Whitespace in filenames in for loop in bash script

Shell Programming and Scripting
1

I'm trying to search all .odt files in a directory for a string in the text of the file.
I've found a bash script that works, except that it can't handle whitespace in the filenames.

#!/bin/bash

if [ $# -ne 1 ]; then
    echo "Usage: searchodt searchterm"
    exit 1
fi

for file in $(ls *.odt); do
    unzip -ca "$file" content.xml | zgrep -ql "$1"
    if [ $? -eq 0 ]; then
        echo "$file"
    fi
done

(Coutesy of [ubuntu] [SOLVED] Search multiple .odt files - Ubuntu Forums)
I've gone through a number of postings on this forum, but simple tricks like quotes, of any kind, don't work. Any quotes I put around

(ls *.odt)

or just

*.odt

stop it working completely

I found this code

find /path/to/some/where/ -name  "*.pdf"  | awk '{print $5}'| uniq -d |while read name ; do

in a thread here

which solves the problem in that context, but I don't see how to integrate it into the script.

2

Try it with:

for file in *.odt; do

instead of:

for file in $(ls *.odt); do
3

Thank you very much, I never thought of just taking that out!

For the completeness of this thread in case others are interested it has also been pointed out by someone on the other thread that all the way over on page *2* of the referenced thread there is a working script for this purpose, which also takes the path as command line arg. :o

#!/bin/bash

if [ $# -ne 2 ]; then
        echo "Usage: searchodt searchpath searchterm"
        exit 1
fi

find $1 -name "*.odt" | while read file
do
        unzip -ca "$file" content.xml | grep -qli "$2"
        if [ $? -eq 0 ]; then
                echo "Found keyword in: " $file
        fi

done

4

For extra completeness:
The reason why $(ls *) fails for files with spaces is that the default word separators for the shell includes spaces. So when the result of the substitution gets expanded by the shell, spaces act as boundaries and filenames get cut over them.
One solution is to remove the space character as a separator, by modifying the special IFS variable:

IFS=$(echo "")

echo always prints a newline unless passed -n, so the above produces a newline portably across platforms.
Now, only newlines define word boundaries and spaces in filenames are no longer a problem.

This way you can substitute things like $(grep string file) for iterating.

5

Thanks, useful know how.