Draft PR: Feature - Add JPG/PNG image support with OCR to document robots, Issue #1110

README.md

@@ -126,13 +126,14 @@ It supports extraction, crawling, scraping, and search — designed to scale fro
 4. **[Search](https://docs.maxun.dev/robot/search/search-introduction)** – Run automated web searches to discover or scrape results, with support for time-based filters.
 5. **[SDK](https://docs.maxun.dev/category/sdk)** – A complete developer toolkit for scraping, extraction, scheduling, and end-to-end data automation.
 6. **[CLI](https://docs.maxun.dev/category/cli)** – Create robots, trigger runs, and retrieve extracted data from your terminal.
+7. **[Document Extraction & Parsing](https://docs.maxun.dev/)** – Extract structured data from documents and images (PDF, DOCX, XLSX, CSV, JPG, PNG) with OCR, or convert them into clean Markdown, HTML, links, or a summary.
 
 
 ## How Does It Work?
 
 Maxun robots are automated tools that help you collect data from websites without writing any code. Think of them as your personal web assistants that can navigate websites, extract information, and organize data just like you would manually - but faster and more efficiently.
 
-There are four types of robots, each designed for a different job.
+There are five types of robots, each designed for a different job.
 
 ### 1. Extract
 Extract emulates real user behavior and captures structured data.
@@ -162,6 +163,13 @@ Run automated web searches to discover or scrape results, with support for time-
 
 Learn more <a href="https://docs.maxun.dev/robot/search/search-introduction">here</a>.
 
+### 5. Document Extraction & Parsing
+Upload a document or image — PDF, DOCX, XLSX, CSV, JPG, or PNG — and Maxun reads the text out of it, using OCR for scanned pages and photos. Convert it into clean Markdown, HTML, a list of links, or a summary, or use AI-powered extraction to pull specific structured fields.
+
+**Use cases:** digitize scanned receipts and invoices, pull fields from a photo of a paper form, turn a screenshot of a table into structured data, or convert a scanned document into clean Markdown for an AI workflow.
+
+Learn more <a href="https://docs.maxun.dev/">here</a>.
+
 ## Quick Start
 
 ### Getting Started
@@ -190,6 +198,7 @@ Maxun can run locally with or without Docker
 - ✨ **Turn Websites to Spreadsheets** – Direct data export to Google Sheets & Airtable
 - ✨ **Adapt To Website Layout Changes** – Auto-recovery from site updates
 - ✨ **Extract Behind Login** – Handle authentication seamlessly
+- ✨ **Extract From Documents & Images** – OCR PDFs, DOCX, XLSX, CSV, JPG, and PNG into structured data, Markdown, HTML, or links
 - ✨ **Integrations** – Connect with your favorite tools
 - ✨ **MCP Support** – Model Context Protocol integration
 - ✨ **LLM-Ready Data** – Clean Markdown for AI applications

server/src/api/sdk.ts

@@ -1485,14 +1485,18 @@ const documentUpload = multer({
     fileFilter: (_req, file, cb) => {
         const allowedMimeTypes = [
             'application/pdf',
+            'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
             'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
             'text/csv',
             'application/csv',
+            'image/jpeg',
+            'image/jpg',
+            'image/png',
         ];
         if (allowedMimeTypes.includes(file.mimetype)) {
             cb(null, true);
         } else {
-            cb(new Error('Only PDF, XLSX, and CSV files are allowed'));
+            cb(new Error('Only PDF, DOCX, XLSX, CSV, JPG, and PNG files are allowed'));
         }
     },
 });
@@ -1509,7 +1513,7 @@ router.post("/sdk/robots/document", requireAPIKey, documentUpload.single('file')
         if (!user) return res.status(401).json({ error: 'Unauthorized' });
 
         const file = (req as any).file as Express.Multer.File | undefined;
-        if (!file) return res.status(400).json({ error: 'A PDF file is required' });
+        if (!file) return res.status(400).json({ error: 'A PDF, DOCX, XLSX, CSV, JPG, or PNG file is required' });
 
         const prompt: string = (req.body.prompt || '').trim();
         if (!prompt) return res.status(400).json({ error: 'prompt is required' });
@@ -1583,7 +1587,7 @@ router.post("/sdk/robots/document-parse", requireAPIKey, documentUpload.single('
         if (!user) return res.status(401).json({ error: 'Unauthorized' });
 
         const file = (req as any).file as Express.Multer.File | undefined;
-        if (!file) return res.status(400).json({ error: 'A PDF file is required' });
+        if (!file) return res.status(400).json({ error: 'A PDF, DOCX, XLSX, CSV, JPG, or PNG file is required' });
 
         const rawFormats = req.body['outputFormats[]'] ?? req.body.outputFormats ?? req.body.formats;
         const requestedFormats: string[] = Array.isArray(rawFormats)

server/src/routes/storage.ts

@@ -51,13 +51,27 @@ const documentUpload = multer({
     if (normalizeDocumentMimeType(file.mimetype, file.originalname)) {
       cb(null, true);
     } else {
-      cb(new Error('Only PDF, DOCX, XLSX, and CSV files are allowed'));
+      cb(new Error('Only PDF, DOCX, XLSX, CSV, JPG, and PNG files are allowed'));
     }
   },
 });
 
