How to insert a text at the beginning of a file?

So far I've been able to find out how to add a line at the beginning of a file but that's not exactly what I want. I'll show it with an example:

File content

some text at the beginning

Result

<added text> some text at the beginning

It's similar but I don't want to create any new line with it...

I would like to do this with sed if possible.

sed can operate on an address:

$ sed -i '1s/^/<added text> /' file

What is this magical 1s you see on every answer here? Line addressing!.

Want to add <added text> on the first 10 lines?

$ sed -i '1,10s/^/<added text> /' file

Or you can use Command Grouping:

$ { echo -n '<added text> '; cat file; } >file.new
$ mv file{.new,}

If you want to add a line at the beginning of a file, you need to add \n at the end of the string in the best solution above.

The best solution will add the string, but with the string, it will not add a line at the end of a file.

sed -i '1s/^/your text\n/' file

If the file is only one line, you can use:

sed 's/^/insert this /' oldfile > newfile

If it's more than one line. one of:

sed '1s/^/insert this /' oldfile > newfile
sed '1,1s/^/insert this /' oldfile > newfile

I've included the latter so that you know how to do ranges of lines. Both of these "replace" the start line marker on their affected lines with the text you want to insert. You can also (assuming your sed is modern enough) use:

sed -i 'whatever command you choose' filename

to do in-place editing.

Use subshell:

echo "$(echo -n 'hello'; cat filename)" > filename

---

Unfortunately, command substitution will remove newlines at the end of file. So as to keep them one can use:

echo -n "hello" | cat - filename > /tmp/filename.tmp
mv /tmp/filename.tmp filename

Neither grouping nor command substitution is needed.

To insert just a newline:

sed '1i\\'

You can use cat -

printf '%s' "some text at the beginning" | cat - filename

To add a line to the top of the file:

sed -i '1iText to add\'

my two cents:

sed  -i '1i /path/of/file.sh' filename

This will work even is the string containing forward slash "/"

Note that on OS X, sed -i <pattern> file, fails. However, if you provide a backup extension, sed -i old <pattern> file, then file is modified in place while file.old is created. You can then delete file.old in your script.

Hi with carriage return:

sed -i '1s/^/your text\n/' file

There is a very easy way:

echo "your header" > headerFile.txt
cat yourFile >> headerFile.txt

PROBLEM: tag a file, at the top of the file, with the base name of the parent directory.

I.e., for

/mnt/Vancouver/Programming/file1

tag the top of file1 with Programming.

SOLUTION 1 -- non-empty files:

bn=${PWD##*/}    ## bn: basename

sed -i '1s/^/'"$bn"'\n/' <file>

1s places the text at line 1 of the file.

SOLUTION 2 -- empty or non-empty files:

The sed command, above, fails on empty files. Here is a solution, based on https://superuser.com/questions/246837/how-do-i-add-text-to-the-beginning-of-a-file-in-bash/246841#246841

printf "${PWD##*/}\n" | cat - <file> > temp && mv -f temp <file>

Note that the - in the cat command is required (reads standard input: see man cat for more information). Here, I believe, it's needed to take the output of the printf statement (to STDIN), and cat that and the file to temp ... See also the explanation at the bottom of http://www.linfo.org/cat.html.

I also added -f to the mv command, to avoid being asked for confirmations when overwriting files.

To recurse over a directory:

for file in *; do printf "${PWD##*/}\n" | cat - $file > temp && mv -f temp $file; done

Note also that this will break over paths with spaces; there are solutions, elsewhere (e.g. file globbing, or find . -type f ... -type solutions) for those.

ADDENDUM: Re: my last comment, this script will allow you to recurse over directories with spaces in the paths:

#!/bin/bash

## https://stackoverflow.com/questions/4638874/how-to-loop-through-a-directory-recursively-to-delete-files-with-certain-extensi

## To allow spaces in filenames,
##   at the top of the script include: IFS=$'\n'; set -f
##   at the end of the script include: unset IFS; set +f

IFS=$'\n'; set -f

# ----------------------------------------------------------------------------
# SET PATHS:

IN="/mnt/Vancouver/Programming/data/claws-test/corpus test/"

# https://superuser.com/questions/716001/how-can-i-get-files-with-numeric-names-using-ls-command

# FILES=$(find $IN -type f -regex ".*/[0-9]*")        ## recursive; numeric filenames only
FILES=$(find $IN -type f -regex ".*/[0-9 ]*")         ## recursive; numeric filenames only (may include spaces)

