lmrtfy

Turn variables into program arguments. Auto-generates a CLI interface and an API using lmrt.fyi.

Introduction

lmrtfy is a tool to share your applications with others without dealing with the tedious things:

• Only minimal changes to your code by annotating your input and output variables
• Automatically generated API that can be shared with others

!!! warning lmrtfy is currently in an early alpha phase. We try to minimize the things that will change in the future, but we cannot make any guarantees at the moment. Please keep that in mind while using the lmrtfy tools.

Currently, the application itself runs on your computer and we provide a tightly controlled interface via our web API to start jobs that make use of the script running on your computer. Follow the quick start guide to get started!

We also have a more comprehensive guide and examples that you might be interested in!

!!! Note If you encounter any obstacles while using our tool or while reading the documentation, please don't hesitate to contact us. Just create an issue on GitHub.

Quickstart

TL;DR

1. intall with pip install lmrtfy
2. login/sign up with lmrtfy login
3. annotate your code's inputs with variable and its outputs with result
4. in terminal 1 run lmrtfy deploy <script.py> --local (returns profile_id)
5. in terminal 2 run lmrtfy submit <profile_id> <input.json> (returns job_id)
6. in terminal 2 run lmrtfy fetch <job_id> <path> to get results

Installation:

$ pip install lmrtfy

First login

Login/sign up to receive access token: $ lmrtfy login. The token is saved in ~/.lmrtfy/auth/token but you should not need to manually open the token.

!!! info Tokens are currently valid for 24 hours. After that you will be requested to login again. That also means that you cannot have scripts deployed more than 24 hours right now. This will change soon so that you can deploy scripts longer than that.

Just run `lmrtfy deploy <script> --local`again after 24h and you are fine if you need longer running
deployments right now.)

Create code annotations

Annotate the inputs and outputs of your script with variable and results and save it as script.py:

# import the required things from lmrtfy
from lmrtfy import variable, result
# annotate an input
time = variable(200., name="time", min=0, max=1000, unit="s")
speed = result(9.81*time, name="speed", min=0, max=9810, unit="m/s")

This calculates the velocity of an object in free fall after time seconds.

Run python script.py to create the profile. This always works, even without access to lmrtfy.

Deploy the script to accept jobs from the generated web API

Run lmrtfy deploy script.py --local to generate the API and start the runner to listen to jobs for your script. You can find the profile id to submit a job in the logs:

INFO Starting deployment of examples/velocity_from_gravity/calc_velocity.py
WARNING Deploying locally.
INFO Profile_id to be used for requests: 7ff68a0cfd8c61122cfdaf0a835c7cd1f94e7db9

Submit jobs with CLI

Open a new terminal and submit a new job (<profile_id> is the profile id you received during the deployment (7ff... in the example above)):

$ lmrtfy submit <profile_id> <input.json>

For the example above save the following in your input.json:

{
  "argument_values": {
    "time": 200.0
  },
  "argument_units" : {
    "time": "s"
  }
}

You will receive a job_id which you will need to fetch the results later on:

INFO Job-id: 14584640-778c-4f91-a288-03cffc2b9c7a

Fetch results

Get the results by calling $ lmrtfy fetch <job_id> <path>. Currently, the results will be saved in <path>/<job_id> as JSON files.

License