# check_offset   Usage:  check_offset filenames

# If you download this from my web page, it is important to
# use the SOURCE FILE.  DON'T mouse it off the screen
# because many of the characters will not be visible.

# Note to myself: remember to maintain the link
# ln check_offset /usr/njas/wwwfiles/sequences/check_offset.txt

# Call this file  check_offset (not check_offset.txt)

if [ "$#" -eq 0 ]
then
echo "Incorrect no of arguments. Usage: check_offset file(s) "
echo "Checks offsets in named files"
echo "Error message indicating empty offset may be caused by blank line in sequence."
echo "Results will be in  report_offset."
exit 1
fi

# choose version of awk to use
AWK=/bin/gawk

echo "Checking offsets in $*. Results will be in  report_offset."

cat $* | $AWK '

BEGIN {
RS = ""; FS = "\n"
type[ 2]="%A"
#type[ 3]="%B"
#type[ 4]="%C"
type[ 5]="%F"
type[ 6]="%I"
type[ 7]="%K"
type[ 8]="%N"
type[ 9]="%O"
type[10]="%r"
type[11]="%R"
type[12]="%S"
type[13]="%T"
type[14]="%U"
type[15]="%W"
type[16]="%Y"
type[17]="%Z"

# clear numbers of all possible types
for (i in type) have[type[i]]=0;

# END OF BEGIN
      }

# START PROCESSING NEXT SEQUENCE:

{								# left 0

seq = ""
seq2 = ""
seq3 = ""
name = ""
ref = ""
offset = ""
cross = ""
memo = ""
formula = ""
callfor = ""

for (i = 1; i <= NF; i++)
           {							# left 1
            if ($i ~ /^%I/)   id = substr( $i, 4, 10 )
            if ($i ~ /^%S/) { seq = $i ; sub( /.........../, "", seq ) }
   if ($i ~ /^%T/) { seq2 = $i ; sub( /.........../, "", seq2 ); seq = seq seq2 }
   if ($i ~ /^%U/) { seq3 = $i ; sub( /.........../, "", seq3 ); seq = seq seq3 }
            if ($i ~ /^%N/) { name = $i ; sub( /.........../, "", name ) }
            if ($i ~ /^%R/)   ref = $i 
            if ($i ~ /^%O/) { offset = $i ; sub( /.........../, "", offset )  }
            if ($i ~ /^%Y/) { cross = $i ; sub( /.........../, "", cross ) }
            if ($i ~ /^%Z/) { memo = $i ; sub( /.........../, "", memo ) }
            if ($i ~ /^%F/) { formula = $i ; sub( /.........../, "", formula ) }

for (j in type ) { if ($i ~ type[j])   ++have[type[j]] };

           };							# right 1

# ACTION TO BE TAKEN ON SEQUENCE
	{
#	print "Seq"
#	print seq
#	print offset
if ( offset !~ /,/ )
	{ print "Error, no comma in offset in", id, ". %O is", offset }
else    {							# left A
	split( offset, oo, "," ); o = oo["2"];
#	print "good %O", o
# now check that claimed offset, o, is correct!
# i will be true offset (k = 0 indicates all terms <= 1)
	k = 0
nnn = split( seq, sequence, "," )
for ( i = 1; i <= nnn; i++)
	if ( sequence[i] > 1 )
		{ k = 1; break}
	if ( k == 0 ) 
		i = 1
	if ( i != o )
		{ print "Error in offset at", id, ". True, claimed = ", i, o }
 	}							# right A
#	print " "
	}

}								# right 0
    ' >report_offset