geo/server/internal/tenant/app/knowledge_prompt_test.go

package app

import (
	"strings"
	"testing"

	"github.com/geo-platform/tenant-api/internal/shared/retrieval"
)

func TestExtractKnowledgePreciseFactsKeepsAddressAndContactLines(t *testing.T) {
	facts := extractKnowledgePreciseFacts(`
# 合肥门店

地址：合肥市蜀山区望江西路 123 号 A 座 8 层
电话：0551-12345678
官网：https://example.com/store
普通介绍段落
`)

	for _, expected := range []string{
		"地址：合肥市蜀山区望江西路 123 号 A 座 8 层",
		"电话：0551-12345678",
		"官网：https://example.com/store",
	} {
		if !containsExactString(facts, expected) {
			t.Fatalf("extractKnowledgePreciseFacts() = %#v, want %q", facts, expected)
		}
	}
}

func TestRenderPromptSectionIncludesPreciseFactsAndSourceURI(t *testing.T) {
	sourceURI := "https://example.com/store"
	output := (&KnowledgeService{}).RenderPromptSection(&KnowledgeContext{
		Snippets: []KnowledgeSnippet{
			{
				GroupName:    "门店信息",
				ItemName:     "合肥门店",
				SourceURI:    &sourceURI,
				Text:         "这里是一段门店介绍。",
				PreciseFacts: []string{"地址：合肥市蜀山区望江西路 123 号", "电话：0551-12345678"},
			},
		},
	})

	for _, expected := range []string{
		"以下为必须严格按原文保留的高精度事实",
		"- 门店信息 / 合肥门店：地址：合肥市蜀山区望江西路 123 号",
		"1. [K1] 门店信息 / 合肥门店",
		"来源链接: https://example.com/store",
		"这里是一段门店介绍。",
	} {
		if !strings.Contains(output, expected) {
			t.Fatalf("RenderPromptSection() = %q, want %q", output, expected)
		}
	}
}

func TestRenderPromptSectionSupportsPreciseFactsWithoutSnippets(t *testing.T) {
	output := (&KnowledgeService{}).RenderPromptSection(&KnowledgeContext{
		PreciseFacts: []string{
			"公司资料 / 合肥门店：地址：合肥市蜀山区望江西路 123 号",
			"公司资料 / 合肥门店：主营业务：全屋定制；橱柜定制",
		},
	})

	for _, expected := range []string{
		"以下为必须严格按原文保留的高精度事实",
		"- 公司资料 / 合肥门店：地址：合肥市蜀山区望江西路 123 号",
		"- 公司资料 / 合肥门店：主营业务：全屋定制；橱柜定制",
	} {
		if !strings.Contains(output, expected) {
			t.Fatalf("RenderPromptSection() = %q, want %q", output, expected)
		}
	}
}

func TestExtractKnowledgePreciseFactsKeepsBusinessScopeSection(t *testing.T) {
	facts := extractKnowledgePreciseFacts(`
# 公司介绍

主营业务
- 全屋定制
- 橱柜定制
- 木门与护墙系统

地址：合肥市蜀山区望江西路 123 号
`)

	for _, expected := range []string{
		"主营业务：全屋定制；橱柜定制；木门与护墙系统",
		"地址：合肥市蜀山区望江西路 123 号",
	} {
		if !containsExactString(facts, expected) {
			t.Fatalf("extractKnowledgePreciseFacts() = %#v, want %q", facts, expected)
		}
	}
}

func TestKnowledgeSnippetsFromSearchPointsSkipsMissingTextAndDedupes(t *testing.T) {
	snippets := dedupeKnowledgeSearchCandidates(knowledgeSnippetsFromSearchPoints([]retrieval.SearchPoint{
		{
			ID:    "point-1",
			Score: 0.91,
			Payload: map[string]any{
				"group_id":   int64(10),
				"group_name": "产品资料",
				"item_id":    int64(100),
				"item_name":  "旗舰款",
				"text":       "  支持全屋定制方案  ",
			},
		},
		{
			ID: "point-2",
			Payload: map[string]any{
				"group_id":   int64(10),
				"group_name": "产品资料",
				"item_id":    int64(100),
				"item_name":  "旗舰款",
				"text":       "支持全屋定制方案",
			},
		},
		{
			ID:      "point-3",
			Payload: map[string]any{},
		},
	}))

	if len(snippets) != 1 {
		t.Fatalf("deduped snippets length = %d, want 1: %#v", len(snippets), snippets)
	}
	if snippets[0].Text != "支持全屋定制方案" {
		t.Fatalf("snippet text = %q, want trimmed text", snippets[0].Text)
	}
}

func TestKnowledgeSearchTextIncludesMetadataForRetrieval(t *testing.T) {
	got := buildKnowledgeSearchText(knowledgeSearchTextInput{
		GroupName:  "产品资料",
		ItemName:   "旗舰款",
		SourceType: "website",
		SourceURI:  "https://example.com/product",
		Text:       "支持全屋定制方案",
	})

