Guest User

Untitled

a guest
Nov 10th, 2011
234
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
Bash 0.99 KB | None | 0 0
  1. #!/bin/sh
  2. DIR="/tmp/wiki"
  3. [ ! -d "$DIR" ] && mkdir "$DIR"
  4. #|| rm -R "$DIR"
  5.  
  6. TMP1=""$DIR"/wiki.tmp"
  7. TMP2=""$DIR"/wiki.categories_list"
  8.  
  9. 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" > $TMP1
  10.  
  11. cat "$TMP1" | grep "<li><a.*/w" | sed s/.*'href="'// | sed s/'<\/a>'.*// | sed s/'\"'.*\>/' \"'/ | sed s/'$'/'\"'/ > "$TMP2"
  12.  
  13. cat "$TMP2" | while read name
  14. do
  15.    a=${name/* \"/\"}
  16.    b=${a//\"/}
  17.    c=${b//\// }
  18.    wget -qO- "http://ru.gentoo-wiki.com${name/ \"*/}" > "$TMP1"
  19.    mkdir -p "wiki/""$c"""
  20.    cat "$TMP1" | grep "<li><a.*/w" | sed s/.*'href="'// | sed s/'<\/a>'.*// | sed s/'\"'.*\>/' \"'/ | sed s/'$'/'\"'/ > "wiki/""$c""/links"
  21. done
  22.  
  23. cd wiki
  24. find * -maxdepth 0 -type d  | while read name
  25. do  
  26.    cat "$name/links" | while read name1
  27.    do
  28.       a=${name1/* \"/\"}
  29.       b=${a//\"/}
  30.       c=${b//\// }
  31.       wget -e robots=off "http://ru.gentoo-wiki.com${name1/ \"*/}" -O """$name"/"$c".html""
  32.    done
  33. done
  34.  
  35.  
  36.  
Advertisement
Add Comment
Please, Sign In to add comment