Understanding how csplit works

ahmedb72 · 07-05-2009, 05:21 AM

Hi,
I tried to split a file into 3 pieces. Following is what I did:

Code:

#ls -lh *
-rw-r--r--  1 root root 156K Jul  5 13:23 file1.txt
#wc -l file1.txt
4617 file1.txt
#csplit file1.txt 10 {3}
279
398
390
310
157767
#ls -lh *
-rw-r--r--  1 root root 156K Jul  5 13:23 file1.txt
-rw-r--r--  1 root root  279 Jul  5 14:14 xx00
-rw-r--r--  1 root root  398 Jul  5 14:14 xx01
-rw-r--r--  1 root root  390 Jul  5 14:14 xx02
-rw-r--r--  1 root root  310 Jul  5 14:14 xx03
-rw-r--r--  1 root root 155K Jul  5 14:14 xx04
#cat xx00 | wc -l
9
#cat xx01 | wc -l
10
#cat xx02 | wc -l
10
#cat xx03 | wc -l
10
#cat xx04 | wc -l
4578

Here are my questions:
1) Why the command generated 5 files, although I put 3 in the braces?
2) Why the first file (xx00) contains 9 lines, although I specified to split the file into 10-line-size sections?

Thanks in advance.

colucix · 07-05-2009, 06:15 AM

From the csplit man page:

Code:

Each PATTERN may be:

       INTEGER
              copy up to but not including specified line number
       {INTEGER}
              repeat the previous pattern specified number of times

Your command copy up to (but not including) the 10th line and repeat this three times. So your final result is 1 file of 9 lines, 3 files of 10 lines, 1 file with the rest. I don't know why the three files have 10 lines instead of 9, but the result is quite consistent with the command you used. What is the result you want to achieve? Maybe split -l is more useful.

ahmedb72 · 07-05-2009, 11:42 PM

Hi colucix,

In my command, the INTEGER=10 and
the {INTEGER}={3}

Therefore, expect to see fours files, not five. Also, I expect all of them should be of same number of lines.

colucix · 07-06-2009, 12:44 AM

You can't split a file with 4617 lines in 4 pieces of 10 lines each! The rest remains in its own fragment. If you want an equal number of lines, try split, since the csplit command is more useful when you want to split a file based on some pattern.