reject invalid hex/unicode escape sequences instead of letting them crash the compiler

Cython/Compiler/Parsing.py

@@ -782,13 +782,19 @@ def p_string_literal(s, kind_override=None):
                 elif c == u'\n':
                     pass
                 elif c == u'x':
+                    if len(systr) == 4:
                         chars.append_charval( int(systr[2:], 16) )
+                    else:
+                        s.error("Invalid hex escape '%s'" % systr, pos=s.position())
                 elif c in u'Uu':
                     if kind in ('u', ''):
+                        if len(systr) in (6,10):
                             chrval = int(systr[2:], 16)
                             if chrval > 1114111: # sys.maxunicode:
                                 s.error("Invalid unicode escape '%s'" % systr,
                                         pos = pos)
+                        else:
+                            s.error("Invalid unicode escape '%s'" % systr, pos=s.position())
                     else:
                         # unicode escapes in plain byte strings are not unescaped
                         chrval = None

tests/errors/invalid_hex_escape0.pyx

+
+'\x'
+
+_ERRORS = '''
+2:1: Invalid hex escape '\x'
+'''

tests/errors/invalid_hex_escape1.pyx

+
+'\x1'
+
+_ERRORS = '''
+2:1: Invalid hex escape '\x'
+'''

tests/errors/invalid_uescape.pyx

+
+u'\uXYZ'
+
+_ERRORS = '''
+2:2: Invalid unicode escape '\u'
+'''

tests/errors/invalid_uescape0.pyx

+
+u'\u'
+
+_ERRORS = '''
+2:1: Invalid unicode escape '\u'
+'''

tests/errors/invalid_uescape2.pyx

+
+u'\u12'
+
+_ERRORS = '''
+2:1: Invalid unicode escape '\u'
+'''

tests/run/strliterals.pyx

@@ -125,6 +125,13 @@ __doc__ = ur"""
     >>> len(uresc)
     9
 
+    >>> bytes_uescape
+    b'\\u1234\\U12345678\\u\\u1\\u12\\uX'
+    >>> bytes_uescape == b'\\u1234\\U12345678\\u\\u1\\u12\\uX'
+    True
+    >>> len(bytes_uescape)
+    28
+
     >>> newlines == "Aaa\n"
     True
     
@@ -165,6 +172,8 @@ sresc =  r'\12\'\"\\'
 bresc = br'\12\'\"\\'
 uresc = ur'\12\'\"\\'
 
+bytes_uescape = b'\u1234\U12345678\u\u1\u12\uX'
+
 newlines = "Aaa\n"
 
 # T640, long literals with escapes