Merge pull request #9 from AikidoSec/string-escape-constant

Only support string escape constant for Postgres, Redshift and generic dialect

Author: willem-delbare

src/dialect/generic.rs

@@ -111,4 +111,8 @@ impl Dialect for GenericDialect {
     fn supports_nested_comments(&self) -> bool {
         true
     }
+
+    fn supports_string_escape_constant(&self) -> bool {
+        true
+    }
 }

src/dialect/mod.rs

@@ -573,6 +573,13 @@ pub trait Dialect: Debug + Any {
     fn supports_nested_comments(&self) -> bool {
         false
     }
+
+    /// Returns true if this dialect supports the E'...' syntax for string literals
+    ///
+    /// Postgres: <https://www.postgresql.org/docs/current/sql-syntax-lexical.html#SQL-SYNTAX-STRINGS-ESCAPE>
+    fn supports_string_escape_constant(&self) -> bool {
+        false
+    }
 }
 
 /// This represents the operators for which precedence must be defined

src/dialect/postgresql.rs

@@ -192,6 +192,10 @@ impl Dialect for PostgreSqlDialect {
     fn supports_nested_comments(&self) -> bool {
         true
     }
+
+    fn supports_string_escape_constant(&self) -> bool {
+        true
+    }
 }
 
 pub fn parse_comment(parser: &mut Parser) -> Result<Statement, ParserError> {

src/dialect/redshift.rs

@@ -68,4 +68,8 @@ impl Dialect for RedshiftSqlDialect {
     fn supports_connect_by(&self) -> bool {
         true
     }
+
+    fn supports_string_escape_constant(&self) -> bool {
+        true
+    }
 }

src/test_utils.rs

@@ -218,13 +218,17 @@ impl TestedDialects {
 
     /// Check that the tokenizer returns the expected tokens for the given SQL.
     pub fn tokenizes_to(&self, sql: &str, expected: Vec<Token>) {
+        if self.dialects.is_empty() {
+            panic!("No dialects to test");
+        }
+
         self.dialects.iter().for_each(|dialect| {
             let mut tokenizer = Tokenizer::new(&**dialect, sql);
             if let Some(options) = &self.options {
                 tokenizer = tokenizer.with_unescape(options.unescape);
             }
             let tokens = tokenizer.tokenize().unwrap();
-            assert_eq!(expected, tokens);
+            assert_eq!(expected, tokens, "Tokenized differently for {:?}", dialect);
         });
     }
 }

src/tokenizer.rs

@@ -790,7 +790,7 @@ impl<'a> Tokenizer<'a> {
                     }
                 }
                 // PostgreSQL accepts "escape" string constants, which are an extension to the SQL standard.
-                x @ 'e' | x @ 'E' => {
+                x @ 'e' | x @ 'E' if self.dialect.supports_string_escape_constant() => {
                     let starting_loc = chars.location();
                     chars.next(); // consume, to check the next char
                     match chars.peek() {
@@ -3244,4 +3244,48 @@ mod tests {
                 ],
             );
     }
+
+    #[test]
+    fn test_string_escape_constant_not_supported() {
+        all_dialects_where(|dialect| !dialect.supports_string_escape_constant()).tokenizes_to(
+            "select e'...'",
+            vec![
+                Token::make_keyword("select"),
+                Token::Whitespace(Whitespace::Space),
+                Token::make_word("e", None),
+                Token::SingleQuotedString("...".to_string()),
+            ],
+        );
+
+        all_dialects_where(|dialect| !dialect.supports_string_escape_constant()).tokenizes_to(
+            "select E'...'",
+            vec![
+                Token::make_keyword("select"),
+                Token::Whitespace(Whitespace::Space),
+                Token::make_word("E", None),
+                Token::SingleQuotedString("...".to_string()),
+            ],
+        );
+    }
+
+    #[test]
+    fn test_string_escape_constant_supported() {
+        all_dialects_where(|dialect| dialect.supports_string_escape_constant()).tokenizes_to(
+            "select e'\\''",
+            vec![
+                Token::make_keyword("select"),
+                Token::Whitespace(Whitespace::Space),
+                Token::EscapedStringLiteral("'".to_string()),
+            ],
+        );
+
+        all_dialects_where(|dialect| dialect.supports_string_escape_constant()).tokenizes_to(
+            "select E'\\''",
+            vec![
+                Token::make_keyword("select"),
+                Token::Whitespace(Whitespace::Space),
+                Token::EscapedStringLiteral("'".to_string()),
+            ],
+        );
+    }
 }