Rewrote bcfg2-repo-validate as bcfg2-lint, which uses a plugin

interface to be lots more flexible and extensible.  Added several more
tests.  If bcfg2-lint is run as bcfg2-repo-validate, it roughly
emulates the functionality of that program.

TODO: Need to figure out correct way to symlink bcfg2-repo-validate to
bcfg2-lint on install.

9 files changed, 743 insertions, 2 deletions

diff --git a/src/lib/Logger.py b/src/lib/Logger.py
index a9c4372b7..ae73a6d41 100644
--- a/src/lib/Logger.py
+++ b/src/lib/Logger.py
@@ -158,17 +158,23 @@ class FragmentingSysLogHandler(logging.handlers.SysLogHandler):
                     pass
 
 
-def setup_logging(procname, to_console=True, to_syslog=True, syslog_facility='daemon', level=0, to_file=None):
+def setup_logging(procname, to_console=True, to_syslog=True,
+                  syslog_facility='daemon', level=0, to_file=None):
     """Setup logging for Bcfg2 software."""
     if hasattr(logging, 'already_setup'):
         return
+
     # add the handler to the root logger
     if to_console:
         console = logging.StreamHandler(sys.stdout)
-        console.setLevel(logging.DEBUG)
+        if to_console is True:
+            console.setLevel(logging.DEBUG)
+        else:
+            console.setLevel(to_console)
         # tell the handler to use this format
         console.setFormatter(TermiosFormatter())
         logging.root.addHandler(console)
+        
     if to_syslog:
         try:
             try:
@@ -186,11 +192,13 @@ def setup_logging(procname, to_console=True, to_syslog=True, syslog_facility='da
             logging.root.error("failed to activate syslogging")
         except:
             print("Failed to activate syslogging")
+            
     if not to_file == None:
         filelog = logging.FileHandler(to_file)
         filelog.setLevel(logging.DEBUG)
         filelog.setFormatter(logging.Formatter('%(asctime)s %(name)s[%(process)d]: %(message)s'))
         logging.root.addHandler(filelog)
+        
     logging.root.setLevel(level)
     logging.already_setup = True
 
diff --git a/src/lib/Server/Lint/Bundles.py b/src/lib/Server/Lint/Bundles.py
new file mode 100644
index 000000000..a1ce631c9
--- /dev/null
+++ b/src/lib/Server/Lint/Bundles.py
@@ -0,0 +1,56 @@
+import lxml.etree
+import Bcfg2.Server.Lint
+
+class Bundles(Bcfg2.Server.Lint.ServerPlugin):
+    """ Perform various bundle checks """
+
+    @Bcfg2.Server.Lint.returnErrors
+    def Run(self):
+        """ run plugin """
+        self.missing_bundles()
+        self.bundle_names()
+        self.sgenshi_groups()
+
+    def missing_bundles(self):
+        """ find bundles listed in Metadata but not implemented in Bundler """
+        groupdata = self.metadata.groups_xml.xdata
+        ref_bundles = set([b.get("name")
+                           for b in groupdata.findall("//Bundle")])
+
+        allbundles = self.core.plugins['Bundler'].entries.keys()
+        for bundle in ref_bundles:
+            xmlbundle = "%s.xml" % bundle
+            genshibundle = "%s.genshi" % bundle
+            if xmlbundle not in allbundles and genshibundle not in allbundles:
+                self.LintError("Bundle %s referenced, but does not exist" %
+                               bundle)
+
+    def bundle_names(self):
+        """ verify bundle name attribute matches filename """
+        for bundle in self.core.plugins['Bundler'].entries.values():
+            if self.HandlesFile(bundle.name):
+                try:
+                    xdata = lxml.etree.XML(bundle.data)
+                except AttributeError:
+                    # genshi template
+                    xdata = lxml.etree.parse(bundle.template.filepath).getroot()
+            
+                fname = bundle.name.split('Bundler/')[1].split('.')[0]
+                bname = xdata.get('name')
+                if fname != bname:
+                    self.LintWarning("Inconsistent bundle name: filename is %s, bundle name is %s" %
+                                     (fname, bname))
+
+    def sgenshi_groups(self):
+        """ ensure that Genshi Bundles do not include <Group> tags,
+        which are not supported  """
+        for bundle in self.core.plugins['Bundler'].entries.values():
+            if self.HandlesFile(bundle.name):
+                if (type(bundle) is
+                    Bcfg2.Server.Plugins.SGenshi.SGenshiTemplateFile):
+                    xdata = lxml.etree.parse(bundle.name)
+                    groups = [self.RenderXML(g)
