Ticket #4594: reverse_helper_cleaned.txt

Index: /home/tobryan1/workspace/django/django/core/urlresolvers.py
===================================================================
--- /home/tobryan1/workspace/django/django/core/urlresolvers.py (revision 5478)
+++ /home/tobryan1/workspace/django/django/core/urlresolvers.py (working copy)
@@ -27,6 +27,18 @@
         return callback, ''
     return callback[:dot], callback[dot+1:]
 
+ESCAPE_CHARS = re.compile(r'([(){}|[\].^$*+?\\])')
+
+def escape(s):
+    """
+    Escapes the characters that will get messed up later
+    """
+    return ESCAPE_CHARS.sub(r'\\\1', s)
+    
+
+CHAR_CLASS_NEEDS_BACKSLASH = re.compile(r'\[([(){}|[\].^$*+?])\]')
+CHAR_CLASS_NO_BACKSLASH = re.compile(r'\[(.)\]')
+
 def reverse_helper(regex, *args, **kwargs):
     """
     Does a "reverse" lookup -- returns the URL for the given args/kwargs.
@@ -39,12 +51,68 @@
         'places/3/'
         >>> reverse_helper(re.compile('^people/(?P<state>\w\w)/(\w+)/$'), 'adrian', state='il')
         'people/il/adrian/'
-
+        >>> reverse_helper(re.compile(r'^prices/less_than_\$(?P<price>\d+)/$'), price='10')
+        'prices/less_than_$10/'
+        >>> reverse_helper(re.compile(r'^prices/less_than_[$](?P<price>\d+)/$'), price='10')
+        'prices/less_than_$10/'
+        >>> reverse_helper(re.compile(r'^headlines/(?P<year>\d+)\.(?P<month>\d+)\.(?P<day>\d+)/$'), year=2007, month=5, day=21)
+        'headlines/2007.5.21/'
+        >>> reverse_helper(re.compile(r'^priests/(?P<name>\w+)\+/$'), name='maynard')
+        'priests/maynard+/'
+        >>> reverse_helper(re.compile(r'^windows_path/(?P<drive_name>[A-Z]):\\\\(?P<path>.+)/$'), drive_name='C', path=r'Documents and Settings\\spam')
+        'windows_path/C:\\\\Documents and Settings\\\\spam/'
+        >>> reverse_helper(re.compile(r'\\Aexpr\\\\b/expr2\\b\\\\Z/$'))
+        'expr\\\\b/expr2\\\\Z/'
+        >>> reverse_helper(re.compile(r'^(?P<name>[^/]+)/\\d+/$'), name='john')
+        Traceback (most recent call last):
+        ...
+        NoReverseMatch: \d must be replaced by an argument in reverse lookup
+        
     Raises NoReverseMatch if the args/kwargs aren't valid for the regex.
-    """
+    """    
     # TODO: Handle nested parenthesis in the following regex.
-    result = re.sub(r'\(([^)]+)\)', MatchChecker(args, kwargs), regex.pattern)
-    return result.replace('^', '').replace('$', '')
+    result = re.sub(r'\(([^)]+)\)', lambda m: escape(MatchChecker(args, kwargs)(m)), regex.pattern)
+    # TODO: octal characters make things even more complicated
+    # you can use a single character class to avoid escaping, e.g. [$] or [.].
+    # normalize to backslash followed by character
+    result = CHAR_CLASS_NEEDS_BACKSLASH.sub(r'\\\1', result)
+    # you can put a single character in brackets (though why you would is
+    # beyond me); removes the brackets
+    result = CHAR_CLASS_NO_BACKSLASH.sub(r'\1', result)
+    # \A, \Z, \b, and \B match the empty string and should be removed, but
+    # only if preceded by an odd number of backslashes, otherwise the backslash
+    # right before is actually the second backslash in the backslash escape \\
+    def delete_if_slashes_odd(m):
+        odd_slash_match = re.match(r'^(\\\\)*\\$', m.group('slashes'))
+        if odd_slash_match:
+            return odd_slash_match.group(1)
+        else:
+            return m.group(0)
+    result = re.sub(r'(?P<slashes>\\+)(?P<char>[AbBZ])', delete_if_slashes_odd, result)
+    # ^ and $ match the empty string and should be removed, but only if
+    # preceded by an even number of backslashes (including none), otherwise the
+    # backslash right before is escaping the literal \^ or \$
+    def delete_if_slashes_even(m):
+        even_slash_match = re.match(r'^(\\\\)*$', m.group('slashes'))
+        if even_slash_match:
+            return m.group('slashes')
+        else:
+            return m.group(0)
+    result = re.sub(r'(?P<slashes>\\*)(?P<char>[$^])', delete_if_slashes_even, result)
+    # many characters are preceded by backslashes in regexes if the literal
+    # character is meant; as we go to a string, the backslash should go away.
+    # We should never find character classes that don't have a single
+    # replacement character. These are \number (the group matching expression),
+    # \d, \D, \s, \S, \w, and \W. If we find these at this point, we raise
+    # an exception.
+    def drop_backslash_if_valid(m):
+        char = m.group(1)
+        if re.match(r'[\ddDsSwW]', char):
+            raise NoReverseMatch(r'\%s must be replaced by an argument in reverse lookup' % char)
+        else:
+            return char
+    result = re.sub(r'\\([[\]{}()^$*+?.\\|\ddDsSwW])', drop_backslash_if_valid, result)
+    return result
 
 class MatchChecker(object):
     "Class used in reverse RegexURLPattern lookup."