[66] | 1 | #!/bin/bash |
---|
| 2 | # Author: WeiYu Chen <waue _at_ nchc org tw> |
---|
| 3 | # License: GPL |
---|
| 4 | # Description: Eazily use for Nutch |
---|
| 5 | # . |
---|
| 6 | |
---|
| 7 | |
---|
| 8 | : ${DIALOG=dialog} |
---|
| 9 | |
---|
[69] | 10 | # set 1 to display more for debug, |
---|
[72] | 11 | VERB=0 |
---|
[66] | 12 | |
---|
[69] | 13 | init_nutchez () { |
---|
| 14 | if ! [ -e ~/.nutchez ] ;then |
---|
| 15 | # copy from /etc/nutch |
---|
| 16 | mkdir ~/.nutchez |
---|
[89] | 17 | cp -rf /etc/nutch/conf ~/.nutchez/ |
---|
[69] | 18 | mkdir ~/.nutchez/log |
---|
[89] | 19 | cp -rf /etc/nutch/sav ~/.nutchez/ |
---|
[69] | 20 | chown -R $LOGNAME:$LOGNAME ~/.nutchez |
---|
| 21 | fi |
---|
[91] | 22 | |
---|
[69] | 23 | } |
---|
| 24 | |
---|
[91] | 25 | echo_dialog_v () { |
---|
[66] | 26 | if [ $VERB -eq 1 ]; then |
---|
| 27 | $DIALOG --msgbox "$1" 16 51 |
---|
| 28 | fi |
---|
| 29 | } |
---|
| 30 | |
---|
| 31 | test_file () { |
---|
| 32 | if ! test -e "$1" ; then |
---|
[91] | 33 | echo_dialog_v "test_file: \n can not find $1" |
---|
[66] | 34 | echo "" > $1 |
---|
| 35 | else |
---|
[91] | 36 | echo_dialog_v "test_file: \n Touch $1 ! \n Its content is \n `cat $1`" |
---|
[66] | 37 | fi |
---|
| 38 | } |
---|
| 39 | |
---|
| 40 | check_if_root() { |
---|
| 41 | if [ ! "$UID" = "0" ]; then |
---|
[91] | 42 | echo_dialog_v "Hi [$LOGNAME] !! " |
---|
| 43 | echo_dialog_v "You need to run this script \"`basename $0`\" as root." |
---|
[66] | 44 | exit 1 |
---|
| 45 | fi |
---|
| 46 | } |
---|
| 47 | |
---|
| 48 | promote_tempfile () { |
---|
[91] | 49 | echo_dialog_v "7. chang tmp as txt" |
---|
[68] | 50 | rm ~/.nutchez/sav/n.*.txt |
---|
[69] | 51 | mv /tmp/n.urls.tmp ~/.nutchez/sav/n.urls.txt |
---|
| 52 | mv /tmp/n.robot.tmp ~/.nutchez/sav/n.robot.txt |
---|
| 53 | mv /tmp/n.crawler.tmp ~/.nutchez/sav/n.crawler.txt |
---|
| 54 | mv /tmp/n.tomcat.tmp ~/.nutchez/sav/n.tomcat.txt |
---|
[66] | 55 | } |
---|
| 56 | |
---|
| 57 | clean_tempfile () { |
---|
[91] | 58 | echo_dialog_v "7. delete tmp" |
---|
[66] | 59 | rm /tmp/n.*.tmp |
---|
| 60 | } |
---|
| 61 | |
---|
[68] | 62 | setup_nutchez () { |
---|
| 63 | if ! [ -e ~/.nutchez/urls ] ; then |
---|
| 64 | # make url list dir |
---|
| 65 | mkdir ~/.nutchez/urls |
---|
[67] | 66 | fi |
---|
[68] | 67 | |
---|
| 68 | if [ -e ~/.nutchez/urls/urls.txt ] ; then |
---|
| 69 | rm ~/.nutchez/urls/urls.txt |
---|
| 70 | fi |
---|
| 71 | |
---|
| 72 | cp ~/.nutchez/sav/n.urls.txt ~/.nutchez/urls/urls.txt |
---|
| 73 | |
---|
[91] | 74 | if [ -e ~/.nutchez/conf/nutch-site.xml ] ; then |
---|
[68] | 75 | # set nutch-site.xml |
---|
[91] | 76 | sed -i -e "4s/<value>[a-zA-Z0-9]*</<value>$ROBOT</" ~/.nutchez/conf/nutch-site.xml |
---|
[68] | 77 | fi |
---|
| 78 | |
---|
[67] | 79 | } |
---|
[66] | 80 | |
---|
