This code automates the process of extracting a LinkedIn user's profile information, including name, title, URL, phone number, and email address.

Run example

npm run import -- "Scrape LinkedIn profile"

Scrape LinkedIn profile

function visitMyProfile() {
    // TODO: add check for needing to go to LinkedIn
    // TODO: add check for needing to login
    return client
        .pause(2000)
        .click('#nav-settings__dropdown-trigger')
        .pause(500)
        .click('.nav-settings__view-profile-link')
        .pause(2000);
}

function readLinkedInProfileInfo() {
    var name, title, url, phone, email;
    return client
        .getUrl().then(url => url.indexOf('/in/') == -1
            ? visitMyProfile()
            : Promise.resolve([]))
        .isExisting('.contact-see-more-less')
        .then(is => is ? client.click('.contact-see-more-less') : client)
        .pause(500)
        .then(() => getAllXPath({
            name: '//*[contains(@class, "pv-top-card-section__name")]//text()',
            title: '//*[contains(@class, "pv-top-card-section__headline")]//text()',
            url: '//*[contains(@class, "ci-vanity-url")]//*[contains(@class, "pv-contact-info__contact-item")]//text()',
            phone: '//*[contains(@class, "ci-phone")]//*[contains(@class, "pv-contact-info__contact-item")]//text()',
            email: '//*[contains(@class, "ci-email")]//*[contains(@class, "pv-contact-info__contact-item")]//text()'
        }))
        .then(r => ({
            name: (typeof r.name === 'string' ? [r.name] : (r.name || [])).join('').trim(),
            title: (typeof r.title === 'string' ? [r.title] : (r.title || [])).join('').trim(),
            url: (typeof r.url === 'string' ? [r.url] : (r.url || [])).join('').trim(),
            phone: (typeof r.phone === 'string' ? [r.phone] : (r.phone || [])).join('').trim(),
            email: (typeof r.email === 'string' ? [r.email] : (r.email || [])).join('').trim()
        }));
};
module.exports = readLinkedInProfileInfo;

/**
 * Function to visit the user's profile page.
 * @param {object} client - The browser client object.
 * @return {Promise} A promise that resolves when the user's profile page is loaded.
 */
function visitMyProfile(client) {
    // Check if we need to go to LinkedIn and handle accordingly.
    if (!isOnLinkedin()) {
        return navigateToLinkedIn(client);
    }

    // Check if we need to login and handle accordingly.
    if (!isLoggedIn()) {
        return login(client);
    }

    // Wait for the navigation menu to load and click on the profile link.
    return client
       .pause(2000)
       .click('#nav-settings__dropdown-trigger')
       .pause(500)
       .click('.nav-settings__view-profile-link')
       .pause(2000);
}

/**
 * Function to read LinkedIn profile info.
 * @param {object} client - The browser client object.
 * @return {Promise} A promise that resolves with an object containing the user's profile info.
 */
function readLinkedInProfileInfo(client) {
    // Check if we are on the user's profile page and return an empty array if not.
    const url = client.getUrl();
    if (url.indexOf('/in/') == -1) {
        return visitMyProfile(client).then(() => []);
    }

    // Wait for the contact info to load and extract the relevant info.
    return client
       .isExisting('.contact-see-more-less')
       .then(is => is? client.click('.contact-see-more-less') : client)
       .pause(500)
       .then(() => getAllXPath({
            name: '//*[contains(@class, "pv-top-card-section__name")]//text()',
            title: '//*[contains(@class, "pv-top-card-section__headline")]//text()',
            url: '//*[contains(@class, "ci-vanity-url")]//*[contains(@class, "pv-contact-info__contact-item")]//text()',
            phone: '//*[contains(@class, "ci-phone")]//*[contains(@class, "pv-contact-info__contact-item")]//text()',
            email: '//*[contains(@class, "ci-email")]//*[contains(@class, "pv-contact-info__contact-item")]//text()'
        }))
       .then(r => ({
            name: extractText(r.name),
            title: extractText(r.title),
            url: extractText(r.url),
            phone: extractText(r.phone),
            email: extractText(r.email)
        }));
}

// Helper function to check if we are on LinkedIn.
function isOnLinkedin() {
    // Implement logic to check if we are on LinkedIn.
}

// Helper function to navigate to LinkedIn.
function navigateToLinkedIn(client) {
    // Implement logic to navigate to LinkedIn.
}

// Helper function to check if we are logged in.
function isLoggedIn() {
    // Implement logic to check if we are logged in.
}

// Helper function to login.
function login(client) {
    // Implement logic to login.
}

// Helper function to extract text from an array or string.
function extractText(text) {
    return (typeof text ==='string'? [text] : (text || [])).join('').trim();
}

module.exports = readLinkedInProfileInfo;

This code defines a function readLinkedInProfileInfo that automates the process of extracting information from a LinkedIn profile.

Here's a breakdown:

1. visitMyProfile Function:

   • This function is intended to navigate to the user's LinkedIn profile.
   • It's currently incomplete and lacks checks for whether the user is already on LinkedIn or needs to log in.

2. readLinkedInProfileInfo Function:

   • This function orchestrates the entire profile information extraction process:
     • It first checks if the user is already on their profile page. If not, it calls visitMyProfile to navigate there.
     • It then expands the "See More" section to reveal all contact information.
     • It uses getAllXPath (a custom function not shown here) to extract the name, title, URL, phone number, and email address from the profile page using XPath selectors.
     • Finally, it returns an object containing the extracted information.

3. Module Export:

   • The readLinkedInProfileInfo function is exported as the main function of this module.

Let me know if you have any more questions!