You can now upload PDFs, CSVs, or images, and the agent will process them accordingly. (#2)

- app.py to support PDF, CSV, and DOC files alongside images, as requested. (3edcefed5e91960b696cd0ab36f9388741e60424)

app.py

@@ -32,8 +32,9 @@ import json
 import warnings
 import logging
 import time
+import base64
 from functools import lru_cache
-from typing import Optional
+from typing import Optional, List
 from datetime import datetime
 
 # Suppress ResourceWarning for cleaner output
@@ -322,6 +323,20 @@ so be accurate and thorough.
 # AGENT CREATION
 # =============================================================================
 
+
+
+class FilePayload(BaseModel):
+    data: str  # Base64 encoded data
+    format: str
+    name: str = "file"
+
+
+class QuestionRequest(BaseModel):
+    question: str
+    files: Optional[List[FilePayload]] = None
+
+
+
 _cached_agent = None
 
 
@@ -382,12 +397,95 @@ def create_enhanced_agent():
     return _cached_agent
 
 
-def query_agent(question: str) -> str:
-    """Process question with the enhanced agent."""
+def query_agent(question: str, files: Optional[List[FilePayload]] = None) -> str:
+    """Process question with the enhanced agent, optionally including files."""
     try:
         logger.info(f"Processing query: {question}")
+        if files:
+            logger.info(f"Query includes {len(files)} files")
+
         agent = create_enhanced_agent()
-        result = agent(question)
+
+        if files:
+            # Try to use official types if available
+            try:
+                from strands.types.content import ImageContent, DocumentContent
+                from strands.types.media import ImageSource, DocumentSource
+                
+                message_content = [{"text": question}]
+                
+                image_formats = {'png', 'jpeg', 'gif', 'webp', 'jpg'}
+                
+                for file_obj in files:
+                    try:
+                        # Remove header if present
+                        base64_data = file_obj.data
+                        if "," in base64_data:
+                            base64_data = base64_data.split(",")[1]
+                        
+                        file_bytes = base64.b64decode(base64_data)
+                        fmt = file_obj.format.lower()
+                        
+                        if fmt in image_formats:
+                            # Handle Image
+                            image_block = ImageContent(
+                                format=fmt if fmt != 'jpg' else 'jpeg', # Normalize jpg
+                                source=ImageSource(bytes=file_bytes)
+                            )
+                            message_content.append({"image": image_block})
+                        else:
+                            # Handle Document
+                            doc_block = DocumentContent(
+                                format=fmt,
+                                name=file_obj.name,
+                                source=DocumentSource(bytes=file_bytes)
+                            )
+                            message_content.append({"document": doc_block})
+                            
+                    except Exception as err:
+                        logger.error(f"Failed to process file {file_obj.name}: {err}")
+                
+            except ImportError:
+                # Fallback for older versions or missing imports
+                logger.info("Using legacy/dict construction (strands.types not found)")
+                message_content = [{"text": question}]
+                
+                image_formats = {'png', 'jpeg', 'gif', 'webp', 'jpg'}
+
+                for file_obj in files:
+                    try:
+                        base64_data = file_obj.data
+                        if "," in base64_data:
+                            base64_data = base64_data.split(",")[1]
+                        
+                        file_bytes = base64.b64decode(base64_data)
+                        fmt = file_obj.format.lower()
+                        
+                        if fmt in image_formats:
+                             message_content.append({
+                                "image": {
+                                    "format": fmt if fmt != 'jpg' else 'jpeg',
+                                    "source": {"bytes": file_bytes},
+                                },
+                            })
+                        else:
+                             message_content.append({
+                                "document": {
+                                    "format": fmt,
+                                    "name": file_obj.name,
+                                    "source": {"bytes": file_bytes},
+                                },
+                            })
+                            
+                    except Exception as err:
+                        logger.error(f"Failed to process file: {err}")
+
+            # Call agent with list payload
+            result = agent(message_content)
+        else:
+            # Standard text-only call
+            result = agent(question)
+
         logger.info("Query completed successfully")
         return str(result)
     except Exception as e:
@@ -411,8 +509,8 @@ app.add_middleware(
 )
 
 
-class QuestionRequest(BaseModel):
-    question: str
+
+
 
 
 class AnswerResponse(BaseModel):
@@ -430,7 +528,7 @@ async def index():
 async def ask_question(request: QuestionRequest):
     """Process a question and return the answer."""
     try:
-        answer = query_agent(request.question)
+        answer = query_agent(request.question, request.files)
         return AnswerResponse(
             answer=answer,
             metrics=_metrics.get_stats()
@@ -731,12 +829,23 @@ def get_ui_html() -> str:
             
             <div class="main-content">
                 <div class="chat-section">
+
                     <div class="input-area">
                         <textarea 
                             id="questionInput" 
                             placeholder="Ask a question about fraud models, applications, or policies..."
                             onkeydown="if(event.key === 'Enter' && !event.shiftKey) {{ event.preventDefault(); askQuestion(); }}"
                         ></textarea>
+                    </div>
+                    
+                    <div class="controls-area" style="display: flex; gap: 10px; margin-bottom: 20px; align-items: center;">
+                         <label for="fileInput" style="cursor: pointer; display: flex; align-items: center; gap: 5px; color: #00d4ff; font-size: 0.9rem;">
+                            <svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M13 2H6a2 2 0 0 0-2 2v16a2 2 0 0 0 2 2h12a2 2 0 0 0 2-2V9z"></path><polyline points="13 2 13 9 20 9"></polyline></svg>
+                            Add File
+                        </label>
+                        <input type="file" id="fileInput" accept="image/*, .pdf, .csv, .doc, .docx, .xls, .xlsx, .txt, .md, .html" style="display: none;" onchange="updateFileLabel()">
+                        <span id="fileName" style="color: #888; font-size: 0.8rem; margin-right: auto;"></span>
+                        
                         <button id="askBtn" onclick="askQuestion()">🔍 Analyze</button>
                     </div>
                     
