hive: add support for array<> (#491)

Signed-off-by: Maciej Obuchowski <obuchowski.maciej@gmail.com>

Co-authored-by: Andrew Lamb <andrew@nerdnetworks.org>

src/parser.rs

@@ -2700,6 +2700,15 @@ impl<'a> Parser<'a> {
                 }
                 Keyword::ENUM => Ok(DataType::Enum(self.parse_string_values()?)),
                 Keyword::SET => Ok(DataType::Set(self.parse_string_values()?)),
+                Keyword::ARRAY => {
+                    // Hive array syntax. Note that nesting arrays - or other Hive syntax
+                    // that ends with > will fail due to "C++" problem - >> is parsed as
+                    // Token::ShiftRight
+                    self.expect_token(&Token::Lt)?;
+                    let inside_type = self.parse_data_type()?;
+                    self.expect_token(&Token::Gt)?;
+                    Ok(DataType::Array(Box::new(inside_type)))
+                }
                 _ => {
                     self.prev_token();
                     let type_name = self.parse_object_name()?;
@@ -2709,7 +2718,8 @@ impl<'a> Parser<'a> {
             unexpected => self.expected("a data type name", unexpected),
         }?;
 
-        // Parse array data types. Note: this is postgresql-specific
+        // Parse array data types. Note: this is postgresql-specific and different from
+        // Keyword::ARRAY syntax from above
         while self.consume_token(&Token::LBracket) {
             self.expect_token(&Token::RBracket)?;
             data = DataType::Array(Box::new(data))

tests/sqlparser_common.rs

@@ -23,8 +23,8 @@ mod test_utils;
 use matches::assert_matches;
 use sqlparser::ast::*;
 use sqlparser::dialect::{
-    AnsiDialect, BigQueryDialect, GenericDialect, MsSqlDialect, PostgreSqlDialect, SQLiteDialect,
+    AnsiDialect, BigQueryDialect, GenericDialect, HiveDialect, MsSqlDialect, PostgreSqlDialect,
-    SnowflakeDialect,
+    SQLiteDialect, SnowflakeDialect,
 };
 use sqlparser::keywords::ALL_KEYWORDS;
 use sqlparser::parser::{Parser, ParserError};
@@ -1755,6 +1755,53 @@ fn parse_create_table() {
         .contains("Expected constraint details after CONSTRAINT <name>"));
 }
 
+#[test]
+fn parse_create_table_hive_array() {
+    // Parsing [] type arrays does not work in MsSql since [ is used in is_delimited_identifier_start
+    let dialects = TestedDialects {
+        dialects: vec![Box::new(PostgreSqlDialect {}), Box::new(HiveDialect {})],
+    };
+    let sql = "CREATE TABLE IF NOT EXISTS something (key int, val array<int>)";
+    match dialects.one_statement_parses_to(
+        sql,
+        "CREATE TABLE IF NOT EXISTS something (key INT, val INT[])",
+    ) {
+        Statement::CreateTable {
+            if_not_exists,
+            name,
+            columns,
+            ..
+        } => {
+            assert!(if_not_exists);
+            assert_eq!(name, ObjectName(vec!["something".into()]));
+            assert_eq!(
+                columns,
+                vec![
+                    ColumnDef {
+                        name: Ident::new("key"),
+                        data_type: DataType::Int(None),
+                        collation: None,
+                        options: vec![],
+                    },
+                    ColumnDef {
+                        name: Ident::new("val"),
+                        data_type: DataType::Array(Box::new(DataType::Int(None))),
+                        collation: None,
+                        options: vec![],
+                    },
+                ],
+            )
+        }
+        _ => unreachable!(),
+    }
+
+    let res = parse_sql_statements("CREATE TABLE IF NOT EXISTS something (key int, val array<int)");
+    assert!(res
+        .unwrap_err()
+        .to_string()
+        .contains("Expected >, found: )"));
+}
+
 #[test]
 fn parse_create_table_with_multiple_on_delete_in_constraint_fails() {
     parse_sql_statements(