+                              for g in xdata.getroottree().findall("//Group")]
+                    if groups:
+                        self.LintError("<Group> tag is not allowed in SGenshi Bundle:\n%s" %
+                                       "\n".join(groups))
diff --git a/src/lib/Server/Lint/Comments.py b/src/lib/Server/Lint/Comments.py
new file mode 100644
index 000000000..0b50df373
--- /dev/null
+++ b/src/lib/Server/Lint/Comments.py
@@ -0,0 +1,183 @@
+import os.path
+import lxml.etree
+import Bcfg2.Server.Lint
+
+class Comments(Bcfg2.Server.Lint.ServerPlugin):
+    """ check files for various required headers """
+    def __init__(self, *args, **kwargs):
+        Bcfg2.Server.Lint.ServerPlugin.__init__(self, *args, **kwargs)
+        self.config_cache = {}
+
+    @Bcfg2.Server.Lint.returnErrors
+    def Run(self):
+        self.check_bundles()
+        self.check_properties()
+        self.check_metadata()
+        self.check_cfg()
+        self.check_infoxml()
+        self.check_probes()
+
+    def required_keywords(self, rtype):
+        """ given a file type, fetch the list of required VCS keywords
+        from the bcfg2-lint config """
+        return self.required_items(rtype, "keyword", default=["Id"])
+
+    def required_comments(self, rtype):
+        """ given a file type, fetch the list of required comments
+        from the bcfg2-lint config """
+        return self.required_items(rtype, "comment")
+
+    def required_items(self, rtype, itype, default=None):
+        """ given a file type and item type (comment or keyword),
+        fetch the list of required items from the bcfg2-lint config """
+        if itype not in self.config_cache:
+            self.config_cache[itype] = {}
+            
+        if rtype not in self.config_cache[itype]:
+            rv = []
+            global_item = "global_%ss" % itype
+            if global_item in self.config:
+                rv.extend(self.config[global_item].split(","))
+            elif default is not None:
+                rv.extend(default)
+            
+            item = "%s_%ss" % (rtype.lower(), itype)
+            if item in self.config:
+                if self.config[item]:
+                    rv.extend(self.config[item].split(","))
+                else:
+                    # config explicitly specifies nothing
+                    rv = []
+            self.config_cache[itype][rtype] = rv
+        return self.config_cache[itype][rtype]
+
+    def check_bundles(self):
+        """ check bundle files for required headers """
+        for bundle in self.core.plugins['Bundler'].entries.values():
+            xdata = None
+            rtype = ""
+            try:
+                xdata = lxml.etree.XML(bundle.data)
+                rtype = "bundler"
+            except AttributeError:
+                xdata = lxml.etree.parse(bundle.template.filepath).getroot()
+                rtype = "sgenshi"
+
+            self.check_xml(bundle.name, xdata, rtype)
+
+    def check_properties(self):
+        """ check properties files for required headers """
+        if 'Properties' in self.core.plugins:
+            props = self.core.plugins['Properties']
+            for propfile, pdata in props.store.entries.items():
+                if os.path.splitext(propfile)[1] == ".xml":
+                    self.check_xml(pdata.name, pdata.data, 'properties')
+
+    def check_metadata(self):
+        """ check metadata files for required headers """
+        metadata = self.core.plugins['Metadata']
+        if self.has_all_xincludes("groups.xml"):
+            self.check_xml(os.path.join(metadata.data, "groups.xml"),
+                           metadata.groups_xml.data,
+                           "metadata")
+        if self.has_all_xincludes("clients.xml"):
+            self.check_xml(os.path.join(metadata.data, "clients.xml"),
+                           metadata.clients_xml.data,
+                           "metadata")
+
+    def check_cfg(self):
+        """ check Cfg files for required headers """
+        for entryset in self.core.plugins['Cfg'].entries.values():
+            for entry in entryset.entries.values():
+                if entry.name.endswith(".genshi"):
+                    rtype = "tgenshi"
+                else:
+                    rtype = "cfg"
+                self.check_plaintext(entry.name, entry.data, rtype)
+
+    def check_infoxml(self):
+        """ check info.xml files for required headers """
+        for entryset in self.core.plugins['Cfg'].entries.items():
