Extract Structured Data From Gmail Attachments to Google Sheets, Gpt Vision

The workflow JSON

Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →

{
  "meta": {
    "templateCredsSetupCompleted": true
  },
  "name": "Extract structured data from Gmail attachments to Google Sheets using GPT-4o Vision",
  "nodes": [
    {
      "id": "762adf7a-01c4-44f8-800b-b084812cf0c3",
      "name": "Workflow Description",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        2384,
        1216
      ],
      "parameters": {
        "width": 380,
        "height": 776,
        "content": "##  Email Attachment Processor Template\n\nAutomatically extract structured information from emails using AI.\n\n### What this workflow does:\n1. **Monitors Gmail** for emails from specified domains\n2. **Classifies emails** into custom categories\n3. **Processes attachments** (Excel, PPTX, PDF, images) using GPT Vision\n4. **Extracts structured data** based on your schema\n5. **Saves to Google Sheets** with classification\n6. **Labels processed emails** in Gmail\n7. **Sends to custom API** (optional)\n8. **Notifies via Slack** (optional)\n\n### Requirements:\n- Gmail OAuth2 credentials\n- OpenAI API key (for GPT-4 Vision)\n- Google Sheets OAuth2 credentials\n- AWS S3 bucket (for image processing)\n- ConvertAPI account (for PPTX/PDF conversion)\n- Slack OAuth2 (optional)\n\n### Setup:\n1. Configure all credentials\n2. Update Google Sheets document ID in Variables\n3. Update S3 bucket name in Variables\n4. Customize domain filter in 'Domain Filter' node\n5. Customize classification keywords as needed"
      },
      "typeVersion": 1
    },
    {
      "id": "1a385d9d-4e9a-4a56-beb3-4d3e7dd9649b",
      "name": "Email Classification Info",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        3040,
        1184
      ],
      "parameters": {
        "color": 7,
        "width": 332,
        "height": 336,
        "content": "###  Email Type Detection\n\n**Customize keywords for your use case:**\n\n**Category A (e.g., Cancellation):**\n- Add your keywords here\n\n**Category B (e.g., Updates):**\n- Add your keywords here\n\n**Category C (e.g., New Items):**\n- Default category\n\nCustomize keywords in the 'Domain Filter' code node."
      },
      "typeVersion": 1
    },
    {
      "id": "579e95c7-8c34-4446-a229-061652df9808",
      "name": "Gmail Labels Info",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        6560,
        1264
      ],
      "parameters": {
        "color": 3,
        "width": 280,
        "height": 216,
        "content": "###  Gmail Labels\n\nProcessed emails are labeled:\n- Processed_CategoryA\n- Processed_CategoryB\n- Processed_CategoryC\n\nLabels are created automatically if they don't exist.\n\n**Customize label names in the code nodes.**"
      },
      "typeVersion": 1
    },
    {
      "id": "e852a330-e3ae-4b4f-b887-eec29afafa25",
      "name": "File Processing Info",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        3696,
        1152
      ],
      "parameters": {
        "color": 7,
        "width": 300,
        "height": 344,
        "content": "###  File Processing Routes\n\n**Excel files (Category C):**\n\u2192 Direct data extraction\n\n**Excel files (Category A/B):**\n\u2192 GPT Vision analysis\n\n**PPTX/PDF files:**\n\u2192 Convert to images \u2192 GPT Vision\n\n**Image attachments:**\n\u2192 Direct GPT Vision analysis\n\n**Text-only emails:**\n\u2192 GPT text analysis"
      },
      "typeVersion": 1
    },
    {
      "id": "eb599c2d-491a-4824-85b9-a503a4b65c47",
      "name": "ConvertAPI Info",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        4176,
        1760
      ],
      "parameters": {
        "color": 3,
        "width": 300,
        "height": 284,
        "content": "###  ConvertAPI Setup\n\nUsed for converting PPTX/PDF to images for GPT Vision analysis.\n\nhttps://www.convertapi.com\n\n*This service is paid but offers free initial conversions.*\n\nReplace the HTTP Header Auth credential with your ConvertAPI secret."
      },
      "typeVersion": 1
    },
    {
      "id": "5b69b7ca-b4e1-42d3-ab8b-cfb915f745a1",
      "name": "Google Sheets Info",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        6128,
        1136
      ],
      "parameters": {
        "color": 3,
        "width": 300,
        "height": 368,
        "content": "###  Google Sheets Output\n\nExtracted data is saved with columns:\n- email_type (Category)\n- field_1 (Date)\n- field_2 (Time)\n- field_3 (Name)\n- field_4 (Amount)\n- And more...\n\n**Customize the output schema in the 'Parse All Results' node.**\n\n**Update the Document ID** in Variables or directly in the Google Sheets node."
      },
      "typeVersion": 1
    },
    {
      "id": "9a32ec51-62fb-49c8-9941-60bf6a1adc13",
      "name": "AWS S3 Info",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        4208,
        1024
      ],
      "parameters": {
        "color": 7,
        "width": 300,
        "height": 288,
        "content": "###  AWS S3 Setup\n\nImages are temporarily uploaded to S3 for GPT Vision processing.\n\n**Required:**\n1. Create an S3 bucket\n2. Configure AWS credentials in n8n\n3. Update bucket name in Variables\n\nImages are used for GPT Vision API calls and can be cleaned up periodically."
      },
      "typeVersion": 1
    },
    {
      "id": "61bd3fd0-9fdd-4cf0-9edf-28721a2358da",
      "name": "Schedule Trigger",
      "type": "n8n-nodes-base.scheduleTrigger",
      "position": [
        2832,
        1536
      ],
      "parameters": {
        "rule": {
          "interval": [
            {
              "field": "minutes",
              "minutesInterval": 30
            }
          ]
        }
      },
      "typeVersion": 1.2
    },
    {
      "id": "11ce6b59-1436-4ef5-a0ef-71f9aa98c763",
      "name": "Gmail - Get Emails",
      "type": "n8n-nodes-base.gmail",
      "position": [
        3040,
        1536
      ],
      "parameters": {
        "limit": 1,
        "filters": {
          "q": "-label:Processed_CategoryA -label:Processed_CategoryB -label:Processed_CategoryC",
          "receivedAfter": "={{ $now.minus({ days: 60 }).toISO() }}"
        },
        "operation": "getAll"
      },
      "typeVersion": 2.1
    },
    {
      "id": "a6dff01e-96d5-4c57-8def-5bd8c1216c26",
      "name": "Domain Filter",
      "type": "n8n-nodes-base.code",
      "position": [
        3264,
        1536
      ],
      "parameters": {
        "mode": "runOnceForEachItem",
