Changeset 94 for nutchez-0.1/bin/nutchez-func.sh
- Timestamp:
- Jul 8, 2009, 2:20:41 PM (16 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
nutchez-0.1/bin/nutchez-func.sh
r91 r94 6 6 7 7 8 : ${DIALOG=dialog} 8 #DIALOG='dialog --backtitle " NutchEz Setup Menu -- powered by NCHC "' 9 DIALOG=dialog 9 10 10 11 # set 1 to display more for debug, … … 39 40 40 41 check_if_root() { 41 if [ ! "$UID" ="0" ]; then42 if [ ! "$UID" -eq "0" ]; then 42 43 echo_dialog_v "Hi [$LOGNAME] !! " 43 44 echo_dialog_v "You need to run this script \"`basename $0`\" as root." … … 71 72 72 73 cp ~/.nutchez/sav/n.urls.txt ~/.nutchez/urls/urls.txt 74 75 if [ $NOCONTINUE -eq 1 ]; then 76 echo_dialog_v " delete the ~/.nutchez/search/*" 77 DATE=`date +%y%m%d%H%M%S` 78 mv ~/.nutchez/search ~/.nutchez/search-$DATE 79 $DIALOG --msgbox "上次搜尋的結果改放到 ~/.nutchez/search-$DATE " 0 0 80 rm -f /tmp/search 81 # rm -rf ~/.nutchez/search/* 82 fi 73 83 74 84 if [ -e ~/.nutchez/conf/nutch-site.xml ] ; then … … 100 110 echo_dialog_v "2.1 test_file ~/.nutchez/sav return : $?" 101 111 # dialog begin 102 dialog --title " The URLS that you want" --editbox ~/.nutchez/sav/n.urls.txt 16 51 2>/tmp/n.urls.tmp112 dialog --title "請輸入你要抓取的網址(一行一個網址)" --editbox ~/.nutchez/sav/n.urls.txt 16 51 2>/tmp/n.urls.tmp 103 113 RET=$? 104 114 echo_dialog_v "2.1 cat url: `cat /tmp/n.urls.tmp`" … … 110 120 echo_dialog_v "3. setup_robot" 111 121 # dialog 112 dialog -- nocancel --inputbox " This agent name \n" 16 51"`cat ~/.nutchez/sav/n.robot.txt`" 2>/tmp/n.robot.tmp122 dialog --title "設定機器人名稱" --nocancel --inputbox " 這個爬網機器人,你要將他取名為:\n\n (ps: 這個設定只是禮貌性宣告,並不會對結果造成影響) \n" 16 55 "`cat ~/.nutchez/sav/n.robot.txt`" 2>/tmp/n.robot.tmp 113 123 echo_dialog_v "3.1 cat robot : `cat /tmp/n.robot.tmp`" 114 124 } … … 117 127 echo_dialog_v "4. setup_crawler" 118 128 test_file ~/.nutchez/sav/n.crawler.txt 119 dialog -- nocancel --inputbox " Depth\n " 16 51 "`cat ~/.nutchez/sav/n.crawler.txt`" 2>/tmp/n.crawler.tmp129 dialog --title "設定抓取深度" --nocancel --inputbox " 對於每個網址,你需要NutchEz爬多深呢?\n\n (ps: 初次體驗建議將深度設為1來感受需要多久) \n " 16 51 "`cat ~/.nutchez/sav/n.crawler.txt`" 2>/tmp/n.crawler.tmp 120 130 echo_dialog_v "4.1 cat robot : `cat /tmp/n.robot.tmp`" 121 131 } … … 124 134 echo_dialog_v "5. setup_tomcat" 125 135 test_file ~/.nutchez/sav/n.tomcat.txt 126 dialog -- nocancel --inputbox " explorer port\n " 16 51 "`cat ~/.nutchez/sav/n.tomcat.txt`" 2>/tmp/n.tomcat.tmp136 dialog --title "設定網頁伺服器" --nocancel --inputbox " 你希望NutchEz將網頁伺服器開在哪個port \n\n (ps: 請選擇一個沒用到的port以免造成衝突 \n 也請盡量不要設成80以免造成你誤以為是apache的混淆) \n " 16 51 "`cat ~/.nutchez/sav/n.tomcat.txt`" 2>/tmp/n.tomcat.tmp 127 137 echo_dialog_v "5.1 cat tomcat : `cat /tmp/n.tomcat.tmp`" 128 138 } 129 139 140 continue_previous () { 141 echo_dialog_v "6. setup_tomcat" 142 $DIALOG --title "清除上次搜尋" --clear \ 143 --yesno "你是否要清除上一次爬網所得的結果,\n否則將加入到URL列裡增加搜尋負擔 \n\n ps: 選no的話,會跑相當相當久,\n 請慎重考慮之\n" 16 51 144 145 case $? in 146 0) 147 NOCONTINUE=1;; 148 1) 149 NOCONTINUE=0;; 150 255) 151 echo "ESC pressed.";; 152 esac 153 echo_dialog_v " 6continue = $CONTINUE" 154 } 155 130 156 final_confirm () { 131 echo_dialog_v " 6. final_confirm : start =0 , back =1 "157 echo_dialog_v "7. final_confirm : start =0 , back =1 " 132 158 tempfile=/tmp/n.finalcheck.tmp 133 159 134 echo " \n 1. The url list is: \n " > $tempfile160 echo " \n 1. 