In bash script,

I have a string which contains several words separated by one or more than one spaces. ie:

Name   Age Sex  ID         Address

If I want to find any of the word, for instance I want to find the index of word "Age", how can I do it?

Is there any command that will return the index number of the word I want directly?



11 Answers 11


Bash performs word splitting in strings all by itself – in fact, more often than not, avoiding that is an issue, and the reason quoting is so important. It’s easy to leverage that in your case: just put your string into an array without quoting it – bash will use word splitting to separate the individual elements. Assuming your string is stored in the variable $str,

ar=($str) # no quotes!

will return an array of 5 elements. Your array index is your word index (counting up from 0, like in most scripting and programming languages), i.e. “Age” is accessed using

${ar[1]}  # 0 => Name, 1 => Age, 2 => Sex, 3 => ID, 4 => Address

or, if you need to find the element index by content, loop over the array, i.e.

function el_index {
    cnt=0; for el in "${ar[@]}"; do
        [[ $el == "$1" ]] && echo $cnt && break
el_index "Age" # => 1
  • wow...i didn't know that without the quotes then it would be an array. thanks!
    – GJ.
    Jun 10, 2012 at 0:08
$ export FOO="Name   Age Sex  ID         Address"

Replace *Age with Age -- this will remove anything before "Age":

$ echo ${FOO/*Age/Age}
Age Sex ID Address

Get anything before "Age"

$ echo ${FOO/Age*/}

Get the length of that string (which is the index of "Age"):

$ BEGIN=${FOO/Age*/}
$ echo ${#BEGIN}
  • Doesn't answer the question, but wow! Slick trick. It even works in ash, and with embedded variables: export L='debug info warn error'; export GTE='warn'; echo ${L/*${GTE}/${GTE}} prints 'warn error' Sep 1, 2018 at 1:16

This is a 7-years-old question, but some may need the answer in pure bash.

STRING="Name   Age Sex  ID         Address"
  • gives error on bash 5.0.17(1) at least: bash: ${#${STRING/Age*/}}: bad substitution but I get it, it is the size of the remaining string Apr 25 at 23:20

You can use bash's native regex

# a function to print the index of a field and its name
printIx() { 
  for ((l=0,i=1;i<$1;i++)) ;do 
  printf '%3s %s\n' $l "$2"

#   Using a zero based index
#   "0----+----1----+----2----+----3----+----4"
str="  Name   Age Sex  ID         Address   "

if [[ $str =~ ^(\ *)(Name)(\ +)(Age)(\ +)(Sex)(\ +()ID)(\ +)(Address)\ *$ ]] ;then
  F=(Name Age Sex ID Address)
  f=(   2   4   6  8      10)  # regex back-references
  for ((g=0;g<${#f[@]};g++)) ;do
     printIx  ${f[g]} "${F[g]}"


  2 Name
  9 Age
 13 Sex
 20 ID
 29 Address

Note: Assuming here that by index you mean you want to know which word it is (starting from 0), not which character in the string the word starts on. Other answers address the latter.

Not that I'm aware of, but you can make one. Two tricks:

  1. Use the inborn abilities of the for construct to split up an unquoted input by whitespace.
  2. Handle the case where you can't find the column you want. In this case, I chose to send the found index to stout and let the status code indicate whether the find was successful. There are other possibilities.


find_index() {
    local str=$1
    local search=$2
    let local n=0
    local retval=1 # here, 1 is failure, 0 success
    for col in $str; do # $str unquoted -> whitespace tokenization!
    if [ $col = $search ]; then
        echo $n
    return $retval

test="Name   Age Sex  ID         Address"
idx=`find_index "$test" Age`
if [ $? -ne 0 ]; then
    echo "Not found!"
    echo "Found: $idx"

Try the following javascript oneliner in a shell (use javascript shell) :

$ js <<< "x = 'Name   Age Sex  ID         Address'; print(x.indexOf('Age'));"

Or with a here-doc:

js <<EOF
x = 'Name   Age Sex  ID         Address';

I found a solution that works fine.

$ string='now is the time'
$ buf=the${string#*the}
$ echo $buf
output: the time
$ index=$(( ${#string} - ${#buf} + 1 ))
$ echo $index
output: 8 -> index of first word "the"

It works similar as function indexOf() in Java which returns first occurrence of an input string.

Found this solution in here http://www.linuxquestions.org/questions/linux-newbie-8/bash-string-manipulation-help-670627/ (last post). This guy saved my day. Credit to him.

Faster way if you want to do substring from first indexof.

$ a="some long string"
$ b="ri"
$ echo ${a/*$b/$b}
$ echo ${a/$b*/$b}
some long stri



If coreutils are available you can do it in following way:

echo ${str/Age//}| cut -d/ -f1| wc -w

Per MariusMatutiae request I'm adding an explanation how this 3 steps operation works:

echo ${str/Age//} 1. replace string which is being searched for unique char (in my case /)

cut -d/ -f1 2. cut off whole part of string which is after unique char

wc -w 3. count and print words which are left this will give us an index number

For references please check:

http://www.tldp.org/LDP/abs/html/parameter-substitution.html (go to: "Variable expansion / Substring replacement")
http://www.gnu.org/software/coreutils/manual/coreutils.html (go to: "The cut command" and "wc invocation"

  • While this solves the problem at hand, such terse replies are frowned upon in these sites. It would be more helpful to spend a few words explaining exactly why this works. Please do so. Jul 9, 2014 at 10:49

A mix of two previously given answers, using pure bash arrays and substring replacement.

The idea is to get a string of all words before the one you want, then count the number of words in that substring by making it into an array.

$ haystack="Name   Age Sex  ID         Address"
$ words_before=( ${haystack%Age*} )     # truncate string, make array
$ echo ${#words_before[*]}              # count words in array

Of course Age can be stored in another variable needle, then use ${haystack%$needle*}. Expect problems if the word you search for is a subset of another word, in which case kopischke's answer is still working.


If you don't have to strictly use bash, but can use other programs commonly found on systems with bash then you could use something like this:

echo "Name   Age Sex ID  Addr" | python -c 'print(raw_input().index("Age"))+1'

Python starts its string indexing at zero, therefore I added +1 to the end of the command.

  • "its string indexing", without the apostrophe. Jun 17, 2020 at 15:58

the size of the remaining string after the erasure using wc -c so that you dont need to create an intermediary string variable

str="Name   Age Sex  ID         Address";
nIndex="`echo -n "${str/Age*/}" |wc -c`";
declare -p nIndex

declare -- nIndex="7"

Btw, this may be useful info: https://www.thegeekstuff.com/2010/07/bash-string-manipulation/

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.