Add support for first, last aggregate function parsing (#882)

* Add order by parsing to functions

* Fix doc error

* minor changes

src/ast/mod.rs

@@ -3366,6 +3366,8 @@ pub struct Function {
     // Some functions must be called without trailing parentheses, for example Postgres
     // do it for current_catalog, current_schema, etc. This flags is used for formatting.
     pub special: bool,
+    // Required ordering for the function (if empty, there is no requirement).
+    pub order_by: Vec<OrderByExpr>,
 }
 
 #[derive(Debug, Copy, Clone, PartialEq, PartialOrd, Eq, Ord, Hash)]
@@ -3392,12 +3394,18 @@ impl fmt::Display for Function {
         if self.special {
             write!(f, "{}", self.name)?;
         } else {
+            let order_by = if !self.order_by.is_empty() {
+                " ORDER BY "
+            } else {
+                ""
+            };
             write!(
                 f,
-                "{}({}{})",
+                "{}({}{}{order_by}{})",
                 self.name,
                 if self.distinct { "DISTINCT " } else { "" },
                 display_comma_separated(&self.args),
+                display_comma_separated(&self.order_by),
             )?;
 
             if let Some(o) = &self.over {

src/ast/visitor.rs

@@ -480,7 +480,7 @@ where
 ///     *expr = Expr::Function(Function {
 ///           name: ObjectName(vec![Ident::new("f")]),
 ///           args: vec![FunctionArg::Unnamed(FunctionArgExpr::Expr(old_expr))],
-///           over: None, distinct: false, special: false,
+///           over: None, distinct: false, special: false, order_by: vec![],
 ///      });
 ///   }
 ///   ControlFlow::<()>::Continue(())

src/parser.rs

@@ -706,6 +706,7 @@ impl<'a> Parser<'a> {
                         over: None,
                         distinct: false,
                         special: true,
+                        order_by: vec![],
                     }))
                 }
                 Keyword::CURRENT_TIMESTAMP
@@ -881,7 +882,7 @@ impl<'a> Parser<'a> {
     pub fn parse_function(&mut self, name: ObjectName) -> Result<Expr, ParserError> {
         self.expect_token(&Token::LParen)?;
         let distinct = self.parse_all_or_distinct()?.is_some();
-        let args = self.parse_optional_args()?;
+        let (args, order_by) = self.parse_optional_args_with_orderby()?;
         let over = if self.parse_keyword(Keyword::OVER) {
             // TBD: support window names (`OVER mywin`) in place of inline specification
             self.expect_token(&Token::LParen)?;
@@ -918,14 +919,15 @@ impl<'a> Parser<'a> {
             over,
             distinct,
             special: false,
+            order_by,
         }))
     }
 
     pub fn parse_time_functions(&mut self, name: ObjectName) -> Result<Expr, ParserError> {
-        let args = if self.consume_token(&Token::LParen) {
+        let (args, order_by) = if self.consume_token(&Token::LParen) {
-            self.parse_optional_args()?
+            self.parse_optional_args_with_orderby()?
         } else {
-            vec![]
+            (vec![], vec![])
         };
         Ok(Expr::Function(Function {
             name,
@@ -933,6 +935,7 @@ impl<'a> Parser<'a> {
             over: None,
             distinct: false,
             special: false,
+            order_by,
         }))
     }
 
@@ -6376,6 +6379,23 @@ impl<'a> Parser<'a> {
         }
     }
 
+    pub fn parse_optional_args_with_orderby(
+        &mut self,
+    ) -> Result<(Vec<FunctionArg>, Vec<OrderByExpr>), ParserError> {
+        if self.consume_token(&Token::RParen) {
+            Ok((vec![], vec![]))
+        } else {
+            let args = self.parse_comma_separated(Parser::parse_function_args)?;
+            let order_by = if self.parse_keywords(&[Keyword::ORDER, Keyword::BY]) {
+                self.parse_comma_separated(Parser::parse_order_by_expr)?
+            } else {
+                vec![]
+            };
+            self.expect_token(&Token::RParen)?;
+            Ok((args, order_by))
+        }
+    }
+
     /// Parse a comma-delimited list of projections after SELECT
     pub fn parse_select_item(&mut self) -> Result<SelectItem, ParserError> {
         match self.parse_wildcard_expr()? {