[68] | 81 | install_tomcat (){ |
---|
| 82 | if ! [ -e ~/.nutchez/tomcat ] ;then |
---|
| 83 | # isntall tomcat to home |
---|
| 84 | cp -rf /opt/nutch/tomcat ~/.nutchez/ |
---|
| 85 | chown -R $LOGNAME:$LOGNAME ~/.nutchez/tomcat/ |
---|
| 86 | # make search dir |
---|
| 87 | if ! [ -e ~/.nutchez/search ] ;then |
---|
| 88 | mkdir ~/.nutchez/search |
---|
| 89 | fi |
---|
[69] | 90 | fi |
---|
[77] | 91 | # change explorer port |
---|
| 92 | sed -i -e "67s/<Connector port=\"[0-9]*\"/<Connector port=\"$PORT\"/" ~/.nutchez/tomcat/conf/server.xml |
---|
[68] | 93 | } |
---|
| 94 | |
---|
| 95 | |
---|
[66] | 96 | show_urls (){ |
---|
| 97 | # show urls : ok =0 ,cancel = 1 |
---|
[91] | 98 | echo_dialog_v "2. show_urls !" |
---|
[69] | 99 | test_file ~/.nutchez/sav/n.urls.txt |
---|
[91] | 100 | echo_dialog_v "2.1 test_file ~/.nutchez/sav return : $?" |
---|
[66] | 101 | # dialog begin |
---|
[75] | 102 | dialog --title "The URLS that you want" --editbox ~/.nutchez/sav/n.urls.txt 16 51 2>/tmp/n.urls.tmp |
---|
[66] | 103 | RET=$? |
---|
[91] | 104 | echo_dialog_v "2.1 cat url: `cat /tmp/n.urls.tmp`" |
---|
[66] | 105 | return $RET |
---|
| 106 | } |
---|
| 107 | |
---|
| 108 | setup_robot () { |
---|
[68] | 109 | test_file ~/.nutchez/sav/n.robot.txt |
---|
[91] | 110 | echo_dialog_v "3. setup_robot" |
---|
[66] | 111 | # dialog |
---|
[68] | 112 | dialog --nocancel --inputbox " This agent name \n" 16 51 "`cat ~/.nutchez/sav/n.robot.txt`" 2>/tmp/n.robot.tmp |
---|
[91] | 113 | echo_dialog_v "3.1 cat robot : `cat /tmp/n.robot.tmp`" |
---|
[66] | 114 | } |
---|
| 115 | |
---|
| 116 | setup_crawler () { |
---|
[91] | 117 | echo_dialog_v "4. setup_crawler" |
---|
[68] | 118 | 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.tmp |
---|
[91] | 120 | echo_dialog_v "4.1 cat robot : `cat /tmp/n.robot.tmp`" |
---|
[66] | 121 | } |
---|
| 122 | |
---|
| 123 | setup_tomcat () { |
---|
[91] | 124 | echo_dialog_v "5. setup_tomcat" |
---|
[68] | 125 | 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.tmp |
---|
[91] | 127 | echo_dialog_v "5.1 cat tomcat : `cat /tmp/n.tomcat.tmp`" |
---|
[66] | 128 | } |
---|
| 129 | |
---|
| 130 | final_confirm () { |
---|
[91] | 131 | echo_dialog_v "6. final_confirm : start =0 , back =1 " |
---|
[69] | 132 | tempfile=/tmp/n.finalcheck.tmp |
---|
[66] | 133 | |
---|
| 134 | echo " \n 1. The url list is : \n " > $tempfile |
---|
[69] | 135 | cat /tmp/n.urls.tmp >> $tempfile |
---|
[66] | 136 | echo " \n 2. The robot name is : \n" >> $tempfile |
---|
| 137 | cat /tmp/n.robot.tmp >> $tempfile |
---|
[67] | 138 | echo " \n 3. The crawled depth is : \n " >> $tempfile |
---|
[66] | 139 | cat /tmp/n.crawler.tmp >> $tempfile |
---|
| 140 | echo " \n 4. The explorer port is : \n " >> $tempfile |
---|
| 141 | cat /tmp/n.tomcat.tmp >> $tempfile |
---|
| 142 | |
---|
