{ "name": "My workflow", "nodes": [ { "parameters": { "httpMethod": "POST", "path": "`search`", "responseMode": "lastNode", "options": {} }, "type": "n8n-nodes-base.webhook", "typeVersion": 2.1, "position": [ 4240, 8576 ], "id": "bcbd7779-5f36-43c2-b189-6048ef13f5bf", "name": "Webhook" }, { "parameters": { "errorMessage": "400 Simple Error " }, "type": "n8n-nodes-base.stopAndError", "typeVersion": 1, "position": [ 4912, 8624 ], "id": "18a14323-8769-466f-95a7-dde5bb307f0b", "name": "Stop and Error" }, { "parameters": { "jsCode": "const items = $input.all();\nconst keywords = items[0].json.keywords;\nconst area = items[0].json.area || '';\nconst region = items[0].json.region || '';\nconst searchId = items[0].json.searchId;\n\n// Arrays de t\u00e9rminos que mejoran la b\u00fasqueda de contactos\nconst contactTerms = [\n 'email contacto',\n 'directorio profesionales',\n 'perfil profesional',\n 'curriculum vitae'\n];\n\nconst queries = [];\n\n// Generar m\u00faltiples queries con diferentes enfoques\ncontactTerms.forEach(term => {\n queries.push({\n query: `${keywords} ${region} ${term} site:.cl OR site:linkedin.com OR site:researchgate.net`,\n searchTerm: keywords,\n area: area,\n region: region,\n searchId: searchId,\n queryType: term\n });\n});\n\n// Query espec\u00edfica para colegios profesionales\nif (area && area.toLowerCase().includes('salud')) {\n queries.push({\n query: `${keywords} ${region} site:colegiomedicochile.cl OR site:minsal.cl contacto`,\n searchTerm: keywords,\n area: area,\n region: region,\n searchId: searchId,\n queryType: 'colegio_profesional'\n });\n}\n\nreturn queries.map(q => ({ json: q }));" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 4912, 8432 ], "id": "34fc3a5b-e66a-4949-9bef-c19305fdf730", "name": "Build Search Query" }, { "parameters": { "url": "https://serpapi.com/search", "sendQuery": true, "queryParameters": { "parameters": [ { "name": "q", "value": "={{$json.query}}" }, { "name": "api_key", "value": "d401a6726f70ad6a019aafe5a4a4496171d4e50becb4f7ec1047346d4c8b1036" }, { "name": "location", "value": "chile" }, { "name": "gl", "value": "cl" }, { "name": "hl", "value": "es" } ] }, "options": { "timeout": 10000 } }, "type": "n8n-nodes-base.httpRequest", "typeVersion": 4.3, "position": [ 5136, 8432 ], "id": "55246d27-a97b-46ca-9960-7b4c639e827d", "name": "Google Custom Search API", "retryOnFail": true, "waitBetweenTries": 2000 }, { "parameters": { "conditions": { "options": { "caseSensitive": true, "leftValue": "", "typeValidation": "loose", "version": 3 }, "conditions": [ { "id": "aa00dba6-61ac-4f88-9dc2-8d9d2793170d", "leftValue": "={{$json.keywords}}", "rightValue": "undefined", "operator": { "type": "string", "operation": "notEquals" } }, { "id": "d82c94fd-6e96-4ee7-b2ca-8fbb5d96ab43", "leftValue": "={{$json.keywords.length}}", "rightValue": 3, "operator": { "type": "number", "operation": "gte" } }, { "id": "f1e94d69-83b8-4c3a-8ea0-0df622386d07", "leftValue": "={{$json.searchId}}", "rightValue": 0, "operator": { "type": "string", "operation": "exists", "singleValue": true } } ], "combinator": "and" }, "looseTypeValidation": true, "options": {} }, "type": "n8n-nodes-base.if", "typeVersion": 2.3, "position": [ 4688, 8576 ], "id": "403711b2-b561-467d-9204-bc9edf739c40", "name": "IF Data is Valid", "alwaysOutputData": false }, { "parameters": { "jsCode": "const items = $input.all();\nconst body = items[0].json.body;\n\nreturn [{\n json: {\n searchId: body.search_id,\n keywords: body.keywords,\n area: body.area,\n region: body.region,\n timestamp: new Date().toISOString(),\n startTime: Date.now()\n }\n}];" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 4464, 8576 ], "id": "31b91048-124b-439b-ae0a-3e9d8d45175e", "name": "Data Structure" }, { "parameters": { "jsCode": "// ==================== PARSER MEJORADO - AUTO-DETECT ESTRUCTURA ====================\n\ntry {\n let inputData = $input.all();\n \n // Si inputData est\u00e1 vac\u00edo o no tiene items, intentar extraer del primer objeto\n if (!inputData || inputData.length === 0) {\n return [{ json: { items: [] } }];\n }\n\n // DEBUG: Ver qu\u00e9 estructura tiene el input\n const firstItem = inputData[0];\n let itemsToProcess = [];\n\n // Intenta detectar d\u00f3nde est\u00e1n realmente los datos\n if (firstItem.json && Array.isArray(firstItem.json)) {\n // Si json es directamente un array\n itemsToProcess = firstItem.json;\n } else if (firstItem.json && Array.isArray(firstItem.json.items)) {\n // Si json tiene un campo \"items\"\n itemsToProcess = firstItem.json.items;\n } else if (firstItem.json && Array.isArray(firstItem.json.organic_results)) {\n // Si es de SerpAPI directamente\n itemsToProcess = firstItem.json.organic_results;\n } else if (Array.isArray(firstItem.json)) {\n // Intenta directamente\n