vllm does not support PDF file URL parsing

[IceBear-huang]

Hello, I have deployed the NanoNets-OCR model using vllm, but vllm does not support PDF file URL parsing. How should I handle this?

[mandalsouvik3333]

You will need to convert the pdfs to images and then send the encoded images.

docext/docext/core/pdf2md/pdf2md.py

Line 108 in 668eee6

"url": f"data:image/jpeg;base64,{encode_image(file_path)}"

You can run the docext app to send pdfs directly https://github.com/NanoNets/docext/blob/main/PDF2MD_README.md#models-with-vllm-linux

[IceBear-huang]

thanks