+const uploadDocument = (req: any, res: any, next: any) => {
+  documentUpload.single('file')(req, res, (err: any) => {
+    if (err) {
+      if (err instanceof multer.MulterError && err.code === 'LIMIT_FILE_SIZE') {
+        const maxMb = Math.round(MAX_FILE_SIZE_BYTES / (1024 * 1024));
+        return res.status(400).json({ error: `File is too large. The maximum size is ${maxMb} MB.` });
+      }
+      return res.status(400).json({ error: err.message || 'Invalid file upload.' });
+    }
+    next();
+  });
+};
+
 //HELPER FUNCTION
 
+
 // const normalizeRobotUrl = (rawUrl: string): string => {
 //   let normalizedUrl: URL;
 //   try {
@@ -2117,21 +2131,21 @@ router.post('/recordings/search', requireSignIn, async (req: AuthenticatedReques
 
 /**
  * POST endpoint for creating a document extraction robot (doc-extract).
- * Accepts a PDF or DOCX upload and an extraction prompt. Uses the configured LLM to generate
- * an extraction schema and stores the document in MinIO.
+ * Accepts a PDF, DOCX, XLSX, CSV, JPG, or PNG upload and an extraction prompt. 
+ * Uses the configured LLM to generate an extraction schema and stores the document in MinIO.
  */
 router.post(
   '/recordings/document',
   requireSignIn,
-  documentUpload.single('file'),
+  uploadDocument,
   async (req: AuthenticatedRequest, res) => {
     try {
       if (!req.user) return res.status(401).json({ error: 'Unauthorized' });
 
       const file = (req as any).file as Express.Multer.File | undefined;
-      if (!file) return res.status(400).json({ error: 'A PDF or DOCX file is required.' });
+      if (!file) return res.status(400).json({ error: 'A PDF, DOCX, XLSX, CSV, JPG, or PNG file is required.' });
       const documentMimeType = normalizeDocumentMimeType(file.mimetype, file.originalname);
-      if (!documentMimeType) return res.status(400).json({ error: 'Only PDF and DOCX files are allowed.' });
+      if (!documentMimeType) return res.status(400).json({ error: 'Only PDF, DOCX, XLSX, CSV, JPG, or PNG files are allowed.' });
 
       const { prompt, name, llmProvider, llmModel, llmApiKey, llmBaseUrl } = req.body;
       if (!prompt || typeof prompt !== 'string' || !prompt.trim()) {
@@ -2191,21 +2205,21 @@ router.post(
 
 /**
  * POST endpoint for creating a document parse robot (doc-parse).
- * Accepts a PDF or DOCX upload and output format list. Parses the document immediately and
- * stores both the document and parsed output in MinIO / database.
+ * Accepts a PDF, DOCX, XLSX, CSV, JPG, or PNG upload and output format list. 
+ * Parses the document immediately and stores both the document and parsed output in MinIO / database.
  */
 router.post(
   '/recordings/document-parse',
   requireSignIn,
-  documentUpload.single('file'),
+  uploadDocument,
   async (req: AuthenticatedRequest, res) => {
     try {
       if (!req.user) return res.status(401).json({ error: 'Unauthorized' });
 
       const file = (req as any).file as Express.Multer.File | undefined;
-      if (!file) return res.status(400).json({ error: 'A PDF or DOCX file is required.' });
+      if (!file) return res.status(400).json({ error: 'A PDF, DOCX, XLSX, CSV, JPG, or PNG file is required.' });
       const documentMimeType = normalizeDocumentMimeType(file.mimetype, file.originalname);
-      if (!documentMimeType) return res.status(400).json({ error: 'Only PDF and DOCX files are allowed.' });
+      if (!documentMimeType) return res.status(400).json({ error: 'Only PDF, DOCX, XLSX, CSV, JPG, or PNG files are allowed.' });
 
       const { name, formats, llmProvider, llmModel, llmApiKey, llmBaseUrl } = req.body;
 
@@ -2218,7 +2232,7 @@ router.post(
         : DOC_PARSE_OUTPUT_FORMAT_OPTIONS.filter((f) => f !== 'summary');
 
