From b5810882e8c6b1e6b76a8239f70a129d415ecee6 Mon Sep 17 00:00:00 2001 From: "Chris St. Pierre" Date: Wed, 20 Apr 2011 09:41:07 -0400 Subject: 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. --- src/lib/Server/Lint/Validate.py | 185 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 185 insertions(+) create mode 100644 src/lib/Server/Lint/Validate.py (limited to 'src/lib/Server/Lint/Validate.py') 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 + -- cgit v1.2.3-1-g7c22