1

ui-src/src/api/settings.js

@@ -0,0 +1,9 @@
+import request from './request.js'
+
+export function getSettings() {
+  return request.get('/system-config/settings')
+}
+
+export function saveSettings(data) {
+  return request.post('/system-config/save-settings', data)
+}

ui-src/src/views/DatasetList.vue

@@ -12,8 +12,10 @@
           <el-tag v-else size="small" type="info">未设置</el-tag>
         </template>
       </el-table-column>
-      <el-table-column label="分块" width="140">
-        <template #default="{ row }">{{ STRATEGY_NAMES[row.chunk_strategy || 'title'] }} {{ row.chunk_size || 800 }}{{ row.chunk_strategy === 'semantic' ? '' : ' / ' + (row.chunk_overlap ?? 150) }}</template>
+      <el-table-column label="分块" width="210">
+        <template #default="{ row }">
+          {{ row.chunk_size || 800 }} / {{ row.chunk_overlap ?? 150 }}<template v-if="row.unit_pattern">（自动识别：{{ humanizePattern(row.unit_pattern) }}）</template>
+        </template>
       </el-table-column>
       <el-table-column prop="created_at" label="创建时间" width="170" />
       <el-table-column label="操作" width="190" fixed="right">
@@ -38,21 +40,13 @@
             <el-option v-for="m in embedders" :key="m.id" :label="m.name" :value="m.id" />
           </el-select>
         </el-form-item>
-        <el-form-item label="分块策略">
-          <el-select v-model="form.chunk_strategy" style="width: 100%">
-            <el-option label="标题分块（推荐）" value="title" />
-            <el-option label="递归字符分块" value="recursive" />
-            <el-option label="语义分块" value="semantic" />
-          </el-select>
-          <div class="ds-tip">标题分块保留标题与段落结构；递归字符分块按分隔符切分，通用文本；语义分块按语义相似度切分，质量更高但解析更慢，且需绑定向量模型</div>
-        </el-form-item>
-        <el-form-item :label="form.chunk_strategy === 'semantic' ? '上限大小' : '分块大小'">
+        <el-form-item label="分块大小">
           <el-input-number v-model="form.chunk_size" :min="50" :max="5000" :step="50" style="width: 100%" />
-          <div class="ds-tip">{{ form.chunk_strategy === 'semantic' ? '语义分块按句子相似度切分，此值为安全上限（max_chunk_size），超过上限的块会再切分；语义分块不支持重叠' : '每块最大字数，超长段落自动按句号/换行切分' }}</div>
+          <div class="ds-tip">每块最大字数。系统自动组合分块策略：优先识别文档结构（条文/章节/编号等）按结构切分，无结构时依次按标题感知、语义、递归切分</div>
         </el-form-item>
-        <el-form-item v-if="form.chunk_strategy !== 'semantic'" label="重叠字数">
+        <el-form-item label="重叠字数">
           <el-input-number v-model="form.chunk_overlap" :min="0" :max="500" :step="10" style="width: 100%" />
-          <div class="ds-tip">相邻分块间的重叠字数，用于保持上下文连贯</div>
+          <div class="ds-tip">相邻分块间的重叠字数，用于保持上下文连贯（语义切分路径自动忽略）</div>
         </el-form-item>
       </el-form>
       <template #footer>
@@ -68,21 +62,41 @@ import { onMounted, ref } from 'vue'
 import { ElMessage, ElMessageBox } from 'element-plus'
 import { listDatasets, saveDataset, deleteDataset } from '../api/dataset.js'
 import { listModelConfigs } from '../api/model_config.js'
+import { getSettings } from '../api/settings.js'
 
-const STRATEGY_NAMES = { title: '标题', recursive: '递归', semantic: '语义' }
+// 把识别出的结构正则转成可读名称（第[一二三四五六七八九十百千]+条 → 条文）
+function humanizePattern(p) {
+  if (!p) return ''
+  if (p.includes('条')) return '条文'
+  if (p.includes('回')) return '章回'
+  if (p.includes('节')) return '小节'
+  if (p.includes('章')) return '章节'
+  if (p.includes('篇') || p.includes('部')) return '篇章'
+  if (p.includes('（') || p.includes('(')) return '序号'
+  if (/\d/.test(p)) return '编号'
+  return '结构单元'
+}
 
