468 lines
22 KiB
JavaScript
468 lines
22 KiB
JavaScript
import { chromium } from 'playwright';
|
|
import pg from 'pg';
|
|
|
|
const { DATABASE_URL } = process.env;
|
|
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;
|
|
|
|
// ==========================================
|
|
// 🧠 MOTOR DE CLASIFICACIÓN DE GREMIOS PRO (DINÁMICO)
|
|
// ==========================================
|
|
|
|
function normalizarTexto(texto) {
|
|
if (!texto) return "";
|
|
return texto.toLowerCase().normalize("NFD").replace(/[\u0300-\u036f]/g, "");
|
|
}
|
|
|
|
// 🛠️ El "Gira-Nombres" EXCLUSIVO para HomeServe
|
|
function arreglarNombre(nombreRaw) {
|
|
if (!nombreRaw) return "";
|
|
let nombreLimpio = nombreRaw.trim();
|
|
if (nombreLimpio.includes(',')) {
|
|
let partes = nombreLimpio.split(',');
|
|
let apellidos = partes[0].trim();
|
|
let nombre = partes[1].trim();
|
|
nombreLimpio = `${nombre} ${apellidos}`.replace(/\s+/g, ' ');
|
|
}
|
|
return nombreLimpio;
|
|
}
|
|
|
|
// 🛡️ DESCODIFICADOR UNIVERSAL PARA HOMESERVE (Arregla los +, Ó, Ñ, etc.)
|
|
function arreglarCodificacion(texto) {
|
|
if (!texto) return "";
|
|
let limpio = texto.replace(/\+/g, ' ');
|
|
try {
|
|
limpio = decodeURIComponent(escape(limpio));
|
|
} catch(e) {}
|
|
return limpio.trim();
|
|
}
|
|
|
|
function clasificarGremio(descripcion, gremiosActivosDB) {
|
|
if (!descripcion || gremiosActivosDB.length === 0) return null;
|
|
|
|
const descNormalizada = normalizarTexto(descripcion);
|
|
|
|
for (const gremio of gremiosActivosDB) {
|
|
const keywords = Array.isArray(gremio.ia_keywords) ? gremio.ia_keywords : [];
|
|
if (keywords.length === 0) continue;
|
|
|
|
const coincide = keywords.some(kw => descNormalizada.includes(normalizarTexto(kw)));
|
|
if (coincide) {
|
|
console.log(` 🧠 Gremio detectado automáticamente: ${gremio.name} (ID: ${gremio.id})`);
|
|
return gremio.id;
|
|
}
|
|
}
|
|
return null;
|
|
}
|
|
|
|
async function gotoWithRetry(page, url, retries = 3) {
|
|
for (let i = 0; i < retries; i++) {
|
|
try {
|
|
await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 45000 });
|
|
return;
|
|
} catch (e) {
|
|
if (i === retries - 1) throw e;
|
|
console.log(`⚠️ Fallo de red detectado al ir a ${url}. Reintentando (${i + 1}/${retries})...`);
|
|
await page.waitForTimeout(3000);
|
|
}
|
|
}
|
|
}
|
|
|
|
async function main() {
|
|
console.log("🤖 ROBOT MODO: CIRUJANO + ASPIRADORA + CLASIFICADOR PRO");
|
|
while (true) {
|
|
const client = await pool.connect();
|
|
try {
|
|
const gremiosRes = await client.query("SELECT id, name, ia_keywords FROM guilds");
|
|
const gremiosDB = gremiosRes.rows;
|
|
|
|
await client.query(`
|
|
DO $$ BEGIN
|
|
IF NOT EXISTS (SELECT 1 FROM information_schema.columns WHERE table_name='scraped_services' AND column_name='is_urgent') THEN
|
|
ALTER TABLE scraped_services ADD COLUMN is_urgent BOOLEAN DEFAULT FALSE;
|
|
END IF;
|
|
IF NOT EXISTS (SELECT 1 FROM information_schema.columns WHERE table_name='guilds' AND column_name='ia_keywords') THEN
|
|
ALTER TABLE guilds ADD COLUMN ia_keywords JSONB DEFAULT '[]';
|
|
END IF;
|
|
END $$;
|
|
`);
|
|
|
|
const res = await client.query("SELECT * FROM provider_credentials WHERE status = 'active'");
|
|
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, gremiosDB);
