knowledge-governance/internal/librarydiff/wiki_test.go

package librarydiff

import (
	"encoding/json"
	"testing"
)

// 两篇文档共同贡献一页：撤销一篇后，页面保留、只掉撤销那篇的来源与引用，
// 并且能说清是哪几类语义变了——这正是"页面级差异"做不到的部分。
func TestCompareWikiReportsContributionRemoval(t *testing.T) {
	base := []map[string]any{
		row(t, "entity/gate", "gate", []string{"doc-a", "doc-b"}, []string{"chunk-a", "chunk-b"},
			"闸门：灰度 30 分钟 ZEBRA-A-7731；阈值 5% ZEBRA-B-4492", []string{"entity/only-a"}, []string{"entity/gate"}),
		row(t, "entity/only-a", "only-a", []string{"doc-a"}, []string{"chunk-a"}, "只属于 A", nil, nil),
	}
	after := []map[string]any{
		row(t, "entity/gate", "gate", []string{"doc-b"}, []string{"chunk-b"},
			"闸门：阈值 5% ZEBRA-B-4492", []string{"entity/gate"}, []string{"entity/gate"}),
	}
	result, err := CompareWiki(base, after)
	if err != nil {
		t.Fatalf("compare: %v", err)
	}
	if result.Updated != 1 || result.Removed != 1 || result.Added != 0 {
		t.Fatalf("counts: %+v", result)
	}
	var kept WikiContribution
	for _, page := range result.Pages {
		if page.Slug == "entity/gate" {
			kept = page
		}
	}
	if kept.Status != "updated" {
		t.Fatalf("kept page status = %q", kept.Status)
	}
	if len(kept.SourcesRemoved) != 1 || kept.SourcesRemoved[0] != "doc-a" {
		t.Fatalf("sources_removed = %v", kept.SourcesRemoved)
	}
	if len(kept.SourcesAdded) != 0 {
		t.Fatalf("sources_added = %v", kept.SourcesAdded)
	}
	want := map[string]bool{"content": true, "sources": true, "citations": true, "out_links": true}
	got := map[string]bool{}
	for _, name := range kept.Changed {
		got[name] = true
	}
	for name := range want {
		if !got[name] {
			t.Fatalf("changed missing %q: %v", name, kept.Changed)
		}
	}
	if kept.CitationsAfter != 1 || kept.CitationsBefore != 2 {
		t.Fatalf("citations %d -> %d", kept.CitationsBefore, kept.CitationsAfter)
	}
	if len(kept.LinksRemoved) != 1 || kept.LinksRemoved[0] != "entity/only-a" {
		t.Fatalf("links_removed = %v", kept.LinksRemoved)
	}
	// 合计口径：共享页掉 1 个来源 + 整页撤销的 only-a 带走 1 个来源 = 2。
	if result.SourcesRemoved != 2 {
		t.Fatalf("result sources_removed = %d", result.SourcesRemoved)
	}
}

func TestCompareWikiIgnoresVolatileFieldsAndUnchangedPages(t *testing.T) {
	base := []map[string]any{row(t, "concept/x", "x", []string{"doc-a"}, []string{"c1"}, "正文", nil, nil)}
	after := []map[string]any{row(t, "concept/x", "x", []string{"doc-a"}, []string{"c1"}, "正文", nil, nil)}
	after[0]["updated_at"] = "2026-10-29T00:00:00Z"
	after[0]["updated_by"] = "someone"
	result, err := CompareWiki(base, after)
	if err != nil {
		t.Fatalf("compare: %v", err)
	}
	if result.Updated+result.Added+result.Removed != 0 || len(result.Pages) != 0 {
		t.Fatalf("noise produced changes: %+v", result)
	}
	if result.State != "computed" || result.BaseCount != 1 || result.TargetCount != 1 {
		t.Fatalf("result header: %+v", result)
	}
}

func TestCompareWikiRejectsRowsWithoutSlug(t *testing.T) {
	if _, err := CompareWiki([]map[string]any{{"title": "no slug"}}, nil); err == nil {
		t.Fatal("expected error for row without slug")
	}
}

func row(t *testing.T, slug, title string, sources, chunks []string, content string,
	outLinks, inLinks []string) map[string]any {
	t.Helper()
	return map[string]any{
		"slug": slug, "title": title, "page_type": "entity", "status": "published",
		"content": content, "summary": "",
		"source_refs": sources, "chunk_refs": chunks,
		"out_links": outLinks, "in_links": inLinks,
		"category_path": []string{"根"}, "folder_id": "",
	}
}

func TestCompareWikiAcceptsJSONEncodedArrays(t *testing.T) {
	// 归档行来自 to_jsonb，数组是 JSON 数组；但历史归档里也出现过字符串化的数组，
	// 两种都要能读，读不出就报错，不能静默当成空集合。
	base := []map[string]any{{"slug": "entity/y", "source_refs": `["doc-a","doc-b"]`,
		"chunk_refs": `["c1"]`, "content": "a", "out_links": `["entity/y"]`}}
	after := []map[string]any{{"slug": "entity/y", "source_refs": `["doc-b"]`,
		"chunk_refs": `["c1"]`, "content": "a", "out_links": `["entity/y"]`}}
	result, err := CompareWiki(base, after)
	if err != nil {
		t.Fatalf("compare: %v", err)
	}
	if len(result.Pages) != 1 || len(result.Pages[0].SourcesRemoved) != 1 {
		raw, _ := json.Marshal(result)
		t.Fatalf("string arrays not parsed: %s", raw)
	}
}