Changeset 67 for nutchez-0.1/bin/nutchez-func.sh

Timestamp:

May 21, 2009, 2:49:49 PM (15 years ago)

Author:

waue

Message:

v2

File:

• nutchez-0.1/bin/nutchez-func.sh (modified) (4 diffs)

nutchez-0.1/bin/nutchez-func.sh

@@ -5 +5 @@
 # .
 
-. $NUTCHEZ_SCRIPT_PATH/conf/hadoop-env.sh
-
-VERB=1
+. /etc/nutch/hadoop-env.sh || . $NUTCHEZ_SCRIPT_PATH/conf/hadoop-env.sh
 
 : ${DIALOG=dialog}
 
-#NEZ_DIR=/home/waue/.nutchez
-#NEZ_DIR_URL
-#NEZ_DIR_OTHER
+VERB=0
 
 
@@ -53 +49 @@
 }
 
+setup_nutchez() {
+  if ! [ -e ~/.nutchez ] ;then
+    cp -rf /etc/nutch/.nutchez ~/
+    chown -R $LOGNAME:$LOGNAME ~/.nutchez
+    if [ -e /tmp/nutch ] ;then
+      rm -rf /tmp/nutch
+    fi
+    mkdir /tmp/nutch
+    ln -sf ~/.nutchez/search /tmp/nutch/
 
+  fi
+}
 
 show_urls (){
@@ -98 +105 @@
   echo " \n 2. The robot name is : \n" >> $tempfile
   cat /tmp/n.robot.tmp >> $tempfile
-  echo " \n 3. The crawled dipth is : \n " >> $tempfile
+  echo " \n 3. The crawled depth is : \n " >> $tempfile
   cat /tmp/n.crawler.tmp >> $tempfile
   echo " \n 4. The explorer port is : \n " >> $tempfile
@@ -113 +120 @@
 }
 
+# define paramaters
+ROBOT=`cat /tmp/n.url.txt`
+URLS=`cat /tmp/n.url.txt`
+DEPTH=`cat /tmp/n.url.txt`
+PORT=`cat /tmp/n.url.txt`
+
 start_crawl () {
   echo_vb "7. start_crawl"
+  setup_nutch
+  /opt/nutch/bin/nutch crawl ~/.nutch/urls -dir ~/.nutch/search -threads 2 -depth $DEPTH 
 }
 
 start_tomcat () {
   echo_vb "8. start_tomcat "
+  /opt/nutch/tomcat/bin/startup.sh
 }