Ldif

------------------

- Fix detection of unsafe strings in ``unparse`` (See `7
<https://github.com/xi/ldif3/pull/7>`_)

------------------

- Ignore non-unicode characters in "dn" in non-strict mode. (Fixes `5
<https://github.com/xi/ldif3/issues/6>`_)

------------------

- Overhaule the unicode support to also support binary data (e.g. images)
encoded in LDIF.

You can now pass an encoding to the parser which will be used to decode
values. If decoding failes, a bytestring will be returned. If you pass an
encoding of ``None``, the parser will not try to do any conversion and
return bytes directly.

This change should be completely backwards compatible, as the parser now
gracefully handles a case where it crashed previously.

(See `4 <https://github.com/xi/ldif3/issues/4>`_)

------------------

- Allow empty values for attributes.

------------------

This is mostly a reaction to `python-ldap 2.4.20
<https://mail.python.org/pipermail/python-ldap/2015q3/003557.html>`_.

- Restore support for ``records_read`` as well as adding ``line_counter`` and
``byte_counter`` that were introduced in python-ldap 2.4.20.
- Stricter order checking of ``dn:``.
- Remove partial support for parsing change records. A more complete
implementation based on improvements made in python-ldap may be included
later. But for now, I don't have the time.

**Breaking change**: ``LDIFParser.parse()`` now yields ``dn, entry`` rather
than ``dn, changetype, entry``.

------------------

- Include documentation source and changelog in source distribution.
(Thanks to Michael Fladischer)
- Add LICENSE file