feat: Agent不支持视觉能力时上传图片提示用户

main_v2.py

@@ -120,6 +120,8 @@ async def get_providers(db: Session = Depends(get_db)):
                 "default_model": p.default_model,
                 "supports_thinking": p.supports_thinking,
                 "thinking_model": p.thinking_model,
+                "supports_vision": p.supports_vision,
+                "vision_model": p.vision_model,
                 "max_tokens": p.max_tokens,
                 "temperature": p.temperature,
                 "is_active": p.is_active,

models_v2.py

@@ -32,6 +32,10 @@ class LLMProvider(Base):
     supports_thinking = Column(Boolean, default=False)  # 是否原生支持思考
     thinking_model = Column(String(100), nullable=True)  # 思考模式模型名（如有单独模型）
     
+    # 视觉能力支持
+    supports_vision = Column(Boolean, default=False)  # 是否支持图片理解（多模态）
+    vision_model = Column(String(100), nullable=True)  # 视觉模型名（如与默认模型不同）
+    
     # 配额和限制
     max_tokens = Column(Integer, default=4096)
     temperature = Column(Float, default=0.7)

templates/admin_v2/index.html

@@ -58,8 +58,8 @@
                 </div>
                 <div class="card-body">
                     <table class="table">
-                        <thead><tr><th>名称</th><th>API地址</th><th>默认模型</th><th>思考支持</th><th>状态</th><th>操作</th></tr></thead>
-                        <tbody id="providers-list"><tr><td colspan="6" class="text-center">加载中...</td></tr></tbody>
+                        <thead><tr><th>名称</th><th>API地址</th><th>默认模型</th><th>思考</th><th>视觉</th><th>状态</th><th>操作</th></tr></thead>
+                        <tbody id="providers-list"><tr><td colspan="7" class="text-center">加载中...</td></tr></tbody>
                     </table>
                 </div>
             </div>
@@ -162,6 +162,8 @@
             <div class="mt-3 form-check"><input type="checkbox" class="form-check-input" id="provider-active" checked><label class="form-check-label">启用</label></div>
             <hr><h6>思考功能</h6>
             <div class="thinking-config"><div class="row"><div class="col-md-6 form-check"><input type="checkbox" class="form-check-input" id="provider-supports-thinking"><label class="form-check-label">支持原生思考</label></div><div class="col-md-6"><label class="form-label">思考模型名</label><input type="text" class="form-control" id="provider-thinking-model"></div></div></div>
+            <hr><h6>视觉能力</h6>
+            <div class="thinking-config"><div class="row"><div class="col-md-6 form-check"><input type="checkbox" class="form-check-input" id="provider-supports-vision"><label class="form-check-label">支持图片理解</label></div><div class="col-md-6"><label class="form-label">视觉模型名</label><input type="text" class="form-control" id="provider-vision-model" placeholder="留空则使用默认模型"></div></div><small class="text-muted mt-2 d-block">启用后可上传图片让AI识别分析内容</small></div>
             <div class="mt-3"><button type="button" class="btn btn-outline-primary" onclick="fetchProviderModels()"><i class="ri-refresh-line"></i> 获取模型</button><button type="button" class="btn btn-outline-secondary" onclick="testProviderConnection()"><i class="ri-link"></i> 测试连接</button></div>
             <div class="mt-2" id="provider-models-preview"></div><div class="mt-2" id="provider-test-result"></div>
         </form></div>
@@ -274,6 +276,7 @@
             tbody.innerHTML = providersData.map(p => `<tr>
                 <td><strong>${p.name}</strong></td><td><small>${p.api_base||'-'}</small></td><td>${p.default_model||'auto'}</td>
                 <td>${p.supports_thinking?'<span class="badge bg-success">支持</span>':'<span class="badge bg-secondary">不支持</span>'}</td>
+                <td>${p.supports_vision?'<span class="badge bg-info">支持</span>':'<span class="badge bg-secondary">不支持</span>'}</td>
                 <td>${p.is_active?'<span class="badge bg-success">启用</span>':'<span class="badge bg-secondary">禁用</span>'}</td>
                 <td><button class="btn btn-sm btn-outline-primary" onclick="editProvider(${p.id})"><i class="ri-edit-line"></i></button>
                     <button class="btn btn-sm btn-outline-danger" onclick="deleteProvider(${p.id},'${p.name}')"><i class="ri-delete-bin-line"></i></button></td>
@@ -290,6 +293,8 @@
             document.getElementById('provider-form').reset();
             document.getElementById('provider-id').value = '';
             document.getElementById('provider-active').checked = true;
+            document.getElementById('provider-supports-thinking').checked = false;
+            document.getElementById('provider-supports-vision').checked = false;
             document.getElementById('provider-models-preview').innerHTML = '';
             document.getElementById('provider-test-result').innerHTML = '';
             new bootstrap.Modal(document.getElementById('providerModal')).show();
@@ -310,6 +315,8 @@
             document.getElementById('provider-active').checked = p.is_active;
             document.getElementById('provider-supports-thinking').checked = p.supports_thinking;
             document.getElementById('provider-thinking-model').value = p.thinking_model || '';
+            document.getElementById('provider-supports-vision').checked = p.supports_vision;
+            document.getElementById('provider-vision-model').value = p.vision_model || '';
             new bootstrap.Modal(document.getElementById('providerModal')).show();
         }
 
