Sorting file names by length of the file name

Question

ls displays the files available in a directory. I want the file names to be displayed based on the length of the file name.

Any help will be highly appreciated. Thanks in Advance

Answer 1

The simplest way is just:

$ ls | perl -e 'print sort { length($b) <=> length($a) } <>'

Answer 2

You can do like this

for i in `ls`; do LEN=`expr length $i`; echo $LEN $i; done | sort -n

Answer 3

make test files:

mkdir -p test; cd test 
touch short-file-name  medium-file-name  loooong-file-name

the script:

ls |awk '{print length($0)"\t"$0}' |sort -n |cut --complement -f1

output:

short-file-name
medium-file-name
loooong-file-name

Answer 4

for i in *; do printf "%d\t%s\n" "${#i}" "$i"; done | sort -n | cut -f2-

Answer 5

TL;DR

Command:

find . -maxdepth 1 -type f -print0 | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | perl -F'/\0/' -ape '$_=join("\n", sort { length($b) <=> length($a) } @F)' | sed 's#/#/\\n/#g'

Alternate version of command that's easier to read:

find . -maxdepth 1 -type f -print0 | \
sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | \
perl -F'/\0/' -ape \
  '$_=join("\n", sort { length($b) <=> length($a) } @F)' | \
sed 's#/#/\\n/#g'

---

Not Parsing ls Output AND Benchmarking

There are good answers here. However, if one wants to follow the advice not to parse the output of ls, here are some ways to get the job done. This will especially take care of the situation where you have spaces in filenames. I'm going to benchmark everything here as well as the paring-ls examples. (Hopefully I get to that, soon.) I've put a bunch of somewhat-random filenames that I've downloaded from different places over the last 25 years or so -- 73 to begin with. All 73 are 'normal' filenames, with only alphanumeric characters, underscores, dots, and hyphens. I'll add 2 more which I make now (in order to show problems with some sorts).

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ mkdir ../dir_w_fnames__spaces

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ cp ./* ../dir_w_fnames__spaces/

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ cd ../dir_w_fnames__spaces/

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces
$ touch "just one file with a really long filename that can throw off some counts bla so there"

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces
$ mkdir ../dir_w_fnames__spaces_and_newlines

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces
$ cp ./* ../dir_w_fnames__spaces_and_newlines/

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces
$ cd ../dir_w_fnames__spaces_and_newlines/

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ touch $'w\nlf.aa'

This one, i.e. the filename,

w
lf.aa

stands for with linefeed - I make it like this to make it easier to see the problems. I don't know why I chose .aa as the file extension, other than the fact that it made this filename length easily visible in the sorts.

Now, I'm going back to the orig_dir_73 directory; just trust me that this directory only contains files. We'll use a surefire way to get the number of files.

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ du --inodes
74      .

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ # The 74th inode is for the current directory, '.'; we have 73 files

There's a more surefire way, which doesn't depend on the directory only having files and doesn't require you to remember the extra '.' inode. I just looked through the man page, did some research, and did some experimentation. This command is

awk -F"\0" '{print NF-1}' < <(find . -maxdepth 1 -type f -print0) | awk '{sum+=$1}END{print sum}'

or, in more-readable fashion,

awk -F"\0" '{print NF-1}' < \
  <(find . -maxdepth 1 -type f -print0) | \
    awk '{sum+=$1}END{print sum}'

Let's find out how many files we have

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ awk -F"\0" '{print NF-1}' < \
  <(find . -maxdepth 1 -type f -print0) | \
    awk '{sum+=$1}END{print sum}'
73

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ cd ../dir_w_fnames__spaces

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces
$ awk -F"\0" '{print NF-1}' < \
  <(find . -maxdepth 1 -type f -print0) | \
    awk '{sum+=$1}END{print sum}'
74

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces
$ cd ../dir_w_fnames__spaces_and_newlines/

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ awk -F"\0" '{print NF-1}' < \
  <(find . -maxdepth 1 -type f -print0) | \
    awk '{sum+=$1}END{print sum}'
75

(See ^{[ 1 ]} for details and an edge case for a previous solution that led to the command here now.)

I'll be switching back and forth between these directories; just make sure you pay attention to the path - I won't note every switch.

---

* Usable even with weird filenames (containing spaces, linefeeds, etc.)

1a. Perl à la @tchrist with Additions

Using find with null separator. Hacking around newlines in a filename.

Command:

find . -maxdepth 1 -type f -print0 | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | perl -F'/\0/' -ape '$_=join("\n", sort { length($b) <=> length($a) } @F)' | sed 's#/#/\\n/#g'

Alternate version of command that's easier to read:

find . -maxdepth 1 -type f -print0 | \
sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | \
perl -F'/\0/' -ape \
  '$_=join("\n", sort { length($b) <=> length($a) } @F)' | \
sed 's#/#/\\n/#g'

I'll actually show part of the sort results to show that the following command works. I'll also show how I'm checking that weird filenames aren't breaking anything.

