dos2unix [options] [FILE ...] [-n INFILE OUTFILE ...] unix2dos [options] [FILE ...] [-n INFILE OUTFILE ...]
In DOS/Windows text files a line break, also known as newline, is a combination of two characters: a Carriage Return (CR) followed by a Line Feed (LF). In Unix text files a line break is a single character: the Line Feed (LF). In Mac text files, prior to Mac OS X, a line break was single Carriage Return (CR) character. Nowadays Mac OS uses Unix style (LF) line breaks.
Besides line breaks Dos2unix can also convert the encoding of files. A few DOS code pages can be converted to Unix Latin-1. And Windows Unicode (UTF-16) files can be converted to Unix Unicode (UTF-8) files.
Binary files are automatically skipped, unless conversion is forced.
Non-regular files, such as directories and FIFOs, are automatically skipped.
Symbolic links and their targets are by default kept untouched. Symbolic links can optionally be replaced, or the output can be written to the symbolic link target. Writing to a symbolic link target is not supported on Windows.
Dos2unix was modelled after dos2unix under SunOS/Solaris. There is one
important difference with the original SunOS/Solaris version. This version does
by default in-place conversion (old file mode), while the original
SunOS/Solaris version only supports paired conversion (new file mode). See also
options
-o
and
-n
. Another difference is that the SunOS/Solaris version
uses by default iso mode conversion while this version uses by default ascii
mode conversion.
dos2unix -- -foo
Or in new file mode:
dos2unix -n -- -foo out.txt
When this option is used, the conversion will not be aborted when the user
and/or group ownership of the original file can't be preserved in old file
mode. Conversion will continue and the converted file will get the same new
ownership as if it was converted in new file mode. See also options
-o
and
-n
. This option is only available if dos2unix has support for preserving
the user and group ownership of files.
This option is only available in dos2unix for Windows with Unicode file name support. This option has no effect on the actual file names read and written, only on how they are displayed.
There are several methods for displaying text in a Windows console based on the encoding of the text. They all have their own advantages and disadvantages.
The disadvantage of ansi is that international file names with characters not inside the system default code page are not displayed properly. You will see a question mark, or a wrong symbol instead. When you don't work with foreign file names this method is OK.
When you use the ConEmu console all text is displayed properly, because ConEmu automatically selects a good font.
The disadvantage of unicode is that it is not compatible with ASCII. The output is not easy to handle when you redirect it to another program.
When method
unicodebom
is used the Unicode text will be preceded with a BOM
(Byte Order Mark). A BOM is required for correct redirection or piping in
PowerShell.
The disadvantage is that when you use the default raster font all non-ASCII characters are displayed wrong. Not only unicode file names, but also translated messages become unreadable. On Windows configured for an East-Asian region you may see a lot of flickering of the console when the messages are displayed.
In a ConEmu console the utf8 encoding method works well.
When method
utf8bom
is used the UTF-8 text will be preceded with a BOM
(Byte Order Mark). A BOM is required for correct redirection or piping in
PowerShell.
The default encoding can be changed with environment variable DOS2UNIX_DISPLAY_ENC
by setting it to
unicode
,
unicodebom
,
utf8
, or
utf8bom
.
The following information is printed, in this order: number of DOS line breaks, number of Unix line breaks, number of Mac line breaks, byte order mark, text or binary, file name.
Example output:
6 0 0 no_bom text dos.txt 0 6 0 no_bom text unix.txt 0 0 6 no_bom text mac.txt 6 6 6 no_bom text mixed.txt 50 0 0 UTF-16LE text utf16le.txt 0 50 0 no_bom text utf8unix.txt 50 0 0 UTF-8 text utf8dos.txt 2 418 219 no_bom binary dos2unix.exe
Note that sometimes a binary file can be mistaken for a text file. See also option
-s
.
Optionally extra flags can be set to change the output. One or more flags can be added.
With the
c
flag dos2unix will print only the files that contain DOS line breaks,
unix2dos will print only file names that have Unix line breaks.
Examples:
Show information for all *.txt files:
dos2unix -i *.txt
Show only the number of DOS line breaks and Unix line breaks:
dos2unix -idu *.txt
Show only the byte order mark:
dos2unix --info=b *.txt
List the files that have DOS line breaks:
dos2unix -ic *.txt
List the files that have Unix line breaks:
unix2dos -ic *.txt
Convert only files that have DOS line breaks and leave the other files untouched:
dos2unix -ic0 *.txt | xargs -0 dos2unix
Find text files that have DOS line breaks:
find -name '*.txt' -print0 | xargs -0 dos2unix -ic
dos2unix: Only DOS line breaks are changed to two Unix line breaks. In Mac mode only Mac line breaks are changed to two Unix line breaks.
unix2dos: Only Unix line breaks are changed to two DOS line breaks. In Mac mode Unix line breaks are changed to two Mac line breaks.
When the input file is UTF-16, and the option
-u
is used, an UTF-16
BOM will be written.
Never use this option when the output encoding is other than UTF-8, UTF-16, or GB18030. See also section UNICODE.
