Telegram chat mining client-side

telegram_scrape.js

(async () => {
    // --- Configuration ---
    const scrollDelay = 2000; // 2 seconds. Increase if your connection is slow.
    const maxRetries = 5; // Will stop after 5 consecutive scrolls with no new messages.

    // --- Script Logic ---
    let allMessages = [];
    const messageContainer = document.querySelector('.MessageList.custom-scroll');
    const seenMessageIds = new Set();

    if (!messageContainer) {
        console.error("Error: Could not find the message container with selector '.MessageList.custom-scroll'. Please make sure the chat is active.");
        return;
    }

    // Function to find new messages, extract text, and add to our set of seen messages
    const getNewMessages = () => {
        const messageElements = document.querySelectorAll('.message-list-item');
        const newMessages = [];

        messageElements.forEach(msgElement => {
            const messageId = msgElement.getAttribute('data-message-id');
            const textContentElement = msgElement.querySelector('.text-content');
            
            // Process only if it's a new message with text content
            if (messageId && !seenMessageIds.has(messageId) && textContentElement) {
                // Clean up the text: remove newlines and extra spaces
                const messageText = textContentElement.innerText.trim().replace(/\s*\n\s*/g, ' '); 
                if (messageText) {
                    newMessages.push(messageText);
                    seenMessageIds.add(messageId);
                }
            }
        });

        return newMessages;
    };

    // Function to scroll the message container to the very bottom
    const scrollToBottom = () => {
        // We set scrollTop to a very large number to ensure it scrolls to the absolute bottom
        messageContainer.scrollTop = messageContainer.scrollHeight;
    };

    const delay = (ms) => new Promise(resolve => setTimeout(resolve, ms));

    let noNewMessagesCount = 0;

    console.log("--- Starting chat mining. Will scroll to the bottom repeatedly. ---");

    while (noNewMessagesCount < maxRetries) {
        const newMessages = getNewMessages();

        if (newMessages.length > 0) {
            allMessages.push(...newMessages); // Append new messages to the end of the array
            noNewMessagesCount = 0; // Reset the counter since we found new content
            console.log(`--- Found ${newMessages.length} new messages. Total collected: ${allMessages.length} ---`);
            console.log("Current full array of messages:", allMessages);
        } else {
            noNewMessagesCount++;
            console.log(`No new messages found. Attempt ${noNewMessagesCount} of ${maxRetries}...`);
        }

        // Scroll down to trigger loading more messages
        scrollToBottom();

        // Wait for the new messages to load
        await delay(scrollDelay);
    }

    console.log("--- Mining finished! ---");
    console.log(`Stopped after ${maxRetries} attempts with no new messages.`);
    console.log("Total messages collected:", allMessages.length);
    console.log("Final array of all messages:", allMessages);
})();

Useful for summarizing large group chats. Telegram has a sliding window mechanism that prevents loading all messages at once to html, so chunking is required.

Scrolling down atm; reversing is easy with:

    const scrollToTop = () => {
        messageContainer.scrollTop = 0;
    };

Then copy the array from the console and use Gemini to summarize.