Workflow 8 min read 19 October 2026

Embedded Files and Attachments in PDF: PDF/A-3, ZUGFeRD, and File Attachment Annotations

PDFs can carry embedded files of any type — XML invoices, spreadsheets, source documents — either via PDF/A-3 associated files or file attachment annotations. This guide explains the two mechanisms, their dictionary structure, and use cases including ZUGFeRD/Factur-X hybrid invoices.

PDFs can carry embedded files of any type — XML invoices, spreadsheets, Word documents, images, CAD drawings, source code — inside the PDF container, accessible alongside or independently from the PDF's visible pages. This capability is at the heart of modern hybrid document workflows: a single PDF file can be both the human-readable presentation of a document and a machine-readable data package. There are two distinct mechanisms for embedding files in a PDF: PDF/A-3 associated files (using the /AF and /AFRelationship entries defined in PDF 2.0) and file attachment annotations (the traditional mechanism). Both result in embedded binary data within the PDF, but they serve different purposes and are supported differently by viewers and validators.

File attachment annotations vs associated files

File Attachment Annotations are the traditional mechanism, introduced in PDF 1.0. They create a small paperclip or push-pin icon on the page, which opens the embedded file when clicked. The embedded file is stored as an embedded file stream in the document's /EmbeddedFiles name tree, and the annotation references it. /Subtype /FileAttachment — The annotation subtype /FS — File specification dictionary referencing the embedded file stream /Name — Icon type: /Paperclip, /PushPin, /Graph, /Tag /Contents — Tooltip text (the attachment description) File attachment annotations are appropriate when the embedded file is supplementary to the PDF content — for example, a spreadsheet supporting a report's tables, or a CAD drawing alongside a technical specification. Associated Files (PDF 2.0 / PDF/A-3) use the /AF entry, which can appear in the document catalogue, a page dictionary, or any other PDF object that the associated file relates to. The associated file's /AFRelationship entry declares the semantic relationship: /Source — The embedded file is the source from which the PDF was generated /Data — The embedded file provides data represented in the PDF /Alternative — The embedded file is an alternative representation of the PDF content /Supplement — The embedded file supplements the PDF content /EncryptedPayload — The embedded file is an encrypted version of the PDF content (PDF 2.0) /FormData — The embedded file contains submitted form data /Unspecified — The relationship is unspecified Associated files are the mechanism used by ZUGFeRD/Factur-X for embedded XML invoices, and are the appropriate choice whenever the embedded file has a defined semantic relationship to the PDF.

ZUGFeRD and Factur-X: hybrid PDF invoices

ZUGFeRD (Zentraler User Guide des Forums elektronische Rechnung Deutschland) and its French equivalent Factur-X are the dominant use cases for PDF/A-3 associated files. A ZUGFeRD 2.1 / Factur-X invoice is a PDF/A-3 document where: 1. The PDF envelope is a conforming PDF/A-3b (or PDF/A-3a) document with all the usual archival constraints — embedded fonts, no encryption, no JavaScript. 2. The embedded XML file is attached with /AFRelationship /Data — it contains the machine-readable invoice data conforming to the EN 16931 European Invoice Core Data Model. 3. The XMP metadata in the PDF includes extensions that identify the ZUGFeRD profile and version. The XML profiles range from MINIMUM (basic identifying information only) through BASIC, EN 16931 (the full CII D16B compliant profile), and EXTENDED (additional fields for specific industries). The German federal XRechnung standard uses a subset of the EN 16931 profile. Creating ZUGFeRD PDFs: - Java: Mustang Project (open-source) - Python: factur-x library - .NET: ZUGFeRD-csharp - SAP, Oracle, and major ERP systems have ZUGFeRD/Factur-X connectors Validating ZUGFeRD PDFs — Mustang Project Validator (online/offline) validates both the PDF/A-3 compliance and the embedded XML Schematron constraints. veraPDF also validates PDF/A-3 compliance of ZUGFeRD containers.

