CODEWITHSUNDEEP
bashinsertsed
Sriram
Sriram

Reputation: 10558

Inserting text from a file to the beginning of another file - bash

I am trying to insert some copy-right information into the beginning of some source files. I know that using sed in the following way:

sed "1iSome copyrighted information." sample.txt

would insert the text "Some copyrighted information" to the beginning of sample.txt.
I need to insert text from a file to the beginning of sample.txt. Is there any way in sed that I could use a cat command for the above purpose, say something like the following?:

sed "1i{cat header.txt}" sample.txt

I have googled for the above and have not found exactly what I have been looking for. Any help on this is most welcome.

Thanks.

Upvotes: 9

Views: 13064

Answers (6)

sdaau
sdaau

Reputation: 38619

Sorry to use up a whole answer - just for the sake of formatting :) Basically, the same answer from @tangens, but without the temporary file:

echo "$(cat header.txt sample.txt)" > sample.txt

Note the quotes (") to preserve the line endings, and the fact that the cat concatenation runs in a subprocess - so there shouldn't be the possibility of overwriting a file (the sample.txt) which is open for reading..

Cheers!

Upvotes: 11

Dennis Williamson
Dennis Williamson

Reputation: 359905

If you have GNU sed:

sed -i -e '2{x;G};1{h;rheader.txt' -e 'd}' sample.txt

sample.txt must be contain at least two lines.

This method works even if the header file contains characters that are special to sed.

Explanation:

  • -i - edit the file in place
  • -e - break the script up into sections - this is necessary in this case to delimit the end of the header filename (it could also be delimited by a newline)
  • 1{h; - on the first line of the file (sample.txt) save it to hold space
  • rheader.txt - read in the header file (header.txt)
  • d} - delete the original first line of the file from pattern space and end processing of line 1 - deleting it here prevents an extra blank line from being inserted at the beginning of the file
  • the header file contents are now output
  • 2{x; - when line 2 of the file (sample.txt) is read, swap it into hold space and swap hold space (containing the original line 1) into pattern space
  • G} - append hold space onto the end of pattern space (which now contains original lines 1 and 2) and complete processing of line 2
  • lines 1 and 2 are now output then processing continues for the rest of the file which consists of simply reading and outputting each line.

Edit: I removed a superfluous command from the version I originally posted.

Upvotes: 10

hluk
hluk

Reputation: 6016

Use ed text editor:

echo -e '0r header.txt\nw' | ed sample.txt

or use vi/ex command:

vi - +'0r header.txt|wq' sample.txt

But I don't see any way to run a command in sed.

Upvotes: 8

Gaius
Gaius

Reputation: 2585

Why not do this the other way round, e.g. append your source files to your copyright file?

The 'cat' command is actually short for 'concatenate', you can simply say

for f in *.c
do
    cp $f /tmp
    cat copyright.txt /tmp/$f >$f
done

Upvotes: 0

codaddict
codaddict

Reputation: 454950

If you really want to do it using sed:

INFO=$(cat header.txt)  # read the contents of file headers.txt into var INFO
sed -i "1i$INFO" sample.txt # insert $INFO, use -i to change the file inplace

I would however use the cat based method.

Upvotes: 1

tangens
tangens

Reputation: 39733

cat header.txt sample.txt > temp.txt
mv temp.txt sample.txt

Upvotes: 5

Related Questions