Merge branch 'advanced_persistence' of code.communitydata.cc:/mediawiki_dump_tools...

[mediawiki_dump_tools.git] / wikiq

diff --git a/wikiq b/wikiq
index 7cf4be26e60db633db9d210e7f5ec74addf1f357..a584390b4eedef9b606ba5dff7ac5b0d0fcccbb6 100755 (executable)
--- a/wikiq
+++ b/wikiq
@@ -63,6 +63,11 @@ class WikiqPage():
     def __init__(self, page, namespace_map, collapse_user=False):
         self.id = page.id
         self.namespace = page.namespace
+        # following mwxml, we assume namespace 0 in cases where
+        # page.namespace is inconsistent with namespace_map
+        if page.namespace not in namespace_map:
+            self.title = page.title
+            page.namespace = 0
         if page.namespace != 0:
             self.title = ':'.join([namespace_map[page.namespace], page.title])
         else:
@@ -368,8 +373,12 @@ parser.add_argument('-p', '--persistence', dest="persist", default=None, const='
 parser.add_argument('-u', '--url-encode', dest="urlencode", action="store_true",
                     help="Output url encoded text strings. This works around some data issues like newlines in editor names. In the future it may be used to output other text data.")
 
-args = parser.parse_args()
+parser.add_argument('-n', '--namespace-include', dest="namespace_filter", type=int, action='append',
+                    help="Id number of namspace to include.")
+
 
+args = parser.parse_args()
+print(args)
 # set persistence method
 
 if args.persist is None:
@@ -404,7 +413,6 @@ if len(args.dumpfiles) > 0:
                                 persist=persist,
                             urlencode=args.urlencode)
 
-
         wikiq.process()
 
         # close things