 const datasets = ref([])
 const embedders = ref([])
+const defaults = ref({ chunk_size: 800, chunk_overlap: 150 })
 const loading = ref(false)
 const saving = ref(false)
 const dialogVisible = ref(false)
 const editing = ref(false)
 let editingRowCfgId = 0
-let editingRowChunk = { chunk_size: 0, chunk_overlap: 0, chunk_strategy: '' }
-const form = ref({ id: 0, name: '', description: '', embedding_cfg_id: 0, chunk_size: 800, chunk_overlap: 150, chunk_strategy: 'title' })
+let editingRowChunk = { chunk_size: 0, chunk_overlap: 0 }
+const form = ref({ id: 0, name: '', description: '', embedding_cfg_id: 0, chunk_size: 800, chunk_overlap: 150 })
 
 onMounted(async () => {
   await load()
+  try {
+    const s = await getSettings()
+    if (s) {
+      defaults.value.chunk_size = s.chunk_size || 800
+      defaults.value.chunk_overlap = s.chunk_overlap ?? 150
+    }
+  } catch { /* 忽略 */ }
   try {
     const m = await listModelConfigs('embedding')
     if (m && m.list) embedders.value = m.list
@@ -107,20 +121,20 @@ function embeddingName(id) {
 function openCreate() {
   editing.value = false
   editingRowCfgId = 0
-  editingRowChunk = { chunk_size: 0, chunk_overlap: 0, chunk_strategy: '' }
-  // 默认选中默认向量模型，其次第一个
+  editingRowChunk = { chunk_size: 0, chunk_overlap: 0 }
+  // 默认选中默认向量模型，其次第一个；分块大小/重叠带出全局设置值
   const def = embedders.value.find(x => x.is_default === 1) || embedders.value[0]
-  form.value = { id: 0, name: '', description: '', embedding_cfg_id: def?.id || 0, chunk_size: 800, chunk_overlap: 150, chunk_strategy: 'title' }
+  form.value = { id: 0, name: '', description: '', embedding_cfg_id: def?.id || 0, chunk_size: defaults.value.chunk_size, chunk_overlap: defaults.value.chunk_overlap }
   dialogVisible.value = true
 }
 
 function openEdit(row) {
   editing.value = true
   editingRowCfgId = row.embedding_cfg_id
-  editingRowChunk = { chunk_size: row.chunk_size || 800, chunk_overlap: row.chunk_overlap ?? 150, chunk_strategy: row.chunk_strategy || 'title' }
+  editingRowChunk = { chunk_size: row.chunk_size || 800, chunk_overlap: row.chunk_overlap ?? 150 }
   form.value = {
     id: row.id, name: row.name, description: row.description, embedding_cfg_id: row.embedding_cfg_id,
-    chunk_size: editingRowChunk.chunk_size, chunk_overlap: editingRowChunk.chunk_overlap, chunk_strategy: editingRowChunk.chunk_strategy,
+    chunk_size: editingRowChunk.chunk_size, chunk_overlap: editingRowChunk.chunk_overlap,
   }
   dialogVisible.value = true
 }
@@ -135,11 +149,11 @@ async function save() {
     return
   }
   const modelChanged = editing.value && form.value.embedding_cfg_id !== editingRowCfgId
-  const chunkChanged = editing.value && (form.value.chunk_size !== editingRowChunk.chunk_size || form.value.chunk_overlap !== editingRowChunk.chunk_overlap || form.value.chunk_strategy !== editingRowChunk.chunk_strategy)
+  const chunkChanged = editing.value && (form.value.chunk_size !== editingRowChunk.chunk_size || form.value.chunk_overlap !== editingRowChunk.chunk_overlap)
   if (modelChanged || chunkChanged) {
     const changed = []
     if (modelChanged) changed.push('向量模型')
-    if (chunkChanged) changed.push('分块策略')
+    if (chunkChanged) changed.push('分块配置')
     try {
       await ElMessageBox.confirm(`${changed.join('与')}已变更，保存后将重新处理该数据集下所有文档（异步进行，可在文档列表查看进度）。是否继续？`, '重新处理文档', { type: 'warning' })
     } catch {

ui-src/src/views/KgGraph.vue

@@ -4,33 +4,44 @@
       <el-select v-model="datasetId" placeholder="选择知识库" style="width: 260px" @change="load">
         <el-option v-for="d in datasets" :key="d.id" :label="d.name" :value="d.id" />
       </el-select>
-      <span class="kg-tip">实体与关系由解析流水线中的 LLM 抽取生成，问答时命中实体自动注入一跳邻居</span>
+      <span class="kg-tip">实体与关系由解析流水线中的 LLM 抽取生成，问答时命中实体自动注入一跳邻居；点击节点高亮其关联实体</span>
     </div>
 
