Bash shell script question

yao_man · 05-15-2008, 06:20 PM

If this isn't the right place I apologize, but it seemed appropriate to me. I have been playing around with my shell script configuration, and I remapped rm and rmdir to move files to the trash instead and I created a delete function that can be used to delete files or directories after checking for confirmation a few times. Now the question, is there a way to iterate over all of the items in a directory. I tried ls -b and ls -1 but both have problems when the filename includes a space. BTW, this is all located in an alias file included in the ~/.bashrc. The alias file is called ~/.bash_addon if that makes any difference. Thank you for any help.

bjagee · 05-15-2008, 06:50 PM

Can you post the contents of the script? I'm guessing that find is what you are looking for, but I would have to see it in context.

pixellany · 05-15-2008, 07:42 PM

You can use a loop, such as:

Code:

for file in *; do
command
command
.
.
done

In the loop you can use the value of "file", eg:
rm $file

calraith · 05-15-2008, 08:12 PM

I'm sure there are less awkward ways to do it, but what I do is as follows:

Code:

for i in `ls -Amp | sed -r -e 's/,//g'`; do {
   command $i
}; done

which will get the dot files as well.

calraith · 05-15-2008, 08:13 PM

You could also use find with the -maxdepth option, and possibly -type f to exclude directories if you wish.

weibullguy · 05-15-2008, 09:41 PM

Quote:

Originally Posted by yao_man

both have problems when the filename includes a space.

When filenames include spaces you need to wrap the name in single quotes. For example

Code:

ls -b 'test file 1.txt'

If you use a loops or other recursive methods that use a variable you may find that commands inside the loop don't work on the files with spaces. Try using the find command with the -exec option. For example

Code:

find ./ -maxdepth 1 -type f -exec ls -b '{}' \; -exec file '{}' \;

Will find every file of type file in the current directory and then execute the ls -b and file command on the found file You can add more than two commands.

rlhartmann · 05-15-2008, 10:20 PM

If your files have spaces in the file name, you must quote the
variable name

ls "$FILE" or
rm -f "$FILE"

also, if you wanted to delete files that were more than say five days
old, you could do something like

find $TRASHDIR -type f -mtime +5 -exec rm -f {} \;

# The find command above will remove all files (files only -type f) more than five days old
out of the directory with name stored in the variable $TRASHDIR.

chrism01 · 05-15-2008, 11:07 PM

Although <space> is a valid char in a filename etc most cmds also use <space> as argument separators, so its highly advisable (for a quiet life) for you to avoid using spaces in filenames...

yao_man · 05-15-2008, 11:41 PM

I had at this point just run a test script to try and access all the file names and print them.

Code:

for file in `ls -b`;do
if [ -d "$file" ];then
echo "\"$file\" is a directory"
elif [ -f "$file" ];then
echo "\"$file\" is a file"
else
echo "\"$file\" is something else"
fi
done

I ran it on a test directory and it returned all of them until the file "test 1" which was read as "test/" and "1".

yao_man · 05-15-2008, 11:45 PM

I tried "for file in *" and it got all of the files correctly, but it couldn't get the hidden files. Is there a way to include hidden files in this?

jschiwal · 05-16-2008, 02:23 AM

You could either use
for file in $(ls -A)
or
for file in $(find . -maxdepth 1 -type f)

The latter will return only files.

Rather than aliasing rm and rmdir you might want to name the command something like "trash" so that the name describes what you want to do. You might also might want to only "trash" the files that you give in arguments instead of all of the contents in the current directory.

yao_man · 05-16-2008, 12:41 PM

I aliased rm and rmdir so that they move the files to the trash instead of removing them completely by using a "trash" function I wrote. I am trying to create a delete function that checks for approval before completely removing the given file and anything within that file (even if it is a directory) so I can easily remove files and directories when I need to.

Quote:

Originally Posted by jschiwal

You could either use
for file in $(ls -A)
or
for file in $(find . -maxdepth 1 -type f)

Both of those have troubles if the file has a space in it.
BTW, this is more of an attempt to teach myself programming problem solving and logic as well as overall programming ability then an attempt at efficiency.

yao_man · 05-16-2008, 05:06 PM

I apologize for this uber long post, but here goes. Ok, I did further searching and found a lot of useful information. I got the function almost working perfectly. The only problem is it will only erase hidden or non-hidden files in one go. If you switch block1 and block2 (they are labeled in the program) it switches which is deleted on the first go and which is deleted on the second go. I commented out two lines so that I could read what the program would execute at completion in the file "delete_tmp" Here is my code

