Parse SET NAMES syntax in Postgres (apache#1752)

Author: mvzink

src/ast/mod.rs

@@ -2956,10 +2956,8 @@ pub enum Statement {
     /// ```sql
     /// SET NAMES 'charset_name' [COLLATE 'collation_name']
     /// ```
-    ///
-    /// Note: this is a MySQL-specific statement.
     SetNames {
-        charset_name: String,
+        charset_name: Ident,
         collation_name: Option<String>,
     },
     /// ```sql
@@ -4684,8 +4682,7 @@ impl fmt::Display for Statement {
                 charset_name,
                 collation_name,
             } => {
-                f.write_str("SET NAMES ")?;
-                f.write_str(charset_name)?;
+                write!(f, "SET NAMES {}", charset_name)?;
 
                 if let Some(collation) = collation_name {
                     f.write_str(" COLLATE ")?;

src/dialect/generic.rs

@@ -155,4 +155,8 @@ impl Dialect for GenericDialect {
     fn supports_match_against(&self) -> bool {
         true
     }
+
+    fn supports_set_names(&self) -> bool {
+        true
+    }
 }

src/dialect/mod.rs

@@ -980,6 +980,16 @@ pub trait Dialect: Debug + Any {
     fn supports_order_by_all(&self) -> bool {
         false
     }
+
+    /// Returns true if the dialect supports `SET NAMES <charset_name> [COLLATE <collation_name>]`.
+    ///
+    /// - [MySQL](https://dev.mysql.com/doc/refman/8.4/en/set-names.html)
+    /// - [Postgres](https://www.postgresql.org/docs/17/sql-set.html)
+    ///
+    /// Note: Postgres doesn't support the `COLLATE` clause, but we permissively parse it anyway.
+    fn supports_set_names(&self) -> bool {
+        false
+    }
 }
 
 /// This represents the operators for which precedence must be defined

src/dialect/mysql.rs

@@ -137,6 +137,10 @@ impl Dialect for MySqlDialect {
     fn supports_match_against(&self) -> bool {
         true
     }
+
+    fn supports_set_names(&self) -> bool {
+        true
+    }
 }
 
 /// `LOCK TABLES`

src/dialect/postgresql.rs

@@ -254,4 +254,8 @@ impl Dialect for PostgreSqlDialect {
     fn supports_geometric_types(&self) -> bool {
         true
     }
+
+    fn supports_set_names(&self) -> bool {
+        true
+    }
 }

src/parser/mod.rs

@@ -10962,14 +10962,14 @@ impl<'a> Parser<'a> {
             OneOrManyWithParens::One(self.parse_object_name(false)?)
         };
 
-        if matches!(&variables, OneOrManyWithParens::One(variable) if variable.to_string().eq_ignore_ascii_case("NAMES")
-            && dialect_of!(self is MySqlDialect | GenericDialect))
-        {
+        let names = matches!(&variables, OneOrManyWithParens::One(variable) if variable.to_string().eq_ignore_ascii_case("NAMES"));
+
+        if names && self.dialect.supports_set_names() {
             if self.parse_keyword(Keyword::DEFAULT) {
                 return Ok(Statement::SetNamesDefault {});
             }
 
-            let charset_name = self.parse_literal_string()?;
+            let charset_name = self.parse_identifier()?;
             let collation_name = if self.parse_one_of_keywords(&[Keyword::COLLATE]).is_some() {
                 Some(self.parse_literal_string()?)
             } else {

tests/sqlparser_common.rs

@@ -14631,3 +14631,11 @@ fn parse_array_type_def_with_brackets() {
     dialects.verified_stmt("SELECT x::INT[]");
     dialects.verified_stmt("SELECT STRING_TO_ARRAY('1,2,3', ',')::INT[3]");
 }
+
+#[test]
+fn parse_set_names() {
+    let dialects = all_dialects_where(|d| d.supports_set_names());
+    dialects.verified_stmt("SET NAMES 'UTF8'");
+    dialects.verified_stmt("SET NAMES 'utf8'");
+    dialects.verified_stmt("SET NAMES UTF8 COLLATE bogus");
+}

tests/sqlparser_mysql.rs

@@ -2696,7 +2696,7 @@ fn parse_set_names() {
     assert_eq!(
         stmt,
         Statement::SetNames {
-            charset_name: "utf8mb4".to_string(),
+            charset_name: "utf8mb4".into(),
             collation_name: None,
         }
     );
@@ -2705,7 +2705,7 @@ fn parse_set_names() {
     assert_eq!(
         stmt,
         Statement::SetNames {
-            charset_name: "utf8mb4".to_string(),
+            charset_name: "utf8mb4".into(),
             collation_name: Some("bogus".to_string()),
         }
     );
@@ -2716,7 +2716,7 @@ fn parse_set_names() {
     assert_eq!(
         stmt,
         vec![Statement::SetNames {
-            charset_name: "utf8mb4".to_string(),
+            charset_name: "utf8mb4".into(),
             collation_name: Some("bogus".to_string()),
         }]
     );