+            if hasattr(entryset, "infoxml") and entryset.infoxml is not None:
+                self.check_xml(entryset.infoxml.name,
+                               entryset.infoxml.pnode.data,
+                               "infoxml")
+
+    def check_probes(self):
+        """ check probes for required headers """
+        if 'Probes' in self.core.plugins:
+            for probe in self.core.plugins['Probes'].probes.entries.values():
+                self.check_plaintext(probe.name, probe.data, "probes")
+
+    def check_xml(self, filename, xdata, rtype):
+        """ check generic XML files for required headers """
+        self.check_lines(filename,
+                         [str(el)
+                          for el in xdata.getiterator(lxml.etree.Comment)],
+                         rtype)
+
+    def check_plaintext(self, filename, data, rtype):
+        """ check generic plaintex files for required headers """
+        self.check_lines(filename, data.splitlines(), rtype)
+
+    def check_lines(self, filename, lines, rtype):
+        """ generic header check for a set of lines """
+        if self.HandlesFile(filename):
+            # found is trivalent:
+            # False == not found
+            # None == found but not expanded
+            # True == found and expanded
+            found = dict((k, False) for k in self.required_keywords(rtype))
+            
+            for line in lines:
+                # we check for both '$<keyword>:' and '$<keyword>$' to see
+                # if the keyword just hasn't been expanded
+                for (keyword, status) in found.items():
+                    if not status:
+                        if '$%s:' % keyword in line:
+                            found[keyword] = True
+                        elif '$%s$' % keyword in line:
+                            found[keyword] = None
+
+            unexpanded = [keyword for (keyword, status) in found.items()
+                          if status is None]
+            if unexpanded:
+                self.LintError("%s: Required keywords(s) found but not expanded: %s" %
+                               (filename, ", ".join(unexpanded)))
+            missing = [keyword for (keyword, status) in found.items()
+                       if status is False]
+            if missing:
+                self.LintError("%s: Required keywords(s) not found: %s" %
+                               (filename, ", ".join(missing)))
+
+            # next, check for required comments.  found is just
+            # boolean
+            found = dict((k, False) for k in self.required_comments(rtype))
+            
+            for line in lines:
+                for (comment, status) in found.items():
+                    if not status:
+                        found[comment] = comment in line
+
+            missing = [comment for (comment, status) in found.items()
+                       if status is False]
+            if missing:
+                self.LintError("%s: Required comments(s) not found: %s" %
+                               (filename, ", ".join(missing)))
+
+    def has_all_xincludes(self, mfile):
+        """ return true if self.files includes all XIncludes listed in
+        the specified metadata type, false otherwise"""
+        if self.files is None:
+            return True
+        else:
+            path = os.path.join(self.metadata.data, mfile)
+            if path in self.files:
+                xdata = lxml.etree.parse(path)
+                for el in xdata.findall('./{http://www.w3.org/2001/XInclude}include'):
+                    if not self.has_all_xincludes(el.get('href')):
+                        self.LintWarning("Broken XInclude chain: could not include %s" % path)
+                        return False
+
+                return True
+
diff --git a/src/lib/Server/Lint/Duplicates.py b/src/lib/Server/Lint/Duplicates.py
new file mode 100644
index 000000000..c8b542025
--- /dev/null
+++ b/src/lib/Server/Lint/Duplicates.py
@@ -0,0 +1,79 @@
+import os.path
+import lxml.etree
+import Bcfg2.Server.Lint
+
+class Duplicates(Bcfg2.Server.Lint.ServerPlugin):
+    """ Find duplicate clients, groups, etc. """
+    def __init__(self, *args, **kwargs):
+        Bcfg2.Server.Lint.ServerPlugin.__init__(self, *args, **kwargs)
+        self.groups_xdata = None
+        self.clients_xdata = None
+        self.load_xdata()
+
+    @Bcfg2.Server.Lint.returnErrors
+    def Run(self):
+        """ run plugin """
+        # only run this plugin if we were not given a list of files.
