Support for files embedded in XML

[MarcMorisse]

Because of deployement of e-invoicing in Europe, that embbeds various files inside the XML, I was wondering if it would be difficult to add XML support to Media-extractor, as it seems perfect to easily extract the embedded files from the XML's.
I can quickly provide many examples of such XML if needed.
And of course, if I misunderstand the purpose of Media-Extractor, just let me know ;-)

[rabanti-github]

Hello,
In principle, you can open the most common archive files with MediExtractor, like:

• zip
• 7z
• jar
• tar
• rar (to some extent)
• office files (docx, xlsx, pptx etc.) --> Office 2003 or newer

But XML itself should be like a text file. Do you mean another format? If you can provide an example, maybe I can help you.

In MediaExtractor, there are also two options to show non-media (images) files.

• The option "Show other embedded Files" will reveal all files in an archive
• The option "Text Preview on unknown File Formats" will force the preview window to show a marked file as text

[MarcMorisse]

Hi Raphael,
Yes you are correct, XML is a text file, but it can be used to embed any binary file using Base64.
In case of e-invoicing, all aspects of the invoice is described by the various tags of the XML, but also attachments can be included, usually a PDF version of the invoice, but it could also be office files (docx, xlsx, pptx, ...), images (jpg, png, ...), text or anything.
The idea would be to support XML by Media-Extrator to extract the Embedded files, since you already have the nice UI, Preview, Drag & Drop and all required functionalities available?
I attached a few sample XML for you:
general_bundle_real_data.xml
general_sample.xml
INVOICE_ubl2-3.xml
ubl_invoice_real_data.xml
ubl_invoice_sample.xml

[rabanti-github]

Hello Marc,
...Oh, that throws me a bit back to the UN/EDIFACT tasks I had to do some years back 😂
Well, the usecase here is a bit special, since MediaExtractor would need to know the tag structure. Currently it is either an image preview for supported image types, a plain text preview or a formatted, raw XML preview (as this was a requirement to inspect XLSX files). An implementation of inline content is a bit difficult, since it would need, beside the tag knowledge, also new viewers for inline PDFs inline DOCX, Inline XLSX, etc.

But the good news are, that the content from these files is very easy to extract by some lines of PowerShell:

$xmlPath = "C:\temp\input\INVOICE_ubl2-3.xml"
$outputFolder = "C:\temp\output"

[xml]$xml = Get-Content -Path $xmlPath -Raw

$ns = New-Object System.Xml.XmlNamespaceManager($xml.NameTable)
$ns.AddNamespace("cbc", "urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2")

$nodes = $xml.SelectNodes("//cbc:EmbeddedDocumentBinaryObject", $ns)

foreach ($node in $nodes) {

    $fileName = $node.GetAttribute("filename")
    $base64 = $node.InnerText.Trim()

    $bytes = [Convert]::FromBase64String($base64)

    $outputPath = Join-Path $outputFolder $fileName

    [System.IO.File]::WriteAllBytes($outputPath, $bytes)

    Write-Host "Extracted: $outputPath"

This is just a hard-coded PoC for embedded PDFs, but based on this, several types of extraction scripts could be created, like:

• Hotfolder processing
• Processing by CLI / command line parameters
• Batch processing where several scrips are used to cover all content types
• etc

For the other examples, probably a multi-line recognition of tags would be necessary to fetch metadata (Filename, MimeType, maybe ContentEncoding) and encoded content (BinaryData).

So, PowerShell could still be easy, but in my opinion, to be more flexible, I would create a very simple console app (maybe with CLI, or hotfolder processing), that looks for a specific chain of tags (e.g. DataExChange -> Payload -> Attachment -> Filename | MimeType | BinaryData )