Nti.contentfragments

==================

- When sanitizing html, disable link creation when already under an anchor.

==================

- Allow ``IAllowedAttributeProvider`` to be registered to provide
additional attributes that would be allowed in sanitized content
fragments.

==================

- Add support for Python 3.8.

- Depend on zope.interface 5.0.

- Update the datrie dependency. See https://github.com/NextThought/nti.contentfragments/issues/24

- Make ``IUnicodeContentFragment`` extend
``zope.interface.common.collections.ISequence`` instead of the
semi-deprecated ``zope.interface.common.sequence.IReadSequence``.

- Replace custom interfaces ``IString``, ``IUnicode`` and ``IBytes``
with aliases for ``INativeString``, ``ITextString`` and
``IByteString`` from ``zope.interface.common.builtins``. These
custom aliases are now deprecated. See https://github.com/NextThought/nti.contentfragments/issues/23.

- Fix unicode normalization breaking schema fields with zope.schema
6.0. See https://github.com/NextThought/nti.contentfragments/issues/26

- Ensure all objects have consistent resolution orders.

==================

- Remove a word from the censored word list. See issue https://github.com/NextThought/nti.contentfragments/issues/22.

==================

- Add support for Python 3.7. Note that ``datrie`` is not yet
available for Python 3.7.

- Add support for PyPy3.

- Add interfaces for all schema fields defined in
``nti.contentfragments.schema`` and make the respective classes
implement them.

==================

- Packaging: Do not use ``html5lib[datrie]`` and instead copy that
dependency into our own dependencies to workaround a buildout error.
See https://github.com/NextThought/nti.contentfragments/issues/17