n8n-nodes-xberg

An n8n community node that runs Xberg document extraction inside your workflows. Point it at a binary file on an incoming item and it returns the extracted text, tables, and metadata — with optional OCR for scans and images.

Extraction runs locally in-process through the @xberg-io/xberg native binding. No API key, no cloud call, no data leaves your n8n instance.

Requirements

• Self-hosted n8n (native N-API addons are not available on n8n Cloud, so this node is self-hosted only).
• Node.js 20.15+ on Linux (x64/arm64, glibc or musl), macOS (arm64), or Windows (x64/arm64) — the platforms for which @xberg-io/xberg ships prebuilt binaries.

Installation

In your n8n instance go to Settings > Community Nodes > Install, enter the package name, and confirm you understand the risks of installing community code:

@xberg-io/n8n-nodes-xberg

Or install manually into a self-hosted instance:

npm install @xberg-io/n8n-nodes-xberg

Operations

The node exposes one resource, Document, with three operations.

Document › Extract

Extract content from one document per incoming item. Set Input Source to read the document from a binary property (uploaded file) or from a URL/local path.

Document › Extract Batch

Extract every incoming item in a single extractBatch call. The native binding schedules the inputs concurrently, so this is substantially faster than a loop of Extract nodes. Use it for file/binary inputs; for crawling a URL use Extract or Map URL instead. Output items are paired back to their input; per-item failures surface through Continue On Fail.

Document › Map URL

List the URLs reachable from a web page or sitemap without extracting them. Set Mode to auto, document, or crawl, and optionally cap results with Max Total URLs. Each output item's JSON holds the seed url and the discovered urls array.

Extract parameters

| Parameter | Description | | --- | --- | | Input Source | Binary Data (a binary property on the item) or URL or Path. | | Input Binary Field | Binary property holding the document (default data). Binary source only. | | URL or Path | HTTP(S) URL or local filesystem path. URL source only. | | Output Format | markdown, plain, html, djot, json, or structured. | | Enable OCR | Run OCR on images and scanned PDF pages (default on). | | Force OCR | OCR every page even when a text layer exists. | | OCR Languages | Comma-separated ISO 639-2 codes, e.g. eng,deu. | | Enable Chunking | Split the content into overlapping chunks for RAG pipelines. | | Chunk Size / Chunk Overlap | Max characters per chunk and overlap between chunks. | | Extract Images | Extract embedded images and report them in the output. | | Enable Quality Processing | Clean up extracted text with post-processing. | | Include Metadata | Attach document metadata (title, author, dates). | | Include Tables | Attach structured table data. | | Include Chunks | Attach the generated chunks (requires Enable Chunking). | | Output Content Field | JSON field to write the extracted text into (default text). | | Return As Binary | Also attach the extracted content as a binary property. |

Output

Each Extract / Extract Batch output item's JSON contains the extracted content under the configured field (default text), plus mimeType, extractionMethod, detectedLanguages, and counts (pages, tables, images). formattedContent, qualityScore, entities, and summary appear when the binding populates them. metadata, tables, and chunks are included when their toggles are set. When Return As Binary is set, the content is also attached as a binary property with a format-appropriate MIME type and extension.

Supported formats

Xberg extracts from 101 formats including PDF, DOCX, PPTX, XLSX, HTML, EPUB, images, and more. See the Xberg documentation for the full list.

Compatibility

Tested against n8n's n8n-workflow 2.x node API (n8nNodesApiVersion 1).

Part of Xberg.io

• Xberg — document intelligence: text, tables, metadata from 101 formats with optional OCR.
• Xberg Enterprise — managed extraction API with SDKs, dashboards, and observability.

License

MIT