The person who starts the conversion in new file (paired) mode will be the owner of the converted file. The read/write permissions of the new file will be the permissions of the original file minus the umask(1) of the person who runs the conversion.
Abort conversion when the user and/or group ownership of the original file
can't be preserved in old file mode. See also options
-o
and
-n
. This
option is only available if dos2unix has support for preserving the user
and group ownership of files.
In old file (in-place) mode the converted file gets the same owner, group, and read/write permissions as the original file. Also when the file is converted by another user who has write permissions on the file (e.g. user root). The conversion will be aborted when it is not possible to preserve the original values. Change of owner could mean that the original owner is not able to read the file any more. Change of group could be a security risk, the file could be made readable for persons for whom it is not intended. Preservation of owner, group, and read/write permissions is only supported on Unix.
To check if dos2unix has support for preserving the user and group ownership of
files type
dos2unix -V
.
Conversion is always done via a temporary file. When an error occurs halfway
the conversion, the temporary file is deleted and the original file stays
intact. When the conversion is successful, the original file is replaced with
the temporary file. You may have write permission on the original file, but no
permission to put the same user and/or group ownership properties on the
temporary file as the original file has. This means you are not able to
preserve the user and/or group ownership of the original file. In this case you
can use option
--allow-chown
to continue with the conversion:
dos2unix --allow-chown foo.txt
Another option is to use new file mode:
dos2unix -n foo.txt foo.txt
The advantage of the
--allow-chown
option is that you can use wildcards,
and the ownership properties will be preserved when possible.
The skipping of binary files is done to avoid accidental mistakes. Be aware that the detection of binary files is not 100% foolproof. Input files are scanned for binary symbols which are typically not found in text files. It is possible that a binary file contains only normal text characters. Such a binary file will mistakenly be seen as a text file.
When there is a Byte Order Mark in the input file the BOM has priority over this option.
When you made a wrong assumption (the input file was not in UTF-16LE format) and the conversion succeeded, you will get an UTF-8 output file with wrong text. You can undo the wrong conversion with iconv(1) by converting the UTF-8 output file back to UTF-16LE. This will bring back the original file.
The assumption of UTF-16LE works as a conversion mode. By switching to the default ascii mode the UTF-16LE assumption is turned off.
This option works the same as option
-ul
.
In Mac mode line breaks are converted from Mac to Unix and vice versa. DOS line breaks are not changed.
To run in Mac mode use the command-line option
-c mac
or use the
commands
mac2unix
or
unix2mac
.
Although the name of this mode is ASCII, which is a 7 bit standard, the actual mode is 8 bit. Use always this mode when converting Unicode UTF-8 files.
When only option
-iso
is used dos2unix will try to determine the active code
page. When this is not possible dos2unix will use default code page CP437,
which is mainly used in the USA. To force a specific code page use options
-437
(US),
-850
(Western European),
-860
(Portuguese),
-863
(French
Canadian), or
-865
(Nordic). Windows code page CP1252 (Western European) is
also supported with option
-1252
. For other code pages use dos2unix in
combination with iconv(1). Iconv can convert between a long list of character
encodings.
Never use ISO conversion on Unicode text files. It will corrupt UTF-8 encoded files.
Some examples:
Convert from DOS default code page to Unix Latin-1:
dos2unix -iso -n in.txt out.txt
Convert from DOS CP850 to Unix Latin-1:
dos2unix -850 -n in.txt out.txt
Convert from Windows CP1252 to Unix Latin-1:
dos2unix -1252 -n in.txt out.txt
Convert from Windows CP1252 to Unix UTF-8 (Unicode):
iconv -f CP1252 -t UTF-8 in.txt | dos2unix > out.txt
Convert from Unix Latin-1 to DOS default code page:
unix2dos -iso -n in.txt out.txt
Convert from Unix Latin-1 to DOS CP850:
unix2dos -850 -n in.txt out.txt
Convert from Unix Latin-1 to Windows CP1252:
unix2dos -1252 -n in.txt out.txt
Convert from Unix UTF-8 (Unicode) to Windows CP1252:
unix2dos < in.txt | iconv -f UTF-8 -t CP1252 > out.txt
See also <http://czyborra.com/charsets/codepages.html> and <http://czyborra.com/charsets/iso8859.html>.
All versions of dos2unix and unix2dos can convert UTF-8 encoded files, because UTF-8 was designed for backward compatibility with ASCII.
Dos2unix and unix2dos with Unicode UTF-16 support, can read little and big
endian UTF-16 encoded text files. To see if dos2unix was built with UTF-16
support type
dos2unix -V
.
On Unix/Linux UTF-16 encoded files are converted to the locale character encoding. Use the locale(1) command to find out what the locale character encoding is. When conversion is not possible a conversion error will occur and the file will be skipped.
On Windows UTF-16 files are by default converted to UTF-8. UTF-8 formatted text files are well supported on both Windows and Unix/Linux.
UTF-16 and UTF-8 encoding are fully compatible, there will no text be lost in the conversion. When an UTF-16 to UTF-8 conversion error occurs, for instance when the UTF-16 input file contains an error, the file will be skipped.
When option
-u
is used, the output file will be written in the same UTF-16
encoding as the input file. Option
-u
prevents conversion to UTF-8.
