Test if element is in array in bash

Question

Is there a nice way of checking if an array has an element in bash (better than looping through)?

Alternatively, is there another way to check if a number or string equals any of a set of predefined constants?

Answer 1

In Bash 4, you can use associative arrays:

# set up array of constants
declare -A array
for constant in foo bar baz
do
    array[$constant]=1
done

# test for existence
test1="bar"
test2="xyzzy"

if [[ ${array[$test1]} ]]; then echo "Exists"; fi    # Exists
if [[ ${array[$test2]} ]]; then echo "Exists"; fi    # doesn't

To set up the array initially you could also do direct assignments:

array[foo]=1
array[bar]=1
# etc.

or this way:

array=([foo]=1 [bar]=1 [baz]=1)

Answer 2

It's an old question, but I think what is the simplest solution has not appeared yet: test ${array[key]+_}. Example:

declare -A xs=([a]=1 [b]="")
test ${xs[a]+_} && echo "a is set"
test ${xs[b]+_} && echo "b is set"
test ${xs[c]+_} && echo "c is set"

Outputs:

a is set
b is set

To see how this work check this.

Answer 3

There is a way to test if an element of an associative array exists (not set), this is different from empty:

isNotSet() {
    if [[ ! ${!1} && ${!1-_} ]]
    then
        return 1
    fi
}

Then use it:

declare -A assoc
KEY="key"
isNotSet assoc[${KEY}]
if [ $? -ne 0 ]
then
  echo "${KEY} is not set."
fi

Answer 4

You can see if an entry is present by piping the contents of the array to grep.

 printf "%s\n" "${mydata[@]}" | grep "^${val}$"

You can also get the index of an entry with grep -n, which returns the line number of a match (remember to subtract 1 to get zero-based index) This will be reasonably quick except for very large arrays.

# given the following data
mydata=(a b c "hello world")

for val in a c hello "hello world"
do
           # get line # of 1st matching entry
    ix=$( printf "%s\n" "${mydata[@]}" | grep -n -m 1 "^${val}$" | cut -d ":" -f1 )

    if [[ -z $ix ]]
    then
        echo $val missing
    else
         # subtract 1.  Bash arrays are zero-based, but grep -n returns 1 for 1st line, not 0 
        echo $val found at $(( ix-1 ))
    fi
done

a found at 0
c found at 2
hello missing
hello world found at 3

explanation:

• $( ... ) is the same as using backticks to capture output of a command into a variable
• printf outputs mydata one element per line
• (all quotes necessary, along with @ instead of *. this avoids splitting "hello world" into 2 lines)
• grep searches for exact string: ^ and $ match beginning and end of line
• grep -n returns line #, in form of 4:hello world
• grep -m 1 finds first match only
• cut extracts just the line number
• subtract 1 from returned line number.

You can of course fold the subtraction into the command. But then test for -1 for missing:

ix=$(( $( printf "%s\n" "${mydata[@]}" | grep -n -m 1 "^${val}$" | cut -d ":" -f1 ) - 1 ))

if [[ $ix == -1 ]]; then echo missing; else ... fi

• $(( ... )) does integer arithmetic

Answer 5

I don't think you can do it properly without looping unless you have very limited data in the array.

Here is one simple variant, this would correctly say that "Super User" exists in the array. But it would also say that "uper Use" is in the array.

MyArray=('Super User' 'Stack Overflow' 'Server Fault' 'Jeff' );
FINDME="Super User"

FOUND=`echo ${MyArray[*]} | grep "$FINDME"`

if [ "${FOUND}" != "" ]; then
  echo Array contains: $FINDME
else
  echo $FINDME not found
fi

#
# If you where to add anchors < and > to the data it could work
# This would find "Super User" but not "uper Use"
#

MyArray2=('<Super User>' '<Stack Overflow>' '<Server Fault>' '<Jeff>' );

FOUND=`echo ${MyArray2[*]} | grep "<$FINDME>"`

if [ "${FOUND}" != "" ]; then
  echo Array contains: $FINDME
else
  echo $FINDME not found
fi

The problem is that there is no easy way to add the anchors (that I can think of) besides looping through the array. Unless you can add them before you put them in the array...