make_index.pl

#!/usr/bin/perl
use strict;
use warnings;
use utf8;
use Encode qw(decode_utf8 encode_utf8);
use File::Basename;
use Cwd qw(cwd abs_path);
use Getopt::Long qw(:config pass_through);

binmode STDOUT, ':encoding(UTF-8)';

# ---------------------------------------------------------------------------
# Translations
# ---------------------------------------------------------------------------
my %i18n = (
    en => { index_of => 'Index of',    folders => 'folder(s)',  files => 'file(s)'      },
    fr => { index_of => 'Index de',    folders => 'dossier(s)', files => 'fichier(s)'   },
    de => { index_of => 'Index von',   folders => 'Ordner',     files => 'Datei(en)'    },
    es => { index_of => 'Índice de',   folders => 'carpeta(s)', files => 'archivo(s)'   },
    it => { index_of => 'Indice di',   folders => 'cartella(e)',files => 'file'         },
    pt => { index_of => 'Índice de',   folders => 'pasta(s)',   files => 'arquivo(s)'   },
    nl => { index_of => 'Index van',   folders => 'map(pen)',   files => 'bestand(en)'  },
    pl => { index_of => 'Indeks',      folders => 'folder(y)',  files => 'plik(i)'      },
    ru => { index_of => 'Индекс',      folders => 'папок',      files => 'файлов'       },
    ja => { index_of => 'インデックス：', folders => 'フォルダ', files => 'ファイル'    },
    zh => { index_of => '索引：',      folders => '个文件夹',   files => '个文件'       },
);

# ---------------------------------------------------------------------------
# Options
# ---------------------------------------------------------------------------
my ($custom_title, $no_recursive, $lang_code, $dark, $help);
GetOptions(
    't|title=s'      => \$custom_title,
    'n|no-recursive' => \$no_recursive,
    'l|language=s'   => \$lang_code,
    'd|dark'         => \$dark,
    'h|help'         => \$help,
);

$lang_code //= 'en';
unless (exists $i18n{$lang_code}) {
    warn "Unknown language '$lang_code'. Available: " . join(', ', sort keys %i18n) . ". Falling back to 'en'.\n";
    $lang_code = 'en';
}
my %t = %{ $i18n{$lang_code} };

if ($help) {
    (my $s = basename($0)) =~ s/\.pl$//;
    print <<HELP;
$s — Generates an HTML index page listing files and directories.

Usage: $s [options] [directory]

Options:
  -t, --title TITLE    Title displayed at the top of the page (default: folder name)
  -n, --no-recursive   Do not recurse into subdirectories (default: recursive)
  -l, --language CODE  Language for page output: en fr de es it pt nl pl ru ja zh (default: en)
  -d, --dark           Use dark theme
  -h, --help           Show this help

Examples:
  $s                                      # current directory, recursive
  $s -n                                   # current directory, top level only
  $s --no-recursive --title "My files" /my/path
  $s -l fr -t "Mes fichiers"
HELP
    exit 0;
}

# ---------------------------------------------------------------------------
# Main
# ---------------------------------------------------------------------------
my $root    = @ARGV ? abs_path($ARGV[0]) : cwd();
$root =~ s|/+$||;

my $title   = $custom_title // basename($root);
my $heading = $custom_title ? $custom_title : "$t{index_of} $title";
my $outfile = "$root/index.html";
my $script  = basename($0);

my ($total_dirs, $total_files) = (0, 0);

sub human_size {
    my ($bytes) = @_;
    my @units = ('B', 'KB', 'MB', 'GB', 'TB');
    my $i = 0;
    while ($bytes >= 1024 && $i < $#units) { $bytes /= 1024; $i++ }
    return $i == 0 ? "${bytes} $units[$i]" : sprintf("%.1f %s", $bytes, $units[$i]);
}

sub file_icon {
    my ($name) = @_;
    my ($ext) = $name =~ /\.([^.]+)$/;
    $ext = lc($ext // '');

    my %icons = (
        # Images
        jpg   => '🖼️', jpeg => '🖼️', png  => '🖼️', gif  => '🖼️',
        bmp   => '🖼️', svg  => '🖼️', webp => '🖼️', ico  => '🖼️',
        tiff  => '🖼️', tif  => '🖼️', heic => '🖼️', raw  => '🖼️',

        # Videos
        mp4   => '🎬', avi  => '🎬', mkv  => '🎬', mov  => '🎬',
        wmv   => '🎬', flv  => '🎬', webm => '🎬', m4v  => '🎬',
        mpg   => '🎬', mpeg => '🎬', ts   => '🎬', divx => '🎬',

        # Audio
        mp3   => '🎵', wav  => '🎵', flac => '🎵', ogg  => '🎵',
        m4a   => '🎵', aac  => '🎵', wma  => '🎵', opus => '🎵',

        # PDF
        pdf   => '📕',

        # Text documents
        doc   => '📝', docx => '📝', odt  => '📝', rtf  => '📝',

        # Spreadsheets
        xls   => '📊', xlsx => '📊', ods  => '📊', csv  => '📊',

        # Presentations
        ppt   => '📋', pptx => '📋', odp  => '📋',

        # Archives
        zip   => '📦', tar  => '📦', gz   => '📦', bz2  => '📦',
        '7z'  => '📦', rar  => '📦', xz   => '📦', zst  => '📦',
        tgz   => '📦', tbz2 => '📦',

        # Code
        pl    => '💻', py   => '💻', js   => '💻', ts   => '💻',
        php   => '💻', c    => '💻', cpp  => '💻', h    => '💻',
        java  => '💻', rb   => '💻', go   => '💻', rs   => '💻',
        cs    => '💻', swift=> '💻', kt   => '💻', lua  => '💻',
        r     => '💻', m    => '💻',

