cannaiq/backend/dist/scripts/platform-id-extractor.js

"use strict";
/**
 * Platform ID Extractor - Standalone script for extracting Dutchie platform IDs
 *
 * This script visits dispensary websites to capture their Dutchie retailerId
 * by intercepting network requests to the Dutchie GraphQL API.
 *
 * It does NOT use the main orchestrator - it's a standalone browser-based tool.
 */
Object.defineProperty(exports, "__esModule", { value: true });
const playwright_1 = require("playwright");
const pg_1 = require("pg");
const pool = new pg_1.Pool({
    connectionString: process.env.DATABASE_URL
});
async function extractPlatformId(browser, dispensary) {
    let capturedId = null;
    let captureSource = null;
    let errorMsg = null;
    const context = await browser.newContext({
        userAgent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36'
    });
    const page = await context.newPage();
    // Patterns to match retailer IDs in various formats
    const idPatterns = [
        /["']retailerId["']\s*:\s*["']([a-f0-9]{24})["']/i,
        /["']dispensaryId["']\s*:\s*["']([a-f0-9]{24})["']/i,
        /retailer["']?\s*:\s*["']([a-f0-9]{24})["']/i,
        /dutchie\.com\/embedded-menu\/([a-f0-9]{24})/i,
        /dutchie\.com\/dispensary\/([a-f0-9]{24})/i,
        /plus\.dutchie\.com\/plus\/([a-f0-9]{24})/i,
        /retailerId=([a-f0-9]{24})/i,
        /\/([a-f0-9]{24})(?:\/|\?|$)/i, // Generic ID in URL path
    ];
    // Intercept network requests
    page.on('request', (request) => {
        if (capturedId)
            return;
        const url = request.url();
        if (url.includes('dutchie') || url.includes('api.dutchie')) {
            // Check URL for retailer ID
            for (const pattern of idPatterns) {
                const match = url.match(pattern);
                if (match && match[1] && match[1].length === 24) {
                    capturedId = match[1];
                    captureSource = 'request_url';
                    break;
                }
            }
            // Check POST data
            const postData = request.postData();
            if (postData && !capturedId) {
                for (const pattern of idPatterns) {
                    const match = postData.match(pattern);
                    if (match && match[1] && match[1].length === 24) {
                        capturedId = match[1];
                        captureSource = 'request_body';
                        break;
                    }
                }
            }
        }
    });
    try {
        console.log(`\n[${dispensary.id}] ${dispensary.name}: ${dispensary.website}`);
        // Load main page
        await page.goto(dispensary.website, {
            waitUntil: 'domcontentloaded',
            timeout: 25000
        });
        await page.waitForTimeout(2000);
        // Check page content
        if (!capturedId) {
            const content = await page.content();
            for (const pattern of idPatterns) {
                const match = content.match(pattern);
                if (match && match[1] && match[1].length === 24) {
                    capturedId = match[1];
                    captureSource = 'page_content';
                    break;
                }
            }
        }
        // Check __NEXT_DATA__
        if (!capturedId) {
            const nextData = await page.evaluate(() => {
                const el = document.getElementById('__NEXT_DATA__');
                return el?.textContent || null;
            });
            if (nextData) {
                for (const pattern of idPatterns) {
                    const match = nextData.match(pattern);
                    if (match && match[1] && match[1].length === 24) {
                        capturedId = match[1];
                        captureSource = '__NEXT_DATA__';
                        break;
                    }
                }
            }
        }
        // Check iframes
        if (!capturedId) {
            const iframes = await page.evaluate(() => {
                return Array.from(document.querySelectorAll('iframe')).map(f => f.src);
            });
            for (const src of iframes) {
                if (src.includes('dutchie')) {
                    const match = src.match(/([a-f0-9]{24})/i);
                    if (match) {
                        capturedId = match[1];
                        captureSource = 'iframe_src';
                        break;
                    }
                }
            }
        }
        // Check scripts
        if (!capturedId) {
            const scripts = await page.evaluate(() => {
                return Array.from(document.querySelectorAll('script'))
                    .map(s => s.src || s.innerHTML?.substring(0, 1000))
                    .filter(Boolean);
            });
            for (const script of scripts) {
                if (script && (script.includes('dutchie') || script.includes('retailerId'))) {
                    for (const pattern of idPatterns) {
                        const match = script.match(pattern);
                        if (match && match[1] && match[1].length === 24) {
                            capturedId = match[1];
                            captureSource = 'script';
                            break;
                        }
                    }
                    if (capturedId)
                        break;
                }
            }
        }
        // Try navigating to menu/shop page
        if (!capturedId) {
            const menuLink = await page.evaluate(() => {
                const links = Array.from(document.querySelectorAll('a'));
                for (const link of links) {
                    const href = link.href?.toLowerCase() || '';
                    const text = link.textContent?.toLowerCase() || '';