itemsToProcess = firstItem.json;\n } else {\n // Si firstItem.json es un objeto individual\n itemsToProcess = inputData.map(item => item.json);\n }\n\n // Si a\u00fan est\u00e1 vac\u00edo\n if (!itemsToProcess || itemsToProcess.length === 0) {\n return [{ json: { items: [], debug: \"No se encontraron datos para procesar\" } }];\n }\n\n // ==================== MAPEO REGIONES CHILE ====================\n const regionKeywords = {\n \"Arica y Parinacota\": [\"arica y parinacota\", \"arica\", \"parinacota\"],\n \"Tarapac\u00e1\": [\"tarapac\u00e1\", \"tarapaca\", \"iquique\"],\n \"Antofagasta\": [\"antofagasta\"],\n \"Atacama\": [\"atacama\", \"copiap\u00f3\", \"copiapo\"],\n \"Coquimbo\": [\"coquimbo\", \"la serena\"],\n \"Valpara\u00edso\": [\"valpara\u00edso\", \"valparaiso\", \"vi\u00f1a\", \"quilpu\u00e9\", \"conc\u00f3n\"],\n \"Santiago\": [\"santiago\", \"metropolitana\", \"rm\", \"\u00f1u\u00f1oa\"],\n \"O'Higgins\": [\"o'higgins\", \"ohiggins\", \"rancagua\"],\n \"Maule\": [\"maule\", \"talca\"],\n \"\u00d1uble\": [\"\u00f1uble\", \"nuble\", \"chill\u00e1n\", \"chillan\"],\n \"Biob\u00edo\": [\"biob\u00edo\", \"biobio\", \"b\u00edo b\u00edo\", \"bio bio\", \"concepci\u00f3n\", \"concepcion\"],\n \"La Araucan\u00eda\": [\"la araucan\u00eda\", \"la araucania\", \"araucan\u00eda\", \"araucania\", \"temuco\"],\n \"Los R\u00edos\": [\"los r\u00edos\", \"los rios\", \"valdivia\"],\n \"Los Lagos\": [\"los lagos\", \"puerto montt\"],\n \"Ays\u00e9n\": [\"ays\u00e9n\", \"aysen\", \"coihaique\", \"coyhaique\"],\n \"Magallanes\": [\"magallanes\", \"punta arenas\"],\n };\n\n const results = [];\n\n // Procesar cada item\n for (let idx = 0; idx < itemsToProcess.length; idx++) {\n const item = itemsToProcess[idx];\n\n // Validar datos b\u00e1sicos\n if (!item.link && !item.title) {\n continue;\n }\n\n // Detectar regi\u00f3n\n let regionInferred = \"Sin especificar\";\n const searchText = ((item.snippet || \"\") + \" \" + (item.title || \"\") + \" \" + (item.link || \"\")).toLowerCase();\n\n for (const [region, regionKeywordsList] of Object.entries(regionKeywords)) {\n let found = false;\n for (const keyword of regionKeywordsList) {\n if (searchText.includes(keyword.toLowerCase())) {\n regionInferred = region;\n found = true;\n break;\n }\n }\n if (found) break;\n }\n\n // Construir resultado\n results.push({\n metadata: {\n title: item.title || null,\n link: item.link || null,\n snippet: item.snippet || null,\n source: item.source || null,\n position: item.position || null,\n },\n region: regionInferred,\n search_id: null,\n keywords: null,\n processed: false,\n html_content: null,\n contact_data: null,\n });\n }\n\n return [{ json: { items: results, count: results.length } }];\n\n} catch (err) {\n let errorMsg = \"Unknown error\";\n if (typeof err === 'string') {\n errorMsg = err;\n } else if (err && typeof err === 'object' && 'message' in err) {\n errorMsg = String(err.message);\n }\n return [{ json: { items: [], error: errorMsg } }];\n}" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 5360, 8432 ], "id": "f2fbe179-3b03-4054-9895-ef9cb14c8711", "name": "Parser1" }, { "parameters": { "jsCode": "try {\n // Obtener todos los items\n const input = $input.first().json;\n let items = [];\n \n // Si viene con estructura de Parser (items array), procesar todos\n if (input.items && Array.isArray(input.items) && input.items.length > 0) {\n items = input.items;\n } else if (input.metadata) {\n // Si ya es un item individual\n items = [input];\n } else {\n throw new Error(\"Invalid input structure: no metadata found\");\n }\n \n // Procesar cada item\n return items.map(item => {\n try {\n const metadata = item.metadata;\n const link = metadata.link.toLowerCase();\n const title = metadata.title.toLowerCase();\n\n // ==================== PATRONES A IGNORAR ====================\n const ignorePatterns = {\n // Marketplaces de servicios\n marketplace: [\n \"cronoshare.cl\",\n \"starofservice.cl\",\n \"uber.com\",\n \"airbnb.com\",\n \"fiverr.com\",\n \"upwork.com\",\n \"freelancer.com\",\n ],\n // Bolsas de empleo\n jobsites: [\n \"computrabajo.cl\",\n \"linkedin.com/jobs\",\n \"trabajos.com\",\n \"buscojob.com\",\n \"indeed.com\",\n \"infojobs.com\",\n ],\n // Instituciones educativas\n education: [\n \"culinary.cl\",\n \".edu.cl\",\n \"universidad\",\n \"escuela\",\n \"curso\",\n \"clase en l\u00ednea\",\n \"diplomado\",\n ],\n // Otros no deseados\n other: [\n \"google.com\",\n \"facebook.com\",\n \"instagram.com\",\n \"tiktok.com\",\n \"youtube.com\",\n \"wikipedia.org\",\n \"wix.com\",\n \"squarespace.com\",\n ],\n };\n\n // ==================== CLASIFICACI\u00d3N ====================\n\n let shouldIgnore = false;\n let ignoreReason = null;\n\n // Verificar si es oferta de trabajo\n if (title.includes(\"oferta\") || title.includes(\"trabajo\") || title.includes(\"empleo\")) {\n if (!link.includes(\"linkedin.com\")) { // LinkedIn profiles s\u00ed son v\u00e1lidos\n shouldIgnore = true;\n ignoreReason = \"job_listing\";\n }\n }\n\n // Verificar patrones a ignorar\n for (const [category, patterns] of Object.entries(ignorePatterns)) {\n for (const pattern of patterns) {\n if (link.includes(pattern.toLowerCase())) {\n shouldIgnore = true;\n ignoreReason = category;\n break;\n }\n }\n if (shouldIgnore) break;\n }\n\n // ==================== VALIDACI\u00d3N POSITIVA ====================\n // Si pas\u00f3 los filtros negativos, verificar que tenga datos de profesional\n\n const validSources = [\n \"linkedin.com\", // LinkedIn profiles\n \"superprof.cl\", // Directorio profesionales\n \"doctoralia.cl\", // M\u00e9dicos\n \"topdoctors.es\", // M\u00e9dicos\n \".cl\", // Sitios chilenos generales\n ];\n\n let isValidSource = false;\n for (const source of validSources) {\n if (link.includes(source)) {\n isValidSource = true;\n break;\n }\n }\n\n // Si no es marketplace/bolsa/escuela Y es fuente v\u00e1lida\n shouldIgnore = shouldIgnore || !isValidSource;\n\n // ==================== RESULTADO ====================\n\n return {\n json: {\n ...item,\n classification: {\n should_process: !shouldIgnore,\n ignore_reason: ignoreReason,\n is_linkedin: link.includes(\"linkedin.com\"),\n source_type: ignoreReason || \"professional\",\n },\n },\n };\n } catch (itemErr) {\n return {\n json: {\n ...item,\n classification: {\n should_process: false,\n ignore_reason: \"error\",\n error: String(itemErr.message || itemErr),\n },\n },\n };\n }\n });\n\n} catch (err) {\n // Error general\n let errorMsg = \"Unknown error\";\n if (typeof err === 'string') {\n errorMsg = err;\n } else if (err && typeof err === 'object' && 'message' in err) {\n errorMsg = String(err.message);\n } else {\n errorMsg = String(err);\n }\n \n return [{ json: { \n classification: {\n should_process: false,\n ignore_reason: \"error\",\n error: errorMsg,\n }\n }}];\n}" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 5584, 8432 ], "id": "43286d55-6c54-4367-9c0e-dad43d509988", "name": "Link Classifier" }, { "parameters": { "jsCode": "try {\n // Procesar TODOS los items que vienen del IF node\n const allItems = $input.all();\n \n return allItems.map(itemWrapper => {\n const item = itemWrapper.json;\n const metadata = item.metadata;\n const title = metadata.title;\n const snippet = metadata.snippet;\n const link = metadata.link;\n\n // ==================== EXTRACCI\u00d3N NOMBRE ====================\n // Formato t\u00edpico LinkedIn: \"Nombre Apellido - Cargo, experiencia...\"\n \n let name = null;\n const nameMatch = title.match(/^([A-Za-z\u00e1\u00e9\u00ed\u00f3\u00fa\u00c1\u00c9\u00cd\u00d3\u00da\u00f1\u00d1\\s]+?)(?:\\s*-|$)/);\n if (nameMatch) {\n name = nameMatch[1].trim();\n // Limpiar si a\u00fan tiene ruido\n if (name.length < 3 || name.toLowerCase().includes(\"error\")) {\n name = null;\n }\n }\n\n // ==================== EXTRACCI\u00d3N CARGO ====================\n // Buscar despu\u00e9s del primer gui\u00f3n\n let position = null;\n const posMatch = title.match(/\\s*-\\s*([^,]+)/);\n if (posMatch) {\n position = posMatch[1].trim();\n // Tomar solo la primera parte (antes de a\u00f1os de experiencia)\n position = position.split(/[,\u00b7]/)[0].trim();\n // Limitar a 50 caracteres\n if (position.length > 50) {\n position = position.split(/\\s+/).slice(0, 3).join(\" \");\n }\n }\n\n // ==================== EMAIL & TEL\u00c9FONO ====================\n const emailRegex = /[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}/g;\n const emails = snippet.match(emailRegex) || [];\n const email = emails.length > 0 ? emails[0] : null;\n\n const phoneRegex = /(\\+?56|0)?[\\s]?9[\\s]?[\\d]{4}[\\s]?[\\d]{4}|(\\+?56|0)?[\\s]?2[\\s]?[\\d]{4}[\\s]?[\\d]{4}/g;\n const phones = snippet.match(phoneRegex) || [];\n const phone = phones.length > 0 ? phones[0] : null;\n\n // ==================== VALIDACI\u00d3N ====================\n // is_valid: true solo si tiene NOMBRE + (CARGO O EMAIL O TEL\u00c9FONO)\n const isValid = !!(name && (position || email || phone));\n \n // validation_score ser\u00e1 calculado en el Backend seg\u00fan criterios finales\n // Aqu\u00ed solo guardamos la informaci\u00f3n extra\u00edda\n const validationScore = isValid ? 