tests/sqlparser_bigquery.rs

@@ -437,6 +437,7 @@ fn parse_map_access_offset() {
                 over: None,
                 distinct: false,
                 special: false,
+                order_by: vec![],
             })],
         })
     );

tests/sqlparser_clickhouse.rs

@@ -52,6 +52,7 @@ fn parse_map_access_expr() {
                     over: None,
                     distinct: false,
                     special: false,
+                    order_by: vec![],
                 })],
             })],
             into: None,
@@ -88,6 +89,7 @@ fn parse_map_access_expr() {
                             over: None,
                             distinct: false,
                             special: false,
+                            order_by: vec![],
                         })]
                     }),
                     op: BinaryOperator::NotEq,
@@ -135,6 +137,7 @@ fn parse_array_fn() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(only(&select.projection))
     );
@@ -189,6 +192,7 @@ fn parse_delimited_identifiers() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[1]),
     );

tests/sqlparser_common.rs

@@ -840,6 +840,7 @@ fn parse_select_count_wildcard() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(only(&select.projection))
     );
@@ -859,6 +860,7 @@ fn parse_select_count_distinct() {
             over: None,
             distinct: true,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(only(&select.projection))
     );
@@ -1696,6 +1698,7 @@ fn parse_select_having() {
                 over: None,
                 distinct: false,
                 special: false,
+                order_by: vec![],
             })),
             op: BinaryOperator::Gt,
             right: Box::new(Expr::Value(number("1"))),
@@ -1729,6 +1732,7 @@ fn parse_select_qualify() {
                 }),
                 distinct: false,
                 special: false,
+                order_by: vec![],
             })),
             op: BinaryOperator::Eq,
             right: Box::new(Expr::Value(number("1"))),
@@ -2076,6 +2080,29 @@ fn parse_array_agg_func() {
     }
 }
 
+#[test]
+fn parse_agg_with_order_by() {
+    let supported_dialects = TestedDialects {
+        dialects: vec![
+            Box::new(GenericDialect {}),
+            Box::new(PostgreSqlDialect {}),
+            Box::new(MsSqlDialect {}),
+            Box::new(AnsiDialect {}),
+            Box::new(HiveDialect {}),
+        ],
+        options: None,
+    };
+
+    for sql in [
+        "SELECT FIRST_VALUE(x ORDER BY x) AS a FROM T",
+        "SELECT FIRST_VALUE(x ORDER BY x) FROM tbl",
+        "SELECT LAST_VALUE(x ORDER BY x, y) AS a FROM T",
+        "SELECT LAST_VALUE(x ORDER BY x ASC, y DESC) AS a FROM T",
+    ] {
+        supported_dialects.verified_stmt(sql);
+    }
+}
+
 #[test]
 fn parse_create_table() {
     let sql = "CREATE TABLE uk_cities (\
@@ -3121,6 +3148,7 @@ fn parse_scalar_function_in_projection() {
                 over: None,
                 distinct: false,
                 special: false,
+                order_by: vec![],
             }),
             expr_from_projection(only(&select.projection))
         );
@@ -3239,6 +3267,7 @@ fn parse_named_argument_function() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(only(&select.projection))
     );
@@ -3277,6 +3306,7 @@ fn parse_window_functions() {
             }),
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[0])
     );
@@ -3675,6 +3705,7 @@ fn parse_at_timezone() {
                 over: None,
                 distinct: false,
                 special: false,
+                order_by: vec![],
             })),
             time_zone: "UTC-06:00".to_string(),
         },
@@ -3701,6 +3732,7 @@ fn parse_at_timezone() {
                             over: None,
                             distinct: false,
                             special: false,
+                            order_by: vec![],
                         },)),
                         time_zone: "UTC-06:00".to_string(),
                     },),),
@@ -3711,6 +3743,7 @@ fn parse_at_timezone() {
                 over: None,
                 distinct: false,
                 special: false,
+                order_by: vec![],
             },),
             alias: Ident {
                 value: "hour".to_string(),
@@ -3868,6 +3901,7 @@ fn parse_table_function() {
                 over: None,
                 distinct: false,
                 special: false,
+                order_by: vec![],
             });
             assert_eq!(expr, expected_expr);
             assert_eq!(alias, table_alias("a"))
@@ -6292,6 +6326,7 @@ fn parse_time_functions() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[0])
     );
@@ -6308,6 +6343,7 @@ fn parse_time_functions() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[0])
     );
@@ -6324,6 +6360,7 @@ fn parse_time_functions() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[0])
     );
@@ -6340,6 +6377,7 @@ fn parse_time_functions() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[0])
     );
@@ -6356,6 +6394,7 @@ fn parse_time_functions() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[0])
     );
@@ -6820,6 +6859,7 @@ fn parse_pivot_table() {
                 over: None,
                 distinct: false,
                 special: false,
+                order_by: vec![],
             }),
             value_column: vec![Ident::new("a"), Ident::new("MONTH")],
             pivot_values: vec![

