xberg-io/xberg

A polyglot document intelligence framework with a Rust core. Extract text, metadata, images, and structured information from PDFs, Office documents, images, and 98+ formats. Available for Rust, Python, Ruby, Java, Go, PHP, Elixir, C#, R, C, TypeScript (Node/Bun/Wasm/Deno)- or use via CLI, REST API, or MCP server.

What it solves

Modern data pipelines often struggle with the complexity of parsing diverse file formats, ranging from PDFs and spreadsheets to audio files and complex source code. Xberg provides a unified engine to extract clean, structured text and metadata without requiring developers to assemble fragmented libraries or set up complex OCR pipelines manually.

How it works

Xberg uses a single core engine to perform intelligent MIME detection and extraction across 98 formats. It supports various processing modes including:

  • Extraction: Converts documents, images, audio, and code into structured formats like Markdown or JSON.
  • OCR & Transcription: Uses backends like Tesseract, PaddleOCR, or Whisper (via ONNX) for visual and audio content.
  • Code Intelligence: Employs tree-sitter to perform syntax-aware chunking and symbol extraction for 306 programming languages.
  • Enrichment: Provides NER, summarization, translation, and table extraction.
  • Deployment: Operates as a library (with 15 language bindings), a CLI, a REST API, or an MCP server.

Who it’s for

  • AI/LLM Developers: Building RAG pipelines that require syntax-aware code chunking or structured document ingestion.
  • Data Engineers: Building automated pipelines to crawl, batch-process, and extract data from massive document archives.
  • Software Engineers: Integrating advanced document parsing and OCR capabilities into existing applications via high-performance language bindings.

Highlights

  • Massive Format Support: Handles 98 formats including Office docs, images, audio, video, and archives.
  • Multi-Language Bindings: Native support for Rust, Python, Node.js, Go, Java, C#, Ruby, PHP, Elixir, Dart, Swift, Zig, and WASM.
  • RAG-Ready: Features syntax-aware code chunking and built-in embedding/reranking support.
  • Flexible Deployment: Can be run as a local library, a Docker container, a REST API, or an MCP server for AI assistants like Claude or Cursor.