11

I have string=substr1-substr2-substr3.substr4 where the substrings are of variable length. I want to extract substr3 from string and am thinking of doing it using the ${string##pattern} and ${string%pattern} string manipulation expansions. Using this approach, it's clear I need to run ${string##*-} and run ${string%.*} on the resulting expansion, or vice versa. My questions are:

  1. is this the best choice?
  2. If yes, how do i run both in a single step? When I try to do something like ${string##*-${string%.*}} or ${string%.*${string##*-}} I just get back the full string. I realize i could always do the extraction in two steps by assigning an intermediate variable to the result of the first step, and then manipulating that; but I want to do it in one step. How do I go about it?

TIA!

Improve this question

2 Answers 2

Reset to default
6

Given

string=substr1-substr2-substr3.substr4

${string%.*} is

substr1-substr2-substr3

so, ${string##*-${string%.*}} is

${string##*-substr1-substr2-substr3}

which means to remove from the start of string, everything up to and including -substr1-substr2-substr3. But there's no such substring in string, as there's no dash in front of substr1, so you get the original string.

What you probably want is

string=substr1-substr2-substr3.substr4
result="${string%.*}"
result="${result##*-}"

which leads to result containg substr3.

In Zsh, you could do it in one step: result=${${string%.*}##*-}, but that doesn't work in Bash.

In Bash, you're limited to using expansions on the right side of # or %, and that doesn't help you here since you need to remove parts from both the start and end of the string.

Alternatively, if $string doesn't contain newline characters, you can use

IFS=.- read -r a b c d <<< "$string"

and read substr3 off from c.

In that case, depending on the shell and the size of the string, the contents of $string will be either stored in a temporary file or fed through a pipe which read will read the first line from and split it according to $IFS.

Improve this answer
4
  • thanks for the explanation; makes sense. I did want to avoid the multiple step extraction... had thought of the read + here string solution, but wanted to see if i could do it with parameter expansion instead. But, yeah I think I am going to go with the alternative. Any particular reason I might need the -r flag though? Commented Aug 15, 2020 at 13:55
  • 1
    @computronium, -r tells read not to treat backslashes specially, otherwise they can be used to make the delimiters, well, not delimit. IFS=.- read a b <<< "foo\-bar" puts foo-bar in a and nothing in b. Commented Aug 15, 2020 at 14:03
  • got that, but in my case the here string will not have any backslashes. so was wondering if there was any other reason you might have suggested it... thanks a lot anyway. Commented Aug 15, 2020 at 14:05
  • @computronium, well, force of habit, and of course I couldn't be sure what you'd have there. Commented Aug 15, 2020 at 14:37
-2

The question was about bash but for anyone who is on Zsh and comes across this question through search... this might be helpful.

An answer for Zsh only...

... and maybe other shells but not bash (5.1.16)

To do multiple string substitutions on Zsh you can nest parameter expansions:

string=substr1-substr2-substr3.substr4
result=${${string##*-}%.*}
Improve this answer
1
  • Please explain the down-vote. Commented Sep 26, 2023 at 7:00

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.