use XML parser that strips whitespace to avoid xml concatenation pretty_print issue

11 files changed, 52 insertions, 24 deletions

diff --git a/src/lib/Server/Core.py b/src/lib/Server/Core.py
index 8f02e0a40..9fa42cfd1 100644
--- a/src/lib/Server/Core.py
+++ b/src/lib/Server/Core.py
@@ -17,6 +17,7 @@ except ImportError:
 
 from Bcfg2.Component import Component, exposed
 from Bcfg2.Server.Plugin import PluginInitError, PluginExecutionError
+import Bcfg2.Server
 import Bcfg2.Server.FileMonitor
 import Bcfg2.Server.Plugins.Metadata
 # Compatibility imports
@@ -413,7 +414,8 @@ class Core(Component):
         # clear dynamic groups
         self.metadata.cgroups[meta.hostname] = []
         try:
-            xpdata = lxml.etree.XML(probedata.encode('utf-8'))
+            xpdata = lxml.etree.XML(probedata.encode('utf-8'),
+                                    parser=Bcfg2.Server.XMLParser)
         except:
             self.logger.error("Failed to parse probe data from client %s" % \
                               (address[0]))
@@ -462,7 +464,8 @@ class Core(Component):
     @exposed
     def RecvStats(self, address, stats):
         """Act on statistics upload."""
-        sdata = lxml.etree.XML(stats.encode('utf-8'))
+        sdata = lxml.etree.XML(stats.encode('utf-8'),
+                               parser=Bcfg2.Server.XMLParser)
         client = self.metadata.resolve_client(address)
         self.process_statistics(client, sdata)
         return "<ok/>"
diff --git a/src/lib/Bcfg2/Server/Lint/Deltas.py b/src/lib/Server/Lint/Deltas.py
index cf91d1d13..cf91d1d13 100644
--- a/src/lib/Bcfg2/Server/Lint/Deltas.py
+++ b/src/lib/Server/Lint/Deltas.py
diff --git a/src/lib/Server/Plugins/BB.py b/src/lib/Server/Plugins/BB.py
index 137142b66..0df6be4e8 100644
--- a/src/lib/Server/Plugins/BB.py
+++ b/src/lib/Server/Plugins/BB.py
@@ -1,4 +1,5 @@
 import lxml.etree
+import Bcfg2.Server
 import Bcfg2.Server.Plugin
 import glob
 import os
@@ -16,7 +17,8 @@ class BBfile(Bcfg2.Server.Plugin.XMLFileBacked):
         """Build data into an xml object."""
 
         try:
-            self.data = lxml.etree.XML(self.data)
+            self.data = lxml.etree.XML(self.data,
+                                       parser=Bcfg2.Server.XMLParser)
         except lxml.etree.XMLSyntaxError:
             Bcfg2.Server.Plugin.logger.error("Failed to parse %s" % self.name)
             return
diff --git a/src/lib/Server/Plugins/Bundler.py b/src/lib/Server/Plugins/Bundler.py
index a58257712..c795d7f90 100644
--- a/src/lib/Server/Plugins/Bundler.py
+++ b/src/lib/Server/Plugins/Bundler.py
@@ -7,7 +7,7 @@ import os
 import os.path
 import re
 import sys
-
+import Bcfg2.Server
 import Bcfg2.Server.Plugin
 
 try:
@@ -53,7 +53,8 @@ class Bundler(Bcfg2.Server.Plugin.Plugin,
             raise Bcfg2.Server.Plugin.PluginInitError
 
     def template_dispatch(self, name):
-        bundle = lxml.etree.parse(name)
+        bundle = lxml.etree.parse(name,
+                                  parser=Bcfg2.Server.XMLParser)
         nsmap = bundle.getroot().nsmap
         if name.endswith('.xml'):
             if have_genshi and \
diff --git a/src/lib/Server/Plugins/FileProbes.py b/src/lib/Server/Plugins/FileProbes.py
index c58040b65..2332a1d40 100644
--- a/src/lib/Server/Plugins/FileProbes.py
+++ b/src/lib/Server/Plugins/FileProbes.py
@@ -11,6 +11,7 @@ import errno
 import binascii
 import lxml.etree
 import Bcfg2.Options
+import Bcfg2.Server
 import Bcfg2.Server.Plugin
 
 probecode = """#!/usr/bin/env python
@@ -104,7 +105,9 @@ class FileProbes(Bcfg2.Server.Plugin.Plugin,
                                   (data.get('name'), metadata.hostname))
             else:
                 try:
-                    self.write_data(lxml.etree.XML(data.text), metadata)
+                    self.write_data(lxml.etree.XML(data.text,
+                                                   parser=Bcfg2.Server.XMLParser),
+                                    metadata)
                 except lxml.etree.XMLSyntaxError:
                     # if we didn't get XML back from the probe, assume
                     # it's an error message
diff --git a/src/lib/Server/Plugins/Metadata.py b/src/lib/Server/Plugins/Metadata.py
index cc1f78456..4f0ca9686 100644
--- a/src/lib/Server/Plugins/Metadata.py
+++ b/src/lib/Server/Plugins/Metadata.py
@@ -8,11 +8,10 @@ import copy
 import fcntl
 import lxml.etree
 import os
-import os.path
 import socket
 import sys
 import time
-
+import Bcfg2.Server
 import Bcfg2.Server.FileMonitor
 import Bcfg2.Server.Plugin
 
