How to split file , .. awk or split

ERBRMN · 07-19-2006, 05:39 AM

Structure of my result file is following :
-----------
11
12
13
14
15
END
21
22
END
31
32
34
45
61
62
END

----------

How can I split it to three text file. File name is free, but indexed.
"END" is split string for files (end of each file).
I want to do it using AWK, but shell is OK.
Have you any idea ???

spirit receiver · 07-19-2006, 06:01 AM

The following will append the corresponding parts to file1, file2, ... . It will read the content from STDIN, so you may invoke it as "command < your_data_file" from the shell.

Code:

#!/bin/bash
FILENAME=file
INDEX=1
while read
do
 if [[ "$REPLY" = "END" ]]
 then
  (( INDEX++ ))
 else
  echo "$REPLY" >> $FILENAME$INDEX
 fi
done

ERBRMN · 07-19-2006, 06:58 AM

Quote:

Originally Posted by spirit receiver

The following will append the corresponding parts to file1, file2, ... . It will read the content from STDIN, so you may invoke it as "command < your_data_file" from the shell.

Code:

#!/bin/bash
FILENAME=file
INDEX=1
while read
do
 if [[ "$REPLY" = "END" ]]
 then
  (( INDEX++ ))
 else
  echo "$REPLY" >> $FILENAME$INDEX
 fi
done

Thank you very much, spirit receiver

It was good job.

konsolebox · 07-20-2006, 11:32 PM

Hello there. I still recommend spirit receiver's script the first but just in case you'll want to use this too.

command.sh file

Code:

#!/bin/bash
FILENAME="$1"
INDEX=1

IFS=$'\n'
for REPLY in $(<$FILENAME); do
 if [[ "$REPLY" = "END" ]]; then
  (( INDEX++ ))
 else
  echo "$REPLY" >> $FILENAME$INDEX
 fi
done

#IFS=$' \t\n'

regards

ERBRMN · 07-21-2006, 12:56 AM

Quote:

Originally Posted by konsolebox

Hello there. I still recommend spirit receiver's script the first but just in case you'll want to use this too.

command.sh file

Code:

#!/bin/bash
FILENAME="$1"
INDEX=1

IFS=$'\n'
for REPLY in $(<$FILENAME); do
 if [[ "$REPLY" = "END" ]]; then
  (( INDEX++ ))
 else
  echo "$REPLY" >> $FILENAME$INDEX
 fi
done

#IFS=$' \t\n'

regards

Thk you, konsolebox

This is also good job.

Tinkster · 08-14-2006, 07:58 PM

A tad late, and not quite perfect, but worth looking at :}

Code:

awk 'BEGIN{RS="END";OFS="\n"}{$1=$1; print > "myfile"NR}' test.txt

I haven't quite figured out why it will produce a 4th file (with only
and empty line in it) even though there's just three records.

Cheers,
Tink

homey · 08-14-2006, 08:21 PM

Tinkster,
On my box, it only makes three files if there is no \n after the last END in the file.txt .

Tinkster · 08-14-2006, 09:07 PM

Ok, looked at it again. Behaves correctly if I do RS="END\n"

Odd.

Cheers,
Tink

frob23 · 08-14-2006, 10:37 PM

Code:

#!/bin/sh
index=1
filebase=foo

while read LINE
do
        if [ "${LINE}" = "END" ];
        then
                index=`expr ${index} + 1`
        else
                echo "${LINE}" >> ${filebase}.${index}
        fi
done

Variation of the first response but is bourne shell compatible (/bin/sh) so it's portable on systems where bash is not in the base. I love bash but, with something this simple, bash specific functionality is not required to reduce complexity and only frustrates portability.

You can see it's not much different from the first reply; you just have to include a variable for the read and use expr (an external command) to do the math because there's no shell arithmatic in the original bourne shell.

Oh, and I named the files foo.# because that's my personal preference... easy to change if you want.

Can anyone tell I am bored?

Tinkster · 08-15-2006, 01:02 AM

Quote:

Originally Posted by Tinkster

Ok, looked at it again. Behaves correctly if I do RS="END\n"

Odd.

Actually, thinking about it some more (and I'm kind of slow due to
a flu :}) it's not that odd at all. If END is the delimiter, the \n does
comprise a 4th record that's unterminated.

Cheers,
Tink