Allow unescaped U+2028 and U+2029 in strings

json5/json5.g

@@ -36,11 +36,13 @@ string         = squote sqchar*:cs squote             -> join('', cs)
 sqchar         = bslash esc_char:c                    -> c
                | bslash eol                           -> ''
                | ~bslash ~squote ~eol anything:c      -> c
+               | '\u2028' | '\u2029'
                | ~?(_strict) '\x00'..'\x1f'
 
 dqchar         = bslash esc_char:c                    -> c
                | bslash eol                           -> ''
                | ~bslash ~dquote ~eol anything:c      -> c
+               | '\u2028' | '\u2029'
                | ~?( _strict ) '\x00'..'\x1f'
 
 bslash         = '\u005C'

json5/parser.py

@@ -480,6 +480,8 @@ def _sqchar_(self):
                 self._sqchar__c1_,
                 self._sqchar__c2_,
                 self._sqchar__c3_,
+                self._sqchar__c4_,
+                self._sqchar__c5_,
             ]
         )
 
@@ -511,14 +513,20 @@ def _sqchar__c2_(self):
         self._pop('sqchar__c2')
 
     def _sqchar__c3_(self):
+        self._ch('\u2028')
+
+    def _sqchar__c4_(self):
+        self._ch('\u2029')
+
+    def _sqchar__c5_(self):
         self._seq(
             [
-                lambda: self._not(self._sqchar__c3__s0_n_),
+                lambda: self._not(self._sqchar__c5__s0_n_),
                 lambda: self._range('\x00', '\x1f'),
             ]
         )
 
-    def _sqchar__c3__s0_n_(self):
+    def _sqchar__c5__s0_n_(self):
         v = self._get('_strict')
         if v:
             self._succeed(v)
@@ -532,6 +540,8 @@ def _dqchar_(self):
                 self._dqchar__c1_,
                 self._dqchar__c2_,
                 self._dqchar__c3_,
+                self._dqchar__c4_,
+                self._dqchar__c5_,
             ]
         )
 
@@ -563,14 +573,20 @@ def _dqchar__c2_(self):
         self._pop('dqchar__c2')
 
     def _dqchar__c3_(self):
+        self._ch('\u2028')
+
+    def _dqchar__c4_(self):
+        self._ch('\u2029')
+
+    def _dqchar__c5_(self):
         self._seq(
             [
-                lambda: self._not(self._dqchar__c3__s0_n_),
+                lambda: self._not(self._dqchar__c5__s0_n_),
                 lambda: self._range('\x00', '\x1f'),
             ]
         )
 
-    def _dqchar__c3__s0_n_(self):
+    def _dqchar__c5__s0_n_(self):
         v = self._get('_strict')
         if v:
             self._succeed(v)

tests/lib_test.py

@@ -325,9 +325,15 @@ def test_strings(self):
         self.check('"foo\\\nbar"', 'foobar')
         self.check("'foo\\\nbar'", 'foobar')
 
+        # U+2028 and U+2029 are allowed unescaped, as in JSON.
+        self.check('"foo
bar
"', 'foo
bar
')
+        self.check("'foo
bar
'", 'foo
bar
')
+        self.check('"foo\\
bar"', 'foobar')
+
         # unterminated string literals.
         self.check_fail('"\n')
         self.check_fail("'\n")
+        self.check_fail('"\r"')
 
         # bad hex literals
         self.check_fail("'\\x0'")