Code:

delete () {	
	if [ "$del_number" == "0" ];then
		dir_list=""
	fi
	
	disp_string=""
	disp_arg=""
	no_exist=""
	
	for arg in $*;do
		if [[ "${arg:0:1}" != "/" ]];then
			arg="$PWD/$arg"
		fi
		if [[ "${arg:0:${#HOME}}" == "$HOME" ]];then
			disp_arg="~${arg:${#HOME}}"
		else
			disp_arg="$arg"
		fi
		if [ -d "$arg" ];then
			disp_string+="$disp_arg/*\n"
		elif [ -r "$arg" ];then
			disp_string+="$disp_arg\n"
		else
			no_exist+="\n$disp_arg"
		fi
	done
	
	if [[ $1 && $disp_string && "${1:0:2}" != "-h" && "${1:0:6}" != "--help" ]];then
		if [[ ! $del_number > 0 ]];then
			echo -e "The following files and all contents will be deleted:\n${disp_string:0:(${#disp_string}-2)}\nOnce completed, action cannot be reversed"
			read -sn 1 -p "Continue? [y,n]" ANS
			echo
			if [[ "$ANS" == "y" || "$ANS" == "Y" ]];then
				echo -e "All contents will be permanently lost\nAction cannot be reversed"
				read -sn 1 -p "Are you sure? [y,n]" ANS
				echo
			fi
		else
			ANS="y"
		fi
		if [[ "$ANS" == "y" || "$ANS" == "Y" ]];then
			del_number=$((del_number + 1))
			for arg in $*;do
				if [ -r "$arg" ];then
					if [ -d "$arg" ];then
						dir_list="/bin/rmdir \"$arg\"\n$dir_list"
						if [ $(ls -A "$arg" | wc -l) != "0" ];then
					#start of block 1
							for file in $arg/.*;do
								if [[ ${file:(${#file}-1):1} == "." || ${file:(${#file}-2):2} == ".." ]];then
									continue
								elif [ -d "$file" ];then
									delete "$file"
								elif [ -r "$file" ];then
									echo "/bin/rm \"$file\"" >> delete_tmp
								fi
							done
					#end of block1
					#start of block2
							for file in $arg/*;do
								if [ -d "$file" ];then
									delete "$file"
								elif [ -r "$file" ];then
									echo "/bin/rm \"$file\"" >> delete_tmp
								fi
							done
					#end of block2	
						fi
					elif [ -r "$arg" ];then
						echo "/bin/rm \"$arg\"" >> delete_tmp
					fi
				else
					if [[ "${arg:0:1}" != "/" ]];then
						arg="$PWD/$arg"
					fi
					if [[ "${arg:0:${#HOME}}" == "$HOME" ]];then
						arg="~${arg:${#HOME}}"
					fi
					echo "\"$arg\" could not be accessed or does not exist"
				fi
			done
			del_number=$((del_number - 1))
			if [ "$del_number" == "0" ];then
				if [ "$dir_list" ];then
					echo -e "${dir_list:0:(${#dir_list}-2)}" >> delete_tmp
					dir_list=""
				fi
				del_number=""
#next 2 lines uncommented for diagnosis
#				cat delete_tmp | bash
#				/bin/rm delete_tmp
			fi
		else
			echo -e "Deletion aborted"
		fi
	elif [[ "${1:0:2}" == "-h" || "${1:0:6}" == "--help" ]];then
		echo -e "del[ete] [-h/--help/file...]\n    -h/--help\tDisplay this help\n    -file...\tCompletely and permanently remove file[s] and contents\n\t\tfrom the system (once removed they cannot be retrieved)"
	else
		echo -e "None of the files exist or could be accessed\nDeletion cancelled"
	fi
	
	if [ $no_exist ];then
		echo -e "The following files did not exist or could not be accessed:$no_exist"
	fi
}

Here is a little breakdown of things in the code:
if -h or --help is the first option, display the help instead of running the delete function call
$del_number is used to tell if this is the initial delete call or if it is a delete call from inside a delete call
$disp_string is used when displaying the arguments offered for deletion
$ANS is used to check for confirmation for deletion on the first call and ignored on any calls from inside that same initial call
...further checks for empty directories ($(ls -A "$arg" | wc -l) != "0") etc
#block1 arranges for deletion of files starting with "." excluding "." and ".."
#block2 arranges for deletion of non hidden files
$del_number check sees if it is the initial call, add the directory deletion info to the end of "delete_tmp", then run "delete_tmp", then delete "delete_tmp"

If block1 is first, it deletes hidden files on the first call and the rest on the second call. If block2 is first, it is the other way around. I have some sample output available if you need it, but this post is too long already. Just ask to see it. Also, it still has problems if an argument contains spaces. e.g. delete 'test 1' tries to delete 'test' and '1' and I don't know how to fix that either. Thank you for any help you may offer. (code optimizations would be nice as well.

rlhartmann · 05-16-2008, 05:22 PM

Sorry, I forgot to mention,

If you are doing a for loop, you MUST redefine
the internal separate field using a command like

IFS="
"

The command above is QUOTING the return key,
type IFS=" then hit the enter key, no spaces,
then the quote, this is critcal for file names
for spaces using your for loop

Quote:

IFS="
"

for file in `ls -b`
do
....

yao_man · 05-16-2008, 05:42 PM

What is the default value of $IFS?

Nevermind, I think I got it. As for the program, still experiencing problems. Using IFS, I have it set up right, but somewhere in the loop, $arg gets lost so it doesn't see the files correctly.

Nevermind again...I switched back to my way, but I need to find out how to keep $arg from getting lost in the first for loop in block1.