danfickle / openhtmltopdf-native

A proof-of-concept PDF generator with GraalVM native-image tool.

OpenHTMLtoPDF with native-image

Aim is to make a blazing fast PDF generator. This is a proof-of-concept. See also danfickle/openhtmltopdf

Instructions

I did this on an Oracle Cloud Ampere ARM compute instance (Oracle Linux 8) with 18GB of memory (always free eligible). I'd recommend an instance with at least 8GB of memory. It may also work on x86 (with a different yum repo below).

Create the instance and SSH into it:

• sudo yum update
• sudo reboot
• sudo yum install git
• sudo yum install graalvm21-ee-11
• sudo yum-config-manager --add-repo=http://yum.oracle.com/repo/OracleLinux/OL8/codeready/builder/aarch64
• sudo yum install graalvm21-ee-11-native-image
• git clone https://github.com/danfickle/openhtmltopdf-native.git
• cd openhtmltopdf-native
• chmod u+x build.sh
• chmod u+x mvnw
• ./build.sh
• cd target
• ./com.example.app to run, will produce test.pdf.
• At the moment testing consists of downloading test.pdf with scp and confirming it opens.

Feedback welcome

• Have you got this working on x86?
• Have you got it working with a web server?
• Serverless?

openhtmltopdf-native's Issues

Unable to make custom fonts to work

I was able to run this as a servless lambda function on AWS using quarkus native build. It is working great but the only issue I was unable to work it out is the fact that even though the font file is loaded properly it is not rendering on the PDF. This issue only occurs with the native build.

I use the following code to load the files from the resource and it does not throw any message.

builder.useUriResolver((baseUri, uri) -> {
	if (baseUri != null) {
		URL resource = RelatorioHandler.class.getClassLoader().getResource(uri);
		if (resource == null)
			throw new RuntimeException("Resource não encontrado");

		return resource.toString();
	}

	return uri;
});

expected.pdf
actual.pdf