Python

You can write your own custom Python codes to manipulate the received data or even process the image before passing to the next block.

def main(data):
  return data

Python blocks can be cascaded.

Installed libraries

There are a few libraries available for use.

Package name	Description
cv2	OpenCV 4.2.1
numpy	Numpy
torch, torchvision	PyTorch

Access the key-value pair

In Python, it is very simple to access the key value pair.

def main(data):
  return {
    "renameTheKey": data["originalKey"]     
  }

Process an image

Since the image is a data URI which is comprised of a base64-encoded string, you can use a base64 package to decode the string and use cv2.imdecode to read a numpy array.

The following sample code converts the image into a black-and-white image.

import base64
import cv2
import numpy as np

def main(data):
  img = cv2.imdecode(np.asarray(bytearray(base64.b64decode(data["image"].split(",")[1])), dtype="uint8"), cv2.IMREAD_COLOR)
  img2 = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # convert to b/w
  imgstr = str(base64.b64encode(cv2.imencode(".png", img2)[1]), "utf-8")

  return {
    "image": "data:image/png;base64," + imgstr,
    "resolution": data["resolution"]
  }

Debugging

You cannot use print() to debug as the function is suppressed. You can always return the debugging values as the result for inspection.