Movatterモバイル変換

[0]ホーム
URL:

Pipedream supportsPython v in workflows. Run any Python code, use anyPyPI package, connect to APIs, and more.

Adding a Python code step

  1. Click the + icon to add a new step
  2. ClickCustom Code
  3. In the new step, select thepython language runtime in language dropdown

Python Code Step Structure

A new Python Code step will have the following structure:
def handler(pd:"pipedream"):  # Reference data from previous steps  print(pd.steps["trigger"]["context"]["id"])  # Return data for use in future steps  return {"foo": {"test":True}}
You can also perform more complex operations, includingleveraging your connected accounts to make authenticated API requests,accessing Data Stores andinstalling PyPI packages.

Logging and debugging

You can useprint at any time in a Python code step to log information as the script is running.The output for theprintlogs will appear in theResults section just beneath the code editor.

Using third party packages

You can use any packages fromPyPI in your Pipedream workflows. This includes popular choices such as:To use a PyPI package, just include it in your step’s code:
import requests
And that’s it. No need to update arequirements.txt or specify elsewhere in your workflow of which packages you need. Pipedream will automatically install the dependency for you.

If your package’simport name differs from its PyPI package name

Pipedream’s package installation usesthepipreqs package to detect package imports and install the associated package for you. Some packages, likepython-telegram-bot, use animport name that differs from their PyPI name:
pip install python-telegram-bot
vs.
import telegram
Use the built inmagic comment system to resolve these mismatches:
# pipedream add-package python-telegram-botimport telegram

Pinning package versions

Each time you deploy a workflow with Python code, Pipedream downloads the PyPi packages youimport in your step.By default, Pipedream deploys the latest version of the PyPi package each time you deploy a change.There are many cases where you may want to specify the version of the packages you’re using. If you’d like to use aspecific version of a package in a workflow, you can add that version in amagic comment, for example:
# pipedream add-package pandas==2.0.0import pandas
Currently, you cannot use different versions of the same package in different steps in a workflow.

Making an HTTP request

We recommend using the popularrequests HTTP client package available in Python to send HTTP requests.No need to runpip install, justimport requests at the top of your step’s code and it’s available for your code to use.See theMaking HTTP Requests with Python docs for more information.

Returning HTTP responses

You can return HTTP responses fromHTTP-triggered workflows using thepd.respond() method:
def handler(pd:"pipedream"):  pd.respond({    "status":200,    "body": {      "message":"Everything is ok"    }  })
Please note to always include at least thebody andstatus keys in yourpd.respond argument. Thebody must also be a JSON serializable object or dictionary.
Unlike the Node.js equivalent, the Pythonpd.respond helper does not yet support responding with Streams.
Don’t forget toconfigure your workflow’s HTTP trigger to allow a custom response. Otherwise your workflow will return the default response.

Sharing data between steps

A step can accept data from other steps in the same workflow, or pass data downstream to others.

Using data from another step

In Python steps, data from the initial workflow trigger and other steps are available in thepd.steps object.In this example, we’ll pretend this data is coming into our workflow’s HTTP trigger via POST request.
// POST <our-workflows-endpoint>.m.pipedream.net{  "id":1,  "name":"Bulbasaur",  "type":"plant"}
In our Python step, we can access this data in thepd.steps object passed into thehandler. Specifically, this data from the POST request into our workflow is available in thetrigger dictionary item.
def handler(pd:"pipedream"):  # retrieve the data from the HTTP request in the initial workflow trigger  pokemon_name= pd.steps["trigger"]["event"]["name"]  pokemon_type= pd.steps["trigger"]["event"]["type"]  print(f"{pokemon_name} is a{pokemon_type} type Pokemon")

Sending data downstream to other steps

To share data created, retrieved, transformed or manipulated by a step to others downstream,return the data in thehandler function:
# This step is named "code" in the workflowimport requestsdef handler(pd:"pipedream"):  r= requests.get("https://pokeapi.co/api/v2/pokemon/charizard")  # Store the JSON contents into a variable called "pokemon"  pokemon= r.json()  # Expose the data to other steps in the "pokemon" key from this step  return {    "pokemon": pokemon  }
Now thispokemon data is accessible to downstream steps withinpd.steps["code"]["pokemon"]
You can only export JSON-serializable data from steps. Things like:
  • strings
  • numbers
  • lists
  • dictionaries

Using environment variables

You can leverage anyenvironment variables defined in your Pipedream account in a Python step. This is useful for keeping your secrets out of code as well as keeping them flexible to swap API keys without having to update each step individually.To access them, use theos module.
import osdef handler(pd:"pipedream"):  token= os.environ["AIRTABLE_API_KEY"]  print(token)
Or an even more useful example, using the stored environment variable to make an authenticated API request.

