替换子目录中文件中的空格

Question

I have about 10,000 directories containing files. I want to write a loop that iterates through each directory, picks out a. txt file, and replaces any spaces with a _ . How can I do this?

for f in *FOLDERS*
do
    cd "$f" && echo "Entering into $f" || { echo "Error: could not enter into $f"; continue; }
    for y in $(ls *.txt | mv "$y" "${y// /_}")
    do
        echo ${y}
    done
done 

But it doesn't work for each directory. What am i doing wrong?

Answer 1

This might be what you're trying to do:

#!/bin/bash

for d in */; do
    cd "$d" || exit
    for t in *\ *.txt; do
        [[ -f $t ]] && mv -i "$t" "${t// /_}"
    done
    cd ..
done

Or, if you want to do it recursively (for all subdirectories in any depth):

#!/bin/bash

shopt -s globstar

for d in **/; do
    cd "$d" || exit
    for t in *\ *.txt; do
        [[ -f $t ]] && mv -i "$t" "${t// /_}"
    done
    cd - >/dev/null
done

Answer 2

1. List all files with spaces in filenames.
2. In that list, duplicate each line. In the second line, change all spaces in the filename for _ .
3. For every two lines, execute mv .

find *FOLDERS* -type f -name '* *.txt' -print0 |
# Duplicate the line. Replace spaces by _ in the second line.
sed -Ez 'h ; s@.*/@@ ; s@ @_@g ; G ; s@([^\x00]*)\x00(.*/)?([^/]*)@\2\3\x00\2\1@' |
# Execute mv for each two arguments.
xargs -0 -n2 echo mv -v

Answer 3

Select files with find , then run bash to perform the filename manipulation.

find . -type f -name '* *.txt' -exec bash -c 'for path; do
    basename="${path##*/}"
    dir="${path%/*}"
    echo mv "$path" "$dir"/"${basename// /_}"
done'  - {} +

This is fully recursive; if you don't want that you can limit the selection depth with find . The other advantage to using find is you can tell for sure what files you will be operating on before you run the dangerous part.

The above is harmless as-is, to make it actually perform its function remove the echo before the mv .

Answer 4

This is why your script is not working. Consider this folder structure:

$ ls
d1  d2  d3

Now lets try to cd into each folder:

$ for d in *; { cd $d; pwd; }
/tmp/d1
bash: cd: d2: No such file or directory
/tmp/d1
bash: cd: d3: No such file or directory
/tmp/d1

You have to go back to 'home' folder first:

$ for d in *; { cd $d; pwd; cd ..; }
/tmp/d1
/tmp/d2
/tmp/d3

Or use full path in $d definition:

$ for d in $PWD/*; { cd $d; pwd; }
/tmp/d1
/tmp/d2
/tmp/d3

Answer 5

For something fast, you will probably want Sorpigal's answer; this will still work, but it's slower.

#!/bin/sh -x

find . -type f -name '* *.txt' > stack

next () {
[[ -s stack ]] && main
end
}

main () { 
line=$(sed -n "1p" stack)
echo "${line}" | tr '/' '\n' > f2
basename=$(sed -n "$p" f2)
sed -i "$d" f2
dirname=$(cat f2 | tr '\n' '/')
newname=$(echo "${basename}" | tr ' ' '_')
mv -v "${dirname}/${basename}" "${dirname}/${newname}"
sed -i "1d" stack
rm -v ./f2
next
}

end () {
rm -v ./stack
exit 0
}

next