package postgres import ( "context" "encoding/json" "errors" "strings" "testing" "time" "github.com/jackc/pgx/v5" "github.com/jackc/pgx/v5/pgconn" "knowledge-governance/internal/governance" ) func TestTriggerPPSourceScanAtomicallyEnqueuesScanJob(t *testing.T) { now := time.Date(2026, 9, 6, 4, 0, 0, 0, time.UTC) db := &recordingDB{row: sourceScanRow{id: "scan-fixed", connectionID: "conn-pp", sourceType: "pp", state: "pending", createdAt: now}} store := NewStore(db) scan, err := store.TriggerPPSourceScan(context.Background(), "conn-pp", now) if err != nil { t.Fatalf("TriggerPPSourceScan() error = %v", err) } if scan.ID != "scan-fixed" || scan.SourceConnectionID != "conn-pp" || scan.SourceType != governance.SourceTypePP || scan.State != governance.SourceScanPending { t.Fatalf("scan = %#v", scan) } sql := strings.ToLower(db.queryRowSQL) for _, required := range []string{ "insert into source_scans", "insert into jobs", "'source.scan.pp'", "'pending'", "not exists", "returning", } { if !strings.Contains(sql, required) { t.Errorf("trigger SQL missing %q", required) } } if strings.Index(sql, "insert into source_scans") > strings.Index(sql, "insert into jobs") { t.Fatal("scan must be created before its job") } if len(db.queryRowArgs) < 4 { t.Fatalf("query args = %#v", db.queryRowArgs) } payload, ok := db.queryRowArgs[2].([]byte) if !ok { t.Fatalf("payload arg type = %T", db.queryRowArgs[2]) } var decoded map[string]string if err := json.Unmarshal(payload, &decoded); err != nil || decoded["scan_id"] == "" { t.Fatalf("payload = %s err=%v", payload, err) } } func TestTriggerPPSourceScanDistinguishesAlreadyActive(t *testing.T) { db := &sourceScanSequenceDB{ recordingDB: &recordingDB{}, rows: []pgx.Row{sourceScanErrorRow{err: pgx.ErrNoRows}, sourceScanActiveRow{sourceType: "pp", active: true}}, } _, err := NewStore(db).TriggerPPSourceScan(context.Background(), "conn-pp", time.Now()) if !errors.Is(err, ErrSourceScanAlreadyActive) { t.Fatalf("TriggerPPSourceScan() error = %v, want ErrSourceScanAlreadyActive", err) } } func TestStartSourceScanTransitionsOrReturnsCompletedForRecovery(t *testing.T) { now := time.Date(2026, 9, 6, 4, 10, 0, 0, time.UTC) db := &recordingDB{row: sourceScanRow{id: "scan-1", connectionID: "conn-pp", sourceType: "pp", state: "running", createdAt: now.Add(-time.Minute)}} scan, err := NewStore(db).StartSourceScan(context.Background(), "scan-1", now) if err != nil { t.Fatalf("StartSourceScan() error = %v", err) } if scan.State != governance.SourceScanRunning { t.Fatalf("state = %q", scan.State) } sql := strings.ToLower(db.queryRowSQL) for _, required := range []string{"update source_scans", "state in ('pending', 'running')", "coalesce(started_at", "state = 'completed'", "union all"} { if !strings.Contains(sql, required) { t.Errorf("start SQL missing %q", required) } } } func TestStoreLoadsOnlyPPStableIdentityAndRevision(t *testing.T) { db := &recordingDB{rows: &ppCurrentRows{values: [][2]string{{"source-a", "revision-a"}, {"source-b", "revision-b"}}}} current, err := NewStore(db).LoadPPCurrentItems(context.Background(), "conn-pp") if err != nil { t.Fatalf("LoadPPCurrentItems() error = %v", err) } if len(current) != 2 || current["source-a"].Revision != "revision-a" || current["source-b"].Revision != "revision-b" { t.Fatalf("current = %#v", current) } sql := strings.ToLower(db.querySQL) if !strings.Contains(sql, "select stable_external_id, revision") || strings.Contains(sql, "normalized_path") || strings.Contains(sql, "raw_record_summary") { t.Fatalf("unexpected