Merge pull request #77 from benesch/count-distinct

Support COUNT(DISTINCT x) and similar

Author: nickolay

src/sqlast/mod.rs

@@ -112,6 +112,8 @@ pub enum ASTNode {
         name: SQLObjectName,
         args: Vec<ASTNode>,
         over: Option<SQLWindowSpec>,
+        // aggregate functions may specify eg `COUNT(DISTINCT x)`
+        distinct: bool,
     },
     /// CASE [<operand>] WHEN <condition> THEN <result> ... [ELSE <result>] END
     /// Note we only recognize a complete single expression as <condition>, not
@@ -190,8 +192,18 @@ impl ToString for ASTNode {
                 format!("{} {}", operator.to_string(), expr.as_ref().to_string())
             }
             ASTNode::SQLValue(v) => v.to_string(),
-            ASTNode::SQLFunction { name, args, over } => {
-                let mut s = format!("{}({})", name.to_string(), comma_separated_string(args));
+            ASTNode::SQLFunction {
+                name,
+                args,
+                over,
+                distinct,
+            } => {
+                let mut s = format!(
+                    "{}({}{})",
+                    name.to_string(),
+                    if *distinct { "DISTINCT " } else { "" },
+                    comma_separated_string(args)
+                );
                 if let Some(o) = over {
                     s += &format!(" OVER ({})", o.to_string())
                 }

src/sqlparser.rs

@@ -270,6 +270,14 @@ impl Parser {
 
     pub fn parse_function(&mut self, name: SQLObjectName) -> Result<ASTNode, ParserError> {
         self.expect_token(&Token::LParen)?;
+        let all = self.parse_keyword("ALL");
+        let distinct = self.parse_keyword("DISTINCT");
+        if all && distinct {
+            return parser_err!(format!(
+                "Cannot specify both ALL and DISTINCT in function: {}",
+                name.to_string(),
+            ));
+        }
         let args = self.parse_optional_args()?;
         let over = if self.parse_keyword("OVER") {
             // TBD: support window names (`OVER mywin`) in place of inline specification
@@ -296,7 +304,12 @@ impl Parser {
             None
         };
 
-        Ok(ASTNode::SQLFunction { name, args, over })
+        Ok(ASTNode::SQLFunction {
+            name,
+            args,
+            over,
+            distinct,
+        })
     }
 
     pub fn parse_window_frame(&mut self) -> Result<Option<SQLWindowFrame>, ParserError> {

tests/sqlparser_common.rs

@@ -211,11 +211,44 @@ fn parse_select_count_wildcard() {
             name: SQLObjectName(vec!["COUNT".to_string()]),
             args: vec![ASTNode::SQLWildcard],
             over: None,
+            distinct: false,
         },
         expr_from_projection(only(&select.projection))
     );
 }
 
+#[test]
+fn parse_select_count_distinct() {
+    let sql = "SELECT COUNT(DISTINCT + x) FROM customer";
+    let select = verified_only_select(sql);
+    assert_eq!(
+        &ASTNode::SQLFunction {
+            name: SQLObjectName(vec!["COUNT".to_string()]),
+            args: vec![ASTNode::SQLUnary {
+                operator: SQLOperator::Plus,
+                expr: Box::new(ASTNode::SQLIdentifier("x".to_string()))
+            }],
+            over: None,
+            distinct: true,
+        },
+        expr_from_projection(only(&select.projection))
+    );
+
+    one_statement_parses_to(
+        "SELECT COUNT(ALL + x) FROM customer",
+        "SELECT COUNT(+ x) FROM customer",
+    );
+
+    let sql = "SELECT COUNT(ALL DISTINCT + x) FROM customer";
+    let res = parse_sql_statements(sql);
+    assert_eq!(
+        ParserError::ParserError(
+            "Cannot specify both ALL and DISTINCT in function: COUNT".to_string()
+        ),
+        res.unwrap_err()
+    );
+}
+
 #[test]
 fn parse_not() {
     let sql = "SELECT id FROM customer WHERE NOT salary = ''";
@@ -676,6 +709,7 @@ fn parse_scalar_function_in_projection() {
             name: SQLObjectName(vec!["sqrt".to_string()]),
             args: vec![ASTNode::SQLIdentifier("id".to_string())],
             over: None,
+            distinct: false,
         },
         expr_from_projection(only(&select.projection))
     );
@@ -704,7 +738,8 @@ fn parse_window_functions() {
                     asc: Some(false)
                 }],
                 window_frame: None,
-            })
+            }),
+            distinct: false,
         },
         expr_from_projection(&select.projection[0])
     );
@@ -776,6 +811,7 @@ fn parse_delimited_identifiers() {
             name: SQLObjectName(vec![r#""myfun""#.to_string()]),
             args: vec![],
             over: None,
+            distinct: false,
         },
         expr_from_projection(&select.projection[1]),
     );