| 143 | MSG=`cat $tempfile` |
---|
[91] | 144 | echo_dialog_v "6.1 final message :\n $MSG" |
---|
[66] | 145 | #read READ |
---|
| 146 | $DIALOG --title "Check It !!" --clear \ |
---|
[75] | 147 | --extra-button --extra-label "reset" --ok-label "ok" --cancel-label "exit" \ |
---|
[69] | 148 | --yesno "$MSG" 26 51 |
---|
[66] | 149 | RET=$? |
---|
[91] | 150 | echo_dialog_v "final return = $RET" |
---|
[66] | 151 | return $RET |
---|
| 152 | } |
---|
| 153 | |
---|
[67] | 154 | # define paramaters |
---|
| 155 | |
---|
[77] | 156 | set_nutchez_p () { |
---|
[69] | 157 | ROBOT=`cat ~/.nutchez/sav/n.robot.txt` |
---|
| 158 | URLS=`cat ~/.nutchez/sav/n.urls.txt` |
---|
| 159 | DEPTH=`cat ~/.nutchez/sav/n.crawler.txt` |
---|
| 160 | PORT=`cat ~/.nutchez/sav/n.tomcat.txt` |
---|
[77] | 161 | } |
---|
| 162 | start_crawl () { |
---|
[68] | 163 | |
---|
[91] | 164 | echo_dialog_v "7. start_crawl" |
---|
[69] | 165 | setup_nutchez |
---|
[68] | 166 | install_tomcat |
---|
[91] | 167 | echo_dialog_v "/opt/nutch/bin/nutch crawl ~/.nutchez/urls -dir ~/.nutchez/search -depth $DEPTH" |
---|
| 168 | echo_dialog_v "nutch conf dir = $NUTCH_CONF_DIR" |
---|
[80] | 169 | /opt/nutch/bin/nutch crawl ~/.nutchez/urls -dir ~/.nutchez/search -depth $DEPTH -topN 5000 -threads 1000 |
---|
[66] | 170 | } |
---|
| 171 | |
---|
| 172 | start_tomcat () { |
---|
[91] | 173 | echo_dialog_v "8. start_tomcat " |
---|
| 174 | echo_dialog_v "/opt/nutch/tomcat/bin/startup.sh" |
---|
[72] | 175 | if [ -e /tmp/search ];then |
---|
[69] | 176 | rm -rf /tmp/search |
---|
| 177 | fi |
---|
| 178 | ln -sf ~/.nutchez/search/ /tmp/ |
---|
[88] | 179 | |
---|
[91] | 180 | pid_tc=$(ps axw -eo pid,command |\ |
---|
| 181 | grep "catalina.startup.Bootstrap" |\ |
---|
| 182 | grep "start" | awk '{print $1}') |
---|
[88] | 183 | if [ -z "$pid_tc" ]; then |
---|
| 184 | echo "starting tomcat" |
---|
| 185 | else |
---|
| 186 | echo "tomcat had been started and the pid is $pid_tc" |
---|
| 187 | echo "stop it first" |
---|
| 188 | kill -9 $pid_tc |
---|
| 189 | if [ -z $? ];then |
---|
| 190 | echo " tomcat ($pid_tc) is killed ..." |
---|
| 191 | else |
---|
| 192 | echo "kill error ..." |
---|
| 193 | fi |
---|
| 194 | fi |
---|
[69] | 195 | ~/.nutchez/tomcat/bin/startup.sh |
---|
[77] | 196 | sleep 3 |
---|
[66] | 197 | } |
---|
| 198 | |
---|
| 199 | show_report () { |
---|
[91] | 200 | echo_dialog_v "9. show_report " |
---|
[68] | 201 | FIREFOX=`which firefox` |
---|
| 202 | RET=$? |
---|
[69] | 203 | if [ $RET == 0 ];then |
---|
[68] | 204 | $FIREFOX -D 0.0 http://localhost:$PORT |
---|
[69] | 205 | RET=$? |
---|
[68] | 206 | fi |
---|
[69] | 207 | if ! [ $RET == 0 ];then |
---|
| 208 | $DIALOG --msgbox "Congratulations! \n you can explore the url: \n http://localhost:$PORT" 0 0 |
---|
| 209 | fi |
---|
[66] | 210 | } |
---|