Bash: Problem with reading files with spaces in the name using a for loop
Recently we were working on a bash script that was supposed to find and process some files that matched certain criteria. The script would process the files one by one and the criteria would be matched using the find
command. To implement our solution, we returned the results of the find
back to the for loop in an attempt to keep it simple and human readable.
Our original code was the following:
(do not use it, see explanation below)
for file in `find $search_path -type f -name '*.kml'`; do
# Formatting KML file to be human friendly.
xmllint --format "$file" > "$output_path/$file";
done
Soon we realized that we had a very nasty bug, the way we formatted the command it would break filenames that had spaces in them into multiple for loop entries and thus we would get incorrect filenames back to process.
To solve this issue we needed a way to force our loop to read the results of find
one line at a time instead of one word at a time. The solution we used in the end was fairly different than the original code as it had the following significant changes:
- the results of the
find
command were piped into the loop - the loop was not longer a
for
loop and awhile
loop was used instead - it used the
read
command that reads one line at a time to fill in the filename variable
(the-r
parameter does not allow backslashes to escape any characters)
Solution
find $search_path -type f -name '*.kml' |
while read -r file; do
# Formatting KML file to be human friendly.
xmllint --format "$file" > "$output_path/$file";
done