        # Web
        html  => '🌐', htm  => '🌐', css  => '🌐', jsx  => '🌐',
        tsx   => '🌐', vue  => '🌐',

        # Shell scripts
        sh    => '⚙️', bash => '⚙️', zsh  => '⚙️', fish => '⚙️',
        bat   => '⚙️', ps1  => '⚙️',

        # Config / data
        json  => '🔧', yaml => '🔧', yml  => '🔧', toml => '🔧',
        xml   => '🔧', ini  => '🔧', conf => '🔧', cfg  => '🔧',
        env   => '🔧',

        # Databases
        db    => '🗄️', sqlite => '🗄️', sql => '🗄️',

        # Plain text / docs
        txt   => '📄', md   => '📄', rst  => '📄', log  => '📄',

        # Fonts
        ttf   => '🔤', otf  => '🔤', woff => '🔤', woff2 => '🔤',

        # Executables / binaries
        exe   => '🖥️', run  => '🖥️', bin  => '🖥️', apk  => '🖥️',
        deb   => '🖥️', rpm  => '🖥️', dmg  => '🖥️', iso  => '🖥️',
    );

    return $icons{$ext} // '📄';
}

sub url_encode {
    my ($str) = @_;                          # decoded Unicode string
    my $bytes = encode_utf8($str);           # back to UTF-8 bytes
    $bytes =~ s/([^A-Za-z0-9\-_.~\/])/sprintf("%%%02X", ord($1))/ge;
    return $bytes;
}

# Scans $abs_dir, writes to $fh, links relative to $root
sub scan_dir {
    my ($fh, $abs_dir, $indent) = @_;
    my $pad = "  " x $indent;

    opendir(my $dh, $abs_dir) or do { warn "Cannot open '$abs_dir': $!\n"; return };
    my @entries = sort grep { !/^\./ } readdir($dh);
    closedir($dh);

    my @dirs  = grep { -d "$abs_dir/$_" } @entries;
    my @files = grep { -f "$abs_dir/$_" } @entries;

    for my $d (@dirs) {
        my $d_display = decode_utf8($d);
        $total_dirs++;
        if (!$no_recursive) {
            print $fh "$pad<li><span class=\"dir\">📁 $d_display</span>\n$pad  <ul>\n";
            scan_dir($fh, "$abs_dir/$d", $indent + 2);
            print $fh "$pad  </ul>\n$pad</li>\n";
        } else {
            print $fh "$pad<li><span class=\"dir\">📁 $d_display</span></li>\n";
        }
    }

    for my $f (@files) {
        next if $abs_dir eq $root && ($f eq 'index.html' || $f eq $script);
        my $rel       = abs_path("$abs_dir/$f") =~ s|^\Q$root\E/||r;
        my $href      = url_encode(decode_utf8($rel));
        my $f_display = decode_utf8($f);
        my $size      = human_size(-s "$abs_dir/$f");
        $total_files++;
        my $icon = file_icon($f);
        print $fh "$pad<li>$icon <a href=\"$href\">$f_display</a><span class=\"size\">$size</span></li>\n";
    }
}

open(my $fh, '>:encoding(UTF-8)', $outfile) or die "Cannot write '$outfile': $!\n";

print $fh <<"HTML";
<!DOCTYPE html>
<html lang="$lang_code">
<head>
  <meta charset="UTF-8">
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
  <title>$heading</title>
  <style>
    :root {
      --bg:      #ffffff;
      --fg:      #222222;
      --link:    #0969da;
      --dir:     #444444;
      --size:    #888888;
      --border:  #cccccc;
      --tree:    #bbbbbb;
      --footer:  #888888;
    }
    body.dark {
      --bg:      #0d1117;
      --fg:      #e6edf3;
      --link:    #58a6ff;
      --dir:     #adbac7;
      --size:    #768390;
      --border:  #30363d;
      --tree:    #4d5a6e;
      --footer:  #768390;
    }
    body  { font-family: sans-serif; max-width: 900px; margin: 2em auto;
            color: var(--fg); background: var(--bg); }
    h1    { border-bottom: 2px solid var(--border); padding-bottom: .4em; }

    ul.tree, ul.tree ul {
      list-style: none;
      padding-left: 1.6em;
      margin: .05em 0;
    }
    ul.tree { padding-left: .4em; }

    ul.tree li {
      position: relative;
      padding: .15em 0;
    }
    /* horizontal arm ─── */
    ul.tree li::before {
      content: "";
      position: absolute;
      left: -1.3em;
      top: .68em;
      width: 1.1em;
      border-top: 1px solid var(--tree);
    }
    /* vertical bar │ — half-height on last child, full on others */
    ul.tree li::after {
      content: "";
      position: absolute;
      left: -1.3em;
      top: 0;
      height: .68em;
      border-left: 1px solid var(--tree);
    }
    ul.tree li:not(:last-child)::after { height: 100%; }

    /* no connector lines on root items */
    ul.tree > li::before,
    ul.tree > li::after { display: none; }

    a       { text-decoration: none; color: var(--link); }
    a:hover { text-decoration: underline; }
    .dir    { font-weight: bold; color: var(--dir); }
    .size   { color: var(--size); font-size: .85em; margin-left: .8em; }
    footer  { margin-top: 2em; font-size: .8em; color: var(--footer); }
  </style>
</head>
<body${\ ($dark ? ' class="dark"' : '')}>
  <h1>$heading</h1>
  <ul class="tree">
HTML

scan_dir($fh, $root, 2);

print $fh <<"HTML";
  </ul>
  <footer>$total_dirs $t{folders}, $total_files $t{files}</footer>
</body>
</html>
HTML

close($fh);
print "index.html generated in: $outfile\n";
print "  $total_dirs folder(s), $total_files file(s)\n";