ThunderAI/api_webchat/controller.js

293 lines
12 KiB
JavaScript

/*
* ThunderAI [https://micz.it/thunderbird-addon-thunderai/]
* Copyright (C) 2024 - 2026 Mic (m@micz.it)
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*
*
* This file contains a modified version of the code from the project at https://github.com/boxabirds/chatgpt-frontend-nobuild
* The original code has been released under the Apache License, Version 2.0.
*/
import { prefs_default, integration_options_config } from '../options/mzta-options-default.js';
import { placeholdersUtils } from '../js/mzta-placeholders.js';
import { getAPIsInitMessageString, convertNewlinesToBr } from '../js/mzta-utils.js';
import { loadPrompt } from '../js/mzta-prompts.js';
// Get the LLM to be used
const urlParams = new URLSearchParams(window.location.search);
const llm = urlParams.get('llm');
const call_id = urlParams.get('call_id');
const ph_def_val = urlParams.get('ph_def_val');
const prompt_id = urlParams.get('prompt_id');
const prompt_name = urlParams.get('prompt_name');
// Data received from the user
let promptData = null;
const messageInput = document.querySelector('message-input');
const messagesArea = document.querySelector('messages-area');
//console.log(">>>>>>>>>> controller.js DOMContentLoaded");
// console.log(">>>>>>>>>>> llm: " + llm);
// console.log(">>>>>>>>>>> call_id: " + call_id);
// The controller wires up all the components and workers together,
// managing the dependencies. A kind of "DI" class.
let worker = null;
const integration = llm.replace('_api', '');
const worker_path_map = {
chatgpt: '../js/workers/model-worker-openai_responses.js',
google_gemini: '../js/workers/model-worker-google_gemini.js',
ollama: '../js/workers/model-worker-ollama.js',
openai_comp: '../js/workers/model-worker-openai_comp.js',
anthropic: '../js/workers/model-worker-anthropic.js',
};
const worker_path = worker_path_map[integration];
if (worker_path) {
worker = new Worker(worker_path, { type: 'module' });
} else {
console.error('[ThunderAI] API WebChat Unknown LLM type:', llm);
}
if (worker) {
messagesArea.init(worker);
messageInput.init(worker);
messageInput.setMessagesArea(messagesArea);
if (integration_options_config[integration]) {
const integration_prefix = integration;
const options_config = integration_options_config[integration];
let prefsToGet = { do_debug: prefs_default.do_debug };
for (const key in options_config) {
prefsToGet[`${integration_prefix}_${key}`] = prefs_default[`${integration_prefix}_${key}`];
}
if (integration === 'openai_comp') {
prefsToGet.openai_comp_chat_name = prefs_default.openai_comp_chat_name;
}
let prefs_api = await browser.storage.sync.get(prefsToGet);
if (prompt_id) {
try {
const prompt = await loadPrompt(prompt_id);
if (prompt && prompt.api_type === llm) {
for (const key in options_config) {
const prefKey = `${integration_prefix}_${key}`;
if (prompt[prefKey] !== undefined) {
prefs_api[prefKey] = prompt[prefKey];
}
}
}
} catch (e) {
console.error("[ThunderAI] Error loading prompt settings:", e);
}
}
let i18nStrings = {};
const i18n_msg_key = integration === 'openai_comp' ? 'OpenAIComp_api_request_failed' : `${integration}_api_request_failed`;
i18nStrings[i18n_msg_key] = browser.i18n.getMessage(i18n_msg_key);
i18nStrings["error_connection_interrupted"] = browser.i18n.getMessage('error_connection_interrupted');
messageInput.setModel(prefs_api[`${integration_prefix}_model`]);
let llmName = "API";
switch(integration) {
case 'chatgpt': llmName = "ChatGPT"; break;
case 'google_gemini': llmName = "Google Gemini"; break;
case 'ollama': llmName = "Ollama Local"; break;
case 'openai_comp': llmName = prefs_api.openai_comp_chat_name || "OpenAI Comp"; break;
case 'anthropic': llmName = "Claude"; break;
}
messagesArea.setLLMName(llmName);
let workerInitMessage = {
type: 'init',
do_debug: prefs_api.do_debug,
i18nStrings: i18nStrings,
};
for (const key in options_config) {
const prefKey = `${integration_prefix}_${key}`;
workerInitMessage[prefKey] = prefs_api[prefKey];
}
worker.postMessage(workerInitMessage);
const additional_messages_config = {
chatgpt: [
{ key: 'store', labelKey: 'ChatGPT_chatgpt_api_store', type: 'boolean' },
{ key: 'developer_messages', labelKey: 'ChatGPT_Developer_Messages', type: 'string' },
{ key: 'temperature', labelKey: 'prefs_api_temperature', type: 'string' }
],
google_gemini: [
{ key: 'system_instruction', labelKey: 'GoogleGemini_SystemInstruction', type: 'string' },
