'^M' character at end of lines

When I run a particular SQL script in Unix environments, I'm am seeing a '^M' character at the end of each line of the SQL script as it is echoed to the command-line. I don't know on which OS the SQL script was originally created.

What is causing this and how do I fix it?

It's caused by the DOS/Windows line-ending characters. Like Andy Whitfield said, the Unix command dos2unix will help fix the problem. If you want more information, you can read the man pages for that command.

Fix line endings in vi by running the following:

:set fileformat=unix

:w

The cause is the difference between how a Windows-based based OS and a Unix based OS store the end-of-line markers.

Windows based operating systems, thanks to their DOS heritage, store an end-of-line as a pair of characters - 0x0D0A (carriage return + line feed). Unix-based operating systems just use 0x0A (a line feed). The ^M you're seeing is a visual representation of 0x0D (a carriage return).

dos2unix will help with this. You probably also need to adjust the source of the scripts to be 'Unix-friendly'.

The easiest way is to use vi. I know that sounds terrible but its simple and already installed on most UNIX environments. The ^M is a new line from Windows/DOS environment.

from the command prompt: $ vi filename

Then press ":" to get to command mode.

Search and Replace all Globally is :%s/^M//g "Press and hold control then press V then M" which will replace ^M with nothing.

Then to write and quit enter ":wq" Done!

Try using dos2unix to strip off the ^M.

In vi, do a :%s/^M//g

To get the ^M hold the CTRL key, press V then M (Both while holding the control key) and the ^M will appear. This will find all occurrences and replace them with nothing.

The SQL script was originally created on a Windows OS. The '^M' characters are a result of Windows and Unix having different ideas about what to use for an end-of-line character. You can use perl at the command line to fix this.

perl -pie 's/\r//g' filename.txt

The ^M is typically caused by the Windows operator newlines, and translated onto Unix looks like a ^M. The command dos2unix should remove them nicely

dos2unix [options] [-c convmode] [-o file ...] [-n infile outfile ...]

C:\tmp\text>dos2unix hello.txt helloUNIX.txt

Sed is even more widely available and can do this kind of thing also if dos2unix is not installed

C:\tmp\text>sed s/\r// hello.txt > helloUNIX.txt  

You could also try tr:

cat hello.txt | tr -d \r > helloUNIX2.txt  

Here are the results:

C:\tmp\text>dumphex hello.txt  
00000000h: 48 61 68 61 0D 0A 68 61 68 61 0D 0A 68 61 68 61 Haha..haha..haha  
00000010h: 0D 0A 0D 0A 68 61 68 61 0D 0A                   ....haha..  

C:\tmp\text>dumphex helloUNIX.txt  
00000000h: 48 61 68 61 0A 68 61 68 61 0A 68 61 68 61 0A 0A Haha.haha.haha..  
00000010h: 68 61 68 61 0A                                  haha.  

C:\tmp\text>dumphex helloUNIX2.txt  
00000000h: 48 61 68 61 0A 68 61 68 61 0A 68 61 68 61 0A 0A Haha.haha.haha..  
00000010h: 68 61 68 61 0A                                  haha.  

An alternative to dos2unix command would be using standard utilities like sed.

For example, dos to unix:

sed 's/\r$//' dos.txt > unix.txt

unix to dos:

sed 's/$/\r/' unix.txt > dos.txt

To replace ^M characters in vi editor use below

open the text file say t1.txt

vi t1.txt

Enter command mode by pressing shift + :

then press keys as mentioned %s/^M/\r/g

in above ^M is not (shift + 6)M instead it is (ctrl + V)(ctrl + M)

You can remove ^M from the files directly via sed command, e.g.:

sed -i'.bak' s/\r//g *.*

If you're happy with the changes, remove the .bak files:

rm -v *.bak

Convert DOS/Windows (\r\n) line endings to Unix (\n) line endings, with tr:

tr '\r\n' '\n' < dosFile.txt > unixFile.txt

Post about replacing newlines from the Unix command line

od -a $file is useful to explore those types of question on Linux (similar to dumphex in the above).

In Perl, if you don't want to set the $/ variable and use chomp() you can also do:

$var =~ /\r\n//g;

My two cents

As already explained, Windows programs like to terminate lines with CRLF, i.e. \r\n instead of the Unix/Linux standard \n. Since I don't need all the features of dos2unix I replaced it by adding the following to my ~/.bashrc which removes the \r:

function win2unix() {
    tmp=$(mktemp) && tr -d '\r' < $1 > $tmp && mv $tmp $1
}

Now when I want to get rid of those ^M characters created e.g. when I export a CSV file from Excel or Calc, I can just do something like:

win2unix filename.csv

You could also use sed or something else, of course. By the way, I use cat -e $filename to visualize the ^M endings.

Another vi command that'll do: :%s/.$// This removes the last character of each line in the file. The drawback to this search and replace command is that it doesn't care what the last character is, so be careful not to call it twice.