       // Summaries need a working LLM. Ollama runs locally and needs no key, but the
-      // hosted providers do — fail early rather than parsing the PDF and then dying.
+      // hosted providers do — fail early rather than parsing the document and then dying.
       const summaryProvider = (llmProvider || 'ollama') as 'anthropic' | 'openai' | 'ollama';
       if (outputFormats.includes('summary') && summaryProvider !== 'ollama') {
         const envKey = summaryProvider === 'anthropic'
@@ -2386,15 +2400,15 @@ router.post('/runs/document-parse-run/:id', requireSignIn, async (req: Authentic
 router.put(
 '/recordings/:id/document',
   requireSignIn,
-  documentUpload.single('file'),
+  uploadDocument,
   async (req: AuthenticatedRequest, res) => {
     try {
       if (!req.user) return res.status(401).json({ error: 'Unauthorized' });
 
       const file = (req as any).file as Express.Multer.File | undefined;
-      if (!file) return res.status(400).json({ error: 'A PDF or DOCX file is required.' });
+      if (!file) return res.status(400).json({ error: 'A PDF, DOCX, XLSX, CSV, JPG, or PNG file is required.' });
       const documentMimeType = normalizeDocumentMimeType(file.mimetype, file.originalname);
-      if (!documentMimeType) return res.status(400).json({ error: 'Only PDF and DOCX files are allowed.' });
+      if (!documentMimeType) return res.status(400).json({ error: 'Only PDF, DOCX, XLSX, CSV, JPG, or PNG files are allowed.' });
 
       const robot = await Robot.findOne({ where: { 'recording_meta.id': req.params.id } });
       if (!robot) return res.status(404).json({ error: 'Robot not found.' });

server/src/utils/document/documentFile.ts

@@ -4,38 +4,54 @@ export const PDF_MIME_TYPE = 'application/pdf';
 export const DOCX_MIME_TYPE = 'application/vnd.openxmlformats-officedocument.wordprocessingml.document';
 export const XLSX_MIME_TYPE = 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet';
 export const CSV_MIME_TYPE = 'text/csv';
+export const JPEG_MIME_TYPE = 'image/jpeg';
+export const PNG_MIME_TYPE = 'image/png';
 
 const ALTERNATE_CSV_MIME_TYPES = new Set<string>(['application/csv', 'text/x-csv']);
+const ALTERNATE_JPEG_MIME_TYPES = new Set<string>(['image/jpg']);
 
 const SUPPORTED_DOCUMENT_MIME_TYPES = new Set<string>([
   PDF_MIME_TYPE,
   DOCX_MIME_TYPE,
   XLSX_MIME_TYPE,
   CSV_MIME_TYPE,
+  JPEG_MIME_TYPE,
+  PNG_MIME_TYPE
 ]);
 
 export const isSupportedDocumentMimeType = (mimeType: string | undefined): boolean =>
-  Boolean(mimeType && (SUPPORTED_DOCUMENT_MIME_TYPES.has(mimeType) || ALTERNATE_CSV_MIME_TYPES.has(mimeType)));
+  Boolean(mimeType && (
+    SUPPORTED_DOCUMENT_MIME_TYPES.has(mimeType) 
+    || ALTERNATE_CSV_MIME_TYPES.has(mimeType)
+    || ALTERNATE_JPEG_MIME_TYPES.has(mimeType)));
+
+export const isImageMimeType = (mimeType: string | undefined): boolean => 
+  mimeType === JPEG_MIME_TYPE || mimeType === PNG_MIME_TYPE;
 
 export const normalizeDocumentMimeType = (
   mimeType: string | undefined,
   originalFileName?: string
 ): string | null => {
   if (mimeType && ALTERNATE_CSV_MIME_TYPES.has(mimeType)) return CSV_MIME_TYPE;
+  if (mimeType && ALTERNATE_JPEG_MIME_TYPES.has(mimeType)) return JPEG_MIME_TYPE;
   if (isSupportedDocumentMimeType(mimeType)) return mimeType as string;
 
   const extension = path.extname(originalFileName || '').toLowerCase();
   if (extension === '.pdf') return PDF_MIME_TYPE;
   if (extension === '.docx') return DOCX_MIME_TYPE;
   if (extension === '.xlsx') return XLSX_MIME_TYPE;
   if (extension === '.csv') return CSV_MIME_TYPE;
+  if (extension === '.jpg' || extension === '.jpeg') return JPEG_MIME_TYPE;
+  if (extension === '.png') return PNG_MIME_TYPE;
   return null;
 };
 
-export const getDocumentExtensionForMimeType = (mimeType: string): '.pdf' | '.docx' | '.xlsx' | '.csv' => {
+export const getDocumentExtensionForMimeType = (mimeType: string): '.pdf' | '.docx' | '.xlsx' | '.csv' | '.jpg' | '.png' => {
   if (mimeType === DOCX_MIME_TYPE) return '.docx';
   if (mimeType === XLSX_MIME_TYPE) return '.xlsx';
   if (mimeType === CSV_MIME_TYPE || ALTERNATE_CSV_MIME_TYPES.has(mimeType)) return '.csv';
+  if (mimeType === JPEG_MIME_TYPE || ALTERNATE_JPEG_MIME_TYPES.has(mimeType)) return '.jpg';
+  if (mimeType === PNG_MIME_TYPE) return '.png';
   return '.pdf';
 };