Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- #!/bin/sh
- DIR="/tmp/wiki"
- [ ! -d "$DIR" ] && mkdir "$DIR"
- #|| rm -R "$DIR"
- TMP1=""$DIR"/wiki.tmp"
- TMP2=""$DIR"/wiki.categories_list"
- wget -qO- "http://ru.gentoo-wiki.com/wiki/%D0%A1%D0%BB%D1%83%D0%B6%D0%B5%D0%B1%D0%BD%D0%B0%D1%8F:Categories" > $TMP1
- cat "$TMP1" | grep "<li><a.*/w" | sed s/.*'href="'// | sed s/'<\/a>'.*// | sed s/'\"'.*\>/' \"'/ | sed s/'$'/'\"'/ > "$TMP2"
- cat "$TMP2" | while read name
- do
- a=${name/* \"/\"}
- b=${a//\"/}
- c=${b//\// }
- wget -qO- "http://ru.gentoo-wiki.com${name/ \"*/}" > "$TMP1"
- mkdir -p "wiki/""$c"""
- cat "$TMP1" | grep "<li><a.*/w" | sed s/.*'href="'// | sed s/'<\/a>'.*// | sed s/'\"'.*\>/' \"'/ | sed s/'$'/'\"'/ > "wiki/""$c""/links"
- done
- cd wiki
- find * -maxdepth 0 -type d | while read name
- do
- cat "$name/links" | while read name1
- do
- a=${name1/* \"/\"}
- b=${a//\"/}
- c=${b//\// }
- wget -e robots=off "http://ru.gentoo-wiki.com${name1/ \"*/}" -O """$name"/"$c".html""
- done
- done
Advertisement
Add Comment
Please, Sign In to add comment