dump() {
	cat - | gawk ' {for(i=2;i<=NF;i++) print $i}'
}

nasacoc7() { cat<<EOF | sed 's/[ \t]+/ /g'
prec			ALL:C2:C3			
flex			ALL:C2:C3			
resl			ALL:C2:C3			
pmat		ALL:C2				
pvol		ALL:C2:C3		C2:C3		
rely	ALL					
data	C2					
plex						
ltex	ALL	ALL:C3	C2			
cplx		C2:C3		C2		
time				C3		ALL
stor				ALL		
acap	C3	C2		C2		
pcap	ALL:C2:C3		C3			
apex					ALL	
pcon		ALL:C2:C3				
site						
ruse		ALL:C2:C3				
docu		ALL:C2:C3				
sced						
tool		ALL			C2:C3	
team	ALL:C2:C3					
EOF
}
xalan7() { cat<<EOF | sed 's/[ \t]+/ /g'
wmc	ALL		C2:C9	
dit		ALL:C0:C1:C2:C3:C4:C5:C7:C8		ALL
noc	C0:C1:C2:C3:C4:C5:C7:C8:C9			C0:C1:C2:C3:C5:C7:C8
cbo	ALL:C0:C1:C3:C4:C5:C7:C8			C0:C1:C2:C4:C8:C9
rfc	ALL:C1:C3:C5:C7:C8			C0:C1:C2:C3:C4:C5:C7:C8
lcom	C9	C0:C1:C2:C3:C4:C5:C7:C8		
ca		ALL	C9	
ce				ALL:C9
npm		C4:C7	C0:C1:C2:C3:C4:C5:C7:C8	ALL
lcom3			ALL:C0:C1:C2:C3:C4:C5:C7:C8	
loc	C0:C1:C3:C4:C5:C7:C8		C9	
dam	C2		C2:C4:C9	
moa			ALL:C0	C9
mfa	C0:C1:C3:C4:C5:C7:C8		ALL:C9	ALL
cam	C9	ALL:C0:C1:C2:C3:C4:C5:C7:C8:C9	C2:C9	
ic	ALL			
cbm				ALL
amc	C0:C1:C2:C3:C4:C5:C7:C8	C0:C1:C2:C3:C4:C5:C7:C8		C9
max_cc		C0:C1:C2:C3:C4:C5:C7:C8		ALL
avg_cc		ALL:C0:C1:C2:C3:C4:C5:C7:C8		C3:C9
EOF
}
china7() { cat<<EOF | sed 's/[ \t]+/ /g'
AFP	C1:C4		C3:C6:C8				C2
Input	C6:C8			ALL:C1:C4:C8	C8		ALL:C1:C4
Output		C1	C2:C8				ALL:C2:C4
Enquiry	ALL:C1:C4				C2:C3:C6		
File	C8	C3:C6:C8	ALL:C1:C8				C2
Interface							C2:C3
Added	C3:C6:C8		ALL:C1:C4:C8	ALL:C1:C4:C8			
Changed				C3:C6:C8			C2
Deleted	ALL:C1:C2:C3:C4:C8						
PDR_AFP		ALL:C1:C4	C2	C8			ALL:C1:C4:C6
PDR_UFP			C2:C6	ALL:C1:C4:C6	C2	C2:C6	C3
NPDR_AFP		ALL:C1	C3:C6	ALL:C1:C4			C1:C4:C6
NPDU_UFP		ALL:C4	C2:C6			ALL:C1:C3:C4:C6	C1:C3:C4:C6
Resource					C2:C3:C8		C2:C3
Dev.Type	ALL:C1:C2:C3:C4:C6:C8						
Duration	ALL:C3:C6:C8				C2:C3:C8		ALL:C3
EOF
}
lucene7() { cat<<EOF | sed 's/[ \t]+/ /g'
wmc						ALL	
dit							
noc	C2:C3						
cbo							
rfc	ALL			C0:C1:C2:C3:C5		C5	
lcom					C0:C1:C2:C3:C5	ALL:C0:C1:C2:C3:C5	
ca				C5			
ce							
npm			C3	ALL		C0:C1:C2:C3	
lcom3			ALL:C0:C1:C2:C3:C5	ALL		C0:C1:C2:C5	
loc					ALL	C1:C3:C5	
dam				ALL	C0:C1:C2:C3:C5		
moa							
mfa	C0:C1:C2:C3:C5					ALL:C0:C1:C2:C3:C5	
cam		ALL:C0:C1:C3:C5			ALL:C0:C1:C2:C3		
ic	ALL						
cbm					C0:C1:C2:C3:C5		
amc	C0:C1:C2:C5	C0:C1:C2:C3:C5		C0:C2			ALL
max_cc						ALL:C0:C1:C2:C3:C5	
avg_cc		ALL			ALL:C0:C1:C2:C3:C5		
EOF
}
clusters() {
	cat - |
	gawk -F: '
        {for(i=1;i<=NF;i++)  {N[$i]++; if($0 ~ "ALL") Overlaps[$i]++}}

	END { 
		print "no overlap" ; 
		for(i in N) {
			o = Overlaps[i] 
			o = o ? o : 0	 
			print i,N[i], o, 100 - int(100*o/N[i]) | "sort -n -k 4"
		}
		#print "others" ; for(i in Diff) if(! (i in Overlaps) )print i,Diff[i]
	}
    '

}	   
worker() {
	echo $1
	$1  | dump | clusters
}

for i in china7 lucene7 xalan7 nasacoc7; do
	worker $i
done