#!/bin/sh

LANGUAGES=en

for LANG in ${LANGUAGES}; do
    rm -f *er?.${LANG}.html
    wget -q -O- http://www.debian.org/CD/index.${LANG}.html | awk -v LANG=$LANG '

        /^<div id="header">/               {state="header1"}
        /^<p id="breadcrumbs">/            {state=""}

        /^<\/div> <!-- end header -->/     {state="header2"}
        /^<div id="maincol">/              {state=""}

        /^<\/div> <!-- end maincol -->/    {state="footer1"}
        /^<div id="pageLang">/             {state=""}

        /\/MailingLists\/disclaimer/       {state="footer2"}
        /^<\/div> <!-- end footermap -->/  {print $0 >> "footer2."LANG".html";state=""}

/.*/ {
         gsub("\"\\.\\./","\"http://www.debian.org/")
         gsub("\"/\"","\"http://www.debian.org/\"")
         gsub("/MailingLists/","http://www.debian.org/MailingLists/")
         if (state == "header1") { print $0 >> "header1."LANG".html" }
         if (state == "header2") { print $0 >> "header2."LANG".html" }
         if (state == "footer1") { print $0 >> "footer1."LANG".html" }
         if (state == "footer2") { print $0 >> "footer2."LANG".html" }
     }'

done