Bash: buffer entire stdin, then output - bash

I need to modify a file in-place using a program prog that doesn't support it.
prog $file > $file.temp
cat $file.temp > $file
rm $file.temp
I want to do this in a single step, without temp files. This looks good but won't work:
cat <(prog $1) > $1
It would work if I had a way of buffering the contents of a pipe (blocking until the write end closes), eg:
cat <(prog $1 | buffer_until_close) > $1
How can I do this, or achieve the desired syntax some other way?

It would work if I had a way of buffering the contents of a pipe (blocking until the write end closes), eg:
cat <(prog $1 | buffer_until_close) > $1
No, it wouldn't. The redirection of stdout (>$1) is performed before any program is started. And as soon as the shell sets up the redirection, it truncates the output file.
However, as mentioned in the comments, sponge will work:
prog $1 | sponge $1
sponge is found in the moreutils package and most linux distros will preinstall it.

Related

Infinite loop when redirecting output to the input file [duplicate]

Basically I want to take as input text from a file, remove a line from that file, and send the output back to the same file. Something along these lines if that makes it any clearer.
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name > file_name
however, when I do this I end up with a blank file.
Any thoughts?
Use sponge for this kind of tasks. Its part of moreutils.
Try this command:
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | sponge file_name
You cannot do that because bash processes the redirections first, then executes the command. So by the time grep looks at file_name, it is already empty. You can use a temporary file though.
#!/bin/sh
tmpfile=$(mktemp)
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name > ${tmpfile}
cat ${tmpfile} > file_name
rm -f ${tmpfile}
like that, consider using mktemp to create the tmpfile but note that it's not POSIX.
Use sed instead:
sed -i '/seg[0-9]\{1,\}\.[0-9]\{1\}/d' file_name
try this simple one
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | tee file_name
Your file will not be blank this time :) and your output is also printed to your terminal.
You can't use redirection operator (> or >>) to the same file, because it has a higher precedence and it will create/truncate the file before the command is even invoked. To avoid that, you should use appropriate tools such as tee, sponge, sed -i or any other tool which can write results to the file (e.g. sort file -o file).
Basically redirecting input to the same original file doesn't make sense and you should use appropriate in-place editors for that, for example Ex editor (part of Vim):
ex '+g/seg[0-9]\{1,\}\.[0-9]\{1\}/d' -scwq file_name
where:
'+cmd'/-c - run any Ex/Vim command
g/pattern/d - remove lines matching a pattern using global (help :g)
-s - silent mode (man ex)
-c wq - execute :write and :quit commands
You may use sed to achieve the same (as already shown in other answers), however in-place (-i) is non-standard FreeBSD extension (may work differently between Unix/Linux) and basically it's a stream editor, not a file editor. See: Does Ex mode have any practical use?
One liner alternative - set the content of the file as variable:
VAR=`cat file_name`; echo "$VAR"|grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' > file_name
Since this question is the top result in search engines, here's a one-liner based on https://serverfault.com/a/547331 that uses a subshell instead of sponge (which often isn't part of a vanilla install like OS X):
echo "$(grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name)" > file_name
The general case is:
echo "$(cat file_name)" > file_name
Edit, the above solution has some caveats:
printf '%s' <string> should be used instead of echo <string> so that files containing -n don't cause undesired behavior.
Command substitution strips trailing newlines (this is a bug/feature of shells like bash) so we should append a postfix character like x to the output and remove it on the outside via parameter expansion of a temporary variable like ${v%x}.
Using a temporary variable $v stomps the value of any existing variable $v in the current shell environment, so we should nest the entire expression in parentheses to preserve the previous value.
Another bug/feature of shells like bash is that command substitution strips unprintable characters like null from the output. I verified this by calling dd if=/dev/zero bs=1 count=1 >> file_name and viewing it in hex with cat file_name | xxd -p. But echo $(cat file_name) | xxd -p is stripped. So this answer should not be used on binary files or anything using unprintable characters, as Lynch pointed out.
