#!/bin/sh

if [ ! -d stores ]; then
   echo "No stores there!"
   exit 1
fi

# small hack, try to find a decent version of sed
# like GNUsed, usually installed in /usr/local/bin 
if [ "$SED" = "" ]; then
    if [ -f /usr/local/bin/sed ]; then
        SED=/usr/local/bin/sed
    elif [ -f /usr/bin/sed ]; then
        SED=/usr/bin/sed
    else
        SED=sed
    fi
fi

echo \*\*\* Store cleaning started at `date`
echo \*\*\* Cleaning backup stores
find stores -name \*.bak -exec rm '{}' \;

echo \*\*\* Removing unused stores

indexFile=`ls stores/*-index.xml | sed 's,.*/,,'`

storeList=`grep repository stores/$indexFile | sed 's/.*>\(.*\)<.*/\1/'`

# try to find any duplicated root resource

rootKey=""
dupKey=0
echo $storeList | sed 's/root\.xml//' | grep root.xml > /dev/null
if [ $? -eq 0 ]; then
    echo \*\*\* Duplicated entry for root resource
    dupKey=1
fi

state=0
for i in `cat stores/root.xml | sed '/^\ \ <attri/ !d'  | sed 's/\ //g'` ; do
    echo $i | grep \>root\< > /dev/null
    if [ $? -eq 0 ]; then
        if [ $state -eq 1 ]; then
            echo WARNING two root entries in root.xml. Inspect manually
	    exit 1
        fi
        state=1
    fi
    echo $i | grep name=\'key\' > /dev/null
    if [ $? -eq 0 ]; then
        if [ $state -eq 1 ]; then
            rootKey=`echo $i | sed 's/.*>\(.*\)<.*/\1/'`
            state=0
        fi
    fi
done

echo \*\*\* RootKey is $rootKey

# remove all the stores in X/st-Y and find all the stores that are only
# in the index 

for i in `find stores -name st-\[0-9\]* -print | grep -v bak | grep -v backup | sed 's,stores/,,'` ; do
    isThere=0
    for j in $storeList ; do
	if [ "$j" = "$i" ]; then
	    isThere=1
	    indexStoreList=`echo $indexStoreList $i`
	    break
        fi
    done
    if [ $isThere -eq 0 ]; then
	echo $i is NOT there... removing it! 
        rm -f stores/$i
    fi
done

# find all the stores in the index that are not in the store directory

for i in $storeList ; do
    if [ "$i" = "root.xml" ]; then
	continue
    fi
    isThere=0
    for j in $indexStoreList ; do
	if [ "$j" = "$i" ]; then
	    isThere=1
	    break
	fi
    done
    if [ $isThere -eq 0 ]; then
	unexistentStoreList=`echo $unexistentStoreList $i`
    fi
done

echo \*\*\* Rebuilding cleanly the index file $indexFile

# rebuild the index file (usually http-server-index.xml) remove all
# the @@NULL@@ stores, the duplicate root resource (if there) and
# all the entries linked to repository that are not there

state=0
echo "<?xml version='1.0' encoding='UTF-8'?>" > stores/$indexFile.new
echo "<jigxml version=\"1.0\" xmlns=\"http://jigsaw.w3.org/JigXML/JigXML1.0\">"  >> stores/$indexFile.new