1 : 0; // Score temporal, Backend lo recalcula\n\n return {\n json: {\n ...item,\n processed: true,\n contact_data: {\n name: name || \"Desconocido\",\n email: email,\n phone: phone,\n position: position,\n organization: \"LinkedIn\",\n region: item.region,\n source_url: link,\n source_type: \"LinkedIn\",\n is_valid: isValid,\n validation_score: validationScore,\n },\n },\n };\n });\n\n} catch (err) {\n // Error general\n let errorMsg = \"Unknown error\";\n if (typeof err === 'string') {\n errorMsg = err;\n } else if (err && typeof err === 'object' && 'message' in err) {\n errorMsg = String(err.message);\n } else {\n errorMsg = String(err);\n }\n \n return [{ json: { \n contact_data: {\n is_valid: false,\n validation_score: 0,\n scraping_source: \"error\",\n scraping_error: errorMsg,\n }\n }}];\n}" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 6928, 8336 ], "id": "ccd244f5-26bf-4fe7-88a7-e3ba4d9e2fb5", "name": "Code in JavaScript6" }, { "parameters": { "url": "={{ $json.metadata.link }}", "sendQuery": true, "queryParameters": { "parameters": [ { "name": "User-Agent", "value": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36" }, { "name": "Accept", "value": "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8" }, { "name": "Accept-Language", "value": "es-ES,es;q=0.9" }, { "name": "Referer", "value": "https://www.google.com/" } ] }, "options": { "allowUnauthorizedCerts": false, "response": { "response": { "responseFormat": "text" } }, "timeout": 10000 } }, "type": "n8n-nodes-base.httpRequest", "typeVersion": 4.3, "position": [ 6256, 8624 ], "id": "564b16c8-fabb-4764-9741-96ce032e2a09", "name": "HTTP Request3", "onError": "continueRegularOutput" }, { "parameters": { "jsCode": "// Combina respuesta HTTP con datos originales\n// Mantiene estructura completa aunque HTTP falle\n// Detecta correctamente cuando HTTP devuelve error\n\ntry {\n const allItems = $input.all();\n \n return allItems.map(itemWrapper => {\n const item = itemWrapper.json;\n \n // Detectar si fue error (tiene status error) o \u00e9xito\n let htmlContent = null;\n let statusCode = null;\n let fetchError = null;\n\n // ==================== DETECTAR ESTRUCTURA DE RESPUESTA ====================\n \n // Si viene con error (HTTP Request con Continue on Error)\n if (item.error) {\n statusCode = item.error.status || item.error.statusCode || 403;\n fetchError = item.error.message || \"HTTP Error\";\n htmlContent = null;\n }\n // Si viene con statusCode y body\n else if (item.statusCode === 200 && item.body) {\n htmlContent = item.body;\n statusCode = 200;\n fetchError = null;\n }\n // Si viene con error status\n else if (item.statusCode === 403) {\n statusCode = 403;\n fetchError = \"CAPTCHA/Blocked (403)\";\n htmlContent = null;\n }\n else if (item.statusCode >= 500) {\n statusCode = item.statusCode;\n fetchError = `Server error (${statusCode})`;\n htmlContent = null;\n }\n else if (item.statusCode) {\n statusCode = item.statusCode;\n fetchError = `HTTP ${statusCode}`;\n htmlContent = null;\n }\n else {\n fetchError = \"Unknown error\";\n statusCode = null;\n }\n\n // ==================== CONSTRUIR RESPUESTA LIMPIA ====================\n // Preservar TODO del item original EXCEPTO el campo \"error\"\n // (destructuring elimina error, spread incluye todo lo dem\u00e1s)\n const { error, ...cleanItem } = item;\n\n return {\n json: {\n ...cleanItem,\n html_fetch: {\n success: htmlContent ? true : false,\n status_code: statusCode,\n error: fetchError,\n html_content: htmlContent,\n },\n },\n };\n });\n\n} catch (err) {\n let errorMsg = \"Unknown error\";\n if (typeof err === 'string') {\n errorMsg = err;\n } else if (err && typeof err === 'object' && 'message' in err) {\n errorMsg = String(err.message);\n } else {\n errorMsg = String(err);\n }\n\n return [{ json: {\n html_fetch: {\n success: false,\n error: errorMsg,\n html_content: null,\n status_code: null,\n }\n }}];\n}" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 6704, 8528 ], "id": "7f605f0f-8d2b-4123-b968-ec42ed9e2a91", "name": "Combine Data" }, { "parameters": { "mode": "combine", "combineBy": "combineByPosition", "options": {} }, "type": "n8n-nodes-base.merge", "typeVersion": 3.2, "position": [ 6480, 8528 ], "id": "69ecf90d-92c0-4425-b53f-e407e79e407e", "name": "Prepare Data" }, { "parameters": { "jsCode": "try {\n const allItems = $input.all();\n \n return allItems.map(itemWrapper => {\n const item = itemWrapper.json;\n const metadata = item.metadata;\n \n // Obtener HTML de la estructura de 3B-1B\n let htmlContent = null;\n let source = \"snippet\";\n let extractionNotes = \"\";\n\n // Verificar estructura html_fetch limpia de 3B-1B\n if (item.html_fetch) {\n if (item.html_fetch.success && item.html_fetch.html_content) {\n htmlContent = item.html_fetch.html_content;\n source = \"html\";\n extractionNotes = \"HTML obtenido exitosamente\";\n } else if (item.html_fetch.error) {\n extractionNotes = `HTTP ${item.html_fetch.status_code}: ${item.html_fetch.error} - usando snippet`;\n } else {\n extractionNotes = \"Ning\u00fan HTML disponible - usando snippet\";\n }\n } else {\n extractionNotes = \"Sin html_fetch - usando snippet\";\n }\n\n // Usar HTML si lo tenemos, sino snippet\n const textToSearch = htmlContent || metadata.snippet || \"\";\n\n const genericNameTerms = new Set([\n \"contacto\", \"inicio\", \"nombre empresa\", \"empresa\", \"servicios\", \"servicio\",\n \"masajes\", \"masajes providencia\", \"masajes corporales\", \"home\", \"about\"\n ]);\n\n const toTitleCase = (value) => value\n .split(\" \")\n .filter(Boolean)\n .map(word => word.charAt(0).toUpperCase() + word.slice(1).toLowerCase())\n .join(\" \" );\n\n const cleanCandidate = (value) => (value || \"\")\n .replace(/\\s*[-|\u2022]\\s*.