Sub-group finding - awk

Question

I have this kind of records (rows):
0 1 4 8 2 3 7 9 3 4 8 9 4 7 9 1 0 0 2 5 8 2 4 5 6 1 0 2 4 8 9 0
Definitions:
group: collection of numbers which are separated by 0-s (zeros)
sub-group: collection of numbers which are separated by local minima in the groups
local minimum: the numbers before and after it are greater
In the above example there are 3 groups and 7 sub-groups, i.e.
groups: 1 4 8 2 3 7 9 3 4 8 9 4 7 9 1 , 2 5 8 2 4 5 6 1 , 2 4 8 9
sub-groups: 1 4 8 , 3 7 9 , 4 8 9 , 7 9 1 , 2 5 8 , 4 5 6 1 , 2 4 8 9 (this last is identical to the group itself)
So, in these kind of records I have to
find the minima (print out: 2, 3, 4, 2)
the size (number of character) of these sub-groups
positions of numbers of the sub-groups in the groups
I have already started to write something, but I am stuck here...
Can anyone help me to solve this?
Here is the code so far:
#!/usr/bin/awk -f
{
db = split($0,a,/( 0)+ */)
for (i=1; i<=db; i++) {
split_at_max(a[i])
for (j=1; j<=ret_count; j++) {
print ""
for (k=1; k<=maximums[j]; k++) {
print ret[j,k]
}
}
}
}
function split_at_max(x) {
m_db = split(x,values," ")
for (mx in ret) {
delete ret[mx]
}
ret_count = 1
ret_curr_db = 0
for (mi=2; mi<m_db; mi++) {
ret_curr_db++
ret[ret_count,ret_curr_db] = values[mi-1]
if ( (values[mi-1] <= values[mi]) &&
(values[mi] >= values[mi+1]) &&
(values[mi+1] <= values[mi+2]) ) {
maximums[ret_count] = ret_curr_db
ret_count++
ret_curr_db = 0
}
}
ret_curr_db++
ret[ret_count,ret_curr_db] = values[mi-1]
ret_curr_db++
ret[ret_count,ret_curr_db] = values[mi]
maximums[ret_count] = ret_curr_db
}

Answer 1

interesting assignment.
wrote a quick and dirty awk script. there should be a lot of room to optimize. I don't know what kind of output are you expecting...
awk -v RS="0" 'NF>1{
delete g;
print "group:";
for(i=1;i<=NF;i++){
printf $i" ";
g[i]=$i
}
print "";
t=1;
delete m;
for(i=2;i<length(g);i++){
if(g[i-1]>g[i] && g[i]<g[i+1]) {
print "found minima:"g[i]
m[t]=i;
t++;
}
}
if(length(m)>0){
s=0;
for(x=1;x<=length(m);x++){
printf "sub-group: "
for(i=s+1;i<m[x];i++){
printf g[i]" "
s=m[x];
}
print "";
if(x+1>length(m)){
printf "sub-group: ";
for(i=s+1;i<=length(g);i++)
printf g[i]" "
print "";
}
}
}else{
print "no minima found. sub-group is the same as group:"
printf "sub-group: "
for(i=1;i<=NF;i++){
printf $i" ";
g[i]=$i
}
}
print "\n-----------------------------"
} yourFile
the output on your example input:
group:
1 4 8 2 3 7 9 3 4 8 9 4 7 9 1
found minima:2
found minima:3
found minima:4
sub-group: 1 4 8
sub-group: 3 7 9
sub-group: 4 8 9
sub-group: 7 9 1
-----------------------------
group:
2 5 8 2 4 5 6 1
found minima:2
sub-group: 2 5 8
sub-group: 4 5 6 1
-----------------------------
group:
2 4 8 9
no minima found. sub-group is the same as group:
sub-group: 2 4 8 9
-----------------------------
update
fixing for those "special" elements like 20,30,40...
still quick and dirty:
change my awk script above to
sed 's/^0$//g' yourFile | awk -v RS="" [following codes are the same as above]......
then the output is:
group:
6 63 81 31 37 44 20
found minima:31
sub-group: 6 63 81
sub-group: 37 44 20
-----------------------------