Adding: Web Access allowing the AI to browse URLs

default.env

@@ -1,16 +1,30 @@
+# Discord Token
 THE_TOKEN = "DISCORD_TOKEN_HERE"
+
+# The Channel IDs the bot will operate in seperated by commas
 CHANNEL_IDS = 1094494101631680653,1094628334727614605
+
+# The INIT prompt for all conversations.
 INIT_PROMPT = "Assistant name: ChatAI. You code, write and provide any information without any mistakes."
-ROOT_PORT = 8000
+
-DATA_DIR = /home/USERNAME/weights
+# Loading Emebed Refresh Timing
-CACHE = 1
+REFRESH_INTERVAL=10
-N_THREADS = 4
+
+# Max Content to fetch from given URLs
+MAX_CONTENT_LENGTH=2000
+
+# Max tokens for Generations
 MAX_TOKENS = 1024
 
-# Bot specific settings
+# END BOT SPECIFIC SETTINGS
-REFRESH_INTERVAL=7
 
 # ROOT_IP is only used when running the bot without docker compose
 ROOT_IP = 192.168.0.15
 
+# PORT  is only used when running the bot without docker compose
+ROOT_PORT = 8000
 
+# Directory to your models (llama.cpp specfic settings)
+DATA_DIR = /home/USERNAME/weights
+CACHE = 1
+N_THREADS = 4

llamabot.js

@@ -175,6 +175,8 @@ client.on('messageCreate', async (message) => {
             conversation.busy = false;
         }
         conversations.set(userID, conversation); // Update user's conversation map in memory
+        console.log(conversation)
+
     } catch (err) {
         console.error(err);
         sendRand(errorMessages)
@@ -184,8 +186,60 @@ client.on('messageCreate', async (message) => {
     }
 });
 
+import cheerio from 'cheerio';
 
 async function generateResponse(conversation, message) {
+
+    // Check if message contains a URL
+    const urlRegex = /(https?:\/\/[^\s]+)/g;
+    const urls = message.content.match(urlRegex);
+
+    if (urls) {
+        // If there are multiple URLs, process them one by one
+        for (const url of urls) {
+            try {
+                const res = await fetch(url);
+                const html = await res.text();
+                const $ = cheerio.load(html);
+
+                // Extract page title, meta description and content
+                const pageTitle = $('head title').text().trim();
+                const pageDescription = $('head meta[name="description"]').attr('content');
+                const pageContent = $('body').text().trim();
+
+                // Construct response message with page details
+                let response = `Title: ${pageTitle}\n`;
+                if (pageDescription) {
+                    response += `Description: ${pageDescription}\n`;
+                }
+                if (pageContent) {
+                    const MAX_CONTENT_LENGTH = 1900;
+                    let plainTextContent = $('<div>').html(pageContent).text().trim().replace(/[\r\n]+/g, ' ');
+
+                    if (plainTextContent.length > MAX_CONTENT_LENGTH) {
+                        plainTextContent = plainTextContent.substring(0, MAX_CONTENT_LENGTH) + '...';
+                        response += `Content: ${plainTextContent}\n`;
+                    } else {
+                        response += `Content: ${plainTextContent}\n`;
+                    }
+                }
+
+                response += `URL: ${url}`;
+
+                // Append bot message to conversation history
+                conversation.messages.push({
+                    role: 'user',
+                    content: "Data from the page is: " + response,
+                });
+
+                console.log("A URL was provided, response: " + response)
+
+            } catch (err) {
+                console.error(err);
+                sendRand(errorMessages);
+            }
+        }
+    }
     const controller = new AbortController();
     const timeout = setTimeout(() => {
         controller.abort();