@@ -816,53 +925,110 @@ Enter your question above and click "Analyze" to get started.</pre>
                 examplesContainer.appendChild(btn);
             }});
             
+            function updateFileLabel() {{
+                const input = document.getElementById('fileInput');
+                const fileName = document.getElementById('fileName');
+                if (input.files && input.files.length > 0) {{
+                    fileName.textContent = input.files[0].name;
+                }} else {{
+                    fileName.textContent = "";
+                }}
+            }}
+            
+            async function convertFileToBase64(file) {{
+                return new Promise((resolve, reject) => {{
+                    const reader = new FileReader();
+                    reader.onload = () => resolve(reader.result);
+                    reader.onerror = error => reject(error);
+                    reader.readAsDataURL(file);
+                }});
+            }}
+
             async function askQuestion() {{
                 const input = document.getElementById('questionInput');
+                const fileInput = document.getElementById('fileInput');
                 const btn = document.getElementById('askBtn');
                 const responseArea = document.getElementById('responseArea');
-                const responseText = document.getElementById('responseText');
                 
                 const question = input.value.trim();
-                if (!question) return;
                 
+                if (!question && (!fileInput.files || fileInput.files.length === 0)) return;
+                
+                // UI updates
+                input.disabled = true;
                 btn.disabled = true;
                 btn.textContent = '⏳ Processing...';
-                responseArea.innerHTML = '<div class="loading"><div class="spinner"></div>Analyzing your question...</div>';
+                responseArea.innerHTML = '<div class="loading"><div class="spinner"></div>Analyzing request...</div>';
                 
                 try {{
+                    const payload = {{ 
+                        question: question,
+                        files: []
+                    }};
+                    
+                    // Handle file upload
+                    if (fileInput.files && fileInput.files.length > 0) {{
+                        const file = fileInput.files[0];
+                        const base64Data = await convertFileToBase64(file);
+                        
+                        // Determine format
+                        let format = "txt"; // default
+                        
+                        // Try from type
+                        if (file.type) {{
+                            const subtype = file.type.split('/')[1];
+                            if (subtype) format = subtype; 
+                        }}
+                        
+                        // Try from extension (more reliable for docs)
+                        const nameParts = file.name.split('.');
+                        if (nameParts.length > 1) {{
+                            format = nameParts[nameParts.length - 1].toLowerCase();
+                        }}
+                        
+                        payload.files.push({{
+                            data: base64Data,
+                            format: format,
+                            name: file.name
+                        }});
+                    }}
+                    
                     const response = await fetch('/api/ask', {{
                         method: 'POST',
                         headers: {{
-                            'Content-Type': 'application/json',
+                            'Content-Type': 'application/json'
                         }},
-                        body: JSON.stringify({{ question }}),
+                        body: JSON.stringify(payload)
                     }});
                     
-                    if (!response.ok) {{
-                        throw new Error(`HTTP error! status: ${{response.status}}`);
-                    }}
-                    
                     const data = await response.json();
                     
-                    responseArea.innerHTML = '<pre id="responseText"></pre>';
-                    document.getElementById('responseText').textContent = data.answer;
-                    
-                    // Update metrics
-                    if (data.metrics) {{
-                        document.getElementById('totalSearches').textContent = data.metrics.total_searches || 0;
-                        document.getElementById('cacheRate').textContent = 
-                            ((data.metrics.cache_hit_rate || 0) * 100).toFixed(0) + '%';
-                        document.getElementById('avgTime').textContent = 
-                            (data.metrics.avg_query_time || 0).toFixed(2) + 's';
-                        document.getElementById('errors').textContent = data.metrics.errors || 0;
+                    if (response.ok) {{
+                        responseArea.innerHTML = '<pre id="responseText"></pre>';
+                        document.getElementById('responseText').textContent = data.answer;
+                        
+                        // Update metrics
+                        if (data.metrics) {{
+                            document.getElementById('totalSearches').textContent = data.metrics.total_searches || 0;
+                            document.getElementById('cacheRate').textContent = 
+                                ((data.metrics.cache_hit_rate || 0) * 100).toFixed(0) + '%';
+                            document.getElementById('avgTime').textContent = 
+                                (data.metrics.avg_query_time || 0).toFixed(2) + 's';
+                            document.getElementById('errors').textContent = data.metrics.errors || 0;
+                        }}
+                    }} else {{
+                        responseArea.innerHTML = `<pre style="color: #ff6b6b">Error: ${{data.detail || 'Unknown error occurred'}}</pre>`;
                     }}
                     
                 }} catch (error) {{
-                    responseArea.innerHTML = '<pre id="responseText" style="color: #ff6b6b;"></pre>';
-                    document.getElementById('responseText').textContent = 'Error: ' + error.message;
+                    responseArea.innerHTML = `<pre style="color: #ff6b6b">Network Error: ${{error.message}}</pre>`;
                 }} finally {{
+                    input.disabled = false;
                     btn.disabled = false;
                     btn.textContent = '🔍 Analyze';
+                    input.focus();
+                    fileInput.value = "";
+                    updateFileLabel();
                 }}
             }}