bash largest folder

trintukaz · 10-15-2011, 11:35 PM

Hello! I have a question. Which command can shows largest folder not by size but by amount of files? I can't find this info.

David the H. · 10-16-2011, 12:24 AM

1) Make a list of directories.
2) Determine the number of files in each directory.
3) Sort the list by that number.
4) The top entry will be your answer.

trintukaz · 10-16-2011, 12:30 AM

But every time new directory can be created and again list need to be updated.

David the H. · 10-16-2011, 01:31 AM

Yeah, so? That's true no matter what command you use.

Set it up as a script or function and tell it to count the files that exist at the time you run it. Even if there was a single command that did just what you want, it would just be doing the same thing internally.

Also, you probably need to define your criteria in more detail. What about sub-directories? Do they count as files, and do you want to recurse through their contents too? What about dotfiles? Symlinks?

trintukaz · 10-16-2011, 01:40 AM

Sub directories does not count dotfiles to.

David the H. · 10-16-2011, 05:25 AM

Ok then. I look forward to seeing what kind of solution you come up with.

trintukaz · 10-17-2011, 02:21 AM

I don't know what to do

Quote:

#!/bin/bash

for i in `find /home/ -type d`
do
find -type f | wc -l
done

Trying something like this.

trintukaz · 10-18-2011, 08:17 AM

anyone?

catkin · 10-18-2011, 10:29 AM

Quote:

Originally Posted by trintukaz

I don't know what to do

Code:

#!/bin/bash

for i in `find /home/arturas -type d`
do
find -type f | wc -l
done

Trying something like this.

The for i in `find /home/arturas -type d` implements David the H's first step, "Make a list of directories". It will bork on directory names with spaces in their names etc. but let's not get distracted by that yet.

The second step is "Determine the number of files in each directory". find -type f | wc -l is going in the right direction but you need to tell it the starting directory (otherwise it starts from the current directory) and to tell it not to look in any subdirectories. find's -maxdepth option is what you need.

BTW, when not familiar with the commands to put in a script it's helpful to run them at the command prompt to explore what they do.

The script will need to keep the number of files in each directory and the directory name. It could keep them in a variable but if there are a lot of directories the script will run slowly and writing them to a temporary file would allow the script to run faster. The >> redirection operator is good for that.

When scripts become large, it is very helpful to have meaningful variable names so for dir in `find /home/arturas -type d` would be a good choice.

Please post back how you got on with these hints.

trintukaz · 10-18-2011, 12:01 PM

I decided to look in subdirectories.

Quote:

#!/bin/bash

for i in `find /home -type d`
do
find "$i" -type f | wc -l
done

Now my script looks like this. It is i little bit better than previous but i think i need make somehow to show only largest directory and also show the name.

catkin · 10-22-2011, 08:54 AM

Quote:

Originally Posted by trintukaz

I decided to look in subdirectories.

That will count files multiple times. OK if that's OK for you.

Quote:

Originally Posted by trintukaz

Now my script looks like this. It is i little bit better than previous but i think i need make somehow to show only largest directory and also show the name.

Code:

#!/bin/bash

for i in `find /home -type d`
do
find "$i" -type f | wc -l
done

True. You've got as far as David the H's steps 1 and 2. Now you need the rest. Step 3 is "Sort the list by that number".

So you need a list. Assuming you want to use a temporary file then you can write the data to file as you generate it. $i already has the directory name. The echo command followed by the >> redirection operator is good for appending to temporary files. The sort command is good for sorting files.

Those are basic techniques explained in bash introductions such as:

BTW if you use CODE tags instead of quote tags then your code will be shown when people quote your post.