Scientific and technical embedded file use cases

Beyond invoicing, embedded files in PDFs are used across scientific and technical domains: Academic publishing — Journal publishers increasingly embed supplementary data files (CSV datasets, R scripts, Python notebooks, raw instrument data) inside PDF articles using associated files. The PDF remains the human-readable article; the embedded files provide the underlying research data for reproducibility. Some publishers use PDF/A-3 for this purpose. Engineering documentation — ISO 32000-2 permits embedding STEP (AP214) CAD files, FEM results, and measurement data alongside technical drawings in a PDF. Manufacturers can distribute a single PDF that contains both the drawing and the associated CAD data — receivers with appropriate software can extract and use the CAD data directly. Medical records — CDA/C-CDA clinical documents can be embedded as associated files in PDF/A-3 medical record PDFs, creating a durable archive that is both human-readable (the PDF) and machine-readable (the CDA XML) in a single file. Software deployment packages — Some software distribution workflows embed checksums, digital signatures, or metadata XML files alongside installer PDFs as an integrity verification mechanism. Limitations: - PDF/A-1 and PDF/A-2 prohibit arbitrary embedded file attachments (only ICC profiles and XMP are allowed) - PDF/A-3 specifically adds embedded file capability to PDF/A-2 with the /AFRelationship constraint - Embedded files increase PDF file size; very large embedded files can make PDF viewers slow to open - Some corporate DLP (Data Loss Prevention) systems flag PDFs with embedded files as potential data exfiltration vectors

Frequently asked questions

How do I extract embedded files from a PDF?

Several methods: (1) Adobe Acrobat Reader/Pro: View > Show/Hide > Navigation Panes > Attachments opens the attachments panel listing all embedded files. Click a file and use the Save icon to extract it. (2) Command-line with pdftk: pdftk input.pdf unpack_files output_directory/ extracts all embedded files. (3) Python with pypdf: iterate over reader.attachments — each entry is a filename-to-bytes mapping. (4) pdfdetach (part of poppler-utils): pdfdetach -list input.pdf shows all attachments; pdfdetach -saveall input.pdf -o output_dir/ extracts them. For ZUGFeRD/Factur-X PDFs specifically, the factur-x Python library has a dedicated extraction function that identifies and extracts the EN 16931 XML.

What is the maximum size for a PDF embedded file?

The PDF specification does not impose a hard limit on embedded file size — theoretically, a PDF can embed files of any size. In practice, the limits are: (1) PDF file size limits imposed by the destination system (email attachment limits, file system limits, document management system configuration). (2) Viewer performance — very large embedded files can cause slow PDF open times, since some viewers load the entire PDF into memory including embedded files. (3) PDF/A-3 conformance is not affected by embedded file size, but very large PDFs may be impractical for archival storage systems. (4) Base64 encoding — if embedding files via some PDF creation libraries that Base64-encode the file data, the embedded stream is approximately 33% larger than the source file. PDF streams should use binary encoding (ASCIIHexDecode or raw binary) not Base64 for efficiency.

Do antivirus tools scan embedded files in PDFs?

Yes, most enterprise antivirus and EDR (Endpoint Detection and Response) solutions are PDF-aware and scan embedded file attachments within PDFs. ClamAV (open-source) has specific signatures for malicious embedded executables in PDFs. Windows Defender, ESET, and Sophos also inspect PDF embedded files. However, detection is not 100%: zero-day malware embedded in PDFs may not be detected by signature-based scanners. For organisations handling PDFs from untrusted sources, a defence-in-depth approach is appropriate: AV scanning + disabling PDF JavaScript (removing the JS attack surface) + sandboxed PDF viewing for high-risk attachments. Corporate DLP systems can also be configured to block PDFs with embedded executable file types (.exe, .msi, .bat, .ps1) from crossing the email perimeter.

Try it yourself

Everything in this article can be done directly in the free PDFTheGuide editor — no account required.

Edit PDF Protect PDF Open editor