Compare two files line-by-line

I'm not sure if there is an easy way to do this, but I thought it was
worth asking. I have two files which contain filenames and sum data. I
would like to compare the files line-by-line and output filenames for
which the sum data is different to a third file. I have no idea how I
can read two files in parallel for comparison, though. Shell scripts
(mine, anyway) are very linear -- they don't exactly lend themselves
to multitasking. Anyone have any pointers?

In article < XXXX@XXXXX.COM >,




exec 3<file1 4<file2

After this, you can use:

read line <&3

to read from file1, and:

read line <&4

to read from file2

-- 
Barry Margolin,  XXXX@XXXXX.COM 
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***

On Tue, 03 Feb 2004 03:44:07 GMT, Barry Margolin < XXXX@XXXXX.COM >




I've never dealt with exec, but the man page makes it sound pretty
simple. I was imagining something using paste to merge the files,
line-for-line, then use awk to compare $2 to $4. Unfortunately, both
your way and mine leave me with a problem I just ran across. This will
only work if the files are the same in both lists. If a file is
present in one list, but not the other, it will throw everything off
from that point forward.

Anything I can do to make the compare smarter? Basically, if a file is
in one list but not the other, log that as a failed checksum. If this
were Java, I would create a recursive loop to do the compares and
break when either a match was found or the end of the file was
reached, but I don't know if that is the way to do it in a shell
script. The first list would always be the master, so I would *only*
care if a file was in list 1 but not list 2, not vice versa. Is a
recursive loop my best option?

Okay, I've got something than seems to work. Again, though, I'm not
certain if it is the best way to go about it. This is just outputting
results to the screen for testing. I'll eventually make it output to a
file, then tar the needed files and transmit and untar them.

exec 3</tmp/master.list

while read line <&3
do
  exec 4</tmp/compare.list
  while read LINE <&4
  do
    if [ "$line" = "$LINE" ]
    then
      echo "$line matches $LINE."
      MATCH=1
      break
    fi
  done
  if [ $MATCH -ne 1 ]
  then
    echo "$line has no match."
  fi
  MATCH=0
done

Am I going in totally the wrong direction here? This seems to work,
but I have a tendency to do things the long way in shell scripts.
Thanks for the feedback.

In article < XXXX@XXXXX.COM >,




Maybe you can make use of "join" or "diff".

-- 
Barry Margolin,  XXXX@XXXXX.COM 
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***

man diff

-- 
William Park, Open Geometry Consulting, < XXXX@XXXXX.COM >
Linux solution for data management and processing. 

> > Anything I can do to make the compare smarter? Basically, if a file is

Or "sdiff" and make "grep" on caracther ">", "<" and "|". On my own script
I'm using sdiff to compare two files that cointain numeric data and
reporting in output difference occuring on each couple file.

Jul