#!/bin/bash

# if this file is executed, /bin/sh is used, as we don't start with #!
# this should work under ash, bash, zsh, ksh, sh style shells.
# $1 should be the collection name
# $2 should be the oai URL
# $3 should the cache directory
# $4 should be the max records
# $5 should be the OAI metadata prefix
# $6 should be GS3 root
# $7 should be the collection directory
# $8 should the log file
# $9 should be the creator
# $10 should be the proxy settings

COL_NAME=$1
OAI_URL=$2
CACHE_DIR=$3
MAX_RECORDS=$4
OAI_PREF=$5
GS3_ROOT=$6
COLLECT_DIR=$7
LOG_FILE=$8
CREATOR=$9
#SHIFT
#PROXY=$9

if [ -x /usr/bin/lsof ] ; then
  lsof=/usr/bin/lsof
else 
  if [ -x /usr/sbin/lsof ] ; then
    lsof=/usr/sbin/lsof
  else 
    echo "Warning: Unable to find 'lsof', skipping test for number of files open"
  fi
fi

TOMCAT_ID=`ps ux  | grep tomcat | grep java | grep -v grep | awk -F" " '{ print $OAI_URL }'`

if [ ! -z "$lsof" ] ; then
  echo Tomcat_ID: $TOMCAT_ID
  #cd /usr/sbin
  NUM_OPEN_FILES=`$lsof -p $TOMCAT_ID | wc -l `
  echo open files: $NUM_OPEN_FILES
fi

cd $GS3_ROOT
 
source gs3-setup.sh  > $LOG_FILE

cd gs2build/
source setup.bash >> $LOG_FILE

perl -S mkcol.pl -gs3mode -buildtype lucene -collectdir $COLLECT_DIR $COL_NAME

echo $COL_NAME $OAI_URL $CACHE_DIR $MAX_RECORDS $OAI_PREF $GS3_ROOT $COLLECT_DIR $LOG_FILE >> $LOG_FILE

if [ ! -z "$lsof" ] ; then
  NUM_OPEN_FILES=`$lsof -p $TOMCAT_ID | wc -l `
  echo open files: $NUM_OPEN_FILES
fi

if [ ! -e $CACHE_DIR ] # be sure the cache directory exists
then
   mkdir $CACHE_DIR
else
   /bin/rm -rf $CACHE_DIR/*
fi

echo "Away to download OAI"  >> $LOG_FILE

if [ ! -z "$lsof" ] ; then
  NUM_OPEN_FILES=`$lsof -p $TOMCAT_ID | wc -l `
  echo open files: $NUM_OPEN_FILES
fi

perl -S downloadfrom.pl -download_mode OAI -cache_dir $CACHE_DIR -url $OAI_URL -max_records $MAX_RECORDS -metadata_prefix $OAI_PREF $CREATOR 2>&1 >> $LOG_FILE

if [ ! -z "$lsof" ] ; then
  NUM_OPEN_FILES=`$lsof -p $TOMCAT_ID | wc -l `
  echo open files: $NUM_OPEN_FILES
fi