• nutchez

#!/bin/bash
# Author: WeiYu Chen <waue _at_ nchc org tw>
# License: GPL
# Description: Eazily use for Nutch
# .

NUTCHEZ_SCRIPT_PATH="${NUTCHEZ_SCRIPT_PATH:-/opt/nutch}"

. ./nutchez-func.sh
# root ?
#check_if_root

# show url lists
CHECK=0
while [ $CHECK -eq 0 ]; do

  LOCK=1
  while [ $LOCK -eq 1 ]; do
    # milestone M1
    show_urls
    URL=$?
    echo_vb "$URL"
    
# add or delete url:  ok , exit
# ok =0 , exit =1 

    if [ $URL -eq 0 ];then
      # go_to_M2
      LOCK=0
    else
      #return_console
      exit
    fi

  done
  # milestone M2

  # setup search engine personality
  #   next, back
  setup_robot

  # setup the crawl paramater
  #   next, back
  setup_crawler

  # setup the tomcat paramater
  #   next, back
  setup_tomcat

  # show the final checklist
  final_confirm
  FC=$?
  echo_vb $FC 
  #   START , back, cancel
  # start =0 , back =1 
  if [ $FC -eq 0 ];then
    #go_to_M3
    CHECK=1
  else
    #go_to_M1
    LOCK=1
    CHECK=0
    #return_console
  fi
done


start_crawl
start_tomcat

# show result message

show_report

# Done 

• nutchez-func.sh

  #!/bin/bash
  # Author: WeiYu Chen <waue _at_ nchc org tw>
  # License: GPL
  # Description: Eazily use for Nutch
  # .
  
  . $NUTCHEZ_SCRIPT_PATH/conf/hadoop-env.sh
  
  VERB=1
  #DIALOG=dialog
  
  #NEZ_DIR=/home/waue/.nutchez
  #NEZ_DIR_URL
  #NEZ_DIR_OTHER
  
  
  echo_vb () {
    if [ $VERB -eq 1 ]; then
      echo $1
    fi
  }
  
  test_file () {
    if ! test -e $1 ; then
      echo "" >  $1
    fi
  }
  
  check_if_root() {
     if [ ! "$UID" = "0" ]; then
       echo_vb "[$LOGNAME] You need to run this script \"`basename $0`\" as root."
       exit 1
     fi
  }
  
  show_urls (){
    echo_vb "show urls : ok =0 ,cancel = 1"
  
    test_file  ~/n.url.txt
  
    # dialog begin
    dialog  --editbox ~/n.url.txt 30 50 2> ~/n.url.tmp
    return $?
  }
  
  setup_robot () {
    test_file ~/n.robot.txt
    echo_vb "setup_robot"
    # dialog
    dialog --inputbox "this agent name \n ex: nutch" 0 0 2> ~/n.robot.tmp
  }
  
  setup_crawler () {
    echo_vb "setup_crawler"
    test_file ~/n.crawler.txt
    dialog --inputbox "Depth  \n ex: 5" 0 0 2> ~/n.crawler.tmp
  }
  
  setup_tomcat () {
    echo_vb "setup_tomcat"
    test_file ~/n.tomcat.txt
    dialog --inputbox "explorer port \n ex:8080 " 0 0 2> ~/n.tomcat.tmp
  }
  
  final_confirm () {
    echo_vb "final_confirm : start =0 , back =1 "
    MSG="urls = \n `cat ~/n.url.txt` \n robot name = \n `~/n.robot.tmp` \n depth = \n `~/n.crawler.tmp` \n explorer port = \n `cat ~/n.tomcat.tmp` \n"
    #read READ
    dialog --msgbox "$MSG" 0 0
    return $READ
  }
  
  start_crawl () {
    echo_vb "start_crawl"
  }
  
  start_tomcat () {
    echo_vb "start_tomcat "
  }
  
  show_report () {
    echo_vb "show_report "
  }