        "jsCode": "// ============================================================\n// DOMAIN FILTER & EMAIL CLASSIFICATION\n// ============================================================\n// Customize this node for your specific use case\n// ============================================================\n\nconst json = $input.item.json;\nconst subject = json.subject || json.Subject || '';\nconst snippet = json.snippet || '';\nconst fromRaw = json.from || json.From || '';\nconst fromText = typeof fromRaw === 'object' ? (fromRaw.text || JSON.stringify(fromRaw)) : String(fromRaw);\nconst text = (subject + ' ' + snippet + ' ' + fromText).toLowerCase();\n\n// ============================================================\n// STEP 1: Configure your allowed domains\n// ============================================================\n// Add the domains you want to monitor (without @ symbol)\n// Example: ['company.com', 'partner.co.jp', 'supplier.net']\nconst allowedDomains = [\n  // Add your domains here\n  // 'example.com',\n  // 'example.co.jp',\n];\n\n// Check if email is from an allowed domain\nconst fromLower = fromText.toLowerCase();\nconst isFromAllowedDomain = allowedDomains.length === 0 || allowedDomains.some(d => fromLower.includes(d));\n\n// ============================================================\n// STEP 2: Configure classification keywords\n// ============================================================\n\n// Category A keywords (e.g., Cancellations, Deletions)\nconst categoryAKeywords = [\n  // Add your Category A keywords here\n  // 'cancel', 'cancelled', 'delete', 'remove'\n];\n\n// Category B keywords (e.g., Updates, Changes)\nconst categoryBKeywords = [\n  // Add your Category B keywords here\n  // 'update', 'change', 'modify', 'reschedule'\n];\n\n// Category C keywords (e.g., New items - this is the default)\nconst categoryCKeywords = [\n  // Add your Category C keywords here (optional)\n  // 'new', 'offer', 'proposal', 'available'\n];\n\n// ============================================================\n// STEP 3: Configure identifier codes (optional)\n// ============================================================\n// If your emails contain specific codes like [ABC] or \u3010XYZ\u3011\nconst identifierCodes = [\n  // Add your identifier codes here (lowercase)\n  // 'abc', 'xyz', 'code1'\n];\n\n// Normalize full-width characters to ASCII\nconst normalizeAscii = value => String(value || '').replace(/[\uff21-\uff3a\uff10-\uff19]/g, ch => String.fromCharCode(ch.charCodeAt(0) - 0xFEE0));\nconst textUpper = normalizeAscii(subject + ' ' + snippet + ' ' + fromText).toUpperCase();\n\n// Check for identifier codes in brackets\nconst hasIdentifierCode = identifierCodes.some(c => \n  textUpper.includes('\u3010' + c.toUpperCase() + '\u3011') || \n  textUpper.includes('[' + c.toUpperCase() + ']')\n);\n\n// ============================================================\n// Classification Logic\n// ============================================================\n\n// Skip if not from allowed domain and no identifier match\nif (!isFromAllowedDomain && !hasIdentifierCode) {\n  return { json: { _skip: true } };\n}\n\n// Check if email matches any category keywords\nconst isCategoryA = categoryAKeywords.some(k => text.includes(k.toLowerCase()));\nconst isCategoryB = categoryBKeywords.some(k => text.includes(k.toLowerCase()));\nconst hasCategoryC = categoryCKeywords.length === 0 || categoryCKeywords.some(k => text.includes(k.toLowerCase()));\n\n// If no keywords configured and domain doesn't match, skip\nif (categoryAKeywords.length === 0 && categoryBKeywords.length === 0 && categoryCKeywords.length === 0) {\n  // Process all emails from allowed domains when no keywords configured\n  if (!isFromAllowedDomain) {\n    return { json: { _skip: true } };\n  }\n}\n\n// Determine email type/category\nlet emailType = 'CategoryC'; // Default\nif (isCategoryA) emailType = 'CategoryA';\nelse if (isCategoryB) emailType = 'CategoryB';\n\nreturn { json: { ...json, _skip: false, _email_type: emailType } };"
      },
      "typeVersion": 2
    },
    {
      "id": "e303d6d7-5c68-4fc6-97d4-4e5e6c171472",
      "name": "Filter Valid Emails",
      "type": "n8n-nodes-base.filter",
      "position": [
        3488,
        1536
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 1,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "operator": {
                "type": "boolean",
                "operation": "notEquals"
              },
              "leftValue": "={{ $json._skip }}",
              "rightValue": true
            }
          ]
        }
      },
      "typeVersion": 2
    },
    {
      "id": "6a426693-2bfc-4d4a-a17a-9f89973f9dc9",
      "name": "Gmail - Get Attachments",
      "type": "n8n-nodes-base.gmail",
      "position": [
        3712,
        1536
      ],
      "parameters": {
        "simple": false,
        "options": {
          "downloadAttachments": true,
          "dataPropertyAttachmentsPrefixName": "attachment_"
        },
        "messageId": "={{ $json.id }}",
        "operation": "get"
      },
      "typeVersion": 2.1
    },
    {
      "id": "7ce1026b-c28f-4e63-95fc-8dcb61814c3f",
      "name": "Detect File Types",
      "type": "n8n-nodes-base.code",
      "position": [
        3920,
        1536
      ],
      "parameters": {
        "mode": "runOnceForEachItem",
        "jsCode": "// ============================================================\n// FILE TYPE DETECTION\n// ============================================================\n// Identifies attachment types and preserves metadata\n// ============================================================\n\nconst item = $input.item;\nconst binary = item.binary || {};\nconst json = item.json;\n\nconst subject = json.Subject || json.subject || '';\nconst fromData = json.From || json.from || '';\n\nlet emailType = json._email_type || '';\nif (!emailType) {\n  try {\n    const currentId = json.id;\n    const filteredItems = $('Filter Valid Emails').all();\n    for (const f of filteredItems) {\n      if (f.json.id === currentId && f.json._email_type) {\n        emailType = f.json._email_type;\n        break;\n      }\n    }\n  } catch(e) {}\n}\nemailType = emailType || 'CategoryC';\n\nlet fileInfo = { excel: null, pptx: null, pdf: null, image: null };\nlet attachments = [];\n\nfor (const key of Object.keys(binary)) {\n  const bin = binary[key];\n  if (!bin || !bin.fileName) continue;\n  const fn = bin.fileName.toLowerCase();\n  const mime = (bin.mimeType || '').toLowerCase();\n  attachments.push({ filename: bin.fileName, mimeType: bin.mimeType, key: key });\n  \n  if (fn.endsWith('.xlsx') || fn.endsWith('.xls')) fileInfo.excel = { key, fileName: bin.fileName };\n  else if (fn.endsWith('.pptx') || fn.endsWith('.ppt')) fileInfo.pptx = { key, fileName: bin.fileName };\n  else if (fn.endsWith('.pdf') || mime.includes('pdf')) fileInfo.pdf = { key, fileName: bin.fileName };\n  else if (fn.endsWith('.png') || fn.endsWith('.jpg') || fn.endsWith('.jpeg') || mime.startsWith('image/')) fileInfo.image = { key, fileName: bin.fileName, mimeType: bin.mimeType || 'image/png' };\n}\n\n// Variable format types need GPT Vision even for Excel\nconst isVariableFormat = emailType === 'CategoryA' || emailType === 'CategoryB';\nconst hasExcelDirect = fileInfo.excel !== null && !isVariableFormat;\nconst hasFileConvert = fileInfo.pptx !== null || fileInfo.pdf !== null || (fileInfo.excel !== null && isVariableFormat);\n\nlet convType = null, convKey = null;\nif (fileInfo.pptx) { convType = 'pptx'; convKey = fileInfo.pptx.key; }\nelse if (fileInfo.pdf) { convType = 'pdf'; convKey = fileInfo.pdf.key; }\nelse if (fileInfo.excel && isVariableFormat) { convType = 'xlsx'; convKey = fileInfo.excel.key; }\n\nconst output = {\n  ...json,\n  id: json.id,\n  Subject: subject,\n  subject: subject,\n  From: fromData,\n  from: fromData,\n  attachments: attachments,\n  has_excel: hasExcelDirect,\n  has_pptx_pdf: hasFileConvert,\n  has_image: fileInfo.image !== null,\n  excel_binary_key: fileInfo.excel ? fileInfo.excel.key : null,\n  conversion_type: convType,\n  conversion_binary_key: convKey,\n  image_binary_key: fileInfo.image ? fileInfo.image.key : null,\n  _email_type: emailType\n};\n\nreturn {\n  json: output,\n  binary: item.binary\n};"
