Fix Hive table comment should be after table column definitions (apache#1413)

Author: git-hulk

src/ast/dml.rs

@@ -226,6 +226,13 @@ impl Display for CreateTable {
             // PostgreSQL allows `CREATE TABLE t ();`, but requires empty parens
             write!(f, " ()")?;
         }
+
+        // Hive table comment should be after column definitions, please refer to:
+        // [Hive](https://cwiki.apache.org/confluence/display/Hive/LanguageManual+DDL#LanguageManualDDL-CreateTable)
+        if let Some(CommentDef::AfterColumnDefsWithoutEq(comment)) = &self.comment {
+            write!(f, " COMMENT '{comment}'")?;
+        }
+
         // Only for SQLite
         if self.without_rowid {
             write!(f, " WITHOUT ROWID")?;
@@ -336,6 +343,8 @@ impl Display for CreateTable {
                 CommentDef::WithoutEq(comment) => {
                     write!(f, " COMMENT '{comment}'")?;
                 }
+                // For CommentDef::AfterColumnDefsWithoutEq will be displayed after column definition
+                CommentDef::AfterColumnDefsWithoutEq(_) => (),
             }
         }
 

src/ast/mod.rs

@@ -6828,12 +6828,18 @@ pub enum CommentDef {
     /// Does not include `=` when printing the comment, as `COMMENT 'comment'`
     WithEq(String),
     WithoutEq(String),
+    // For Hive dialect, the table comment is after the column definitions without `=`,
+    // so we need to add an extra variant to allow to identify this case when displaying.
+    // [Hive](https://cwiki.apache.org/confluence/display/Hive/LanguageManual+DDL#LanguageManualDDL-CreateTable)
+    AfterColumnDefsWithoutEq(String),
 }
 
 impl Display for CommentDef {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         match self {
-            CommentDef::WithEq(comment) | CommentDef::WithoutEq(comment) => write!(f, "{comment}"),
+            CommentDef::WithEq(comment)
+            | CommentDef::WithoutEq(comment)
+            | CommentDef::AfterColumnDefsWithoutEq(comment) => write!(f, "{comment}"),
         }
     }
 }

src/parser/mod.rs

@@ -5612,6 +5612,17 @@ impl<'a> Parser<'a> {
 
         // parse optional column list (schema)
         let (columns, constraints) = self.parse_columns()?;
+        let mut comment = if dialect_of!(self is HiveDialect)
+            && self.parse_keyword(Keyword::COMMENT)
+        {
+            let next_token = self.next_token();
+            match next_token.token {
+                Token::SingleQuotedString(str) => Some(CommentDef::AfterColumnDefsWithoutEq(str)),
+                _ => self.expected("comment", next_token)?,
+            }
+        } else {
+            None
+        };
 
         // SQLite supports `WITHOUT ROWID` at the end of `CREATE TABLE`
         let without_rowid = self.parse_keywords(&[Keyword::WITHOUT, Keyword::ROWID]);
@@ -5722,15 +5733,14 @@ impl<'a> Parser<'a> {
 
         let strict = self.parse_keyword(Keyword::STRICT);
 
-        let comment = if self.parse_keyword(Keyword::COMMENT) {
+        // Excludes Hive dialect here since it has been handled after table column definitions.
+        if !dialect_of!(self is HiveDialect) && self.parse_keyword(Keyword::COMMENT) {
             let _ = self.consume_token(&Token::Eq);
             let next_token = self.next_token();
-            match next_token.token {
+            comment = match next_token.token {
                 Token::SingleQuotedString(str) => Some(CommentDef::WithoutEq(str)),
                 _ => self.expected("comment", next_token)?,
             }
-        } else {
-            None
         };
 
         // Parse optional `AS ( query )`

tests/sqlparser_hive.rs

@@ -16,7 +16,7 @@
 //! is also tested (on the inputs it can handle).
 
 use sqlparser::ast::{
-    ClusteredBy, CreateFunctionBody, CreateFunctionUsing, CreateTable, Expr, Function,
+    ClusteredBy, CommentDef, CreateFunctionBody, CreateFunctionUsing, CreateTable, Expr, Function,
     FunctionArgumentList, FunctionArguments, Ident, ObjectName, OneOrManyWithParens, OrderByExpr,
     SelectItem, Statement, TableFactor, UnaryOperator, Use, Value,
 };
@@ -115,6 +115,39 @@ fn create_table_like() {
     hive().verified_stmt(like);
 }
 
+#[test]
+fn create_table_with_comment() {
+    let sql = concat!(
+        "CREATE TABLE db.table_name (a INT, b STRING)",
+        " COMMENT 'table comment'",
+        " PARTITIONED BY (a INT, b STRING)",
+        " CLUSTERED BY (a, b) SORTED BY (a ASC, b DESC)",
+        " INTO 4 BUCKETS"
+    );
+    match hive().verified_stmt(sql) {
+        Statement::CreateTable(CreateTable { comment, .. }) => {
+            assert_eq!(
+                comment,
+                Some(CommentDef::AfterColumnDefsWithoutEq(
+                    "table comment".to_string()
+                ))
+            )
+        }
+        _ => unreachable!(),
+    }
+
+    // negative test case
+    let invalid_sql = concat!(
+        "CREATE TABLE db.table_name (a INT, b STRING)",
+        " PARTITIONED BY (a INT, b STRING)",
+        " COMMENT 'table comment'",
+    );
+    assert_eq!(
+        hive().parse_sql_statements(invalid_sql).unwrap_err(),
+        ParserError::ParserError("Expected: end of statement, found: COMMENT".to_string())
+    );
+}
+
 #[test]
 fn create_table_with_clustered_by() {
     let sql = concat!(