将文件递归重命名为ASCII标准

Question

So we have a problem where we need to crawl through hundreds of thousands of images and rename all of them to comply with ASCII standards. After doing a lot of research online, we found this handy piece of code:

mv 'file' $(echo 'file' | sed -e 's/[^A-Za-z0-9._-]/_/g')

sourced from: How to remove invalid characters from filenames

I have tried merging it into a recursive find command, to be run whilst in our main images directory:

find . -print0 | xargs -0 mv $(echo | sed -e 's/[^A-Za-z0-9._-]/_/g')

But i cant seem to get it to run. The closest i've gotten, is the above code throwing a lot of "mv: target filename Is not a directory"

So can anyone help with this ?

Answer 1

I guess this would be better:

find . -type f -exec bash -c 'for f do d=${f%/*} b=${f##*/} nb=${b//[^A-Za-z0-9._-]/_}; [[ $b = "$nb" ]] || echo mv "$f" "$d/$nb"; done' _ {} +

find will find all files ( -type f ), pass them as positional arguments to this Bash snippet:

for f do
    d=${f%/*} b=${f##*/} nb=${b//[^A-Za-z0-9._-]/_}
    [[ $b = "$nb" ]] || echo mv "$f" "$d/$nb"
done

We split the filename into dirname d and basename b . We use parameter expansions to replace all the unwanted characters with underscores and save that expansion into variable nb . We check that the expansions $b and $nb are distinct (so as to avoid an error with mv ), and if they are distinct, perform the renaming.

I left echo so that nothing is actually performed, the commands are only echoed. Remove the echo if it looks good.

Note that this can overwrite files, eg, files a&b and a_b .

Answer 2

Put this in a shell script, say fixname.sh:

#!/bin/sh

dir=$(dirname "$1")
name=$(basename "$1")

newname=$(echo "$name" | sed -e 's/[^A-Za-z0-9._-]/_/g')
if [ "$name" != "$newname" ]; then
    if [ ! -e "$newname" ]; then
        mv "$1" "$dir/$newname"
    else
        echo >&2 "$newname already exist for $1"
    fi
fi

Then use find like so:

find . -type f -exec sh fixname.sh {} \;