Public | Automated Build

Last pushed: 3 years ago
Short Description
Short description is empty for this repo.
Full Description

Latex-Online

Try it

Latex-Online

Thanks to @udalov for deployment server

Contents

About

This is a small service developed to
compile latex documents online. The slogan is: "You give it a link, it gives you
PDF.", but the service evolved and now you can give it a git repo as well.

Additionally the service has a command-line utility that allows you for compiling
local documents.
See "usage" section for more information.

Capabilities

  • Compile .TEX file via link. (Limitation: includes will be ignored)
  • Compile GIT repo via link.
  • Compile local files or git repo via command-line interface.
  • REST API for compiling.

API

HTTP Response Codes

The service will return HTTP.2xx on success and compiled PDF file. Otherwise
a HTTP.4xx code will be returned with a compilation error log in response body.

Compile url

Format:

/compile?url=<url to tex file>

Example:

latex.aslushnikov.com/compile?url=https://raw.githubusercontent.com/aslushnikov/latex-online/master/sample/sample.tex

Limitation: this command will ignore all includes during compiling

Compile git repo

Format:

/compile?git=<repo>&target=<target file>

This will fetch git repo and compile the target. target should be a relative path to the repository root.

Example:

latex.aslushnikov.com/compile?git=https://github.com/aslushnikov/diplom-latex&target=diplom.tex

Optional request arguments

For every compilation request you can pass the following additional arguments:

  • force=true This will force cache skipping and document recompilation
  • download=sample.pdf This will initiate downloading of the resulting PDF
    into the file with the name "sample.pdf"

Command-line interface

The command-line interface makes it possible to edit TeX documents in
vim/emacs and compile them whenever you want from the command-line. To do so, you will
need a tool called laton.

Installation

curl -L https://raw.githubusercontent.com/aslushnikov/latex-online/master/util/laton > laton && chmod 755 laton

This command will result in a laton script created in a current directory. Put it somewhere
your $PATH references to make it available around the system.

Example 1: compile single file

If you've got a single file (say main.tex) that doesn't have any includes,
then you can compile it like this:

laton main.tex

After compiling file main.pdf will be created in the current dir

Example 2: compile files with dependencies

If your paper includes some files, you have to declare them to the
laton.

laton main.tex some-image.jpg some-cool-file.tex

NB The first file should be the file you want to compile, all others are
supporting files.

Example 3: compile local git repo

In case you store all your .tex and supporting files in a git repo, you
can compile the project with the command

laton -g main.tex

The script will behave as if main.tex includes each file stored in your git repo.

How it works?

In this section a brief description of the service from the inside is given.

Generally speaking the service is made out of two parts

  • Node.js part which with the help of express.js provides the REST API, and
    with the help of mc bridges the gap with memcached instance
  • Bash scripts which handle all the jobs related to the service (fetching
    files, hashSumming them, compiling e.t.c)

There are three types of requests:

  1. Given a link to a .TEX file, compile it
  2. Given a tarball with files structure, compile it
  3. Given a git repo, compile it

Every request is handled in a bit special way

1. Given a link to a .TEX file, compile it

  1. The URL of the given file is fetched and saved locally
  2. The hash sum of the file is counted in some way
  3. Check in cache if we've got a PDF for the hashsum
  4. If yes, then just return the precompiled PDF from cache
  5. If no, then compile the file, cache the result and return it to user

2. Given a tarball with files structure, compile it

  1. The tarball is saved locally
  2. The hash sum of the tarball is counted in some way
  3. Check in cache if we've got a PDF for the hashsum
  4. If yes, then just return the precompiled PDF from cache
  5. If no, then extract file structure from the tarball, compile it,
    cache the result and return it to user

3. Given a git repo, compile it

This kind of request is handled in a bit different way, as we can
get a hashSum of the repo without cloning the entire repository.

  1. Using git ls-remote extracting sha1 of the master branch
  2. Check in cache if we've got a PDF for the given sha1
  3. If yes, then just return the precompiled PDF from cache
  4. If no, then do a shallow copy of the given git repo
  5. Compile it, cache the result and return it to user

DEPLOYMENT

This part is for you if you'd like to deploy the service on your
own machine. There are two ways to do it:

  • easy one, with the help of awesome project Docker
  • the hard one - to do everything manually.

Deploy with Docker

Recently a Dockerfile was added which you can just build and run.

  1. docker build -t latex .
  2. docker run -d -p 2700:2700 -t latex

Deploy manually

Dependencies

  • pdflatex command to compile documents
  • bc to estimate some values in scripts
  • curl to fetch documents from web
  • node.js to run server
  • npm to install node dependencies
  • memcached for caching compiled documents
  • md5 or md5sum utility to hash documents according to their value
  • rubber latex build system is used to build files

Installation

  1. git clone git@github.com:aslushnikov/latex-online.git to clone repo
  2. sh util/check.sh to check if all dependencies are satisfied and create
    tmp/ dir
  3. npm install to install node dependencies

Installation of pdflatex is beyond the scope of the document.

Running

  1. node app.js - runs node server
  2. memcached - runs local memcached instance
Docker Pull Command
Owner
mattfullerton
Source Repository