PP current query: %s", sql) } } func TestCompleteSourceScanAtomicallyPersistsChangesAndHealth(t *testing.T) { now := time.Date(2026, 9, 6, 4, 20, 0, 0, time.UTC) db := &recordingDB{row: sourceScanCompletionRow{completed: 1, changes: 2, health: 1}} result := governance.SourceScanResult{ SnapshotSHA256: strings.Repeat("f", 64), ObservedItems: 613, UnchangedItems: 558, UpdatedItems: 1, OverLimitItems: 54, MissingItems: 1, PlannedUploads: 1, Changes: []governance.SourceScanChange{ {StableExternalID: strings.Repeat("a", 64), ChangeType: governance.SourceScanUpdated, PreviousRevision: "old", ObservedRevision: "new", ObservedSizeBytes: scanInt64Pointer(10)}, {StableExternalID: strings.Repeat("b", 64), ChangeType: governance.SourceScanUnavailable, PreviousRevision: "old"}, }, } if err := NewStore(db).CompleteSourceScan(context.Background(), "scan-1", result, now); err != nil { t.Fatalf("CompleteSourceScan() error = %v", err) } sql := strings.ToLower(db.queryRowSQL) for _, required := range []string{ "update source_scans", "state = 'running'", "state = 'completed'", "insert into source_scan_changes", "jsonb_to_recordset", "left join source_items", "update source_connections", "jsonb_build_object", "'healthy'", } { if !strings.Contains(sql, required) { t.Errorf("complete SQL missing %q", required) } } for _, forbidden := range []string{"normalized_path", "raw_record_summary", "base_url", "secret_ref"} { if strings.Contains(sql, forbidden) { t.Errorf("complete SQL includes forbidden field %q", forbidden) } } } func TestFailSourceScanStoresOnlyFixedCodeAndUnhealthyState(t *testing.T) { now := time.Date(2026, 9, 6, 4, 30, 0, 0, time.UTC) db := &recordingDB{row: sourceScanFailureRow{failed: 1, health: 1}} if err := NewStore(db).FailSourceScan(context.Background(), "scan-1", "pp_snapshot_invalid", now); err != nil { t.Fatalf("FailSourceScan() error = %v", err) } sql := strings.ToLower(db.queryRowSQL) for _, required := range []string{"update source_scans", "state = 'running'", "state = 'failed'", "error_code = $2", "update source_connections", "'unhealthy'"} { if !strings.Contains(sql, required) { t.Errorf("failure SQL missing %q", required) } } if len(db.queryRowArgs) != 3 || db.queryRowArgs[1] != "pp_snapshot_invalid" { t.Fatalf("failure args = %#v", db.queryRowArgs) } } func TestListSourceScansReturnsOnlySafeSummaryColumns(t *testing.T) { now := time.Date(2026, 9, 6, 6, 0, 0, 0, time.UTC) db := &recordingDB{rows: &sourceScanSummaryRows{values: []sourceScanSummaryRow{{ id: "scan-safe", state: "completed", observed: 613, unchanged: 559, overLimit: 54, createdAt: now.Add(-time.Minute), completedAt: &now, }}}} page, err := NewStore(db).ListSourceScans(context.Background(), "conn-pp", 20, 0) if err != nil { t.Fatalf("ListSourceScans() error = %v", err) } if page.Total != 1 || len(page.Items) != 1 || page.Items[0].ID != "scan-safe" || page.Items[0].OverLimitItems != 54 { t.Fatalf("page = %#v", page) } sql := strings.ToLower(db.querySQL) for _, required := range []string{"from source_scans", "source_connection_id = $1", "count(*) over()", "limit $2 offset $3"} { if !strings.Contains(sql, required) { t.Errorf("list SQL missing %q", required) } } for _, forbidden := range []string{"snapshot_sha256", "stable_external_id", "previous_revision", "observed_revision"} { if strings.Contains(sql, forbidden) { t.Errorf("list SQL includes forbidden field %q", forbidden) } } } func