knowledge-governance/internal/httpapi/library_snapshots_test.go

package httpapi

import (
	"bytes"
	"fmt"
	"os"
	"path/filepath"
	"strings"
	"testing"
	"time"

	"knowledge-governance/internal/librarysnapshot"
)

// The 整库版本 page reads a sealed archive through this exact function. A
// synthetic archive keeps the assertion about what the page shows independent of
// whether a real capture has run on this host.
func TestReadLibrarySnapshotsReportsSealedTotals(t *testing.T) {
	root, err := os.MkdirTemp(os.TempDir(), "kglib-")
	if err != nil {
		t.Fatal(err)
	}
	defer os.RemoveAll(root)
	if err = os.Chmod(root, 0700); err != nil {
		t.Fatal(err)
	}
	archive, err := librarysnapshot.Open(root, 1024*1024)
	if err != nil {
		t.Fatal(err)
	}
	// The archive re-derives row counts from the stored bytes, so the bodies must
	// carry exactly the declared number of rows.
	rows := func(count int) string {
		if count == 0 {
			return "[]"
		}
		items := make([]string, count)
		for i := range items {
			items[i] = fmt.Sprintf(`{"id":"00000000-0000-4000-8000-%012d"}`, i)
		}
		return "[" + strings.Join(items, ",") + "]"
	}
	parts := []struct {
		kind, name string
		count      int64
		body       string
	}{
		{"database", "knowledges", 4, rows(4)},
		{"database", "chunks", 38, rows(38)},
		{"database", "resources", 21, rows(21)},
		{"database", "resource_bindings", 4, rows(4)},
		{"database", "knowledge_tags", 0, rows(0)},
		{"database", "knowledge_tag_relations", 0, rows(0)},
		{"index", "chunks", 122, rows(122)},
		{"wiki", "wiki_pages", 31, rows(31)},
		{"wiki", "wiki_folders", 21, rows(21)},
		{"wiki", "wiki_page_issues", 0, rows(0)},
		{"wiki", "wiki_log_entries", 11, rows(11)},
	}
	// Every resource row must have its own archived object: the archive refuses a
	// snapshot whose resource dependencies are incomplete.
	for i := 0; i < 21; i++ {
		parts = append(parts, struct {
			kind, name string
			count      int64
			body       string
		}{"resource", fmt.Sprintf("00000000-0000-4000-8000-%012d", i), 1, fmt.Sprintf("object-%d", i)})
	}
	manifest := librarysnapshot.Manifest{ValidationPolicy: librarysnapshot.ArtifactAcceptancePolicy,
		Format: librarysnapshot.Format, Scope: "internal", TenantID: 10000,
		KnowledgeBaseID: "0c84a098-d3ed-4b82-a165-e6de9691d6a8", CreatedAt: time.Now().UTC(),
		Compatibility: librarysnapshot.Compatibility{WeKnoraVersion: "v0.7.1", SchemaVersion: 74,
			RetrieverEngine: "elasticsearch_v8", EmbeddingModel: "text-embedding-v4", EmbeddingDimension: 1024,
			IndexMappingSHA256: "137cde2965f3b56521f5fb722b4e0617f46d0e509e571b217fe01e090d5e4d01"}}
	for _, part := range parts {
		stored, err := archive.Put(bytes.NewReader([]byte(part.body)))
		if err != nil {
			t.Fatal(err)
		}
		stored.Kind, stored.Name, stored.Count = part.kind, part.name, part.count
		manifest.Parts = append(manifest.Parts, stored)
	}
	snapshotID, err := archive.Seal(manifest)
	if err != nil {
		t.Fatal(err)
	}
	if err = archive.Close(); err != nil {
		t.Fatal(err)
	}
	list, err := readLibrarySnapshots(t.Context(), root)
	if err != nil {
		t.Fatal(err)
	}
	if !list.Available || len(list.Snapshots) != 1 || list.Snapshots[0].SnapshotID != snapshotID {
		t.Fatalf("unexpected list %+v", list)
	}
	snapshot := list.Snapshots[0]
	want := map[string]int64{"documents": 4, "chunks": 38, "index_documents": 122, "resources": 21, "wiki_pages": 31}
	for key, value := range want {
		if snapshot.Totals[key] != value {
			t.Fatalf("total %s = %d, want %d", key, snapshot.Totals[key], value)
		}
	}
	// The archive-wide content figure deduplicates blobs across versions, so it
	// can be smaller than a version's logical sum but never larger.
	if snapshot.LogicalBytes <= 0 || list.ContentBytes <= 0 || list.ContentBytes > snapshot.LogicalBytes {
		t.Fatalf("bytes not reported: logical=%d content=%d", snapshot.LogicalBytes, list.ContentBytes)
	}
	if list.BlobAllocatedBytes < list.BlobFileBytes || list.ManifestFileBytes <= 0 {
		t.Fatalf("filesystem facts not separated: %+v", list)
	}
	if snapshot.Compatibility.EmbeddingDimension != 1024 || snapshot.ValidationPolicy != librarysnapshot.ArtifactAcceptancePolicy {
		t.Fatalf("contract not reported: %+v", snapshot)
	}
	if list.CaptureNote == "" {
		t.Fatal("page must explain who runs the capture")
	}
	if _, err = os.Stat(filepath.Join(root, "versions", snapshotID+".json")); err != nil {
		t.Fatalf("sealed manifest missing: %v", err)
	}
}

func TestReadLibrarySnapshotsWithoutArchiveIsExplicitlyUnwired(t *testing.T) {
	list, err := readLibrarySnapshots(t.Context(), "")
	if err != nil {
		t.Fatal(err)
	}
	if list.Available || list.Reason == "" || len(list.Snapshots) != 0 {
		t.Fatalf("empty configuration must be reported as unwired, got %+v", list)
	}
}