*/, \"\")\n .replace(/[\\d()]/g, \"\")\n .replace(/\\s+/g, \" \")\n .trim();\n\n const isGenericCandidate = (value) => {\n const normalized = (value || \"\").toLowerCase().trim();\n if (!normalized || normalized.length < 3) return true;\n if (genericNameTerms.has(normalized)) return true;\n if (/^(contacto|inicio|servicios?|home|masajes?)$/.test(normalized)) return true;\n return false;\n };\n\n const domainToName = (domainValue) => toTitleCase(\n (domainValue || \"\")\n .replace(/^https?:\\/\\//i, \"\")\n .replace(/^www\\./i, \"\")\n .split(\"/\")[0]\n .replace(/\\.(cl|com|net|org|io|app|co|es)$/i, \"\")\n .replace(/[-_]+/g, \" \")\n .trim()\n );\n\n const normalizeOrganization = (sourceValue, linkValue, titleValue) => {\n let org = (sourceValue || \"\").replace(/^LinkedIn\\s*\u00b7\\s*/i, \"\").trim();\n\n if (!org || org.includes(\".\")) {\n try {\n const host = new URL(linkValue || \"\").hostname;\n if (host) {\n org = domainToName(host);\n }\n } catch (_) {}\n } else {\n org = cleanCandidate(org);\n }\n\n if ((!org || isGenericCandidate(org)) && titleValue) {\n const titleOrg = cleanCandidate(titleValue);\n if (!isGenericCandidate(titleOrg)) {\n org = toTitleCase(titleOrg);\n }\n }\n\n return org && !isGenericCandidate(org) ? org : null;\n };\n\n // ==================== NOMBRE ====================\n let name = cleanCandidate(metadata.title);\n const personMatch = (metadata.title || \"\").match(/\\b([A-Z\u00c1\u00c9\u00cd\u00d3\u00da\u00d1][a-z\u00e1\u00e9\u00ed\u00f3\u00fa\u00f1]+(?:\\s+[A-Z\u00c1\u00c9\u00cd\u00d3\u00da\u00d1][a-z\u00e1\u00e9\u00ed\u00f3\u00fa\u00f1]+){1,3})\\b/);\n if (personMatch) {\n name = personMatch[1].trim();\n }\n if (isGenericCandidate(name)) {\n name = null;\n }\n\n // ==================== CARGO ====================\n let position = null;\n const positionKeywords = [\n \"chef\", \"cocinero\", \"pastelero\", \"panadero\", \"sommelier\",\n \"profesor\", \"instructor\", \"tutor\", \"capacitador\", \"consultor\",\n \"m\u00e9dico\", \"doctor\", \"ingeniero\", \"arquitecto\", \"abogado\", \"contador\",\n \"electricista\", \"plomero\", \"carpintero\", \"pintor\", \"mec\u00e1nico\",\n \"gerente\", \"director\", \"administrador\", \"coordinador\", \"supervisor\",\n \"dise\u00f1ador\", \"desarrollador\", \"programador\", \"analista\", \"t\u00e9cnico\",\n ];\n\n for (const keyword of positionKeywords) {\n const regex = new RegExp(`\\\\b${keyword}\\\\b`, \"i\");\n if (regex.test(textToSearch)) {\n position = keyword.charAt(0).toUpperCase() + keyword.slice(1);\n break;\n }\n }\n\n // ==================== EMAIL ====================\n const emailRegex = /[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}/g;\n const emails = textToSearch.match(emailRegex) || [];\n const email = emails.length > 0 ? emails[0] : null;\n\n // ==================== TEL\u00c9FONO ====================\n const phoneRegex = /(\\+?56|0)?[\\s]?9[\\s]?[\\d]{4}[\\s]?[\\d]{4}|(\\+?56|0)?[\\s]?2[\\s]?[\\d]{4}[\\s]?[\\d]{4}/g;\n const phones = textToSearch.match(phoneRegex) || [];\n let phone = phones.length > 0 ? phones[0] : null;\n\n if (phone) {\n phone = phone.replace(/[\\s\\-().]/g, \"\");\n if (!phone.startsWith(\"+56\")) {\n if (phone.startsWith(\"0\")) {\n phone = \"+56\" + phone.substring(1);\n } else if (!phone.startsWith(\"56\")) {\n phone = \"+56\" + phone;\n } else {\n phone = \"+\" + phone;\n }\n }\n }\n\n // ==================== ORGANIZACI\u00d3N ====================\n let organization = normalizeOrganization(metadata.source, metadata.link, metadata.title);\n\n if (!name && organization) {\n name = organization;\n }\n\n // ==================== VALIDACI\u00d3N ====================\n // Para Superprof: requiere NOMBRE + (EMAIL O TEL\u00c9FONO)\n // Si no hay email/tel\u00e9fono en snippet \u2192 is_valid = false\n const isValid = !!(name && (email || phone));\n const validationScore = isValid ? 