[66] | 1 | #!/bin/bash |
---|
| 2 | # Author: WeiYu Chen <waue _at_ nchc org tw> |
---|
| 3 | # License: GPL |
---|
| 4 | # Description: Eazily use for Nutch |
---|
| 5 | # . |
---|
| 6 | |
---|
[67] | 7 | . /etc/nutch/hadoop-env.sh || . $NUTCHEZ_SCRIPT_PATH/conf/hadoop-env.sh |
---|
[66] | 8 | |
---|
| 9 | : ${DIALOG=dialog} |
---|
| 10 | |
---|
[68] | 11 | # display more for debug |
---|
[67] | 12 | VERB=0 |
---|
[66] | 13 | |
---|
| 14 | echo_vb () { |
---|
| 15 | if [ $VERB -eq 1 ]; then |
---|
| 16 | $DIALOG --msgbox "$1" 16 51 |
---|
| 17 | fi |
---|
| 18 | } |
---|
| 19 | |
---|
| 20 | test_file () { |
---|
| 21 | if ! test -e "$1" ; then |
---|
| 22 | echo_vb "test_file: \n can not find $1" |
---|
| 23 | echo "" > $1 |
---|
| 24 | else |
---|
| 25 | echo_vb "test_file: \n Touch $1 ! \n Its content is \n `cat $1`" |
---|
| 26 | fi |
---|
| 27 | } |
---|
| 28 | |
---|
| 29 | check_if_root() { |
---|
| 30 | if [ ! "$UID" = "0" ]; then |
---|
| 31 | echo_vb "Hi [$LOGNAME] !! " |
---|
| 32 | echo_vb "You need to run this script \"`basename $0`\" as root." |
---|
| 33 | exit 1 |
---|
| 34 | fi |
---|
| 35 | } |
---|
| 36 | |
---|
| 37 | promote_tempfile () { |
---|
| 38 | echo_vb "7. chang tmp as txt" |
---|
[68] | 39 | rm ~/.nutchez/sav/n.*.txt |
---|
| 40 | mv /tmp/n.url.tmp ~/.nutchez/sav/ |
---|
| 41 | mv /tmp/n.robot.tmp ~/.nutchez/sav/ |
---|
| 42 | mv /tmp/n.crawler.tmp ~/.nutchez/sav/ |
---|
| 43 | mv /tmp/n.tomcat.tmp ~/.nutchez/sav/ |
---|
[66] | 44 | } |
---|
| 45 | |
---|
| 46 | clean_tempfile () { |
---|
| 47 | echo_vb "7. delete tmp" |
---|
| 48 | rm /tmp/n.*.tmp |
---|
| 49 | } |
---|
| 50 | |
---|
[68] | 51 | init_nutchez () { |
---|
[67] | 52 | if ! [ -e ~/.nutchez ] ;then |
---|
[68] | 53 | # copy from /etc/nutch |
---|
| 54 | cp -rf /etc/nutch/* ~/.nutchez |
---|
[67] | 55 | chown -R $LOGNAME:$LOGNAME ~/.nutchez |
---|
[68] | 56 | fi |
---|
| 57 | } |
---|
[66] | 58 | |
---|
[68] | 59 | setup_nutchez () { |
---|
| 60 | if ! [ -e ~/.nutchez/urls ] ; then |
---|
| 61 | # make url list dir |
---|
| 62 | mkdir ~/.nutchez/urls |
---|
[67] | 63 | fi |
---|
[68] | 64 | |
---|
| 65 | if [ -e ~/.nutchez/urls/urls.txt ] ; then |
---|
| 66 | rm ~/.nutchez/urls/urls.txt |
---|
| 67 | fi |
---|
| 68 | |
---|
| 69 | cp ~/.nutchez/sav/n.urls.txt ~/.nutchez/urls/urls.txt |
---|
| 70 | |
---|
| 71 | if [ -e ~/.nutchez/nutch-site.xml ] ; then |
---|
| 72 | # set nutch-site.xml |
---|
| 73 | sed -ie "s/>user</>$ROBOT</" ~/.nutchez/nutch-site.xml |
---|
| 74 | fi |
---|
| 75 | |
---|
[67] | 76 | } |
---|
[66] | 77 | |
---|
[68] | 78 | install_tomcat (){ |
---|
| 79 | if ! [ -e ~/.nutchez/tomcat ] ;then |
---|
| 80 | # isntall tomcat to home |
---|
| 81 | cp -rf /opt/nutch/tomcat ~/.nutchez/ |
---|
| 82 | chown -R $LOGNAME:$LOGNAME ~/.nutchez/tomcat/ |
---|
| 83 | # make search dir |
---|
| 84 | if ! [ -e ~/.nutchez/search ] ;then |
---|
| 85 | mkdir ~/.nutchez/search |
---|
| 86 | fi |
---|
| 87 | # change explorer port |
---|
| 88 | sed -ie "s/8080/>$PORT</" ~/.nutchez/tomcat/conf/server.xml |
---|
| 89 | } |
---|
| 90 | |
---|
| 91 | |
---|
[66] | 92 | show_urls (){ |
---|
| 93 | # show urls : ok =0 ,cancel = 1 |
---|
| 94 | echo_vb "2. show_urls !" |
---|
[68] | 95 | test_file ~/.nutchez/sav/n.url.txt |
---|
| 96 | echo_vb "2.1 test_file ~/.nutchez/sav return : $?" |
---|
[66] | 97 | # dialog begin |
---|
[68] | 98 | dialog --editbox ~/.nutchez/sav/n.url.txt 16 51 2>/tmp/n.url.tmp |
---|
[66] | 99 | RET=$? |
---|
