9

I have an array list in a bash script, and a variable var. I know that $var appears in ${list[@]}, but have no easy way of determining its index. I'd like to remove it from list.

This answer achieves something very close to what I need, except that list retains an empty element where $var once was. Note, e.g.:

$ list=(one two three)
$ var="two"
$ list=( "${list[@]/$var}" )
$ echo ${list[@]}
one three
$ echo ${#list[@]}
3

The same thing happens if I use delete=( "$var" ) and replace $var for $delete in the third line. Also, doing list=( "${list[@]/$var/}" ) makes no difference either. (I'll note that, experimenting with the comment to that answer, I managed to match only whole words using list=( "${list[@]/%$var}" ), omitting the #.)

I also saw this answer proposing a nice trick to keep track of index and use unset, but that is unfeasible in my case. Finally, the same issue also appeared here, except that OP was satisfied with the result and probably didn't run into the problem empty elements create for me later on in my script, when I iterate through list. I tried to negate that problem by using expansion as follows, without any apparent effect:

for item in "${list[@]}"; do
  if [ -n ${item:+'x'} ];then
    ...
  fi
done

It's the same when I do [ ${#item} > 0 ], and I'm running out of ideas. Suggestions?

EDIT:

I have no understanding of why this happens, but @l0b0's comment made me notice something. Using the above preamble, I get:

$ for item in "${list[@]}"; do echo "Here!"; done
Here!
Here!
Here!

but:

$ for item in ${list[@]}; do echo "Here!"; done
Here!
Here!

I'm not sure I can omit the quotes in my script, though, as items are considerably more complicated there (file names and paths, both containing spaces and odd characters).

Improve this question
6
  • Regarding your edit, I am unable to reproduce it. I get three instances of Here! for both runs. Commented May 5, 2014 at 3:08
  • @JS웃 very odd. It's completely reproducible on my machine (running trusty). Commented May 5, 2014 at 10:27
  • 1
    The reason your edit gets two instances in the second case is that the list ("one" "" "three") gets expanded by the shell to for item in one three instead of for item in "one" "" "three" Commented Nov 23, 2015 at 12:34
  • @JoshuaGoldberg have you noticed that that question was the first link I provided, then explained why it doesn't suffice? (To wit, the empty element.) Commented Nov 1, 2016 at 18:56
  • @JonathanY., I'm very sorry, I did miss that... although it might be more of a limitation of the bulk of the answers there, rather than the question. (I posted a somewhat hacky answer there that doesn't happen to leave the blank slot.) I'm on the fence about whether it would be more helpful to people looking for answers to mark dup or just leave linked. Commented Nov 1, 2016 at 21:24

3 Answers 3

Reset to default
9

You can delete an element from existing array though the whole process isn't very straightforward and may appear like a hack. 

#!/bin/bash

list=( "one" "two" "three" "four" "five" )
var1="two"
var2="four"

printf "%s\n" "Before:"
for (( i=0; i<${#list[@]}; i++ )); do 
    printf "%s = %s\n" "$i" "${list[i]}"; 
done

for (( i=0; i<${#list[@]}; i++ )); do 
    if [[ ${list[i]} == $var1 || ${list[i]} == $var2 ]]; then
        list=( "${list[@]:0:$i}" "${list[@]:$((i + 1))}" )
        i=$((i - 1))
    fi
done

printf "\n%s\n" "After:"
for (( i=0; i<${#list[@]}; i++ )); do 
    printf "%s = %s\n" "$i" "${list[i]}"; 
done

This script outputs: 

Before:
0 = one
1 = two
2 = three
3 = four
4 = five

After:
0 = one
1 = three
2 = five

Key part of the script is:

list=( "${list[@]:0:$i}" "${list[@]:$((i + 1))}" )

Here we re-construct your existing array by specifying the index and length to remove the element from array completely and re-order the indices.

Improve this answer
Sign up to request clarification or add additional context in comments.

1 Comment

This post is incorrect. If you use var1="two" var2="three" it will fail because the indexes are shifted while you're looping over the array. So it will be index 1, remove "two", then go index 2, which is now "four" thus skipping testing three. To correct this i=$((i - 1)) should be inserted in the if statement.
7

If you want to delete the array element & shift the indices, you can use answer by l0b0 or JS웃.

However, if you don't want to shift the indices, you can use below script-let: (Particularly useful for associative arrays)

$ list=(one two three)
$ delete_me=two
$ for i in ${!list[@]};do
    if [ "${list[$i]}" == "$delete_me" ]; then
        unset list[$i]
    fi 
done

$ for i in ${!list[@]};do echo "$i = ${list[$i]}"; done
0 = one
2 = three

If you want to shift the indices to make them continuous, re-construct the array as this:

$ list=("${list[@]}")
$ for i in ${!list[@]};do echo "$i = ${list[$i]}"; done
0 = one
1 = three
Improve this answer

Comments

3

If you want to remove by value and shift the indexes I think you have to create a new array:

list=(one two three)
new_list=() # Not strictly necessary, but added for clarity
var="two"
for item in ${list[@]}
do
    if [ "$item" != "$var" ]
    then
        new_list+=("$item")
    fi
done
list=("${new_list[@]}")
unset new_list

Test:

$ echo "${list[@]}"
one three
$ echo "${#list[@]}"
2
Improve this answer

1 Comment

Thanks! However, as I might need to go through this process many times in one run, I'd really prefer to avoid this approach if at all possible. (Not that I expect the complexity of what I'm doing to matter in any real way, but there's the principle of the thing :))

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.