# Functions quickstart

### What are deepmux functions

DeepMux functions are serverless functions like AWS Lambda, Google Cloud Functions but with GPU support.

### What are we going to do

We are going to create a demo python funciton step-by-step with DeepMux platform.

## Make sure deepmux-cli is installed

Check out the [Installation guide](https://hackmd.io/uB5Z3Kt9QwiXtPZ1aTf98g)

## Setup project

We recommend creating your project in a separate directory. For example:

mkdir -p myproject
cd myproject

List function environments and choose:

deepmux env

We are going to use `python3.7` environment

Than initialize your project

deepmux init --name yourproject --env python3.7

Will create an empty function named `myproject` and `deepmux.yaml` file at your working directory.

name: myproject
env: python3.7
python:
  call: <required, module:function to call>
  requirements: <optional, path to requirements.txt>

We are going to fill those sections later.

## Implement function
Now it's time to implement the function itself.

In this example we will implement a functions that simply reverses it's input.

Create `main.py` file in project directory and fill it with following code:

Functions accept `bytes` and should return `bytes`

def reverse_function(data):
    return data[::-1]

Let's save this function to a file `main.py` and add it as an entrypoint.

You need to open `deepmux.yaml` and add `main:reverse_function` to the `call:` section.

At this point `deepmux.yaml` would look like:

name: myproject
env: python3.7
python:
  call: main:reverse_function
  requirements: <optional, path to requirements.txt>

And the project structure should look like:

myproject/
myproject/deepmux.yaml
myproject/main.py

## Add requirements (Optional)

You can add requirements for your function. Let's say we need `numpy` package somewhere.

Write`requirements.txt` file

numpy==1.19.4

And add path to the file in your `deepmux.yaml`:

name: yourproject
env: python3.7
python:
  call: main:reverse_function
  requirements: requirements.txt

## Upload function

Once we've finished with function, requirements and `deepmux.yaml` we are able to upload function to the platform via cli.

deepmux upload

## Run the function

You can use `deepmux list` command to get your functions and their statuses or simply check your functions via Web UI.

Once your function has finished processing and is in `READY` state you can call it with an HTTP request.

Here's an example using curl:

curl -X POST \
     -H "X-Token: <YOUR API TOKEN>" \
     http://api.deepmux.com/v1/function/myproject/run \
     --data "Hello!"

You should see the following on the screen:

!olleH