[Question]: Need help with RAG over YAML/OPENAPI Specification API documentation

[hpsc3]

Question Validation

• I have searched both the documentation and discord for an answer.

Question

I want to chat with an openapi yaml specification file. The idea is that the user will ask questions about the api and the chatbot should answer from the knowledge about the api from the api specification. For eg. 'How do get a list of all accounts'-> A query engine should retrieve this. I am trying to figure our a loader for YAML files, and I found OPENAPITOOL on llama_hub, but it appears to be designed to work with agents and not query engines. I tried searching the document but did not find any YAML loaders for doing normal RAG. Any assistance over this would be appreciated. If there is any existing code examples that achieve this, that would be great to.

[dosubot[bot]]

🤖

Hello,

Thank you for reaching out with your question. Currently, LlamaIndex does not have a built-in YAML loader for regular RAG. The supported file types are listed in the DEFAULT_FILE_READER_CLS dictionary, which includes the following file types: ".hwp", ".pdf", ".docx", ".pptx", ".jpg", ".png", ".jpeg", ".mp3", ".mp4", ".csv", ".epub", ".md", ".md", ".mbox", ".ipynb". Unfortunately, YAML is not included in this list.

However, you can create your own YAML loader by creating a class that inherits from the BaseReader class and implements the load_data method. Once you have this, you can use the OnDemandLoaderTool utility tool provided by LlamaHub to turn your YAML loader into a tool that an agent can use.

Here is an example of how you can use it:

from llama_hub.wikipedia.base import WikipediaReader
from llama_index.tools.ondemand_loader_tool import OnDemandLoaderTool

# Assuming you have a YAMLReader class that reads your OpenAPI YAML specification file
from your_module import YAMLReader

reader = YAMLReader('path_to_your_yaml_file')

tool = OnDemandLoaderTool.from_defaults(
	reader,
	name="YAML Tool",
	description="A tool for loading data and querying from YAML file"
)

In this example, replace 'path_to_your_yaml_file' with the path to your OpenAPI YAML specification file and your_module with the module where your YAMLReader class is defined.

Please note that the OnDemandLoaderTool is designed to work with any data loader that inherits from the BaseReader class, not just agents. The OnDemandLoaderTool can be called with all the parameters needed to trigger load_data from the data loader, along with a natural language query string. During execution, it first loads data from the data loader, indexes it (for instance with a vector store), and then queries it “on-demand”. All three of these steps happen in a single tool call.

You can find more information about this in the LlamaHub Tools Guide.

I hope this helps! If you have any other questions, feel free to ask.

Best,
Dosu

Sources

• docs/core_modules/agent_modules/tools/llamahub_tools_guide.md
• llama_index/readers/file/base.py

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.