[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 |
---|
[72] | 22 | # export NUTCH_CONF_DIR=~/.nutchez |
---|
| 23 | # export HADOOP_CONF_DIR=~/.nutchez |
---|
| 24 | # export HADOOP_LOG_DIR=~/.nutchez/log |
---|
| 25 | # . ~/.nutchez/hadoop-env.sh || . /etc/nutch/hadoop-env.sh |
---|
[69] | 26 | } |
---|
| 27 | |
---|
[66] | 28 | echo_vb () { |
---|
| 29 | if [ $VERB -eq 1 ]; then |
---|
| 30 | $DIALOG --msgbox "$1" 16 51 |
---|
| 31 | fi |
---|
| 32 | } |
---|
| 33 | |
---|
| 34 | test_file () { |
---|
| 35 | if ! test -e "$1" ; then |
---|
| 36 | echo_vb "test_file: \n can not find $1" |
---|
| 37 | echo "" > $1 |
---|
| 38 | else |
---|
| 39 | echo_vb "test_file: \n Touch $1 ! \n Its content is \n `cat $1`" |
---|
| 40 | fi |
---|
| 41 | } |
---|
| 42 | |
---|
| 43 | check_if_root() { |
---|
| 44 | if [ ! "$UID" = "0" ]; then |
---|
| 45 | echo_vb "Hi [$LOGNAME] !! " |
---|
| 46 | echo_vb "You need to run this script \"`basename $0`\" as root." |
---|
| 47 | exit 1 |
---|
| 48 | fi |
---|
| 49 | } |
---|
| 50 | |
---|
| 51 | promote_tempfile () { |
---|
| 52 | echo_vb "7. chang tmp as txt" |
---|
[68] | 53 | rm ~/.nutchez/sav/n.*.txt |
---|
[69] | 54 | mv /tmp/n.urls.tmp ~/.nutchez/sav/n.urls.txt |
---|
| 55 | mv /tmp/n.robot.tmp ~/.nutchez/sav/n.robot.txt |
---|
| 56 | mv /tmp/n.crawler.tmp ~/.nutchez/sav/n.crawler.txt |
---|
| 57 | mv /tmp/n.tomcat.tmp ~/.nutchez/sav/n.tomcat.txt |
---|
[66] | 58 | } |
---|
| 59 | |
---|
| 60 | clean_tempfile () { |
---|
| 61 | echo_vb "7. delete tmp" |
---|
| 62 | rm /tmp/n.*.tmp |
---|
| 63 | } |
---|
| 64 | |
---|
[68] | 65 | setup_nutchez () { |
---|
| 66 | if ! [ -e ~/.nutchez/urls ] ; then |
---|
| 67 | # make url list dir |
---|
| 68 | mkdir ~/.nutchez/urls |
---|
[67] | 69 | fi |
---|
[68] | 70 | |
---|
| 71 | if [ -e ~/.nutchez/urls/urls.txt ] ; then |
---|
| 72 | rm ~/.nutchez/urls/urls.txt |
---|
| 73 | fi |
---|
| 74 | |
---|
| 75 | cp ~/.nutchez/sav/n.urls.txt ~/.nutchez/urls/urls.txt |
---|
| 76 | |
---|
| 77 | if [ -e ~/.nutchez/nutch-site.xml ] ; then |
---|
| 78 | # set nutch-site.xml |
---|
[69] | 79 | sed -i -e "4s/<value>[a-zA-Z0-9]*</<value>$ROBOT</" ~/.nutchez/nutch-site.xml |
---|
[68] | 80 | fi |
---|
| 81 | |
---|
[67] | 82 | } |
---|
[66] | 83 | |
---|
[68] | 84 | install_tomcat (){ |
---|
| 85 | if ! [ -e ~/.nutchez/tomcat ] ;then |
---|
| 86 | # isntall tomcat to home |
---|
| 87 | cp -rf /opt/nutch/tomcat ~/.nutchez/ |
---|
| 88 | chown -R $LOGNAME:$LOGNAME ~/.nutchez/tomcat/ |
---|
| 89 | # make search dir |
---|
| 90 | if ! [ -e ~/.nutchez/search ] ;then |
---|
| 91 | mkdir ~/.nutchez/search |
---|
| 92 | fi |
---|
[69] | 93 | fi |