The general solution (albiet slightly slower, more memory intensive and still stripping unprintable characters) is:
(v=$(cat file_name; printf x); printf '%s' ${v%x} > file_name)
Test from https://askubuntu.com/a/752451:
printf "hello\nworld\n" > file_uniquely_named.txt && for ((i=0; i<1000; i++)); do (v=$(cat file_uniquely_named.txt; printf x); printf '%s' ${v%x} > file_uniquely_named.txt); done; cat file_uniquely_named.txt; rm file_uniquely_named.txt
Should print:
hello
world
Whereas calling cat file_uniquely_named.txt > file_uniquely_named.txt in the current shell:
printf "hello\nworld\n" > file_uniquely_named.txt && for ((i=0; i<1000; i++)); do cat file_uniquely_named.txt > file_uniquely_named.txt; done; cat file_uniquely_named.txt; rm file_uniquely_named.txt
Prints an empty string.
I haven't tested this on large files (probably over 2 or 4 GB).
I have borrowed this answer from Hart Simha and kos.
This is very much possible, you just have to make sure that by the time you write the output, you're writing it to a different file. This can be done by removing the file after opening a file descriptor to it, but before writing to it:
exec 3<file ; rm file; COMMAND <&3 >file ; exec 3>&-
Or line by line, to understand it better :
exec 3<file # open a file descriptor reading 'file'
rm file # remove file (but fd3 will still point to the removed file)
COMMAND <&3 >file # run command, with the removed file as input
exec 3>&- # close the file descriptor
It's still a risky thing to do, because if COMMAND fails to run properly, you'll lose the file contents. That can be mitigated by restoring the file if COMMAND returns a non-zero exit code :
exec 3<file ; rm file; COMMAND <&3 >file || cat <&3 >file ; exec 3>&-
We can also define a shell function to make it easier to use :
# Usage: replace FILE COMMAND
replace() { exec 3<$1 ; rm $1; ${#:2} <&3 >$1 || cat <&3 >$1 ; exec 3>&- }
Example :
$ echo aaa > test
$ replace test tr a b
$ cat test
bbb
Also, note that this will keep a full copy of the original file (until the third file descriptor is closed). If you're using Linux, and the file you're processing on is too big to fit twice on the disk, you can check out this script that will pipe the file to the specified command block-by-block while unallocating the already processed blocks. As always, read the warnings in the usage page.
The following will accomplish the same thing that sponge does, without requiring moreutils:
shuf --output=file --random-source=/dev/zero
The --random-source=/dev/zero part tricks shuf into doing its thing without doing any shuffling at all, so it will buffer your input without altering it.
However, it is true that using a temporary file is best, for performance reasons. So, here is a function that I have written that will do that for you in a generalized way:
# Pipes a file into a command, and pipes the output of that command
# back into the same file, ensuring that the file is not truncated.
# Parameters:
# $1: the file.
# $2: the command. (With $3... being its arguments.)
# See https://stackoverflow.com/a/55655338/773113
siphon()
{
local tmp file rc=0
[ "$#" -ge 2 ] || { echo "Usage: siphon filename [command...]" >&2; return 1; }
file="$1"; shift
tmp=$(mktemp -- "$file.XXXXXX") || return
"$#" <"$file" >"$tmp" || rc=$?
mv -- "$tmp" "$file" || rc=$(( rc | $? ))
return "$rc"
}
There's also ed (as an alternative to sed -i):
# cf. http://wiki.bash-hackers.org/howto/edit-ed
printf '%s\n' H 'g/seg[0-9]\{1,\}\.[0-9]\{1\}/d' wq | ed -s file_name
You can use slurp with POSIX Awk:
!/seg[0-9]\{1,\}\.[0-9]\{1\}/ {
q = q ? q RS $0 : $0
}
END {
print q > ARGV[1]
}
Example
This does the trick pretty nicely in most of the cases I faced:
cat <<< "$(do_stuff_with f)" > f
Note that while $(…) strips trailing newlines, <<< ensures a final newline, so generally the result is magically satisfying.
(Look for “Here Strings” in man bash if you want to learn more.)
Full example:
#! /usr/bin/env bash
get_new_content() {
sed 's/Initial/Final/g' "${1:?}"
}
echo 'Initial content.' > f
cat f
cat <<< "$(get_new_content f)" > f
cat f
This does not truncate the file and yields:
Initial content.