@@ -74,7 +73,8 @@ class XMLMetadataConfig(object):
     def load_xml(self):
         """Load changes from XML"""
         try:
-            xdata = lxml.etree.parse("%s/%s" % (self.basedir, self.basefile))
+            xdata = lxml.etree.parse(os.path.join(self.basedir, self.basefile),
+                                     parser=Bcfg2.Server.XMLParser)
         except lxml.etree.XMLSyntaxError:
             self.logger.error('Failed to parse %s' % (self.basefile))
             return
@@ -146,10 +146,13 @@ class XMLMetadataConfig(object):
             """Try to find the data in included files"""
             for included in self.extras:
                 try:
-                    xdata = lxml.etree.parse("%s/%s" % (self.basedir, included))
+                    xdata = lxml.etree.parse(os.path.join(self.basedir,
+                                                          included),
+                                             parser=Bcfg2.Server.XMLParser)
                     cli = xdata.xpath(xpath)
                     if len(cli) > 0:
-                        return {'filename': "%s/%s" % (self.basedir, included),
+                        return {'filename': os.path.join(self.basedir,
+                                                         included),
                                 'xmltree': xdata,
                                 'xquery': cli}
                 except lxml.etree.XMLSyntaxError:
@@ -283,7 +286,8 @@ class Metadata(Bcfg2.Server.Plugin.Plugin,
 
     def get_groups(self):
         '''return groups xml tree'''
-        groups_tree = lxml.etree.parse(self.data + "/groups.xml")
+        groups_tree = lxml.etree.parse(self.data + "/groups.xml",
+                                       parser=Bcfg2.Server.XMLParser)
         root = groups_tree.getroot()
         return root
 
@@ -341,7 +345,8 @@ class Metadata(Bcfg2.Server.Plugin.Plugin,
 
     def add_bundle(self, bundle_name):
         """Add bundle to groups.xml."""
-        tree = lxml.etree.parse(self.data + "/groups.xml")
+        tree = lxml.etree.parse(self.data + "/groups.xml",
+                                parser=Bcfg2.Server.XMLParser)
         root = tree.getroot()
         element = lxml.etree.Element("Bundle", name=bundle_name)
         node = self.search_group(bundle_name, tree)
@@ -364,7 +369,8 @@ class Metadata(Bcfg2.Server.Plugin.Plugin,
 
     def remove_bundle(self, bundle_name):
         """Remove a bundle."""
-        tree = lxml.etree.parse(self.data + "/groups.xml")
+        tree = lxml.etree.parse(self.data + "/groups.xml",
+                                parser=Bcfg2.Server.XMLParser)
         root = tree.getroot()
         node = self.search_group(bundle_name, tree)
         if node == None:
@@ -815,7 +821,8 @@ class Metadata(Bcfg2.Server.Plugin.Plugin,
         def include_group(group):
             return not only_client or group in clientmeta.groups
         
-        groups_tree = lxml.etree.parse(self.data + "/groups.xml")
+        groups_tree = lxml.etree.parse(self.data + "/groups.xml",
+                                       parser=Bcfg2.Server.XMLParser)
         try:
             groups_tree.xinclude()
         except lxml.etree.XIncludeError:
diff --git a/src/lib/Server/Plugins/NagiosGen.py b/src/lib/Server/Plugins/NagiosGen.py
index 8a76c130d..287e1b0d3 100644
--- a/src/lib/Server/Plugins/NagiosGen.py
+++ b/src/lib/Server/Plugins/NagiosGen.py
@@ -7,7 +7,7 @@ import glob
 import socket
 import logging
 import lxml.etree
-
+import Bcfg2.Server
 import Bcfg2.Server.Plugin
 
