Rasa custom component - rasa.nlu.components

Rasa Open Source
1

I want to create a custom component for a Multilingual bot but I get the error ‘No module named rasa.nlu.components’ when I try from rasa.nlu.components import Components

image
image815×128 5.68 KB
image
image976×50 3.16 KB

2

I got the same error and I don’t see any solution for it. @minju?

I think mostly it is a version mismatch error…

But I don’t know which version would be most suitable. I am not even getting any help from the github developer community.

Please let me know as soon as you get the solution…

3

I know it worked when I downgraded rasa to 2.8.3 or 2.6 ! But I need to work with the latest release of rasa so I just don’t know how to find an alternative to that problem… Finally, for the moment, I just gave up with the custom component

4

@minji - please migrate your custom components to graph component in Rasa 3.x

5

same issue in 3.5.1

6

Lets say you wanted to make a component, that would lower_case an incoming query.

\middleware\lower_case.py

from typing import List

from rasa.engine.graph import GraphComponent
from rasa.shared.nlu.constants import TEXT
from rasa.shared.nlu.training_data.message import Message
from rasa.engine.recipes.default_recipe import DefaultV1Recipe


@DefaultV1Recipe.register(
    component_types=[DefaultV1Recipe.ComponentType.INTENT_CLASSIFIER],
    is_trainable=False,
)
class TextCorrectionMiddleware(GraphComponent):

    def __init__(self, component_config=None):
        super().__init__(component_config)

    def process(self, messages: List[Message], **kwargs) -> List[Message]:
        for message in messages:
            text = message.get(TEXT)
            if text:
                print(f"Original Text: {text}")  # Print original text
                message.set(TEXT, text.lower())
                print(f"Lowercased Text: {message.get(TEXT)}")
        return messages

in config.yml:


pipeline:
  - name: middleware.lower_case.TextCorrectionMiddleware
  - name: WhitespaceTokenizer
  - name: RegexFeaturizer
***** REST OF THE CODE ******