+        # not only is it marginally silly to run this plugin with a
+        # partial list of files, it turns out to be really freaking
+        # hard to get only a fragment of group or client metadata
+        if self.groups_xdata is not None:
+            self.duplicate_groups()
+            self.duplicate_defaults()
+        if self.clients_xdata is not None:
+            self.duplicate_clients()
+
+    def load_xdata(self):
+        """ attempt to load XML data for groups and clients.  only
+        actually load data if all documents reference in XIncludes can
+        be found in self.files"""
+        if self.has_all_xincludes("groups.xml"):
+            self.groups_xdata = self.metadata.clients_xml.xdata
+        if self.has_all_xincludes("clients.xml"):
+            self.clients_xdata = self.metadata.clients_xml.xdata
+            
+    def duplicate_groups(self):
+        """ find duplicate groups """
+        self.duplicate_entries(self.clients_xdata.xpath('//Groups/Group'),
+                               'group')
+
+    def duplicate_clients(self):
+        """ find duplicate clients """
+        self.duplicate_entries(self.clients_xdata.xpath('//Clients/Client'),
+                               'client')
+
+    def duplicate_entries(self, data, etype):
+        """ generic duplicate entry finder """
+        seen = {}
+        for el in data:
+            if el.get('name') not in seen:
+                seen[el.get('name')] = el
+            else:
+                self.LintError("Duplicate %s '%s':\n%s\n%s" %
+                               (etype, el.get('name'),
+                                self.RenderXML(seen[el.get('name')]),
+                                self.RenderXML(el)))
+
+    def duplicate_defaults(self):
+        """ check for multiple default group definitions """
+        default_groups = [g for g in self.groups_xdata.findall('.//Group')
+                          if g.get('default') == 'true']
+        if len(default_groups) > 1:
+            self.LintError("Multiple default groups defined: %s" %
+                           ",".join(default_groups))
+
+    def has_all_xincludes(self, mfile):
+        """ return true if self.files includes all XIncludes listed in
+        the specified metadata type, false otherwise"""
+        if self.files is None:
+            return True
+        else:
+            path = os.path.join(self.metadata.data, mfile)
+            if path in self.files:
+                xdata = lxml.etree.parse(path)
+                for el in xdata.findall('./{http://www.w3.org/2001/XInclude}include'):
+                    if not self.has_all_xincludes(el.get('href')):
+                        self.LintWarning("Broken XInclude chain: could not include %s" % path)
+                        return False
+
+                return True
diff --git a/src/lib/Server/Lint/InfoXML.py b/src/lib/Server/Lint/InfoXML.py
new file mode 100644
index 000000000..097c2d6f9
--- /dev/null
+++ b/src/lib/Server/Lint/InfoXML.py
@@ -0,0 +1,38 @@
+import os.path
+import Bcfg2.Options
+import Bcfg2.Server.Lint
+
+class InfoXML(Bcfg2.Server.Lint.ServerPlugin):
+    """ ensure that all config files have an info.xml file"""
+
+    @Bcfg2.Server.Lint.returnErrors
+    def Run(self):
+        for filename, entryset in self.core.plugins['Cfg'].entries.items():
+            infoxml_fname = os.path.join(entryset.path, "info.xml")
+            if self.HandlesFile(infoxml_fname):
+                if (hasattr(entryset, "infoxml") and
+                    entryset.infoxml is not None):
+                    xdata = entryset.infoxml.pnode.data
+                    for info in xdata.getroottree().findall("//Info"):
+                        required = ["owner", "group", "perms"]
+                        if "required" in self.config:
+                            required = self.config["required"].split(",")
+
+                        missing = [attr for attr in required
+                                   if info.get(attr) is None]
+                        if missing:
+                            self.LintError("Required attribute(s) %s not found in %s:%s" %
+                                           (",".join(missing), infoxml_fname,
+                                            self.RenderXML(info)))
+
+                        if ("require_paranoid" in self.config and
+                            self.config["require_paranoid"].lower() == "true" and