 LOGGER = logging.getLogger('Bcfg2.Plugins.NagiosGen')
@@ -85,7 +85,8 @@ class NagiosGen(Bcfg2.Server.Plugin.Plugin,
             LOGGER.warn("Parsing deprecated NagiosGen/parents.xml. "
                         "Update to the new-style config with "
                         "nagiosgen-convert.py.")
-            parents = lxml.etree.parse(pfile)
+            parents = lxml.etree.parse(pfile,
+                                       parser=Bcfg2.Server.XMLParser)
             for el in parents.xpath("//Depend[@name='%s']" % metadata.hostname):
                 if 'parent' in xtra:
                     xtra['parent'] += "," + el.get("on")
diff --git a/src/lib/Server/Plugins/Probes.py b/src/lib/Server/Plugins/Probes.py
index ae1ed4c2b..2c24d6cc4 100644
--- a/src/lib/Server/Plugins/Probes.py
+++ b/src/lib/Server/Plugins/Probes.py
@@ -2,6 +2,8 @@ import time
 import lxml.etree
 import operator
 import re
+import os
+import Bcfg2.Server
 
 try:
     import json
@@ -93,7 +95,8 @@ class ProbeData(object):
     def xdata(self):
         if self._xdata is None:
             try:
-                self._xdata = lxml.etree.XML(self.data)
+                self._xdata = lxml.etree.XML(self.data,
+                                             parser=Bcfg2.Server.XMLParser)
             except lxml.etree.XMLSyntaxError:
                 pass
         return self._xdata
@@ -221,7 +224,8 @@ class Probes(Bcfg2.Server.Plugin.Plugin,
 
     def load_data(self):
         try:
-            data = lxml.etree.parse(self.data + '/probed.xml').getroot()
+            data = lxml.etree.parse(os.path.join(self.data, 'probed.xml'),
+                                    parser=Bcfg2.Server.XMLParser).getroot()
         except:
             self.logger.error("Failed to read file probed.xml")
             return
diff --git a/src/lib/Server/Plugins/SGenshi.py b/src/lib/Server/Plugins/SGenshi.py
index f6a98c141..ae2623d29 100644
--- a/src/lib/Server/Plugins/SGenshi.py
+++ b/src/lib/Server/Plugins/SGenshi.py
@@ -8,7 +8,7 @@ import logging
 import copy
 import sys
 import os.path
-
+import Bcfg2.Server
 import Bcfg2.Server.Plugin
 import Bcfg2.Server.Plugins.TGenshi
 
@@ -29,7 +29,8 @@ class SGenshiTemplateFile(Bcfg2.Server.Plugins.TGenshi.TemplateFile,
         try:
             stream = self.template.generate(metadata=metadata).filter( \
                 Bcfg2.Server.Plugins.TGenshi.removecomment)
-            data = lxml.etree.XML(stream.render('xml', strip_whitespace=False))
+            data = lxml.etree.XML(stream.render('xml', strip_whitespace=False),
+                                  parser=Bcfg2.Server.XMLParser)
             bundlename = os.path.splitext(os.path.basename(self.name))[0]
             bundle = lxml.etree.Element('Bundle', name=bundlename)
             for item in self.Match(metadata, data):
diff --git a/src/lib/Server/Plugins/SSLCA.py b/src/lib/Server/Plugins/SSLCA.py
index 7b4a08ae1..e5bc38fba 100644
--- a/src/lib/Server/Plugins/SSLCA.py
+++ b/src/lib/Server/Plugins/SSLCA.py
@@ -43,14 +43,16 @@ class SSLCA(Bcfg2.Server.Plugin.GroupSpool):
         if event.filename.endswith('.xml'):
             if action in ['exists', 'created', 'changed']:
                 if event.filename.endswith('key.xml'):
-                    key_spec = dict(list(lxml.etree.parse(epath).find('Key').items()))
+                    key_spec = dict(list(lxml.etree.parse(epath,
+                                                          parser=Bcfg2.Server.XMLParser).find('Key').items()))
                     self.key_specs[ident] = {
                         'bits': key_spec.get('bits', 2048),
                         'type': key_spec.get('type', 'rsa')
                     }
                     self.Entries['Path'][ident] = self.get_key
                 elif event.filename.endswith('cert.xml'):
-                    cert_spec = dict(list(lxml.etree.parse(epath).find('Cert').items()))
+                    cert_spec = dict(list(lxml.etree.parse(epath,
+                                                           parser=Bcfg2.Server.XMLParser).find('Cert').items()))
                     ca = cert_spec.get('ca', 'default')
                     self.cert_specs[ident] = {
                         'ca': ca,
diff --git a/src/lib/Server/__init__.py b/src/lib/Server/__init__.py
index 25f397565..bca73ded7 100644
--- a/src/lib/Server/__init__.py
+++ b/src/lib/Server/__init__.py
@@ -1,6 +1,10 @@
 # $Id$
 """This is the set of modules for Bcfg2.Server."""
 
+import lxml.etree
+
 __revision__ = '$Revision$'
 __all__ = ["Admin", "Core", "FileMonitor", "Plugin", "Plugins",
-           "Hostbase", "Reports", "Snapshots"]
+           "Hostbase", "Reports", "Snapshots", "XMLParser"]
+
+XMLParser = lxml.etree.XMLParser(remove_blank_text=True)