Note that one wouldn't usually use head or tail if one wants the whole, sorted list (hopefully, it's not a sordid list). I'm using those commands for demonstration.

First, 'normal' filenames.

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ find . -maxdepth 1 -type f -print0 | \
sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | \
perl -F'/\0/' -ape \
  '$_=join("\n", sort { length($b) <=> length($a) } @F)' | \
sed 's#/#/\\n/#g' | head -n 5
68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt
oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt
17f09d51d6280fb8393d5f321f344f616c461a57a8b9cf9cc3099f906b567c992.txt

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ find . -maxdepth 1 -type f -print0 | \
sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | \
perl -F'/\0/' -ape \
  '$_=join("\n", sort { length($b) <=> length($a) } @F)' | \
sed 's#/#/\\n/#g' | tail -n 5
137.csv
13.csv
o6.dat
3.csv
a.dat

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ # No spaces in fnames, so...

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ find . -maxdepth 1 -type f | wc -l
73

• Works for normal filenames

Next: spaces

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces
$ find . -maxdepth 1 -type f -print0 | \
sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | \
perl -F'/\0/' -ape \
  '$_=join("\n", sort { length($b) <=> length($a) } @F)' | \
sed 's#/#/\\n/#g' | head -n 5
68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt
oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
just one file with a really long filename that can throw off some counts bla so there
79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt

• Works for filenames containing spaces

Next: newline

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ find . -maxdepth 1 -type f -print0 | \
sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | \
perl -F'/\0/' -ape \
  '$_=join("\n", sort { length($b) <=> length($a) } @F)' | \
sed 's#/#/\\n/#g' | tail -8
Lk3f.png
LOqU.txt
137.csv
w/\n/lf.aa
13.csv
o6.dat
3.csv
a.dat

If you prefer, you can also change this command a bit, so the filename comes out with the linefeed "evaluated".

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ find . -maxdepth 1 -type f -print0 | \
sed 's#\./.*/\([^/]\+\)\./$#\1#g' | tr '\n' '/' | \
perl -F'/\0/' -ape \
  '$_=join("\n", sort { length($b) <=> length($a) } @F)' | \
sed 's#/#\n#g' | tail -8
LOqU.txt
137.csv
w
lf.aa
13.csv
o6.dat
3.csv
a.dat

In either case, you will know, due to what we've been doing, that the list is sorted, even though it doesn't appear so.

(Visual on not appearing sorted by filename length)

********
********
*******
**********       <-- Visual Problem
*****
*****
****
****

OR

********
*******
*                <-- Visual
****             <-- Problems
*****
*****
****
****

• Works for filenames containing newlines

---

* 2a. Very Close, but Doesn't Keep Newline Filename Together - à la @cpasm

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ for i in *; do printf "%d\t%s\n" "${#i}" "$i"; done | sort -n | cut -f2- | head
lf.aa
3.csv
a.dat
13.csv
o6.dat
137.csv
w
1UG5.txt
1uWj.txt
2Ese.txt

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ for i in *; do printf "%d\t%s\n" "${#i}" "$i"; done | sort -n | cut -f2- | tail -5
83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt
79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
just one file with a really long filename that can throw off some counts bla so there
oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt

Note, for the head part, that the w in

w(\n)
lf.aa

is in the correct, sorted position for the 6-character-long filename that it is. However, the lf.aa is not in a logical place.

---

* Less-Easily Breakable (only '\n' and possibly command characters could be a problem)

1b. Perl à la @tchrist with find, not ls

Using find with null separator and xargs.

Command:

find . -maxdepth 1 -type f -print0 | xargs -I'{}' -0 echo "{}" | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | perl -e 'print sort { length($b) <=> length($a) } <>'

Alternate version of command that's easier to read:

find . -maxdepth 1 -type f -print0 | \
  xargs -I'{}' -0 \
    echo "{}" | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | \
        perl -e 'print sort { length($b) <=> length($a) } <>'

Let's go for it.

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ find . -maxdepth 1 -type f -print0 | \
  xargs -I'{}' -0 \
    echo "{}" | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | \
      perl -e 'print sort { length($b) <=> length($a) } <>' | head -n 5
68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt
oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt
17f09d51d6280fb8393d5f321f344f616c461a57a8b9cf9cc3099f906b567c992.txt

bballdave025@MY-MACHINE /home/bballdave025/orig_dir_73
$ find . -maxdepth 1 -type f -print0 | \
  xargs -I'{}' -0 \
    echo "{}" | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | \
      perl -e 'print sort { length($b) <=> length($a) } <>' | tail -8
IKlT.txt
Lk3f.png
LOqU.txt
137.csv
13.csv
o6.dat
3.csv
a.dat

• Works for normal filenames

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces
$ find . -maxdepth 1 -type f -print0 | \
  xargs -I'{}' -0 \
    echo "{}" | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | \
      perl -e 'print sort { length($b) <=> length($a) } <>' | head -n 5
68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt
oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
just one file with a really long filename that can throw off some counts bla so there
79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt

• Works for filenames containing spaces

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ find . -maxdepth 1 -type f -print0 | \
  xargs -I'{}' -0 \
    echo "{}" | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | \
      perl -e 'print sort { length($b) <=> length($a) } <>' | head -n 5
68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt
oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
just one file with a really long filename that can throw off some counts bla so there
79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ find . -maxdepth 1 -type f -print0 | \
  xargs -I'{}' -0 \
    echo "{}" | sed 's#\./.*/\([^/]\+\)\./$#\1#g' | 
      perl -e 'print sort { length($b) <=> length($a) } <>' | tail -8
LOqU.txt
137.csv
13.csv
o6.dat
3.csv
a.dat
lf.aa
w

WARNING

• BREAKS for filenames containing newlines

1c. Good for normal filenames and filenames with spaces, but breakable with filenames containing newlines - à la @tchrist

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ ls | perl -e 'print sort { length($b) <=> length($a) } <>' | head -n 5
68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt
oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
just one file with a really long filename that can throw off some counts bla so there
79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ ls | perl -e 'print sort { length($b) <=> length($a) } <>' | tail -8
LOqU.txt
137.csv
13.csv
o6.dat
3.csv
a.dat
lf.aa
w

3a. Good for normal filenames and filenames with spaces, but breakable with filenames containing newlines - à la @Peter_O

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ ls | awk '{print length($0)"\t"$0}' | sort -n | cut --complement -f1 | head -n 8
w
3.csv
a.dat
lf.aa
13.csv
o6.dat
137.csv
1UG5.txt

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ ls | awk '{print length($0)"\t"$0}' | sort -n | cut --complement -f1 | tail -5
83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt
79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
just one file with a really long filename that can throw off some counts bla so there
oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt

---

* More-Easily Breakable

4a. Good for normal filenames - à la @Raghuram

This version is breakable with filenames containing either spaces or newlines (or both).

I do want to add that I do like the display of the actual string length, if just for analysis purposes.

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ for i in `ls`; do LEN=`expr length $i`; echo $LEN $i; done | sort -n | head -n 20
1 a
1 w
2 so
3 bla
3 can
3 off
3 one
4 file
4 just
4 long
4 some
4 that
4 with
5 3.csv
5 a.dat
5 lf.aa
5 there
5 throw
6 13.csv
6 counts

bballdave025@MY-MACHINE /home/bballdave025/dir_w_fnames__spaces_and_newlines
$ for i in `ls`; do LEN=`expr length $i`; echo $LEN $i; done | sort -n | tail -5
69 17f09d51d6280fb8393d5f321f344f616c461a57a8b9cf9cc3099f906b567c992.txt
70 83dfee2e0f8560dbd2a681a5a40225fd260d3b428b962dcfb75d17e43a5fdec9_1.txt
76 79496d6167652f71526c586e654345744a365939773d3d2d343538sfp6m8o1m53hlwlfja.dat
87 oinwrxK2ea1sfp6m8o49255f679496d6167652f71526c586e654345744a365939773d3d2d343538b3e0.csv
238 68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f776174747061642d6d656469612d736572766963652f53746f7279496d6167652f71526c586e654345744a365939773d3d2d3435383139353437362e313464633462356336326266656365303439363432373931333139382e676966.txt

---

Explanation of Some Commands

For now, I'll only note that, with the works-for-all find command, I used '/' for the newline substitute because it is the only character that is illegal in a filename both on *NIX and Windows.

---

Note(s)

^{[ 1 ]} The command used,

du --inodes --files0-from=<(find . -maxdepth 1 -type f -print0) | \
awk '{sum+=int($1)}END{print sum}'

will work in this case, because when there is a file with a newline, and therefore an "extra" line in the output of the find command, awk's int function will evaluate to 0 for the text of that link. Specifically, for our newline-containing filename, w\nlf.aa, i.e.

w
lf.aa

we will get

$ awk '{print int($1)}' < <(echo "lf.aa")
0

If you have a situation where the filename is something like

firstline\n3 and some other\n1\n2\texciting\n86stuff.jpg

i.e.

firstline
3 and some other
1
2     exciting
86stuff.jpg

well, I guess the computer has beaten me. If anyone has a solution, I'd be glad to hear it.

Edit I think I'm way too deep into this question. from this SO answer and experimentation, I got this command (I don't understand all the details, but I've tested it pretty well.)

awk -F"\0" '{print NF-1}' < <(find . -maxdepth 1 -type f -print0) | awk '{sum+=$1}END{print sum}'

More readably:

awk -F"\0" '{print NF-1}' < \
  <(find . -maxdepth 1 -type f -print0) | \
    awk '{sum+=$1}END{print sum}'

Answer 6

You can use

ls --color=never --indicator-style=none | awk '{print length, $0}' |
sort -n | cut -d" " -f2-

To see it in action, create some files

% touch a ab abc

and some directories

% mkdir d de def

Output of the normal ls command

% ls
a  ab  abc  d/  de/  def/

Output from the proposed command

% ls --color=never --indicator-style=none | awk '{print length, $0}' |
sort -n | cut -d" " -f2-
a
d
ab
de
abc
def