diff --git a/robot.js b/robot.js index d9f7451..8660c59 100644 --- a/robot.js +++ b/robot.js @@ -2,28 +2,20 @@ 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); -} +if (!DATABASE_URL) { console.error("❌ Error: No hay DATABASE_URL."); 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) { + for (const cred of res.rows) { 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') { @@ -31,80 +23,58 @@ async function main() { } await client.query("UPDATE provider_credentials SET last_sync = NOW() WHERE id = $1", [cred.id]); } - } catch (e) { - console.error("❌ Error ciclo:", e.message); - } finally { - client.release(); - } + } catch (e) { console.error("❌ Error ciclo:", e.message); } + finally { client.release(); } console.log("\n💤 Durmiendo 15 minutos..."); await new Promise(r => setTimeout(r, 15 * 60 * 1000)); } } +// ========================================== +// 🏥 MULTIASISTENCIA (FUNCIONANDO) +// ========================================== async function runMultiasistencia(ownerId, user, pass) { const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] }); const context = await browser.newContext(); const page = await context.newPage(); - try { await page.goto('https://web.multiasistencia.com/w3multi/acceso.php', { timeout: 60000 }); await page.fill('input[name="usuario"]', user); await page.fill('input[type="password"]', pass); await page.click('input[type="submit"]'); await page.waitForTimeout(4000); - await page.goto('https://web.multiasistencia.com/w3multi/frepasos_new.php?refresh=1'); await page.waitForTimeout(2000); - const expedientes = await page.evaluate(() => { const links = Array.from(document.querySelectorAll('a[href*="reparacion="]')); return Array.from(new Set(links.map(a => a.href.match(/reparacion=(\d+)/)?.[1]).filter(Boolean))); }); - for (const ref of expedientes) { await page.goto(`https://web.multiasistencia.com/w3multi/repasos1.php?reparacion=${ref}`, { waitUntil: 'domcontentloaded' }); await page.waitForTimeout(1500); - let scrapData = null; - - // RECORRIDO DE FRAMES + EXTRACCIÓN POR TABLAS for (const frame of page.frames()) { try { scrapData = await frame.evaluate(() => { const clean = (t) => t ? t.replace(/\s+/g, ' ').trim() : ""; const body = document.body?.innerText || ""; if (!body.includes("Nombre Cliente") && !body.includes("Asegurado")) return null; - - // Seleccionamos todas las celdas de cabecera y detalle const cabeceras = Array.from(document.querySelectorAll('.tcab')); const detalles = Array.from(document.querySelectorAll('.tdet')); - const findByCab = (texto) => { const idx = cabeceras.findIndex(el => el.innerText.includes(texto)); return idx !== -1 && detalles[idx] ? clean(detalles[idx].innerText) : null; }; - - // 1. DESCRIPCIÓN QUIRÚRGICA - // Buscamos específicamente en la tabla de "Más información del servicio" let rawDesc = ""; const descHeader = Array.from(document.querySelectorAll('td.tcab')).find(td => td.innerText.includes("Descripción de la Reparación")); - if (descHeader && descHeader.nextElementSibling) { - rawDesc = clean(descHeader.nextElementSibling.innerText); - } - - // Cortamos la descripción en la primera fecha detectada (historial) + if (descHeader && descHeader.nextElementSibling) rawDesc = clean(descHeader.nextElementSibling.innerText); const idxDate = rawDesc.search(/\b\d{2}\/\d{2}\/\d{4}\b/); const cleanDesc = idxDate !== -1 ? rawDesc.substring(0, idxDate).trim() : rawDesc; - - // 2. SEPARACIÓN DE C.P. Y POBLACIÓN const rawCPField = findByCab("Distrito Postal") || ""; const cpMatch = rawCPField.match(/\b\d{5}\b/); const cpOnly = cpMatch ? cpMatch[0] : ""; const popOnly = rawCPField.replace(cpOnly, '').replace('-', '').trim(); - - // 3. TELÉFONO LIMPIO const phoneMatch = body.match(/[6789]\d{8}/); - return { "Expediente": findByCab("Número Reparación") || "", "Nombre Cliente": findByCab("Nombre Cliente") || "", @@ -120,97 +90,98 @@ async function runMultiasistencia(ownerId, user, pass) { if (scrapData && scrapData['Nombre Cliente']) break; } catch (e) { continue; } } - - if (scrapData && scrapData['Nombre Cliente']) { - await saveServiceToDB(ownerId, 'multiasistencia', ref, scrapData); - } + if (scrapData && scrapData['Nombre Cliente']) await saveServiceToDB(ownerId, 'multiasistencia', ref, scrapData); } - } catch (e) { console.error("❌ Error Multi:", e.message); } - finally { await browser.close(); } + } catch (e) { console.error("❌ Error Multi:", e.message); } finally { await browser.close(); } } +// ========================================== +// 