tests/sqlparser_hive.rs

@@ -346,6 +346,7 @@ fn parse_delimited_identifiers() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[1]),
     );

tests/sqlparser_mssql.rs

@@ -177,6 +177,7 @@ fn parse_delimited_identifiers() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[1]),
     );

tests/sqlparser_mysql.rs

@@ -791,6 +791,7 @@ fn parse_insert_with_on_duplicate_update() {
                             over: None,
                             distinct: false,
                             special: false,
+                            order_by: vec![],
                         })
                     },
                     Assignment {
@@ -803,6 +804,7 @@ fn parse_insert_with_on_duplicate_update() {
                             over: None,
                             distinct: false,
                             special: false,
+                            order_by: vec![],
                         })
                     },
                     Assignment {
@@ -815,6 +817,7 @@ fn parse_insert_with_on_duplicate_update() {
                             over: None,
                             distinct: false,
                             special: false,
+                            order_by: vec![],
                         })
                     },
                     Assignment {
@@ -827,6 +830,7 @@ fn parse_insert_with_on_duplicate_update() {
                             over: None,
                             distinct: false,
                             special: false,
+                            order_by: vec![],
                         })
                     },
                     Assignment {
@@ -839,6 +843,7 @@ fn parse_insert_with_on_duplicate_update() {
                             over: None,
                             distinct: false,
                             special: false,
+                            order_by: vec![],
                         })
                     },
                 ])),
@@ -1182,6 +1187,7 @@ fn parse_table_colum_option_on_update() {
                             over: None,
                             distinct: false,
                             special: false,
+                            order_by: vec![],
                         })),
                     },],
                 }],

tests/sqlparser_postgres.rs

@@ -2057,7 +2057,8 @@ fn test_composite_value() {
                 )))],
                 over: None,
                 distinct: false,
-                special: false
+                special: false,
+                order_by: vec![],
             }))))
         }),
         select.projection[0]
@@ -2219,6 +2220,7 @@ fn parse_current_functions() {
             over: None,
             distinct: false,
             special: true,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[0])
     );
@@ -2229,6 +2231,7 @@ fn parse_current_functions() {
             over: None,
             distinct: false,
             special: true,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[1])
     );
@@ -2239,6 +2242,7 @@ fn parse_current_functions() {
             over: None,
             distinct: false,
             special: true,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[2])
     );
@@ -2249,6 +2253,7 @@ fn parse_current_functions() {
             over: None,
             distinct: false,
             special: true,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[3])
     );
@@ -2503,6 +2508,7 @@ fn parse_delimited_identifiers() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[1]),
     );

tests/sqlparser_redshift.rs

@@ -132,6 +132,7 @@ fn parse_delimited_identifiers() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[1]),
     );

tests/sqlparser_snowflake.rs

@@ -249,6 +249,7 @@ fn parse_delimited_identifiers() {
             over: None,
             distinct: false,
             special: false,
+            order_by: vec![],
         }),
         expr_from_projection(&select.projection[1]),
     );