Final content.
Note that I used a function here for the sake of clarity and extensibility, but that’s not a requirement.
A common usecase is JSON edition:
echo '{ "a": 12 }' > f
cat f
cat <<< "$(jq '.a = 24' f)" > f
cat f
This yields:
{ "a": 12 }
{
"a": 24
}
Try this
echo -e "AAA\nBBB\nCCC" > testfile
cat testfile
AAA
BBB
CCC
echo "$(grep -v 'AAA' testfile)" > testfile
cat testfile
BBB
CCC
I usually use the tee program to do this:
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | tee file_name
It creates and removes a tempfile by itself.

Editing multiple files using a bash script [duplicate]

Basically I want to take as input text from a file, remove a line from that file, and send the output back to the same file. Something along these lines if that makes it any clearer.
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name > file_name
however, when I do this I end up with a blank file.
Any thoughts?
Use sponge for this kind of tasks. Its part of moreutils.
Try this command:
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | sponge file_name
You cannot do that because bash processes the redirections first, then executes the command. So by the time grep looks at file_name, it is already empty. You can use a temporary file though.
#!/bin/sh
tmpfile=$(mktemp)
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name > ${tmpfile}
cat ${tmpfile} > file_name
rm -f ${tmpfile}
like that, consider using mktemp to create the tmpfile but note that it's not POSIX.
Use sed instead:
sed -i '/seg[0-9]\{1,\}\.[0-9]\{1\}/d' file_name
try this simple one
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | tee file_name
Your file will not be blank this time :) and your output is also printed to your terminal.
You can't use redirection operator (> or >>) to the same file, because it has a higher precedence and it will create/truncate the file before the command is even invoked. To avoid that, you should use appropriate tools such as tee, sponge, sed -i or any other tool which can write results to the file (e.g. sort file -o file).
Basically redirecting input to the same original file doesn't make sense and you should use appropriate in-place editors for that, for example Ex editor (part of Vim):
ex '+g/seg[0-9]\{1,\}\.[0-9]\{1\}/d' -scwq file_name
where:
'+cmd'/-c - run any Ex/Vim command
g/pattern/d - remove lines matching a pattern using global (help :g)
-s - silent mode (man ex)
-c wq - execute :write and :quit commands
You may use sed to achieve the same (as already shown in other answers), however in-place (-i) is non-standard FreeBSD extension (may work differently between Unix/Linux) and basically it's a stream editor, not a file editor. See: Does Ex mode have any practical use?
One liner alternative - set the content of the file as variable:
VAR=`cat file_name`; echo "$VAR"|grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' > file_name
Since this question is the top result in search engines, here's a one-liner based on https://serverfault.com/a/547331 that uses a subshell instead of sponge (which often isn't part of a vanilla install like OS X):
echo "$(grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name)" > file_name
The general case is:
echo "$(cat file_name)" > file_name
Edit, the above solution has some caveats:
printf '%s' <string> should be used instead of echo <string> so that files containing -n don't cause undesired behavior.
Command substitution strips trailing newlines (this is a bug/feature of shells like bash) so we should append a postfix character like x to the output and remove it on the outside via parameter expansion of a temporary variable like ${v%x}.
Using a temporary variable $v stomps the value of any existing variable $v in the current shell environment, so we should nest the entire expression in parentheses to preserve the previous value.
Another bug/feature of shells like bash is that command substitution strips unprintable characters like null from the output. I verified this by calling dd if=/dev/zero bs=1 count=1 >> file_name and viewing it in hex with cat file_name | xxd -p. But echo $(cat file_name) | xxd -p is stripped. So this answer should not be used on binary files or anything using unprintable characters, as Lynch pointed out.
