feat: Allows for expanded configuration

* Adds the ability to use your model of preference
* Adds the ability to use custom base URLS.
  * This is helpful for running local LLMs or using Azure based openAI calls
* Adds the ability to add additional parameters to the API call for increased customization
* Adds a sample file
* Backwards compatible with old api_key.lua configuration
* Updates README with new documentation and use cases

.gitignore

@@ -1 +1,2 @@
-/api_key.lua
+/api_key.lua
+/configuration.lua

README.md

@@ -10,17 +10,36 @@ Get [KoReader](https:/koreader/koreader) installed on your e-reader.
 
 If you want to do this on a Kindle, you are going to have to jailbreak it. I recommend following [this guide](https://www.mobileread.com/forums/showthread.php?t=320564) to jailbreak your Kindle.
 
-An API key from OpenAI. Once you have your API key, create a `api_key.lua` file in the following structure:
+Acquire an API key from an API account on OpenAI (with credits). Once you have your API key, create a `configuration.lua` file in the following structure or modify and rename the `configuration.lua.sample` file:
+
+> **Note:** The prior `api_key.lua` style configuration is deprecated. Please use the new `configuration.lua` style configuration.
+
+```lua
+local CONFIGURATION = {
+ api_key = "YOUR_API_KEY",
+ model = "gpt-4o-mini",
+ base_url = "https://api.openai.com/v1/chat/completions",
+}
+
+return CONFIGURATION
+```
+
+In this new format you can specify the model you want to use, the API key, and the base URL for the API. The model is optional and defaults to `gpt-4o-mini`. The base URL is also optional and defaults to `https://api.openai.com/v1/chat/completions`. This is useful if you want to use a different model or a different API endpoint (such as via Azure or another LLM that uses the same API style as OpenAI).
+
+For example, you could use a local API via a tool like (Ollama)[https://ollama.com/blog/openai-compatibility] and set the base url to point to your computers IP address and port.
 
 ```lua
-local API_KEY = {
- key = "your_api_key",
+local CONFIGURATION = {
+ api_key = "ollama",
+ model = "zephyr",
+ base_url = "http://192.168.1.87:11434/v1/chat/completions",
+ additional_parameters = {},
 }
 
-return API_KEY
+return CONFIGURATION
 ```
 
-If you clone this project, you should be able to put the directory, `askgpt.koplugin`, in the `koreader/plugins` directory and it should work. If you want to use the plugin without cloning the project, you can download the zip file from the releases page and extract the `askgpt.koplugin` directory to the `koreader/plugins` directory.
+If you clone this project, you should be able to put the directory, `askgpt.koplugin`, in the `koreader/plugins` directory and it should work. If you want to use the plugin without cloning the project, you can download the zip file from the releases page and extract the `askgpt.koplugin` directory to the `koreader/plugins` directory. If for some reason you extract the files of this repository in another directory, rename it before moving it to the `koreader/plugins` directory.
 
 ## How To Use
 

configuration.lua.sample

@@ -0,0 +1,8 @@
+local CONFIGURATION = {
+ api_key = "YOUR_API_KEY",
+ model = "gpt-4o-mini",
+ base_url = "https://api.openai.com/v1/chat/completions",
+ additional_parameters = {},
+}
+
+return CONFIGURATION

gpt_query.lua

@@ -1,25 +1,62 @@
-local API_KEY = require("api_key")
+local api_key = nil
+local CONFIGURATION = nil
+
+-- Attempt to load the api_key module. IN A LATER VERSION, THIS WILL BE REMOVED
+local success, result = pcall(function() return require("api_key") end)
+if success then
+ api_key = result.key
+else
+ print("api_key.lua not found, skipping...")
+end
+
+-- Attempt to load the configuration module
+success, result = pcall(function() return require("configuration") end)
+if success then
+ CONFIGURATION = result
+else
+ print("configuration.lua not found, skipping...")
+end
+
+-- Define your queryChatGPT function
 local https = require("ssl.https")
+local http = require("socket.http")
 local ltn12 = require("ltn12")
 local json = require("json")
 
 local function queryChatGPT(message_history)
- local api_key = API_KEY.key
- local api_url = "https://api.openai.com/v1/chat/completions"
+ -- Use api_key from CONFIGURATION or fallback to the api_key module
+ local api_key_value = CONFIGURATION and CONFIGURATION.api_key or api_key
+ local api_url = CONFIGURATION and CONFIGURATION.base_url or "https://api.openai.com/v1/chat/completions"
+ local model = CONFIGURATION and CONFIGURATION.model or "gpt-4o-mini"
+
+ -- Determine whether to use http or https
+ local request_library = api_url:match("^https://") and https or http
+
+ -- Start building the request body
+ local requestBodyTable = {
+ model = model,
+ messages = message_history,
+ }
+
+ -- Add additional parameters if they exist
+ if CONFIGURATION and CONFIGURATION.additional_parameters then
+ for key, value in pairs(CONFIGURATION.additional_parameters) do
+ requestBodyTable[key] = value
+ end
+ end
+
+ -- Encode the request body as JSON
+ local requestBody = json.encode(requestBodyTable)
 
  local headers = {
  ["Content-Type"] = "application/json",
- ["Authorization"] = "Bearer " .. api_key,
+ ["Authorization"] = "Bearer " .. api_key_value,
  }
 
- local requestBody = json.encode({
- model = "gpt-4o",
- messages = message_history,
- })
-
  local responseBody = {}
 
- local res, code, responseHeaders = https.request {
+ -- Make the HTTP/HTTPS request
+ local res, code, responseHeaders = request_library.request {
  url = api_url,
  method = "POST",
  headers = headers,
@@ -35,4 +72,4 @@ local function queryChatGPT(message_history)
  return response.choices[1].message.content
 end
 
-return queryChatGPT
+return queryChatGPT