Support CHARSET and ENGINE clauses on CREATE TABLE for mysql (#392)

* Add support for SET and ENUM types

See https://dev.mysql.com/doc/refman/8.0/en/set.html
and https://dev.mysql.com/doc/refman/8.0/en/enum.html

* Add support for mysql ENGINE and DEFAULT CHARSET in CREATE TABLE

See https://dev.mysql.com/doc/refman/8.0/en/create-table.html

* Add support for COMMENT and CHARACTER SET field attributes

See https://dev.mysql.com/doc/refman/8.0/en/create-table.html

Author: antialize

src/ast/data_type.rs

@@ -11,14 +11,16 @@
 // limitations under the License.
 
 #[cfg(not(feature = "std"))]
-use alloc::boxed::Box;
+use alloc::{boxed::Box, string::String, vec::Vec};
 use core::fmt;
 
 #[cfg(feature = "serde")]
 use serde::{Deserialize, Serialize};
 
 use crate::ast::ObjectName;
 
+use super::value::escape_single_quote_string;
+
 /// SQL data types
 #[derive(Debug, Clone, PartialEq, Eq, Hash)]
 #[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
@@ -75,6 +77,10 @@ pub enum DataType {
     Custom(ObjectName),
     /// Arrays
     Array(Box<DataType>),
+    /// Enums
+    Enum(Vec<String>),
+    /// Set
+    Set(Vec<String>),
 }
 
 impl fmt::Display for DataType {
@@ -116,6 +122,26 @@ impl fmt::Display for DataType {
             DataType::Bytea => write!(f, "BYTEA"),
             DataType::Array(ty) => write!(f, "{}[]", ty),
             DataType::Custom(ty) => write!(f, "{}", ty),
+            DataType::Enum(vals) => {
+                write!(f, "ENUM(")?;
+                for (i, v) in vals.iter().enumerate() {
+                    if i != 0 {
+                        write!(f, ", ")?;
+                    }
+                    write!(f, "'{}'", escape_single_quote_string(v))?;
+                }
+                write!(f, ")")
+            }
+            DataType::Set(vals) => {
+                write!(f, "SET(")?;
+                for (i, v) in vals.iter().enumerate() {
+                    if i != 0 {
+                        write!(f, ", ")?;
+                    }
+                    write!(f, "'{}'", escape_single_quote_string(v))?;
+                }
+                write!(f, ")")
+            }
         }
     }
 }

src/ast/ddl.rs

@@ -14,12 +14,13 @@
 //! (commonly referred to as Data Definition Language, or DDL)
 
 #[cfg(not(feature = "std"))]
-use alloc::{boxed::Box, string::ToString, vec::Vec};
+use alloc::{boxed::Box, string::String, string::ToString, vec::Vec};
 use core::fmt;
 
 #[cfg(feature = "serde")]
 use serde::{Deserialize, Serialize};
 
+use crate::ast::value::escape_single_quote_string;
 use crate::ast::{display_comma_separated, display_separated, DataType, Expr, Ident, ObjectName};
 use crate::tokenizer::Token;
 
@@ -338,7 +339,9 @@ pub enum ColumnOption {
     /// `DEFAULT <restricted-expr>`
     Default(Expr),
     /// `{ PRIMARY KEY | UNIQUE }`
-    Unique { is_primary: bool },
+    Unique {
+        is_primary: bool,
+    },
     /// A referential integrity constraint (`[FOREIGN KEY REFERENCES
     /// <foreign_table> (<referred_columns>)
     /// { [ON DELETE <referential_action>] [ON UPDATE <referential_action>] |
@@ -356,6 +359,8 @@ pub enum ColumnOption {
     /// - MySQL's `AUTO_INCREMENT` or SQLite's `AUTOINCREMENT`
     /// - ...
     DialectSpecific(Vec<Token>),
+    CharacterSet(ObjectName),
+    Comment(String),
 }
 
 impl fmt::Display for ColumnOption {
@@ -388,6 +393,8 @@ impl fmt::Display for ColumnOption {
             }
             Check(expr) => write!(f, "CHECK ({})", expr),
             DialectSpecific(val) => write!(f, "{}", display_separated(val, " ")),
+            CharacterSet(n) => write!(f, "CHARACTER SET {}", n),
+            Comment(v) => write!(f, "COMMENT '{}'", escape_single_quote_string(v)),
         }
     }
 }

