Apollo Scraper

apollo.js

function clickAccessEmailIfAvailable() {
  const tbodys = document.querySelectorAll(
    ".finder-results-list-panel-content table tbody"
  );

  for (let i = 0; i < tbodys.length; i++) {
    const tbody = tbodys[i];
    const buttons = tbody.querySelectorAll("button");
    // Create an array to store buttons with text including "Access Email"
    const filteredButtons = [];
    // Loop through the selected buttons and filter them
    buttons.forEach((button) => {
      if (button.textContent.includes("Access Email")) {
        filteredButtons.push(button);
      }
    });
    if (filteredButtons.length === 1) {
      filteredButtons[0].click();
    }
  }
}

function scrapePeople() {
  const people = [];
  const tbodys = document.querySelectorAll(
    ".finder-results-list-panel-content table tbody"
  );

  for (let i = 0; i < tbodys.length; i++) {
    const tbody = tbodys[i];
    const name = tbody.querySelector("tr td:nth-child(1)").textContent;
    const firstName = name?.split(" ")?.[0];
    const lastName = name?.split(" ")?.[1];
    const title = tbody.querySelector("tr td:nth-child(2)").textContent;
    const company = tbody.querySelector(
      "tr td:nth-child(3) > span > div > div"
    ).textContent;
    const location = tbody.querySelector("tr td:nth-child(5)").textContent;
    const employeeCount = tbody.querySelector("tr td:nth-child(6)").textContent;
    const email = tbody.querySelector("tr td:nth-child(7)").textContent;
    const industry = tbody.querySelector("tr td:nth-child(8)").textContent;
    const keywords = tbody.querySelector("tr td:nth-child(9)").textContent;
    people.push({
      name,
      firstName,
      lastName,
      title,
      company,
      location,
      employeeCount,
      email,
      industry,
      keywords,
    });
  }
  console.log("people", people);
  return people;
}

function createCSV(jsonData, fileName) {
  // Convert JSON to CSV
  const csvData = [];

  // Extract the headers
  const headers = Object.keys(jsonData[0]);
  csvData.push(headers.join(","));

  jsonData.forEach((item) => {
    const row = [];
    for (const key in item) {
      if (item.hasOwnProperty(key)) {
        const value = item[key]?.includes(",") ? `"${item[key]}"` : item[key];
        row.push(value);
      }
    }
    csvData.push(row.join(","));
  });

  // Create a Blob containing the CSV data
  const csvBlob = new Blob([csvData.join("\n")], {
    type: "text/csv;charset=utf-8",
  });

  // Create a URL for the Blob
  const csvUrl = URL.createObjectURL(csvBlob);

  // Create a link element
  const link = document.createElement("a");
  link.href = csvUrl;
  link.target = "_blank";
  link.download = fileName;

  // Append the link to the body
  document.body.appendChild(link);

  // Trigger a click event on the link
  link.click();

  // Remove the link and revoke the Blob URL
  document.body.removeChild(link);
  URL.revokeObjectURL(csvUrl);
}

async function scrapeApollo() {
  const people = [];
  let page = 1;
  let nextButton = document.querySelector("button[aria-label='right-arrow']");
  let disabled = nextButton.disabled;

  while (!disabled) {
    console.log(
      `Scraping page ${page}. If you need anything else web scraped, email me: adrian@thewebscrapingguy.com`
    );
    const peeps = scrapePeople();
    people.push(...peeps);
    nextButton = document.querySelector("button[aria-label='right-arrow']");
    disabled = nextButton.disabled;
    nextButton.click();
    await new Promise((resolve) => setTimeout(resolve, 2000));
    page++;
  }

  console.log(
    `Congrats! 🎉 You just scraped ${people.length} profiles! If you want more leads, or want anything else scraped, email me: adrian@thewebscrapingguy.com`
  );

  createCSV(people, `apollo_${new Date().getTime()}.csv`);
}

await scrapeApollo();