     <el-row :gutter="12">
-      <el-col :span="10">
+      <el-col :span="15">
         <el-card shadow="never" class="kg-card">
-          <template #header>实体（{{ entityTotal }}）</template>
-          <el-table :data="entities" size="small" v-loading="loading">
-            <el-table-column prop="name" label="实体" min-width="140" show-overflow-tooltip />
-            <el-table-column prop="entity_type" label="类型" width="110" />
-            <el-table-column prop="chunk_id" label="来源分块" width="100" />
-          </el-table>
-          <el-pagination class="kg-page-bar" small layout="prev, pager, next" :total="entityTotal"
-                         :page-size="entityPageSize" v-model:current-page="entityPage" @current-change="loadEntities" />
+          <template #header>
+            <div class="kg-graph-head">
+              <span>关系图谱（{{ graphCountText }}）</span>
+              <el-tag v-if="truncated" size="small" type="warning">实体过多，仅展示度数最高的前 {{ MAX_NODES }} 个</el-tag>
+            </div>
+          </template>
+          <div ref="graphEl" class="kg-graph" v-loading="loading"></div>
+          <el-empty v-if="!loading && graphNodes.length === 0" description="暂无图谱数据，上传文档并完成解析后自动生成" />
         </el-card>
       </el-col>
-      <el-col :span="14">
+      <el-col :span="9">
         <el-card shadow="never" class="kg-card">
-          <template #header>关系（{{ relationTotal }}）</template>
-          <el-table :data="relations" size="small" v-loading="loading">
-            <el-table-column prop="head" label="主体" min-width="140" show-overflow-tooltip />
-            <el-table-column prop="relation" label="关系" min-width="100" show-overflow-tooltip />
-            <el-table-column prop="tail" label="客体" min-width="140" show-overflow-tooltip />
-            <el-table-column prop="chunk_id" label="来源分块" width="100" />
-          </el-table>
-          <el-pagination class="kg-page-bar" small layout="prev, pager, next" :total="relationTotal"
-                         :page-size="relationPageSize" v-model:current-page="relationPage" @current-change="loadRelations" />
+          <el-tabs v-model="sideTab">
+            <el-tab-pane :label="`实体（${entityTotal}）`" name="entity">
+              <el-table :data="entities" size="small" height="440" v-loading="loading">
+                <el-table-column prop="name" label="实体" min-width="120" show-overflow-tooltip />
+                <el-table-column prop="entity_type" label="类型" width="100" />
+                <el-table-column prop="chunk_id" label="来源分块" width="90" />
+              </el-table>
+              <el-pagination class="kg-page-bar" small layout="prev, pager, next" :total="entityTotal"
+                             :page-size="entityPageSize" v-model:current-page="entityPage" @current-change="loadEntities" />
+            </el-tab-pane>
+            <el-tab-pane :label="`关系（${relationTotal}）`" name="relation">
+              <el-table :data="relations" size="small" height="440" v-loading="loading">
+                <el-table-column prop="head" label="主体" min-width="110" show-overflow-tooltip />
+                <el-table-column prop="relation" label="关系" min-width="90" show-overflow-tooltip />
+                <el-table-column prop="tail" label="客体" min-width="110" show-overflow-tooltip />
+              </el-table>
+              <el-pagination class="kg-page-bar" small layout="prev, pager, next" :total="relationTotal"
+                             :page-size="relationPageSize" v-model:current-page="relationPage" @current-change="loadRelations" />
+            </el-tab-pane>
+          </el-tabs>
         </el-card>
       </el-col>
     </el-row>
@@ -38,13 +49,23 @@
 </template>
 
 <script setup>
-import { onMounted, ref } from 'vue'
+import { computed, nextTick, onBeforeUnmount, onMounted, ref } from 'vue'
+import * as echarts from 'echarts/core'
+import { GraphChart } from 'echarts/charts'
+import { TooltipComponent, LegendComponent } from 'echarts/components'
+import { CanvasRenderer } from 'echarts/renderers'
 import { listDatasets } from '../api/dataset.js'
 import { listEntities, listRelations } from '../api/kg.js'
 
+echarts.use([GraphChart, TooltipComponent, LegendComponent, CanvasRenderer])
+
+const MAX_NODES = 400 // 力导向图节点上限，超过取度数最高者，避免布局卡顿
+
 const datasets = ref([])
 const datasetId = ref(null)
 const loading = ref(false)
+const graphEl = ref(null)
+const truncated = ref(false)
 
 const entities = ref([])
 const entityTotal = ref(0)
@@ -56,6 +77,14 @@ const relationTotal = ref(0)
 const relationPage = ref(1)
 const relationPageSize = 20
 
