From 79f5fce2a387f5dfc13e10146b36cb620a5bf166 Mon Sep 17 00:00:00 2001 From: "Chris St. Pierre" Date: Tue, 28 Jan 2014 15:52:42 -0500 Subject: Client: let lxml.etree XML implementation parse very large documents --- src/lib/Bcfg2/Client/XML.py | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'src/lib/Bcfg2/Client/XML.py') diff --git a/src/lib/Bcfg2/Client/XML.py b/src/lib/Bcfg2/Client/XML.py index 561cd25a8..e35ce4497 100644 --- a/src/lib/Bcfg2/Client/XML.py +++ b/src/lib/Bcfg2/Client/XML.py @@ -5,17 +5,25 @@ # pylint: disable=E0611,W0611,W0613,C0103 try: - from lxml.etree import Element, SubElement, tostring + from lxml.etree import Element, SubElement, tostring, XMLParser from lxml.etree import XMLSyntaxError as ParseError from lxml.etree import XML as _XML from Bcfg2.Compat import wraps driver = 'lxml' + try: + # libxml2 2.9.0+ doesn't parse 10M+ documents by default: + # https://mail.gnome.org/archives/commits-list/2012-August/msg00645.html + _parser = XMLParser(huge_tree=True) + except TypeError: + _parser = XMLParser() + @wraps(_XML) def XML(val, **kwargs): """ unicode strings w/encoding declaration are not supported in recent lxml.etree, so we try to read XML, and if it fails we try encoding the string. """ + kwargs.setdefault('parser', _parser) try: return _XML(val, **kwargs) except ValueError: -- cgit v1.2.3-1-g7c22