tumblr_download

#!/bin/bash

cd ~/play/tumblr

true && for (( j = 0; j < 100; j = j + 50 )) ; do
	while curl "http://bobbyatto.tumblr.com/api/read?start=$j&num=50" > t && \
			fgrep -q 'Our servers are over capacity' t 
	do
		echo "Tumblr overload; try later" >&2
		sleep 1h
	done

	num=$(xmllint --xpath 'count(//tumblr/posts/post/photo-url)' t)
	for ((i=1;i<=num;i++)) ; do
		xmllint --xpath "(//tumblr/posts/post/photo-url)[$i]/text()" t
		printf '\n'
	done 
done > all_urls

{
	while read line ; do

		declare -A size
		declare -A url

		x=${line##*/}
		x=${x#tumblr_}
		y="a" ; z=0

		[[ $x = *75sq.??? ]] && continue

		if [[ $x = *1_???.??? ]] ; then
			z="${x##*1_}" ; z="${z%.???}"
			y="${x%?1_???.???}"
		else
			z=10000
			y=$x
		fi
		false && printf "%s %s %s %s\n" "$x" "$y" "$z" "$line"
		[[ -z ${size[$y]} ]] && size[$y]=0
		if (( ${size[$y]} < z )) ; then
			size[$y]=$z
			url[$y]=$line
		fi

	done < all_urls
	printf "%s\n" "${url[@]}"
} > urls
wget -i urls -nc --wait=1