Merge pull request #54 from nickolay/windows-newlines

Support \r and \r\n line breaks in tokenizer

src/sqltokenizer.rs

@@ -260,16 +260,15 @@ impl<'a> Tokenizer<'a> {
         //println!("next_token: {:?}", chars.peek());
         match chars.peek() {
             Some(&ch) => match ch {
-                ' ' => {
+                ' ' => self.consume_and_return(chars, Token::Whitespace(Whitespace::Space)),
-                    chars.next();
+                '\t' => self.consume_and_return(chars, Token::Whitespace(Whitespace::Tab)),
-                    Ok(Some(Token::Whitespace(Whitespace::Space)))
+                '\n' => self.consume_and_return(chars, Token::Whitespace(Whitespace::Newline)),
-                }
+                '\r' => {
-                '\t' => {
+                    // Emit a single Whitespace::Newline token for \r and \r\n
-                    chars.next();
-                    Ok(Some(Token::Whitespace(Whitespace::Tab)))
-                }
-                '\n' => {
                     chars.next();
+                    if let Some('\n') = chars.peek() {
+                        chars.next();
+                    }
                     Ok(Some(Token::Whitespace(Whitespace::Newline)))
                 }
                 'N' => {
@@ -749,6 +748,26 @@ mod tests {
         compare(expected, tokens);
     }
 
+    #[test]
+    fn tokenize_newlines() {
+        let sql = String::from("line1\nline2\rline3\r\nline4\r");
+
+        let dialect = GenericSqlDialect {};
+        let mut tokenizer = Tokenizer::new(&dialect, &sql);
+        let tokens = tokenizer.tokenize().unwrap();
+        let expected = vec![
+            Token::make_word("line1", None),
+            Token::Whitespace(Whitespace::Newline),
+            Token::make_word("line2", None),
+            Token::Whitespace(Whitespace::Newline),
+            Token::make_word("line3", None),
+            Token::Whitespace(Whitespace::Newline),
+            Token::make_word("line4", None),
+            Token::Whitespace(Whitespace::Newline),
+        ];
+        compare(expected, tokens);
+    }
+
     fn compare(expected: Vec<Token>, actual: Vec<Token>) {
         //println!("------------------------------");
         //println!("tokens   = {:?}", actual);