#!/bin/bash
# Opensource.ORG quick&dirty mirror script.
# (C) 2007 Marcello Barnaba <vjt@openssl.it>
# Released under the terms of the DWTFYW License.

# The absolute base path of the htdocs directory
MIRROR_BASE="/home/httpd/antifork.org/htdocs"

# The directory under which the contents will be downloaded
MIRROR_DIR="opensource.antifork.org"

# The URI to mirror
MIRROR_URI="http://opensource.org"

# A directory containing a placeholder page to show while
# the script is running
WIP_DIR="${MIRROR_DIR}-updating"

# A Temporary working directory
WORK_DIR="${MIRROR_DIR}-$RANDOM"

# The path to the wget(1) program
WGET="/usr/bin/wget --quiet"

# Abort on errors .. yuck!
set -e

# .................................................
# NO USER SERVICEABLE PARTS BELOW THE DOTTED LINE .
# .................................................

pushd "$MIRROR_BASE" > /dev/null

# wget madness downloads robots.txt continuously,
# so erase and re-download every time. for this kind
# of site, it's ok.
# 
rm -rf $MIRROR_DIR
ln -s $WIP_DIR $MIRROR_DIR
mkdir $WORK_DIR

# Download the HTML and IMG stuff ..
# 
$WGET --domains=opensource.org --convert-links --level=0  \
    --mirror --page-requisites --no-host-directories  \
    $MIRROR_URI --exclude-directories '/user,/event'  \
    --directory-prefix="$WORK_DIR" --html-extension

# Get all the stylesheets from the ToS page, one that will never
# disappear, hopefully
# 
STYLESHEETS=$(grep css $WORK_DIR/ToS.html | sed 's#.*import "\([a-z0-9/\.-]*\)".*#\1#')

# Download all the images referenced in the CSS style sheets
# 
for stylesheet in $STYLESHEETS; do
    base=`dirname $stylesheet`
    mkdir -p "$WORK_DIR/$base"
    $WGET -O "$WORK_DIR/$stylesheet" "$MIRROR_URI/$stylesheet"
    sed -n 's#.*url(\(.*\)).*#\1#p' < "$WORK_DIR/$stylesheet" | sort | uniq | while read image; do
        $WGET -O $WORK_DIR/$base/$image $MIRROR_URI/$base/$image
    done
done

# Remove the placeholder page
rm -f $MIRROR_DIR

# Put the mirror online
mv $WORK_DIR $MIRROR_DIR

popd > /dev/null

# EOF