Loader

Loaders fetch data into BroxiAI from various sources, such as databases, websites, and local files.

Use a loader component in a flow​

This flow creates a question-and-answer chatbot for documents that are loaded into the flow. The Unstructured.io loader component loads files from your local machine, and then parses them into a list of structured Data objects. This loaded data informs the Open AI component's responses to your questions.

Sample Flow retrieving data with unstructured

Confluence​

The Confluence component integrates with the Confluence wiki collaboration platform to load and process documents. It utilizes the ConfluenceLoader from LangChain to fetch content from a specified Confluence space.

Inputs​

Name	Display Name	Info
url	Site URL	The base URL of the Confluence Space (e.g., https://company.atlassian.net/wiki)
username	Username	Atlassian User E-mail (e.g., [e-mail])
api_key	API Key	Atlassian API Key (Create an API key at: Atlassian)
space_key	Space Key	The key of the Confluence space to access
cloud	Use Cloud?	Whether to use Confluence Cloud (default: true)
content_format	Content Format	Specify content format (default: STORAGE)
max_pages	Max Pages	Maximum number of pages to retrieve (default: 1000)

Outputs​

Name	Display Name	Info
data	Data	List of Data objects containing the loaded Confluence documents

GitLoader​

The GitLoader component uses the GitLoader from LangChain to fetch and load documents from a specified Git repository.

Inputs​

Name	Display Name	Info
repo_path	Repository Path	The local path to the Git repository
clone_url	Clone URL	The URL to clone the Git repository from (optional)
branch	Branch	The branch to load files from (default: 'main')
file_filter	File Filter	Patterns to filter files (e.g., '.py' to include only .py files, '!.py' to exclude .py files)
content_filter	Content Filter	A regex pattern to filter files based on their content

Outputs​

Name	Display Name	Info
data	Data	List of Data objects containing the loaded Git repository documents

Unstructured​

This component uses the Unstructured.io Serverless API to load and parse files into a list of structured Data objects.

Inputs​

Name	Display Name	Info
file	File	The path to the file to be parsed (supported types are listed here)
api_key	API Key	Unstructured.io Serverless API Key
api_url	Unstructured.io API URL	Optional URL for the Unstructured API
chunking_strategy	Chunking Strategy	Strategy for chunking the document (options: "", "basic", "by_title", "by_page", "by_similarity")
unstructured_args	Additional Arguments	Optional dictionary of additional arguments for the Unstructured.io API

Outputs​

Name	Display Name	Info
data	Data	List of Data objects containing the parsed content from the input file