Search code examples
bashunixawksedtr

tr -d to remove an exact string of characters from a string

I am trying to make tr -d remove a string of characters from an existing string, without it removing characters everywhere else.

For example, I want tr to remove : OK from the end of every string in foo.txt.

Contents of foo.txt:

BROKEN BONES: OK
Kefen Odvora: OK
BOOKS_FOR_MUM: OK
E: OK Amded: OK

This is the command I run:

cat foo.txt | tr -d ": OK$"

I want it to output this:

BROKEN BONES
Kefen Odvora
BOOKS_FOR_MUM
E: OK Amded

But instead I get this, which I don't want:

BRENBNES
efendvora
BS_FR_MUM
EAmded

How can I fix this?

Solution

  • You are using the wrong tool. You want sed, not tr:

    cat foo.txt | sed 's/: OK$//'

    or preferably

    sed 's/: OK$//' foo.txt

    when your input really is just a file, not a more complicated command.

    tr -d removes all occurrences of any character found in the argument to -d; it does not treat it as a regular expression to match and remove. Specifically, you are removing all occurrences of :, , O, K, and $ from each line.