Local AI Web Hook

This project extends Local AI with pre-query functionality.

---

Usage

Installation

Clone this repository to any location of your choice.

Start/Stop

Open a terminal in the project directory and run:

python app.py

A simple server will start on port 10001. You can specify an alternate port by passing it as a parameter to the command line:

python app.py 10002

Note: You may need to use python3 instead of python in this command depending on your configuration.

Hooks

As this is a Python project, all files must be Python scripts and must be located inside the api/ directory; its contents are otherwise unrestricted.

Calls

To list all available hooks, send a GET request to the root URL:

http://localhost:10001/

assuming the default port is used.

The response will be a list containing all available files and their functions.

   1. file: test.py; func: ['helloWorld']
   2. file: parsefile.py; func: ['do_Parse']

Note: Regardless of the requested data format, the response will have a Content-Type header set to text/plain.

Any other request must call the server api/ endpoint (See Examples below).

Consuming the hook

Requests expect four components, which are optional except for one:

1. File (required)
2. Function name (optional)
3. Post data (optional)
4. Attachment (optional)

You can provide any or all of these components when making a request.

Examples

1. Calling a file:

http://localhost:10001/api/test

In this scenario, test.py file must define a default function that can handle the request.

2. To invoke a specific function, include its name in the URL path, separated from the file name by a slash ('/').

http://localhost:10001/api/test/helloWorld

Both scenarios are handled as both GET and POST requests, with the result returned as plain text.

When sending data via POST, you can provide it in either a key=value format (as FormData) or JSON format. Files must be sent using form-multipart encoding.

Reading Web Resource

A simple example of how to fetch web content which can be passed next to the model as a context is available here

Usage

curl -X POST http://localhost:10001/api/readweb -H "Content-Type: application/json" -d '{"resource":"https://github.com/ivostoykov/localAI"}'