|
|
} else if (cred.provider === 'homeserve') {
|
|
await runHomeserve(cred.owner_id, cred.username, password, gremiosDB);
|
|
}
|
|
|
|
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 (PAGINACIÓN ANTIBALAS Y DOBLE RECARGA)
|
|
// ==========================================
|
|
// --- BUCLE DE PAGINACIÓN INFALIBLE ---
|
|
let todosExpedientes = new Set();
|
|
let paginaActual = 1;
|
|
|
|
console.log(`📄 [Multi] Iniciando lectura por páginas...`);
|
|
while (true) {
|
|
console.log(`\n📄 [Multi] Escaneando página ${paginaActual}...`);
|
|
|
|
// 🛑 NUEVO: Esperamos a que los enlaces existan o damos 5 seg. de margen extra
|
|
await page.waitForSelector('a[href*="reparacion="]', { timeout: 6000 }).catch(() => {});
|
|
await page.waitForTimeout(1000);
|
|
|
|
// 1. Extraemos los enlaces de esta página
|
|
const expedientesPagina = await page.evaluate(() => {
|
|
const links = Array.from(document.querySelectorAll('a[href*="reparacion="]'));
|
|
return links.map(a => a.href.match(/reparacion=(\d+)/)?.[1]).filter(Boolean);
|
|
});
|
|
|
|
const unicosPagina = [...new Set(expedientesPagina)];
|
|
console.log(` 👉 Encontrados ${unicosPagina.length} expedientes únicos en la pág. ${paginaActual}.`);
|
|
|
|
if (unicosPagina.length === 0) {
|
|
console.log(" ⚠️ Página en blanco o cargando demasiado lento. Terminamos aquí.");
|
|
break;
|
|
}
|
|
|
|
unicosPagina.forEach(ref => todosExpedientes.add(ref));
|
|
|
|
// 2. BUSCAMOS Y PULSAMOS EL BOTÓN "SIGUIENTE"
|
|
const clicked = await page.evaluate(() => {
|
|
const links = Array.from(document.querySelectorAll('a.lnkheader'));
|
|
const btn = links.find(a => a.innerText.toLowerCase().includes('siguiente'));
|
|
if (btn) {
|
|
btn.click();
|
|
return true;
|
|
}
|
|
return false;
|
|
});
|
|
|
|
if (clicked) {
|
|
console.log(` ➡️ Botón 'Siguiente' detectado. Pulsando y esperando a que cargue...`);
|
|
// 🛑 EL FRENO: Congelamos el robot 5 segundos fijos para que el servidor responda
|
|
await page.waitForTimeout(5000);
|
|
paginaActual++;
|
|
|
|
if(paginaActual > 20) {
|
|
console.log("⚠️ [Multi] Límite de 20 páginas alcanzado por seguridad.");
|
|
break;
|
|
}
|
|
} else {
|
|
console.log("🛑 [Multi] Fin de la lista. No hay enlace a página siguiente.");
|
|
break;
|
|
}
|
|
}
|
|
|
|
const expedientesUnicos = Array.from(todosExpedientes);
|
|
console.log(`\n✅ [Multi] TOTAL EXPEDIENTES LEÍDOS EN WEB: ${expedientesUnicos.length}`);
|
|
|
|
// SEGURO DE VIDA PARA MULTIASISTENCIA
|
|
if (expedientesUnicos.length > 5 || paginaActual === 1) {
|
|
console.log(`🧹 [Multi] Llamando a la escoba con ${expedientesUnicos.length} refs...`);
|
|
await syncAndArchive(ownerId, 'multiasistencia', expedientesUnicos);
|
|
} else {
|
|
console.log("⚠️ [Multi] Menos de 5 expedientes detectados. ABORTANDO ARCHIVADO DE SEGURIDAD.");
|
|
}
|
|
|
|
// BUCLE DE GUARDADO
|
|
for (const ref of expedientesUnicos) {
|
|
await gotoWithRetry(page, `https://web.multiasistencia.com/w3multi/repasos1.php?reparacion=${ref}`);
|
|
await page.waitForTimeout(1500);
|
|
let scrapData = null;
|
|
|
|
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;
|
|
|
|
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;
|
|
};
|
|
|
|
