+-- fuzzer will fail, enable when fixed
+-- SELECT DISTINCT ON (a, b) a, b, c FROM t1 LIMIT 1 BY a, b; -- { clientError 62 }
+
+-- SELECT DISTINCT ON a, b a, b FROM t1; -- { clientError 62 }
+-- SELECT DISTINCT ON a a, b FROM t1; -- { clientError 62 }
+
+-- "Code: 47. DB::Exception: Missing columns: 'DISTINCT'" - error can be better
+-- SELECT DISTINCT ON (a, b) DISTINCT a, b FROM t1; -- { serverError 47 }
+-- SELECT DISTINCT DISTINCT ON (a, b) a, b FROM t1; -- { clientError 62 }
+
+-- SELECT ALL DISTINCT ON (a, b) a, b FROM t1; -- { clientError 62 }
+-- SELECT DISTINCT ON (a, b) ALL a, b FROM t1; -- { clientError 62 }
+
+DROP TABLE IF EXISTS t1;