1 files changed, 21 insertions, 1 deletions
diff --git a/src/lib/Bcfg2/Client/XML.py b/src/lib/Bcfg2/Client/XML.py
index 91d4ac5c6..4ba06abae 100644
--- a/src/lib/Bcfg2/Client/XML.py
+++ b/src/lib/Bcfg2/Client/XML.py
@@ -5,9 +5,29 @@
 # pylint: disable=E0611,W0611,W0613,C0103
 
 try:
-    from lxml.etree import Element, SubElement, XML, 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'
+
+    # libxml2 2.9.0+ doesn't parse 10M+ documents by default:
+    # https://mail.gnome.org/archives/commits-list/2012-August/msg00645.html
+    try:
+        _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:
+            return _XML(val.encode(), **kwargs)
 except ImportError:
     # lxml not available
     from xml.parsers.expat import ExpatError as ParseError