PostgreSQL: Support SQL standard ARRAY keyword in type declarations

src/ast/data_type.rs

@@ -720,6 +720,8 @@ impl fmt::Display for DataType {
                 ArrayElemTypeDef::SquareBracket(t, Some(size)) => write!(f, "{t}[{size}]"),
                 ArrayElemTypeDef::AngleBracket(t) => write!(f, "ARRAY<{t}>"),
                 ArrayElemTypeDef::Parenthesis(t) => write!(f, "Array({t})"),
+                ArrayElemTypeDef::Qualified(t, None) => write!(f, "{t} ARRAY"),
+                ArrayElemTypeDef::Qualified(t, Some(size)) => write!(f, "{t} ARRAY[{size}]"),
             },
             DataType::Custom(ty, modifiers) => {
                 if modifiers.is_empty() {
@@ -1163,6 +1165,8 @@ pub enum ArrayElemTypeDef {
     SquareBracket(Box<DataType>, Option<u64>),
     /// Parenthesis style, e.g. `Array(Int64)`.
     Parenthesis(Box<DataType>),
+    /// Qualified by a data type and optional size, e.g. `INT ARRAY` or `INT ARRAY[4]`.
+    Qualified(Box<DataType>, Option<u64>),
 }
 
 /// Represents different types of geometric shapes which are commonly used in

src/ast/mod.rs

@@ -1110,12 +1110,6 @@ pub enum Expr {
         expr: Box<Expr>,
         /// Target data type.
         data_type: DataType,
-        /// [MySQL] allows CAST(... AS type ARRAY) in functional index definitions for InnoDB
-        /// multi-valued indices. It's not really a datatype, and is only allowed in `CAST` in key
-        /// specifications, so it's a flag here.
-        ///
-        /// [MySQL]: https://dev.mysql.com/doc/refman/8.4/en/cast-functions.html#function_cast
-        array: bool,
         /// Optional CAST(string_expression AS type FORMAT format_string_expression) as used by [BigQuery]
         ///
         /// [BigQuery]: https://cloud.google.com/bigquery/docs/reference/standard-sql/format-elements#formatting_syntax
@@ -1976,14 +1970,10 @@ impl fmt::Display for Expr {
                 kind,
                 expr,
                 data_type,
-                array,
                 format,
             } => match kind {
                 CastKind::Cast => {
                     write!(f, "CAST({expr} AS {data_type}")?;
-                    if *array {
-                        write!(f, " ARRAY")?;
-                    }
                     if let Some(format) = format {
                         write!(f, " FORMAT {format}")?;
                     }

src/ast/spans.rs

@@ -1606,7 +1606,6 @@ impl Spanned for Expr {
                 kind: _,
                 expr,
                 data_type: _,
-                array: _,
                 format: _,
             } => expr.span(),
             Expr::AtTimeZone {

