递归区分bash中的文件和目录

Question

Bash here. Looking to write a script that:

• Takes a absolute/relative target directory as its sole input argument and loops through its contents
• If one of its direct children is a file, I want it to echo " Found a file called $file! " (where $file is the name of the file)
• Else, if one of its direct children is itself a directory, I want it to echo " Found a directory named $dir " (where $dir is the name of the directory), but then I want it to recursively execute with this same exact logic inside that directory, and finally I want it to continue looping through the rest of the target directory

Hence given the following target directory structure:

~/testDir/
    1.txt
    2.txt
    childDirA/
        foo.png
    3.txt
    childDirB/
        buzz.gif
        childDirC/
            foo.bar
    4.txt

The output of the script would be:

Found a file called 1.txt!
Found a file called 2.txt!
Found a directory named childDirA
Found a file called foo.png!
Found a file called 3.txt!
Found a directory named childDirB
Found a file called buzz.gif!
Found a directory named childDirC
Found a file called foo.bar!
Found a file called 4.txt!

So far the best I've been able to come up with is:

#!/bin/bash
for file in $1;
  do echo "Found a file called $file";
done

However if I point this at my testDir the only output I get is:

Found a file called testDir

Any ideas where I'm going awry?

Answer 1

Here's a script that'll do that.

function listDir() {
    for file in "$1"/*;
    do
        if [ -f "$file" ]; then
            echo "Found file $(basename "$file")"
        elif [ -d "$file" ]; then
            echo "Found directory $(basename "$file")"
            listDir "$file"
        fi
    done
}

listDir "$1"

Answer 2

As Tripp mentioned in the comments, find can do the 'recursive' work for you; then you just need to tweak the output into the format you want.

Assuming you only need to worry about files and directories:

#!/usr/bin/bash

while IFS= read -d "" -r tgt
do
        # for printing purposes, strip off directory info; basically simulate
        # `basename` without the overhead of spawning an expensive sub-process

        tgtname=${tgt##*/}

        # determine tgt's type : 'file' or 'directory'

        tgttype='file'
        [ -d "${tgt}" ] && tgttype='directory'

        echo "Found a ${tgttype} named ${tgtname}"

done < <(find "$1" -print0)

NOTES:

• run find $1 by itself to see the format of the data being fed into the loop
• add additional tests to determine tgttype if you're dealing with links, devices, etc (perhaps use a case statement?)
• consider additional command line args/params to find to fine tune the list that's fed into the loop

Answer 3

Find

#!/bin/sh
find "$1" -mindepth 1 -type f -printf 'Found a file called %f!\n' -o -type d -printf 'Found a directory named %f\n'

Explanation:

By default, find recurses through subdirectories.

Arguments to find :

• -mindepth 1 Exclude the base directory
• -type f ... If a normal file is found, do the following
• -o -type d ... Or, if a directory is found, do the following
• -printf ... Print according to the given format string
• %f The basename of the file (dirs are files in this context)