      },
      "typeVersion": 2
    },
    {
      "id": "64aef15e-2374-4845-acc6-0dfe45d35c8f",
      "name": "IF - Has Excel?",
      "type": "n8n-nodes-base.if",
      "position": [
        4144,
        1536
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 1,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "operator": {
                "type": "boolean",
                "operation": "equals"
              },
              "leftValue": "={{ $json.has_excel }}",
              "rightValue": true
            }
          ]
        }
      },
      "typeVersion": 2
    },
    {
      "id": "7c133bed-216b-49fe-a6d9-8e7292391f87",
      "name": "Extract from Excel",
      "type": "n8n-nodes-base.extractFromFile",
      "position": [
        4368,
        1344
      ],
      "parameters": {
        "options": {},
        "operation": "xlsx",
        "binaryPropertyName": "={{ $json.excel_binary_key || 'attachment_0' }}"
      },
      "typeVersion": 1
    },
    {
      "id": "ae047edd-7921-47cd-84c1-98f9fd5b3ae2",
      "name": "Parse Excel Data",
      "type": "n8n-nodes-base.code",
      "position": [
        5008,
        1216
      ],
      "parameters": {
        "jsCode": "// ============================================================\n// EXCEL DATA EXTRACTION\n// ============================================================\n// Customize this node to parse your specific Excel format\n// ============================================================\n\n// Map codes to full names (customize for your use case)\nconst codeToNameMap = {\n  // 'CODE1': 'Full Name 1',\n  // 'CODE2': 'Full Name 2',\n};\n\nfunction excelTimeToHHMM(fraction) {\n  if (!fraction && fraction !== 0) return '';\n  const totalMinutes = Math.round(fraction * 24 * 60);\n  const hours = Math.floor(totalMinutes / 60);\n  const minutes = totalMinutes % 60;\n  return String(hours).padStart(2, '0') + ':' + String(minutes).padStart(2, '0');\n}\n\nlet excelSourceItems = [];\ntry { excelSourceItems = $('IF - Has Excel?').all(); } catch(e) {}\n\nconst allRows = $input.all().map(i => i.json);\nconst results = [];\n\nlet hintYear = new Date().getFullYear();\nlet hintMonth = new Date().getMonth() + 1;\nlet currentName = '';\nlet currentType = 'Regular';\n\nconst dataBySource = {};\n\nfor (const row of allRows) {\n  const allText = JSON.stringify(row);\n  const firstKey = Object.keys(row)[0] || '';\n  const firstVal = row[firstKey] || '';\n  \n  // Extract year/month from text (customize pattern as needed)\n  const ymMatch = allText.match(/(20\\d{2})[-\\/\u5e74]\\s*(\\d{1,2})/);\n  if (ymMatch) {\n    hintYear = parseInt(ymMatch[1]);\n    hintMonth = parseInt(ymMatch[2]);\n  }\n  \n  // Extract source name from brackets (customize pattern as needed)\n  const nameMatch = allText.match(/\u3010([A-Z]+)[^\u3011]*\u3011/i);\n  if (nameMatch) {\n    const code = nameMatch[1].toUpperCase();\n    currentName = codeToNameMap[code] || code;\n  }\n  \n  // Check for codes in the text\n  for (const code of Object.keys(codeToNameMap)) {\n    if (allText.toUpperCase().includes(code)) {\n      currentName = codeToNameMap[code] || code;\n      break;\n    }\n  }\n  \n  // Skip header rows (customize as needed)\n  if (allText.includes('Header1') && allText.includes('Header2')) continue;\n\n  // ============================================================\n  // CUSTOMIZE: Add your Excel parsing logic here\n  // ============================================================\n  \n  // Example: Parse rows with specific structure\n  if (row.__EMPTY_1 !== undefined && typeof row.__EMPTY_1 === 'number') {\n    const startTime = typeof row.__EMPTY_2 === 'number'\n      ? excelTimeToHHMM(row.__EMPTY_2)\n      : excelTimeToHHMM(row.__EMPTY_1);\n    const endTime = typeof row.__EMPTY_4 === 'number'\n      ? excelTimeToHHMM(row.__EMPTY_4)\n      : excelTimeToHHMM(row.__EMPTY_2);\n\n    if (startTime) {\n      const sourceName = currentName || 'Unknown';\n      const amount = typeof row.__EMPTY_7 === 'number' ? row.__EMPTY_7 : (row.__EMPTY_5 || 0);\n      const dataItem = {\n        date: hintYear + '-' + String(hintMonth).padStart(2,'0') + '-' + String(row.__EMPTY_1).padStart(2,'0'),\n        day_of_week: firstVal,\n        start_time: startTime,\n        end_time: endTime,\n        amount: amount,\n        type: currentType,\n        name: '',\n        source_name: sourceName\n      };\n      if (!dataBySource[sourceName]) dataBySource[sourceName] = [];\n      dataBySource[sourceName].push(dataItem);\n    }\n  }\n}\n\n// Build results\nfor (const srcItem of excelSourceItems) {\n  const meta = srcItem.json;\n  const emailId = meta.id || '';\n  const emailType = meta._email_type || 'CategoryC'; \n  \n  let defaultSource = '';\n  const subjectText = (meta.Subject || meta.subject || '');\n  const sourceMatch = subjectText.match(/\u3010([A-Za-z]+)\u3011/i);\n  if (sourceMatch) defaultSource = codeToNameMap[sourceMatch[1].toUpperCase()] || sourceMatch[1].toUpperCase();\n\n  let dataForEmail = dataBySource[defaultSource] || [];\n\n  if (!defaultSource && dataForEmail.length == 0) {\n    const allSources = Object.keys(dataBySource);\n    if (allSources.length === 1) {\n      dataForEmail = dataBySource[allSources[0]] || [];\n    } else if (allSources.length > 1) {\n      dataForEmail = allSources.flatMap(name => dataBySource[name] || []);\n    }\n  }\n\n  results.push({\n    json: {\n      _meta_subject: meta.Subject || meta.subject,\n      _meta_from: meta.from || meta.From,\n      _meta_id: emailId,\n      email_id: emailId, \n      _meta_source_name: defaultSource,\n      _email_type: emailType,\n      choices: [{ message: { content: JSON.stringify(dataForEmail.length ? dataForEmail : [{notes: '\u26a0\ufe0fNo data found'}]) } }]\n    }\n  });\n}\n\nreturn results.length ? results : [{ json: { _no_data: true } }];"
      },
      "typeVersion": 2
    },
    {
      "id": "edbfd491-dbab-411d-a3d2-c674ee34020d",
      "name": "IF - Has PPTX/PDF?",
      "type": "n8n-nodes-base.if",
      "position": [
        4368,
        1536
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 1,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "operator": {
                "type": "boolean",
                "operation": "equals"
              },
              "leftValue": "={{ $json.has_pptx_pdf }}",
              "rightValue": true
            }
          ]
        }
      },
      "typeVersion": 2
    },
    {
      "id": "49b24d00-f787-4b0d-8162-d41f62f6ca0d",
      "name": "ConvertAPI to PNG",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        4592,
        1488
      ],
      "parameters": {
        "url": "=https://v2.convertapi.com/convert/{{ $json.conversion_type }}/to/png",
        "method": "POST",
        "options": {},
        "sendBody": true,
        "contentType": "multipart-form-data",
        "authentication": "genericCredentialType",
        "bodyParameters": {
          "parameters": [
            {
              "name": "File",
              "parameterType": "formBinaryData",
              "inputDataFieldName": "={{ $json.conversion_binary_key }}"
            }
          ]
        },
        "genericAuthType": "httpHeaderAuth"
