Skip to content

tr

Replaces, squeezes or deletes single characters in a text stream.

tr output

Syntax

tr [OPTION]... SET1 [SET2]

Options

Option Description
(none) Replace each character from SET1 with the matching one from SET2
-d, --delete Delete all characters from SET1
-s, --squeeze-repeats Squeeze repeated characters into one
-c, --complement Use all characters that are not in SET1

Useful character sets: a-z, A-Z, 0-9, [:upper:], [:lower:], [:digit:], [:space:], '\n' (line break), '\r' (Windows line ending), '\t' (tab).

Examples

Convert to upper case:

echo "Hello World" | tr 'a-z' 'A-Z'
HELLO WORLD

Convert to lower case:

echo "Hello World" | tr '[:upper:]' '[:lower:]'
hello world

Remove Windows line endings from a script:

tr -d '\r' < deploy.sh > deploy-fixed.sh
file deploy-fixed.sh
deploy-fixed.sh: Bourne-Again shell script, ASCII text executable

Squeeze several spaces into one:

echo "too    many   spaces" | tr -s ' '
too many spaces

Show a colon-separated list line by line:

echo "/usr/local/bin:/usr/bin:/bin" | tr ':' '\n'
/usr/local/bin
/usr/bin
/bin

Generate a random password:

tr -dc 'A-Za-z0-9' < /dev/urandom | head -c 20; echo
q8ZkT3vNw1LpXa7RcY0m

-dc deletes every character that is not a letter or digit, so only those remain.

When do I need it?

  • To fix files with Windows line endings (tr -d '\r')
  • To normalise upper and lower case before comparing or counting
  • To split lists like $PATH into lines, or to join lines into a list (tr '\n' ',')
  • To clean up output with multiple spaces before passing it to cut

One entry per line

echo "$PATH" | tr ':' '\n' shows every directory in your search path on its own line. That makes it much easier to read.

tr only reads from standard input

tr does not take a file name. tr a b users.csv fails with tr: extra operand ‘users.csv’. Use tr a b < users.csv or a pipe instead.

No umlauts

GNU tr works byte by byte. Characters like ä or é consist of several bytes in UTF-8 and are not converted, or are even broken. Use sed for such text.

cut · file · sed