I want to delete a batch from file - shell

Question

I have a file and contents are like :
|T1234
010000000000
02123456878
05122345600000000000000
07445678920000000000000
09000000000123000000000
10000000000000000000000
.T1234
|T798
013457829
0298365799
05600002222222222222222
09348977722220000000000
10000057000004578933333
.T798
Here one complete batch means it will start from |T and end with .T.
In the file i have 2 batches.
I want to edit this file to delete a batch for record 10(position1-2),if from position 3 till position 20 is 0 then delete the batch.
Please let me know how i can achieve this by writing a shell script or syncsort or sed or awk .

Answer 1

I am still a little unclear about exactly what you want, but I think I have it enough to give you an outline on a bash solution. The part I was unclear on is exactly which line contained the first two characters of 10 and remaining 0's, but it looks like that is the last line in each batch. Not knowing exactly how you wanted the batch (with the matching 10) handled, I have simply written the remaining wanted batch(es) out to a file called newbatch.txt in the current working directory.
The basic outline of the script is to read each batch into a temporary array. If during the read, the 10 and 0's match is found, it sets a flag to delete the batch. After the last line is read, it checks the flag, if set simply outputs the batch number to delete. If the flag is not set, then it writes the batch to ./newbatch.txt.
Let me know if your requirements are different, but this should be fairly close to a solution. The code is fairly well commented. If you have questions, just drop a comment.
#!/bin/bash
ifn=${1:-dat/batch.txt} # input filename
ofn=./newbatch.txt # output filename
:>"$ofn" # truncate output filename
declare -i bln=0 # batch line number
declare -i delb=0 # delete batch flag
declare -a ba # temporary batch array
[ -r "$ifn" ] || { # test input file readable
printf "error: file not readable. usage: %s filename\n" "${0//*\//}"
exit 1
}
## read each line in input file
while read -r line || test -n "$line"; do
printf " %d %s\n" $bln "$line"
ba+=( "$line" ) # add line to array
## if chars 1-2 == 10 and chars 3 on == 00...
if [ ${line:0:2} == 10 -a ${line:3} == 00000000000000000000 ]; then
delb=1 # set delete flag
fi
((bln++)) # increment line number
## if the line starts with '.'
if [ ${line:0:1} == '.' ]; then
## if the delete batch flag is set
if [ $delb -eq 1 ]; then
## do nothing (but show batch no. to delete)
printf " => deleting batch : %s\n" "${ba[0]}"
## if delb not set, then write the batch to output file
else
printf "%s\n" ${ba[#]} >> "$ofn"
fi
## reset line no., flags, and uset array.
bln=0
delb=0
unset ba
fi
done <"$ifn"
exit 0
Output (to stdout)
$ bash batchdel.sh
0 |T1234
1 010000000000
2 02123456878
3 05122345600000000000000
4 07445678920000000000000
5 09000000000123000000000
6 10000000000000000000000
7 .T1234
=> deleting batch : |T1234
0 |T798
1 013457829
2 0298365799
3 05600002222222222222222
4 09348977722220000000000
5 10000057000004578933333
6 .T798
Output (to newbatch.txt)
$ cat newbatch.txt
|T798
013457829
0298365799
05600002222222222222222
09348977722220000000000
10000057000004578933333
.T798