How to remove new lines from files or pipe streams under Linux? This post contains simple examples that show how to use common Linux shell tools such as tr, awk/gawk, perl, sed and many others to delete new line characters. C and C++ source codes are also provided. They can be compiled into a binary tool that removes new lines. To get started, here is an example text file: days.txt.
Lets have a look at its content by running the following command from shell.
cat days.txt ©2007-2008 dsplabs.com.au
The output is shown below.
Mon Tue Wed Thu Fri Sat Sun
If the new lines were removed from days.txt then its content would look like this:
MonTueWedThuFriSatSun
If instead of simply removing the new line characters they were replaced with spaces then the output would look as follows.
Mon Tue Wed Thu Fri Sat Sun
Below are simple examples of two types: one to remove new lines and the other to replace them with spaces. The outputs of these examples are as shown above.
Either of the following commands can be used to delete new lines using the tr utility.
tr -d '\n' < days.txt cat days.txt | tr -d '\n'
While the new line characters can be replaced with spaces using the tr program as follows.
tr '\n' ' ' < days.txt cat days.txt | tr '\n' ' '
Either of the following commands can be used to delete new lines using awk or gawk.
awk '{ printf "%s", $0 }' days.txt cat days.txt | awk '{ printf "%s", $0 }'
While the new line characters can be replaced with spaces using either of the following commands.
awk '{ printf "%s ", $0 }' days.txt cat days.txt | awk '{ printf "%s ", $0 }'
Please be careful when using printf's: never pass any unvalidated inputs to printf as a format string as it is a security risk. That is, use: { printf "%s ", $0 } but not: { printf $0 }.
Either of the following commands can be used to delete new lines using perl.
perl -e 'while (<>) { chomp; print; }; exit;' days.txt cat days.txt | perl -e 'while (<>) { chomp; print; }; exit;'
While the new line characters can be replaced with spaces using either of the following commands.
perl -e 'while (<>) { chomp; print; print " "; }; exit;' days.txt cat days.txt | perl -e 'while (<>) { chomp; print; print " "; }; exit;'
Here are some more perl examples:
perl -p -e 's/\s+$/ /g' days.txt cat days.txt | perl -p -e 's/\s+$/ /g'
The above regex approach is actually much simpler and neater!
You could also use sed to remove new lines. The solution using sed is not very readable but it works. Use either of the following commands.
sed ':a;N;$!ba;s/\n//g' days.txt cat days.txt | sed ':a;N;$!ba;s/\n//g'
Or, to replace the new line characters with spaces use either of the following commands.
sed ':a;N;$!ba;s/\n/ /g' days.txt cat days.txt | sed ':a;N;$!ba;s/\n/ /g'
James of crypto.dsplabs.com.au suggested this, easier to read, sed solution on the Linux Blog Forums:
sed '{:q;N;s/\n//g;t q}' days.txt
If you fancy none of the above approaches then create your own binary tool for this task using for example C or C++. Here is an example C source code: rmnl.c also courtesy of James.
#includeint main(int argc, char *argv[]) { int k; // using int here and not char since EOF is an int while((k = getchar()) != EOF) if(k != 0x0D && k != 0x0A) putchar(k); // else putchar(' '); // replace newlines with spaces putchar(0x0A); return 0; }
While my attempt at writing a new line removal tool in C++, rmnl.cpp, is listed below.
#includeusing namespace std; int main() { string line; while( getline(cin, line) && !cin.eof() ) cout << line; // delete newlines // while( getline(cin, line) && !cin.eof() ) cout << line << ' '; // replace n with spaces cout << endl; return 0; }
Another way in C++ is to use stream iterators like so:
#include#include int main () { std::remove_copy( std::istream_iterator ( std::cin ), std::istream_iterator (), std::ostream_iterator ( std::cout ), '\n' ); }
To compile the above sources with GNU gcc or g++ compiler use the following commands, respectively.
gcc rmnl.c -o rmnl g++ rmnl.cpp -o rmnl
Either of the following commands can be used to run the newly created rmnl binary assuming that rmnl is in your path.
cat days.txt | rmnl rmnl < days.txt
If rmnl binary in not in your path variable, but it is in your current directory, then replace rmnl above with ./rmnl like so:
cat days.txt | ./rmnl ./rmnl < days.txt
Here is the compiled binary: rmnl and some information about it:
$>file rmnl rmnl: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.9, not stripped $>ldd rmnl linux-gate.so.1 => (0x0041d000) libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x00b21000) libm.so.6 => /lib/libm.so.6 (0x0086c000) libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x009ec000) libc.so.6 => /lib/libc.so.6 (0x0072a000) /lib/ld-linux.so.2 (0x0070d000)
Replacing new lines with spaces using xargs:
xargs echo < days.txt cat days.txt | xargs echo cat days.txt | xargs echo -n
Removing new lines using Haskell, f.e. with the Glasgow Haskell Compiler, ghci:
cat days.txt | ghci -e "interact (concat . lines)"
Removing new lines or substituting them with spaces using sam text editor:
printf ", x/.*/nq" | sam -d days.txt 2>/dev/null (echo ', x/\n/ c/ /'; echo ', p'; echo q) | sam -d days.txt 2>/dev/null
Or its stream version ssam:
ssam -n ', x/.*/ p' < days.txt 2>/dev/null ssam ', x/\n/ c/ /' < days.txt 2>/dev/null
Another approach is to use GNU Bash like so:
while read; do echo -n "$REPLY "; done < days.txt
Or even simpler:
echo -n `cat days.txt`
Or, Python like this:
python -c 'import sys; print sys.stdin.read().replace("\n", " ")' < days.txt