jnayak1 / pdflatex-lambda Goto Github PK

Create a script that:

• Downloads the TeXLive installer
• Runs the text gui installer allowing user to select what packages they want to install, where they want to install them to, etc.
• Adds /path/to/package/bin/x86_64-linux to PATH

Also:

• Look into the TeXLive perltk installer. That might give more options/customizability than the text gui installer.

Once we get all the packages onto lambda, how do we get them to work together properly?

• Will modifying PATH be enough to get everything working?
• Look into TEXMFHOME environment variable

Approach:

• Launch fresh ec2 instance, clone github repo
• Add tlsb-packages-extracted/usr/bin to PATH variable
• Test pdflatex

• Write a script that downloads all rpm of the packges in texlive-scheme-basic.
• Put all rpm packages in repo

One issue is that TeXLive, which compiles LaTeX documents, is very large. The full, portable version is about 500MB according to this post 4.6GB.

This is a problem since Lambda limits the size of the zip file you can upload to 250MB unzipped.

However, pdflatex depends on just a subset of the full version.

TODO:

• Determine the subset of the full version of TeXLive needed to run pdflatex
• How big is this subset?
• How can we download and move this subset?

Approach:

• Research
• Boot up EC2 Amazon Linux instances and test

Create a script that launches a lambda function with the permissions we need:

• ZIPs up a specified directory
• Creates an S3 bucket using aws cli
• Creates a role with read, write permissions with aws cli
• Uploads the zip to lambda using aws cli with the role just created
• Initializes a trigger on the lambda function for when a file is uploaded to the s3 bucket using aws cli

I have seen this issue here where it is mentioned that pdflatex does not work. So what does the PdfLaTeX.zip file contain?
Also, since the main problem is the size limit of the package, have you considered implementing this on Azure Functions?
Thanks.

Unzip configure and compile each rpm package on Amazon EC2 Linux Instance and upload them to repo. Do not install them though.

AWS lambda's size limit for the ZIP file is 50MB. texlive-scheme-basic zipped up is a little over 59MB. The error message I get when I try to upload the ZIP says that I should use s3 to upload the ZIP instead.

• Look into how to use s3 to upload the zip
• Look into see if the package can be further compressed. I had used zip -9 but there might be something else that can compress the package further.
• Look into if there are additional files we can delete in the texlive-scheme-basic package

TeXCaller is a package that provides an interface to TeXLive in many different languages (C++, Python, etc.). The nice thing about it is that it provides error handling.

TODO:

• Script to install TeXCaller locally (just the portion we need, just python part, delete the rest)
• Test to make sure that TeXCaller can be run locally on Amazon Linux EC2 instance (without sudo)

• What AWS services are needed for this?
• Can you respond to HTTP request with lambda function?
• What authentication is needed?