marsalva/robot-proveedores
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Actualizar robot.js

Browse Source
This commit is contained in:
marsalva
2026-02-13 09:31:36 +00:00
parent 5b3f36b888
commit 92b081befa
1 changed files with 85 additions and 124 deletions
Download Patch File Download Diff File Expand all files Collapse all files

197
robot.js
View File

@@ -1,7 +1,6 @@
import { chromium } from 'playwright'; import { chromium } from 'playwright';
import pg from 'pg'; import pg from 'pg';
// En Coolify las variables se inyectan solas
const { DATABASE_URL } = process.env; const { DATABASE_URL } = process.env;
if (!DATABASE_URL) { if (!DATABASE_URL) {
@@ -10,57 +9,41 @@ if (!DATABASE_URL) {
} }
const pool = new pg.Pool({ connectionString: DATABASE_URL, ssl: false }); const pool = new pg.Pool({ connectionString: DATABASE_URL, ssl: false });
const HEADLESS = true;
// CONFIGURACIÓN
const HEADLESS = true; // En servidor SIEMPRE true
async function main() { async function main() {
console.log("🤖 INICIANDO ROBOT UNIFICADO (MEJORADO)..."); console.log("🤖 ROBOT MODO: ASPIRADORA INDUSTRIAL (CAPTURA TODO)");
while (true) { while (true) {
const client = await pool.connect(); const client = await pool.connect();
try { try {
// 1. Obtener credenciales activas
const res = await client.query("SELECT * FROM provider_credentials WHERE status = 'active'"); const res = await client.query("SELECT * FROM provider_credentials WHERE status = 'active'");
const credentials = res.rows; const credentials = res.rows;
console.log(`📋 Cuentas activas: ${credentials.length}`);
console.log(`📋 Procesando ${credentials.length} cuentas de proveedores...`);
for (const cred of credentials) { for (const cred of credentials) {
let password = ""; let password = Buffer.from(cred.password_hash, 'base64').toString('utf-8');
try { console.log(`\n🔄 Escaneando ${cred.provider.toUpperCase()} (ID: ${cred.owner_id})...`);
password = Buffer.from(cred.password_hash, 'base64').toString('utf-8');
} catch (e) {
console.error(`Error decodificando pass para usuario ${cred.owner_id}`);
continue;
}
console.log(`\n🔄 Sincronizando ${cred.provider.toUpperCase()} para usuario ID ${cred.owner_id}...`);
if (cred.provider === 'multiasistencia') { if (cred.provider === 'multiasistencia') {
await runMultiasistencia(cred.owner_id, cred.username, password); await runMultiasistencia(cred.owner_id, cred.username, password);
} else if (cred.provider === 'homeserve') { } else if (cred.provider === 'homeserve') {
await runHomeserve(cred.owner_id, cred.username, password); await runHomeserve(cred.owner_id, cred.username, password);
} }
// Actualizar timestamp
await client.query("UPDATE provider_credentials SET last_sync = NOW() WHERE id = $1", [cred.id]); await client.query("UPDATE provider_credentials SET last_sync = NOW() WHERE id = $1", [cred.id]);
} }
} catch (e) { } catch (e) {
console.error("❌ Error en ciclo principal:", e); console.error("❌ Error ciclo:", e.message);
} finally { } finally {
client.release(); client.release();
} }
// ESPERA DE 15 MINUTOS
console.log("\n💤 Durmiendo 15 minutos..."); console.log("\n💤 Durmiendo 15 minutos...");
await new Promise(r => setTimeout(r, 15 * 60 * 1000)); await new Promise(r => setTimeout(r, 15 * 60 * 1000));
} }
} }
// ========================================== // ==========================================
// 🛠️ ROBOT MULTIASISTENCIA (CORREGIDO) // 🧹 MULTIASISTENCIA (EXTRACCIÓN TOTAL)
// ========================================== // ==========================================
async function runMultiasistencia(ownerId, user, pass) { async function runMultiasistencia(ownerId, user, pass) {
const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] }); const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] });
@@ -68,10 +51,9 @@ async function runMultiasistencia(ownerId, user, pass) {
const page = await context.newPage(); const page = await context.newPage();
try { try {
console.log("🌍 [Multi] Conectando..."); console.log("🌍 [Multi] Entrando...");
await page.goto('https://web.multiasistencia.com/w3multi/acceso.php', { timeout: 60000 }); await page.goto('https://web.multiasistencia.com/w3multi/acceso.php', { timeout: 60000 });
// Login
const userInput = await page.$('input[name="usuario"]') || await page.$('input[type="text"]'); const userInput = await page.$('input[name="usuario"]') || await page.$('input[type="text"]');
if(userInput) { if(userInput) {
await userInput.fill(user); await userInput.fill(user);
@@ -80,93 +62,67 @@ async function runMultiasistencia(ownerId, user, pass) {
await page.waitForTimeout(4000); await page.waitForTimeout(4000);