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);
|
|
|
|
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();
|
|
|
|
let telefonoReal = "";
|
|
const titulosDiv = Array.from(document.querySelectorAll('.subtitulo'));
|
|
const divTelefono = titulosDiv.find(div => div.innerText && div.innerText.includes('Teléfono'));
|
|
if (divTelefono && divTelefono.nextElementSibling) {
|
|
const celdaNum = divTelefono.nextElementSibling.querySelector('.tdet');
|
|
if (celdaNum) telefonoReal = celdaNum.innerText.replace(/[^0-9]/g, '');
|
|
}
|
|
if (!telefonoReal || telefonoReal.length < 9) {
|
|
telefonoReal = (document.body.innerText.match(/[6789]\d{8}/) || [])[0] || "";
|
|
}
|
|
|
|
return {
|
|
"Expediente": findByCab("Número Reparación") || "",
|
|
"Nombre Cliente": findByCab("Nombre Cliente") || "",
|
|
"Dirección": findByCab("Dirección") || "",
|
|
"Población": popOnly,
|
|
"Código Postal": cpOnly,
|
|
"Compañía": "MULTI - " + (findByCab("Procedencia") || "MULTIASISTENCIA"),
|
|
"Descripción": rawDesc,
|
|
"Teléfono": telefonoReal,
|
|
"Estado": findByCab("Estado") || "",
|
|
"Urgente": findByCab("Urgente") || "No"
|
|
};
|
|
});
|
|
if (scrapData && scrapData['Nombre Cliente']) break;
|
|
} catch (e) { continue; }
|
|
}
|
|
|
|
if (scrapData && scrapData['Nombre Cliente']) {
|
|
const idGremioDetectado = clasificarGremio(scrapData['Descripción'], gremiosDB);
|
|
if (idGremioDetectado) {
|
|
scrapData['guild_id'] = idGremioDetectado;
|
|
}
|
|
await saveServiceToDB(ownerId, 'multiasistencia', ref, scrapData);
|
|
}
|
|
} catch (e) {
|
|
console.error("❌ Error Multi:", e.message);
|
|
} finally {
|
|
await browser.close();
|
|
}
|
|
}
|
|
|
|
// ==========================================
|
|
// 🧹 HOMESERVE (MEJORADO CON DECODIFICADOR, ICONOS Y URGENCIA)
|
|
// ==========================================
|
|
async function runHomeserve(ownerId, user, pass, gremiosDB) {
|
|
const browser = await chromium.launch({
|
|
headless: HEADLESS,
|
|
args: ['--no-sandbox', '--disable-setuid-sandbox', '--disable-dev-shm-usage', '--disable-gpu']
|
|
});
|
|
const page = await browser.newPage();
|
|
try {
|
|
console.log("🌍 [HomeServe] Entrando...");
|
|
await gotoWithRetry(page, 'https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=PROF_PASS');
|
|
|
|
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 gotoWithRetry(page, 'https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=lista_servicios_total');
|
|
await page.waitForTimeout(3000);
|
|
|
|
// 🛑 EXTRAEMOS REFS E ICONOS
|
|
const listaConIconos = await page.evaluate(() => {
|
|
const results = [];
|
|
const rows = Array.from(document.querySelectorAll('table[bgcolor="#FCF4D6"] tr'));
|
|
rows.forEach(tr => {
|
|
const firstTd = tr.querySelector('td');
|
|
if (!firstTd) return;
|
|
const textMatch = firstTd.innerText.trim().match(/^15\d{6}$/);
|
|
const aMatch = firstTd.querySelector('a') ? firstTd.querySelector('a').innerText.trim().match(/^15\d{6}$/) : null;
|
|
const ref = textMatch ? textMatch[0] : (aMatch ? aMatch[0] : null);
|
|
if (ref) {
|
|
const hasLock = firstTd.querySelector('img[src*="candado.gif"]') !== null;
|
|
const hasEyes = firstTd.querySelector('img[src*="ojos.gif"]') !== null;
|
|
results.push({ ref, hasLock, hasEyes });
|
|
}