TestGetSourceScanDetailReturnsSafeChangeProjection(t *testing.T) { now := time.Date(2026, 9, 6, 6, 10, 0, 0, time.UTC) db := &sourceScanReadDB{ recordingDB: &recordingDB{}, row: sourceScanDetailRow{id: "scan-safe", state: "completed", observed: 2, added: 1, missing: 1, plannedUploads: 1, createdAt: now, completedAt: &now}, rows: &sourceScanChangeRows{values: []sourceScanChangeRow{ {id: "chg-1", sourceItemID: scanStringPointer("si-safe"), changeType: "updated", observedSize: scanInt64Pointer(42), createdAt: now}, {id: "chg-2", changeType: "unavailable", createdAt: now}, }}, } detail, err := NewStore(db).GetSourceScanDetail(context.Background(), "conn-pp", "scan-safe") if err != nil { t.Fatalf("GetSourceScanDetail() error = %v", err) } if detail.ID != "scan-safe" || len(detail.Changes) != 2 || detail.Changes[0].SourceItemID == nil || *detail.Changes[0].SourceItemID != "si-safe" { t.Fatalf("detail = %#v", detail) } joined := strings.ToLower(db.queryRowSQL + "\n" + db.querySQL) for _, required := range []string{"source_connection_id = $1", "id = $2", "from source_scan_changes", "source_scan_id = $2"} { if !strings.Contains(joined, required) { t.Errorf("detail SQL missing %q", required) } } for _, forbidden := range []string{"snapshot_sha256", "stable_external_id", "previous_revision", "observed_revision"} { if strings.Contains(joined, forbidden) { t.Errorf("detail SQL includes forbidden field %q", forbidden) } } } func scanInt64Pointer(value int64) *int64 { return &value } func scanStringPointer(value string) *string { return &value } type sourceScanSequenceDB struct { *recordingDB rows []pgx.Row index int } func (d *sourceScanSequenceDB) QueryRow(_ context.Context, _ string, _ ...any) pgx.Row { row := d.rows[d.index] d.index++ return row } type sourceScanErrorRow struct{ err error } func (r sourceScanErrorRow) Scan(...any) error { return r.err } type sourceScanActiveRow struct { sourceType string active bool } func (r sourceScanActiveRow) Scan(dest ...any) error { *dest[0].(*string) = r.sourceType *dest[1].(*bool) = r.active return nil } type sourceScanRow struct { id string connectionID string sourceType string state string createdAt time.Time } func (r sourceScanRow) Scan(dest ...any) error { *dest[0].(*string) = r.id *dest[1].(*string) = r.connectionID *dest[2].(*string) = r.sourceType *dest[3].(*string) = r.state *dest[4].(*time.Time) = r.createdAt return nil } type sourceScanCompletionRow struct{ completed, changes, health int } func (r sourceScanCompletionRow) Scan(dest ...any) error { *dest[0].(*int) = r.completed *dest[1].(*int) = r.changes *dest[2].(*int) = r.health return nil } type sourceScanFailureRow struct{ failed, health int } func (r sourceScanFailureRow) Scan(dest ...any) error { *dest[0].(*int) = r.failed *dest[1].(*int) = r.health return nil } type ppCurrentRows struct { values [][2]string index int } func (r *ppCurrentRows) Close() {} func (r *ppCurrentRows) Err() error { return nil } func (r *ppCurrentRows) CommandTag() pgconn.CommandTag { return pgconn.CommandTag{} } func (r *ppCurrentRows) FieldDescriptions() []pgconn.FieldDescription { return nil } func (r *ppCurrentRows) RawValues() [][]byte { return nil } func (r *ppCurrentRows) Conn() *pgx.Conn { return nil } func (r *ppCurrentRows) Next() bool { if r.index >= len(r.values) { return false } r.index++ return true } func (r *ppCurrentRows) Scan(dest ...any) error { if r.index == 0 || r.index > len(r.values) { return errors.New("scan outside row") } *dest[0].