# echo '$FILES:'                                      ## single-quoted, (literally) prints: $FILES:
# echo "$FILES"                                       ## double-quoted, prints path/, filename (one per line)

# ----------------------------------------------------------------------------
# MAIN LOOP:

for f in $FILES
do

  # Tag top of file with basename of current dir:
  printf "[top] Tag: ${PWD##*/}\n\n" | cat - $f > temp && mv -f temp $f

  # Tag bottom of file with basename of current dir:
  printf "\n[bottom] Tag: ${PWD##*/}\n" >> $f
done

unset IFS; set +f

Just for fun, here is a solution using ed which does not have the problem of not working on an empty file. You can put it into a shell script just like any other answer to this question.

ed Test <<EOF
a

.
0i
<added text>
.
1,+1 j
$ g/^$/d
wq
EOF

The above script adds the text to insert to the first line, and then joins the first and second line. To avoid ed exiting on error with an invalid join, it first creates a blank line at the end of the file and remove it later if it still exists.

Limitations: This script does not work if <added text> is exactly equal to a single period.

echo -n "text to insert " ;tac filename.txt| tac > newfilename.txt

The first tac pipes the file backwards (last line first) so the "text to insert" appears last. The 2nd tac wraps it once again so the inserted line is at the beginning and the original file is in its original order.

Another solution with aliases. Add to your init rc/ env file:

addtail () { find . -type f ! -path "./.git/*" -exec sh -c "echo $@ >> {}" \; }
addhead () { find . -type f ! -path "./.git/*" -exec sh -c  "sed -i '1s/^/$@\n/' {}" \; }

Usage:

addtail "string to add at the beginning of file"
addtail "string to add at the end of file"

With the echo approach, if you are on macOS/BSD like me, lose the -n switch that other people suggest. And I like to define a variable for the text.

So it would be like this:

Header="my complex header that may have difficult chars \"like these quotes\" and line breaks \n\n "

{ echo "$Header"; cat "old.txt"; } > "new.txt"
mv new.txt old.txt

The simplest solution I found is:

echo -n "<text to add>" | cat - myFile.txt | tee myFile.txt

Notes:

• Remove | tee myFile.txt if you don't want to change the file contents.
• Remove the -n parameter if you want to append a full line.
• Add &> /dev/null to the end if you don't want to see the output (the generated file).
• This can be used to append a shebang to the file. Example:

  # make it executable (use u+x to allow only current user)
  chmod +x cropImage.ts
  # append the shebang
  echo '#''!'/usr/bin/env ts-node | cat - cropImage.ts | tee cropImage.ts &> /dev/null
  # execute it
  ./cropImage.ts myImage.png
  

TL;dr -

Consider using ex. Since you want the front of a given line, then the syntax is basically the same as what you might find for sed but the option of "in place editing" is built-in.

I cannot imagine an environment where you have sed but not ex/vi, unless it is a MS Windows box with some special "sed.exe", maybe.

sed & grep sort of evolved from ex / vi, so it might be better to say sed syntax is the same as ex.

You can change the line number to something besides #1 or search for a line and change that one.

source=myFile.txt
Front="This goes IN FRONT "
man true > $source
ex -s ${source} <<EOF
1s/^/$Front/ 
wq
EOF

$ head -n 3 $source
This goes IN FRONT TRUE(1)                                                    User Commands                                                    TRUE(1)

NAME

Long version, I recommend ex (or ed if you are one of the cool kids).

I like ex because it is portable, extremely powerful, allows me to write in-place, and/or make backups all without needing GNU (or even BSD) extensions.

Additionally, if you know the ex way, then you know how to do it in vi - and probably vim if that is your jam.

source=myFile.txt
topString="******  This goes on TOP  ******"
man true > "$source"
ex -s ${source} <<-EOF
0r!echo "$topString"
wq
EOF

The 0r is shorthand for :0read! or :0r! that you have likely used in vi mode (it is literally the same thing) but the : is optional here.
Writing to the special line #0 automatically pushes everything "down", and then you just :wq to save your changes.

$ head -n 5 "$source"
******  This goes on TOP  ******
TRUE(1)                                                    User Commands                                                    TRUE(1)

NAME
       true - do nothing, successfully

Also, some older sed implementations do not seem to have extensions (like \U&) that ex should have by default.