      },
      "typeVersion": 4.2
    },
    {
      "id": "bdd00767-8b17-4d7f-bc38-588ba5a923b9",
      "name": "Prepare for S3 Upload",
      "type": "n8n-nodes-base.code",
      "position": [
        4800,
        1488
      ],
      "parameters": {
        "jsCode": "// ============================================================\n// PREPARE IMAGES FOR S3 UPLOAD\n// ============================================================\n\nconst allItems = $input.all();\nconst results = [];\n\nlet pptxPdfItems = [];\ntry {\n  const allFileItems = $('Detect File Types').all();\n  pptxPdfItems = allFileItems.filter(item => item.json.has_pptx_pdf === true);\n} catch(e) {}\n\nlet filterItems = [];\ntry { filterItems = $('Filter Valid Emails').all(); } catch(e) {}\n\nfor (let idx = 0; idx < allItems.length; idx++) {\n  const item = allItems[idx];\n  const response = item.json;\n  \n  let emailId='', subject='', fromData='', attachments=[], emailType='';\n  \n  if (pptxPdfItems[idx]) {\n    const meta = pptxPdfItems[idx].json;\n    emailId = meta.id || '';\n    subject = meta.Subject || '';\n    fromData = meta.From || '';\n    attachments = meta.attachments || [];\n    emailType = meta._email_type || '';\n  }\n  \n  if (!emailType && emailId) {\n    for (const f of filterItems) {\n      if (f.json.id === emailId && f.json._email_type) {\n        emailType = f.json._email_type;\n        break;\n      }\n    }\n  }\n  emailType = emailType || 'CategoryC';\n  \n  if (response.Files && response.Files.length > 0) {\n    const maxPages = Math.min(response.Files.length, 15);\n    const ts = Date.now();\n    \n    for (let i = 0; i < maxPages; i++) {\n      const fileData = response.Files[i];\n      if (fileData && fileData.FileData) {\n        const fileName = 'convert-' + emailId + '-page-' + (i+1) + '-' + ts + '.png';\n        results.push({\n          json: {\n            _meta_subject: subject,\n            _meta_from: fromData,\n            _meta_id: emailId,\n            _meta_attachments: attachments,\n            _meta_process_type: 'pptx_pdf',\n            _email_type: emailType,\n            _page_index: i,\n            _page_filename: fileName,\n            _total_pages: maxPages\n          },\n          binary: {\n            data: {\n              data: fileData.FileData,\n              mimeType: 'image/png',\n              fileName: fileName,\n              fileExtension: 'png'\n            }\n          }\n        });\n      }\n    }\n  }\n}\n\nif (results.length === 0) return [{ json: { _skip: true } }];\nreturn results;"
      },
      "typeVersion": 2
    },
    {
      "id": "9a844a79-cb50-439f-abcf-ea9ef20d3510",
      "name": "S3 Upload (PPTX)",
      "type": "n8n-nodes-base.awsS3",
      "position": [
        5024,
        1488
      ],
      "parameters": {
        "fileName": "={{ $json._page_filename }}",
        "operation": "upload",
        "bucketName": "={{ $vars.S3_BUCKET_NAME || 'your-bucket-name' }}",
        "additionalFields": {}
      },
      "typeVersion": 2
    },
    {
      "id": "22f24843-db51-4bc5-aac7-bf9cf8034222",
      "name": "Collect S3 URLs",
      "type": "n8n-nodes-base.code",
      "position": [
        5248,
        1488
      ],
      "parameters": {
        "jsCode": "// ============================================================\n// COLLECT S3 URLs FOR GPT VISION\n// ============================================================\n\nlet prepItems = [];\ntry { prepItems = $('Prepare for S3 Upload').all(); } catch(e) {}\n\nconst bucket = $vars.S3_BUCKET_NAME || 'your-bucket-name';\nconst region = $vars.AWS_REGION || 'us-east-1';\nconst emailGroups = new Map();\n\nfor (const prep of prepItems) {\n  const p = prep.json;\n  const emailId = p._meta_id || '';\n  const filename = p._page_filename || '';\n  if (!emailId || !filename) continue;\n  \n  const s3Url = 'https://' + bucket + '.s3.' + region + '.amazonaws.com/' + filename;\n  \n  if (!emailGroups.has(emailId)) {\n    emailGroups.set(emailId, {\n      subject: p._meta_subject || '',\n      from: p._meta_from || '',\n      emailId: emailId,\n      attachments: p._meta_attachments || [],\n      emailType: p._email_type || 'CategoryC',\n      sourceName: p._meta_source_name || '',\n      urls: []\n    });\n  }\n  emailGroups.get(emailId).urls.push({ url: s3Url, index: p._page_index || 0 });\n}\n\nconst results = [];\nfor (const [emailId, data] of emailGroups) {\n  data.urls.sort((a, b) => a.index - b.index);\n  results.push({\n    json: {\n      _meta_subject: data.subject,\n      _meta_from: data.from,\n      _meta_id: data.emailId,\n      _meta_attachments: data.attachments,\n      _meta_source_name: data.sourceName,\n      _email_type: data.emailType,\n      _s3_urls: data.urls.map(u => u.url)\n    }\n  });\n}\n\nif (results.length === 0) return [{ json: { _skip: true } }];\nreturn results;"
      },
      "typeVersion": 2
    },
    {
      "id": "f92643b3-d9a2-4053-b2ca-8f76294e0e67",
      "name": "Prepare GPT Request (PPTX)",
      "type": "n8n-nodes-base.code",
      "position": [
        5472,
        1488
      ],
      "parameters": {
        "mode": "runOnceForEachItem",
        "jsCode": "// ============================================================\n// PREPARE GPT VISION REQUEST FOR PPTX/PDF\n// ============================================================\n// Customize the system prompt and extraction fields\n// ============================================================\n\n// Map codes to full names (customize for your use case)\nconst codeToNameMap = {\n  // 'CODE1': 'Full Name 1',\n  // 'CODE2': 'Full Name 2',\n};\n\nconst item = $input.item.json;\nconst subject = item._meta_subject || '';\nconst emailId = item._meta_id || '';\nconst s3Urls = item._s3_urls || [];\nconst emailType = item._email_type || 'CategoryC';\n\n// Extract source name from subject\nconst sourceMatch = subject.match(/\u3010([A-Za-z]+)\u3011/i);\nconst sourceCode = sourceMatch ? sourceMatch[1].toUpperCase() : '';\nconst sourceName = codeToNameMap[sourceCode] || sourceCode;\n\n// Build prompt based on email type\nlet prompt = '';\nif (emailType === 'CategoryA') {\n  prompt = 'Subject: ' + subject + '\\n\\nThis is a CANCELLATION notice. Extract cancellation details.';\n} else if (emailType === 'CategoryB') {\n  prompt = 'Subject: ' + subject + '\\n\\nThis is an UPDATE notice. Extract the changes.';\n} else {\n  prompt = 'Subject: ' + subject + '\\n\\nExtract all relevant information from this image.';\n}\n\nlet userContent = [{ type: 'text', text: prompt }];\nfor (const url of s3Urls) {\n  userContent.push({ type: 'image_url', image_url: { url: url, detail: 'high' } });\n}\n\n// ============================================================\n// CUSTOMIZE: Update the system prompt and fields for your use case\n// ============================================================\nconst systemPrompt = `You are a data extraction expert. Extract structured information accurately from images.\n\n\u3010Fields to extract\u3011\n- date (YYYY-MM-DD format)\n- day_of_week\n- start_time (HH:MM format)\n- end_time (HH:MM format)\n- source_code\n- name\n- amount (numbers only)\n- category\n- type\n- notes\n\n\u3010Output\u3011JSON array only, no explanation`;\n\nconst gptBody = {\n  model: 'gpt-4o',\n  messages: [\n    { role: 'system', content: systemPrompt },\n    { role: 'user', content: userContent }\n  ],\n  max_tokens: 4096,\n  temperature: 0.1\n};\n\nreturn {\n  json: {\n    _meta_subject: subject,\n    _meta_from: item._meta_from,\n    _meta_id: emailId,\n    _meta_attachments: item._meta_attachments || [],\n    _meta_process_type: 'pptx_pdf',\n    _meta_source_name: sourceName,\n    _email_type: emailType,\n    gpt_request_body: JSON.stringify(gptBody)\n  }\n};"
