xml2rfc is vulnerable to arbitrary file reads through prepped files

@@ -755,6 +755,11 @@ def getroot(self):
     def getpis(self):
         """ Returns a list of the XML processing instructions """
         return self.pis.copy()
+
+    def sanitize(self):
+        """ Sanitize the document """
+        # Strip link attachments
+        xml2rfc.utils.strip_link_attachments(self.tree)
     
     def validate(self, dtd_path=None):
         """ Validate the document with its default dtd, or an optional one 

@@ -577,6 +577,9 @@ def main():
             xml2rfc.log.exception('Unable to validate the XML document: ' + args[0], errors)
             sys.exit(1)
 
+    # sanitize the document
+    xmlrfc.sanitize()
+
     if options.filename:
         xml2rfc.log.warn("The -f and --filename options are deprecated and will"
                         " go away in version 3.0 of xml2rfc.  Use -o instead")

@@ -28,7 +28,7 @@
 from xml2rfc.util.file import can_access, FileAccessError
 from xml2rfc.util.name import short_author_ascii_name_parts, full_author_name_expansion, short_author_name_parts
 from xml2rfc.util.unicode import is_svg
-from xml2rfc.utils import namespaces, find_duplicate_ids, slugify, strip_link_attachments
+from xml2rfc.utils import find_duplicate_ids, namespaces, slugify
 
 
 SUBSERIES = {
@@ -2117,9 +2117,6 @@ def validate(self, when='', warn=False):
         # a duplicate xsd:ID attribute.  So we check all attributes with
         # content specified as xsd:ID first, and give better messages:
 
-        # Strip link attachments
-        strip_link_attachments(self.tree)
-
         # Get the attributes we need to check
         if when and not when.startswith(' '):
             when = ' '+when