---|
[77] | 94 | # change explorer port |
---|
| 95 | sed -i -e "67s/<Connector port=\"[0-9]*\"/<Connector port=\"$PORT\"/" ~/.nutchez/tomcat/conf/server.xml |
---|
[68] | 96 | } |
---|
| 97 | |
---|
| 98 | |
---|
[66] | 99 | show_urls (){ |
---|
| 100 | # show urls : ok =0 ,cancel = 1 |
---|
| 101 | echo_vb "2. show_urls !" |
---|
[69] | 102 | test_file ~/.nutchez/sav/n.urls.txt |
---|
[68] | 103 | echo_vb "2.1 test_file ~/.nutchez/sav return : $?" |
---|
[66] | 104 | # dialog begin |
---|
[75] | 105 | dialog --title "The URLS that you want" --editbox ~/.nutchez/sav/n.urls.txt 16 51 2>/tmp/n.urls.tmp |
---|
[66] | 106 | RET=$? |
---|
[69] | 107 | echo_vb "2.1 cat url: `cat /tmp/n.urls.tmp`" |
---|
[66] | 108 | return $RET |
---|
| 109 | } |
---|
| 110 | |
---|
| 111 | setup_robot () { |
---|
[68] | 112 | test_file ~/.nutchez/sav/n.robot.txt |
---|
[66] | 113 | echo_vb "3. setup_robot" |
---|
| 114 | # dialog |
---|
[68] | 115 | dialog --nocancel --inputbox " This agent name \n" 16 51 "`cat ~/.nutchez/sav/n.robot.txt`" 2>/tmp/n.robot.tmp |
---|
[66] | 116 | echo_vb "3.1 cat robot : `cat /tmp/n.robot.tmp`" |
---|
| 117 | } |
---|
| 118 | |
---|
| 119 | setup_crawler () { |
---|
| 120 | echo_vb "4. setup_crawler" |
---|
[68] | 121 | test_file ~/.nutchez/sav/n.crawler.txt |
---|
| 122 | dialog --nocancel --inputbox " Depth \n " 16 51 "`cat ~/.nutchez/sav/n.crawler.txt`" 2>/tmp/n.crawler.tmp |
---|
[66] | 123 | echo_vb "4.1 cat robot : `cat /tmp/n.robot.tmp`" |
---|
| 124 | } |
---|
| 125 | |
---|
| 126 | setup_tomcat () { |
---|
| 127 | echo_vb "5. setup_tomcat" |
---|
[68] | 128 | test_file ~/.nutchez/sav/n.tomcat.txt |
---|
| 129 | dialog --nocancel --inputbox " explorer port \n " 16 51 "`cat ~/.nutchez/sav/n.tomcat.txt`" 2>/tmp/n.tomcat.tmp |
---|
[66] | 130 | echo_vb "5.1 cat tomcat : `cat /tmp/n.tomcat.tmp`" |
---|
| 131 | } |
---|
| 132 | |
---|
| 133 | final_confirm () { |
---|
| 134 | echo_vb "6. final_confirm : start =0 , back =1 " |
---|
[69] | 135 | tempfile=/tmp/n.finalcheck.tmp |
---|
[66] | 136 | |
---|
| 137 | echo " \n 1. The url list is : \n " > $tempfile |
---|
[69] | 138 | cat /tmp/n.urls.tmp >> $tempfile |
---|
[66] | 139 | echo " \n 2. The robot name is : \n" >> $tempfile |
---|
| 140 | cat /tmp/n.robot.tmp >> $tempfile |
---|
[67] | 141 | echo " \n 3. The crawled depth is : \n " >> $tempfile |
---|
[66] | 142 | cat /tmp/n.crawler.tmp >> $tempfile |
---|
| 143 | echo " \n 4. The explorer port is : \n " >> $tempfile |
---|
| 144 | cat /tmp/n.tomcat.tmp >> $tempfile |
---|
| 145 | |
---|
| 146 | MSG=`cat $tempfile` |
---|
| 147 | echo_vb "6.1 final message :\n $MSG" |
---|
| 148 | #read READ |