      },
      "typeVersion": 2
    },
    {
      "id": "eeb04c4a-6278-4f95-b7a1-270ea8052e46",
      "name": "GPT Vision Analysis",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        5664,
        1264
      ],
      "parameters": {
        "url": "https://api.openai.com/v1/chat/completions",
        "method": "POST",
        "options": {
          "timeout": 300000
        },
        "jsonBody": "={{ $json.gpt_request_body }}",
        "sendBody": true,
        "specifyBody": "json",
        "authentication": "predefinedCredentialType",
        "nodeCredentialType": "openAiApi"
      },
      "typeVersion": 4.2
    },
    {
      "id": "59b8220a-6ea1-42d6-88c4-91225668facf",
      "name": "IF - Has Image?",
      "type": "n8n-nodes-base.if",
      "position": [
        4592,
        1648
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 1,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "operator": {
                "type": "boolean",
                "operation": "equals"
              },
              "leftValue": "={{ $json.has_image }}",
              "rightValue": true
            }
          ]
        }
      },
      "typeVersion": 2
    },
    {
      "id": "68b29f63-ef7c-437a-806b-19ae1e9e1be3",
      "name": "Prepare Image Metadata",
      "type": "n8n-nodes-base.code",
      "position": [
        4800,
        1616
      ],
      "parameters": {
        "mode": "runOnceForEachItem",
        "jsCode": "// ============================================================\n// PREPARE IMAGE METADATA FOR S3 UPLOAD\n// ============================================================\n\nconst item = $input.item;\nconst json = item.json;\nconst subject = json.Subject || '';\nconst emailId = json.id || 'unknown';\n\nlet emailType = json._email_type || '';\nif (!emailType && emailId) {\n  try {\n    const filteredItems = $('Filter Valid Emails').all();\n    for (const f of filteredItems) {\n      if (f.json.id === emailId && f.json._email_type) {\n        emailType = f.json._email_type;\n        break;\n      }\n    }\n  } catch(e) {}\n}\nemailType = emailType || 'CategoryC';\n\nconst timestamp = Date.now();\nconst fileName = 'email-image-' + emailId + '-' + timestamp + '.png';\n\nreturn {\n  json: {\n    _meta_subject: subject,\n    _meta_from: json.From,\n    _meta_id: json.id,\n    _meta_attachments: json.attachments || [],\n    _meta_process_type: 'image',\n    _email_type: emailType,\n    _image_key: json.image_binary_key || 'attachment_0',\n    _s3_filename: fileName\n  },\n  binary: item.binary\n};"
      },
      "typeVersion": 2
    },
    {
      "id": "45f24d85-b9d3-4532-997c-55d6036aaa4a",
      "name": "S3 Upload (Image)",
      "type": "n8n-nodes-base.awsS3",
      "position": [
        5024,
        1648
      ],
      "parameters": {
        "fileName": "={{ $json._s3_filename }}",
        "operation": "upload",
        "bucketName": "={{ $vars.S3_BUCKET_NAME || 'your-bucket-name' }}",
        "additionalFields": {},
        "binaryPropertyName": "={{ $json._image_key }}"
      },
      "typeVersion": 2
    },
    {
      "id": "e6c8c547-bd98-4e4e-9c7f-8efb515349f1",
      "name": "Prepare GPT Request (Image)",
      "type": "n8n-nodes-base.code",
      "position": [
        5248,
        1648
      ],
      "parameters": {
        "mode": "runOnceForEachItem",
        "jsCode": "// ============================================================\n// PREPARE GPT VISION REQUEST FOR IMAGE ATTACHMENTS\n// ============================================================\n\nconst codeToNameMap = {\n  // 'CODE1': 'Full Name 1',\n};\n\nconst item = $input.item.json;\nlet meta = {};\ntry { meta = $('Prepare Image Metadata').first().json; } catch(e) {}\n\nconst subject = meta._meta_subject || '';\nconst emailId = meta._meta_id || '';\nconst emailType = meta._email_type || 'CategoryC';\n\nconst sourceMatch = subject.match(/\u3010([A-Za-z]+)\u3011/i);\nconst sourceCode = sourceMatch ? sourceMatch[1].toUpperCase() : '';\nconst sourceName = codeToNameMap[sourceCode] || sourceCode;\n\nconst imageUrl = item.Location || '';\nif (!imageUrl) return { json: { _skip: true } };\n\nlet prompt = '';\nif (emailType === 'CategoryA') {\n  prompt = 'Subject: ' + subject + '\\n\\nThis is a CANCELLATION notice.';\n} else if (emailType === 'CategoryB') {\n  prompt = 'Subject: ' + subject + '\\n\\nThis is an UPDATE notice.';\n} else {\n  prompt = 'Subject: ' + subject + '\\n\\nExtract all relevant data from this image.';\n}\n\nconst gptBody = {\n  model: 'gpt-4o',\n  messages: [\n    { role: 'system', content: 'You are a data extraction expert. Extract structured information from images.\\n\\n\u3010Fields\u3011date(YYYY-MM-DD), day_of_week, start_time(HH:MM), end_time(HH:MM), source_code, name, amount(numbers only), category, type, notes\\n\\n\u3010Output\u3011JSON array only' },\n    { role: 'user', content: [{ type: 'text', text: prompt }, { type: 'image_url', image_url: { url: imageUrl, detail: 'high' } }] }\n  ],\n  max_tokens: 4096\n};\n\nreturn {\n  json: {\n    _meta_subject: subject,\n    _meta_from: meta._meta_from || '',\n    _meta_id: emailId,\n    _meta_attachments: meta._meta_attachments || [],\n    _meta_process_type: 'image',\n    _meta_source_name: sourceName,\n    _email_type: emailType,\n    gpt_request_body: JSON.stringify(gptBody)\n  }\n};"
      },
      "typeVersion": 2
    },
    {
      "id": "16937348-70e9-46da-89eb-9ab047956476",
      "name": "GPT Image Analysis",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        5472,
        1760
      ],
      "parameters": {
        "url": "https://api.openai.com/v1/chat/completions",
        "method": "POST",
        "options": {},
        "jsonBody": "={{ $json.gpt_request_body }}",
        "sendBody": true,
        "specifyBody": "json",
        "authentication": "predefinedCredentialType",
        "nodeCredentialType": "openAiApi"
      },
      "typeVersion": 4.2
    },
    {
      "id": "3b289245-1499-4dd3-849f-2b39cf04f65a",
      "name": "Prepare GPT Request (Text)",
      "type": "n8n-nodes-base.code",
      "position": [
        4800,
        1840
      ],
      "parameters": {
        "mode": "runOnceForEachItem",