|
|
});
|
|
return results;
|
|
});
|
|
|
|
const refs = [...new Set(listaConIconos.map(item => item.ref))];
|
|
|
|
if (refs.length > 0) {
|
|
await syncAndArchive(ownerId, 'homeserve', refs);
|
|
}
|
|
|
|
console.log(`🔍 [HomeServe] ${refs.length} expedientes detectados.`);
|
|
|
|
for (const ref of refs) {
|
|
await gotoWithRetry(page, `https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=ver_servicioencurso&Servicio=${ref}`);
|
|
await page.waitForTimeout(2000);
|
|
|
|
const scrapData = await page.evaluate(() => {
|
|
const d = {};
|
|
let isUrgent = "No";
|
|
|
|
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 inputEl = cells[1].querySelector('textarea');
|
|
const txt = inputEl ? inputEl.value : (cells[1].innerText || "");
|
|
|
|
// 🔥 DETECTOR DE URGENCIA 🔥
|
|
const txtLower = txt.toLowerCase().normalize("NFD").replace(/[\u0300-\u036f]/g, "");
|
|
if (txtLower.includes("atencion presencial urgencias") || txtLower.includes("atencion de la urgencia") || txtLower.includes("urgente")) {
|
|
isUrgent = "Sí";
|
|
}
|
|
|
|
const cleanDesc = txt.split('\n').filter(line => {
|
|
const l = line.toUpperCase();
|
|
return !["ESTADO ASIGNADO", "SMS NO ENVIADO", "CONTACTO CON PROF", "0000"].some(b => l.includes(b));
|
|
}).join('\n').trim();
|
|
d['Descripción'] = cleanDesc;
|
|
} else if (k.length > 1 && v.length > 0 && !k.includes("MENU")) {
|
|
d[k] = v;
|
|
}
|
|
}
|
|
});
|
|
|
|
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'] || "";
|
|
d['Urgente'] = isUrgent;
|
|
|
|
return d;
|
|
});
|
|
|
|
if (scrapData && scrapData['Nombre Cliente']) {
|
|
// 🪄 Arreglamos codificación y nombres de HomeServe
|
|
scrapData['Nombre Cliente'] = arreglarNombre(arreglarCodificacion(scrapData['Nombre Cliente']));
|
|
scrapData['Dirección'] = arreglarCodificacion(scrapData['Dirección']);
|
|
scrapData['Población'] = arreglarCodificacion(scrapData['Población']);
|
|
scrapData['Descripción'] = arreglarCodificacion(scrapData['Descripción']);
|
|
|
|
// 🛑 INYECTAMOS ICONOS CAZADOS
|
|
const iconInfo = listaConIconos.find(item => item.ref === ref);
|
|
if (iconInfo) {
|
|
scrapData['has_lock'] = iconInfo.hasLock;
|
|
scrapData['has_eyes'] = iconInfo.hasEyes;
|
|
}
|
|
|
|
const idGremioDetectado = clasificarGremio(scrapData['Descripción'], gremiosDB);
|
|
if (idGremioDetectado) {
|
|
scrapData['guild_id'] = idGremioDetectado;
|
|
}
|
|
await saveServiceToDB(ownerId, 'homeserve', ref, scrapData);
|
|
}
|
|
}
|
|
} catch (e) { console.error("❌ [HomeServe] Error:", e.message); }
|
|
finally { await browser.close(); }
|
|
}
|
|
|
|
async function syncAndArchive(ownerId, provider, currentWebRefs) {
|
|
const client = await pool.connect();
|
|
try {
|
|
const { rows: dbServices } = await client.query(
|
|
`SELECT id, service_ref, assigned_to, raw_data FROM scraped_services
|
|
WHERE owner_id = $1 AND provider = $2 AND status IN ('pending', 'imported')`,
|
|
[ownerId, provider]
|
|
);
|
|
|
|
const missingServices = dbServices.filter(s => !currentWebRefs.includes(s.service_ref));
|
|
const refsToArchive = missingServices.map(s => s.service_ref);
|
|
|
|
console.log(`\n📊 [ESCOBA ${provider.toUpperCase()}] Refs en Web: ${currentWebRefs.length} | Pendientes en BD: ${dbServices.length}`);
|
|
