repoze.xmliter 2.0

Wrapper for ``lxml`` trees which serializes to string upon iteration.

Overview

This package provides a wrapper for lxml trees which serializes to string on iteration, but otherwise makes the tree available in an attribute.

The primary for this is WSGI middleware which may avoid needless XML parsing and serialization.

Usage

It’s recommend to use the lazy decorator on your application method. This allows you to return an lxml tree object, which is then automatically turned into an XMLSerializer.

>>> from repoze.xmliter import lazy

>>> @lazy
... def application(environ, start_response)
...     return some_lxml_tree

You may provide a serializer function, which will be used when the XMLSerializer is eventually iterated over (i.e. when the response is rendered):

>>> @lazy(serializer=lxml.html.tostring)
... def application(environ, start_response)
...     return some_lxml_tree

Middleware can use isinstance to test if the result is an XML iterable:

>>> from repoze.xmliter.serializer import XMLSerializer
>>> isinstance(result, XMLSerializer)

In this case, the middleware can simply access the tree attribute of the result.

There are two convenience methods which can be used to parse a WSGI iterable of strings and build an XMLSerializer object, but avoids re-building the serializer if the input iterable is already an instance of XMLSerializer:

>>> from repoze.xmliter.utils import getXMLSerializer
>>> result = getXMLSerializer(result)

Or, if you are parsing HTML:

>>> from repoze.xmliter.utils import getHTMLSerializer
>>> result = getHTMLSerializer(result)

If result is not an XMLSerializer, it will be parsed using a feed parser, turned into an lxml tree, and wrapped up in an XMLSerializer, which is returned.

Changelog

2.0 (2026-05-08)

• Make final release, no code changes. [maurits]

2.0a2 (2025-12-22)

• Let getXMLSerializer raise an XMLSyntaxError when passed an empty iterable. Then the result with lxml 6 is the same as it was with lxml 5, instead of throwing an AttributeError. [maurits]

2.0a1 (2025-12-22)

• Replace pkg_resources namespace with PEP 420 native namespace. [maurits]

1.0 (2025-10-29)

• Add support for Python 3.8 - 3.14. [maurits]

• Drop support for Python 3.7: it is almost two years out of support and cannot easily be tested. [maurits]

1.0b1 (2024-01-31)

• Fix serialization of a tree to bytes PR. [maurits]

• Removed unused future dependency. Fixes issue 10. [maurits]

• Drop support for Python 2.6, 2.7, 3.3, 3.4, 3.5, and 3.6. [tseaver, mborch]

• Add support for Python 3.7, 3.8, 3.9, 3.10, and 3.11. [tseaver, mborch]

0.6.1 (2022-01-14)

• Fixed tests with lxml 4.7.1 or higher. Fixes issue 8. [maurits]

0.6 - 2014-09-21

• Python 3 compatibility [Lennart Regebro]

0.5 - 2012-01-25

• Add __len__ to serializer to help WSGI servers. [Laurence]

• Serializer should iter the entire string in one go. [Laurence]

0.4 - 2011-06-16

• Ensure trailing space is removed when replacing doctype with empty string. [Laurence]

0.3 - 2011-06-03

• Add doctype option to replace doctype on serialization. [Laurence]

0.2 - 2010-09-11

• Use document encoding by default. (This fixes test failure on Ubuntu 10.04.) [Laurence]

• Defer to xsl:output settings when serializing an XSLResultTree. [Laurence]

• Turn off pretty printing by default for HTML to avoid affecting rendering on the browser. [Laurence]

0.1 - 2010-04-21

• Initial release