        "jsCode": "// ============================================================\n// PREPARE GPT REQUEST FOR TEXT-ONLY EMAILS\n// ============================================================\n\nconst item = $input.item.json;\nconst subject = item.Subject || '';\nconst emailId = item.id || '';\n\nlet emailType = item._email_type || '';\nif (!emailType && emailId) {\n  try {\n    const filteredItems = $('Filter Valid Emails').all();\n    for (const f of filteredItems) {\n      if (f.json.id === emailId && f.json._email_type) {\n        emailType = f.json._email_type;\n        break;\n      }\n    }\n  } catch(e) {}\n}\nemailType = emailType || 'CategoryC';\n\nconst textBody = item.textPlain || item.text || '';\nlet userPrompt = 'Subject: ' + subject + '\\n\\nBody:\\n' + textBody.substring(0, 3000);\nif (emailType === 'CategoryA') userPrompt = '\u3010CANCELLATION\u3011' + userPrompt;\nelse if (emailType === 'CategoryB') userPrompt = '\u3010UPDATE\u3011' + userPrompt;\n\nconst gptBody = {\n  model: 'gpt-4o-mini',\n  messages: [\n    { role: 'system', content: 'You are a data extraction expert. Extract structured information from email text.\\n\\n\u3010Fields\u3011date, day_of_week, start_time, end_time, source_code, name, amount, category, type, notes\\n\\n\u3010Output\u3011JSON array only' },\n    { role: 'user', content: userPrompt }\n  ],\n  max_tokens: 2048\n};\n\nreturn {\n  json: {\n    _meta_subject: subject,\n    _meta_from: item.From,\n    _meta_id: emailId,\n    _meta_attachments: item.attachments || [],\n    _meta_process_type: 'text',\n    _email_type: emailType,\n    gpt_request_body: JSON.stringify(gptBody)\n  }\n};"
      },
      "typeVersion": 2
    },
    {
      "id": "27d9b48b-9762-43e5-b624-f810651664e3",
      "name": "GPT Text Analysis",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        5024,
        1840
      ],
      "parameters": {
        "url": "https://api.openai.com/v1/chat/completions",
        "method": "POST",
        "options": {},
        "jsonBody": "={{ $json.gpt_request_body }}",
        "sendBody": true,
        "specifyBody": "json",
        "authentication": "predefinedCredentialType",
        "nodeCredentialType": "openAiApi"
      },
      "typeVersion": 4.2
    },
    {
      "id": "f0ac5f4d-bf3b-4a45-ab48-b1cf17f43ce1",
      "name": "Merge All Paths",
      "type": "n8n-nodes-base.merge",
      "position": [
        5904,
        1536
      ],
      "parameters": {},
      "typeVersion": 3,
      "alwaysOutputData": true
    },
    {
      "id": "0a1f3790-23fa-4b48-a1c6-857c88758b86",
      "name": "Parse All Results",
      "type": "n8n-nodes-base.code",
      "position": [
        6128,
        1536
      ],
      "parameters": {
        "jsCode": "// ============================================================\n// PARSE ALL GPT RESPONSES AND FORMAT FOR OUTPUT\n// ============================================================\n// Customize the output schema for your Google Sheets\n// ============================================================\n\n// Map codes to full names (customize for your use case)\nconst codeToNameMap = {\n  // 'CODE1': 'Full Name 1',\n  // 'CODE2': 'Full Name 2',\n};\n\nconst allItems = $input.all();\nconst results = [];\n\nlet pptxPrepItems = [], imagePrepItems = [], textPrepItems = [];\ntry { pptxPrepItems = $('Prepare GPT Request (PPTX)').all(); } catch(e) {}\ntry { imagePrepItems = $('Prepare GPT Request (Image)').all(); } catch(e) {}\ntry { textPrepItems = $('Prepare GPT Request (Text)').all(); } catch(e) {}\n\nlet pptxPrepIndex = 0, imagePrepIndex = 0, textPrepIndex = 0;\n\nfor (let i = 0; i < allItems.length; i++) {\n  const item = allItems[i];\n  const data = item.json;\n  \n  const hasMetadata = !!data._meta_id;\n  let subject, from, emailId, attachments, sourceName, emailType;\n  \n  if (hasMetadata) {\n    subject = data._meta_subject || '';\n    from = data._meta_from || '';\n    emailId = data._meta_id || '';\n    attachments = data._meta_attachments || [];\n    sourceName = data._meta_source_name || '';\n    emailType = data._email_type || 'CategoryC';\n  } else {\n    const modelName = data.model || '';\n    const isTextModel = modelName.includes('gpt-4o-mini');\n    let meta = null;\n    if (isTextModel && textPrepIndex < textPrepItems.length) {\n      meta = textPrepItems[textPrepIndex].json;\n      textPrepIndex++;\n    } else if (imagePrepIndex < imagePrepItems.length) {\n      meta = imagePrepItems[imagePrepIndex].json;\n      imagePrepIndex++;\n    } else if (pptxPrepIndex < pptxPrepItems.length) {\n      meta = pptxPrepItems[pptxPrepIndex].json;\n      pptxPrepIndex++;\n    }\n\n    if (meta) {\n      subject = meta._meta_subject || '';\n      from = meta._meta_from || '';\n      emailId = meta._meta_id || '';\n      attachments = meta._meta_attachments || [];\n      sourceName = meta._meta_source_name || '';\n      emailType = meta._email_type || 'CategoryC';\n    } else {\n      subject = ''; from = ''; emailId = ''; attachments = []; sourceName = ''; emailType = 'CategoryC';\n    }\n  }\n  \n  if (from && typeof from === 'object') {\n    from = from.text || from.address || (from.value && from.value[0] ? from.value[0].address : '') || '';\n  }\n  from = String(from || '');\n  \n  let content = '';\n  if (data.choices && data.choices[0] && data.choices[0].message) {\n    content = data.choices[0].message.content || '';\n  }\n  \n  // Check for unreadable content\n  const unreadablePatterns = ['sorry', 'cannot read', 'unable to', 'not able to'];\n  const isUnreadable = unreadablePatterns.some(p => content.toLowerCase().includes(p));\n  \n  if (isUnreadable) {\n    results.push({\n      json: {\n        email_type: emailType,\n        date: '', day_of_week: '', start_time: '', end_time: '',\n        source_name: sourceName, name: '', amount: '0', category: '', type: '',\n        status: 'Needs Review', client: '',\n        notes: '\u26a0\ufe0fFile unreadable - manual review required',\n        created_at: new Date().toISOString(),\n        email_subject: subject, sender: from, email_id: emailId,\n        attachments: attachments.map(a => a.filename || '').join(', '),\n        _unreadable: true\n      }\n    });\n    continue;\n  }\n  \n  // Parse JSON from GPT response\n  let items = [];\n  try {\n    let jsonStr = content;\n    const codeMatch = content.match(/```json\\s*([\\s\\S]*?)\\s*```/);\n    if (codeMatch) jsonStr = codeMatch[1];\n    else {\n      const arrMatch = content.match(/\\[[\\s\\S]*\\]/);\n      if (arrMatch) jsonStr = arrMatch[0];\n    }\n    const parsed = JSON.parse(jsonStr);\n    items = Array.isArray(parsed) ? parsed : [parsed];\n  } catch(e) { \n    if (emailId) {\n      results.push({\n        json: {\n          email_type: emailType,\n          date: '', day_of_week: '', start_time: '', end_time: '',\n          source_name: sourceName, name: '', amount: '0', category: '', type: '',\n          status: 'Needs Review', client: '',\n          notes: '\u26a0\ufe0fParse failed - manual review required',\n          created_at: new Date().toISOString(),\n          email_subject: subject, sender: from, email_id: emailId,\n          attachments: attachments.map(a => a.filename || '').join(', '),\n          _unreadable: true\n        }\n      });\n    }\n    continue;\n  }\n  \n  if (items.length === 0) continue;\n  \n  // ============================================================\n  // CUSTOMIZE: Map extracted data to your output schema\n  // ============================================================\n  for (const s of items) {\n    let sourceNameFinal = sourceName || s.source_name || '';\n    if (s.source_code) {\n      const code = s.source_code.toUpperCase();\n      sourceNameFinal = codeToNameMap[code] || sourceNameFinal || code;\n    }\n    \n    // Determine category from time (customize as needed)\n    let category = s.category || '';\n    if (!category && s.start_time) {\n      const h = parseInt(s.start_time.split(':')[0]) || 0;\n      if (h >= 5 && h < 10) category = 'Morning';\n      else if (h >= 10 && h < 15) category = 'Afternoon';\n      else if (h >= 15 && h < 23) category = 'Evening';\n      else category = 'Night';\n    }\n\n    results.push({\n      json: {\n        email_type: emailType,\n        date: s.date || '',\n        day_of_week: s.day_of_week || '',\n        start_time: s.start_time || '',\n        end_time: s.end_time || '',\n        source_name: sourceNameFinal,\n        name: s.name || '',\n        amount: String(s.amount || 0),\n        category: category,\n        type: s.type || 'Regular',\n        status: 'Pending',\n        client: '',\n        notes: s.notes || '',\n        created_at: new Date().toISOString(),\n        email_subject: subject,\n        sender: from,\n        email_id: emailId,\n        attachments: attachments.map(a => a.filename || '').join(', ')\n      }\n    });\n  }\n}\n\nif (results.length === 0) return [{ json: { _no_data: true } }];\nreturn results;"
