Support serdeproperties for CREATE TABLE with HIVE (#1152)

Co-authored-by: Andrew Lamb <[email protected]>

Author: jonathanlehto

src/ast/mod.rs

@@ -3205,6 +3205,7 @@ impl fmt::Display for Statement {
 
                 if let Some(HiveFormat {
                     row_format,
+                    serde_properties,
                     storage,
                     location,
                 }) = hive_formats
@@ -3229,6 +3230,13 @@ impl fmt::Display for Statement {
                         }
                         _ => (),
                     }
+                    if let Some(serde_properties) = serde_properties.as_ref() {
+                        write!(
+                            f,
+                            " WITH SERDEPROPERTIES ({})",
+                            display_comma_separated(serde_properties)
+                        )?;
+                    }
                     if !*external {
                         if let Some(loc) = location {
                             write!(f, " LOCATION '{loc}'")?;
@@ -4886,6 +4894,7 @@ pub enum HiveIOFormat {
 #[cfg_attr(feature = "visitor", derive(Visit, VisitMut))]
 pub struct HiveFormat {
     pub row_format: Option<HiveRowFormat>,
+    pub serde_properties: Option<Vec<SqlOption>>,
     pub storage: Option<HiveIOFormat>,
     pub location: Option<String>,
 }

src/keywords.rs

@@ -596,6 +596,7 @@ define_keywords!(
     SEQUENCEFILE,
     SEQUENCES,
     SERDE,
+    SERDEPROPERTIES,
     SERIALIZABLE,
     SESSION,
     SESSION_USER,

src/parser/mod.rs

@@ -4353,7 +4353,12 @@ impl<'a> Parser<'a> {
     pub fn parse_hive_formats(&mut self) -> Result<HiveFormat, ParserError> {
         let mut hive_format = HiveFormat::default();
         loop {
-            match self.parse_one_of_keywords(&[Keyword::ROW, Keyword::STORED, Keyword::LOCATION]) {
+            match self.parse_one_of_keywords(&[
+                Keyword::ROW,
+                Keyword::STORED,
+                Keyword::LOCATION,
+                Keyword::WITH,
+            ]) {
                 Some(Keyword::ROW) => {
                     hive_format.row_format = Some(self.parse_row_format()?);
                 }
@@ -4375,6 +4380,16 @@ impl<'a> Parser<'a> {
                 Some(Keyword::LOCATION) => {
                     hive_format.location = Some(self.parse_literal_string()?);
                 }
+                Some(Keyword::WITH) => {
+                    self.prev_token();
+                    let properties = self
+                        .parse_options_with_keywords(&[Keyword::WITH, Keyword::SERDEPROPERTIES])?;
+                    if !properties.is_empty() {
+                        hive_format.serde_properties = Some(properties);
+                    } else {
+                        break;
+                    }
+                }
                 None => break,
                 _ => break,
             }

tests/sqlparser_hive.rs

@@ -27,9 +27,11 @@ use sqlparser::test_utils::*;
 fn parse_table_create() {
     let sql = r#"CREATE TABLE IF NOT EXISTS db.table (a BIGINT, b STRING, c TIMESTAMP) PARTITIONED BY (d STRING, e TIMESTAMP) STORED AS ORC LOCATION 's3://...' TBLPROPERTIES ("prop" = "2", "asdf" = '1234', 'asdf' = "1234", "asdf" = 2)"#;
     let iof = r#"CREATE TABLE IF NOT EXISTS db.table (a BIGINT, b STRING, c TIMESTAMP) PARTITIONED BY (d STRING, e TIMESTAMP) STORED AS INPUTFORMAT 'org.apache.hadoop.hive.ql.io.orc.OrcInputFormat' OUTPUTFORMAT 'org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat' LOCATION 's3://...'"#;
+    let serdeproperties = r#"CREATE EXTERNAL TABLE IF NOT EXISTS db.table (a STRING, b STRING, c STRING) PARTITIONED BY (d STRING, e STRING) ROW FORMAT SERDE 'org.apache.hadoop.hive.serde.config' WITH SERDEPROPERTIES ('prop_a' = 'a', 'prop_b' = 'b') STORED AS TEXTFILE LOCATION 's3://...' TBLPROPERTIES ('prop_c' = 'c')"#;
 
     hive().verified_stmt(sql);
     hive().verified_stmt(iof);
+    hive().verified_stmt(serdeproperties);
 }
 
 fn generic(options: Option<ParserOptions>) -> TestedDialects {