(*string) = r.values[r.index-1][0] *dest[1].(*string) = r.values[r.index-1][1] return nil } func (r *ppCurrentRows) Values() ([]any, error) { return nil, nil } type sourceScanSummaryRow struct { id, state string observed, unchanged, added, updated, overLimit, missing int plannedUploads int errorCode string createdAt time.Time completedAt *time.Time } type sourceScanSummaryRows struct { values []sourceScanSummaryRow index int } func (r *sourceScanSummaryRows) Close() {} func (r *sourceScanSummaryRows) Err() error { return nil } func (r *sourceScanSummaryRows) CommandTag() pgconn.CommandTag { return pgconn.CommandTag{} } func (r *sourceScanSummaryRows) FieldDescriptions() []pgconn.FieldDescription { return nil } func (r *sourceScanSummaryRows) RawValues() [][]byte { return nil } func (r *sourceScanSummaryRows) Conn() *pgx.Conn { return nil } func (r *sourceScanSummaryRows) Next() bool { if r.index >= len(r.values) { return false } r.index++ return true } func (r *sourceScanSummaryRows) Scan(dest ...any) error { v := r.values[r.index-1] *dest[0].(*string), *dest[1].(*string) = v.id, v.state *dest[2].(*int), *dest[3].(*int), *dest[4].(*int), *dest[5].(*int) = v.observed, v.unchanged, v.added, v.updated *dest[6].(*int), *dest[7].(*int), *dest[8].(*int) = v.overLimit, v.missing, v.plannedUploads *dest[9].(*string), *dest[10].(*time.Time), *dest[11].(**time.Time) = v.errorCode, v.createdAt, v.completedAt *dest[12].(*int) = len(r.values) return nil } func (r *sourceScanSummaryRows) Values() ([]any, error) { return nil, nil } type sourceScanDetailRow sourceScanSummaryRow func (r sourceScanDetailRow) Scan(dest ...any) error { v := sourceScanSummaryRow(r) *dest[0].(*string), *dest[1].(*string) = v.id, v.state *dest[2].(*int), *dest[3].(*int), *dest[4].(*int), *dest[5].(*int) = v.observed, v.unchanged, v.added, v.updated *dest[6].(*int), *dest[7].(*int), *dest[8].(*int) = v.overLimit, v.missing, v.plannedUploads *dest[9].(*string), *dest[10].(*time.Time), *dest[11].(**time.Time) = v.errorCode, v.createdAt, v.completedAt return nil } type sourceScanChangeRow struct { id string sourceItemID *string changeType string observedSize *int64 createdAt time.Time } type sourceScanChangeRows struct { values []sourceScanChangeRow index int } func (r *sourceScanChangeRows) Close() {} func (r *sourceScanChangeRows) Err() error { return nil } func (r *sourceScanChangeRows) CommandTag() pgconn.CommandTag { return pgconn.CommandTag{} } func (r *sourceScanChangeRows) FieldDescriptions() []pgconn.FieldDescription { return nil } func (r *sourceScanChangeRows) RawValues() [][]byte { return nil } func (r *sourceScanChangeRows) Conn() *pgx.Conn { return nil } func (r *sourceScanChangeRows) Next() bool { if r.index >= len(r.values) { return false } r.index++ return true } func (r *sourceScanChangeRows) Scan(dest ...any) error { v := r.values[r.index-1] *dest[0].(*string), *dest[1].(**string), *dest[2].(*string) = v.id, v.sourceItemID, v.changeType *dest[3].(**int64), *dest[4].(*time.Time) = v.observedSize, v.createdAt return nil } func (r *sourceScanChangeRows) Values() ([]any, error) { return nil, nil } type sourceScanReadDB struct { *recordingDB row pgx.Row rows pgx.Rows } func (d *sourceScanReadDB) QueryRow(_ context.Context, sql string, args ...any) pgx.Row { d.queryRowSQL, d.queryRowArgs = sql, args return d.row } func (d *sourceScanReadDB) Query(_ context.Context, sql string, args ...any) (pgx.Rows, error) { d.querySQL, d.queryArgs = sql, args return d.rows, nil }