1 : 0;\n\n return {\n json: {\n ...item,\n processed: true,\n contact_data: {\n name: name || \"Desconocido\",\n email: email,\n phone: phone,\n position: position,\n organization: organization,\n region: item.region,\n source_url: metadata.link,\n source_type: metadata.source,\n is_valid: isValid,\n validation_score: validationScore,\n },\n extraction_info: {\n source_used: source,\n notes: extractionNotes,\n http_status: item.html_fetch?.status_code || null,\n },\n },\n };\n });\n\n} catch (err) {\n let errorMsg = \"Unknown error\";\n if (typeof err === 'string') {\n errorMsg = err;\n } else if (err && typeof err === 'object' && 'message' in err) {\n errorMsg = String(err.message);\n } else {\n errorMsg = String(err);\n }\n \n return [{ json: { \n contact_data: {\n is_valid: false,\n validation_score: 0,\n error: errorMsg,\n }\n }}];\n}" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 6928, 8528 ], "id": "03ee11a1-43c2-433e-88e8-d3fa8445c2d4", "name": "Code in JavaScript8" }, { "parameters": { "conditions": { "options": { "caseSensitive": true, "leftValue": "", "typeValidation": "loose", "version": 3 }, "conditions": [ { "id": "2cd8af3f-f9dd-4dfb-971d-dea61bd52d7a", "leftValue": "={{ !!$json.contact_data }}", "rightValue": "true", "operator": { "type": "string", "operation": "equals", "name": "filter.operator.equals" } } ], "combinator": "and" }, "looseTypeValidation": true, "options": {} }, "type": "n8n-nodes-base.if", "typeVersion": 2.3, "position": [ 7824, 8384 ], "id": "9632a364-8bc5-433c-a727-959dd8f4ee66", "name": "Valid?" }, { "parameters": {}, "type": "n8n-nodes-base.merge", "typeVersion": 3.2, "position": [ 7152, 8384 ], "id": "b7e78d2c-ff2b-42e9-8f30-1716ce3742c7", "name": "Merge Branches" }, { "parameters": { "jsCode": "const genericTerms = new Set(['contacto','inicio','empresa','servicio','servicios','home','about','nosotros','quienes somos','sitio oficial']);\nconst companyTerms = new Set(['spa','ltda','limitada','sa','s.a','eirl','sociedad','empresa','group','grupo','holding','consultora','consultores','clinica','cl\u00ednica']);\nconst connectorWords = new Set(['de','del','la','las','los','y']);\n\nconst clean = (value) => (value || '').toString().replace(/\\s+/g, ' ').replace(/\\s*[-|\u2022]\\s*.*/, '').trim();\nconst titleCase = (value) => clean(value).split(' ').filter(Boolean).map(w => w.charAt(0).toUpperCase() + w.slice(1).toLowerCase()).join(' ');\nconst isGeneric = (value) => {\n const v = clean(value).toLowerCase();\n return !v || v.length < 3 || genericTerms.has(v) || /^(contacto|inicio|servicios?|home|masajes?)$/.test(v);\n};\nconst fromUrl = (rawUrl) => {\n try {\n const host = new URL(rawUrl || '').hostname.replace(/^www\\./i, '');\n return titleCase(host.replace(/\\.(cl|com|net|org|io|app|co|es)$/i, '').replace(/[-_]+/g, ' '));\n } catch (_) {\n return null;\n }\n};\nconst isLikelyPerson = (value) => {\n const candidate = clean(value);\n if (isGeneric(candidate) || /\\d/.test(candidate)) return false;\n const words = candidate.split(/\\s+/).filter(Boolean);\n const core = words.filter(w => !connectorWords.has(w.toLowerCase()));\n if (core.length < 2 || core.length > 4) return false;\n return core.every(w => /^[A-Z\u00c1\u00c9\u00cd\u00d3\u00da\u00d1][a-z\u00e1\u00e9\u00ed\u00f3\u00fa\u00f1'\u2019-]+$/.test(w) && !companyTerms.has(w.toLowerCase().replace(/\\./g, '')));\n};\n\nreturn $input.all().map((itemWrapper) => {\n const item = itemWrapper.json;\n const contact = item.contact_data || {};\n const title = clean(item.metadata?.title || item.title || '');\n const source = clean(item.metadata?.source || contact.source_type || '');\n const url = contact.source_url || item.metadata?.link || '';\n\n let name = clean(contact.name);\n let organization = clean(contact.organization);\n\n if (!isLikelyPerson(name)) {\n const match = title.match(/\\b([A-Z\u00c1\u00c9\u00cd\u00d3\u00da\u00d1][a-z\u00e1\u00e9\u00ed\u00f3\u00fa\u00f1]+(?:\\s+[A-Z\u00c1\u00c9\u00cd\u00d3\u00da\u00d1][a-z\u00e1\u00e9\u00ed\u00f3\u00fa\u00f1]+){1,3})\\b/);\n const titlePerson = match ? clean(match[1]) : '';\n name = isLikelyPerson(titlePerson) ? titlePerson : null;\n }\n\n if (!organization || isGeneric(organization) || isLikelyPerson(organization) || organization.includes('.')) {\n if (source && !isGeneric(source) && !isLikelyPerson(source) && !source.includes('.')) {\n organization = titleCase(source);\n }\n }\n\n if (!organization || isGeneric(organization) || isLikelyPerson(organization)) {\n const domainOrg = fromUrl(url);\n if (domainOrg && !isGeneric(domainOrg) && !isLikelyPerson(domainOrg)) {\n organization = domainOrg;\n }\n }\n\n if (!name && organization) {\n name = organization;\n }\n\n let score = Number(contact.validation_score);\n const isValid = !!contact.is_valid;\n if (!Number.isFinite(score)) score = isValid ? 