src/ast/mod.rs

@@ -734,6 +734,8 @@ pub enum Statement {
         query: Option<Box<Query>>,
         without_rowid: bool,
         like: Option<ObjectName>,
+        engine: Option<String>,
+        default_charset: Option<String>,
     },
     /// SQLite's `CREATE VIRTUAL TABLE .. USING <module_name> (<module_args>)`
     CreateVirtualTable {
@@ -1182,6 +1184,8 @@ impl fmt::Display for Statement {
                 query,
                 without_rowid,
                 like,
+                default_charset,
+                engine,
             } => {
                 // We want to allow the following options
                 // Empty column list, allowed by PostgreSQL:
@@ -1307,6 +1311,12 @@ impl fmt::Display for Statement {
                 if let Some(query) = query {
                     write!(f, " AS {}", query)?;
                 }
+                if let Some(engine) = engine {
+                    write!(f, " ENGINE={}", engine)?;
+                }
+                if let Some(default_charset) = default_charset {
+                    write!(f, " DEFAULT CHARSET={}", default_charset)?;
+                }
                 Ok(())
             }
             Statement::CreateVirtualTable {

src/keywords.rs

@@ -118,6 +118,7 @@ define_keywords!(
     CHAR,
     CHARACTER,
     CHARACTER_LENGTH,
+    CHARSET,
     CHAR_LENGTH,
     CHECK,
     CLOB,
@@ -194,6 +195,8 @@ define_keywords!(
     END_EXEC = "END-EXEC",
     END_FRAME,
     END_PARTITION,
+    ENGINE,
+    ENUM,
     EQUALS,
     ERROR,
     ESCAPE,

src/parser.rs

@@ -1522,6 +1522,8 @@ impl<'a> Parser<'a> {
             query: None,
             without_rowid: false,
             like: None,
+            default_charset: None,
+            engine: None,
         })
     }
 
@@ -1696,6 +1698,26 @@ impl<'a> Parser<'a> {
             None
         };
 
+        let engine = if self.parse_keyword(Keyword::ENGINE) {
+            self.expect_token(&Token::Eq)?;
+            match self.next_token() {
+                Token::Word(w) => Some(w.value),
+                unexpected => self.expected("identifier", unexpected)?,
+            }
+        } else {
+            None
+        };
+
+        let default_charset = if self.parse_keywords(&[Keyword::DEFAULT, Keyword::CHARSET]) {
+            self.expect_token(&Token::Eq)?;
+            match self.next_token() {
+                Token::Word(w) => Some(w.value),
+                unexpected => self.expected("identifier", unexpected)?,
+            }
+        } else {
+            None
+        };
+
         Ok(Statement::CreateTable {
             name: table_name,
             temporary,
@@ -1713,6 +1735,8 @@ impl<'a> Parser<'a> {
             query,
             without_rowid,
             like,
+            engine,
+            default_charset,
         })
     }
 
@@ -1778,8 +1802,15 @@ impl<'a> Parser<'a> {
     }
 
     pub fn parse_optional_column_option(&mut self) -> Result<Option<ColumnOption>, ParserError> {
-        if self.parse_keywords(&[Keyword::NOT, Keyword::NULL]) {
+        if self.parse_keywords(&[Keyword::CHARACTER, Keyword::SET]) {
+            Ok(Some(ColumnOption::CharacterSet(self.parse_object_name()?)))
+        } else if self.parse_keywords(&[Keyword::NOT, Keyword::NULL]) {
             Ok(Some(ColumnOption::NotNull))
+        } else if self.parse_keywords(&[Keyword::COMMENT]) {
+            match self.next_token() {
+                Token::SingleQuotedString(value, ..) => Ok(Some(ColumnOption::Comment(value))),
+                unexpected => self.expected("string", unexpected),
+            }
         } else if self.parse_keyword(Keyword::NULL) {
             Ok(Some(ColumnOption::Null))
         } else if self.parse_keyword(Keyword::DEFAULT) {
@@ -2301,6 +2332,8 @@ impl<'a> Parser<'a> {
                     let (precision, scale) = self.parse_optional_precision_scale()?;
                     Ok(DataType::Decimal(precision, scale))
                 }
+                Keyword::ENUM => Ok(DataType::Enum(self.parse_string_values()?)),
+                Keyword::SET => Ok(DataType::Set(self.parse_string_values()?)),
                 _ => {
                     self.prev_token();
                     let type_name = self.parse_object_name()?;
@@ -2311,6 +2344,23 @@ impl<'a> Parser<'a> {
         }
     }
 
