cronman-en

blog generator是一个基于Catty 2引擎的工具

#!/bin/sh

cd cron-en
touch ../data/learned

for i in *; do

  LEARNED=`grep -F "[$i]" ../data/learned`

  if [ ! "$LEARNED" = "" ]; then
    echo "[*] Already searched for $i"
    rm -f "$i"
    continue
  fi

  echo "[*] Trying to learn about $i..."
  echo "[$i]" >>../data/learned
  rm -f "$i"

  (sleep 40;killall -9 lynx ) &>/dev/null &
  
  # filter=0
  lynx -connect_timeout=30 -dump "http://www.google.com/search?q=%22$i%22+site:diaryland.com&num=15" \
  | grep -F '. http://' | grep -vF 'google.com' | grep -vF 'cache:' | grep -vFi '.doc' | grep -vFi '.pdf' | awk -F '\\. ' '{print $2}' | \
  awk -F '^^FIKU^^' '{print "./getwww \"" $0 "\" >>./data/blog-en"}' >.cronman

  killall -9 sleep &>/dev/null

  cd ..
  . cron-en/.cronman
  cd cron-en

  rm -f .cronman

done

cd ..
cat data/blog-en | sort | uniq >data/.tmpk
cat data/.tmpk >data/blog-en
rm -f data/.tmpk

echo "[+] Done. You have `grep -c . data/blog-en` phrases."

# restart catty