File size distribution: Difference between revisions

Is your file system predominantly devoted to a large number of smaller files, or a smaller number of huge files?

<br><br>

 

=={{header|C}}==

The platform independent way to get the file size in C involves opening every file and reading the size. The implementation below works for Windows and utilizes command scripts to get size information quickly even for a large number of files, recursively traversing a large number of directories. Both textual and graphical ( ASCII ) outputs are shown. The same can be done for Linux by a combination of the find, ls and stat commands and my plan was to make it work on both OS types, but I don't have access to a Linux system right now. This would also mean either abandoning scaling the graphical output in order to fit the console buffer or porting that as well, thus including windows.h selectively.

===Windows===

#include<windows.h>

#include<string.h>

}

}

Invocation and textual output :

<pre>

{{libheader|POSIX}}

This works on macOS 10.15. It should be OK for Linux as well.

#include <locale.h>

#include <stdint.h>

printf("Total file size: %'lu\n", total_size);

return EXIT_SUCCESS;

 

{{out}}

 

=={{header|C++}}==

#include <array>

#include <filesystem>

}

return EXIT_SUCCESS;

 

{{out}}

{{libheader| Winapi.Windows}}

{{Trans|Go}}

program File_size_distribution;

 

fileSizeDistribution('.');

readln;

 

=={{header|Factor}}==

{{works with|Factor|0.99 2020-03-02}}

io.files.types io.pathnames kernel math math.functions

math.statistics namespaces sequences ;

current-directory get file-size-histogram dup

[ "Count of files < 10^%d bytes: %4d\n" printf ] assoc-each

{{out}}

<pre>

=={{header|Go}}==

{{trans|Kotlin}}

 

import (

func main() {

fileSizeDistribution("./")

 

{{out}}

Uses a grouped frequency distribution. Program arguments are optional. Arguments include starting directory and initial frequency distribution group size. After the first frequency distribution is computed it further breaks it down for any group that exceeds 25% of the total file count, when possible.

</p>

 

import Control.Concurrent (forkIO, setNumCapabilities)

mapM_ (displayFrequency fileCount) $ Map.assocs results

where

{{out}}

<pre style="height: 50rem;">$ filedist ~/Music

16.00MB <-> 18.67MB = 3 0.436%: ▍

24.00MB <-> 26.66MB = 1 0.145%: ▍

</pre>

 

{{works with|Julia|0.6}}

 

 

function sizelist(path::AbstractString)

end

 

 

{{out}}

 

=={{header|Kotlin}}==

 

import java.io.File

fun main(args: Array<String>) {

fileSizeDistribution("./") // current directory

 

{{out}}

Number of inaccessible files : 0

</pre>

 

=={{header|Nim}}==

 

const

echo fmt"Size in {rangeString: 14} {count:>7} {100 * count / total:5.2f}%"

echo ""

 

{{out}}

=={{header|Perl}}==

{{trans|Raku}}

use List::Util qw(max);

 

 

sub fsize { $fsize{ log10( (lstat($_))[7] ) }++ }

{{out}}

<pre>File size distribution in bytes for directory: .

=={{header|Phix}}==

Works on Windows and Linux. Uses "proper" sizes, ie 1MB==1024KB. Can be quite slow at first, but is pretty fast on the second and subsequent runs, that is once the OS has cached its (low-level) directory reads.

{{out}}

<pre>

The distribution is stored in a '''collections.Counter''' object (like a dictionary with automatic 0 value when a key is not found, useful when incrementing). Anything could be done with this object, here the number of files is printed for increasing sizes. No check is made during the directory walk: usually, safeguards would be needed or the program will fail on any unreadable file or directory (depending on rights, or too deep paths, for instance). Here links are skipped, so it should avoid cycles.

 

from collections import Counter

 

print("Total %d bytes for %d files" % (s, n))

 

 

=={{header|Racket}}==

 

 

(define (file-size-distribution (d (current-directory)) #:size-group-function (sgf values))

(module+ test

(call-with-values (λ () (file-size-distribution #:size-group-function log10-or-so))

 

{{out}}

By default, process the current and all readable sub-directories, or, pass in a directory path at the command line.

 

sub log10 (Int $s) { $s ?? $s.log(10).Int !! 0 }

my %fsize;

my ($end, $bar) = $scaled.polymod(8);

(@blocks[8] x $bar * 8) ~ (@blocks[$end] if $end) ~ "\n"

 

{{out}}

 

Also, some Windows versions of the &nbsp; '''dir''' &nbsp; command insert commas into numbers, so code was added to elide them.

numeric digits 30 /*ensure enough decimal digits for a #.*/

parse arg ds . /*obtain optional argument from the CL.*/

exit /*stick a fork in it, we're all done. */

/*──────────────────────────────────────────────────────────────────────────────────────*/

This REXX program makes use of &nbsp; '''LINESIZE''' &nbsp; REXX program (or BIF) which is used to determine the screen width (or linesize) of the terminal (console) so as to maximize the width of the histogram.

 

{{libheader|walkdir}}

{{works with|Rust|2018}}

use std::error::Error;

use std::marker::PhantomData;

}

}

{{out}}

<pre>

 

=={{header|Sidef}}==

dir.open(\var dir_h) || return nil

 

}

 

{{out}}

<pre>

=={{header|Tcl}}==

This is with the '''fileutil::traverse''' package from Tcllib to do the tree walking, a '''glob''' based alternative ignoring links but not hidden files is possible but would add a dozen of lines.

namespace path {::tcl::mathfunc ::tcl::mathop}

 

foreach key [lsort -int [dict keys $hist]] {

puts "[? {$key == -1} 0 {1e$key}]\t[dict get $hist $key]"

{{out}}

<pre>0 1

{{works with|Bourne Shell}}

Use POSIX conformant code unless the environment variable GNU is set to anything not empty.

set -eu

 

if [ ${GNU:-} ]

then

# Use a subshell to remove the last "total" line per each ARG_MAX

find -- "${1:-.}" -type f -exec sh -c 'wc -c -- "$@" | sed \$d' argv0 {} +

{

++hist[$1 ? length($1) - 1 : -1]

}

END {

for (i in hist)

{{out}}

<pre>$ time ~/fsd.sh

$ time GNU=1 ~/fsd.sh

 

=={{header|zkl}}==

// hoover all files in tree, don't return directories

fcn(pipe,dir){ File.globular(dir,"*",True,8,pipe); }

println("%15s : %s".fmt(szchrs[idx,*], "*"*(scale*cnt).round().toInt()));

idx-=1 + comma();

{{out}}

<pre>