OP

0.1

Types/OpenPackage/get_LanguagePercent.ps1

                                <#

.SYNOPSIS

    Gets the language percentages of a package

.DESCRIPTION

    Gets the language percentages present in the package.

.NOTES

    Definitions of what constitutes a language have been quite contentious.

    For the purposes of accurately identifying what lies within a package, we want a very broad definition.

    If you believe a language should be included, file an issue.

    If you believe any given file format is or is not a language, do not file an issue.

#>

$LanguagesByLength = [Ordered]@{}

$totalLength = 0

$fileSizes = $this.FileSize

foreach ($part in $this.GetParts()) {            

    $partLength = $fileSizes[$part.Uri]

    $recognizedLanguage =

        switch -regex ($part.Uri) {

            '\.3mf$' { '3MF'}

            '\.astro' { 'Astro' }

            '\.c$' { 'C' }

            '\.cast$' { 'Asciiema' }

            '\.clixml$' { 'Clixml'}

            '\.cjs$' { 'Common JavaScript'}            

            '\.cpp$' { 'C++' }

            '\.cs$' { 'C# '}

            '\.csv$' { 'Comma Separated Values' }            

            '\.csh$' { 'CShell'}

            '\.css$' { 'Cascading Stylesheets' }

            '(?>/word/.+?\.xml|\.docx?)$' { 'Word '}

            '\.dll$' { 'Binary' } 

            '\.exe$' { 'Binary' }

            '\.gif$' { 'GIF' }

            '\.go$' { 'Go Language' }

            '\.h$' { 'C Header' }

            '\.html?$' { 'Hypertext Markup Language' }

            '\.java$' {'Java' }

            '\.jpe?g$' { 'Joint Pictures Expert Group'}

            '\.json$' {'JavaScript Object Notation' }

            '\.jsonc$' {'Commented JavaScript Object Notation' }

            '\.jsonl$' {'JavaScript Object Notation Lines' }

            '\.js$' { 'Javascript'}

            '\.jsx$' { 'JavaScript XML'}

            '\.(?>md|mdx|markdown)$' { 'Markdown' }

            '\.midi?$' { 'MIDI' }

            '\.(?>jsm|mjs)$' { 'JavaScript Module'}

            '\.mkv$' { 'Matroska Video'}

            '\.mka$' { 'Matroska Audio'}

            '\.mks$' { 'Matroska Subtitle'}

            '\.mk3d$' { 'Matroska Stereoscopic Video'}

            '\.mp3$' { 'MP3' }

            '\.mp4$' { 'MP4' }

            '\.nix$' { 'Nix' }

            '\.oog$' { 'OOG' }

            '\.pl$' { 'Perl' }

            '\.png$' { 'Portable Network Graphics' }

            '(?>/ppt/.+?\.xml|\.pptx?)$' { 'PowerPoint'}

            '\.psm?1$' { 'PowerShell' }

            '\.psd1$' {'PowerShell Data Language' }

            '\.ps1xml$' { 'PowerShell Xml' }

            '\.py$' { 'Python' }            

            '\.rs$' { 'Rust '}

            '\.rss$' { 'RSS' }

            '\.sh$' { 'BourneShell'}

            '\.stl$' { 'STL'}

            '\.svg$' { 'SVG' }

            '\.tar$' { 'Tarfile' }

            '(?>\.tar\.gz|\.tgz)$' { 'GZippedTarfile' }

            '\.tsx?$' { 'TypeScript' }

            '\.tsv$' { 'Tab Separated Values' }

            '\.toml$' { 'Tom''s Obvious Minimal Language' }

            '\.xhtml$' { 'XHTML' }

            '(?>/xl/.+?\.xml|\.xlsx?)$' { 'Excel'}

            '\.xsl$' { 'XSL' }

            '\.xml$' { 'XML' }

            '\.ya?ml$' { 'Yaml' }

            '\.zip$' { 'Zip' }

            '\.webm' { 'Web Movie' }

            '\.weba' { 'Web Audio' }

            '\.webp' { 'Web Photo' }

            default { 'Unknown' }

        }

    if (-not $recognizedLanguage) {

        continue

    }

    if (-not $LanguagesByLength[$recognizedLanguage]) {        

        $LanguagesByLength[$recognizedLanguage] = 0    

    }

    $LanguagesByLength[$recognizedLanguage]+=$partLength

    $totalLength += $partLength

}

$SortedByLength = [Ordered]@{}

foreach ($keyValue in $languagesByLength.GetEnumerator() | 

    Sort-Object Value -Descending

) {

    $SortedByLength[$keyValue.Key] = $keyValue.Value / $totalLength

}

return $SortedByLength