robot-proveedores/robot.js

import { chromium } from 'playwright';
import pg from 'pg';

const { DATABASE_URL } = process.env;

if (!DATABASE_URL) {
    console.error("❌ Error: No hay DATABASE_URL definida.");
    process.exit(1);
}

const pool = new pg.Pool({ connectionString: DATABASE_URL, ssl: false });
const HEADLESS = true; 

async function main() {
    console.log("🤖 ROBOT MODO: CIRUJANO + ASPIRADORA INTELIGENTE");
    
    while (true) {
        const client = await pool.connect();
        try {
            const res = await client.query("SELECT * FROM provider_credentials WHERE status = 'active'");
            const credentials = res.rows;

            for (const cred of credentials) {
                let password = Buffer.from(cred.password_hash, 'base64').toString('utf-8');
                console.log(`\n🔄 Procesando ${cred.provider.toUpperCase()}...`);

                if (cred.provider === 'multiasistencia') {
                    await runMultiasistencia(cred.owner_id, cred.username, password);
                } else if (cred.provider === 'homeserve') {
                    await runHomeserve(cred.owner_id, cred.username, password);
                }
                await client.query("UPDATE provider_credentials SET last_sync = NOW() WHERE id = $1", [cred.id]);
            }
        } catch (e) {
            console.error("❌ Error ciclo:", e.message);
        } finally {
            client.release();
        }
        console.log("\n💤 Durmiendo 15 minutos...");
        await new Promise(r => setTimeout(r, 15 * 60 * 1000));
    }
}

// ==========================================
// 🏥 MULTIASISTENCIA V7 (EXTRACCIÓN POR PATRONES DE TEXTO)
// ==========================================
async function runMultiasistencia(ownerId, user, pass) {
    const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] });
    const context = await browser.newContext();
    const page = await context.newPage();

    try {
        console.log("🌍 [Multi] Accediendo...");
        await page.goto('https://web.multiasistencia.com/w3multi/acceso.php', { timeout: 60000 });

        const userInput = await page.$('input[name="usuario"]') || await page.$('input[type="text"]');
        if(userInput) {
            await userInput.fill(user);
            await page.fill('input[type="password"]', pass);
            await page.click('input[type="submit"]');
            await page.waitForTimeout(5000);
        }

        await page.goto('https://web.multiasistencia.com/w3multi/frepasos_new.php?refresh=1', { waitUntil: 'domcontentloaded' });
        
        const expedientes = await page.evaluate(() => {
            const links = Array.from(document.querySelectorAll('a[href*="reparacion="]'));
            return Array.from(new Set(links.map(a => a.href.match(/reparacion=(\d+)/)?.[1]).filter(Boolean)));
        });

        console.log(`🔍 [Multi] Procesando ${expedientes.length} expedientes.`);

        for (const ref of expedientes) {
            await page.goto(`https://web.multiasistencia.com/w3multi/repasos1.php?reparacion=${ref}`, { waitUntil: 'domcontentloaded' });
            
            const fullData = await page.evaluate(() => {
                const text = document.body.innerText;
                const data = {};

                // Función para extraer texto entre dos palabras clave
                const extract = (start, end) => {
                    const regex = new RegExp(`${start}\\s*:?\\s*([\\s\\S]*?)(?=${end}|$)`, 'i');
                    const match = text.match(regex);
                    return match ? match[1].trim() : null;
                };

                // CAPTURA POR BLOQUES DE TEXTO (IGNORANDO TABLAS)
                data['Número Reparación'] = extract('Número Reparación', 'Número de Póliza');
                data['Número de Póliza'] = extract('Número de Póliza', 'Nombre Cliente');
                data['Nombre Cliente'] = extract('Nombre Cliente', 'Dirección');
                data['Dirección'] = extract('Dirección', 'Baremo');
                data['Distrito Postal'] = extract('Distrito Postal', 'Más información');
                data['Procedencia'] = extract('Procedencia', 'Forma de pago');
                data['Descripción'] = extract('Descripción de la Reparación', 'Urgente');
                data['Tramitador'] = extract('Tramitador', 'Teléfono del Cliente');
                data['Urgente'] = extract('Urgente', 'Fecha realización');
                data['Fecha realización'] = extract('Fecha realización', 'Estado');

