flickr-get.sh

#!/bin/sh
# flickr-get, version 0.5.
#
# Downloads entire flickr photostreams with the help of wget.
# This is somewhat hacky, and will probably break when flickr
# changes its website layout.
#
# Licenced under the Public Domain or whatever.
 
# Temporary directory location.
tempdir=".flickr-get.$$"
 
# Check for command-line arguments.
if [ -z "$1" ]
then
        echo "Usage: $0 [flickr-url] [cookies-file]"
        exit 1
fi
 
# Check if [cookies-file] was specified. If it was, set $cookies to the
# absolute path of the cookies file.
if [ -n "$2" ]
then
        cd "`dirname "$2"`"
        cookies="`pwd`/`basename "$2"`"
        cd "$OLDPWD"
fi
 
mkdir $tempdir
cd $tempdir
 
# Download first page, check if it returns a photo-gallery. If not, retry
# with the use of a cookies file (if specified).
wget -q -E "$1/page1"
grep photo_container page1.html &>/dev/null
if [ $? -eq 1 ]
then
        if [ -e "$cookies" ]
        then
                wget_cookies="--load-cookies "$cookies""
                rm page1.html
                wget -q -E $wget_cookies "$1/page1"
        else
                echo -e "\033[1mDownloading this photostream requires a valid cookies.txt file.\033[0m"
                rm -Rf $tempdir
                exit 1
        fi
fi
 
# Find the last page in a photostream.
lastpage=`grep -A 2 '<span class="break">...</span>' page1.html | tail -1 | cut -d">" -f2 | cut -d"<" -f1`
 
# Start downloading all pages up to the last one. Cookies will be used
# if they were needed for the first page.
echo -e "\033[1mDownloading temporary files...\033[0m"
for i in `seq 2 $lastpage`
do
        wget -q -E $wget_cookies "$1/page$i"
done
 
# Extract a list of photo URLs, removing the size specification (along
# with the extension) and dumping the resulting list in 'flickr-photos'.
grep -h photo_container *.html | cut -d'"' -f8 | sed 's/_m.jpg//' > flickr-photos
 
cd "$OLDPWD"
 
# Start downloading photos, first trying for original (or large) size
# images and moving to medium size images if the originals are unavailable.
echo -e "\033[1mDownloading photos...\033[0m"
for i in `cat $tempdir/flickr-photos`
do
        wget -q -nc "${i}_b.jpg"
        if [ -f "photo_unavailable.gif" ]
        then
                rm "photo_unavailable.gif"
                wget -q -nc "${i}_z.jpg"
        fi
done
 
# Clean up and exit.
#rm -Rf $tempdir
echo -e "\033[1mDone!\033[0m"
 
exit 0