Pdfmathtranslate is a specialized tool for translating scientific papers and other PDF documents, focusing on preserving complex layouts and mathematical formulas. Built primarily with Python, it offers flexible distribution, including a Python package installable via pip or uv, Windows executable files in zip archives, and Docker images. Users can interact with Pdfmathtranslate through a Graphic User Interface (GUI), a Web UI, or its command-line interface. This program is designed to connect with various AI translation services, providing adaptable solutions for researchers and students.

What does Pdfmathtranslate connect to?

Pdfmathtranslate connects to a diverse array of AI translation backends and services, including Google, DeepL, Ollama, OpenAI, Anthropic's Claude, and many others that comply with OpenAI protocols. It also utilizes a layout detection model, DocLayout-YOLO-DocStructBench-onnx, to accurately process document structures. The tool itself functions as an MCP server, with the filesystem serving as a required MCP server component for locating PDF files. Its translate_pdf function operates as an independent MCP server component.

Core features and tools

Pdfmathtranslate exposes several callable tools for translation. The primary command-line interface tool is pdf2zh, used for operations like pdf2zh document.pdf or --share configurations. For programmatic integration, Python API functions such as translate and translate_stream are available. Additionally, it provides a dedicated Graphic User Interface (GUI) for macOS and Windows, and a Web UI accessible as an online service. A community-contributed Zotero Plugin extends its functionality to reference management workflows.

Who is it for?

Pdfmathtranslate is for researchers, students, and academics who need to translate scientific papers and other PDF documents. It is particularly useful for those who require the preservation of complex layouts, mathematical formulas, and scientific notation during the translation process. The tool supports various interaction methods, from command-line operations to a graphical interface, catering to different user preferences. Skip this if you do not require preserving complex layouts and formulas in translations.

How to get started

The server authenticates using an authorization file containing username and password pairs, specified via the --authorized command-line argument or a config.toml file. API keys for various translation services are typically set via environment variables or within a configuration file.

Use cases

• Translating a single PDF document via the command-line interface: Users execute pdf2zh document.pdf to generate translated and bilingual versions.
• Batch translation of PDF documents using the graphical user interface: Users upload multiple PDFs, select languages and translation services, and download outputs.
• Integrating PDF translation directly into a Zotero reference management workflow: Researchers utilize the Zotero plugin to translate PDFs within their existing library.

Pdfmathtranslate offers a versatile solution for academic and scientific PDF translation, connecting to numerous AI services and providing multiple user interfaces. Its ability to preserve document structure and mathematical formulas makes it a valuable asset for its target audience. The program's flexible deployment options, from Python packages to Docker images, enhance its accessibility. Start by translating a PDF document using the pdf2zh command-line interface.