BASH remove specific tokens from a word

Question

I am trying to find te longest word in a given file. Before I check the lengtgh of each word I need to remove all of the following tokens {,.:} that may be attached (once or more) to the word. so for example, for this text:

:,cat dog, encyclopedia; remove:.,

i need the result:

cat dog encyclopedia remove

I am trying this, but I get a "command not found":

longest=0
for word in $(<$1)
do
    #new_word = $(echo "${word//[.,:]/}")
    new_word = "${word//[.,:]/}"
    len=${#new_word}

    if (( len > longest ))
    then
        longest=$len
        longword=$new_word
    fi
done
echo The longest word is $longword and its length is $longest.

thank you.

And by the way, [don't read lines with `for`](http://mywiki.wooledge.org/DontReadLinesWithFor) — gniourf_gniourf, Dec 24 '17 at 18:07
If your file contains `*`, for example, it'll be replaced with a list of filenames in the current directory -- surely not what you want. Consider instead: `read -r -a words <"$1"; for word in "${words[@]}"; do` — Charles Duffy, Dec 24 '17 at 18:08
The problem with `new_word = ...anything...` is that you can't have spaces around the `=` in an assignment; once you have a space, it's running `new_word` as a command with `=` as its first argument. — Charles Duffy, Dec 24 '17 at 18:09
BTW, http://shellcheck.net/ would have identified that for you automatically. — Charles Duffy, Dec 24 '17 at 18:10

score 0 · Accepted Answer · answered Dec 24 '17 at 18:09

Your use of parameter expansion replacement pattern is correct.

The problem is that there must not be any whitespace around = while declaring variables in bash (any shell in general).

So, the following should work:

new_word="${word//[.,:]/}"

As an aside, use a while read ... construct to loop over the lines in a file, using for is pretty fragile.

BASH remove specific tokens from a word

1 Answers1