The general solution (albiet slightly slower, more memory intensive and still stripping unprintable characters) is:
(v=$(cat file_name; printf x); printf '%s' ${v%x} > file_name)
Test from https://askubuntu.com/a/752451:
printf "hello\nworld\n" > file_uniquely_named.txt && for ((i=0; i<1000; i++)); do (v=$(cat file_uniquely_named.txt; printf x); printf '%s' ${v%x} > file_uniquely_named.txt); done; cat file_uniquely_named.txt; rm file_uniquely_named.txt
Should print:
hello
world
Whereas calling cat file_uniquely_named.txt > file_uniquely_named.txt in the current shell:
printf "hello\nworld\n" > file_uniquely_named.txt && for ((i=0; i<1000; i++)); do cat file_uniquely_named.txt > file_uniquely_named.txt; done; cat file_uniquely_named.txt; rm file_uniquely_named.txt
Prints an empty string.
I haven't tested this on large files (probably over 2 or 4 GB).
I have borrowed this answer from Hart Simha and kos.
This is very much possible, you just have to make sure that by the time you write the output, you're writing it to a different file. This can be done by removing the file after opening a file descriptor to it, but before writing to it:
exec 3<file ; rm file; COMMAND <&3 >file ; exec 3>&-
Or line by line, to understand it better :
exec 3<file # open a file descriptor reading 'file'
rm file # remove file (but fd3 will still point to the removed file)
COMMAND <&3 >file # run command, with the removed file as input
exec 3>&- # close the file descriptor
It's still a risky thing to do, because if COMMAND fails to run properly, you'll lose the file contents. That can be mitigated by restoring the file if COMMAND returns a non-zero exit code :
exec 3<file ; rm file; COMMAND <&3 >file || cat <&3 >file ; exec 3>&-
We can also define a shell function to make it easier to use :
# Usage: replace FILE COMMAND
replace() { exec 3<$1 ; rm $1; ${#:2} <&3 >$1 || cat <&3 >$1 ; exec 3>&- }
Example :
$ echo aaa > test
$ replace test tr a b
$ cat test
bbb
Also, note that this will keep a full copy of the original file (until the third file descriptor is closed). If you're using Linux, and the file you're processing on is too big to fit twice on the disk, you can check out this script that will pipe the file to the specified command block-by-block while unallocating the already processed blocks. As always, read the warnings in the usage page.
The following will accomplish the same thing that sponge does, without requiring moreutils:
shuf --output=file --random-source=/dev/zero
The --random-source=/dev/zero part tricks shuf into doing its thing without doing any shuffling at all, so it will buffer your input without altering it.
However, it is true that using a temporary file is best, for performance reasons. So, here is a function that I have written that will do that for you in a generalized way:
# Pipes a file into a command, and pipes the output of that command
# back into the same file, ensuring that the file is not truncated.
# Parameters:
# $1: the file.
# $2: the command. (With $3... being its arguments.)
# See https://stackoverflow.com/a/55655338/773113
siphon()
{
local tmp file rc=0
[ "$#" -ge 2 ] || { echo "Usage: siphon filename [command...]" >&2; return 1; }
file="$1"; shift
tmp=$(mktemp -- "$file.XXXXXX") || return
"$#" <"$file" >"$tmp" || rc=$?
mv -- "$tmp" "$file" || rc=$(( rc | $? ))
return "$rc"
}
There's also ed (as an alternative to sed -i):
# cf. http://wiki.bash-hackers.org/howto/edit-ed
printf '%s\n' H 'g/seg[0-9]\{1,\}\.[0-9]\{1\}/d' wq | ed -s file_name
You can use slurp with POSIX Awk:
!/seg[0-9]\{1,\}\.[0-9]\{1\}/ {
q = q ? q RS $0 : $0
}
END {
print q > ARGV[1]
}
Example
This does the trick pretty nicely in most of the cases I faced:
cat <<< "$(do_stuff_with f)" > f
Note that while $(…) strips trailing newlines, <<< ensures a final newline, so generally the result is magically satisfying.
(Look for “Here Strings” in man bash if you want to learn more.)
Full example:
#! /usr/bin/env bash
get_new_content() {
sed 's/Initial/Final/g' "${1:?}"
}
echo 'Initial content.' > f
cat f
cat <<< "$(get_new_content f)" > f
cat f
This does not truncate the file and yields:
Initial content.
Final content.
Note that I used a function here for the sake of clarity and extensibility, but that’s not a requirement.
