tr - translate or delete characters
purpose
the tr command is for “translating” (replacing or deleting) characters in a string. Accepts input from stdin, and puts output on stdout. I am bold about characters because it’s not got a way to specify that you want to replace one string with another string (use sed for that).
examples
single character replacement
The basic usage on tr is to just replace one character with another. It’s case-sensitive:
echo "aAbBcCdD1234" | tr 1 R
aAbBcCdDR234 #only the '1' was replaced with 'R'
character set replacement
One can specify a set of characters to be replaced by a set of others:
echo "Johnny Apple Seed" | tr abcde ZYXWV
Johnny ApplV SVVW
Something interesting to also note is what happens if the size of each set of characters differ:
echo "abcd1234" | tr 12 abcd #[[ L1 < L2 ]]: L2's extra chars take no effect
abcdab34
echo "abcd1234" | tr 1234 ab #[[ L1 > L2 ]]: L1's extra chars are replaced by last char in L2
abcdabbb
You can eliminate the weird side effect in the 2nd example above by “truncating” set1 (tr -t):
echo "abcd1234" | tr -t 1234 ab
abcdab34 # notice the '3' and '4' are still there.
delete characters
echo "aAbBcCdD1234" | tr -d aC3
AbBcdD124
I keep seeing this specifically pop up to “translate” microsoft newline/carriage returns (’\r\n’) and turn them into regular *nix newlines (’\n’):
file dosplaintext.txt
#dosplaintext.txt: ASCII text, with CRLF line terminators
tr -d '\r' < dosplaintext.txt > plaintext.txt
file plaintext.txt
#plaintext.txt: ASCII text
complement
using the -c makes it so that the complement (all characters OTHER than the ones specified) gets translated. Here’s a normal one followed by an inverted one:
echo "abcd1234" | tr 12 ab
abcdab34
echo "abcd1234" | tr -c 12 ab
bbbb12bbb
compress
using -s makes it so that it “compresses” sequences of the same character (as well as translate, if a second set is given):
echo "aabbccdd11223344" | tr -s 1
aabbccdd1223344 # notice only a single '1' is left behind
echo "aabbccdd11223344" | tr -s 1 a
abbccdda223344 # that same '1' got left behind, then translated to an 'a'.
interpreted sequences
Characters are taken literally; a=a, 1=1, b=b, etc. But some strings passed in sets are interpreted. There are several. Refer to the man for the full list, but there are things like [:space:] to represent white space, or [:upper:] to represent all upper case chars, etc.