                    if (href.includes('menu') || href.includes('shop') || href.includes('order') ||
                        text.includes('menu') || text.includes('shop') || text.includes('order')) {
                        return link.href;
                    }
                }
                return null;
            });
            if (menuLink && !menuLink.startsWith('javascript:')) {
                try {
                    console.log(`  -> Following menu link: ${menuLink.substring(0, 60)}...`);
                    await page.goto(menuLink, { waitUntil: 'domcontentloaded', timeout: 20000 });
                    await page.waitForTimeout(3000);
                    // Recheck all sources on new page
                    const newContent = await page.content();
                    for (const pattern of idPatterns) {
                        const match = newContent.match(pattern);
                        if (match && match[1] && match[1].length === 24) {
                            capturedId = match[1];
                            captureSource = 'menu_page_content';
                            break;
                        }
                    }
                    // Check iframes on new page
                    if (!capturedId) {
                        const newIframes = await page.evaluate(() => {
                            return Array.from(document.querySelectorAll('iframe')).map(f => f.src);
                        });
                        for (const src of newIframes) {
                            if (src.includes('dutchie')) {
                                const match = src.match(/([a-f0-9]{24})/i);
                                if (match) {
                                    capturedId = match[1];
                                    captureSource = 'menu_page_iframe';
                                    break;
                                }
                            }
                        }
                    }
                }
                catch (navError) {
                    // Menu navigation failed, continue
                }
            }
        }
        // Final wait for async content
        if (!capturedId) {
            await page.waitForTimeout(3000);
            // Final iframe check
            const finalIframes = await page.evaluate(() => {
                return Array.from(document.querySelectorAll('iframe')).map(f => f.src);
            });
            for (const src of finalIframes) {
                if (src.includes('dutchie')) {
                    const match = src.match(/([a-f0-9]{24})/i);
                    if (match) {
                        capturedId = match[1];
                        captureSource = 'delayed_iframe';
                        break;
                    }
                }
            }
        }
        if (capturedId) {
            console.log(`  ✓ Found: ${capturedId} (${captureSource})`);
        }
        else {
            console.log(`  ✗ Not found`);
        }
    }
    catch (e) {
        errorMsg = e.message.substring(0, 100);
        console.log(`  ✗ Error: ${errorMsg}`);
    }
    finally {
        await context.close();
    }
    return {
        id: dispensary.id,
        name: dispensary.name,
        website: dispensary.website,
        platformId: capturedId,
        source: captureSource,
        error: errorMsg
    };
}
async function main() {
    // Get specific dispensary ID from command line, or process all missing
    const targetId = process.argv[2] ? parseInt(process.argv[2], 10) : null;
    let query;
    let params = [];
    if (targetId) {
        query = `
      SELECT id, name, website
      FROM dispensaries
      WHERE id = $1
        AND website IS NOT NULL AND website != ''
    `;
        params = [targetId];
    }
    else {
        query = `
      SELECT id, name, website
      FROM dispensaries
      WHERE state = 'AZ'
        AND menu_type = 'dutchie'
        AND (platform_dispensary_id IS NULL OR platform_dispensary_id = '')
        AND website IS NOT NULL AND website != ''
      ORDER BY name
    `;
    }
    const result = await pool.query(query, params);
    if (result.rows.length === 0) {
        console.log('No dispensaries to process');
        await pool.end();
        return;
    }
    console.log(`\n=== Platform ID Extractor ===`);
    console.log(`Processing ${result.rows.length} dispensaries...\n`);
    const browser = await playwright_1.chromium.launch({
        headless: true,
        args: ['--no-sandbox', '--disable-setuid-sandbox']
    });
    const results = [];
    for (const dispensary of result.rows) {
        const extractionResult = await extractPlatformId(browser, dispensary);
        results.push(extractionResult);
        // Update database immediately if found
        if (extractionResult.platformId) {
            await pool.query('UPDATE dispensaries SET platform_dispensary_id = $1 WHERE id = $2', [extractionResult.platformId, extractionResult.id]);
        }
    }
    await browser.close();
    // Summary
    console.log('\n' + '='.repeat(60));
    console.log('SUMMARY');
    console.log('='.repeat(60));
    const found = results.filter(r => r.platformId);
    const notFound = results.filter(r => !r.platformId);
    console.log(`\nFound: ${found.length}/${results.length}`);
    if (found.length > 0) {
        console.log('\nSuccessful extractions:');
        found.forEach(r => console.log(`  [${r.id}] ${r.name} -> ${r.platformId} (${r.source})`));
    }
    if (notFound.length > 0) {
        console.log(`\nNot found: ${notFound.length}`);
        notFound.forEach(r => {
            const reason = r.error || 'No Dutchie ID detected';
            console.log(`  [${r.id}] ${r.name}: ${reason}`);
        });
    }
    await pool.end();
}
main().catch(e => {
    console.error('Fatal error:', e);
    process.exit(1);
});