Pipeline Feasibility for Python Library: Kornia

[xp1632]

• In theory, what we did in the pipeline is just to extract and rearrange the information into the JSON format that VP project could recognize

• Without the java-python, type incompatibility problem we are facing now

• Auto generation for other libraries could be even easier

[xp1632]

Where to get parameter information for a python method:

• help()
• annnotation since python 3.5

---

• We can get all parameter name information from a python method via help()
• And if a python method is defined with annotation standard, we can also get its datatype via .__annotation__

---

when checking the source code for kornia method: morphology.dilation: https://github.com/kornia/kornia/blob/main/kornia/morphology/morphology.py

• we can see it's defined with clear annotation, which makes it easy for us to get its datatype

[xp1632]

Then we just iterate over the Kornia Library via inspect

• sample example:

import kornia
import inspect

# Iterate over the attributes of the kornia module
for name, obj in inspect.getmembers(kornia):
    # Check if the attribute is a function
    if inspect.isfunction(obj):
        print(f"Function name: {name}")

        # Get the function's signature
        sig = inspect.signature(obj)

        # Iterate over the parameters of the function
        for param_name, param in sig.parameters.items():
            print(f"  Parameter: {param_name}")

            # Check if the parameter has a type annotation
            if param.annotation is not param.empty:
                print(f"    Type: {param.annotation}")
            else:
                print(f"    Type: Unknown")