201 lines
9.9 KiB
JavaScript
201 lines
9.9 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 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 = $1", [cred.id]);
|
|
}
|
|
} catch (e) {
|
|
console.error("❌ Error ciclo:", e.message);
|
|
} finally {
|
|
client.release();
|
|
}
|
|
console.log("\n💤 Durmiendo 15 minutos...");
|
|
await new Promise(r => setTimeout(r, 15 * 60 * 1000));
|
|
}
|
|
}
|
|
|
|
async function runMultiasistencia(ownerId, user, pass) {
|
|
const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] });
|
|
const context = await browser.newContext();
|
|
const page = await context.newPage();
|
|
|
|
try {
|
|
await page.goto('https://web.multiasistencia.com/w3multi/acceso.php', { timeout: 60000 });
|
|
await page.fill('input[name="usuario"]', user);
|
|
await page.fill('input[type="password"]', pass);
|
|
await page.click('input[type="submit"]');
|
|
await page.waitForTimeout(4000);
|
|
|
|
await page.goto('https://web.multiasistencia.com/w3multi/frepasos_new.php?refresh=1');
|
|
await page.waitForTimeout(2000);
|
|
|
|
const expedientes = await page.evaluate(() => {
|
|
const links = Array.from(document.querySelectorAll('a[href*="reparacion="]'));
|
|
return Array.from(new Set(links.map(a => a.href.match(/reparacion=(\d+)/)?.[1]).filter(Boolean)));
|
|
});
|
|
|
|
for (const ref of expedientes) {
|
|
await page.goto(`https://web.multiasistencia.com/w3multi/repasos1.php?reparacion=${ref}`, { waitUntil: 'domcontentloaded' });
|
|
await page.waitForTimeout(1500);
|
|
|
|
let scrapData = null;
|
|
|
|
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 allCells = Array.from(document.querySelectorAll('td, th, b, div'));
|
|
const getH = (keys) => {
|
|
const header = allCells.find(el => keys.some(k => el.innerText.toUpperCase().includes(k.toUpperCase())));
|
|
return header && header.nextElementSibling ? clean(header.nextElementSibling.innerText) : null;
|
|
};
|
|
const getV = (keys) => {
|
|
const header = allCells.find(el => keys.some(k => el.innerText.trim().toUpperCase() === k.toUpperCase()));
|
|
if (!header) return null;
|
|
const idx = header.cellIndex;
|
|
const row = header.parentElement;
|
|
let nextRow = row.nextElementSibling;
|
|
if (!nextRow) {
|
|
const table = header.closest('table');
|
|
const tbody = table ? table.querySelector('tbody') : null;
|
|
nextRow = tbody ? tbody.rows[0] : null;
|
|
}
|
|
return nextRow && nextRow.cells[idx] ? clean(nextRow.cells[idx].innerText) : null;
|
|
};
|
|
|
|
// --- LÓGICA DE LIMPIEZA EN ORIGEN ---
|
|
|
|
// 1. DESCRIPCIÓN TÉCNICA (Corte por fecha)
|
|
let rawDesc = getH(["Descripción de la Reparación", "Descripción", "Daños"]) || "";
|
|
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 = getV(["Distrito Postal", "C.P", "Distrito"]) || "";
|
|
const cpMatch = rawCPField.match(/\b\d{5}\b/);
|
|
const cpOnly = cpMatch ? cpMatch[0] : "";
|
|
// Extraemos la población eliminando el CP y el guion
|
|
const popOnly = rawCPField.replace(cpOnly, '').replace('-', '').trim();
|
|
|
|
// 3. TELÉFONO LIMPIO (Solo 9 dígitos)
|
|
const phoneMatch = body.match(/[6789]\d{8}/);
|
|
|
|
return {
|
|
"Expediente": getH(["Número Reparación", "Número de Servicio"]) || "",
|
|
"Nombre Cliente": getV(["Nombre Cliente", "Asegurado"]) || "",
|
|
"Dirección": getV(["Dirección", "Domicilio"]) || "",
|
|
"Población": popOnly, // Nueva clave limpia
|
|
"Código Postal": cpOnly, // Nueva clave limpia
|
|
"Compañía": getH(["Procedencia", "Compañía"]) || "MULTI",
|
|
"Descripción": cleanDesc,
|
|
"Teléfono": phoneMatch ? phoneMatch[0] : "",
|
|
"Urgente": getH(["Urgente"]) || "No"
|
|
};
|
|
});
|
|
if (scrapData && scrapData['Nombre Cliente']) break;
|
|
} catch (e) { continue; }
|
|
}
|
|
|
|
if (scrapData && scrapData['Nombre Cliente']) {
|
|
await saveServiceToDB(ownerId, 'multiasistencia', ref, scrapData);
|
|
}
|
|
}
|
|
} catch (e) { console.error("❌ Error Multi:", e.message); }
|
|
finally { await browser.close(); }
|
|
}
|
|
|
|
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(); |