#! /bin/sh
site=http://www.softerra.ru
cp1251='þàáöäåôãõèéêëìíîïÿðñòóæâüûçøýù÷úÞÀÁÖÄÅÔÃÕÈÉÊËÌÍÎÏßÐÑÒÓÆÂÜÛÇØÝÙ×Ú¸«»–'
koi8r='À-ßà-ÿ£""-'
# function - getin path-to-file
getin () {
	snarf -p $site/$1 -|
	sed 's/><br></><br>\
</g'|	tr $cp1251 $koi8r > page$n.html
	n=$(($n+1))
}
# function - update directory-at-site num-pages
update () {
n=1	# page counter
getin $1/index.html
for j in $(seq 2 $2)
do  getin $1/index.html?page=$j
done
list=$(perl -wne '
print "$1\n" if m#href="/'$1'/(\d{4,5})/"#
' page[0-9]*.html|sort|uniq)
rm -f page[0-9]*.html
# page list make and get
for i in $list
do	test -s nn/$i.html || {
	snarf -p $site/$1/$i/print.html -|
	sed 's/><!--/>\
<!--/g;s/--></-->\
</g;s#href="/#href="#g;s#src="http://img.softerra.ru/#src="#g'|
sed '/^<!--ctsearch_noindex-->$/,/^<!--\/ctsearch_noindex-->$/d'|
	tr $cp1251 $koi8r > $i.html
	}
done
# get pics
list=$(perl -wne '
print "'$site'/$1\n" if m#src="(pubimages/[^"]+\.(?:gif|jpg))"#
' [0-9][0-9][0-9][0-9]*.html)
test -z "$list" || { 
	echo $list |
	xargs snarf -p
}
mv -f *.gif *.jpg pubimages
for i in [0-9][0-9][0-9][0-9]*.html
# move pages and make symlinks
do fn=$(grep '<h1 align="center">' $i|
  sed 's#^.*<h1 align="center">\(.*\)</h1>.*$#\1#;s/ /_/g;s/\&.*;//g;s/[!?\/–«»¹…]/~/g;s/\.*$//;s/,//g;s/"//g')
	mv $i nn/$i
	ln -s nn/$i ${fn}.html
done
}
rm -f page[0-9]*.html
update freeos 2
update program 2