feat(auto-reply): 优化自动回复逻辑和知识库功能

- 将默认回复详细程度从"detailed"调整为"medium"，前后端保持一致 - 新增话题切换检测逻辑，当用户主动要求换话题时提供引导回复 - 优化上下文处理机制，仅在指代型追问时注入历史对话，避免模型复读旧内容 - 改进知识库检索逻辑，区分自包含问题和指代型问题的上下文需求 - 完善知识库完整性指令，确保回复详细程度与知识展开程度一致 - 重构知识库重建逻辑，支持递归扫描子目录中的文件，修复索引为空的问题 - 增强素材匹配算法，引入强信号检测机制，避免仅凭模糊匹配误发素材 - 新增素材开场白AI生成功能，支持图片、视频、文档等类型智能描述 - 改进知识库重建通知，显示具体的文件数、分片数及失败统计信息
2026-06-26 14:25:35 +08:00
parent 1517be2a25
commit 849090a627
12 changed files with 809 additions and 40 deletions
--- a/helper/auto_reply_test.go
+++ b/helper/auto_reply_test.go
@@ -473,7 +473,7 @@ func TestSyncAutoReplyMaterialsAddsRemovesAndKeepsExistingConfig(t *testing.T) {
 		t.Fatalf("write existing index: %v", err)
 	}

-	result, err := syncAutoReplyMaterials(dir, indexPath)
+	result, err := syncAutoReplyMaterials(dir, indexPath, nil)
 	if err != nil {
 		t.Fatalf("syncAutoReplyMaterials failed: %v", err)
 	}
@@ -847,6 +847,214 @@ func TestSpecificMaterialRequestSendsOnlyBestMatch(t *testing.T) {
 	}
 }

+// TestFuzzyOnlyMatchDoesNotSendMaterial 锁住“强信号门槛”：
+// 客户问句只与素材长标题切出的 2-gram 片段（如“数字”）模糊相交，
+// 没有整词关键词/整串标题命中时，不应误发素材。
+func TestFuzzyOnlyMatchDoesNotSendMaterial(t *testing.T) {
+	dir := t.TempDir()
+	if err := os.WriteFile(filepath.Join(dir, "企业级AI数字员工宣传手册.pptx"), []byte("file"), 0644); err != nil {
+		t.Fatalf("write material: %v", err)
+	}
+	indexPath := filepath.Join(dir, "materials.json")
+	materials := autoReplyMaterialsFile{Materials: []AutoReplyMaterial{{
+		ID:           "ai-worker-brochure",
+		Title:        "企业级AI数字员工宣传手册",
+		Keywords:     []string{"企业级AI数字员工", "AI数字员工", "宣传手册"},
+		MaterialType: "file",
+		Path:         "企业级AI数字员工宣传手册.pptx",
+		Enabled:      true,
+	}}}
+	data, err := json.Marshal(materials)
+	if err != nil {
+		t.Fatalf("marshal materials: %v", err)
+	}
+	if err := os.WriteFile(indexPath, data, 0644); err != nil {
+		t.Fatalf("write materials index: %v", err)
+	}
+
+	cfg := config.NewDefaultAutoReplyConfig()
+	cfg.Materials.Directory = dir
+	cfg.Materials.IndexPath = indexPath
+	cfg.Materials.MaxPerReply = 2
+	engine := testAutoReplyEngine(cfg)
+
+	// “数字证书”与标题只在“数字”这个 2-gram 上相交，属于弱命中，应被门槛挡掉。
+	if matches := engine.matchMaterials("发我数字证书的资料", "发我数字证书的资料", nil); len(matches) != 0 {
+		t.Fatalf("expected fuzzy-only match to be rejected, got %#v", matches)
+	}
+}
+
+func TestMaterialNeedsCaptionGeneration(t *testing.T) {
+	cases := []struct {
+		name     string
+		material AutoReplyMaterial
+		want     bool
+	}{
+		{"manual never regenerated", AutoReplyMaterial{Caption: "随手写的", CaptionSource: "manual"}, false},
+		{"ai not regenerated", AutoReplyMaterial{Caption: "已生成", CaptionSource: "ai"}, false},
+		{"empty caption needs", AutoReplyMaterial{MaterialType: "file"}, true},
+		{"typed default needs", AutoReplyMaterial{Caption: defaultMaterialCaption("image"), MaterialType: "image"}, true},
+		{"legacy generic needs", AutoReplyMaterial{Caption: "我把相关资料直接发你。"}, true},
