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 09:40:28 +00:00
parent 92b081befa
commit e6e2602f90
1 changed files with 54 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files

105
robot.js
View File

@@ -12,18 +12,17 @@ const pool = new pg.Pool({ connectionString: DATABASE_URL, ssl: false });
const HEADLESS = true; const HEADLESS = true;
async function main() { async function main() {
console.log("🤖 ROBOT MODO: ASPIRADORA INDUSTRIAL (CAPTURA TODO)"); console.log("🤖 ROBOT MODO: CIRUJANO + ASPIRADORA INTELIGENTE");
while (true) { while (true) {
const client = await pool.connect(); const client = await pool.connect();
try { try {
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}`);
for (const cred of credentials) { for (const cred of credentials) {
let password = Buffer.from(cred.password_hash, 'base64').toString('utf-8'); let password = Buffer.from(cred.password_hash, 'base64').toString('utf-8');
console.log(`\n🔄 Escaneando ${cred.provider.toUpperCase()} (ID: ${cred.owner_id})...`); console.log(`\n🔄 Procesando ${cred.provider.toUpperCase()}...`);
if (cred.provider === 'multiasistencia') { if (cred.provider === 'multiasistencia') {
await runMultiasistencia(cred.owner_id, cred.username, password); await runMultiasistencia(cred.owner_id, cred.username, password);
@@ -43,7 +42,7 @@ async function main() {
} }
// ========================================== // ==========================================
// 🧹 MULTIASISTENCIA (EXTRACCIÓN TOTAL) // 🏥 MULTIASISTENCIA V3 (CIRUJANO)
// ========================================== // ==========================================
async function runMultiasistencia(ownerId, user, pass) { async function runMultiasistencia(ownerId, user, pass) {
const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] }); const browser = await chromium.launch({ headless: HEADLESS, args: ['--no-sandbox'] });
@@ -51,7 +50,7 @@ async function runMultiasistencia(ownerId, user, pass) {
const page = await context.newPage(); const page = await context.newPage();
try { try {
console.log("🌍 [Multi] Entrando..."); console.log("🌍 [Multi] Conectando...");
await page.goto('https://web.multiasistencia.com/w3multi/acceso.php', { timeout: 60000 }); await page.goto('https://web.multiasistencia.com/w3multi/acceso.php', { timeout: 60000 });
const userInput = await page.$('input[name="usuario"]') || await page.$('input[type="text"]'); const userInput = await page.$('input[name="usuario"]') || await page.$('input[type="text"]');
@@ -69,45 +68,71 @@ async function runMultiasistencia(ownerId, user, pass) {
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.`); console.log(`🔍 [Multi] ${expedientes.length} expedientes a analizar.`);
for (const ref of expedientes) { for (const ref of expedientes) {
await page.goto(`https://web.multiasistencia.com/w3multi/repasos1.php?reparacion=${ref}`, { waitUntil: 'domcontentloaded' }); await page.goto(`https://web.multiasistencia.com/w3multi/repasos1.php?reparacion=${ref}`, { waitUntil: 'domcontentloaded' });
// --- LÓGICA AGRESIVA DE CAPTURA ---
const fullData = await page.evaluate(() => { const fullData = await page.evaluate(() => {
const data = {}; const data = {};
const bodyText = document.body.innerText;
// 1. Recorrer TODAS las filas de TODAS las tablas // --- 1. CIRUJANO (Extracción manual de lo importante) ---
// Intentamos sacar la dirección de la cabecera típica de Multi
const boldElements = Array.from(document.querySelectorAll('b, strong, .titulo'));
// Buscamos nombre: Suele estar cerca de "Asegurado" o es un texto en mayúsculas aislado
let clientNameCandidate = "";
// Buscamos dirección: Suele contener "CL", "AV", "PZ"
let addressCandidate = "";
// Estrategia por palabras clave vecinas (más seguro)
const findNeighbor = (keywords) => {
const cells = Array.from(document.querySelectorAll('td'));
for (let i = 0; i < cells.length; i++) {
const txt = cells[i].innerText.toUpperCase();
if (keywords.some(k => txt.includes(k))) {
// Devolvemos el texto de la siguiente celda que no esté vacía
for(let j=1; j<=3; j++) {
if(cells[i+j] && cells[i+j].innerText.trim().length > 2) return cells[i+j].innerText.trim();
}
}
}
return "";
};
data.clientName_fixed = findNeighbor(['ASEGURADO', 'NOMBRE CLIENTE', 'CONTACTO']);
data.address_fixed = findNeighbor(['DIRECCIÓN', 'DOMICILIO', 'RIESGO', 'UBICACIÓN']);
data.phone_fixed = (bodyText.match(/[6789]\d{8}/) || [])[0] || "";
data.description_fixed = findNeighbor(['DESCRIPCIÓN', 'DAÑOS', 'AVERÍA']);
// --- 2. ASPIRADORA INTELIGENTE (Tablas) ---
const rows = document.querySelectorAll('tr'); const rows = document.querySelectorAll('tr');
rows.forEach(row => { rows.forEach(row => {
const cells = Array.from(row.querySelectorAll('td, th')); const cells = Array.from(row.querySelectorAll('td'));
// Intentamos emparejar: Celda 1 (Clave) -> Celda 2 (Valor)
for (let i = 0; i < cells.length - 1; i++) { for (let i = 0; i < cells.length - 1; i++) {
let key = cells[i].innerText.trim().replace(':', ''); let key = cells[i].innerText.trim().replace(':', '');