🧹 HOMESERVE (CORREGIDO) +// ========================================== 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 }); + + // Login 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); } + + // Navegar al listado real (Usando la URL de tu log) await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=lista_servicios_total'); + await page.waitForTimeout(3000); + + // EXTRAER LISTA DE EXPEDIENTES const refs = await page.evaluate(() => { - const filas = Array.from(document.querySelectorAll('table tr')); + const cells = Array.from(document.querySelectorAll('td')); const found = []; - filas.forEach(tr => { - const txt = tr.innerText; const match = txt.match(/(\d{6,10})/); - if (match) found.push(match[1]); + cells.forEach(td => { + const match = td.innerText.trim().match(/^15\d{6}$/); // Refs de HomeServe suelen ser 15XXXXXX + if (match) found.push(match[0]); }); return [...new Set(found)]; }); - console.log(`🔍 [HomeServe] ${refs.length} expedientes.`); + + console.log(`🔍 [HomeServe] ${refs.length} expedientes detectados.`); + for (const ref of refs) { - try { - await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=lista_servicios_total'); - const link = await page.getByText(ref).first(); - if (await link.isVisible()) { - await link.click(); await page.waitForTimeout(1500); - // ... dentro de la función runHomeserve, en el evaluate del detalle: + // Ir al detalle directo + await page.goto(`https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=ver_servicioencurso&Servicio=${ref}`); + await page.waitForTimeout(2000); -// ... dentro de la función runHomeserve ... + const scrapData = await page.evaluate(() => { + const d = {}; + const clean = (t) => t ? t.replace(/\s+/g, ' ').trim() : ""; + 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.includes("COMENTARIOS")) { + const txt = cells[1].querySelector('textarea')?.value || ""; + // Filtro quirúrgico de descripción + const cleanDesc = txt.split('\n').filter(line => { + const l = line.toUpperCase(); + return !["CAMBIO DE ESTADO", "ESTADO ASIGNADO", "SMS NO ENVIADO", "CONTACTO CON PROF", "0000"].some(b => l.includes(b)); + }).slice(0, 3).join('\n').trim(); + d['Descripción'] = cleanDesc; + } else if (k.length > 1 && v.length > 0 && !k.includes("MENU")) { + d[k] = v; + } + } + }); -const fullData = await page.evaluate(() => { - const d = {}; - const clean = (t) => t ? t.replace(/\s+/g, ' ').trim() : ""; - - // 1. EXTRAER CAMPOS NORMALES (CLIENTE, DOMICILIO, ETC.) - 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(); - - // 2. CAPTURA ESPECIAL DEL CUADRO DE COMENTARIOS (LA DESCRIPCIÓN) - if (k.includes("COMENTARIOS")) { - const textarea = cells[1].querySelector('textarea'); - if (textarea) { - const lines = textarea.value.split('\n'); - - // Filtro inteligente para quitar basura de sistema - const cleanLines = lines.filter(line => { - const l = line.toUpperCase(); - if (l.includes("CAMBIO DE ESTADO")) return false; - if (l.includes("ESTADO ASIGNADO")) return false; - if (l.includes("SMS NO ENVIADO")) return false; - if (l.includes("CONTACTO CON PROF")) return false; - if (l.includes("tipo de cobro es banco")) return false; - if (l.includes("0000")) return false; - return true; - }); + // Separar Población y CP de HomeServe + const rawPop = d['POBLACION-PROVINCIA'] || ""; + const cpMatch = rawPop.match(/\b\d{5}\b/); + d['Código Postal'] = cpMatch ? cpMatch[0] : ""; + d['Población'] = rawPop.replace(d['Código Postal'], '').replace('-', '').trim(); + d['Compañía'] = "HOME - " + (d['COMPAÑIA'] || "HOMESERVE"); + d['Nombre Cliente'] = d['CLIENTE'] || ""; + d['Dirección'] = d['DOMICILIO'] || ""; - // Guardamos las primeras líneas útiles como "Descripción" - d['Descripción'] = cleanLines.slice(0, 3).join('\n').trim(); - } - } else if (k.length > 1 && v.length > 0 && !k.includes("MENU")) { - d[k] = v; + return d; + }); + + if (scrapData && scrapData['Nombre Cliente']) { + await saveServiceToDB(ownerId, 'homeserve', ref, scrapData); } } - }); - 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(); } + } catch (e) { console.error("❌ [HomeServe] Error:", e.message); } + finally { await browser.close(); } } async function saveServiceToDB(ownerId, provider, ref, data) { - console.log(`💾 Guardando ${ref}...`); + console.log(`💾 Guardando ${provider.toUpperCase()} ${ref}...`); await pool.query(` INSERT INTO scraped_services (owner_id, provider, service_ref, raw_data, status) VALUES ($1, $2, $3, $4, 'pending')