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