Linux: Extract the first line of a file

Question

I am working with OpenWrt and a very small amount of space.

Trying to extract the first line from a file. The line needs to go into a variable and be removed from the file. I can use head to put it into a variable but can't use tail because as far as I understand I would have to do tail file > newFile and I do not have room for that second file.

Does some one know if a better technic?

Answer 1

使用

sed -i -e '1 w /dev/stdout' -e '1d' file

Answer 2

Edit: you can't use my old answer (see below) with OpenWrt since OpenWrt doesn't ship with ed . What a shame. So here are two methods:

The vi way

vi is a genuine editor too, so the following will work:

vi -c ':1d' -c ':wq' file > /dev/null

We open the file with vi , and use the commands :1d to delete the first line and :wq to save and quit, redirecting all output to /dev/null . Cool, clean, short and simple.

Oh, you will of course run:

firstline=$(head -n1 file)

before running this vi command to get the first line of the file into the variable firstline .

Note. On a system with very little memory, and when file is huge, this method fails.

The dd way

dd is a cool tool for this. The dd methods given in other answers are really great, yet they rely on the truncate utility which does not ship with OpenWrt. Here's a workaround:

firstline=$(head -n1 file)
linelength=$(head -n1 file | wc -c)
newsize=$(( $(wc -c < file) - $linelength ))
dd if=file of=file bs=1 skip=$linelength conv=notrunc
dd if=/dev/null of=file bs=1 count=0 seek=$newsize

This will work even with huge files and very little memory! The last dd command plays the role of the truncate command given in the other answers.

---

Old answer was:

You can use ed for this:

firstline=$(printf '%s\n' 1p d wq | ed -s file.txt)

At each call, you'll get the first line of the file file.txt in the variable firstline , and this line is removed from the file.

Answer 3

Show first line and delete it:

 head -1 file
 sed -i 1d file

This does, however, implicitly create a temporary file. A better solution could be sponge as pointed out in this article about “In-place” editing of files .

Answer 4

I wouldn't call it elegant, but here is a potential way:

# Find the number of characters in the first line
BYTES=$(head -n 1 file | wc -c)
# Shift all the data in-place
dd if=file of=file bs=1 skip=$BYTES conv=notrunc
# Remove the extra bytes on the end
truncate -s -$BYTES file

Answer 5

There are no standard utilities (except of dd ) to modify a file inplace without any temporary file.

One solution which might work is:

LINELENGTH=$(head -1 test.txt | wc -c)
tail -n +2 test.txt | dd of=test.txt conv=notrunc
truncate -s -$LINELENGTH test.txt

See https://unix.stackexchange.com/a/11074/15241 for discussion of this topic.

EDIT:

As the file is downloaded first with curl (see comments). There is another option:

curl URL | sed -e "1w first_line.txt" -e "1d" > rest.txt
firstline=$(cat first_line.txt)
rm first_line.txt

Answer 6

If you have enough ram:

contents=$(cat file) 
echo "$contents" | sed -n '1 !p' > file

Alternatively, use ed:

ed file
2,$wq