你所選擇要爬取的網址為 : \n " > $tempfile 135 161 cat /tmp/n.urls.tmp >> $tempfile 136 echo " \n 2. The robot name is: \n" >> $tempfile162 echo " \n\n 2. 對於這個爬網機器人,你取名為 : \n" >> $tempfile 137 163 cat /tmp/n.robot.tmp >> $tempfile 138 echo " \n 3. The crawled depth is: \n " >> $tempfile164 echo " \n\n 3. 爬網的深度,你設定為 : \n " >> $tempfile 139 165 cat /tmp/n.crawler.tmp >> $tempfile 140 echo " \n 4. The explorer port is: \n " >> $tempfile166 echo " \n\n 4. NutchEz將會把你的搜尋結果呈現在這個Port : \n " >> $tempfile 141 167 cat /tmp/n.tomcat.tmp >> $tempfile 142 168 if [ $NOCONTINUE -eq 0 ];then 169 echo " \n\n 5. 是否要清除上一次的收尋結果 : \n " >> $tempfile 170 echo_dialog_v " 7continue = $CONTINUE" 171 echo "NO" >> $tempfile 172 elif [ $NOCONTINUE -eq 1 ];then 173 echo " \n\n 5. 是否要清除上一次的收尋結果繼續搜尋 : \n " >> $tempfile 174 echo_dialog_v " 7continue = $CONTINUE" 175 echo "YES" >> $tempfile 176 else 177 echo_dialog_v " 無資料可匯入 " 178 fi 143 179 MSG=`cat $tempfile` 144 echo_dialog_v " 6.1 final message :\n $MSG"180 echo_dialog_v "7.1 final message :\n $MSG" 145 181 #read READ 146 $DIALOG --title " Check It !!" --clear \182 $DIALOG --title "請檢查你的選擇 ! \n\n 若所有的設定都是正確的,你可以按 \"ok\",\n 若你按了 \"reset\" 則會重頭開始設定, \n 若你選擇 \"exit\" 則會跳出NutchEz的設定選單 \n ps: reset 與 exit都不會把資料記成預設值,請放心使用 " --clear \ 147 183 --extra-button --extra-label "reset" --ok-label "ok" --cancel-label "exit" \ 148 184 --yesno "$MSG" 26 51 … … 179 215 180 216 pid_tc=$(ps axw -eo pid,command |\ 181 grep "catalina.startup.Bootstrap" |\182 grep "start" | awk '{print $1}')217 grep "catalina" | grep "java" |\ 218 grep "start" | awk '{print $1}') 183 219 if [ -z "$pid_tc" ]; then 184 echo "starting tomcat"220 echo_dialog_v "no another tomcat is running" 185 221 else 186 echo "tomcat had been started and the pid is $pid_tc"187 echo "stop it first"222 echo_dialog_v "tomcat had been started and the pid is $pid_tc" 223 echo_dialog_v "stop it first" 188 224 kill -9 $pid_tc 189 225 if [ -z $? ];then 190 echo " tomcat ($pid_tc) is killed ..."226 echo_dialog_v " tomcat ($pid_tc) is killed ..." 191 227 else 192 echo "kill error ..."228 echo_dialog_v "kill error ..." 193 229 fi 194 230 fi 231 echo "Starting Tomcat ...." 195 232 ~/.nutchez/tomcat/bin/startup.sh 196 233 sleep 3 … … 201 238 FIREFOX=`which firefox` 202 239 RET=$? 203 if [ $RET ==0 ];then240 if [ $RET -eq 0 ];then 204 241 $FIREFOX -D 0.0 http://localhost:$PORT 205 242 RET=$? 206 243 fi 207 if ! [ $RET ==0 ];then208 $DIALOG --msgbox " Congratulations! \n you can explore the url: \n http://localhost:$PORT" 0 0209 fi 210 } 244 if ! [ $RET -eq 0 ];then 245 $DIALOG --msgbox "恭喜你已經完成了! \n 你可以用瀏覽器瀏覽: \n http://host_ip:$PORT" 0 0 246 fi 247 }
Note: See TracChangeset
for help on using the changeset viewer.