#!/bin/sh

## back up a URL to S3, by downloading it and then 

# s3://flickr-backup.mortia.org.uk
BUCKET="$1"
SRCURL="$2"

fatal() {
    echo >&2 "$*"
    exit 2
}

if test -z "$BUCKET" -o -z "$SRCURL"; then
    echo >&2 "Usage: $0 bucket srcurl"
    exit 2
fi

# remove the http: and also the hostname, for services that use farms
# of different hosts serving the "same" content
DESTKEY=`echo "$SRCURL" | sed \
    -e 's;^https*://*[^/]*//*;;' \
    -e 's;\&;;g' \
    `

DEST=$BUCKET/$DESTKEY

# is it already there?
ONS3=`s3cmd ls $DEST | grep -v "Bucket '"`
if test -z "$ONS3"; then
    echo "Source: $SRCURL"
    echo "Dest:   $DEST"

    SRCLOCAL=`echo "$SRCURL" | sed -e 's;[/\\*+: &=?];_;g'`
    wget -c -O "$SRCLOCAL" "$SRCURL" || fatal "Download failed"
    s3cmd put "$SRCLOCAL" "$DEST" || fatal "Upload failed"
    rm "$SRCLOCAL"
else
    echo "Already there: $DEST"
fi