1 : 0.3;\n if (!isValid && score < 0.3) score = 0.3;\n\n contact.name = name || contact.name || organization || 'Desconocido';\n contact.organization = organization || null;\n contact.is_valid = isValid;\n contact.validation_score = score;\n\n item.contact_data = contact;\n return { json: item };\n});" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 7376, 8384 ], "id": "4de8e1e3-4938-4461-a02e-6d535a9cbd94", "name": "Contact Final Validation" }, { "parameters": { "jsCode": "const regionRules = [\n { region: 'Arica y Parinacota', keys: ['arica y parinacota','arica','parinacota'] },\n { region: 'Tarapac\u00e1', keys: ['tarapac\u00e1','tarapaca','iquique'] },\n { region: 'Antofagasta', keys: ['antofagasta'] },\n { region: 'Atacama', keys: ['atacama','copiap\u00f3','copiapo'] },\n { region: 'Coquimbo', keys: ['coquimbo','la serena'] },\n { region: 'Valpara\u00edso', keys: ['valpara\u00edso','valparaiso','vi\u00f1a del mar','vina del mar','quilpu\u00e9','quilpue','conc\u00f3n','concon'] },\n { region: 'Santiago', keys: ['santiago','metropolitana','rm','providencia','las condes','\u00f1u\u00f1oa','nunoa'] },\n { region: \"O'Higgins\", keys: [\"o'higgins\",'ohiggins','rancagua'] },\n { region: 'Maule', keys: ['maule','talca'] },\n { region: '\u00d1uble', keys: ['\u00f1uble','nuble','chill\u00e1n','chillan'] },\n { region: 'Biob\u00edo', keys: ['biob\u00edo','biobio','b\u00edo b\u00edo','bio bio','concepci\u00f3n','concepcion'] },\n { region: 'La Araucan\u00eda', keys: ['la araucan\u00eda','la araucania','araucan\u00eda','araucania','temuco'] },\n { region: 'Los R\u00edos', keys: ['los r\u00edos','los rios','valdivia'] },\n { region: 'Los Lagos', keys: ['los lagos','puerto montt'] },\n { region: 'Ays\u00e9n', keys: ['ays\u00e9n','aysen','coihaique','coyhaique'] },\n { region: 'Magallanes', keys: ['magallanes','punta arenas'] }\n];\n\nreturn $input.all().map((itemWrapper) => {\n const item = itemWrapper.json;\n const contact = item.contact_data || {};\n const text = [\n contact.source_url || '',\n contact.organization || '',\n item.metadata?.title || '',\n item.metadata?.snippet || '',\n item.snippet || ''\n ].join(' ').toLowerCase();\n\n let inferred = null;\n for (const rule of regionRules) {\n if (rule.keys.some(key => text.includes(key))) {\n inferred = rule.region;\n break;\n }\n }\n\n if (inferred) {\n contact.region = inferred;\n item.region = inferred;\n }\n\n item.contact_data = contact;\n return { json: item };\n});" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 7600, 8384 ], "id": "e0052a9f-9eda-4a5f-a215-1d4e4da592f1", "name": "Region Correction" }, { "parameters": { "method": "POST", "url": "=http://backend:8081/api/v1/searches/{{ Number($(\"Webhook\").first().json.body.search_id) }}/callback", "sendHeaders": true, "headerParameters": { "parameters": [ { "name": "X-N8N-API-KEY", "value": "=Unab.2026" } ] }, "sendBody": true, "bodyParameters": { "parameters": [ { "name": "search_id", "value": "={{$json.search_id}}" }, { "name": "status", "value": "={{$json.status}}" }, { "name": "results_count", "value": "={{$json.results_count}}" }, { "name": "duplicates_count", "value": "={{$json.duplicates_count}}" }, { "name": "execution_time_ms", "value": "={{$json.execution_time_ms}}" }, { "name": "avg_validation_score", "value": "={{$json.avg_validation_score}}" }, { "name": "stats", "value": "={{$json.stats}}" }, { "name": "failed_searches", "value": "={{$json.failed_searches}}" }, { "name": "high_quality_contacts", "value": "={{$json.high_quality_contacts}}" } ] }, "options": {} }, "type": "n8n-nodes-base.httpRequest", "typeVersion": 4.3, "position": [ 8496, 8384 ], "id": "f8e58d8f-3c14-4d37-aa5d-8718e8a78604", "name": "Notify Backend" }, { "parameters": { "operation": "executeQuery", "query": "SET @sid = CAST('{{ \n $items(\"Webhook\")[0].json.search_id \n || $items(\"Webhook\")[0].json.body?.search_id \n || 0 \n}}' AS UNSIGNED);\n\nINSERT INTO contacts\n(name, email, phone, position, organization, region, source_url, source_type, research_lines, is_valid, validation_score)\nVALUES\n(\n '{{$json.contact_data.name.replace(/'/g, \"''\")}}',\n {{$json.contact_data.email ? `'${$json.contact_data.email.replace(/'/g, \"''\")}'` : 'NULL'}},\n {{$json.contact_data.phone ? `'${$json.contact_data.phone.replace(/'/g, \"''\")}'` : 'NULL'}},\n {{$json.contact_data.position ? `'${$json.contact_data.position.replace(/'/g, \"''\")}'` : 'NULL'}},\n {{$json.contact_data.organization ? `'${$json.contact_data.organization.replace(/'/g, \"''\")}'` : 'NULL'}},\n {{$json.contact_data.region ? `'${$json.contact_data.region.replace(/'/g, \"''\")}'` : 'NULL'}},\n '{{$json.contact_data.source_url.replace(/'/g, \"''\")}}',\n {{$json.contact_data.source_type ? `'${$json.contact_data.source_type.replace(/'/g, \"''\")}'` : 'NULL'}},\n NULL,\n 1,\n {{ Number($json.contact_data.validation_score || 1) }}\n)\nON DUPLICATE KEY UPDATE\n id = LAST_INSERT_ID(id),\n updated_at = CURRENT_TIMESTAMP;\n\nSET @cid = LAST_INSERT_ID();\n\nINSERT INTO search_results (search_id, contact_id, relevance_score)\nSELECT\n @sid,\n @cid,\n COALESCE((SELECT validation_score FROM contacts