[PerlWikiBot] SF.net SVN: perlwikibot:[86] trunk/no-interwiki/prepare_noiw_list.pl

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

Revision: 86
          http://perlwikibot.svn.sourceforge.net/perlwikibot/?rev=86&view=rev
Author:   amire80
Date:     2010-03-25 15:02:28 +0000 (Thu, 25 Mar 2010)

Log Message:
-----------
Update of empty pages handling.

Modified Paths:
--------------
    trunk/no-interwiki/prepare_noiw_list.pl

Modified: trunk/no-interwiki/prepare_noiw_list.pl
===================================================================
--- trunk/no-interwiki/prepare_noiw_list.pl	2010-03-09 12:44:41 UTC (rev 85)
+++ trunk/no-interwiki/prepare_noiw_list.pl	2010-03-25 15:02:28 UTC (rev 86)
@@ -280,9 +280,7 @@
 my $SECTION_LINK_RE   = qr{(?<!&)\#}xms;
 my $LOWERCASE_LINK_RE = qr{\A[[:lower:]]}xms;
 
-## no critic (RegularExpressions::ProhibitEscapedMetacharacters)
 my $TRUE_TEMPLATE_RE = qr/\{ $RE{balanced}{-parens=>'{}'} \}/xms;
-## use critic (RegularExpressions::ProhibitEscapedMetacharacters)
 
 # get_string() cannot be used in re
 my $string_exclude_lowercase = get_string('exclude_lowercase');
@@ -420,23 +418,9 @@
 exit;
 
 sub load_dump {
-    Readonly my $WORKING_PMWD_VER => 0.91;
-    Readonly my $FUTURE_PMWD_VER  => 0.94;
-
     my $dump;
 
-    if ($Parse::MediaWikiDump::VERSION == $WORKING_PMWD_VER) {
-        $dump = Parse::MediaWikiDump::Pages->new($dump_fn);
-    }
-    else {
-        if ($Parse::MediaWikiDump::VERSION < $FUTURE_PMWD_VER) {
-            carp(     'You are running Parse::MediaWikiDump version '
-                    . $Parse::MediaWikiDump::VERSION
-                    . ".\n Redirect handling may be broken\n");
-        }
-        my $pmwd = Parse::MediaWikiDump->new();
-        $dump = $pmwd->revisions($dump_fn);
-    }
+    $dump = Parse::MediaWikiDump::Pages->new($dump_fn);
 
     return $dump;
 }
@@ -471,6 +455,12 @@
 
         my $page_text_ref = $page_ref->text();
 
+        if (${$page_text_ref} !~ /\S/xmsi
+            and not is_in_namespace($page_ref, 'User', 'User talk'))
+        {
+            special_cases_file('empty_page', $page_ref);
+        }
+
         if (
             $page_counter < $option{start_from}
             or not defined ${$page_text_ref}  # must be tested before redirect
@@ -1266,6 +1256,12 @@
 sub print_multi_links_by_foreign {
     LANG_CODE:
     foreach my $lang_code (sort keys %found_links) {
+        say "processing $lang_code";
+        my $lang_reftype = ref $found_links{$lang_code};
+        if ($lang_reftype ne 'HASH') {
+            carp('$lang_code is $lang_reftype, not hashref!');
+            next LANG_CODE;
+        }
         my $filename         = "$MULTI_DIR/$lang_code.$WIKITEXT_EXT";
         my @foreign_articles = sort keys %{ $found_links{$lang_code} };
         say format_link_table($lang_code, scalar @foreign_articles);
@@ -1274,6 +1270,14 @@
         foreach my $foreign_article (@foreign_articles) {
             my @local_articles =
                 keys %{ $found_links{$lang_code}->{$foreign_article} };
+
+            my $article_reftype =
+                ref $found_links{$lang_code}->{$foreign_article};
+            if ($article_reftype ne 'HASH') {
+                carp("$foreign_article is $article_reftype, not hashref!");
+                next FOREIGN_ARTICLE;
+            }
+
             if (scalar @local_articles > 1) {
                 my $links = join q{ | }, sort map { make_link($_) }
                     keys %{ $found_links{$lang_code}->{$foreign_article} };


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