} }
// Ir a listado
await page.goto('https://web.multiasistencia.com/w3multi/frepasos_new.php?refresh=1', { waitUntil: 'domcontentloaded' }); await page.goto('https://web.multiasistencia.com/w3multi/frepasos_new.php?refresh=1', { waitUntil: 'domcontentloaded' });
// Sacar expedientes
const expedientes = await page.evaluate(() => { const expedientes = await page.evaluate(() => {
const links = Array.from(document.querySelectorAll('a[href*="reparacion="]')); 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))); return Array.from(new Set(links.map(a => a.href.match(/reparacion=(\d+)/)?.[1]).filter(Boolean)));
}); });
console.log(`🔍 [Multi] ${expedientes.length} expedientes encontrados.`); console.log(`🔍 [Multi] ${expedientes.length} expedientes.`);
for (const ref of expedientes) { for (const ref of expedientes) {
await page.goto(`https://web.multiasistencia.com/w3multi/repasos1.php?reparacion=${ref}`, { waitUntil: 'domcontentloaded' }); await page.goto(`https://web.multiasistencia.com/w3multi/repasos1.php?reparacion=${ref}`, { waitUntil: 'domcontentloaded' });
// --- NUEVA LÓGICA DE EXTRACCIÓN POR CELDAS --- // --- LÓGICA AGRESIVA DE CAPTURA ---
const data = await page.evaluate(() => { const fullData = await page.evaluate(() => {
const data = {};
// Función auxiliar para buscar en celdas vecinas // 1. Recorrer TODAS las filas de TODAS las tablas
const findValueByHeader = (keywords) => { const rows = document.querySelectorAll('tr');
const cells = Array.from(document.querySelectorAll('td, th')); rows.forEach(row => {
for (const cell of cells) { const cells = Array.from(row.querySelectorAll('td, th'));
const text = (cell.innerText || "").toUpperCase().trim(); // Intentamos emparejar: Celda 1 (Clave) -> Celda 2 (Valor)
// Si la celda contiene una de las palabras clave for (let i = 0; i < cells.length - 1; i++) {
if (keywords.some(k => text.includes(k.toUpperCase()))) { let key = cells[i].innerText.trim().replace(':', '');
// Devolvemos el texto de la celda SIGUIENTE let val = cells[i+1]?.innerText.trim();
const nextCell = cell.nextElementSibling;
if (nextCell) return nextCell.innerText.trim(); // Filtros de limpieza básicos
if (key.length > 2 && key.length < 60 && val && val.length > 0) {
// Si la clave ya existe, no la machacamos (o le añadimos un indice)
if (!data[key]) {
data[key] = val;
// Saltamos la celda de valor para no leerla como clave en la siguiente vuelta
i++;
} }
} }
return "";
};
// Función auxiliar para descripción (busca bloques de texto grandes)
const findDescription = () => {
// Intento 1: Buscar celda "Descripción"
let desc = findValueByHeader(['Descripción', 'Daños', 'Solicitud']);
if(desc) return desc;
// Intento 2: Buscar en el cuerpo general si falla la tabla
const body = document.body.innerText;
const match = body.match(/Descripción\s*[:\-]?\s*([^\n]+)/i);
return match ? match[1].trim() : "";
} }
// Extracción más precisa
let address = findValueByHeader(['Dirección', 'Domicilio', 'Riesgo']);
let pob = findValueByHeader(['Población', 'Localidad']);
let cp = findValueByHeader(['C.P.', 'Postal']);
// Limpieza de dirección (quitar "Baremo" si se cuela)
if (address.includes("Baremo")) address = address.split("Baremo")[0].trim();
// Unir dirección completa
let fullAddress = address;
if (pob) fullAddress += `, ${pob}`;
if (cp) fullAddress += ` (${cp})`;
return {
clientName: findValueByHeader(['Nombre Cliente', 'Asegurado']) || "Desconocido",
address: fullAddress || "Sin dirección detectada",
phone: (document.body.innerText.match(/[6789]\d{8}/) || [])[0] || "",
description: findDescription()
};
}); });
if (data.clientName) { // Asegurar mínimos (fallback)
// Solo guardamos si tenemos al menos un nombre if (!data.clientName) data.clientName = data["Nombre Cliente"] || data["Asegurado"] || "Desconocido";
await saveServiceToDB(ownerId, 'multiasistencia', ref, data); if (!data.address) data.address = data["Dirección"] || data["Domicilio"] || data["Riesgo"] || "";
} if (!data.phone) data.phone = (document.body.innerText.match(/[6789]\d{8}/) || [])[0] || "";
}
} catch (e) { return data;
console.error("❌ [Multi] Error:", e.message); });
} finally {
await browser.close(); if (fullData && Object.keys(fullData).length > 2) {
await saveServiceToDB(ownerId, 'multiasistencia', ref, fullData);
} }
}
} catch (e) { console.error("❌ [Multi]", e.message); } finally { await browser.close(); }
} }
// ========================================== // ==========================================
// 🛠️ ROBOT HOMESERVE (MANTENIDO) // 🧹 HOMESERVE (EXTRACCIÓN TOTAL)