| 100 | echo_vb "2.1 cat url: `cat /tmp/n.url.tmp`" |
---|
| 101 | return $RET |
---|
| 102 | } |
---|
| 103 | |
---|
| 104 | setup_robot () { |
---|
[68] | 105 | test_file ~/.nutchez/sav/n.robot.txt |
---|
[66] | 106 | echo_vb "3. setup_robot" |
---|
| 107 | # dialog |
---|
[68] | 108 | dialog --nocancel --inputbox " This agent name \n" 16 51 "`cat ~/.nutchez/sav/n.robot.txt`" 2>/tmp/n.robot.tmp |
---|
[66] | 109 | echo_vb "3.1 cat robot : `cat /tmp/n.robot.tmp`" |
---|
| 110 | } |
---|
| 111 | |
---|
| 112 | setup_crawler () { |
---|
| 113 | echo_vb "4. setup_crawler" |
---|
[68] | 114 | test_file ~/.nutchez/sav/n.crawler.txt |
---|
| 115 | dialog --nocancel --inputbox " Depth \n " 16 51 "`cat ~/.nutchez/sav/n.crawler.txt`" 2>/tmp/n.crawler.tmp |
---|
[66] | 116 | echo_vb "4.1 cat robot : `cat /tmp/n.robot.tmp`" |
---|
| 117 | } |
---|
| 118 | |
---|
| 119 | setup_tomcat () { |
---|
| 120 | echo_vb "5. setup_tomcat" |
---|
[68] | 121 | test_file ~/.nutchez/sav/n.tomcat.txt |
---|
| 122 | dialog --nocancel --inputbox " explorer port \n " 16 51 "`cat ~/.nutchez/sav/n.tomcat.txt`" 2>/tmp/n.tomcat.tmp |
---|
[66] | 123 | echo_vb "5.1 cat tomcat : `cat /tmp/n.tomcat.tmp`" |
---|
| 124 | } |
---|
| 125 | |
---|
| 126 | final_confirm () { |
---|
| 127 | echo_vb "6. final_confirm : start =0 , back =1 " |
---|
| 128 | tempfile=`tempfile 2>/dev/null` || tempfile=/tmp/n.finalcheck.tmp |
---|
| 129 | #trap "rm -f $tempfile" 0 1 2 5 15 |
---|
| 130 | |
---|
| 131 | echo " \n 1. The url list is : \n " > $tempfile |
---|
| 132 | cat /tmp/n.url.tmp >> $tempfile |
---|
| 133 | echo " \n 2. The robot name is : \n" >> $tempfile |
---|
| 134 | cat /tmp/n.robot.tmp >> $tempfile |
---|
[67] | 135 | echo " \n 3. The crawled depth is : \n " >> $tempfile |
---|
[66] | 136 | cat /tmp/n.crawler.tmp >> $tempfile |
---|
| 137 | echo " \n 4. The explorer port is : \n " >> $tempfile |
---|
| 138 | cat /tmp/n.tomcat.tmp >> $tempfile |
---|
| 139 | |
---|
| 140 | MSG=`cat $tempfile` |
---|
| 141 | echo_vb "6.1 final message :\n $MSG" |
---|
| 142 | #read READ |
---|
| 143 | $DIALOG --title "Check It !!" --clear \ |
---|
| 144 | --yesno "$MSG" 16 51 |
---|
| 145 | RET=$? |
---|
| 146 | echo_vb "final return = $RET" |
---|
| 147 | return $RET |
---|
| 148 | } |
---|
| 149 | |
---|
[67] | 150 | # define paramaters |
---|
| 151 | |
---|
[66] | 152 | start_crawl () { |
---|
[68] | 153 | |
---|
| 154 | ROBOT=`cat ~/.nutchez/sav` |
---|
| 155 | URLS=`cat ~/.nutchez/sav` |
---|
| 156 | DEPTH=`cat ~/.nutchez/sav` |
---|
| 157 | PORT=`cat ~/.nutchez/sav` |
---|
| 158 | |
---|
[66] | 159 | echo_vb "7. start_crawl" |
---|
[67] | 160 | setup_nutch |
---|
[68] | 161 | install_tomcat |
---|
| 162 | # /opt/nutch/bin/nutch crawl ~/.nutchez/urls -dir ~/.nutchez/search -depth $DEPTH |
---|
| 163 | echo "/opt/nutch/bin/nutch crawl ~/.nutchez/urls -dir ~/.nutchez/search -depth $DEPTH" |
---|
[66] | 164 | } |
---|
| 165 | |
---|
| 166 | start_tomcat () { |
---|
| 167 | echo_vb "8. start_tomcat " |
---|
[67] | 168 | /opt/nutch/tomcat/bin/startup.sh |
---|
[66] | 169 | } |
---|
| 170 | |
---|
| 171 | show_report () { |
---|
| 172 | echo_vb "9. show_report " |
---|
[68] | 173 | FIREFOX=`which firefox` |
---|
| 174 | RET=$? |
---|
| 175 | if [ RET == 0 ];then |
---|
| 176 | $FIREFOX -D 0.0 http://localhost:$PORT |
---|
| 177 | else |
---|
| 178 | $DIALOG --msgbox "Congratulations! \n you can explore the url: \n http://localhost:8080" 0 0 |
---|
| 179 | fi |
---|
[66] | 180 | } |
---|