+                            not Bcfg2.Options.MDATA_PARANOID.value and
+                            info.get("paranoid").lower() != "true"):
+                            self.LintError("Paranoid must be true in %s:%s" %
+                                           (infoxml_fname,
+                                            self.RenderXML(info)))
+                elif ("require" in self.config and
+                      self.config["require"].lower != "false"):
+                    self.LintError("No info.xml found for %s" % filename)
+
diff --git a/src/lib/Server/Lint/Pkgmgr.py b/src/lib/Server/Lint/Pkgmgr.py
new file mode 100644
index 000000000..28ca698bd
--- /dev/null
+++ b/src/lib/Server/Lint/Pkgmgr.py
@@ -0,0 +1,33 @@
+import Bcfg2.Server.Lint
+
+class Pkgmgr(Bcfg2.Server.Lint.ServerPlugin):
+    """ find duplicate Pkgmgr entries with the same priority """
+
+    @Bcfg2.Server.Lint.returnErrors
+    def Run(self):
+        pset = set()
+        for plist in self.core.plugins['Pkgmgr'].entries.values():
+            if self.HandlesFile(plist.name):
+                xdata = plist.data
+                # get priority, type, group
+                priority = xdata.getroot().get('priority')
+                ptype = xdata.getroot().get('type')
+                for pkg in xdata.findall("//Package"):
+                    if pkg.getparent().tag == 'Group':
+                        grp = pkg.getparent().get('name')
+                        if (type(grp) is not str and
+                            grp.getparent().tag == 'Group'):
+                            pgrp = grp.getparent().get('name')
+                        else:
+                            pgrp = 'none'
+                    else:
+                        grp = 'none'
+                        pgrp = 'none'
+                    ptuple = (pkg.get('name'), priority, ptype, grp, pgrp)
+                    # check if package is already listed with same
+                    # priority, type, grp
+                    if ptuple in pset:
+                        self.LintWarning("Duplicate Package %s, priority:%s, type:%s" %
+                                         (pkg.get('name'), priority, ptype))
+                    else:
+                        pset.add(ptuple)
diff --git a/src/lib/Server/Lint/RequiredAttrs.py b/src/lib/Server/Lint/RequiredAttrs.py
new file mode 100644
index 000000000..7215fe163
--- /dev/null
+++ b/src/lib/Server/Lint/RequiredAttrs.py
@@ -0,0 +1,69 @@
+import os.path
+import lxml.etree
+import Bcfg2.Server.Lint
+
+class RequiredAttrs(Bcfg2.Server.Lint.ServerPlugin):
+    """ verify attributes for configuration entries (as defined in
+    doc/server/configurationentries) """
+
+    def __init__(self, *args, **kwargs):
+        Bcfg2.Server.Lint.ServerPlugin.__init__(self, *args, **kwargs)
+        self.required_attrs = {
+            'device': ['name', 'owner', 'group', 'dev_type'],
+            'directory': ['name', 'owner', 'group', 'perms'],
+            'file': ['name', 'owner', 'group', 'perms'],
+            'hardlink': ['name', 'to'],
+            'symlink': ['name', 'to'],
+            'ignore': ['name'],
+            'nonexistent': ['name'],
+            'permissions': ['name', 'owner', 'group', 'perms']}
+
+    @Bcfg2.Server.Lint.returnErrors
+    def Run(self):
+        self.check_rules()
+        self.check_bundles()
+
+    def check_rules(self):
+        """ check Rules for Path entries with missing attrs """
+        if 'Rules' in self.core.plugins:
+            for rules in self.core.plugins['Rules'].entries.values():
+                xdata = rules.pnode.data
+                for path in xdata.xpath("//Path"):
+                    self.check_entry(path, os.path.join(self.config['repo'],
+                                                        rules.name))
+
+    def check_bundles(self):
+        """ check bundles for BoundPath entries with missing attrs """
+        for bundle in self.core.plugins['Bundler'].entries.values():
+            try:
+                xdata = lxml.etree.XML(bundle.data)
+            except AttributeError:
+                xdata = lxml.etree.parse(bundle.template.filepath).getroot()
+
+            for path in xdata.xpath("//BoundPath"):
+                self.check_entry(path, bundle.name)
+
+    def check_entry(self, entry, filename):
+        """ generic entry check """
+        if self.HandlesFile(filename):