                // Teléfonos (Regex de 9 dígitos limpia)
                const phones = text.match(/[6789]\d{8}/g);
                if (phones) data['phone_fixed'] = [...new Set(phones)].join(' / ');

                // Limpieza final: si el valor capturado contiene el nombre de otra etiqueta, está mal
                const labels = ["Número Reparación", "Dirección", "Nombre Cliente", "Urgente", "Estado"];
                Object.keys(data).forEach(k => {
                    if (data[k]) {
                        // Cortamos si se nos ha colado la siguiente etiqueta
                        labels.forEach(l => {
                            if (data[k].includes(l)) data[k] = data[k].split(l)[0].trim();
                        });
                        // Quitamos saltos de línea molestos
                        data[k] = data[k].replace(/\n/g, ' ').replace(/\s+/g, ' ');
                    }
                });

                return data;
            });

            if (fullData && Object.keys(fullData).length > 2) {
                await saveServiceToDB(ownerId, 'multiasistencia', ref, fullData);
            }
        }
    } catch (e) { 
        console.error("❌ [Multi] Error Crítico:", e.message); 
    } finally { 
        await browser.close(); 
    }
}

// ... (Mantén el resto del archivo igual)
// ==========================================
// 🧹 HOMESERVE (MANTENIDO)
// ==========================================
async function runHomeserve(ownerId, user, pass) {
    const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] });
    const page = await browser.newPage();
    try {
        console.log("🌍 [HomeServe] Entrando...");
        await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=PROF_PASS', { timeout: 60000 });
        if (await page.isVisible('input[name="CODIGO"]')) {
            await page.fill('input[name="CODIGO"]', user);
            await page.fill('input[type="password"]', pass);
            await page.keyboard.press('Enter');
            await page.waitForTimeout(5000);
        }
        await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=lista_servicios_total');
        const refs = await page.evaluate(() => {
            const filas = Array.from(document.querySelectorAll('table tr'));
            const found = [];
            filas.forEach(tr => {
                const txt = tr.innerText; const match = txt.match(/(\d{6,10})/);
                if (match) found.push(match[1]);
            });
            return [...new Set(found)];
        });
        console.log(`🔍 [HomeServe] ${refs.length} expedientes.`);
        for (const ref of refs) {
            try {
                await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=lista_servicios_total');
                const link = await page.getByText(ref).first();
                if (await link.isVisible()) {
                    await link.click(); await page.waitForTimeout(1500);
                    const fullData = await page.evaluate(() => {
                        const d = {};
                        const rows = Array.from(document.querySelectorAll('tr'));
                        rows.forEach(r => {
                            const cells = r.querySelectorAll('td');
                            if(cells.length >= 2) {
                                const k = cells[0].innerText.toUpperCase().trim().replace(':', '');
                                const v = cells[1].innerText.trim();
                                if(k.length > 1 && v.length > 0) d[k] = v;
                            }
                            if(cells.length >= 4) {
                                const k2 = cells[2].innerText.toUpperCase().trim().replace(':', '');
                                const v2 = cells[3].innerText.trim();
                                if(k2.length > 1 && v2.length > 0) d[k2] = v2;
                            }
                        });
                        return d;
                    });
                    if (fullData) await saveServiceToDB(ownerId, 'homeserve', ref, fullData);
                }
            } catch (errRef) { console.error(`⚠️ Error ref ${ref}`); }
        }
    } catch (e) { console.error("❌ [HomeServe]", e.message); } finally { await browser.close(); }
}

async function saveServiceToDB(ownerId, provider, ref, data) {
    console.log(`💾 Guardando ${ref}...`);
    await pool.query(`
        INSERT INTO scraped_services (owner_id, provider, service_ref, raw_data, status)
        VALUES ($1, $2, $3, $4, 'pending')
        ON CONFLICT (owner_id, provider, service_ref) 
        DO UPDATE SET raw_data = EXCLUDED.raw_data, created_at = NOW()
    `, [ownerId, provider, ref, JSON.stringify(data)]);
}

main();