parse twitter

parse-twitter.js

// go to x.com and copy paste below code to console
// use https://react-tweet.vercel.app/ to render tweets in your app using tweet id

function delay(ms) {
  return new Promise(resolve => setTimeout(resolve, ms));
}

function extractItems() {
  const tweets = document.querySelectorAll('[data-testid="tweet"]');
  const items = [];

  for (let tweet of tweets) {
    const text = tweet.querySelector('[data-testid="tweet"] div[lang]')
      ?.innerText;
    const link = tweet.querySelector('[data-testid="tweet"] a[aria-label]')
      ?.href;
    
    // This version limits the ID to 1-19 digits (Twitter's status IDs are currently 19 digits).
    const tweetId = link.match(/\/status\/([0-9]{1,19})/)?.[1];
      
    const img = tweet.querySelector('[data-testid="tweetPhoto"] img')?.src;
    const createdAt = tweet.querySelector('[data-testid="tweet"] time')
      ?.dateTime;

    const avatar = tweet.querySelector('[data-testid="tweet"] a div img')?.src;

    const userName = tweet.querySelector('[data-testid="tweet"] [dir=ltr]')
      ?.innerText;

    const replies = tweet.querySelector('[data-testid="reply"]')?.innerText;
    const likes = tweet.querySelector('[data-testid*="like"]')?.innerText;
    const retweets = tweet.querySelector('[data-testid="retweet"]')?.innerText;
    

    items.push({
      tweetId,
      userName,
      avatar,
      text,
      link,
      img,
      createdAt,
      replies,
      likes,
      retweets,
    });
  }
  return items;
}

let items = []

for(let i=0; i<5; i++){
    const newItems = extractItems()
    items = [...items, ...newItems];

    window.scrollBy(0, 3000)
    await delay(1500)
}

console.log({items})