A common usecase is JSON edition:
echo '{ "a": 12 }' > f
cat f
cat <<< "$(jq '.a = 24' f)" > f
cat f
This yields:
{ "a": 12 }
{
"a": 24
}
Try this
echo -e "AAA\nBBB\nCCC" > testfile
cat testfile
AAA
BBB
CCC
echo "$(grep -v 'AAA' testfile)" > testfile
cat testfile
BBB
CCC
I usually use the tee program to do this:
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | tee file_name
It creates and removes a tempfile by itself.

Can envsubst not do in-place substitution?

I have a config file which contains some ENV_VARIABLE styled variables.
This is my file.
It might contain $EXAMPLES of text.
Now I want that variable replaced with a value which is saved in my actual environment variables. So I'm trying this:
export EXAMPLES=lots
envsubst < file.txt > file.txt
But it doesn't work when the input file and output file are identical. The result is an empty file of size 0.
There must be a good reason for this, some bash basics that I'm not aware of?
How do I achieve what I want to do, ideally without first outputting to a different file and then replacing the original file with it?
I know that I can do it easily enough with sed, but when I discovered the envsubst command I thought that it should be perfect for my use case, so I'd like to use that.
Here is the solution that I use:
originalfile="file.txt"
tmpfile=$(mktemp)
cp --attributes-only --preserve $originalfile $tmpfile
cat $originalfile | envsubst > $tmpfile && mv $tmpfile $originalfile
Be careful with other solutions that do not use a temporary file. Pipes are asynchronous, so the file will occasionally be read after it has already been truncated.
Redirects are handled by the shell, not the program being executed, and they are set up before the program is invoked.
The redirect >output.file has the effect of creating output.file if it doesn't exist and emptying it if it does. Either way, you end up with an empty file, and that is what the program's output is redirected to.
Programs like sed which are capable of "in-place" modification must take the filename as a command-line argument, not as a redirect.
In your case, I would suggest using a temporary file and then renaming it if all goes OK.
envsubst < file.txt | tee file.txt
I found another shortcut to put into temp file and then rename it to original file.
envsubst < in.txt > out.txt && mv out.txt in.txt
To avoid creating a temporary file, use sponge not tee:
envsubst < file.txt | sponge file.txt
From https://linux.die.net/man/1/sponge:
sponge reads standard input and writes it out to the specified file. Unlike a shell redirect, sponge soaks up all its input before opening the output file. This allows constricting pipelines that read from and write to the same file.
You can achieve in-place substitution by calling envsubst from gnu sed with the "e" command:
EXAMPLES=lots sed -i 's/.*/echo & | envsubst/e' file.txt
It's worth noting that the mv solution won't maintain file permissions. Using cp -pf would be preferable in the case that you're modifying an executable file.
tmpfile=$(mktemp)
cat file.txt | envsubst > "$tmpfile" && cp -pf "$tmpfile" file.txt
rm -f "$tmpfile"
This answer was framed from two other answers. I guess this is the best solution.
originalFile=file.txt
tmpfile=$(mktemp)
cat $originalFile | envsubst > "$tmpfile" && cp -pf "$tmpfile" $originalFile
rm -f "$tmpfile"
Updated 20221011 - Using 1 sed command
sed -i -r 's/["`]|\$\(/\\&/g; s/.*/echo "&"/ e' ./input.txt
Updated 20221007 - Using 2 sed commands
sed -i -r 's/["`]|\$\(/\\&/g' input.txt
sed -i -r 's/.*/echo "&"/ e' input.txt
Do it without envsubst
envsubst_file () {
local original_file=$1
local temp_file=$(mktemp)
trap "rm -f ${temp_file}" 0 2 3 15
cp -p ${original_file} ${temp_file}
cat ${original_file} | sed -r 's/["`]|\$\(/\\&/g' | sed -r 's/.*/echo "&"/g' | sh > ${temp_file}
mv ${temp_file} ${original_file}
}
envsubst_file 'input.txt'
First using sed to escapes double quotes("), backtick(`) and command $( by prefixing with backslash(\),then using sed again replace with
echo "&"
Finally executing the shell script and redirecting to ${temp_file}
If you use bash, check this:
a=`<file.txt` && envsubst <<<"$a" >file.txt
Tested on 500mb file, works as expected.
