Changed the extension filtering from using a blacklist to a whitelist, to avoid a lot of incorrect types.

 - Legacy-Id: 13487

Author: levkowetz

ietf/settings.py

@@ -558,7 +558,7 @@ def skip_unreadable_post(record):
 INTERNET_ALL_DRAFTS_ARCHIVE_DIR = '/a/www/www6s/archive/id'
 MEETING_RECORDINGS_DIR = '/a/www/audio'
 
-DOCUMENT_FORMAT_BLACKLIST = ["tar", "dtd", "p7s"]
+DOCUMENT_FORMAT_WHITELIST = ["txt", "ps", "pdf", "xml", "html", ]
 
 # Mailing list info URL for lists hosted on the IETF servers
 MAILING_LIST_INFO_URL = "https://www.ietf.org/mailman/listinfo/%(list_addr)s"

ietf/stats/views.py

@@ -328,7 +328,8 @@ def generate_canonical_names(docalias_qs):
                     continue
 
                 basename, ext = t
-                if any(ext.lower().endswith(blacklisted_ext.lower()) for blacklisted_ext in settings.DOCUMENT_FORMAT_BLACKLIST):
+                ext = ext.lower()
+                if not any(ext==whitelisted_ext for whitelisted_ext in settings.DOCUMENT_FORMAT_WHITELIST):
                     continue
 
                 canonical_name = doc_names_with_missing_types.get(basename)