py3: fix conversion of text to bytes in LDIF parser

src/ldaptools/ldif_utils.py

@@ -2,7 +2,8 @@ import ldap
 import ldif
 from ldap.dn import dn2str
 
-from ldaptools.utils import idict, str2dn, bytes2str_entry, str2bytes_entry
+from ldaptools.utils import idict, str2dn, str2bytes_entry, bytes2str_entry
+
 
 class AddError(Exception):
     pass

src/ldaptools/utils.py

@@ -1,5 +1,5 @@
 import ldap.dn
-from six import string_types
+import six
 
 
 # Copied from http://code.activestate.com/recipes/194371-case-insensitive-strings/
@@ -76,17 +76,19 @@ class idict(dict):
         dict.__init__(self)
         self._keydict = {}  # not self.__keydict because I want it to be easily accessible by subclasses
         for entry in indict:
-            self[entry] = indict[entry]  # not dict.__setitem__(self, entry, indict[entry]) becasue this causes errors (phantom entries) where indict has overlapping keys
+            # not dict.__setitem__(self, entry, indict[entry]) becasue this
+            # causes errors (phantom entries) where indict has overlapping keys
+            self[entry] = indict[entry]
 
     def findkey(self, item):
         """A caseless way of checking if a key exists or not.
         It returns None or the correct key."""
-        if not isinstance(item, string_types):
+        if not isinstance(item, six.string_types):
             raise TypeError('Keywords for this object must be strings. You supplied %s' % type(item))
         key = item.lower()
         try:
             return self._keydict[key]
-        except:
+        except Exception:
             return None
 
     def changekey(self, item):
@@ -148,13 +150,13 @@ class idict(dict):
 
     def has_key(self, item):
         """A case insensitive test for keys."""
-        if not isinstance(item, string_types):
+        if not isinstance(item, six.string_types):
             return False               # should never have a non-string key
         return item.lower() in self._keydict           # does the key exist
 
     def __contains__(self, item):
         """A case insensitive __contains__."""
-        if not isinstance(item, string_types):
+        if not isinstance(item, six.string_types):
             return False               # should never have a non-string key
         return item.lower() in self._keydict           # does the key exist
 
@@ -202,14 +204,14 @@ class idict(dict):
     def __eq__(self, other):
         for k in self:
             if k not in other:
-                 return False
+                return False
             if self[k] != other[k]:
-                 return False
+                return False
         for k in other:
             if k not in self:
-                 return False
+                return False
             if self[k] != other[k]:
-                 return False
+                return False
         return True
 
     def __ne__(self, other):
@@ -240,15 +242,23 @@ def str2dn(s):
     return tuple(map(tuple, ldap.dn.str2dn(s)))
 
 
+def decode_if_possible(s):
+    try:
+        return s.decode('utf-8')
+    except UnicodeDecodeError:
+        return s
+
+
 def bytes2str_entry(entry):
     str_entry = {}
     for key, values in entry.items():
-        str_entry[key] = [v.decode('utf-8') if isinstance(v, bytes) else v for v in values]
+        str_entry[key] = [decode_if_possible(v) if isinstance(v, bytes) else v for v in values]
     return str_entry
 
+
 def str2bytes_entry(entry):
     bytes_entry = {}
 
     for key, values in entry.items():
-        bytes_entry[key] = [v.encode('utf-8') if isinstance(v, string_types) else v for v in values]
+        bytes_entry[key] = [v.encode('utf-8') if isinstance(v, six.text_type) else v for v in values]
     return bytes_entry

tests/test_ldif_utils.py

@@ -9,9 +9,11 @@ from ldaptools.ldif_utils import ListLDIFParser
 def test_ldifparser():
     parser = ListLDIFParser(StringIO('''dn: o=orga
 objectClass: organization
+jpegPhoto:: E+o9UYDeUDNblBzchRD/1+2HMdI=
 
 '''))
     parser.parse()
     assert len(list(parser)) == 1
     assert list(parser)[0][0] == 'o=orga'
-    assert list(parser)[0][1] == {'objectClass': ['organization']}
+    assert list(parser)[0][1]['objectClass'] == ['organization']
+    assert list(parser)[0][1]['jpegPhoto']