migrate

tools/claims-migrate/migrate_test.go

@@ -373,3 +373,71 @@ func TestHalfFloat(t *testing.T) {
 		}
 	}
 }
+
+// Two records listing the same cid in cids/* with no cid: index row (the same
+// content stored twice): the first keeps it, a record left with no cid of its
+// own folds into it, and the duplicate (cid, key) claims are skipped.
+func TestSharedUnindexedCIDFolds(t *testing.T) {
+	dir := t.TempDir()
+	oldPath := filepath.Join(dir, "v1.db")
+	db, err := sql.Open("sqlite", "file:"+oldPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if _, err := db.Exec(`CREATE TABLE kv(k TEXT PRIMARY KEY, v TEXT NOT NULL) WITHOUT ROWID`); err != nil {
+		t.Fatal(err)
+	}
+	for k, v := range map[string]string{
+		"meta-core:schema-version":         "2",
+		"file:" + ridA + "/cids/" + cidSHA: "true",
+		"file:" + ridA + "/title":          "A",
+		"file:" + ridB + "/cids/" + cidSHA: "true",
+		"file:" + ridB + "/title":          "B",
+		"file:" + ridB + "/genres/Drama":   "true",
+		"file:" + ridC + "/cids/" + cidSHA: "true",
+		"file:" + ridC + "/cids/" + cidB:   "true",
+		"file:" + ridC + "/title":          "C",
+	} {
+		if _, err := db.Exec(`INSERT INTO kv(k, v) VALUES(?, ?)`, k, v); err != nil {
+			t.Fatal(err)
+		}
+	}
+	db.Close()
+	truthPath := filepath.Join(dir, "truth.db")
+	st, err := migrate(options{oldPath: oldPath, truthPath: truthPath, workers: 2})
+	if err != nil {
+		t.Fatalf("migrate: %v", err)
+	}
+	if st.Rids != 2 || st.RecordsFolded != 1 || st.CIDShared != 2 || st.DupClaims != 1 {
+		t.Errorf("rids %d folded %d shared %d dups %d", st.Rids, st.RecordsFolded, st.CIDShared, st.DupClaims)
+	}
+	out, err := sql.Open("sqlite", "file:"+truthPath+"?mode=ro")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer out.Close()
+	groups := map[string]string{}
+	rows, err := out.Query(`SELECT cid, rid FROM cid_group`)
+	if err != nil {
+		t.Fatal(err)
+	}
+	for rows.Next() {
+		var c, r string
+		if err := rows.Scan(&c, &r); err != nil {
+			t.Fatal(err)
+		}
+		groups[c] = r
+	}
+	rows.Close()
+	if groups[cidSHA] != ridA || groups[cidB] != ridC || len(groups) != 2 {
+		t.Errorf("cid_group %v", groups)
+	}
+	var n int
+	if err := out.QueryRow(`SELECT count(*) FROM claim WHERE cid = ? AND key IN ('title', 'genres/Drama')`, cidSHA).Scan(&n); err != nil || n != 2 {
+		t.Errorf("claims on the shared cid: %d (%v)", n, err)
+	}
+	var title string
+	if err := out.QueryRow(`SELECT value FROM claim WHERE cid = ? AND key = 'title'`, cidB).Scan(&title); err != nil || title != "C" {
+		t.Errorf("record C title on its own cid: %q (%v)", title, err)
+	}
+}