      },
      "typeVersion": 2
    },
    {
      "id": "3b91edb4-d360-4441-ac58-bcaf1da85514",
      "name": "Filter Valid Data",
      "type": "n8n-nodes-base.filter",
      "position": [
        6352,
        1536
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 1,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "operator": {
                "type": "string",
                "operation": "notEmpty",
                "singleValue": true
              },
              "leftValue": "={{ $json[\"email_type\"] }}",
              "rightValue": true
            }
          ]
        }
      },
      "typeVersion": 2
    },
    {
      "id": "07942913-2997-42f5-a93f-b4459ca4c119",
      "name": "Save to Google Sheets",
      "type": "n8n-nodes-base.googleSheets",
      "position": [
        6560,
        1536
      ],
      "parameters": {
        "columns": {
          "value": {},
          "mappingMode": "autoMapInputData"
        },
        "options": {},
        "operation": "append",
        "sheetName": {
          "__rl": true,
          "mode": "name",
          "value": "Sheet1"
        },
        "documentId": {
          "__rl": true,
          "mode": "id",
          "value": "={{ $vars.GOOGLE_SHEET_ID || 'your-google-sheet-id' }}"
        }
      },
      "typeVersion": 4.4
    },
    {
      "id": "93650748-aff3-423d-afa6-48dd53b5230b",
      "name": "Extract Email IDs",
      "type": "n8n-nodes-base.code",
      "position": [
        6784,
        1536
      ],
      "parameters": {
        "jsCode": "// ============================================================\n// EXTRACT UNIQUE EMAIL IDs FOR LABELING\n// ============================================================\n\nconst allItems = $input.all();\nconst emailMap = new Map();\n\nfunction addEmail(emailId, emailType, isUnreadable) {\n  if (!emailId) return;\n  if (!emailMap.has(emailId)) {\n    emailMap.set(emailId, { type: isUnreadable ? 'NeedsReview' : emailType });\n  }\n  if (isUnreadable) emailMap.get(emailId).type = 'NeedsReview';\n}\n\nfor (const item of allItems) {\n  const emailId = item.json.email_id;\n  const emailType = item.json.email_type || 'CategoryC';\n  const isUnreadable = item.json._unreadable === true;\n  addEmail(emailId, emailType, isUnreadable);\n}\n\ntry {\n  const filteredItems = $('Filter Valid Emails').all();\n  for (const f of filteredItems) {\n    addEmail(f.json.id, f.json._email_type || 'CategoryC', false);\n  }\n} catch (e) {}\n\nreturn Array.from(emailMap.entries()).map(([id, data]) => ({\n  json: { email_id: id, email_type: data.type }\n}));"
      },
      "typeVersion": 2
    },
    {
      "id": "613a3c8a-e875-48f5-b75b-1b121b92f173",
      "name": "Get Gmail Labels",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        6784,
        1760
      ],
      "parameters": {
        "url": "https://gmail.googleapis.com/gmail/v1/users/me/labels",
        "options": {},
        "authentication": "predefinedCredentialType",
        "nodeCredentialType": "gmailOAuth2"
      },
      "typeVersion": 4.2
    },
    {
      "id": "dab2e312-225d-4df4-a45d-4b1ab1a37558",
      "name": "Check Labels",
      "type": "n8n-nodes-base.code",
      "position": [
        7008,
        1760
      ],
      "parameters": {
        "jsCode": "// ============================================================\n// CHECK WHICH LABELS EXIST\n// ============================================================\n// Customize label names for your use case\n// ============================================================\n\nconst response = $input.first().json;\nconst labels = response.labels || [];\n\n// Define your label names here\nconst labelNames = [\n  'Processed_CategoryA',\n  'Processed_CategoryB', \n  'Processed_CategoryC'\n];\n\nconst existingLabels = {};\nconst missingLabels = [];\n\nfor (const name of labelNames) {\n  const found = labels.find(l => l.name === name);\n  if (found) existingLabels[name] = found.id;\n  else missingLabels.push(name);\n}\n\nreturn [{ \n  json: { \n    existing_labels: existingLabels, \n    missing_labels: missingLabels, \n    needs_creation: missingLabels.length > 0 \n  } \n}];"
      },
      "typeVersion": 2
    },
    {
      "id": "f169d45b-e952-4df5-bd93-7e61a4f07980",
      "name": "IF - Need Labels?",
      "type": "n8n-nodes-base.if",
      "position": [
        7232,
        1760
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 1,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "operator": {
                "type": "boolean",
                "operation": "equals"
              },
              "leftValue": "={{ $json.needs_creation }}",
              "rightValue": true
            }
          ]
        }
      },
      "typeVersion": 2
    },
    {
      "id": "137c484a-e4ce-4996-b603-4f2164309dc3",
      "name": "Prepare Label Creation",
      "type": "n8n-nodes-base.code",
      "position": [
        7440,
        1664
      ],
      "parameters": {
        "jsCode": "// Prepare label creation requests\nconst items = $input.all();\nconst results = [];\nfor (const item of items) {\n  const data = item.json || {};\n  const missingLabels = data.missing_labels || [];\n  const existingLabels = data.existing_labels || {};\n  if (missingLabels.length === 0) results.push({ json: { label_name: '', existing_labels: existingLabels, _skip_creation: true } });\n  else for (const name of missingLabels) results.push({ json: { label_name: name, existing_labels: existingLabels, _skip_creation: false } });\n}\nif (results.length === 0) return [{ json: { label_name: '', existing_labels: {}, _skip_creation: true } }];\nreturn results;"
      },
      "typeVersion": 2
    },
    {
      "id": "28717cfd-d0dc-4040-b3fa-dd2982e05819",
      "name": "Filter Labels to Create",
      "type": "n8n-nodes-base.filter",
      "position": [
        7664,
        1664
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 1,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "operator": {
                "type": "boolean",
                "operation": "notEquals"
              },
              "leftValue": "={{ $json._skip_creation }}",
              "rightValue": true
            }
          ]
        }
      },
      "typeVersion": 2
    },
    {
      "id": "5f7f79dc-0cb8-43b8-b938-6539881c02d8",
      "name": "Create Gmail Label",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        7888,