+		{"hand-written kept", AutoReplyMaterial{Caption: "这是AgentBox产线实拍图，您看下整体布局～"}, false},
+	}
+	for _, tc := range cases {
+		if got := materialNeedsCaptionGeneration(tc.material); got != tc.want {
+			t.Fatalf("%s: materialNeedsCaptionGeneration = %v, want %v", tc.name, got, tc.want)
+		}
+	}
+}
+
+func TestSanitizeMaterialCaption(t *testing.T) {
+	if got, ok := sanitizeMaterialCaption("  “这是产品图，您看下～”  "); !ok || got != "这是产品图，您看下～" {
+		t.Fatalf("expected quotes/space stripped, got %q ok=%v", got, ok)
+	}
+	if got, ok := sanitizeMaterialCaption("第一行\n第二行"); !ok || got != "第一行 第二行" {
+		t.Fatalf("expected newline collapsed to single line, got %q ok=%v", got, ok)
+	}
+	if _, ok := sanitizeMaterialCaption("   "); ok {
+		t.Fatal("expected blank input to be rejected")
+	}
+	if _, ok := sanitizeMaterialCaption("NO_ANSWER"); ok {
+		t.Fatal("expected NO_ANSWER token to be rejected")
+	}
+	long := strings.Repeat("描述", 50)
+	got, ok := sanitizeMaterialCaption(long)
+	if !ok || len([]rune(got)) > 60 {
+		t.Fatalf("expected long caption truncated to <=60 runes, got %d runes ok=%v", len([]rune(got)), ok)
+	}
+}
+
+func TestApplyMaterialCaptionsOnlyFillsTargets(t *testing.T) {
+	materials := []AutoReplyMaterial{
+		{Path: "a.jpg", MaterialType: "image", Caption: defaultMaterialCaption("image")},
+		{Path: "b.jpg", MaterialType: "image", Caption: "运营手写不能动", CaptionSource: "manual"},
+		{Path: "c.jpg", MaterialType: "image", Caption: "上次生成的", CaptionSource: "ai"},
+	}
+	calls := 0
+	generate := func(material AutoReplyMaterial, absPath string) (string, bool) {
+		calls++
+		return "生成给-" + material.Path, true
+	}
+	applyMaterialCaptions(materials, t.TempDir(), generate)
+
+	if calls != 1 {
+		t.Fatalf("expected generator called once (only the default-caption item), got %d", calls)
+	}
+	if materials[0].Caption != "生成给-a.jpg" || materials[0].CaptionSource != "ai" {
+		t.Fatalf("expected a.jpg regenerated and marked ai, got %#v", materials[0])
+	}
+	if materials[1].Caption != "运营手写不能动" || materials[1].CaptionSource != "manual" {
+		t.Fatalf("manual caption must be preserved, got %#v", materials[1])
+	}
+	if materials[2].Caption != "上次生成的" {
+		t.Fatalf("existing ai caption must not be regenerated, got %#v", materials[2])
+	}
+}
+
+func TestApplyMaterialCaptionsKeepsOriginalOnFailure(t *testing.T) {
+	materials := []AutoReplyMaterial{
+		{Path: "a.jpg", MaterialType: "image", Caption: defaultMaterialCaption("image")},
+	}
+	generate := func(material AutoReplyMaterial, absPath string) (string, bool) {
+		return "", false // 模拟生成失败
+	}
+	applyMaterialCaptions(materials, t.TempDir(), generate)
+
+	if materials[0].Caption != defaultMaterialCaption("image") || materials[0].CaptionSource != "" {
+		t.Fatalf("expected failed generation to leave caption untouched, got %#v", materials[0])
+	}
+}
+
+// TestSyncAutoReplyMaterialsGeneratesCaptionsEndToEnd 串起整条同步链路（mock 生成器，不调真实 AI）：
+// 真实落盘文件 → 扫描发现 → 经过生成器 → 写回 materials.json → 重新加载校验。