+            pathname = entry.get('name')
+            pathtype = entry.get('type')
+            pathset = set(entry.attrib.keys())
+            try:
+                required_attrs = set(self.required_attrs[pathtype] + ['type'])
+            except KeyError:
+                self.LintError("Unknown path type %s: %s" %
+                               (pathtype, self.RenderXML(entry)))
+
+            if 'dev_type' in required_attrs:
+                dev_type = entry.get('dev_type')
+                if dev_type in ['block', 'char']:
+                    # check if major/minor are specified
+                    required_attrs |= set(['major', 'minor'])
+            if not pathset.issuperset(required_attrs):
+                self.LintError("The required attributes %s are missing for %s %sin %s:\n%s" %
+                               (",".join([attr
+                                          for attr in
+                                          required_attrs.difference(pathset)]),
+                                entry.tag, pathname, filename,
+                                self.RenderXML(entry)))
diff --git a/src/lib/Server/Lint/Validate.py b/src/lib/Server/Lint/Validate.py
new file mode 100644
index 000000000..bb5af93f4
--- /dev/null
+++ b/src/lib/Server/Lint/Validate.py
@@ -0,0 +1,185 @@
+import glob
+import lxml.etree
+import os
+import fnmatch
+import Bcfg2.Options
+import Bcfg2.Server.Lint
+from subprocess import Popen, PIPE, STDOUT
+
+class Validate(Bcfg2.Server.Lint.ServerlessPlugin):
+    """ Ensure that the repo validates """
+
+    def __init__(self, *args, **kwargs):
+        Bcfg2.Server.Lint.ServerlessPlugin.__init__(self, *args, **kwargs)
+        self.filesets = {"metadata:groups":"%s/metadata.xsd",
+                         "metadata:clients":"%s/clients.xsd",
+                         "info":"%s/info.xsd",
+                         "%s/Bundler/*.{xml,genshi}":"%s/bundle.xsd",
+                         "%s/Pkgmgr/*.xml":"%s/pkglist.xsd",
+                         "%s/Base/*.xml":"%s/base.xsd",
+                         "%s/Rules/*.xml":"%s/rules.xsd",
+                         "%s/etc/report-configuration.xml":"%s/report-configuration.xsd",
+                         "%s/Svcmgr/*.xml":"%s/services.xsd",
+                         "%s/Deps/*.xml":"%s/deps.xsd",
+                         "%s/Decisions/*.xml":"%s/decisions.xsd",
+                         "%s/Packages/config.xml":"%s/packages.xsd",
+                         "%s/GroupPatterns/config.xml":"%s/grouppatterns.xsd"}
+
+        self.filelists = {}
+        self.get_filelists()
+
+    @Bcfg2.Server.Lint.returnErrors
+    def Run(self):
+        self.schemadir = self.config['schema']
+        
+        for schemaname, path in self.filesets.items():
+            try:
+                filelist = self.filelists[path]
+            except KeyError:
+                filelist = []
+                
+            if filelist:
+                # avoid loading schemas for empty file lists
+                try:
+                    schema = lxml.etree.XMLSchema(lxml.etree.parse(schemaname %
+                                                                   schemadir))
+                except:
+                    self.LintWarning("Failed to process schema %s",
+                                     schemaname % schemadir)
+                    continue
+                for filename in filelist:
+                    self.validate(filename, schemaname % schemadir,
+                                  schema=schema)
+
+        self.check_properties()
+
+    def check_properties(self):
+        """ check Properties files against their schemas """
+        alert = self.logger.debug
+        if "properties_schema" in self.config:
+            if self.config['properties_schema'].lower().startswith('warn'):
+                alert = self.LintWarning
+            elif self.config['properties_schema'].lower().startswith('require'):
+                alert = self.LintError
+            
+        for filename in self.filelists['props']:
+            schemafile = "%s.xsd" % os.path.splitext(filename)[0]
+            if os.path.exists(schemafile):
+                self.validate(filename, schemafile)
+            else:
+                alert("No schema found for %s" % filename)
+
+    def validate(self, filename, schemafile, schema=None):
+        """validate a file against the given lxml.etree.Schema.
