repoman: check metadata.xml doctype, bug #328113

diff --git a/bin/repoman b/bin/repoman
index a77b5de780ca0aafee15f2e25c3baa3a64d76741..07f0adc030b55e5dceb80a261c35782f096886c1 100755 (executable)
--- a/bin/repoman
+++ b/bin/repoman
@@ -508,6 +508,7 @@ suspect_virtual = {
        "dev-libs/libusb-compat":"virtual/libusb",
 }
 
+metadata_doctype_name = 'pkgmetadata'
 metadata_dtd_uri = 'http://www.gentoo.org/dtd/metadata.dtd'
 # force refetch if the local copy creation time is older than this
 metadata_dtd_ctime_interval = 60 * 60 * 24 * 7 # 7 days
@@ -1278,8 +1279,12 @@ class _MetadataTreeBuilder(xml.etree.ElementTree.TreeBuilder):
        Implements doctype() as required to avoid deprecation warnings with
        >=python-2.7.
        """
+       def __init__(self, data):
+               xml.etree.ElementTree.TreeBuilder.__init__(self)
+               self._portage_data = data
+
        def doctype(self, name, pubid, system):
-               pass
+               self._portage_data["DOCTYPE"] = (name, pubid, system)
 
 try:
        herd_base = make_herd_base(os.path.join(repoman_settings["PORTDIR"], "metadata/herds.xml"))
@@ -1638,6 +1643,7 @@ for x in effective_scanlist:
        # metadata.xml parse check
        else:
                metadata_bad = False
+               xml_info = {}
 
                # read metadata.xml into memory
                try:
@@ -1645,13 +1651,33 @@ for x in effective_scanlist:
                                _unicode_encode(os.path.join(checkdir, "metadata.xml"),
                                encoding=_encodings['fs'], errors='strict'),
                                parser=xml.etree.ElementTree.XMLParser(
-                                       target=_MetadataTreeBuilder()))
+                                       target=_MetadataTreeBuilder(xml_info)))
                except (ExpatError, SyntaxError, EnvironmentError) as e:
                        metadata_bad = True
                        stats["metadata.bad"] += 1
                        fails["metadata.bad"].append("%s/metadata.xml: %s" % (x, e))
                        del e
                else:
+                       if "DOCTYPE" not in xml_info:
+                               metadata_bad = True
+                               stats["metadata.bad"] += 1
+                               fails["metadata.bad"].append("%s/metadata.xml: %s" % (x,
+                                       "DOCTYPE is missing"))
+                       else:
+                               doctype_name, doctype_pubid, doctype_system = \
+                                       xml_info["DOCTYPE"]
+                               if doctype_system != metadata_dtd_uri:
+                                       stats["metadata.bad"] += 1
+                                       fails["metadata.bad"].append("%s/metadata.xml: "
+                                               "DOCTYPE: SYSTEM should refer to '%s', not '%s'" %
+                                               (x, metadata_dtd_uri, doctype_system))
+
+                               if doctype_name != metadata_doctype_name:
+                                       stats["metadata.bad"] += 1
+                                       fails["metadata.bad"].append("%s/metadata.xml: "
+                                               "DOCTYPE: name should be '%s', not '%s'" %
+                                               (x, metadata_doctype_name, doctype_name))
+
                        # load USE flags from metadata.xml
                        try:
                                musedict = utilities.parse_metadata_use(_metadata_xml)