let val = cells[i+1]?.innerText.trim(); let val = cells[i+1]?.innerText.trim();
// Filtros de limpieza básicos // FILTRO DE SEGURIDAD PARA ETIQUETAS
if (key.length > 2 && key.length < 60 && val && val.length > 0) { // 1. La clave no puede ser larguísima (eso es una descripción o dirección)
// Si la clave ya existe, no la machacamos (o le añadimos un indice) if (key.length > 35) continue;
if (!data[key]) { // 2. La clave no puede contener dígitos (ej: "28001" no es una clave)
data[key] = val; if (/\d/.test(key) && key.length > 10) continue;
// Saltamos la celda de valor para no leerla como clave en la siguiente vuelta // 3. La clave no debe empezar por tipo de vía
i++; if (/^(CL|AV|PZ|UR|CJ)\s/.test(key.toUpperCase())) continue;
}
if (key.length > 2 && val && val.length > 0) {
// Guardamos normalizando la clave
data[key] = val;
} }
} }
}); });
// Asegurar mínimos (fallback)
if (!data.clientName) data.clientName = data["Nombre Cliente"] || data["Asegurado"] || "Desconocido";
if (!data.address) data.address = data["Dirección"] || data["Domicilio"] || data["Riesgo"] || "";
if (!data.phone) data.phone = (document.body.innerText.match(/[6789]\d{8}/) || [])[0] || "";
return data; return data;
}); });
if (fullData && Object.keys(fullData).length > 2) { if (fullData) {
await saveServiceToDB(ownerId, 'multiasistencia', ref, fullData); await saveServiceToDB(ownerId, 'multiasistencia', ref, fullData);
} }
} }
@@ -115,78 +140,56 @@ async function runMultiasistencia(ownerId, user, pass) {
} }
// ========================================== // ==========================================
// 🧹 HOMESERVE (EXTRACCIÓN TOTAL) // 🧹 HOMESERVE (SE MANTIENE IGUAL, FUNCIONABA BIEN)
// ========================================== // ==========================================
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] Entrando..."); 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"]')) {
await page.fill('input[name="CODIGO"]', user); await page.fill('input[name="CODIGO"]', user);
await page.fill('input[type="password"]', pass); await page.fill('input[type="password"]', pass);
await page.keyboard.press('Enter'); await page.keyboard.press('Enter');
await page.waitForTimeout(5000); await page.waitForTimeout(5000);
} }
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');
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'));
const found = []; const found = [];
filas.forEach(tr => { filas.forEach(tr => {
const txt = tr.innerText; const txt = tr.innerText; const match = txt.match(/(\d{6,10})/);
const match = txt.match(/(\d{6,10})/);
if (match) found.push(match[1]); if (match) found.push(match[1]);
}); });
return [...new Set(found)]; return [...new Set(found)];
}); });
console.log(`🔍 [HomeServe] ${refs.length} expedientes.`); 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');
const link = await page.getByText(ref).first(); const link = await page.getByText(ref).first();
if (await link.isVisible()) { if (await link.isVisible()) {
await link.click(); await link.click(); await page.waitForTimeout(1500);
await page.waitForTimeout(1500);
const fullData = await page.evaluate(() => { const fullData = await page.evaluate(() => {
const d = {}; const d = {};
// HomeServe suele ser tablas limpias: TR -> TD (Clave) | TD (Valor)
const rows = Array.from(document.querySelectorAll('tr')); const rows = Array.from(document.querySelectorAll('tr'));
rows.forEach(r => { rows.forEach(r => {
const cells = r.querySelectorAll('td'); const cells = r.querySelectorAll('td');
// Si tiene al menos 2 celdas
if(cells.length >= 2) { if(cells.length >= 2) {
const k = cells[0].innerText.toUpperCase().trim().replace(':', ''); const k = cells[0].innerText.toUpperCase().trim().replace(':', '');
const v = cells[1].innerText.trim(); const v = cells[1].innerText.trim();
if(k.length > 1 && v.length > 0) d[k] = v; if(k.length > 1 && v.length > 0) d[k] = v;
} }
// A veces son tablas de 4 columnas (Clave | Valor | Clave | Valor)
if(cells.length >= 4) { if(cells.length >= 4) {
const k2 = cells[2].innerText.toUpperCase().trim().replace(':', ''); const k2 = cells[2].innerText.toUpperCase().trim().replace(':', '');
const v2 = cells[3].innerText.trim(); const v2 = cells[3].innerText.trim();
if(k2.length > 1 && v2.length > 0) d[k2] = v2; 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 (fullData) await saveServiceToDB(ownerId, 'homeserve', ref, fullData);
if (fullData) {
await saveServiceToDB(ownerId, 'homeserve', ref, fullData);
}
} }
} catch (errRef) { console.error(`⚠️ Error ref ${ref}`); } } catch (errRef) { console.error(`⚠️ Error ref ${ref}`); }
} }
@@ -194,7 +197,7 @@ async function runHomeserve(ownerId, user, pass) {
} }
async function saveServiceToDB(ownerId, provider, ref, data) { async function saveServiceToDB(ownerId, provider, ref, data) {
console.log(`💾 Guardando ${ref} con ${Object.keys(data).length} variables encontradas.`); console.log(`💾 Guardando ${ref}...`);
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')