+const sideTab = ref('entity')
+let chart = null
+let resizeObserver = null
+
+const graphCountText = computed(() => `${truncated.value ? '≈' : ''}${graphNodes.value.length} 实体 / ${graphLinks.value.length} 关系`)
+const graphNodes = ref([])
+const graphLinks = ref([])
+
 onMounted(async () => {
   try {
     const d = await listDatasets()
@@ -65,10 +94,107 @@ onMounted(async () => {
   await load()
 })
 
+onBeforeUnmount(() => {
+  resizeObserver?.disconnect()
+  chart?.dispose()
+})
+
 async function load() {
   entityPage.value = 1
   relationPage.value = 1
-  await Promise.all([loadEntities(), loadRelations()])
+  loading.value = true
+  try {
+    const [allEntities, allRelations] = await Promise.all([
+      listEntities({ dataset_id: datasetId.value, page: 1, page_size: 100000 }),
+      listRelations({ dataset_id: datasetId.value, page: 1, page_size: 100000 }),
+    ])
+    buildGraph(allEntities?.list || [], allRelations?.list || [])
+    await Promise.all([loadEntities(), loadRelations()])
+  } finally {
+    loading.value = false
+  }
+}
+
+// 实体按名字去重合并（同一实体可能出现在多个分块），按出入度决定节点大小
+function buildGraph(entitiesList, relationsList) {
+  const byName = new Map()
+  for (const e of entitiesList) {
+    const prev = byName.get(e.name)
+    if (prev) {
+      prev.chunks = [...new Set([...prev.chunks, e.chunk_id])]
+    } else {
+      byName.set(e.name, { name: e.name, entity_type: e.entity_type, chunks: [e.chunk_id] })
+    }
+  }
+  const degree = new Map()
+  const edges = []
+  const seenEdges = new Set()
+  for (const r of relationsList) {
+    if (!byName.has(r.head) || !byName.has(r.tail)) continue
+    const key = `${r.head}|${r.relation}|${r.tail}`
+    if (seenEdges.has(key)) continue
+    seenEdges.add(key)
+    edges.push({ source: r.head, target: r.tail, relation: r.relation })
+    degree.set(r.head, (degree.get(r.head) || 0) + 1)
+    degree.set(r.tail, (degree.get(r.tail) || 0) + 1)
+  }
+  let nodes = [...byName.values()]
+  truncated.value = nodes.length > MAX_NODES
+  if (truncated.value) {
+    nodes.sort((a, b) => (degree.get(b.name) || 0) - (degree.get(a.name) || 0))
+    const keep = new Set(nodes.slice(0, MAX_NODES).map(n => n.name))
+    nodes = nodes.filter(n => keep.has(n.name))
+    for (const e of [...edges]) {
+      if (!keep.has(e.source) || !keep.has(e.target)) edges.splice(edges.indexOf(e), 1)
+    }
+  }
+  const minD = Math.min(...nodes.map(n => degree.get(n.name) || 0))
+  const maxD = Math.max(...nodes.map(n => degree.get(n.name) || 0))
+  const sizeFor = d => (maxD === minD ? 30 : 20 + ((d - minD) / (maxD - minD)) * 36)
+  graphNodes.value = nodes.map(n => ({
+    id: n.name,
+    name: n.name,
+    category: n.entity_type || '未分类',
+    symbolSize: sizeFor(degree.get(n.name) || 0),
+  }))
+  graphLinks.value = edges
+  renderGraph()
+}
+
+function renderGraph() {
+  if (!graphEl.value) return
+  chart ??= echarts.init(graphEl.value)
+  const types = [...new Set(graphNodes.value.map(n => n.category))]
+  chart.setOption({
+    legend: { bottom: 0, type: 'scroll', data: types, textStyle: { fontSize: 11 } },
+    tooltip: {
+      trigger: 'item',
+      formatter: p => {
+        if (p.dataType === 'edge') {
+          const e = graphLinks.value[p.dataIndex]
+          return `${e.source} → ${e.target}<br/>关系：${e.relation}`
+        }
+        return `${p.name}<br/>类型：${p.data.category}`
+      },
+    },
+    series: [{
+      type: 'graph',
+      layout: 'force',
+      roam: true,
+      draggable: true,
+      data: graphNodes.value,
+      links: graphLinks.value,
+      categories: types.map(t => ({ name: t })),
+      force: { repulsion: 260, edgeLength: [40, 110], gravity: 0.08 },
+      label: { show: true, position: 'right', fontSize: 10, color: '#606266' },
+      edgeLabel: { show: false },
+      emphasis: {
+        focus: 'adjacency', // 点击节点自动高亮一跳邻居并淡化其余
+        label: { fontSize: 12, fontWeight: 600 },
+        lineStyle: { width: 2 },
+      },
+    }],
+  }, true)
 }
 
