UNIX for Dummies Questions & Answers

Well,
I don't think this is the real input and this is important because the code should be based on patterns. Anyway, given your sample you could use something like this with your original input file(a very quick, dirty and not idiomatic solution):

(the code should work even with the old Solaris awk)

Code:

awk '/student|class/ {
  if ($0 ~ /student/)
    s[++c] = $0 
  else {
    if (c > 1)
      while (++i <= c)
        printf "%s\n%s\n", s[i], $0
    else
      printf "%s\n%s\n", s[1], $0    
    split("", s)
    c = i = 0    
    }
}' filename

well thanks radoulov, but yes, your solution does depend on the pattern "student" or "class" .. however, the file doesn't have the actual words "student" and "class"!!! it's student names and class names.. that's why i was trying to use line by line editing and counters and such

Now you have the logic, you just need to adjust the patterns and that's left as an exercise.

well thanks to all for the help..

I've written a nawk program for your problem..

BEGIN {
counter=0;
store[1]="";
}

{
if(match($0,"class"))
{
for(i=1;i<=counter;i++)
{
printf("%s\n",store[i]);
printf("%s\n",$0);
}

for(i=1;i<=counter;i++)
{
delete store[counter];
}
counter=0;
}
else {
store[++counter]=$0;
}
}

Data is:
$ cat data
student1
classa
student2
classb
student3
student4
classc
student5
student6
student7
classd

$ nawk -f class.nawk data
student1
classa
student2
classb
student3
classc
student4
classc
student5
classd
student6
classd
student7
classd

you can redirect the output as needed.
I ain't a pro, but this solution works!

I've been able to figure out a solution myself too. thanks for the help..