filename=`basename $1`
min_seconds=${2:-100}

if [ ${filename: -3} == ".gz" ] ; then
  cat="zcat"
  date=`echo $filename  | awk 'match($0, /hiveserver2Interactive.log.([0-9-]+)_.*.gz/, a) {print a[1]}' `
else
  cat="cat"
  date=`date -I`
fi

declare queries=$($cat $filename | awk -v fname=$filename -v min_seconds=$min_seconds 'match($0, /queryId=([a-zA-Z_0-9-]+).*Time taken: +([0-9.]+)/, a) { if (a[2] > min_seconds) { print fname, a[1], a[2] }}')


if [ -z "$queries" ] ; then
  exit 0
fi

while read -r line; do
# sample line: hiveserver2Interactive.log.2018-12-04_1.gz hive_20181204021011_06d9474f-587c-4f36-9456-85a568c9919a 4660.311
   params=( $line )
   filename=${params[0]}
   queryId=${params[1]}
   queryTime=${params[2]}

   queryString=$($cat "$filename" | sed -n "/Compiling command(queryId=$queryId)/,/$date/{
        s/.*Compiling command(queryId=$queryId): \(.*\)/\1/
        /$date/!p
   }"
   echo ";"
   )
   echo  "$filename", "$queryId", "$queryTime" , "${queryString}" | tr "\n" " "
   echo ""
done <<< "${queries%x}"