Overview

In a client/server environment dealing with various types of content from users, it’s important to know what not just the Python type of a particular string is, but also what the semantic type of the string is: HTML, plain text, LaTeX, etc.

This package defines interfaces and classes to be able to record this information. It also features a framework for transforming between the various supported semantic types (e.g., HTML to plain text).

Other features:

• Support for making arbitrary incoming HTML safe (sanitizing it).

• Support for very configurable (optionally) event-based profanity censoring that integrates with nti.schema/zope.schema.

See the documentation for more details.

1.9.0 (2021-10-26)

• Fix adapting base string input to plain text to behave more like 1.7 by only running the HTML to plain text algorithm if the input looks like it may contain HTML markup. Note that in some instances where characters like ‘<’ were previously escaped to ‘&lt;’, this will no longer happen if the rest of the string doesn’t look like HTML. See issue 44.

• Add schema fields VerbatimPlainText and VerbatimPlainTextLine to assume any incoming unicode value already represents a plain text content fragment, instead of (possibly) passing it through the HTML to plain text algorithm.

1.8.0 (2021-10-06)

• Add support for Python 3.9 and 3.10.

• Move to Github Actions from Travis CI.

• The algorithm for converting HTML to plain text has been changed and produces higher quality output. For example, links are preserved in a human-readable fashion. See issue 39.

• Fix an error getting link text when there was no link formatter utility installed. See PR 42.

1.7.0 (2020-10-07)

• Allow conversion of reStructuredText fragments to plain text.

1.6.1 (2020-09-14)

• Ensure disallowed tags nested within anchors do not raise. See issue 34.

1.6.0 (2020-09-02)

• Add support for reStructuredText content fragments and corresponding fields.

1.5.0 (2020-07-23)

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

1.4.0 (2020-06-17)

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

1.3.0 (2020-04-06)

• 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.

1.2.1 (2019-11-07)

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

1.2.0 (2018-10-15)

• 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.

1.1.1 (2018-06-29)

• 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

1.1.0 (2017-06-14)

• Remove dependency of dolmen.builtins. The interfaces IUnicode, IBytes and IString are now always defined by this package.

• Add support for Python 3.6.

1.0.0 (2016-08-19)

• Add support for Python 3.

• Stop configuring plone.i18n. It’s a big dependency and doesn’t work on Python 3.

• Introduce our own interfaces for IUnicode and IString, subclassing dolmen.builtins.IUnicode and IString, respectively, if possible.

• The word lists used in censoring are cached in memory.

• nti.contentfragments.html._Serializer has been renamed and is no longer public.

• Depend on zope.mimetype >= 2.1.0 for better support of Python 3.