for i in `cat stores/$indexFile | sed -e 's/.*repository.*>\(.*\)<.*/repository=\1/' -e 's/.*key.*>\(.*\)<.*/key=\1/' -e 's/<resource.*/<resource>/'` ; do
    if [ "$i" = "</resource>" ]; then
        if [ $state -eq 0 ]; then
           echo Unstarted resource, aborting
           break
        fi
	if [ "$currentStore" = "@@NULL@@" ]; then
	    echo Removing @@NULL@@ entry
	    state=0
            currentStore=""
	    nullKeys=`echo $nullKeys $currentKey`
            currentKey=""
	    continue
	fi
        if [ "$currentStore" = "root.xml" ]; then
	    if [ "$currentKey" != "$rootKey" ]; then
		if [ $dupKey -eq 1 ]; then
		    echo Removing duplicate root entry $currentKey
		    state=0
		    currentStore=""
		    nullKeys=`echo $nullKeys $currentKey`
		    currentKey=""
		    continue
		fi
		echo Fixing wrong reference for root key
		currentKey=$rootKey
	    fi
	fi
	for j in $unexistentStoreList ; do
	    if [ "$j" = "$currentStore" ]; then
		echo Removing non-existent store $j
		state=0
		currentStore=""
		nullKeys=`echo $nullKeys $currentKey`
		currentKey=""
		continue
	    fi
	done
	if [ $state -eq 0 ]; then
	    continue
	fi
	echo " <resource class='org.w3c.tools.resources.store.NewStoreEntry'>" >> stores/$indexFile.new
	echo "  <attribute name='repository' flag='4' class='org.w3c.tools.resources.StringAttribute'>"$currentStore"</attribute>" >> stores/$indexFile.new
	echo "  <attribute name='key' flag='4' class='org.w3c.tools.resources.IntegerAttribute'>"$currentKey"</attribute>" >> stores/$indexFile.new
	echo " </resource>" >> stores/$indexFile.new
       	continue
    fi
   
    echo $i | grep key > /dev/null
    if [ $? -eq 0 ]; then
        currentKey=`echo $i | sed 's/.*=//'`
        state=1
        continue
    fi
    echo $i | grep repository > /dev/null
    if [ $? -eq 0 ]; then
        currentStore=`echo $i | sed 's/.*=//'`
        state=1
        continue
    fi
done
echo "</jigxml>" >> stores/$indexFile.new

mv stores/$indexFile stores/$indexFile.old
mv stores/$indexFile.new stores/$indexFile

echo \*\*\* Creating the sed script to clean all stores with unreferenced keys

# we are now creating a sed script that will clean every store
# it accepts all the known keys and destroy all container with an unknown key

keyList=`grep key stores/$indexFile | sed 's/.*>\(.*\)<.*/\1/'`
cat > resource.sed << EOF
/^ <resource/ {
            b Res
        }
        p
        d
:Bogus
/^ <\/resource>/ {
	w resource.rej
        d
}
/\n <\/resource>$/ {
	w resource.rej
        d
}
        N
        b Bogus
:Res
/^ <\/resource>$/ {
             p
             d
        }
/\n <\/resource>$/ {
            p
            d
        }
/'key'/ {
	    b Key
	}
	N
	b Res
:Key
EOF
for i in $keyList ; do
    echo "/  <attribute name='key'.*>"$i"<\/attribute>$/ {" >> resource.sed
    cat >> resource.sed << EOF
            p
            d
        }
EOF
done
cat >> resource.sed << EOF
/  <attribute name='key'/ {
            b Bogus
        }
        N
        b Res
EOF
$SED -n -f resource.sed resource.sed 2>/dev/null > /dev/null

# if the sed implementation is not good enough to handle more than 256 
# commands, ask people to manually inspect the stores

if [ $? -eq 0 ]; then
    echo \*\*\* Purging all stores
    for i in stores/root.xml `find stores -name st-\[0-9\]* -print | grep -v backup` ; do
        $SED -n -f resource.sed $i > $i.new
	if [ $? -eq 0 ]; then
	    mv $i.new $i
	else
	    echo store `echo $i | sed 's,stores/,,'` was not processed due to errors
	fi
	if [ -s resource.rej ]; then
	    mv resource.rej $i.rej
	fi
    done
else
    echo WARNING your sed implementation does not allow enough commands

    echo \*\*\* Checking for null repositories linked $nullKeys

    for i in `find stores -name st-\[0-9\]* -print | grep -v backup` ; do
	for j in `echo $nullKeys` ; do
	    grep \>$j\< $i > /dev/null
	    if [ $? -eq 0 ]; then
		echo \*\*\* Store `echo $i | sed 's,stores/,,'` needs inspection for key $j
	    fi
	done
    done
fi
# do some cleanup
rm -f resource.sed resource.rej 2>/dev/null > /dev/null

echo \*\*\* Store cleaning finished at `date`