+func TestSyncAutoReplyMaterialsGeneratesCaptionsEndToEnd(t *testing.T) {
+	dir := t.TempDir()
+	// 新图片素材：无既有索引，应被生成器赋予描述。
+	if err := os.WriteFile(filepath.Join(dir, "产线实拍.jpg"), []byte("jpg"), 0644); err != nil {
+		t.Fatalf("write image material: %v", err)
+	}
+	// 运营手写 caption 的素材：必须原样保留，不被生成覆盖。
+	if err := os.WriteFile(filepath.Join(dir, "报价单.pdf"), []byte("pdf"), 0644); err != nil {
+		t.Fatalf("write manual material: %v", err)
+	}
+	indexPath := filepath.Join(dir, "materials.json")
+	existing := autoReplyMaterialsFile{Materials: []AutoReplyMaterial{{
+		ID:            "manual-quote",
+		Title:         "报价单",
+		Keywords:      []string{"报价单", "报价"},
+		MaterialType:  "file",
+		Path:          "报价单.pdf",
+		Caption:       "这是最新报价，您过下目～",
+		CaptionSource: "manual",
+		Priority:      5,
+		Enabled:       true,
+	}}}
+	data, err := json.Marshal(existing)
+	if err != nil {
+		t.Fatalf("marshal existing: %v", err)
+	}
+	if err := os.WriteFile(indexPath, data, 0644); err != nil {
+		t.Fatalf("write existing index: %v", err)
+	}
+
+	// mock 生成器：记录被生成的素材路径，返回可识别的描述。
+	var generated []string
+	generate := func(material AutoReplyMaterial, absPath string) (string, bool) {
+		generated = append(generated, material.Path)
+		if _, statErr := os.Stat(absPath); statErr != nil {
+			t.Errorf("generator got unreadable absPath %q: %v", absPath, statErr)
+		}
+		return "看下这张「" + material.Title + "」～", true
+	}
+
+	result, err := syncAutoReplyMaterials(dir, indexPath, generate)
+	if err != nil {
+		t.Fatalf("sync failed: %v", err)
+	}
+	if result.Total != 2 {
+		t.Fatalf("expected 2 materials total, got %#v", result)
+	}
+
+	// 只有新图片应触发生成，手写素材不触发。
+	if len(generated) != 1 || generated[0] != "产线实拍.jpg" {
+		t.Fatalf("expected only the new image to be generated, got %#v", generated)
+	}
+
+	got, err := loadAutoReplyMaterials(indexPath)
+	if err != nil {
+		t.Fatalf("reload synced materials: %v", err)
+	}
+	byPath := make(map[string]AutoReplyMaterial, len(got))
+	for _, item := range got {
+		byPath[item.Path] = item
+	}
+
+	image, ok := byPath["产线实拍.jpg"]
+	if !ok {
+		t.Fatalf("image material missing after sync: %#v", got)
+	}
+	if image.Caption != "看下这张「产线实拍」～" || image.CaptionSource != "ai" {
+		t.Fatalf("expected generated caption marked ai, got %#v", image)
+	}
+
+	manual, ok := byPath["报价单.pdf"]
+	if !ok {
+		t.Fatalf("manual material missing after sync: %#v", got)
+	}
+	if manual.Caption != "这是最新报价，您过下目～" || manual.CaptionSource != "manual" {
+		t.Fatalf("manual caption must survive sync untouched, got %#v", manual)
+	}
+
+	// 再同步一次：ai 描述已存在，不应重复调用生成器。
+	generated = nil
+	if _, err := syncAutoReplyMaterials(dir, indexPath, generate); err != nil {
+		t.Fatalf("second sync failed: %v", err)
+	}
+	if len(generated) != 0 {
+		t.Fatalf("expected no regeneration on second sync, got %#v", generated)
+	}
+}
+
 func TestPromptLeakageAnswerIsSanitized(t *testing.T) {
