feat: add PDF processing support and enhance document handling

examples/attachment-demo-server.ts

@@ -9,7 +9,7 @@ When users send you attachments, analyze them carefully and provide helpful, det
 
 For images: Describe what you see in detail.
 For documents: Analyze and summarize the content, structure, or data as appropriate.
-Supported document types: DOCX, XLSX, CSV, TXT, JSON, ZIP files.`,
+Supported document types: PDF, DOCX, XLSX, CSV, TXT, JSON, ZIP files.`,
   modelConfig: {
     name: 'claude-sonnet-4', 
     temperature: 0.7,
@@ -144,6 +144,48 @@ server.start().then(() => {
     ]
   }'\n`);
 
+  console.log('5.5. PDF document:');
+  console.log(`curl -X POST http://localhost:3002/chat \\
+  -H "Content-Type: application/json" \\
+  -d '{
+    "agentName": "attachment-analyst",
+    "messages": [
+      {
+        "role": "user",
+        "content": "Please analyze this PDF document",
+        "attachments": [
+          {
+            "kind": "document",
+            "mimeType": "application/pdf",
+            "name": "sample-document.pdf",
+            "url": "https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf"
+          }
+        ]
+      }
+    ]
+  }'\n`);
+
+  console.log('5.6. PDF with base64 data:');
+  console.log(`curl -X POST http://localhost:3002/chat \\
+  -H "Content-Type: application/json" \\
+  -d '{
+    "agentName": "attachment-analyst",
+    "messages": [
+      {
+        "role": "user",
+        "content": "Extract text from this PDF document",
+        "attachments": [
+          {
+            "kind": "document",
+            "mimeType": "application/pdf",
+            "name": "test-document.pdf",
+            "data": "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"
+          }
+        ]
+      }
+    ]
+  }'\n`);
+
   console.log('6. CSV data:');
   console.log(`curl -X POST http://localhost:3002/chat \\
   -H "Content-Type: application/json" \\
@@ -286,7 +328,7 @@ server.start().then(() => {
   console.log('Configuration:');
   console.log('- Use Ctrl+C to stop the server');
   console.log('- Image attachments: Full visual analysis');
-  console.log('- Document attachments: Text extraction and analysis for DOCX, XLSX, CSV, TXT, JSON, ZIP');
+  console.log('- Document attachments: Text extraction and analysis for PDF, DOCX, XLSX, CSV, TXT, JSON, ZIP');
   console.log('- LiteLLM format: Use "useLiteLLMFormat": true for efficient large file processing');
   console.log('  * Large documents: No context window waste, native model processing');
   console.log('  * Better layout understanding, tables, images preserved');

jest.config.js

@@ -18,6 +18,8 @@ module.exports = {
   ],
   moduleNameMapper: {
     '^(\\.{1,2}/.*)\\.js$': '$1',
+    '^pdfjs-dist/legacy/build/pdf\\.mjs$': '<rootDir>/src/__mocks__/pdfjs-dist.ts',
+    '^canvas$': '<rootDir>/src/__mocks__/canvas.ts',
   },
   collectCoverageFrom: [
     'src/**/*.ts',

package.json

@@ -143,13 +143,15 @@
     "@modelcontextprotocol/sdk": "^1.17.4",
     "@types/yauzl": "^2.10.3",
     "ai": "^5.0.35",
+    "canvas": "^3.2.0",
     "eventsource": "^2.0.2",
     "express": "^4.18.2",
     "fastify": "^4.29.1",
     "mammoth": "^1.10.0",
     "mathjs": "^14.6.0",
     "openai": "^4.0.0",
     "papaparse": "^5.5.3",
+    "pdfjs-dist": "^5.4.149",
     "tunnel": "^0.0.6",
     "uuid": "^9.0.0",
     "xlsx": "^0.18.5",