+    pub fn parse_string_values(&mut self) -> Result<Vec<String>, ParserError> {
+        self.expect_token(&Token::LParen)?;
+        let mut values = Vec::new();
+        loop {
+            match self.next_token() {
+                Token::SingleQuotedString(value) => values.push(value),
+                unexpected => self.expected("a string", unexpected)?,
+            }
+            match self.next_token() {
+                Token::Comma => (),
+                Token::RParen => break,
+                unexpected => self.expected(", or }", unexpected)?,
+            }
+        }
+        Ok(values)
+    }
+
     /// Parse `AS identifier` (or simply `identifier` if it's not a reserved keyword)
     /// Some examples with aliases: `SELECT 1 foo`, `SELECT COUNT(*) AS cnt`,
     /// `SELECT ... FROM t1 foo, t2 bar`, `SELECT ... FROM (...) AS bar`

tests/sqlparser_mysql.rs

@@ -155,6 +155,93 @@ fn parse_create_table_auto_increment() {
     }
 }
 
+#[test]
+fn parse_create_table_set_enum() {
+    let sql = "CREATE TABLE foo (bar SET('a', 'b'), baz ENUM('a', 'b'))";
+    match mysql().verified_stmt(sql) {
+        Statement::CreateTable { name, columns, .. } => {
+            assert_eq!(name.to_string(), "foo");
+            assert_eq!(
+                vec![
+                    ColumnDef {
+                        name: Ident::new("bar"),
+                        data_type: DataType::Set(vec!["a".to_string(), "b".to_string()]),
+                        collation: None,
+                        options: vec![],
+                    },
+                    ColumnDef {
+                        name: Ident::new("baz"),
+                        data_type: DataType::Enum(vec!["a".to_string(), "b".to_string()]),
+                        collation: None,
+                        options: vec![],
+                    }
+                ],
+                columns
+            );
+        }
+        _ => unreachable!(),
+    }
+}
+
+#[test]
+fn parse_create_table_engine_default_charset() {
+    let sql = "CREATE TABLE foo (id INT(11)) ENGINE=InnoDB DEFAULT CHARSET=utf8mb3";
+    match mysql().verified_stmt(sql) {
+        Statement::CreateTable {
+            name,
+            columns,
+            engine,
+            default_charset,
+            ..
+        } => {
+            assert_eq!(name.to_string(), "foo");
+            assert_eq!(
+                vec![ColumnDef {
+                    name: Ident::new("id"),
+                    data_type: DataType::Int(Some(11)),
+                    collation: None,
+                    options: vec![],
+                },],
+                columns
+            );
+            assert_eq!(engine, Some("InnoDB".to_string()));
+            assert_eq!(default_charset, Some("utf8mb3".to_string()));
+        }
+        _ => unreachable!(),
+    }
+}
+
+#[test]
+fn parse_create_table_comment_character_set() {
+    let sql = "CREATE TABLE foo (s TEXT CHARACTER SET utf8mb4 COMMENT 'comment')";
+    match mysql().verified_stmt(sql) {
+        Statement::CreateTable { name, columns, .. } => {
+            assert_eq!(name.to_string(), "foo");
+            assert_eq!(
+                vec![ColumnDef {
+                    name: Ident::new("s"),
+                    data_type: DataType::Text,
+                    collation: None,
+                    options: vec![
+                        ColumnOptionDef {
+                            name: None,
+                            option: ColumnOption::CharacterSet(ObjectName(vec![Ident::new(
+                                "utf8mb4"
+                            )]))
+                        },
+                        ColumnOptionDef {
+                            name: None,
+                            option: ColumnOption::Comment("comment".to_string())
+                        }
+                    ],
+                },],
+                columns
+            );
+        }
+        _ => unreachable!(),
+    }
+}
+
 #[test]
 fn parse_quote_identifiers() {
     let sql = "CREATE TABLE `PRIMARY` (`BEGIN` INT PRIMARY KEY)";