WHERE id = @cid), 1)\nWHERE @sid > 0\nON DUPLICATE KEY UPDATE\n relevance_score = VALUES(relevance_score);\n\nSELECT @sid AS search_id, @cid AS contact_id;", "options": {} }, "type": "n8n-nodes-base.mySql", "typeVersion": 2.5, "position": [ 8048, 8384 ], "id": "fe5726ea-8a31-4e10-979a-606d1e84235c", "name": "Insert Valid Contacts", "credentials": { "mySql": { "name": "" } } }, { "parameters": { "jsCode": "const webhookData = $(\"Webhook\").first().json;\nconst inserted = $(\"Insert Valid Contacts\").all().length;\n\n// En Webhook de n8n normalmente viene en body\nconst rawSearchId = webhookData?.body?.search_id ?? webhookData?.search_id;\n\nconst searchId = Number(rawSearchId);\nif (!Number.isInteger(searchId) || searchId <= 0) {\n throw new Error(`search_id inv\u00e1lido. Valor recibido: ${JSON.stringify(rawSearchId)}`);\n}\n\nreturn [{\n json: {\n search_id: searchId,\n status: \"completed\",\n results_count: inserted,\n duplicates_count: 0,\n execution_time_ms: 0,\n avg_validation_score: 1,\n stats: {},\n failed_searches: 0,\n high_quality_contacts: inserted\n }\n}];" }, "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 8272, 8384 ], "id": "061530b4-cf1d-4db0-a195-43285ad7f416", "name": "Build Callback" }, { "parameters": { "conditions": { "options": { "caseSensitive": true, "leftValue": "", "typeValidation": "loose", "version": 3 }, "conditions": [ { "id": "f44c22a0-7f19-420e-a27b-3ebf11cc17e3", "leftValue": "={{ $json.classification.should_process }}", "rightValue": "true", "operator": { "type": "string", "operation": "equals", "name": "filter.operator.equals" } } ], "combinator": "and" }, "looseTypeValidation": true, "options": {} }, "type": "n8n-nodes-base.if", "typeVersion": 2.3, "position": [ 5808, 8432 ], "id": "1b15e3f3-014e-4aae-a4cd-1a76bb9770ae", "name": "Data Valid?" }, { "parameters": { "conditions": { "options": { "caseSensitive": true, "leftValue": "", "typeValidation": "loose", "version": 3 }, "conditions": [ { "id": "523301ab-f2f3-4d0b-9cdc-0528dd2b1d6d", "leftValue": "={{ $json.classification.is_linkedin }}", "rightValue": "true", "operator": { "type": "string", "operation": "equals", "name": "filter.operator.equals" } } ], "combinator": "and" }, "looseTypeValidation": true, "options": {} }, "type": "n8n-nodes-base.if", "typeVersion": 2.3, "position": [ 6032, 8432 ], "id": "ac59c7a5-7502-400c-a0d1-3c115e270b07", "name": "Linkedin?" } ], "connections": { "Webhook": { "main": [ [ { "node": "Data Structure", "type": "main", "index": 0 } ] ] }, "Build Search Query": { "main": [ [ { "node": "Google Custom Search API", "type": "main", "index": 0 } ] ] }, "Google Custom Search API": { "main": [ [ { "node": "Parser1", "type": "main", "index": 0 } ] ] }, "IF Data is Valid": { "main": [ [ { "node": "Build Search Query", "type": "main", "index": 0 } ], [ { "node": "Stop and Error", "type": "main", "index": 0 } ] ] }, "Data Structure": { "main": [ [ { "node": "IF Data is Valid", "type": "main", "index": 0 } ] ] }, "Parser1": { "main": [ [ { "node": "Link Classifier", "type": "main", "index": 0 } ] ] }, "Link Classifier": { "main": [ [ { "node": "Data Valid?", "type": "main", "index": 0 } ] ] }, "HTTP Request3": { "main": [ [ { "node": "Prepare Data", "type": "main", "index": 1 } ] ] }, "Prepare Data": { "main": [ [ { "node": "Combine Data", "type": "main", "index": 0 } ] ] }, "Combine Data": { "main": [ [ { "node": "Code in JavaScript8", "type": "main", "index": 0 } ] ] }, "Code in JavaScript8": { "main": [ [ { "node": "Merge Branches", "type": "main", "index": 1 } ] ] }, "Code in JavaScript6": { "main": [ [ { "node": "Merge Branches", "type": "main", "index": 0 } ] ] }, "Valid?": { "main": [ [ { "node": "Insert Valid Contacts", "type": "main", "index": 0 } ] ] }, "Merge Branches": { "main": [ [ { "node": "Contact Final Validation", "type": "main", "index": 0 } ] ] }, "Contact Final Validation": { "main": [ [ { "node": "Region Correction", "type": "main", "index": 0 } ] ] }, "Region Correction": { "main": [ [ { "node": "Valid?", "type": "main", "index": 0 } ] ] }, "Insert Valid Contacts": { "main": [ [ { "node": "Build Callback", "type": "main", "index": 0 } ] ] }, "Build Callback": { "main": [ [ { "node": "Notify Backend", "type": "main", "index": 0 } ] ] }, "Data Valid?": { "main": [ [ { "node": "Linkedin?", "type": "main", "index": 0 } ] ] }, "Linkedin?": { "main": [ [ { "node": "Code in JavaScript6", "type": "main", "index": 0 } ], [ { "node": "HTTP Request3", "type": "main", "index": 0 }, { "node": "Prepare Data", "type": "main", "index": 0 } ] ] } }, "active": true, "settings": { "executionOrder": "v1", "availableInMCP": false }, "versionId": "56043686-1c91-4f2c-bfe6-71dfe9240ca5", "meta": { "templateCredsSetupCompleted": true }, "id": "Mng5-VRYT5d6W3iWV0o6H", "tags": [] }