This projects aims to create RELAX NG schemas to validate XMP metadata packets in PDF and PDF/A compliances. The source is a custom RELAX NG schema with simple extensions to allow to customize the template for different flavors (eg. different the PDF/A parts/revisions). The main one is a condition attribute that can be used to select nodes using a XSLT compatible expression, so for example the pdf:Trapped property below:

<rng:optional condition="$IsPDFA4OrGreater">
    <rng:ref name="pdf.Trapped"/>
</rng:optional>

... is enabled for a PDF/A-4 schema, while all the pdfaExtension properties below:

<rng:interleave>
    <!-- ... -->
    <rng:ref name="XMP_Properties-pdfaExtension" condition="$IsPDFA1 or $IsPDFA2 or $IsPDFA3"/>
</rng:interleave>

... are enabled for PDF/A-1, PDF/A-2 and PDF/A-3 schemas. The schemas are then garbage collected so only recursivelly referenced nodes from the <rng:start> element are preserved in the final generated schema.

If you want to look at the template development follows the steps below:

• Download and install .NET SDK¹ and Powershell² (Linux, macOS, and Windows);
• Run pwsh generate-schemas.ps1
• Have a look at the schema template starting with Schemas/XMP_FullPacket.rng, plus the RNGMerger and XMPSchema.Test C# projects.

The current snapshots targeting PDF/A compliances can be found at the following locations:

• PDF/A-1: ISO19005-1-XMP_Packet.rng
• PDF/A-2 and PDF/A-3: ISO19005-2_3-XMP_Packet.rng
• PDF/A-4: ISO19005-4-XMP_Packet.rng

The schemas are meant to be used to actually validate XMP packets following the normalization procedure as explained in chapter 5 "Canonical serialization of XMP" of ISO 16684-2:2014. An implementation of the algorithm can be found in the PoDoFo library.