Support MySQL size variants for BLOB and TEXT columns (apache#1564)

Author: mvzink

src/ast/data_type.rs

@@ -76,6 +76,18 @@ pub enum DataType {
     /// [standard]: https://jakewheat.github.io/sql-overview/sql-2016-foundation-grammar.html#binary-large-object-string-type
     /// [Oracle]: https://docs.oracle.com/javadb/10.8.3.0/ref/rrefblob.html
     Blob(Option<u64>),
+    /// [MySQL] blob with up to 2**8 bytes
+    ///
+    /// [MySQL]: https://dev.mysql.com/doc/refman/9.1/en/blob.html
+    TinyBlob,
+    /// [MySQL] blob with up to 2**24 bytes
+    ///
+    /// [MySQL]: https://dev.mysql.com/doc/refman/9.1/en/blob.html
+    MediumBlob,
+    /// [MySQL] blob with up to 2**32 bytes
+    ///
+    /// [MySQL]: https://dev.mysql.com/doc/refman/9.1/en/blob.html
+    LongBlob,
     /// Variable-length binary data with optional length.
     ///
     /// [bigquery]: https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types#bytes_type
@@ -275,6 +287,18 @@ pub enum DataType {
     Regclass,
     /// Text
     Text,
+    /// [MySQL] text with up to 2**8 bytes
+    ///
+    /// [MySQL]: https://dev.mysql.com/doc/refman/9.1/en/blob.html
+    TinyText,
+    /// [MySQL] text with up to 2**24 bytes
+    ///
+    /// [MySQL]: https://dev.mysql.com/doc/refman/9.1/en/blob.html
+    MediumText,
+    /// [MySQL] text with up to 2**32 bytes
+    ///
+    /// [MySQL]: https://dev.mysql.com/doc/refman/9.1/en/blob.html
+    LongText,
     /// String with optional length.
     String(Option<u64>),
     /// A fixed-length string e.g [ClickHouse][1].
@@ -355,6 +379,9 @@ impl fmt::Display for DataType {
                 format_type_with_optional_length(f, "VARBINARY", size, false)
             }
             DataType::Blob(size) => format_type_with_optional_length(f, "BLOB", size, false),
+            DataType::TinyBlob => write!(f, "TINYBLOB"),
+            DataType::MediumBlob => write!(f, "MEDIUMBLOB"),
+            DataType::LongBlob => write!(f, "LONGBLOB"),
             DataType::Bytes(size) => format_type_with_optional_length(f, "BYTES", size, false),
             DataType::Numeric(info) => {
                 write!(f, "NUMERIC{info}")
@@ -486,6 +513,9 @@ impl fmt::Display for DataType {
             DataType::JSONB => write!(f, "JSONB"),
             DataType::Regclass => write!(f, "REGCLASS"),
             DataType::Text => write!(f, "TEXT"),
+            DataType::TinyText => write!(f, "TINYTEXT"),
+            DataType::MediumText => write!(f, "MEDIUMTEXT"),
+            DataType::LongText => write!(f, "LONGTEXT"),
             DataType::String(size) => format_type_with_optional_length(f, "STRING", size, false),
             DataType::Bytea => write!(f, "BYTEA"),
             DataType::Array(ty) => match ty {

src/keywords.rs

@@ -453,6 +453,8 @@ define_keywords!(
     LOCKED,
     LOGIN,
     LOGS,
+    LONGBLOB,
+    LONGTEXT,
     LOWCARDINALITY,
     LOWER,
     LOW_PRIORITY,
@@ -471,7 +473,9 @@ define_keywords!(
     MAXVALUE,
     MAX_DATA_EXTENSION_TIME_IN_DAYS,
     MEASURES,
+    MEDIUMBLOB,
     MEDIUMINT,
+    MEDIUMTEXT,
     MEMBER,
     MERGE,
     METADATA,
@@ -765,7 +769,9 @@ define_keywords!(
     TIMEZONE_HOUR,
     TIMEZONE_MINUTE,
     TIMEZONE_REGION,
+    TINYBLOB,
     TINYINT,
+    TINYTEXT,
     TO,
     TOP,
     TOTALS,

src/parser/mod.rs

@@ -8129,6 +8129,9 @@ impl<'a> Parser<'a> {
                 Keyword::BINARY => Ok(DataType::Binary(self.parse_optional_precision()?)),
                 Keyword::VARBINARY => Ok(DataType::Varbinary(self.parse_optional_precision()?)),
                 Keyword::BLOB => Ok(DataType::Blob(self.parse_optional_precision()?)),
+                Keyword::TINYBLOB => Ok(DataType::TinyBlob),
+                Keyword::MEDIUMBLOB => Ok(DataType::MediumBlob),
+                Keyword::LONGBLOB => Ok(DataType::LongBlob),
                 Keyword::BYTES => Ok(DataType::Bytes(self.parse_optional_precision()?)),
                 Keyword::UUID => Ok(DataType::Uuid),
                 Keyword::DATE => Ok(DataType::Date),
@@ -8188,6 +8191,9 @@ impl<'a> Parser<'a> {
                     Ok(DataType::FixedString(character_length))
                 }
                 Keyword::TEXT => Ok(DataType::Text),
+                Keyword::TINYTEXT => Ok(DataType::TinyText),
+                Keyword::MEDIUMTEXT => Ok(DataType::MediumText),
+                Keyword::LONGTEXT => Ok(DataType::LongText),
                 Keyword::BYTEA => Ok(DataType::Bytea),
                 Keyword::NUMERIC => Ok(DataType::Numeric(
                     self.parse_exact_number_optional_precision_scale()?,

tests/sqlparser_mysql.rs

@@ -3014,3 +3014,20 @@ fn parse_bitstring_literal() {
         ))]
     );
 }
+
+#[test]
+fn parse_longblob_type() {
+    let sql = "CREATE TABLE foo (bar LONGBLOB)";
+    let stmt = mysql_and_generic().verified_stmt(sql);
+    if let Statement::CreateTable(CreateTable { columns, .. }) = stmt {
+        assert_eq!(columns.len(), 1);
+        assert_eq!(columns[0].data_type, DataType::LongBlob);
+    } else {
+        unreachable!()
+    }
+    mysql_and_generic().verified_stmt("CREATE TABLE foo (bar TINYBLOB)");
+    mysql_and_generic().verified_stmt("CREATE TABLE foo (bar MEDIUMBLOB)");
+    mysql_and_generic().verified_stmt("CREATE TABLE foo (bar TINYTEXT)");
+    mysql_and_generic().verified_stmt("CREATE TABLE foo (bar MEDIUMTEXT)");
+    mysql_and_generic().verified_stmt("CREATE TABLE foo (bar LONGTEXT)");
+}