Perform schema checks for all Xinclude files in groups.xml

git-svn-id: https://svn.mcs.anl.gov/repos/bcfg/trunk/bcfg2@4890 ce84e21b-d406-0410-9b95-82705330c041

1 files changed, 32 insertions, 13 deletions

diff --git a/src/sbin/bcfg2-repo-validate b/src/sbin/bcfg2-repo-validate
index d067d4e74..459032346 100755
--- a/src/sbin/bcfg2-repo-validate
+++ b/src/sbin/bcfg2-repo-validate
@@ -10,6 +10,14 @@ import glob, lxml.etree, os, sys
 
 import Bcfg2.Options
 
+try:
+    pdlist = set
+except:
+    class pdlist(list):
+        def add(self, item):
+            if item not in self:
+                self.append(item)
+
 if __name__ == '__main__':
     opts = {'repo': Bcfg2.Options.SERVER_REPOSITORY,
             'prefix': Bcfg2.Options.INSTALL_PREFIX,
@@ -31,16 +39,31 @@ if __name__ == '__main__':
             if filename.endswith('info.xml'):
                 info_list.append(os.path.join(root, filename))
 
-    # get all XIncluded bundles
-    ref_bundles = []
-    unref_bundles = []
+    # get metadata list (with all included files)
+    metadata_list = glob.glob("%s/Metadata/groups.xml" % repo)
+    ref_bundles = pdlist()
     xdata = lxml.etree.parse("%s/Metadata/groups.xml" % repo)
+    included = pdlist([ent.get('href') for ent in \
+                    xdata.findall('./{http://www.w3.org/2001/XInclude}include')])
+    while included:
+        try:
+            filename = included.pop()
+        except KeyError:
+            continue
+        metadata_list.append("%s/Metadata/%s" % (repo, filename))
+        groupdata = lxml.etree.parse("%s/Metadata/%s" % (repo, filename))
+        group_ents = [ent.get('href') for ent in \
+                      groupdata.findall('./{http://www.w3.org/2001/XInclude}include')]
+        for ent in group_ents:
+            included.add(ent)
+        included.discard(filename)
+
+    # get all XIncluded bundles
     xdata.xinclude()
     for bundle in xdata.findall("//Bundle"):
-	ref_bundles.append("%s/Bundler/%s.xml" % (repo,bundle.get('name')))
+        ref_bundles.add("%s/Bundler/%s.xml" % (repo, bundle.get('name')))
 
     # get lists of all other xml files to validate
-    metadata_list = glob.glob("%s/Metadata/groups.xml" % repo)
     clients_list = glob.glob("%s/Metadata/clients.xml" % repo)
     bundle_list = glob.glob("%s/Bundler/*.xml" % repo)
     pkg_list = glob.glob("%s/Pkgmgr/*.xml" % repo)
@@ -50,11 +73,6 @@ if __name__ == '__main__':
     services_list = glob.glob("%s/Svcmgr/*.xml" % repo)
     deps_list = glob.glob("%s/Deps/*.xml" % repo)
 
-    # find all unreferenced bundles
-    for bundle in ref_bundles:
-        if bundle not in bundle_list:
-            unref_bundles.append(bundle)
-
     filesets = {'metadata':(metadata_list, "%s/metadata.xsd"),
                 'clients':(clients_list, "%s/clients.xsd"),
                 'info':(info_list, "%s/info.xsd"),
@@ -104,8 +122,9 @@ if __name__ == '__main__':
     # print out missing bundle information
     if verbose:
         print("")
-        for bundle in unref_bundles:
-            print ("*** Warning: Bundle %s referenced, but does not "
-                                "exist." % bundle)
+        for bundle in ref_bundles:
+            if bundle not in bundle_list:
+                print ("*** Warning: Bundle %s referenced, but does not "
+                                    "exist." % bundle)
 
     raise SystemExit, failures