	for _, expected := range []string{
		"目录: 产品资料",
		"标题: 旗舰款",
		"来源类型: website",
		"来源链接: https://example.com/product",
		"内容:\n支持全屋定制方案",
	} {
		if !strings.Contains(got, expected) {
			t.Fatalf("buildKnowledgeSearchText() = %q, want %q", got, expected)
		}
	}
}

func TestKnowledgeSnippetsFromSearchPointsKeepsSearchTextForRerank(t *testing.T) {
	snippets := knowledgeSnippetsFromSearchPoints([]retrieval.SearchPoint{
		{
			ID:    "point-1",
			Score: 0.91,
			Payload: map[string]any{
				"group_id":    int64(10),
				"group_name":  "产品资料",
				"item_id":     int64(100),
				"item_name":   "旗舰款",
				"source_type": "website",
				"source_uri":  "https://example.com/product",
				"text":        "支持全屋定制方案",
				"search_text": "目录: 产品资料\n标题: 旗舰款\n内容:\n支持全屋定制方案",
			},
		},
	})

	if len(snippets) != 1 {
		t.Fatalf("snippets length = %d, want 1", len(snippets))
	}
	if snippets[0].Text != "支持全屋定制方案" {
		t.Fatalf("Text = %q, want original chunk text", snippets[0].Text)
	}
	if snippets[0].RerankText() != "目录: 产品资料\n标题: 旗舰款\n内容:\n支持全屋定制方案" {
		t.Fatalf("RerankText() = %q, want search text", snippets[0].RerankText())
	}
	if snippets[0].SourceURI == nil || *snippets[0].SourceURI != "https://example.com/product" {
		t.Fatalf("SourceURI = %#v, want payload source uri", snippets[0].SourceURI)
	}
}

func TestDedupeKnowledgeSearchCandidatesUsesMetadataAwareSearchText(t *testing.T) {
	candidates := []KnowledgeSnippet{
		{
			PointID:         "p1",
			KnowledgeItemID: 1,
			ItemName:        "A 款",
			Text:            "支持全屋定制方案",
			SearchText:      "标题: A 款\n内容:\n支持全屋定制方案",
		},
		{
			PointID:         "p2",
			KnowledgeItemID: 2,
			ItemName:        "B 款",
			Text:            "支持全屋定制方案",
			SearchText:      "标题: B 款\n内容:\n支持全屋定制方案",
		},
	}

	deduped := dedupeKnowledgeSearchCandidates(candidates)
	if len(deduped) != 2 {
		t.Fatalf("deduped length = %d, want 2: %#v", len(deduped), deduped)
	}
}

func TestSelectKnowledgeRerankedSnippetsFillsAndDiversifiesResults(t *testing.T) {
	candidates := []KnowledgeSnippet{
		{PointID: "p1", KnowledgeItemID: 1, Text: "同一文档片段 1", Score: 0.90},
		{PointID: "p2", KnowledgeItemID: 1, Text: "同一文档片段 2", Score: 0.89},
		{PointID: "p3", KnowledgeItemID: 1, Text: "同一文档片段 3", Score: 0.88},
		{PointID: "p4", KnowledgeItemID: 2, Text: "第二文档片段", Score: 0.80},
		{PointID: "p5", KnowledgeItemID: 3, Text: "第三文档片段", Score: 0.70},
	}

	selected := selectKnowledgeRerankedSnippets(candidates, []retrieval.RerankResult{
		{Index: 2, RelevanceScore: 0.99},
		{Index: 99, RelevanceScore: 1},
	}, 4)

	if len(selected) != 4 {
		t.Fatalf("selected length = %d, want 4: %#v", len(selected), selected)
	}
	if selected[0].PointID != "p3" || selected[0].Score != 0.99 {
		t.Fatalf("selected[0] = %#v, want reranked p3 with rerank score", selected[0])
	}
	if selected[1].PointID != "p1" || selected[2].PointID != "p4" || selected[3].PointID != "p5" {
		t.Fatalf("selected order = %#v, want reranked result filled with diverse vector candidates", selected)
	}
}

func TestSelectKnowledgeRerankedSnippetsKeepsOriginalChunkText(t *testing.T) {
	candidates := []KnowledgeSnippet{
		{
			PointID:         "p1",
			KnowledgeItemID: 1,
			Text:            "原始正文片段",
			SearchText:      "目录: 产品资料\n标题: 旗舰款\n内容:\n原始正文片段",
			Score:           0.90,
		},
	}

	selected := selectKnowledgeRerankedSnippets(candidates, []retrieval.RerankResult{
		{
			Index:          0,
			RelevanceScore: 0.99,
			Text:           "目录: 产品资料\n标题: 旗舰款\n内容:\n原始正文片段",
		},
	}, 1)

	if len(selected) != 1 {
		t.Fatalf("selected length = %d, want 1", len(selected))
	}
	if selected[0].Text != "原始正文片段" {
		t.Fatalf("selected text = %q, want original chunk text", selected[0].Text)
	}
	if selected[0].Score != 0.99 {
		t.Fatalf("selected score = %v, want rerank score", selected[0].Score)
	}
}

func containsExactString(items []string, target string) bool {
	for _, item := range items {
		if item == target {
			return true
		}
	}
	return false
}