+        return True on success, False on failure """
+        if schema is None:
+            # if no schema object was provided, instantiate one
+            try:
+                schema = lxml.etree.XMLSchema(lxml.etree.parse(schemafile))
+            except:
+                self.LintWarning("Failed to process schema %s" % schemafile)
+                return False
+
+        try:
+            datafile = lxml.etree.parse(filename)
+        except SyntaxError:
+            lint = Popen(["xmllint", filename], stdout=PIPE, stderr=STDOUT)
+            self.LintError("%s fails to parse:\n%s" % (filename,
+                                                       lint.communicate()[0]))
+            lint.wait()
+            return False
+        except IOError:
+            self.LintError("Failed to open file %s" % filename)
+            return False
+    
+        if not schema.validate(datafile):
+            cmd = ["xmllint"]
+            if self.files is None:
+                cmd.append("--xinclude")
+            cmd.extend(["--noout", "--schema", schemafile, filename])
+            lint = Popen(cmd, stdout=PIPE, stderr=STDOUT)
+            output = lint.communicate()[0]
+            if lint.wait():
+                self.LintError("%s fails to verify:\n%s" % (filename, output))
+                return False
+        return True
+
+    def get_filelists(self):
+        """ get lists of different kinds of files to validate """
+        if self.files is not None:
+            listfiles = lambda p: fnmatch.filter(self.files, p % "*")
+        else:
+            listfiles = lambda p: glob.glob(p % self.config['repo'])
+
+        for path in self.filesets.keys():
+            if path.startswith("metadata:"):
+                mtype = path.split(":")[1]
+                self.filelists[path] = self.get_metadata_list(mtype)
+            elif path == "info":
+                if self.files is not None:
+                    self.filelists[path] = \
+                                         [f for f in self.files
+                                          if os.path.basename(f) == 'info.xml']
+                else: # self.files is None
+                    self.filelists[path] = []
+                    for infodir in ['Cfg', 'TGenshi', 'TCheetah']:
+                        for root, dirs, files in os.walk('%s/%s' %
+                                                         (self.config['repo'],
+                                                          infodir)):
+                            self.filelists[path].extend([os.path.join(root, f)
+                                                         for f in files
+                                                         if f == 'info.xml'])
+            else:
+                self.filelists[path] = listfiles(path)
+
+        self.filelists['props'] = listfiles("%s/Properties/*.xml")
+        all_metadata = listfiles("%s/Metadata/*.xml")
+
+        # if there are other files in Metadata that aren't xincluded
+        # from clients.xml or groups.xml, we can't verify them.  warn
+        # about those.
