​
About Environment Variables

Environment variables allow you to access information uniformly across public apps on Sieve and custom-deployed Sieve functions. These can be accessed through os.environ. Several public apps, like Dubbing and Text-To-Speech, use environment variables to let users supply custom API Keys for external services like OpenAI and ElevenLabs. If users set secrets, they will be mapped automatically to any environment variable with a matching name. For more information on secrets, click here.

​
Adding Environment Variables to a Sieve Function

We can add one or more environment variables to a sieve Function using the header parameter environment_variables. Each environment variable you would supply is packaged as a sieve.Env object, which is structured as follows: 
class Env:
  name: str # the name of the environment variable (e.g. OPENAI_API_KEY)
  description: str # what the environment variable is used for in the function
  default: Optional[str] # a default value for the environment variable, if not provided by the user.
To use environment variables, you would pass in a list of sieve.Env objects to the environment_variables parameter of the header. An example function is pasted below. 
import sieve

@sieve.function(
  name="test_environment_variables",
  environment_variables=[
    sieve.Env(name="OPENAI_API_KEY", description="Used to communication with the OpenAI API."),
    sieve.Env(name="OPTIONAL_CREDENTIALS", description="An optional credential used for a backup operation.", default="")
  ],
)
def test_environment_variables(a: str) -> bool:
  import os
  return (os.environ["OPENAI_API_KEY"] is not "") and (os.environ["OPTIONAL_CREDENTIALS"] is not "")

​
Example: Generate an image with Dall-E 3

Let’s build a simple function to generate an image using OpenAI’s Dall-E 3 model. 
# openai_image_generator/main.py

import sieve

@sieve.function(
  name="openai_image_generator",
  python_packages=[
    "openai"
  ],
)
def openai_image_generator(prompt: str) -> sieve.File:
  from openai import OpenAI
  client = OpenAI()

  response = client.images.generate(
    model="dall-e-3",
    prompt=prompt,
    size="1024x1024",
    quality="standard",
    n=1,
  )

  image_url = response.data[0].url

  return sieve.File(image_url)
If you run this function via the Sieve dashboard, you will get an error similar to the following: 
openai.OpenAIError: The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable
We don’t want to leak credentials in our source code, so let’s use secrets and environment variables to inject our API key into our function. First, sign up for an OpenAI API key here. Then, let’s set the API key as an environment variable via the settings page on the dashboard. Now, let’s add an environment variable to the code to use this secret. 
# openai_image_generator/main.py

import sieve

@sieve.function(
  name="openai_image_generator",
  python_packages=[
    "openai"
  ],
  environment_variables=[
    sieve.Env(name="OPENAI_API_KEY", description="API key for OpenAI"),
  ],
)
def openai_image_generator(prompt: str) -> sieve.File:
  from openai import OpenAI
  client = OpenAI()

  response = client.images.generate(
    model="dall-e-3",
    prompt=prompt,
    size="1024x1024",
    quality="standard",
    n=1,
  )

  image_url = response.data[0].url

  return sieve.File(image_url)
Now, if you run it on the Sieve dashboard, you should get an output!