JAVASCRIPT extract.js

// Google Maps search results extractor

() => {
    const cards = document.querySelectorAll('div.Nv2PK');
    const results = [];

    for (let card of cards) {
        const mainLink = card.querySelector('a.hfpxzc');
        if (!mainLink) continue;

        const name = mainLink.getAttribute('aria-label') || '';
        if (!name) continue;

        // Get all links in this card
        const links = Array.from(card.querySelectorAll('a'));
        let website = '', phone = '', mapsUrl = '';

        for (let link of links) {
            const href = link.getAttribute('href') || '';

            // Maps URL
            if (href.includes('/maps/place/') && !href.startsWith('http')) {
                mapsUrl = 'https://www.google.com' + href.split('?')[0];
            } else if (href.includes('/maps/place/')) {
                mapsUrl = href.split('?')[0];
            }

            // Phone from tel: link
            if (href.startsWith('tel:')) {
                phone = href.replace('tel:', '');
            }
        }

        // Website: external link that's not Google Maps, social, or booking
        for (let link of links) {
            const href = link.getAttribute('href') || '';
            const text = (link.textContent || '').trim().toLowerCase();

            if (href.startsWith('http') &&
                !href.includes('google.com') &&
                !href.includes('g.page') &&
                !href.includes('fb.com') &&
                !href.includes('facebook.com') &&
                !href.includes('instagram.com') &&
                !href.includes('yelp.com') &&
                !href.includes('recreateai.com') &&
                !href.includes('servicetitan.com') &&
                !text.includes('reviews') &&
                !text.includes('directions') &&
                !text.includes('book online') &&
                !text.includes('save') &&
                !text.includes('share')) {
                website = href;
                break;
            }
        }

        // Phone fallback: regex from card text
        if (!phone) {
            const phoneMatch = card.textContent.match(/\(?(\d{3})\)?[\s.-](\d{3})[\s.-](\d{4})/);
            if (phoneMatch) phone = phoneMatch[0];
        }

        // Rating: look for span with star emoji or number followed by review count
        let rating = '';
        let reviews = '';
        // Google Maps uses a specific structure: a div containing the rating number
        // followed by (N reviews) text
        const cardText = card.textContent;
        const ratingMatch = cardText.match(/(\d\.\d)\s+\(([\d,]+)\s*review[s]?\)/);
        if (ratingMatch) {
            rating = ratingMatch[1];
            reviews = ratingMatch[2];
        } else {
            // Try simpler: just find a decimal number that looks like a rating
            const simpleRating = cardText.match(/(\d\.\d)\s*\(/);
            if (simpleRating) {
                rating = simpleRating[1];
            }
        }

        // Address: look for line starting with number and containing street info
        let address = '';
        const lines = cardText.split('\n').map(l => l.trim()).filter(l => l);
        for (let line of lines) {
            if (/^\d+\s+[A-Za-z]/.test(line) && line.length > 10 && line.length < 80) {
                if (!/\d{3,4}$/.test(line) && !line.includes('(')) {
                    address = line;
                    break;
                }
            }
        }

        results.push({
            name,
            phone,
            website,
            rating,
            reviews,
            address,
            maps_url: mapsUrl,
        });
    }

    return results;
}