Merge pull request #8 from AikidoSec/national-strings

National strings: check if dialect supports backslash escape

Author: willem-delbare

src/test_utils.rs

@@ -33,7 +33,7 @@ use core::fmt::Debug;
 
 use crate::dialect::*;
 use crate::parser::{Parser, ParserError};
-use crate::tokenizer::Tokenizer;
+use crate::tokenizer::{Token, Tokenizer};
 use crate::{ast::*, parser::ParserOptions};
 
 #[cfg(test)]
@@ -215,6 +215,18 @@ impl TestedDialects {
     pub fn verified_expr(&self, sql: &str) -> Expr {
         self.expr_parses_to(sql, sql)
     }
+
+    /// Check that the tokenizer returns the expected tokens for the given SQL.
+    pub fn tokenizes_to(&self, sql: &str, expected: Vec<Token>) {
+        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);
+        });
+    }
 }
 
 /// Returns all available dialects.

src/tokenizer.rs

@@ -776,7 +776,10 @@ impl<'a> Tokenizer<'a> {
                     match chars.peek() {
                         Some('\'') => {
                             // N'...' - a <national character string literal>
-                            let s = self.tokenize_single_quoted_string(chars, '\'', true)?;
+                            let backslash_escape =
+                                self.dialect.supports_string_literal_backslash_escape();
+                            let s =
+                                self.tokenize_single_quoted_string(chars, '\'', backslash_escape)?;
                             Ok(Some(Token::NationalStringLiteral(s)))
                         }
                         _ => {
@@ -1890,6 +1893,7 @@ mod tests {
     use crate::dialect::{
         BigQueryDialect, ClickHouseDialect, HiveDialect, MsSqlDialect, MySqlDialect, SQLiteDialect,
     };
+    use crate::test_utils::all_dialects_where;
     use core::fmt::Debug;
 
     #[test]
@@ -3214,4 +3218,30 @@ mod tests {
         let expected = vec![Token::SingleQuotedString("''".to_string())];
         compare(expected, tokens);
     }
+
+    #[test]
+    fn test_national_strings_backslash_escape_not_supported() {
+        all_dialects_where(|dialect| !dialect.supports_string_literal_backslash_escape())
+            .tokenizes_to(
+                "select n'''''\\'",
+                vec![
+                    Token::make_keyword("select"),
+                    Token::Whitespace(Whitespace::Space),
+                    Token::NationalStringLiteral("''\\".to_string()),
+                ],
+            );
+    }
+
+    #[test]
+    fn test_national_strings_backslash_escape_supported() {
+        all_dialects_where(|dialect| dialect.supports_string_literal_backslash_escape())
+            .tokenizes_to(
+                "select n'''''\\''",
+                vec![
+                    Token::make_keyword("select"),
+                    Token::Whitespace(Whitespace::Space),
+                    Token::NationalStringLiteral("'''".to_string()),
+                ],
+            );
+    }
 }