TypeError: 'dict' object is not callable when trying to test Open API tool

[Chance-Obondo]

I am getting the mentioned error when trying to test the OpenAPI tool, not sure how to go about it.

My code: ` import os from haystack.dataclasses import ChatMessage from haystack_experimental.components.tools.openapi import OpenAPITool, LLMProvider from haystack.utils import Secret os.environ["OPENAI_API_KEY"]=""

os.environ["SERPERDEV_API_KEY"]=""

tool = OpenAPITool(generator_api=LLMProvider.OPENAI, spec="https://raw.githubusercontent.com/open-meteo/open-meteo/main/openapi.yml")

tool.run(messages=[ChatMessage.from_user("Weather in San Francisco, US")])`

The error:

TypeError: 'dict' object is not callable

[vblagoje]

@Chance-Obondo this works for me on the main branch but it could indeed not be working on old release of haystack-experimental (v0.1.1). We are about to release a new version of experimental but until then please try with pip install git+https://github.com/deepset-ai/haystack-experimental, export OPENAI_API_KEY and then run:

from haystack.dataclasses import ChatMessage
from haystack_experimental.components.tools.openapi import OpenAPITool, LLMProvider

tool = OpenAPITool(generator_api=LLMProvider.OPENAI, spec="https://raw.githubusercontent.com/open-meteo/open-meteo/main/openapi.yml")
tool.run(messages=[ChatMessage.from_user("Weather in San Francisco, US")])

I just tried it and I get:

{'service_response': [ChatMessage(content='{"latitude": 37.763283, "longitude": -122.41286, "generationtime_ms": 0.07796287536621094, "utc_offset_seconds": 0, "timezone": "GMT", "timezone_abbreviation": "GMT", "elevation": 18.0, "current_weather_units": {"time": "iso8601", "interval": "seconds", "temperature": "\\u00b0C", "windspeed": "km/h", "winddirection": "\\u00b0", "is_day": "", "weathercode": "wmo code"}, "current_weather": {"time": "2024-09-24T09:15", "interval": 900, "temperature": 16.3, "windspeed": 6.5, "winddirection": 289, "is_day": 0, "weathercode": 0}}', role=<ChatRole.USER: 'user'>, name=None, meta={})]}

Let us know 🙏

[Chance-Obondo]

Thank you! It has worked on my end, is there a way maybe the documentation on colab can be updated and also on github so that anyone else trying to experiment can know there's some updates?

[vblagoje]

Thank you! It has worked on my end, is there a way maybe the documentation on colab can be updated and also on github so that anyone else trying to experiment can know there's some updates?

It would be a moot update as the new release is imminent - should be literally today. Please keep trying OpenAPITool @Chance-Obondo - your feedback is going to be immensely valuable.

[vblagoje]

Recommend closing @julian-risch

[vblagoje]

Thank you! It has worked on my end, is there a way maybe the documentation on colab can be updated and also on github so that anyone else trying to experiment can know there's some updates?

It would be a moot update as the new release is imminent - should be literally today. Please keep trying OpenAPITool @Chance-Obondo - your feedback is going to be immensely valuable.

Also, even though sometimes an OpenAPI spec for a certain REST endpoint is not available - you can usually generate it using its documentation and AI generation tools, making it easier to integrate REST endpoints with your Haystack-based LLM app.

[vblagoje]

A new release of haystack-experimental is out. This issue is now officially fixed, closing as "Fixed"!