I have a output file which contains n number of document.Each document has n number of segments and identified using below points

The starting segment is ISA and Ending segment is IEA
Each document has unique number and it will be passed in REF*D9 segment

Each line in sample file is called segment and each segment is group of field delimited with (*)

Since i cant paste the entire contents i have attached the sample file.

Refer the sample file for exact file structure.

If i need to display particular segment from large sample file

For eg N3 segment which is under NM1*85 segments for following documents alone

12000000010
12000000011
12000000012
12000000013
12000000014
12000000015
12000000016
12000000017

1.awk 'a-->0;/NM1\*85/{a=1}/REF\*D9/' file > file1
2.nawk '{s=s?s"\t"$0:$0}/REF/{print s;s=""}' file1|sed -n '/12000000010/,/12000000017/p'

3.awk 'a-->0;/NM1\*85/{a=2}/REF\*D9/' file|egrep -v "N3" > file2
4.nawk '{s=s?s"\t"$0:$0}/REF/{print s;s=""}' file2|sed -n '/12000000010/,/12000000017/p'

So I need something user interactive

i.e it should prompt 
for search pattern and no of lines to be displayed after that pattern as i mentioned in 1st command
for segment to be omitted as i mentioned in 3rd command(egrep)
and also for start and end document no. as i mentioned in 4rd command(sed)
all these validation need to be done in tmp file or something
and as soon i provided start and end document no it should display the output

OS - SunOS viper 5.9 Generic_117171-07 sun4u sparc SUNW,Sun-Fire-480R
Shell - csh

Attached Files

file.txt (520.9 KB, 9 views)

A shell loop can ask for the pattern, capture matching lines in a temp file, count the lines at the same time, and if any lines, turn you over to vi, pg, less in the temp file to view and play in the subset. When you exit the viewer, it asks you if you want to view again, and if not, loops. You can make any copies you want from inside the viewer.

Now, only a select few actually prefer the csh. The bash and ksh93 are much nicer and there are more experts to help.

PS: I use the ksh/bash 'set -o vi' option to do command recall, so I do not type commands over and over. I never even leave my home dir except inside (...) so I return automatically. It makes the history more valuable. I also save the history file occasionally where my tools can find it, and set a huge HISTSIZE of 32767. I consider one of the best CUI's.

ok i will be happy if i get the solution based on any shell

In bash, something like this? File names go on command line.

#!/bin/bash
 
ans=y
 
while [[ "$ans" == '[Yy]*' ]]
 do
  echo "Enter pattern: \c"
  read ptn
  echo 'Searching . . . \c'
  lc=$( grep -E "$ptn" "$@" | tee /tmp/patgrep.$$ | wc -l )
  echo "$lc lines found."
 
  if [ $lc != 0 ]
   then
    sleep 6
    vi /tmp/patgrep.$$
   fi
 
  echo 'Another (y/n)? \c'
  read ans
 done