[Pywikipedia-l] SVN: [5342] branches/rewrite/pywikibot/site.py

[nicdumz at svn.wikimedia.org]

Revision: 5342
Author:   nicdumz
Date:     2008-05-09 11:35:15 +0000 (Fri, 09 May 2008)

Log Message:
-----------
* Adding a batch parameter to categorymembers to set the numbers of pages retrieved at once
* Trying to address NoPage issues

Modified Paths:
--------------
    branches/rewrite/pywikibot/site.py

Modified: branches/rewrite/pywikibot/site.py
===================================================================
--- branches/rewrite/pywikibot/site.py	2008-05-09 10:56:35 UTC (rev 5341)
+++ branches/rewrite/pywikibot/site.py	2008-05-09 11:35:15 UTC (rev 5342)
@@ -744,7 +744,7 @@
                                                       for ns in namespaces)
         return tlgen
 
-    def categorymembers(self, category, namespaces=None):
+    def categorymembers(self, category, namespaces=None, batch=None):
         """Iterate members of specified category.
 
         @param category: The Category to iterate.
@@ -754,6 +754,8 @@
             however, that the iterated values are always Page objects, even
             if in the Category or Image namespace.
         @type namespaces: list of ints
+        @param batch: the number of pages to fetch each time.
+        @type batch: int
 
         """
         if category.namespace() != 14:
@@ -766,6 +768,11 @@
         if namespaces is not None:
             cmgen.request[u"gcmnamespace"] = u"|".join(unicode(ns)
                                                       for ns in namespaces)
+        if batch is not None:
+            if batch > 5000:
+                logging.debug("No more than 5000 rows can be fetched at once.")
+                batch=5000
+            cmgen.request[u'cmlimit'] = str(batch)
         return cmgen
 
     def loadrevisions(self, page=None, getText=False, revids=None,
@@ -881,6 +888,8 @@
                     raise Error(
                         u"loadrevisions: Query on %s returned data on '%s'"
                         % (page, pagedata['title']))
+                if pagedata.has_key('missing'):
+                    raise NoPage(u'Page %s does not exist' % page.title(asLink=True)) 
             else:
                 page = Page(self, pagedata['title'])
             api.update_page(page, pagedata)