#!/usr/bin/ksh
#
# wget_usgs_gmag.ksh
#
# Author: Lydia Philpott (lphilpott@igpp.ucla.edu)
#
# Based heavily on wget_intermagnet_gmag.ksh
#
# This script uses wget to download data from USGS magnetometer sites
#
# This script is initially being set up to download data from BRW (Barrow). With the intention 
# that it be extendable to other sites in the future.
#
# $1 = site
# $2 = year
# $3 = month

#
## Set environment variables
#
  if [[ -z $THMSOC ]]
  then
    THMSOC=/disks/socware/thmsoc_dp_current
    #THMSOC=/home/thmsoc/lphilpotts_stuff
    export THMSOC
  fi
	. $THMSOC/src/config/soc_it_to_me.config
	. $THMSOC/src/config/wget_usgs_gmag.config
    #. $THMSOC/soc_it_to_me.config
    #. $THMSOC/wget_usgs_gmag.config
#
## Handle input arguments
#
site=${1}
year=${2}
month=${3}

#
## Set logfiles
#
	processdate=$(date '+%y%m%d')
	LOGFILE=${LOGDIR}/webdownloads/usgs_gmag_log_${processdate}
	WGETLOG=${LOGDIR}/webdownloads/wget_output_usgs_gmag_${processdate}_${site}_$$


	echo ":$$:usgs_gmag_log:" >> $LOGFILE
	echo ":$$:usgs_gmag_log:Starting script wget_usgs_gmag at $(date)" >> $LOGFILE

#
## Run wget to download data from usgs for given site
#
	if [[ $os = "SunOS" ]]
  then
  	siteupper=$(echo $site | nawk '{print toupper($1)}')
 	else
    siteupper=$(echo $site | awk '{print toupper($1)}')
  fi

	wgetTime=$(date '+%Y-%m-%d %T')
	url=${WEBSITE}/$siteupper/OneSecond/
	sitequery="$site$year$month*"
	/usr/bin/wget -r -nH --cut-dirs=5 -N -o $WGETLOG -P $LOCALWEBMIRROR/${site}/${year}/${month} -A $sitequery -np $url

	echo ":$$:usgs_gmag_log: Download Complete: $site" >> $LOGFILE

#
## Parse through wget output to see what was downloaded
#
	grep saved $WGETLOG | grep -v 'index\|robot' | awk '{print $6 " " $8}' > /tmp/wget_usgs_gmag$$
	while read line
	do
	        echo "line: $line"
		path=$(echo $line |awk '{print $1}')
		size=$(echo $line |awk '{print $2}')

		fileName=$(basename $path)
		fileName=${fileName%*\'}

		# Parse date information from filename
		year=$(echo $fileName |awk -F_ '{print substr($1,4,4)}')
		mon=$(echo $fileName |awk -F_ '{print substr($1,8,2)}')
		day=$(echo $fileName |awk -F_ '{print substr($1,10,2)}')
		fileTime="${year}-${mon}-${day} 00:00:00"

		# Verify file exists
		if [[ ! -a ${LOCALWEBMIRROR}/${site}/${year}/${month}/${fileName} ]]
		then
    	echo "$$:usgs_gmag_log:" >> $LOGFILE
    	echo "$$:intermagnet_gmag_log: ${LOCALWEBMIRROR}/${site}/${year}/${month}/${fileName} not found. " >> $LOGFILE
    	echo "$$:intermagnet_gmag_log: continuing..................." >> $LOGFILE

		fi

		fileSize=${size%\]}
		fileSize=${fileSize#\[}

		# Update database with stats
		${PHP_SCRIPTS}/wget_usgs_gmag.php $fileName $wgetTime $fileTime $fileSize

		# Create a file used for CDF processing
		cp ${LOCALWEBMIRROR}/${site}/${year}/${month}/${fileName} ${GMAGWORKDIR}/$fileName

	done < /tmp/wget_usgs_gmag$$

#
## Cleanup
#
	rm -f /tmp/wget_usgs_gmag$$


exit 0