working

h4ck3rm1k3 · Jan 23, 2016 · 8adb792 · 8adb792
1 parent c2591c3
commit 8adb792
Show file tree

Hide file tree

Showing 4 changed files with 21 additions and 5 deletions.
diff --git a/dumpgenerator.py b/dumpgenerator.py
@@ -135,9 +135,15 @@ def getNamespaces(config={}):
 
 def getSDTitles(site):
     titles = []
-    for x in (        
+    for x in (
+            'AfD_debates',
+            'Articles for deletion',
+            'Miscellaneous_pages_for_deletion',
+            'Templates_for_deletion',
+            'All articles proposed for deletion',
             'Declined_AfC_submissions',
             "Proposed_deletion",
+            "Proposed deletions needing attention",
             "Expired_proposed_deletions",
             'Candidates_for_speedy_deletion_as_hoaxes',
             'Candidates_for_speedy_deletion_as_importance_or_significance_not_asserted',

diff --git a/families/wikipedia_family.py b/families/wikipedia_family.py
@@ -950,6 +950,17 @@ def __init__(self):
             'en': u'Draft talk',
         }
 
+        self.namespaces[710] = { 
+            'en': u'TimedText', 
+            'pt': u'TimedText', 
+        } 
+
+        self.namespaces[711] = { 
+            'en': u'TimedText talk', 
+            'pt': u'TimedText talk', 
+        }
+
+
         self.category_redirect_templates = {
             '_default': (),
             'ar': (u'تحويل تصنيف',

diff --git a/process.py b/process.py
@@ -8,7 +8,7 @@
 path = './'
 for infile in glob.glob( os.path.join(path, '*') ):
     print "current file is: " + infile
-    match = re.search(r'(\d\d\d\d)(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)',infile)
+    #match = re.search(r'(\d\d\d\d)(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)',infile)
     match = re.search(r'w\-(\d+)\-wiki',infile)
 #    match = re.search(r'.+(dddddddddddddd).+',infile)
 #    match = re.search(r'.+(\dddddddddddddd).+',infile)

diff --git a/runexport.sh b/runexport.sh
@@ -1,12 +1,11 @@
 #set -e 
 
 export PYTHONPATH=$HOME/experiments/wikiteam/
-cd ${HOME}/experiments/wikiteam/data
+pushd ${HOME}/experiments/wikiteam/data
 python $HOME/experiments/wikiteam/dumpgenerator.py
-
 echo going to run process
-cd ${HOME}/experiments/wikiteam/data
 #pwd
 python $HOME/experiments/wikiteam/process.py
+popd
 rm -rf  ${HOME}/experiments/wikiteam/data/enwikipedia*
 #rm -rf  ${HOME}/experiments/wikiteam/data/Api*