Dos2unix and unix2dos have no option to convert UTF-8 files to UTF-16.
ISO and 7-bit mode conversion do not work on UTF-16 files.
On Unix Unicode files typically don't have a BOM. It is assumed that text files are encoded in the locale character encoding.
Dos2unix can only detect if a file is in UTF-16 format if the file has a BOM. When an UTF-16 file doesn't have a BOM, dos2unix will see the file as a binary file.
Use option
-ul
or
-ub
to convert an UTF-16 file without BOM.
Dos2unix writes by default no BOM in the output file. With option
-b
Dos2unix writes a BOM when the input file has a BOM.
Unix2dos writes by default a BOM in the output file when the input file has a
BOM. Use option
-r
to remove the BOM.
Dos2unix and unix2dos write always a BOM when option
-m
is used.
There are some issues with displaying Unicode file names in a Windows console.
See option
-D
,
--display-enc
. The file names may be displayed wrongly in
the console, but the files will be written with the correct name.
dos2unix -n in.txt out.txt
Convert from Windows UTF-16LE (without BOM) to Unix UTF-8:
dos2unix -ul -n in.txt out.txt
Convert from Unix UTF-8 to Windows UTF-8 with BOM:
unix2dos -m -n in.txt out.txt
Convert from Unix UTF-8 to Windows UTF-16:
unix2dos < in.txt | iconv -f UTF-8 -t UTF-16 > out.txt
GB18030 is fully compatible with Unicode, and can be considered an unicode transformation format. Like UTF-8, GB18030 is compatible with ASCII. GB18030 is also compatible with Windows code page 936, also known as GBK.
On Unix/Linux UTF-16 files are converted to GB18030 when the locale encoding is
set to GB18030. Note that this will only work if the locale is supported by the
system. Use command
locale -a
to get the list of supported locales.
On Windows you need to use option
-gb
to convert UTF-16 files to GB18030.
GB18030 encoded files can have a Byte Order Mark, like Unicode files.
dos2unix < a.txt cat a.txt | dos2unix
Convert and replace a.txt. Convert and replace b.txt:
dos2unix a.txt b.txt dos2unix -o a.txt b.txt
Convert and replace a.txt in ascii conversion mode:
dos2unix a.txt
Convert and replace a.txt in ascii conversion mode, convert and replace b.txt in 7bit conversion mode:
dos2unix a.txt -c 7bit b.txt dos2unix -c ascii a.txt -c 7bit b.txt dos2unix -ascii a.txt -7 b.txt
Convert a.txt from Mac to Unix format:
dos2unix -c mac a.txt mac2unix a.txt
Convert a.txt from Unix to Mac format:
unix2dos -c mac a.txt unix2mac a.txt
Convert and replace a.txt while keeping original date stamp:
dos2unix -k a.txt dos2unix -k -o a.txt
Convert a.txt and write to e.txt:
dos2unix -n a.txt e.txt
Convert a.txt and write to e.txt, keep date stamp of e.txt same as a.txt:
dos2unix -k -n a.txt e.txt
Convert and replace a.txt, convert b.txt and write to e.txt:
dos2unix a.txt -n b.txt e.txt dos2unix -o a.txt -n b.txt e.txt
Convert c.txt and write to e.txt, convert and replace a.txt, convert and replace b.txt, convert d.txt and write to f.txt:
dos2unix -n c.txt e.txt -o a.txt b.txt -n d.txt f.txt
find . -name '*.txt' -print0 |xargs -0 dos2unix
The find(1) option
-print0
and corresponding xargs(1) option
-0
are needed
when there are files with spaces or quotes in the name. Otherwise these options
can be omitted. Another option is to use find(1) with the
-exec
option:
find . -name '*.txt' -exec dos2unix {} \;
In a Windows Command Prompt the following command can be used:
for /R %G in (*.txt) do dos2unix "%G"
PowerShell users can use the following command in Windows PowerShell:
get-childitem -path . -filter '*.txt' -recurse | foreach-object {dos2unix $_.Fullname}
export LANG=nl Dutch export LANG=nl_NL Dutch, The Netherlands export LANG=nl_BE Dutch, Belgium export LANG=es_ES Spanish, Spain export LANG=es_MX Spanish, Mexico export LANG=en_US.iso88591 English, USA, Latin-1 encoding export LANG=en_GB.UTF-8 English, UK, UTF-8 encoding
For a complete list of language and country codes see the gettext manual: <http://www.gnu.org/software/gettext/manual/html_node/Usual-Language-Codes.html>
On Unix systems you can use the command locale(1) to get locale specific information.
If you select a language which is not available you will get the standard English messages.
Example (POSIX shell):
export DOS2UNIX_LOCALEDIR=$HOME/share/locale
The return value is always zero in quiet mode, except when wrong command-line options are used.
<http://en.wikipedia.org/wiki/Carriage_return>
<http://en.wikipedia.org/wiki/Newline>
<http://en.wikipedia.org/wiki/Unicode>
Project page: <http://waterlan.home.xs4all.nl/dos2unix.html>
SourceForge page: <http://sourceforge.net/projects/dos2unix/>