{ key: 'temperature', labelKey: 'prefs_api_temperature', type: 'string' },
{ key: 'thinking_budget', labelKey: 'prefs_google_gemini_thinking_budget', type: 'string' }
],
ollama: [
{ key: 'think', labelKey: 'prefs_ollama_think', type: 'boolean' },
{ key: 'temperature', labelKey: 'prefs_api_temperature', type: 'string' },
{ key: 'num_ctx', labelKey: 'prefs_ollama_num_ctx', type: 'number_gt_zero' }
],
openai_comp: [
{ key: 'temperature', labelKey: 'prefs_api_temperature', type: 'string' }
],
anthropic: [
{ key: 'system_prompt', labelKey: 'Anthropic_System_Prompt', type: 'string' },
{ key: 'max_tokens', labelKey: 'prefs_OptionText_anthropic_max_tokens', type: 'number_gt_zero' },
{ key: 'temperature', labelKey: 'prefs_api_temperature', type: 'string' }
]
};
const getAdditionalMessages = (integration, prefs) => {
const messages = [];
const config = additional_messages_config[integration];
if (!config) return messages;
for (const item of config) {
const prefKey = `${integration}_${item.key}`;
const value = prefs[prefKey];
if (value !== undefined && value !== null && value !== '') {
let displayValue;
let shouldAdd = false;
switch (item.type) {
case 'boolean':
displayValue = value ? 'Yes' : 'No';
shouldAdd = true;
break;
case 'string':
if (value.length > 0) {
displayValue = value;
shouldAdd = true;
}
break;
case 'number_gt_zero':
if (value > 0) {
displayValue = value;
shouldAdd = true;
}
break;
}
if (shouldAdd) {
messages.push({ label: browser.i18n.getMessage(item.labelKey), value: displayValue });
}
}
}
return messages;
};
let additional_text_elements = [];
additional_text_elements.push({label: browser.i18n.getMessage("prompt_string"), value: '[' + prompt_id + '] ' + decodeURIComponent(prompt_name)});
additional_text_elements.push(...getAdditionalMessages(integration, prefs_api));
const api_strings = {
chatgpt: "ChatGPT API",
google_gemini: "Google Gemini API",
ollama: "Ollama API",
openai_comp: "OpenAI Compatible API",
anthropic: "Claude API"
};
messagesArea.appendUserMessage(getAPIsInitMessageString({
api_string: api_strings[integration],
model_string: prefs_api[`${integration_prefix}_model`],
host_string: prefs_api[`${integration_prefix}_host`],
version_string: prefs_api[`${integration_prefix}_version`],
additional_messages: additional_text_elements
}), "info");
//console.log(`>>>>>>>>>>>>> command: ${llm}_ready_${call_id}`,)
browser.runtime.sendMessage({
command: `${llm}_ready_${call_id}`,
window_id: (await browser.windows.getCurrent()).id
});
}
}
//let prefs_ph = await browser.storage.sync.get({placeholders_use_default_value: false});
// Event listeners for worker messages
worker.onmessage = async function(event) {
const { type, payload } = event.data;
switch (type) {
case 'messageSent':
messageInput.handleMessageSent();
break;
case 'newToken':
messagesArea.handleNewToken(payload.token);
messageInput.setStatusMessage(browser.i18n.getMessage("apiwebchat_receiving_data") + '...');
break;
case 'tokensDone':
await messagesArea.handleTokensDone(promptData);
messageInput.enableInput();
break;
case 'error':
messagesArea.appendBotMessage(payload,'error');
messageInput.enableInput();
break;
default:
console.error('[ThunderAI] Unknown event type from API worker:', type);
}
};
// handling commands from the backgound page
browser.runtime.onMessage.addListener((message, sender, sendResponse) => {
switch (message.command) {
case "api_send":
promptData = message;
//send the received prompt to the llm api
if(message.do_custom_text=="1") {
messageInput._showCustomTextField();
}else{
sendPrompt(message);
}
break;
case 'api_send_custom_text':
let userInput = message.custom_text;
if(userInput !== null) {
if(!placeholdersUtils.hasPlaceholder(promptData.prompt, 'additional_text')){
// no additional_text placeholder, do as usual
promptData.prompt += " " + userInput;
}else{
// we have the additional_text placeholder, do the magic!
let finalSubs = {};
finalSubs["additional_text"] = userInput;
promptData.prompt = placeholdersUtils.replacePlaceholders({
text: promptData.prompt,
replacements: finalSubs,
use_default_value: ph_def_val==='1'
})
}
sendPrompt(promptData);
}
break;
case "api_error":
messagesArea.appendBotMessage(message.error,'error');
messageInput.enableInput();
break;
}
});
function sendPrompt(message){
messageInput._setMessageInputValue(convertNewlinesToBr(message.prompt));
messageInput._handleNewChatMessage();
}