Only allow newlines in triple-quoted literals.

Fixes #114.

Author: RubenVerborgh

lib/N3Lexer.js

@@ -49,8 +49,8 @@ N3Lexer.prototype = {
 
   _iri: /^<((?:[^ <>{}\\]|\\[uU])+)>[ \t]*/, // IRI with escape sequences; needs sanity check after unescaping
   _unescapedIri: /^<([^\x00-\x20<>\\"\{\}\|\^\`]*)>[ \t]*/, // IRI without escape sequences; no unescaping
-  _unescapedString: /^"[^"\\]+"(?=[^"\\])/, // non-empty string without escape sequences
-  _singleQuotedString: /^"[^"\\]*(?:\\.[^"\\]*)*"(?=[^"\\])|^'[^'\\]*(?:\\.[^'\\]*)*'(?=[^'\\])/,
+  _unescapedString: /^"[^"\\\r\n]+"/, // non-empty string without escape sequences
+  _singleQuotedString: /^"(?:[^"\\\r\n]|\\.)*"(?=[^"])|^'(?:[^'\\\r\n]|\\.)*'(?=[^'])/,
   _tripleQuotedString: /^""("[^"\\]*(?:(?:\\.|"(?!""))[^"\\]*)*")""|^''('[^'\\]*(?:(?:\\.|'(?!''))[^'\\]*)*')''/,
   _langcode: /^@([a-z]+(?:-[a-z0-9]+)*)(?=[^a-z0-9\-])/i,
   _prefix: /^((?:[A-Za-z\xc0-\xd6\xd8-\xf6\xf8-\u02ff\u0370-\u037d\u037f-\u1fff\u200c\u200d\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff\uf900-\ufdcf\ufdf0-\ufffd]|[\ud800-\udb7f][\udc00-\udfff])(?:\.?[\-0-9A-Z_a-z\xb7\xc0-\xd6\xd8-\xf6\xf8-\u037d\u037f-\u1fff\u200c\u200d\u203f\u2040\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff\uf900-\ufdcf\ufdf0-\ufffd]|[\ud800-\udb7f][\udc00-\udfff])*)?:(?=[#\s<])/,

test/N3Lexer-test.js

@@ -265,6 +265,14 @@ describe('N3Lexer', function () {
                      { type: 'literal', value: '"string"', line: 1 },
                      { type: 'eof', line: 1 }));
 
+    it('should not tokenize a quoted string literal with a newline',
+      shouldNotTokenize('"abc\ndef" ',
+                        'Unexpected ""abc" on line 1.'));
+
+    it('should not tokenize a quoted string literal with a carriage return',
+      shouldNotTokenize('"abc\rdef" ',
+                        'Unexpected ""abc" on line 1.'));
+
     it('should tokenize a triple quoted string literal',
       shouldTokenize('"""string"""',
                      { type: 'literal', value: '"string"', line: 1 },
@@ -325,6 +333,14 @@ describe('N3Lexer', function () {
                      { type: 'literal', value: '"string"', line: 1 },
                      { type: 'eof', line: 1 }));
 
+    it('should not tokenize a single-quoted string literal with a newline',
+      shouldNotTokenize("'abc\ndef' ",
+                        'Unexpected "\'abc" on line 1.'));
+
+    it('should not tokenize a single-quoted string literal with a carriage return',
+      shouldNotTokenize("'abc\rdef' ",
+                        'Unexpected "\'abc" on line 1.'));
+
     it('should tokenize a triple single-quoted string literal with quotes newlines inside',
       shouldTokenize("'''st'r\ni''ng'''",
                      { type: 'literal', value: '"st\'r\ni\'\'ng"', line: 1 },