In the end I found that using envsubst was too dangerous after all. My files might contain dollar signs in places where I don't want any substitution to happen, and envsubst will just replace them with empty strings if no corresponding environment variable is defined. Not cool.

How to process lines which is read from standard input in UNIX shell script?

I get stuck by this problem:
I wrote a shell script and it gets a large file with many lines from stdin, that's how it is executed:
./script < filename
I want use the file as an input to another operation in the script, however I don't know how to store this file's name in a variable.
It is a script that takes a file from stdin as argument and then do awk operation in this file it self. Say if I write in script:
script:
#!/bin/sh
...
read file
...
awk '...' < "$file"
...
it only reads first line of the input file.
And I find a way to write like this:
Min=-1
while read line; do
n=$(echo $line | awk -F$delim '{print NF}')
if [ $Min -eq -1 ] || [ $n -lt $Min ];then
Min=$n
fi
done
it would take very very long time to wait for processing, it seems awk takes much time.
So how to improve this?
/dev/stdin can be quite useful here.
In fact, it's just a chain of links to your input.
So, writing cat /dev/stdin will give you all input from your file and you can deny using input filename at all.
Now answer to question :) Recursively read links, beginning at /dev/stdin, and you will get filename. Bash code:
r(){
l=`readlink $1`
if [ $? -ne 0 ]
then
echo $1
else
r $l
fi
}
filename=`r /dev/stdin`
echo $filename
UPD:
in Ubuntu I found an option -f to readlink. i.e. readlink -f /dev/stdin gives the same output. This option may absent in some systems.
UPD2:tests (test.sh is code above):
$ ./test.sh <input # that is a file
/home/sfedorov/input
$ ./test.sh <<EOF
> line
> EOF
/tmp/sh-thd-214216298213
$ echo 1 | ./test.sh
pipe:[91219]
$ readlink -f /dev/stdin < input
/home/sfedorov/input
$ readlink -f /dev/stdin << EOF
> line
> EOF
/tmp/sh-thd-3423766239895 (deleted)
$ echo 1 | readlink -f /dev/stdin
/proc/18489/fd/pipe:[92382]
You're overdoing this. The way you invoke your script:
the file contents are the script's standard input
the script receives no argument
But awk already takes input from stdin by default, so all you need to do to make this work is:
not give awk any file name argument, it's going to be the wrapping shell's stdin automatically
not consume any of that input before the wrapping script reaches the awk part. Specifically: no read
If that's all there is to your script, it reduces to the awk invocation, so you might consider doing away with it altogether and just call awk directly. Or make your script directly an awk one instead of a sh one.
Aside: the reason your while read line/multiple awk variant (the one in the question) is slow is because it spawns an awk process for each and every line of the input, and process spawning is order of magnitudes slower than awk processing a single line. The reason why the generate tmpfile/single awk variant (the one in your answer) is still a bit slow is because it's generating the tmpfile line by line, reopening to append every time.
Modify your script to that it takes the input file name as an argument, then read from the file in your script:
$ ./script filename
In script:
filename=$1
awk '...' < "$filename"
If your script just reads from standard input, there is no guarantee that there is a named file providing the input; it could just as easily be reading from a pipe or a network socket.
How about invoking the script differently pipe standard output of YourFilename into
your scriptName as follows (the standard output of the cat filename now becomes standard
input to you script, actually in this case to the awk command
For I have filename Names.data and script showNames.sh execute as follows
cat Names.data | ./showNames.sh
Contents of filename Names.data
Huckleberry Finn
Jack Spratt
Humpty Dumpty
Contents of scrip;t showNames.sh
#!/bin/bash
#whatever awk commands you need
awk "{ print }"
Well I finally find this way to solve my problem, although it will take several seconds.
grep '.*' >> /tmp/tmpfile
Min=$(awk -F$delim 'NF < min || min == "" { min = NF };END {printmin}'</tmp/tmpfile)
Just append each line into a temporary file so that after reading from stdin, the tmpfile is the same as input file.

