How can I add a line to a file in a shell script?

Question

I want to add a row of headers to an existing CSV file, editing in place. How can I do this?

echo 'one, two, three' > testfile.csv

and I want to end up with

column1, column2, column3
one,     two,     three

(nb changing the initial CSV output is out of my hands)

EDIT: I originally had this as 'using sed' but any standard command will do. The important thing is the file is edited in place, and the line is inserted at the beginning of the file. Thanks,

Answer 1

This doesn't use sed, but using >> will append to a file. For example:

echo 'one, two, three' >> testfile.csv

Edit: To prepend to a file, try something like this:

echo "text"|cat - yourfile > /tmp/out && mv /tmp/out yourfile

I found this through a quick Google search.

Answer 2

sed is line based, so I'm not sure why you want to do this with sed. The paradigm is more processing one line at a time( you could also programatically find the # of fields in the CSV and generate your header line with awk) Why not just

echo "c1, c2, ... " >> file
cat testfile.csv >> file

?

Answer 3

As far as I understand, you want to prepend column1, column2, column3 to your existing one, two, three.

I would use ed in place of sed, since sed write on the standard output and not in the file.

The command:

printf '0a\ncolumn1, column2, column3\n.\nw\n' | ed testfile.csv

should do the work.

perl -i is worth taking a look as well.

Answer 4

Note that for large text files, this is going to cause some real problems in terms of disk fragmentation and speed.

Answer 5

This adds custom text at the beginning of your file:

echo 'your_custom_escaped_content' > temp_file.csv
cat testfile.csv >> temp_file.csv
mv temp_file.csv testfile.csv

Answer 6

Use perl -i, with a command that replaces the beginning of line 1 with what you want to insert (the .bk will have the effect that your original file is backed up):

perl -i.bk -pe 's/^/column1, column2, column3\n/ if($.==1)' testfile.csv

Answer 7

To answer your original question, here's how you do it with sed:

sed -i '1icolumn1, column2, column3' testfile.csv

The -i option also takes an optional argument to create a backup file, for example

sed -i~ '1icolumn1, column2, column3' testfile.csv

would keep the original file in "testfile.csv~".