Loop through the directories to get the image files

Question

I have a directory which includes multiple sub-directories. I would like to go through the directories and subdirectories and find the jpg files and convert the size using mogrify command. I would like to do it as dynamic as possible that's why I wrote a script. The $1 is the first argument that I pass through when executing the bash script. After running the script, it gives me an error about 'mogrify can not read [@]%'. I guess something is very wrong with my code and I am not mature in bash. Can anyone tell me how to do this script dynamically so that would be fast.

p.s: the name of jpg files are not in especial format...just bunch of numbers.

for folder in $1/*
do
    for file in "$folder"/*
    do
        if [  -e "${file[@]%.jpg}" ]; then
            mogrify -resize 112x112! "${file[@]%.jpg}"
        fi

    done
done

Answer 1

If you're open to using find, then this becomes pretty easy:

#!/usr/bin/env bash
find "$1" \( -iname \*.jpg -o -iname \*.jpeg \) -print0 | while read -r -d $'\0' file; do
  # base="${file##*/}" $base is the file name with all the directory stuff stripped off
  # dir="${file%/*}    $dir is the directory with the file name stripped off
  mogify -resize '112x112!' "$file"
done

Put that in a file named mymog.bash then

$ chmod 755 mymog.bash
$ mymog.bash /some/dir

Notes:

• ! is special to bash, so putting that in the single quotes make it "unspecial", passing it along to the mogrify command unmolested.
• The double quotes around $1 and $file are needed in case a directory or file name has spaces in it. If you had a directory named /Users/alice/my pictures and didn't use the quotes, mogrify would get one argument named /Users/alice/my and another one named pictures.
• Make sure you use the \( and \) for find. That makes the whole condition ("match *.jpg" OR "match *.jpeg") apply to the action -print0.
• I used find 's -print0 action which prints each matching file name with a null-terminated (zero-terminated) string. You can have filenames that have newline characters in the middle. This protects against that.
• bash 's built-in read command reads until a newline by default. I used the -d $'\0' to make it read each "line" or "record" (filename) up to the null (zero) character at its end. (Each ends with null because of the -print0.)

This solution (one of many) has two parts:

1. It uses the find utility to find (under the directory given) all files that end in .jpg or .jpeg, ignoring the case of the filenames. [So it will match .JPG or even \.JpEg.]
   • It spits out one record for each file.
   • If you give it an absolute path like /some/dir, it will find /some/dir/a.jpg and /some/dir/sub1/sub2/sub3/b.jpg.
   • If you give it a relative path like ../../nearby/dir, it will find ../../nearby/dir/c.jpg and ../../nearby/dir/sub1/sub2/sub3/d.jpeg.
2. The find part ends with the first | on that line. After that, it is a bash while…do loop.
   • The variable file takes on the value of each record spit out by find.
   • The loop (everything between do and done) runs once for each value that file takes on.
   • The two rows that start with # are comments. They contain commands that are ignored (skipped). You can remove the # to have bash run those commands too. I included them as examples in case you needed the directory part or just the filename part of the record.

Answer 2

find "$1" -type f -name "*.jpg" -exec mogrify -resize 112x112! {} \;

Answer 3

Try find and a while read loop:

 find "$1" -type f -name '*.jpg' -print | while read fname
 do
    ....
 done

If your filenames may contain special chars like line feed, the use:

find "$1" -type f -name '*.jpg' -print0 | while IFS= read -r -d '' fname
do
    ....
done

There are many more options to tune the search.