// ========================================== // ==========================================
async function runHomeserve(ownerId, user, pass) { async function runHomeserve(ownerId, user, pass) {
const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] }); const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] });
const page = await browser.newPage(); const page = await browser.newPage();
try { try {
console.log("🌍 [HomeServe] Conectando..."); console.log("🌍 [HomeServe] Entrando...");
await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=PROF_PASS', { timeout: 60000 }); await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=PROF_PASS', { timeout: 60000 });
if (await page.isVisible('input[name="CODIGO"]')) { if (await page.isVisible('input[name="CODIGO"]')) {
@@ -180,60 +136,65 @@ async function runHomeserve(ownerId, user, pass) {
const refs = await page.evaluate(() => { const refs = await page.evaluate(() => {
const filas = Array.from(document.querySelectorAll('table tr')); const filas = Array.from(document.querySelectorAll('table tr'));
return filas.map(tr => tr.querySelector('td')?.innerText.trim()).filter(t => /^\d{4,}$/.test(t)); 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 encontrados.`); console.log(`🔍 [HomeServe] ${refs.length} expedientes.`);
for (const ref of refs) { for (const ref of refs) {
try { try {
await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=lista_servicios_total'); await page.goto('https://www.clientes.homeserve.es/cgi-bin/fccgi.exe?w3exec=lista_servicios_total');
await page.click(`text="${ref}"`, { timeout: 5000 }); const link = await page.getByText(ref).first();
await page.waitForTimeout(1000); if (await link.isVisible()) {
await link.click();
await page.waitForTimeout(1500);
const data = await page.evaluate(() => { const fullData = await page.evaluate(() => {
const rows = Array.from(document.querySelectorAll('tr'));
const d = {}; const d = {};
// HomeServe suele ser tablas limpias: TR -> TD (Clave) | TD (Valor)
const rows = Array.from(document.querySelectorAll('tr'));
rows.forEach(r => { rows.forEach(r => {
const cells = r.querySelectorAll('td'); const cells = r.querySelectorAll('td');
if(cells.length > 1) { // Si tiene al menos 2 celdas
const k = cells[0].innerText.toUpperCase(); if(cells.length >= 2) {
const v = cells[1].innerText; const k = cells[0].innerText.toUpperCase().trim().replace(':', '');
if(k.includes("CLIENTE")) d.clientName = v; const v = cells[1].innerText.trim();
if(k.includes("DOMICILIO")) d.address = v; if(k.length > 1 && v.length > 0) d[k] = v;
if(k.includes("POBLACION")) d.address += ", " + v; }
if(k.includes("TELEFONOS")) d.phone = v; // A veces son tablas de 4 columnas (Clave | Valor | Clave | Valor)
if(k.includes("COMENTARIOS") || k.includes("AVERIA")) d.description = 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;
} }
}); });
// Mapeo mínimo
d.clientName = d["CLIENTE"] || d["ASEGURADO"] || "Desconocido";
d.phone = d["TELEFONOS"] || d["MOVIL"] || "";
d.address = d["DOMICILIO"] || d["DIRECCION"] || "";
if(d["POBLACION"]) d.address += ", " + d["POBLACION"];
return d; return d;
}); });
if (data.clientName) { if (fullData) {
await saveServiceToDB(ownerId, 'homeserve', ref, data); await saveServiceToDB(ownerId, 'homeserve', ref, fullData);
} }
} catch (e) { console.error(`Error leyendo ${ref}: ${e.message}`); }
} }
} catch (errRef) { console.error(`⚠️ Error ref ${ref}`); }
} catch (e) {
console.error("❌ [HomeServe] Error:", e.message);
} finally {
await browser.close();
} }
} catch (e) { console.error("❌ [HomeServe]", e.message); } finally { await browser.close(); }
} }
// ==========================================
// 💾 GUARDADO EN BASE DE DATOS
// ==========================================
async function saveServiceToDB(ownerId, provider, ref, data) { async function saveServiceToDB(ownerId, provider, ref, data) {
// Verificar si ya existe en servicios reales (para no duplicar trabajo) console.log(`💾 Guardando ${ref} con ${Object.keys(data).length} variables encontradas.`);
const exists = await pool.query("SELECT id FROM services WHERE company_ref = $1 AND owner_id = $2", [ref, ownerId]);
if (exists.rowCount > 0) return;
// Verificar si ya existe en buzón, si existe ACTUALIZAMOS los datos (por si hemos mejorado el scraping)
console.log(`💾 Guardando/Actualizando: ${ref} (${data.address})`);
await pool.query(` await pool.query(`
INSERT INTO scraped_services (owner_id, provider, service_ref, raw_data, status) INSERT INTO scraped_services (owner_id, provider, service_ref, raw_data, status)
VALUES ($1, $2, $3, $4, 'pending') VALUES ($1, $2, $3, $4, 'pending')