#!/bin/bash

if [ "X$2" == X ]; then
	echo "Usage: compress-html <language> <edition>"
	exit
fi


lang=$1
edition=$2
site=wikipedia

base=/mnt/static
sitebase=$base/$site
dest=$base/downloads/$edition/$lang
p7zip=$base/scripts/7za-readdir-hack

mkdir -p $dest

echo Finding files...
cd $sitebase
find $lang -name \*.html > $dest/html.lst

#echo Filtering...
#php scripts/filterNamespaces.php $lang wikipedia $dest/html.lst > $dest/reduced.lst

find $lang/skins $lang/images -follow -type f \
	-not \( -name \*.php -or -name \*.xcf -or -name .\* -or -name \*~ -or \
	-path \*/CVS/\* -or -path \*/.svn\* -or -path \*/wikimania\* \) \
	> $dest/skins.lst
find $lang/raw -type f >> $dest/skins.lst
echo $lang/upload/b/bc/Wiki.png >> $dest/skins.lst

echo Creating HTML archive...
rm -f $dest/wikipedia-$lang-html.7z

# Set chunk size to 8MB for faster random access
$p7zip -l -ms8m a $dest/wikipedia-$lang-html.7z @$dest/html.lst @$dest/skins.lst

#if [ ! -e $dest/wikipedia-$lang-reduced.7z ];then
#	echo
#	echo Creating reduced archive...
#	7z -l a $dest/wikipedia-$lang-reduced.7z @$dest/reduced.lst @$dest/skins.lst
#fi

fileCount=`wc -l /mnt/static/downloads/April_2007/aa/html.lst | awk '{print $1}'` 
if [ $fileCount -gt 2000000 ]; then
	echo "Creating split archives"
	$base/scripts/compress-volumes "$lang" "$edition"
fi