How can I use a file in a command and redirect output to the same file without truncating it?

Basically I want to take as input text from a file, remove a line from that file, and send the output back to the same file. Something along these lines if that makes it any clearer.
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name > file_name
however, when I do this I end up with a blank file.
Any thoughts?
Use sponge for this kind of tasks. Its part of moreutils.
Try this command:
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | sponge file_name
You cannot do that because bash processes the redirections first, then executes the command. So by the time grep looks at file_name, it is already empty. You can use a temporary file though.
#!/bin/sh
tmpfile=$(mktemp)
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name > ${tmpfile}
cat ${tmpfile} > file_name
rm -f ${tmpfile}
like that, consider using mktemp to create the tmpfile but note that it's not POSIX.
Use sed instead:
sed -i '/seg[0-9]\{1,\}\.[0-9]\{1\}/d' file_name
try this simple one
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | tee file_name
Your file will not be blank this time :) and your output is also printed to your terminal.
You can't use redirection operator (> or >>) to the same file, because it has a higher precedence and it will create/truncate the file before the command is even invoked. To avoid that, you should use appropriate tools such as tee, sponge, sed -i or any other tool which can write results to the file (e.g. sort file -o file).
Basically redirecting input to the same original file doesn't make sense and you should use appropriate in-place editors for that, for example Ex editor (part of Vim):
ex '+g/seg[0-9]\{1,\}\.[0-9]\{1\}/d' -scwq file_name
where:
'+cmd'/-c - run any Ex/Vim command
g/pattern/d - remove lines matching a pattern using global (help :g)
-s - silent mode (man ex)
-c wq - execute :write and :quit commands
You may use sed to achieve the same (as already shown in other answers), however in-place (-i) is non-standard FreeBSD extension (may work differently between Unix/Linux) and basically it's a stream editor, not a file editor. See: Does Ex mode have any practical use?
One liner alternative - set the content of the file as variable:
VAR=`cat file_name`; echo "$VAR"|grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' > file_name
Since this question is the top result in search engines, here's a one-liner based on https://serverfault.com/a/547331 that uses a subshell instead of sponge (which often isn't part of a vanilla install like OS X):
echo "$(grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name)" > file_name
The general case is:
echo "$(cat file_name)" > file_name
Edit, the above solution has some caveats:
printf '%s' <string> should be used instead of echo <string> so that files containing -n don't cause undesired behavior.
Command substitution strips trailing newlines (this is a bug/feature of shells like bash) so we should append a postfix character like x to the output and remove it on the outside via parameter expansion of a temporary variable like ${v%x}.
Using a temporary variable $v stomps the value of any existing variable $v in the current shell environment, so we should nest the entire expression in parentheses to preserve the previous value.
Another bug/feature of shells like bash is that command substitution strips unprintable characters like null from the output. I verified this by calling dd if=/dev/zero bs=1 count=1 >> file_name and viewing it in hex with cat file_name | xxd -p. But echo $(cat file_name) | xxd -p is stripped. So this answer should not be used on binary files or anything using unprintable characters, as Lynch pointed out.
The general solution (albiet slightly slower, more memory intensive and still stripping unprintable characters) is:
(v=$(cat file_name; printf x); printf '%s' ${v%x} > file_name)
Test from https://askubuntu.com/a/752451:
printf "hello\nworld\n" > file_uniquely_named.txt && for ((i=0; i<1000; i++)); do (v=$(cat file_uniquely_named.txt; printf x); printf '%s' ${v%x} > file_uniquely_named.txt); done; cat file_uniquely_named.txt; rm file_uniquely_named.txt
Should print:
hello
world
Whereas calling cat file_uniquely_named.txt > file_uniquely_named.txt in the current shell:
printf "hello\nworld\n" > file_uniquely_named.txt && for ((i=0; i<1000; i++)); do cat file_uniquely_named.txt > file_uniquely_named.txt; done; cat file_uniquely_named.txt; rm file_uniquely_named.txt
Prints an empty string.
I haven't tested this on large files (probably over 2 or 4 GB).