src/parser/mod.rs

@@ -1831,7 +1831,6 @@ impl<'a> Parser<'a> {
                         kind: CastKind::Cast,
                         expr: Box::new(parser.parse_expr()?),
                         data_type: DataType::Binary(None),
-                        array: false,
                         format: None,
                     })
                 }
@@ -2905,14 +2904,12 @@ impl<'a> Parser<'a> {
         let expr = self.parse_expr()?;
         self.expect_keyword_is(Keyword::AS)?;
         let data_type = self.parse_data_type()?;
-        let array = self.parse_keyword(Keyword::ARRAY);
         let format = self.parse_optional_cast_format()?;
         self.expect_token(&Token::RParen)?;
         Ok(Expr::Cast {
             kind,
             expr: Box::new(expr),
             data_type,
-            array,
             format,
         })
     }
@@ -4186,7 +4183,6 @@ impl<'a> Parser<'a> {
                 kind: CastKind::DoubleColon,
                 expr: Box::new(expr),
                 data_type: self.parse_data_type()?,
-                array: false,
                 format: None,
             })
         } else if Token::ExclamationMark == *tok && self.dialect.supports_factorial_operator() {
@@ -4443,7 +4439,6 @@ impl<'a> Parser<'a> {
             kind: CastKind::DoubleColon,
             expr: Box::new(expr),
             data_type: self.parse_data_type()?,
-            array: false,
             format: None,
         })
     }
@@ -13150,6 +13145,20 @@ impl<'a> Parser<'a> {
                 data = DataType::Array(ArrayElemTypeDef::SquareBracket(Box::new(data), size))
             }
         }
+
+        // Type-qualified array, e.g. `INT ARRAY` or `INT ARRAY[3]`. One-dimensional
+        // with a single optional size. The `ARRAY` keyword is unambiguous everywhere.
+        if self.parse_keyword(Keyword::ARRAY) {
+            let size = if self.consume_token(&Token::LBracket) {
+                let size = self.maybe_parse(|p| p.parse_literal_uint())?;
+                self.expect_token(&Token::RBracket)?;
+                size
+            } else {
+                None
+            };
+            data = DataType::Array(ArrayElemTypeDef::Qualified(Box::new(data), size));
+        }
+
         Ok((data, trailing_bracket))
     }
 

tests/sqlparser_common.rs

@@ -3172,7 +3172,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::BigInt(None),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3185,7 +3184,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::TinyInt(None),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3217,7 +3215,6 @@ fn parse_cast() {
                 length: 50,
                 unit: None,
             })),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3230,7 +3227,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::Clob(None),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3243,7 +3239,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::Clob(Some(50)),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3256,7 +3251,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::Binary(Some(50)),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3269,7 +3263,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::Varbinary(Some(BinaryLength::IntegerLength { length: 50 })),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3282,7 +3275,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::Blob(None),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3295,7 +3287,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::Blob(Some(50)),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3308,7 +3299,6 @@ fn parse_cast() {
             kind: CastKind::Cast,
             expr: Box::new(Expr::Identifier(Ident::new("details"))),
             data_type: DataType::JSONB,
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -3324,7 +3314,6 @@ fn parse_try_cast() {
             kind: CastKind::TryCast,
             expr: Box::new(Expr::Identifier(Ident::new("id"))),
             data_type: DataType::BigInt(None),
-            array: false,
             format: None,
         },
         expr_from_projection(only(&select.projection))
@@ -6698,7 +6687,6 @@ fn interval_disallow_interval_expr_double_colon() {
                 fractional_seconds_precision: None,
             })),
             data_type: DataType::Text,
-            array: false,
             format: None,
         }
     )
@@ -6716,7 +6704,6 @@ fn parse_text_type_modifier_double_colon_cast() {
                 ObjectName::from(vec![Ident::new("TEXT")]),
                 vec!["16777216".to_string()]
             ),
-            array: false,
             format: None,
         }
     );
@@ -9482,7 +9469,6 @@ fn parse_double_colon_cast_at_timezone() {
                         .with_empty_span()
                 )),
                 data_type: DataType::Timestamp(None, TimezoneInfo::None),
-                array: false,
                 format: None
             }),
             time_zone: Box::new(Expr::Value(
@@ -13897,7 +13883,6 @@ fn test_dictionary_syntax() {
                         (Value::SingleQuotedString("2023-04-01".to_owned())).with_empty_span(),
                     )),
                     data_type: DataType::Timestamp(None, TimezoneInfo::None),
-                    array: false,
                     format: None,
                 }),
             },
@@ -13909,7 +13894,6 @@ fn test_dictionary_syntax() {
                         (Value::SingleQuotedString("2023-04-05".to_owned())).with_empty_span(),
                     )),
                     data_type: DataType::Timestamp(None, TimezoneInfo::None),
-                    array: false,
                     format: None,
                 }),
             },
@@ -14206,7 +14190,6 @@ fn test_extract_seconds_ok() {
                     fields: None,
                     precision: None
                 },
-                array: false,
                 format: None,
             }),
         }
@@ -14237,7 +14220,6 @@ fn test_extract_seconds_ok() {
                         fields: None,
                         precision: None,
                     },
-                    array: false,
                     format: None,
                 }),
             })],
@@ -14295,7 +14277,6 @@ fn test_extract_seconds_single_quote_ok() {
                     fields: None,
                     precision: None
                 },
-                array: false,
                 format: None,
             }),
         }

