#!/bin/sh -e

# find databases with lost extFiles.  creates a directory
# gbRoot/misc/lostExtFiles/$host with the details, as well as a list of
# databases in gbRoot/misc/lostExtFiles/$host.lost
# and a log in gbRoot/misc/lostExtFiles/$host.log

HOST=`hostname`

dbsFile=/genbank/etc/rr.dbs
if [ $HOST = "hgwbeta" ] ; then
    dbsFile=/genbank/etc/hgwbeta.dbs
fi
if [ $HOST = "hgwdev" ] ; then
    dbsFile=/cluster/data/genbank/etc/hgwdev.dbs
fi
   
gbRoot=`dirname $dbsFile`
gbRoot=`dirname $gbRoot`
outDir=$gbRoot/misc/lostExtFiles/$HOST
mkdir -p $outDir
logFile=$outDir.log
lostFile=$outDir.lost
exec </dev/null >$logFile 2>&1

dbs="`egrep '^[a-z]' $dbsFile`"

for  db in  $dbs ; do
    echo checking $db
    hgc-sql -N -e 'select acc,gbExtFile from gbSeq left join gbExtFile on gbExtFile.id = gbSeq.gbExtFile where gbExtFile.id is null' $db >$outDir/$db.lost
    if [ -s $outDir/$db.lost ] ; then
        echo $db >>$lostFile
    fi
done
echo done
