From 148dcf2e431f9db55da2a13fe0af6cb3d2febb43 Mon Sep 17 00:00:00 2001
From: Wenjie Zhang <xerrors@163.com>
Date: Mon, 29 Dec 2025 11:25:03 +0800
Subject: [PATCH] =?UTF-8?q?feat(know-tool-calling):=20=E5=A2=9E=E5=BC=BA?=
 =?UTF-8?q?=E7=9F=A5=E8=AF=86=E5=BA=93=E5=B7=A5=E5=85=B7=E7=BB=93=E6=9E=9C?=
 =?UTF-8?q?=E6=98=BE=E7=A4=BA=E5=8A=9F=E8=83=BD?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

支持 get_mindmap 操作的纯文本显示
优化搜索结果的展示样式和参数解析
移除未使用的 FileTextOutlined 图标
---
 docs/latest/changelog/roadmap.md              |  4 +-
 .../ToolCallingResult/ToolCallRenderer.vue    |  5 +-
 .../tools/KnowledgeBaseTool.vue               | 80 ++++++++++++++++++-
 3 files changed, 83 insertions(+), 6 deletions(-)

diff --git a/docs/latest/changelog/roadmap.md b/docs/latest/changelog/roadmap.md
index e7a1605e..a1cdd33a 100644
--- a/docs/latest/changelog/roadmap.md
+++ b/docs/latest/changelog/roadmap.md
@@ -14,7 +14,9 @@
 - 考虑修改附件的处理逻辑，考虑使用文件系统，将附件解析后放到文件系统中，智能体按需读取，用户可以使用 @ 来引用附件，例如 @file:reports.md 相较于现在的处理逻辑感觉会更加自然一点。在此之前可能还要考虑文件系统的后端支持问题
 - skills 如何实现还需要继续调研
 - 优化 paddle 的命名，paddlex 有歧义，修改为 PP-StructureV3
-- 增加 paddle-vl 以及 deepseek-ocr 的支持
+- 增加 paddle-vl 以及 deepseek-ocr 的支持（deepseek-ocr 已支持）
+- RAG 检索的时候，支持限定文件类型
+- 将现有的 Milvus 的命名调整为通用 RAG
 
 ### Bugs
 - 部分异常状态下，智能体的模型名称出现重叠[#279](https://github.com/xerrors/Yuxi-Know/issues/279)
diff --git a/web/src/components/ToolCallingResult/ToolCallRenderer.vue b/web/src/components/ToolCallingResult/ToolCallRenderer.vue
index 0056478b..9ae38c6d 100644
--- a/web/src/components/ToolCallingResult/ToolCallRenderer.vue
+++ b/web/src/components/ToolCallingResult/ToolCallRenderer.vue
@@ -97,8 +97,9 @@ const isKnowledgeBaseResult = computed(() => {
     const hasKnowledgebaseTag = metadata.tag && metadata.tag.includes('knowledgebase');
     const isNotLightrag = metadata.kb_type !== 'lightrag';
     if (hasKnowledgebaseTag && isNotLightrag) {
-      const data = parseData(props.toolCall.tool_call_result?.content);
-      return Array.isArray(data) && data.length > 0;
+      // const data = parseData(props.toolCall.tool_call_result?.content);
+      // return Array.isArray(data) && data.length > 0;
+      return true
     }
   }
   return false;
diff --git a/web/src/components/ToolCallingResult/tools/KnowledgeBaseTool.vue b/web/src/components/ToolCallingResult/tools/KnowledgeBaseTool.vue
index b813ed56..f0c97b42 100644
--- a/web/src/components/ToolCallingResult/tools/KnowledgeBaseTool.vue
+++ b/web/src/components/ToolCallingResult/tools/KnowledgeBaseTool.vue
@@ -1,7 +1,22 @@
 <template>
-  <BaseToolCall :tool-call="toolCall">
+  <BaseToolCall :tool-call="toolCall" :hide-params="true">
+    <template #header>
+      <div class="sep-header">
+        <span class="note">{{ operationLabel }}</span>
+        <span class="separator" v-if="queryText">|</span>
+        <span class="description">"{{ queryText }}"</span>
+      </div>
+    </template>
     <template #result="{ resultContent }">
-      <div class="knowledge-base-result">
+      <!-- get_mindmap 操作：纯文本显示 -->
+      <div v-if="operation === 'get_mindmap'" class="knowledge-base-result">
+        <div class="mindmap-result">
+          <pre class="mindmap-content">{{ formatMindmapResult(resultContent) }}</pre>
+        </div>
+      </div>
+
+      <!-- search 操作：原有的文件分组显示 -->
+      <div v-else class="knowledge-base-result">
         <div class="result-summary">
           找到 {{ parsedData(resultContent).length }} 个相关文档片段，来自 {{ fileGroups(parsedData(resultContent)).length }} 个文件
         </div>
@@ -109,7 +124,7 @@
 <script setup>
 import BaseToolCall from '../BaseToolCall.vue';
 import { ref, computed } from 'vue'
-import { FileTextOutlined, FileOutlined, DownOutlined, EyeOutlined, DatabaseOutlined } from '@ant-design/icons-vue'
+import { FileOutlined, DownOutlined, EyeOutlined, DatabaseOutlined } from '@ant-design/icons-vue'
 
 const props = defineProps({
   toolCall: {
@@ -118,6 +133,38 @@ const props = defineProps({
   }
 })
 
+// 解析参数
+const args = computed(() => {
+  const args = props.toolCall.args || props.toolCall.function?.arguments;
+  if (!args) return {};
+
+  if (typeof args === 'object') return args;
+  try {
+    return JSON.parse(args);
+  } catch (e) {
+    return {};
+  }
+});
+
+// 获取操作类型
+const operation = computed(() => {
+  return args.value.operation || 'search';
+});
+
+// 获取操作标签
+const operationLabel = computed(() => {
+  const labels = {
+    search: `${props.toolCall.name} 搜索`,
+    get_mindmap: props.toolCall.name
+  };
+  return labels[operation.value] || operation.value;
+});
+
+// 获取查询文本
+const queryText = computed(() => {
+  return args.value.query_text || '';
+});
+
 const parseData = (content) => {
   if (typeof content === 'string') {
     try {
@@ -194,6 +241,17 @@ const getScoreColor = (score) => {
   if (score >= 0.5) return '#faad14'  // 橙色 - 中等相关性
   return '#ff4d4f'  // 红色 - 低相关性
 }
+
+// 格式化思维导图结果
+const formatMindmapResult = (content) => {
+  if (typeof content === 'string') {
+    return content;
+  }
+  if (typeof content === 'object') {
+    return JSON.stringify(content, null, 2);
+  }
+  return String(content);
+}
 </script>
 
 <style lang="less" scoped>
@@ -202,6 +260,22 @@ const getScoreColor = (score) => {
   border-radius: 8px;
   // border: 1px solid var(--gray-200);
 
+  .mindmap-result {
+    padding: 12px 16px;
+    max-height: 300px;
+    overflow-y: auto;
+
+    .mindmap-content {
+      margin: 0;
+      font-size: 13px;
+      line-height: 1.6;
+      color: var(--gray-700);
+      white-space: pre-wrap;
+      word-break: break-word;
+      font-family: 'Monaco', 'Menlo', 'Ubuntu Mono', monospace;
+    }
+  }
+
   .result-summary {
     padding: 12px 16px;
     background: var(--gray-25);