        1664
      ],
      "parameters": {
        "url": "https://gmail.googleapis.com/gmail/v1/users/me/labels",
        "method": "POST",
        "options": {},
        "jsonBody": "={\"name\": \"{{ $json.label_name }}\", \"labelListVisibility\": \"labelShow\", \"messageListVisibility\": \"show\"}",
        "sendBody": true,
        "specifyBody": "json",
        "authentication": "predefinedCredentialType",
        "nodeCredentialType": "gmailOAuth2"
      },
      "typeVersion": 4.2
    },
    {
      "id": "13873411-e770-4f62-a957-f0dd3690b26d",
      "name": "Get New Label IDs",
      "type": "n8n-nodes-base.code",
      "position": [
        8112,
        1664
      ],
      "parameters": {
        "jsCode": "// Collect newly created label IDs\nconst allItems = $input.all();\nconst createdLabels = {};\nfor (const item of allItems) {\n  const data = item.json;\n  if (data.id && data.name) createdLabels[data.name] = data.id;\n}\nlet existingLabels = {};\ntry { const labelCheckData = $('Check Labels').first().json; existingLabels = labelCheckData.existing_labels || {}; } catch(e) {}\nreturn [{ json: { all_labels: { ...existingLabels, ...createdLabels } } }];"
      },
      "typeVersion": 2
    },
    {
      "id": "8ca950ac-7f74-4596-adf7-ca74e6b191d3",
      "name": "Pass Existing Labels",
      "type": "n8n-nodes-base.code",
      "position": [
        7440,
        1856
      ],
      "parameters": {
        "jsCode": "// Pass through existing labels when no creation needed\nconst items = $input.all();\nlet existingLabels = {};\nfor (const item of items) {\n  const data = item.json || {};\n  if (data.existing_labels) existingLabels = { ...existingLabels, ...data.existing_labels };\n}\nreturn [{ json: { all_labels: existingLabels } }];"
      },
      "typeVersion": 2
    },
    {
      "id": "bebcee89-dc11-4bab-8a9c-e72744778b07",
      "name": "Merge Label IDs",
      "type": "n8n-nodes-base.code",
      "position": [
        8112,
        1856
      ],
      "parameters": {
        "jsCode": "// Merge all label IDs\nconst items = $input.all();\nlet allLabels = {};\nfor (const item of items) {\n  const data = item.json || {};\n  const labels = data.all_labels || data.existing_labels || {};\n  allLabels = { ...allLabels, ...labels };\n}\nreturn [{ json: { all_labels: allLabels } }];"
      },
      "typeVersion": 2
    },
    {
      "id": "ec6d8a00-ade2-4047-84fa-2f448cceb979",
      "name": "Prepare Label Application",
      "type": "n8n-nodes-base.code",
      "position": [
        8320,
        1776
      ],
      "parameters": {
        "jsCode": "// ============================================================\n// PREPARE LABEL APPLICATION FOR EACH EMAIL\n// ============================================================\n// Customize the type-to-label mapping\n// ============================================================\n\nconst inputItems = $input.all();\nconst labelInfo = inputItems[0]?.json?.all_labels\n  ? inputItems[0].json\n  : (() => {\n      try { return $('Merge Label IDs').first().json; } catch (e) { return {}; }\n    })();\n\nlet emailItems = [];\ntry { emailItems = $('Extract Email IDs').all(); } catch (e) { emailItems = []; }\nif (emailItems.length === 0) {\n  try {\n    const filteredItems = $('Filter Valid Emails').all();\n    emailItems = filteredItems\n      .map(item => ({\n        json: {\n          email_id: item.json.id || '',\n          email_type: item.json._email_type || 'CategoryC'\n        }\n      }))\n      .filter(item => item.json.email_id);\n  } catch (e) {\n    emailItems = [];\n  }\n}\n\nconst allLabels = labelInfo.all_labels || {};\n\n// Map email types to label names (customize these)\nconst typeToLabel = { \n  'CategoryA': 'Processed_CategoryA', \n  'CategoryB': 'Processed_CategoryB', \n  'CategoryC': 'Processed_CategoryC' \n};\n\nconst results = [];\n\nfor (const item of emailItems) {\n  const emailType = item.json.email_type || 'CategoryC';\n  const labelName = typeToLabel[emailType] || 'Processed_CategoryC';\n  const labelId = allLabels[labelName];\n  if (labelId) {\n    results.push({\n      json: {\n        email_id: item.json.email_id,\n        email_type: emailType,\n        label_name: labelName,\n        label_id: labelId,\n        request_body: JSON.stringify({ addLabelIds: [labelId] })\n      }\n    });\n  }\n}\n\nif (results.length === 0) return [{ json: { _no_emails: true } }];\nreturn results;"
      },
      "typeVersion": 2
    },
    {
      "id": "b6a78bf7-4974-47ee-8500-440c30991c39",
      "name": "Has Emails?",
      "type": "n8n-nodes-base.filter",
      "position": [
        8544,
        1776
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 1,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "operator": {
                "type": "boolean",
                "operation": "notEquals"
              },
              "leftValue": "={{ $json._no_emails }}",
              "rightValue": true
            }
          ]
        }
      },
      "typeVersion": 2
    },
    {
      "id": "f1e17c5b-24d1-4168-9b06-f24be083c53f",
      "name": "Apply Gmail Labels",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        8768,
        1776
      ],
      "parameters": {
        "url": "=https://gmail.googleapis.com/gmail/v1/users/me/messages/{{ $json.email_id }}/modify",
        "method": "POST",
        "options": {},
        "jsonBody": "={{ $json.request_body }}",
        "sendBody": true,
        "specifyBody": "json",
        "authentication": "predefinedCredentialType",
        "nodeCredentialType": "gmailOAuth2"
      },
      "typeVersion": 4.2
    },
    {
      "id": "config-variables-note",
      "name": "Configuration Variables",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        2384,
        2040
      ],
      "parameters": {
        "color": 4,
        "width": 500,
        "height": 300,
        "content": "###  Required Variables\n\nSet these in n8n Settings > Variables:\n\n| Variable | Description |\n|----------|-------------|\n| `S3_BUCKET_NAME` | Your AWS S3 bucket name |\n| `AWS_REGION` | AWS region (e.g., us-east-1) |\n| `GOOGLE_SHEET_ID` | Google Sheets document ID |\n\n### Required Credentials\n\n- **Gmail OAuth2** - For email access\n- **OpenAI API** - For GPT Vision\n- **AWS S3** - For image storage\n- **Google Sheets OAuth2** - For data output\n- **ConvertAPI HTTP Header Auth** - For file conversion"
      },
      "typeVersion": 1
    }
  ],
  "connections": {
    "Has Emails?": {
      "main": [
        [
          {
            "node": "Apply Gmail Labels",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Check Labels": {
      "main": [
        [
          {
            "node": "IF - Need Labels?",
            "type": "m

About this workflow

Automatically extract structured information from emails using AI-powered document analysis. This workflow processes emails from specified domains, classifies them by type, and extracts structured data from various attachment formats.

Source: https://n8n.io/workflows/13004/ — original creator credit. Request a take-down →

More Email & Gmail workflows → · Browse all categories →

Related workflows

Workflows that share integrations, category, or trigger type with this one. All free to copy and import.