tests/sqlparser_databricks.rs

@@ -421,7 +421,6 @@ fn data_type_timestamp_ntz() {
                 "created_at".into()
             )))),
             data_type: DataType::TimestampNtz(None),
-            array: false,
             format: None
         }
     );

tests/sqlparser_duckdb.rs

@@ -388,7 +388,6 @@ fn test_duckdb_specific_int_types() {
                     Value::Number("123".parse().unwrap(), false).with_empty_span()
                 )),
                 data_type: data_type.clone(),
-                array: false,
                 format: None,
             },
             expr_from_projection(&select.projection[0])

tests/sqlparser_mysql.rs

@@ -886,7 +886,6 @@ fn test_functional_key_part() {
                 )),
             }),
             data_type: DataType::Unsigned,
-            array: false,
             format: None,
         })),
     );
@@ -903,8 +902,10 @@ fn test_functional_key_part() {
                     Value::SingleQuotedString("$.fields".to_string()).with_empty_span()
                 )),
             }),
-            data_type: DataType::Unsigned,
-            array: true,
+            data_type: DataType::Array(ArrayElemTypeDef::Qualified(
+                Box::new(DataType::Unsigned),
+                None,
+            )),
             format: None,
         })),
     );
@@ -4278,12 +4279,47 @@ fn parse_cast_integers() {
 
 #[test]
 fn parse_cast_array() {
-    mysql().verified_expr("CAST(foo AS SIGNED ARRAY)");
+    // The element type may be any type accepted by CAST().
+    for ty in [
+        "SIGNED",
+        "UNSIGNED",
+        "CHAR",
+        "CHAR(10)",
+        "BINARY",
+        "BINARY(5)",
+        "DATE",
+        "TIME",
+        "DATETIME",
+        "DECIMAL",
+        "DECIMAL(10,2)",
+        "DOUBLE",
+        "FLOAT",
+        "YEAR",
+    ] {
+        mysql().verified_expr(&format!("CAST(foo AS {ty} ARRAY)"));
+    }
+
+    // `ARRAY` on its own is not a valid CAST target type.
     mysql()
         .run_parser_method("CAST(foo AS ARRAY)", |p| p.parse_expr())
         .expect_err("ARRAY alone is not a type");
 }
 
+#[test]
+fn parse_multi_valued_index() {
+    // `CAST(... AS <type> ARRAY)` key part in CREATE TABLE, CREATE INDEX, and
+    // ALTER TABLE. See https://dev.mysql.com/doc/refman/8.0/en/create-index.html
+    mysql_and_generic().verified_stmt(
+        "CREATE TABLE customers (id BIGINT, custinfo JSON, INDEX zips ((CAST(custinfo -> '$.zipcode' AS UNSIGNED ARRAY))))",
+    );
+    mysql_and_generic().verified_stmt(
+        "CREATE INDEX zips ON customers((CAST(custinfo -> '$.zipcode' AS UNSIGNED ARRAY)))",
+    );
+    mysql_and_generic().verified_stmt(
+        "ALTER TABLE customers ADD INDEX zips ((CAST(custinfo -> '$.zipcode' AS UNSIGNED ARRAY)))",
+    );
+}
+
 #[test]
 fn parse_match_against_with_alias() {
     let sql = "SELECT tbl.ProjectID FROM surveys.tbl1 AS tbl WHERE MATCH (tbl.ReferenceID) AGAINST ('AAA' IN BOOLEAN MODE)";