+        for fname in all_metadata:
+            if (fname not in self.filelists['metadata:groups'] and
+                fname not in self.filelists['metadata:clients']):
+                self.LintWarning("Broken XInclude chain: Could not determine file type of %s" % fname)
+
+    def get_metadata_list(self, mtype):
+        """ get all metadata files for the specified type (clients or
+        group) """
+        if self.files is not None:
+            rv = fnmatch.filter(self.files, "*/Metadata/%s.xml" % mtype)
+        else:
+            rv = glob.glob("%s/Metadata/%s.xml" % (self.config['repo'], mtype))
+
+        # attempt to follow XIncludes.  if the top-level files aren't
+        # listed in self.files, though, there's really nothing we can
+        # do to guess what a file in Metadata is
+        if rv:
+            rv.extend(self.follow_xinclude(rv[0]))
+
+        return rv
+
+    def follow_xinclude(self, xfile):
+        """ follow xincludes in the given file """
+        xdata = lxml.etree.parse(xfile)
+        included = set([ent.get('href') for ent in
+                        xdata.findall('./{http://www.w3.org/2001/XInclude}include')])
+        rv = []
+        
+        while included:
+            try:
+                filename = included.pop()
+            except KeyError:
+                continue
+
+            path = os.path.join(os.path.dirname(xfile), filename)
+            if self.HandlesFile(path):
+                rv.append(path)
+                groupdata = lxml.etree.parse(path)
+                [included.add(el.get('href'))
+                 for el in
+                 groupdata.findall('./{http://www.w3.org/2001/XInclude}include')]
+                included.discard(filename)
+
+        return rv
+
diff --git a/src/lib/Server/Lint/__init__.py b/src/lib/Server/Lint/__init__.py
new file mode 100644
index 000000000..4e6d03fb5
--- /dev/null
+++ b/src/lib/Server/Lint/__init__.py
@@ -0,0 +1,90 @@
+__revision__ = '$Revision$'
+
+__all__ = ['Bundles',
+           'Comments',
+           'Duplicates',
+           'InfoXML',
+           'Pkgmgr',
+           'RequiredAttrs',
+           'Validate']
+
+import logging
+import os.path
+from copy import copy
+import lxml.etree
+import Bcfg2.Logger
+
+def returnErrors(fn):
+    """ Decorator for Run method that returns error counts """
+    def run(self, *args, **kwargs):
+        fn(self, *args, **kwargs)
+        return (self.error_count, self.warning_count)
+
+    return run
+
+class Plugin (object):
+    """ base class for ServerlessPlugin and ServerPlugin """
+    def __init__(self, config, files=None):
+        self.files = files
+        self.error_count = 0
+        self.warning_count = 0
+        self.config = config
+        Bcfg2.Logger.setup_logging('bcfg2-info', to_syslog=False)
+        self.logger = logging.getLogger('bcfg2-lint')
+
+    def Run(self):
+        """ run the plugin.  must be overloaded by child classes """
+        pass
+
+    def HandlesFile(self, fname):
+        """ returns true if the given file should be handled by the
+        plugin according to the files list, false otherwise """
+        return (self.files is None or
+                fname in self.files or
+                os.path.join(self.config['repo'], fname) in self.files or
+                os.path.abspath(fname) in self.files or
+                os.path.abspath(os.path.join(self.config['repo'],
+                                             fname)) in self.files)
+    
+    def LintError(self, msg):
+        """ log an error condition """
+        self.error_count += 1
+        lines = msg.splitlines()
+        self.logger.error("ERROR: %s" % lines.pop())
+        [self.logger.error("  %s" % l) for l in lines]
+
+    def LintWarning(self, msg):
+        """ log a warning condition """
+        self.warning_count += 1
+        lines = msg.splitlines()
+        self.logger.warning("WARNING: %s" % lines.pop())
+        [self.logger.warning("  %s" % l) for l in lines]
+
+    def RenderXML(self, element):
+        """render an XML element for error output -- line number
+        prefixed, no children"""
+        xml = None
+        if len(element) or element.text:
+            el = copy(element)
+            if el.text:
+                el.text = '...'
+            [el.remove(c) for c in el.iterchildren()]
+            xml = lxml.etree.tostring(el).strip()
+        else:
+            xml = lxml.etree.tostring(element).strip()
+        return "   line %s: %s" % (element.sourceline, xml)
+
+class ServerlessPlugin (Plugin):
+    """ base class for plugins that are run before the server starts
+    up (i.e., plugins that check things that may prevent the server
+    from starting up) """
+    pass
+
+class ServerPlugin (Plugin):
+    """ base class for plugins that check things that require the
+    running Bcfg2 server """
+    def __init__(self, lintCore, config, files=None):
+        Plugin.__init__(self, config, files=files)
+        self.core = lintCore
+        self.logger = self.core.logger
+        self.metadata = self.core.metadata