[PerlWikiBot] SF.net SVN: perlwikibot:[83] trunk/no-interwiki

[<am...@us...>]

Revision: 83
          http://perlwikibot.svn.sourceforge.net/perlwikibot/?rev=83&view=rev
Author:   amire80
Date:     2009-09-29 13:25:11 +0000 (Tue, 29 Sep 2009)

Log Message:
-----------
Refactoring empty pages handling.

Modified Paths:
--------------
    trunk/no-interwiki/cv.strings.txt
    trunk/no-interwiki/prepare_noiw_list.pl

Modified: trunk/no-interwiki/cv.strings.txt
===================================================================
--- trunk/no-interwiki/cv.strings.txt	2009-09-27 12:41:34 UTC (rev 82)
+++ trunk/no-interwiki/cv.strings.txt	2009-09-29 13:25:11 UTC (rev 83)
@@ -23,7 +23,8 @@
 type	тĕс
 
 # MW specials
-REDIRECT	куçару
+#REDIRECT	куçару
+REDIRECT	перенаправление	
 
 # Namespaces
 User	Хутшăнакан

Modified: trunk/no-interwiki/prepare_noiw_list.pl
===================================================================
--- trunk/no-interwiki/prepare_noiw_list.pl	2009-09-27 12:41:34 UTC (rev 82)
+++ trunk/no-interwiki/prepare_noiw_list.pl	2009-09-29 13:25:11 UTC (rev 83)
@@ -463,55 +463,40 @@
             say $page_counter;
         }
 
-        last PAGE
-            if ($option{stop_after}
-            and $page_counter > $option{stop_after});
-
-        next PAGE
-            if ($page_counter < $option{start_from});
-
-        my $page_namespace = namespace($page_ref);
-        my $page_title     = $page_ref->title();
-        my $page_text_ref  = $page_ref->text();
-
-        if (not defined $page_text_ref) {
-            WARN('ref to page text undefined, skipping');
-            $statistics{'ref to page text undefined'}++;
-            next PAGE;
+        if (    $option{stop_after}
+            and $page_counter > $option{stop_after})
+        {
+            last PAGE;
         }
 
-        if (not defined ${$page_text_ref}) {
-            WARN('page text undefined, skipping');
-            $statistics{'page text undefined'}++;
-            next PAGE;
-        }
+        my $page_text_ref = $page_ref->text();
 
-        if (${$page_text_ref} eq q()) {
-            WARN('page text empty, skipping');
-            $statistics{'page text empty'}++;
+        if (
+            $page_counter < $option{start_from}
+            or not defined ${$page_text_ref}  # must be tested before redirect
+            or not is_in_namespace($page_ref, @INCLUDE_NAMESPACES)
+            or is_redirect($page_ref)
+            )
+        {
             next PAGE;
         }
 
-        if (not is_in_namespace($page_ref, @INCLUDE_NAMESPACES)) {
-            next PAGE;
-        }
+        my $page_namespace = namespace($page_ref);
+        $namespace_count{$page_namespace}++;
 
-        if (my $redirect_type = is_redirect($page_ref)) {
-            $statistics{"redirect - $redirect_type"}++;
-            DEBUG("\n$page_title - $redirect_type redirect");
-            next PAGE;
-        }
+        my $page_title = $page_ref->title();
 
+        INFO("\n* processing $page_counter - ", $page_title);
+
         # TODO: Be more precise here.
         # Portal pages which have a '/' in their name are probably
         # internal and do not need interwiki links.
         if (is_in_namespace($page_ref, 'Portal') and $page_title =~ m{/}xms) {
+            INFO('internal portal, skipping');
+            $statistics{'internal portal'}++;
             next PAGE;
         }
 
-        $namespace_count{$page_namespace}++;
-        INFO("\n* processing $page_counter - ", $page_title);
-
         # A simple sanity check: is the no_iw template anywhere around here?
         my $has_template_no_iw =
             (${$page_text_ref} =~ $SIMPLE_NO_IW_CHECK_RE);


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.