From: Nate E TeBlunthuis <nathante@mox2.hyak.local>
Date: Fri, 24 Aug 2018 01:02:56 +0000 (-0700)
Subject: add namespace filter parameter
X-Git-Url: https://code.communitydata.science/mediawiki_dump_tools.git/commitdiff_plain/5b6aaad86232a364f071f4705d18aaa01a7581f6?hp=-c

add namespace filter parameter
---

5b6aaad86232a364f071f4705d18aaa01a7581f6
diff --git a/wikiq b/wikiq
index 7cf4be2..47dd687 100755
--- 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:
@@ -368,6 +373,9 @@ 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.")
 
+parser.add_argument('-ns', '--namespace-filter', dest="namespace_filter", type=str, help="Comma-seperate list of namespaces numbers to include")
+
+
 args = parser.parse_args()
 
 # set persistence method