@@ -326,7 +333,9 @@
                 description: document.getElementById('provider-description').value,
                 is_active: document.getElementById('provider-active').checked,
                 supports_thinking: document.getElementById('provider-supports-thinking').checked,
-                thinking_model: document.getElementById('provider-thinking-model').value
+                thinking_model: document.getElementById('provider-thinking-model').value,
+                supports_vision: document.getElementById('provider-supports-vision').checked,
+                vision_model: document.getElementById('provider-vision-model').value
             };
             const res = await fetch(id ? `/api/v2/providers/${id}` : '/api/v2/providers', { method: id ? 'PUT' : 'POST', headers: {'Content-Type':'application/json'}, body: JSON.stringify(data) });
             const result = await res.json();

templates/index.html

@@ -260,6 +260,7 @@
         let currentConversationId = null;
         let currentAgentId = null;
         let agents = [];
+        let providers = [];  // 大模型池数据
         let quickPhrases = [];
         let lastSentMessage = null;  // 记录最后发送的消息
         let lastSentFiles = null;  // 记录发送的文件
@@ -272,6 +273,7 @@
         let messageVersions = {}; // 存储每个assistant消息的多个版本 { messageId: [{content, thinking}] }
         
         document.addEventListener('DOMContentLoaded', () => {
+            loadProviders();  // 加载大模型池
             loadAgents();
             loadQuickPhrases();
             connectWebSocket();
@@ -279,6 +281,23 @@
             setupTextarea();
         });
         
+        // 加载大模型池
+        async function loadProviders() {
+            try {
+                const res = await fetch('/api/v2/providers');
+                const data = await res.json();
+                providers = data.providers || [];
+            } catch (e) { console.error('加载Provider失败:', e); }
+        }
+        
+        // 检查当前Agent是否支持视觉
+        function checkAgentVisionSupport() {
+            const agent = agents.find(a => a.id === currentAgentId);
+            if (!agent) return false;
+            const provider = providers.find(p => p.id === agent.llm_provider_id);
+            return provider?.supports_vision || false;
+        }
+        
         // 加载Agent
         async function loadAgents() {
             try {
@@ -868,6 +887,16 @@
             // 如果没有消息且没有文件，不发送
             if (!msg && pendingFiles.length === 0) return;
             
+            // 检查是否有图片上传
+            const hasImages = pendingFiles.some(f => f.type && f.type.startsWith('image/'));
+            if (hasImages && !checkAgentVisionSupport()) {
+                const agent = agents.find(a => a.id === currentAgentId);
+                const agentName = agent?.display_name || agent?.name || '当前Agent';
+                alert(`⚠️ ${agentName} 不支持图片识别功能\n\n请选择支持视觉能力的Agent（如 vlm-agent），或者移除图片后再发送。`);
+                document.getElementById('sendBtn').disabled = false;
+                return;
+            }
+            
             document.getElementById('sendBtn').disabled = true;
             input.value = '';
             input.style.height = 'auto';
@@ -1014,6 +1043,21 @@
                     }
                     
                     previewArea.appendChild(previewItem);
+                    
+                    // 如果上传的是图片且当前Agent不支持视觉，显示警告提示
+                    if (file.type.startsWith('image/') && !checkAgentVisionSupport()) {
+                        const agent = agents.find(a => a.id === currentAgentId);
+                        const agentName = agent?.display_name || agent?.name || '当前Agent';
+                        const warningDiv = document.createElement('div');
+                        warningDiv.className = 'vision-warning';
+                        warningDiv.id = 'vision-warning-tip';
+                        warningDiv.style.cssText = 'margin-top:8px;padding:8px 12px;background:#fff3cd;border:1px solid #ffc107;border-radius:6px;font-size:13px;color:#856404;';
+                        warningDiv.innerHTML = `<i class="ri-alert-line"></i> <strong>${agentName}</strong> 不支持图片识别，请选择支持视觉的Agent或移除图片`;
+                        // 避免重复添加
+                        if (!document.getElementById('vision-warning-tip')) {
+                            previewArea.appendChild(warningDiv);
+                        }
+                    }
                 };
                 
                 // 根据文件类型选择读取方式
@@ -1035,6 +1079,13 @@
             pendingFiles = pendingFiles.filter(f => f.id !== fileId);
             const preview = document.getElementById(fileId + '-preview');
             if (preview) preview.remove();
+            
+            // 如果没有图片了，移除视觉警告提示
+            const hasImages = pendingFiles.some(f => f.type && f.type.startsWith('image/'));
+            if (!hasImages) {
+                const warning = document.getElementById('vision-warning-tip');
+                if (warning) warning.remove();
+            }
         }
         
         // 发送消息