---|
| 149 | $DIALOG --title "Check It !!" --clear \ |
---|
[75] | 150 | --extra-button --extra-label "reset" --ok-label "ok" --cancel-label "exit" \ |
---|
[69] | 151 | --yesno "$MSG" 26 51 |
---|
[66] | 152 | RET=$? |
---|
| 153 | echo_vb "final return = $RET" |
---|
| 154 | return $RET |
---|
| 155 | } |
---|
| 156 | |
---|
[67] | 157 | # define paramaters |
---|
| 158 | |
---|
[77] | 159 | set_nutchez_p () { |
---|
[69] | 160 | ROBOT=`cat ~/.nutchez/sav/n.robot.txt` |
---|
| 161 | URLS=`cat ~/.nutchez/sav/n.urls.txt` |
---|
| 162 | DEPTH=`cat ~/.nutchez/sav/n.crawler.txt` |
---|
| 163 | PORT=`cat ~/.nutchez/sav/n.tomcat.txt` |
---|
[77] | 164 | } |
---|
| 165 | start_crawl () { |
---|
[68] | 166 | |
---|
[66] | 167 | echo_vb "7. start_crawl" |
---|
[69] | 168 | setup_nutchez |
---|
[68] | 169 | install_tomcat |
---|
[69] | 170 | echo_vb "/opt/nutch/bin/nutch crawl ~/.nutchez/urls -dir ~/.nutchez/search -depth $DEPTH" |
---|
[75] | 171 | echo_vb "nutch conf dir = $NUTCH_CONF_DIR" |
---|
[80] | 172 | /opt/nutch/bin/nutch crawl ~/.nutchez/urls -dir ~/.nutchez/search -depth $DEPTH -topN 5000 -threads 1000 |
---|
[66] | 173 | } |
---|
| 174 | |
---|
| 175 | start_tomcat () { |
---|
| 176 | echo_vb "8. start_tomcat " |
---|
[69] | 177 | echo_vb "/opt/nutch/tomcat/bin/startup.sh" |
---|
[72] | 178 | if [ -e /tmp/search ];then |
---|
[69] | 179 | rm -rf /tmp/search |
---|
| 180 | fi |
---|
| 181 | ln -sf ~/.nutchez/search/ /tmp/ |
---|
[88] | 182 | |
---|
| 183 | pid_tc=$(ps axw -eo pid,command | \ |
---|
| 184 | grep "org.apache.catalina.startup.Bootstrap"\ |
---|
| 185 | | grep "start" | awk '{print $1}') |
---|
| 186 | if [ -z "$pid_tc" ]; then |
---|
| 187 | echo "starting tomcat" |
---|
| 188 | else |
---|
| 189 | echo "tomcat had been started and the pid is $pid_tc" |
---|
| 190 | echo "stop it first" |
---|
| 191 | kill -9 $pid_tc |
---|
| 192 | if [ -z $? ];then |
---|
| 193 | echo " tomcat ($pid_tc) is killed ..." |
---|
| 194 | else |
---|
| 195 | echo "kill error ..." |
---|
| 196 | fi |
---|
| 197 | fi |
---|
[69] | 198 | ~/.nutchez/tomcat/bin/startup.sh |
---|
[77] | 199 | sleep 3 |
---|
[66] | 200 | } |
---|
| 201 | |
---|
| 202 | show_report () { |
---|
| 203 | echo_vb "9. show_report " |
---|
[68] | 204 | FIREFOX=`which firefox` |
---|
| 205 | RET=$? |
---|
[69] | 206 | if [ $RET == 0 ];then |
---|
[68] | 207 | $FIREFOX -D 0.0 http://localhost:$PORT |
---|
[69] | 208 | RET=$? |
---|
[68] | 209 | fi |
---|
[69] | 210 | if ! [ $RET == 0 ];then |
---|
| 211 | $DIALOG --msgbox "Congratulations! \n you can explore the url: \n http://localhost:$PORT" 0 0 |
---|
| 212 | fi |
---|
[66] | 213 | } |
---|