I have borrowed this answer from Hart Simha and kos.
This is very much possible, you just have to make sure that by the time you write the output, you're writing it to a different file. This can be done by removing the file after opening a file descriptor to it, but before writing to it:
exec 3<file ; rm file; COMMAND <&3 >file ; exec 3>&-
Or line by line, to understand it better :
exec 3<file # open a file descriptor reading 'file'
rm file # remove file (but fd3 will still point to the removed file)
COMMAND <&3 >file # run command, with the removed file as input
exec 3>&- # close the file descriptor
It's still a risky thing to do, because if COMMAND fails to run properly, you'll lose the file contents. That can be mitigated by restoring the file if COMMAND returns a non-zero exit code :
exec 3<file ; rm file; COMMAND <&3 >file || cat <&3 >file ; exec 3>&-
We can also define a shell function to make it easier to use :
# Usage: replace FILE COMMAND
replace() { exec 3<$1 ; rm $1; ${#:2} <&3 >$1 || cat <&3 >$1 ; exec 3>&- }
Example :
$ echo aaa > test
$ replace test tr a b
$ cat test
bbb
Also, note that this will keep a full copy of the original file (until the third file descriptor is closed). If you're using Linux, and the file you're processing on is too big to fit twice on the disk, you can check out this script that will pipe the file to the specified command block-by-block while unallocating the already processed blocks. As always, read the warnings in the usage page.
The following will accomplish the same thing that sponge does, without requiring moreutils:
shuf --output=file --random-source=/dev/zero
The --random-source=/dev/zero part tricks shuf into doing its thing without doing any shuffling at all, so it will buffer your input without altering it.
However, it is true that using a temporary file is best, for performance reasons. So, here is a function that I have written that will do that for you in a generalized way:
# Pipes a file into a command, and pipes the output of that command
# back into the same file, ensuring that the file is not truncated.
# Parameters:
# $1: the file.
# $2: the command. (With $3... being its arguments.)
# See https://stackoverflow.com/a/55655338/773113
siphon()
{
local tmp file rc=0
[ "$#" -ge 2 ] || { echo "Usage: siphon filename [command...]" >&2; return 1; }
file="$1"; shift
tmp=$(mktemp -- "$file.XXXXXX") || return
"$#" <"$file" >"$tmp" || rc=$?
mv -- "$tmp" "$file" || rc=$(( rc | $? ))
return "$rc"
}
There's also ed (as an alternative to sed -i):
# cf. http://wiki.bash-hackers.org/howto/edit-ed
printf '%s\n' H 'g/seg[0-9]\{1,\}\.[0-9]\{1\}/d' wq | ed -s file_name
You can use slurp with POSIX Awk:
!/seg[0-9]\{1,\}\.[0-9]\{1\}/ {
q = q ? q RS $0 : $0
}
END {
print q > ARGV[1]
}
Example
This does the trick pretty nicely in most of the cases I faced:
cat <<< "$(do_stuff_with f)" > f
Note that while $(…) strips trailing newlines, <<< ensures a final newline, so generally the result is magically satisfying.
(Look for “Here Strings” in man bash if you want to learn more.)
Full example:
#! /usr/bin/env bash
get_new_content() {
sed 's/Initial/Final/g' "${1:?}"
}
echo 'Initial content.' > f
cat f
cat <<< "$(get_new_content f)" > f
cat f
This does not truncate the file and yields:
Initial content.
Final content.
Note that I used a function here for the sake of clarity and extensibility, but that’s not a requirement.
A common usecase is JSON edition:
echo '{ "a": 12 }' > f
cat f
cat <<< "$(jq '.a = 24' f)" > f
cat f
This yields:
{ "a": 12 }
{
"a": 24
}
Try this
echo -e "AAA\nBBB\nCCC" > testfile
cat testfile
AAA
BBB
CCC
echo "$(grep -v 'AAA' testfile)" > testfile
cat testfile
BBB
CCC
I usually use the tee program to do this:
grep -v 'seg[0-9]\{1,\}\.[0-9]\{1\}' file_name | tee file_name
It creates and removes a tempfile by itself.

Resources