diff --git a/src/test_utils.rs b/src/test_utils.rs index 914be7d9f..51e4fd748 100644 --- a/src/test_utils.rs +++ b/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)] @@ -237,6 +237,18 @@ 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) { + 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. diff --git a/src/tokenizer.rs b/src/tokenizer.rs index 39ca84c9f..08e233b66 100644 --- a/src/tokenizer.rs +++ b/src/tokenizer.rs @@ -971,7 +971,10 @@ impl<'a> Tokenizer<'a> { match chars.peek() { Some('\'') => { // N'...' - a - 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))) } _ => { @@ -2155,6 +2158,7 @@ mod tests { use crate::dialect::{ BigQueryDialect, ClickHouseDialect, HiveDialect, MsSqlDialect, MySqlDialect, SQLiteDialect, }; + use crate::test_utils::all_dialects_where; use core::fmt::Debug; #[test] @@ -3543,4 +3547,30 @@ mod tests { ]; 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()), + ], + ); + } }