 async function loadEntities() {
@@ -94,6 +220,12 @@ async function loadRelations() {
     loading.value = false
   }
 }
+
+onMounted(() => {
+  // 布局稳定后渲染一次（图表容器尺寸依赖 el-row 布局），并监听容器尺寸变化
+  resizeObserver = new ResizeObserver(() => nextTick(renderGraph))
+  resizeObserver.observe(graphEl.value)
+})
 </script>
 
 <style scoped>
@@ -107,6 +239,14 @@ async function loadRelations() {
   font-size: 13px;
   color: #909399;
 }
+.kg-graph {
+  height: 620px;
+}
+.kg-graph-head {
+  display: flex;
+  align-items: center;
+  gap: 10px;
+}
 .kg-page-bar {
   margin-top: 10px;
   justify-content: flex-end;

ui-src/src/views/Settings.vue

@@ -1,5 +1,17 @@
 <template>
   <div class="settings-page">
+    <div class="block-card">
+      <div class="block-title">分块默认值</div>
+      <div class="block-tip">新建数据集时自动带出，单个数据集仍可在表单中修改</div>
+      <div class="block-row">
+        <span class="field-label">分块大小</span>
+        <el-input-number v-model="chunkForm.chunk_size" :min="50" :max="5000" :step="50" />
+        <span class="field-label">重叠字数</span>
+        <el-input-number v-model="chunkForm.chunk_overlap" :min="0" :max="500" :step="10" />
+        <el-button type="primary" :loading="chunkSaving" @click="saveChunkDefaults">保存</el-button>
+      </div>
+    </div>
+
     <div class="section-head">
       <el-radio-group v-model="modelType" @change="loadModels">
         <el-radio-button value="chat">对话模型</el-radio-button>
@@ -65,8 +77,32 @@
 import { onMounted, ref } from 'vue'
 import { ElMessage, ElMessageBox } from 'element-plus'
 import { listModelConfigs, saveModelConfig, deleteModelConfig, testModelConfig, setDefaultModelConfig } from '../api/model_config.js'
+import { getSettings, saveSettings } from '../api/settings.js'
 
 const modelType = ref('chat')
+
+const chunkForm = ref({ chunk_size: 800, chunk_overlap: 150 })
+const chunkSaving = ref(false)
+
+async function loadChunkDefaults() {
+  try {
+    const s = await getSettings()
+    if (s) {
+      chunkForm.value.chunk_size = s.chunk_size || 800
+      chunkForm.value.chunk_overlap = s.chunk_overlap ?? 150
+    }
+  } catch { /* 忽略 */ }
+}
+
+async function saveChunkDefaults() {
+  chunkSaving.value = true
+  try {
+    await saveSettings(chunkForm.value)
+    ElMessage.success('分块默认值已保存')
+  } finally {
+    chunkSaving.value = false
+  }
+}
 const models = ref([])
 const allModels = ref([])
 const modelLoading = ref(false)
@@ -78,6 +114,7 @@ const modelForm = ref({ id: 0, name: '', model_type: 'chat', model_name: '', end
 
 onMounted(async () => {
   await loadModels()
+  await loadChunkDefaults()
   try { const all = await listModelConfigs(''); if (all && all.list) allModels.value = all.list } catch { /* 忽略 */ }
 })
 
@@ -180,4 +217,28 @@ async function test(row) {
   align-items: center;
   margin-bottom: 12px;
 }
+.block-card {
+  border: 1px solid #e4e7ed;
+  border-radius: 6px;
+  padding: 14px 16px;
+  margin-bottom: 16px;
+}
+.block-title {
+  font-weight: 600;
+  margin-bottom: 4px;
+}
+.block-tip {
+  font-size: 12px;
+  color: #909399;
+  margin-bottom: 10px;
+}
+.block-row {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+}
+.field-label {
+  color: #606266;
+  font-size: 14px;
+}
 </style>