|
|
if (refsToArchive.length > 0) {
|
|
console.log(`🚨 ATENCIÓN: Se van a ARCHIVAR ${refsToArchive.length} expedientes porque NO están en la web:`);
|
|
console.log(`💀 LISTA:`, refsToArchive.join(', '));
|
|
|
|
await client.query(
|
|
`UPDATE scraped_services
|
|
SET status = 'archived'
|
|
WHERE owner_id = $1
|
|
AND provider = $2
|
|
AND service_ref = ANY($3)`,
|
|
[ownerId, provider, refsToArchive]
|
|
);
|
|
|
|
// ========================================================
|
|
// 🛡️ EL ESCUDO ANTI-VIAJES EN BALDE (Apagado para Multiasistencia)
|
|
// ========================================================
|
|
const statusQ = await client.query("SELECT id FROM service_statuses WHERE owner_id = $1 AND name ILIKE '%anulado%' LIMIT 1", [ownerId]);
|
|
const idAnulado = statusQ.rowCount > 0 ? statusQ.rows[0].id : null;
|
|
|
|
if (idAnulado && provider !== 'multiasistencia') {
|
|
for (const svc of missingServices) {
|
|
if (svc.assigned_to) {
|
|
const raw = svc.raw_data || {};
|
|
const currentStatusId = String(raw.status_operativo || "");
|
|
|
|
if (currentStatusId !== String(idAnulado)) {
|
|
let isFinal = false;
|
|
if (currentStatusId) {
|
|
const checkStatusQ = await client.query("SELECT is_final FROM service_statuses WHERE id = $1", [currentStatusId]);
|
|
isFinal = checkStatusQ.rowCount > 0 ? checkStatusQ.rows[0].is_final : false;
|
|
}
|
|
|
|
if (!isFinal) {
|
|
raw.status_operativo = idAnulado;
|
|
await client.query("UPDATE scraped_services SET raw_data = $1 WHERE id = $2", [JSON.stringify(raw), svc.id]);
|
|
|
|
await client.query(
|
|
"INSERT INTO scraped_service_logs (scraped_id, user_name, action, details) VALUES ($1, $2, $3, $4)",
|
|
[svc.id, "Sistema Robot", "Cancelación Automática", "La compañía ha retirado el expediente. Se pasa a Anulado."]
|
|
);
|
|
console.log(`🛡️ [ESCUDO] Expediente ${svc.service_ref} anulado automáticamente (Técnico salvado).`);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
console.log(`📦 [${provider.toUpperCase()}] Archivados ${refsToArchive.length} expedientes desaparecidos.`);
|
|
}
|
|
} catch (error) {
|
|
console.error(`❌ Error archivando ${provider}:`, error.message);
|
|
} finally {
|
|
client.release();
|
|
}
|
|
}
|
|
|
|
async function saveServiceToDB(ownerId, provider, ref, data) {
|
|
console.log(`💾 Guardando/Actualizando ${provider.toUpperCase()} ${ref}...`);
|
|
|
|
const isUrgent = (data['Urgente'] && (data['Urgente'].toLowerCase().trim() === 'sí' || data['Urgente'].toLowerCase().trim() === 'si')) ? true : false;
|
|
|
|
await pool.query(`
|
|
INSERT INTO scraped_services (owner_id, provider, service_ref, raw_data, status, is_urgent)
|
|
VALUES ($1, $2, $3, $4, 'pending', $5)
|
|
ON CONFLICT (owner_id, provider, service_ref)
|
|
DO UPDATE SET
|
|
raw_data = (scraped_services.raw_data || EXCLUDED.raw_data) ||
|
|
jsonb_build_object('guild_id', COALESCE(scraped_services.raw_data->'guild_id', EXCLUDED.raw_data->'guild_id')),
|
|
|
|
is_urgent = EXCLUDED.is_urgent,
|
|
status = CASE
|
|
WHEN scraped_services.status = 'archived' THEN 'archived'
|
|
WHEN scraped_services.status = 'imported' THEN 'imported'
|
|
ELSE 'pending'
|
|
END
|
|
`, [ownerId, provider, ref, JSON.stringify(data), isUrgent]);
|
|
}
|
|
|
|
main(); |