 	answer := "您好，我是企业微信智能客服。\n话语规则：只用第一人称，不要说本系统、本AI。你的目标是让客户感觉自己在和这家公司的人对话。根据知识库回答。"
 	cfg := config.NewDefaultAutoReplyConfig()
@@ -1585,6 +1793,64 @@ func TestContextualSearchTextIncludesRecentQuestion(t *testing.T) {
 	}
 }

+// 自包含问题（如“今天星期几”）不应把上一个话题的对话拼进检索 query，
+// 否则会把旧话题的知识高分召回，导致顺着旧话题继续答。
+func TestContextualSearchTextSkipsContextForSelfContainedQuestion(t *testing.T) {
+	withTestContextCachePath(t)
+	cfg := config.NewDefaultAutoReplyConfig()
+	engine := testAutoReplyEngine(cfg)
+	prev := autoReplyMessage{ClientID: 7, RobotID: "robot-user", ConversationID: "S:robot-user_customer-user", FromWxID: "customer-user", Content: "IRB 1200是什么"}
+	engine.rememberUserMessage(prev)
+	engine.rememberAssistantMessage(prev, "IRB 1200是一款紧凑型6轴工业机器人，重复定位精度±0.02mm。")
+
+	question := "今天星期几"
+	searchText := engine.contextualSearchText(question, autoReplyMessage{ClientID: 7, RobotID: "robot-user", ConversationID: "S:robot-user_customer-user", FromWxID: "customer-user", Content: question})
+	if searchText != question {
+		t.Fatalf("self-contained question should not carry previous topic into search, got %q", searchText)
+	}
+	if strings.Contains(searchText, "IRB") {
+		t.Fatalf("search text leaked previous topic: %q", searchText)
+	}
+}
+
+func TestQuestionReferencesContext(t *testing.T) {
+	cases := []struct {
+		question string
+		want     bool
+	}{
+		{"它多少钱", true},        // 它多少钱
+		{"这个怎么用", true},  // 这个怎么用
+		{"刚才那个再说说", true}, // 刚才那个再说说
+		{"继续", true},                    // 继续
+		{"今天星期几", false}, // 今天星期几
+		{"你们有什么产品", false}, // 你们有什么产品
+		{"换个话题吧", false}, // 换个话题吧
+	}
+	for _, c := range cases {
+		if got := questionReferencesContext(c.question); got != c.want {
+			t.Errorf("questionReferencesContext(%q)=%v, want %v", c.question, got, c.want)
+		}
+	}
+}
+
+func TestIsPureTopicSwitchMessage(t *testing.T) {
+	cases := []struct {
+		content string
+		want    bool
+	}{
+		{"换个话题吧", true},          // 换个话题吧
+		{"我们聊点别的", true},    // 我们聊点别的
+		{"不聊这个了", true},          // 不聊这个了
+		{"换个话题，你们产品多少钱", false}, // 换个话题，你们产品多少钱（带了新问题）
+		{"今天星期几", false},         // 今天星期几
+	}
+	for _, c := range cases {
+		if got := isPureTopicSwitchMessage(c.content); got != c.want {
+			t.Errorf("isPureTopicSwitchMessage(%q)=%v, want %v", c.content, got, c.want)
+		}
+	}
+}
+
 func TestImageRecognitionContentEntersNormalReplyFlow(t *testing.T) {
 	withTestContextCachePath(t)
 	restoreClients := setTestIdentifiedClients(t, map[uint32]string{7: "robot-user"})
@@ -2363,8 +2629,8 @@ func TestFastAutoReplyDefaults(t *testing.T) {
 	if cfg.AI.MaxTokens != 700 {
 		t.Fatalf("expected 700 max tokens, got %d", cfg.AI.MaxTokens)
 	}
-	if cfg.AI.ReplyDetail != "detailed" {
-		t.Fatalf("expected detailed reply detail, got %s", cfg.AI.ReplyDetail)
+	if cfg.AI.ReplyDetail != "medium" {
+		t.Fatalf("expected medium reply detail, got %s", cfg.AI.ReplyDetail)
 	}
 	if cfg.AI.EnableThinking {
 		t.Fatal("expected thinking to be disabled by default")