Using API key authentication

If an particular service requires you to use an API key, you can pass it via the headers of the request.This proves your identity to the service so you can interact with it:
import requestsimport osdef handler(pd:"pipedream"):  token= os.environ["AIRTABLE_API_KEY"]  url= "https://api.airtable.com/v0/your-airtable-base/your-table"  headers= {"Authorization":f"Bearer{token}"}  r= requests.get(url,headers=headers)  print(r.text)
There are 2 different ways of using theos module to access your environment variables.os.environ["ENV_NAME_HERE"] will raise an error that stops your workflow if that key doesn’t exist in your Pipedream account.Whereasos.environ.get("ENV_NAME_HERE") willnot throw an error and instead returns an empty string.If your code relies on the presence of a environment variable, consider usingos.environ["ENV_NAME_HERE"] instead.

Handling errors

You may need to exit a workflow early. In a Python step, just araise an error to halt a step’s execution.
raise NameError("Something happened that should not. Exiting early.")
All exceptions from your Python code will appear in thelogs area of the results.

Ending a workflow early

Sometimes you want to end your workflow early, or otherwise stop or cancel the execution of a workflow under certain conditions. For example:
  • You may want to end your workflow early if you don’t receive all the fields you expect in the event data.
  • You only want to run your workflow for 5% of all events sent from your source.
  • You only want to run your workflow for users in the United States. If you receive a request from outside the U.S., you don’t want the rest of the code in your workflow to run.
  • You may use theuser_id contained in the event to look up information in an external API. If you can’t find data in the API tied to that user, you don’t want to proceed.
In any code step, callingreturn pd.flow.exit() will end the execution of the workflow immediately. No remaining code in that step, and no code or destination steps below, will run for the current event.
It’s a good practice to usereturn pd.flow.exit() to immediately exit the workflow. In contrast,pd.flow.exit() on its own will end the workflow only after executing all remaining code in the step.
def handler(pd:"pipedream"):  return pd.flow.exit("reason")  print("This code will not run, since pd.flow.exit() was called above it")
You can pass any string as an argument topd.flow.exit():
def handler(pd:"pipedream"):  return pd.flow.exit("Exiting early. Goodbye.")  print("This code will not run, since pd.flow.exit() was called above it")
Or exit the workflow early within a conditional:
import randomdef handler(pd:"pipedream"):  # Flip a coin, running pd.flow.exit() for 50% of events  if random.randint(0,100)<= 50:    return pd.flow.exit("reason")  print("This code will only run 50% of the time");

File storage

You can also store and read files with Python steps. This means you can upload photos, retrieve datasets, accept files from an HTTP request and more.The/tmp directory is accessible from your workflow steps for saving and retrieving files.You have full access to read and write both files in/tmp.See theWorking with the filesystem in Python docs for more information.

FAQ

What’s the difference betweendef handler(pd) and thepipedream package for Python code steps?

The pd object passed to the handler method lets you exit theworkflow early,integrate a Data Store, anduse connected accounts into your Python code steps.However, at this time there are issues with our Python interpreter that is causing anECONNRESET error.If you needto use data from other steps orexport data to other steps in your workflow, we recommend using thepipedream package module.If you need to use a Data Store in your workflow, we recommend using apre-built action to retrieve or store data orNode.js’s Data Store capabilities.

I’ve tried installing a Python package with a normal import and the magic comment system, but I still can’t. What can I do?

Some Python packages require binaries present within the environment in order to function properly. Or they include binaries but those binaries are not compatible with the Pipedream workflow environment.Unfortunately we cannot support these types of packages at this time, but if you have an issue importing a PyPI package into a Python code stepplease open a issue.

Can I publish my Python code as a reusable pre-built action or trigger like you can with Node.js?

Not at this time. Pipedream only supports Python as a code step language. The Components system only supports Node.js at this time.You can still duplicate Python code steps within the same workflow, but to reuse a code step, you’ll need to copy and paste the Python code to another workflow.
[8]ページ先頭
©2009-2025 Movatter.jp