# This is a script to identify pages with identical (but differently cased) names in the main namespace
# It's intended to be used before changing the mediawiki namespace to a case-insensitive collation
#
# Usage : ./find_wiki_dups.sh <database name> <namespace id> <URL path to wiki pages>

DBNAME=$1
NAMESPACE=$2
URL=$3

#!/bin/bash
for page in `mysql $DBNAME -s -N -e "select page_title  from mediawiki_page where page_namespace=$NAMESPACE group by lower(page_title) having count(lower(page_title)) > 1 order by page_touched asc;"`
do
        echo "DUP FOUND [$page] - Bottom one is the more recent"
        for dup in `mysql $DBNAME -s -N -e "select page_title from mediawiki_page where lower(page_title) = lower('$page')order by page_touched asc;"`
        do
                echo "$URL?title=$dup&action=delete"
        done

done