// Copyright (c) 2012, Suryandaru Triandana <syndtr@gmail.com>
// All rights reserved.
//
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

package leveldb

import (
	"bytes"
	"container/list"
	crand "crypto/rand"
	"encoding/binary"
	"fmt"
	"math/rand"
	"os"
	"path/filepath"
	"runtime"
	"strings"
	"sync"
	"sync/atomic"
	"testing"
	"time"
	"unsafe"

	"github.com/onsi/gomega"

	"github.com/syndtr/goleveldb/leveldb/comparer"
	"github.com/syndtr/goleveldb/leveldb/errors"
	"github.com/syndtr/goleveldb/leveldb/filter"
	"github.com/syndtr/goleveldb/leveldb/iterator"
	"github.com/syndtr/goleveldb/leveldb/opt"
	"github.com/syndtr/goleveldb/leveldb/storage"
	"github.com/syndtr/goleveldb/leveldb/testutil"
	"github.com/syndtr/goleveldb/leveldb/util"
)

func tkey(i int) []byte {
	return []byte(fmt.Sprintf("%016d", i))
}

func tval(seed, n int) []byte {
	r := rand.New(rand.NewSource(int64(seed)))
	return randomString(r, n)
}

func testingLogger(t *testing.T) func(log string) {
	return func(log string) {
		t.Log(log)
	}
}

func testingPreserveOnFailed(t *testing.T) func() (preserve bool, err error) {
	return func() (preserve bool, err error) {
		preserve = t.Failed()
		return
	}
}

type dbHarness struct {
	t *testing.T

	stor *testutil.Storage
	db   *DB
	o    *opt.Options
	ro   *opt.ReadOptions
	wo   *opt.WriteOptions
}

func newDbHarnessWopt(t *testing.T, o *opt.Options) *dbHarness {
	h := new(dbHarness)
	h.init(t, o)
	return h
}

func newDbHarness(t *testing.T) *dbHarness {
	return newDbHarnessWopt(t, &opt.Options{})
}

func (h *dbHarness) init(t *testing.T, o *opt.Options) {
	gomega.RegisterTestingT(t)
	h.t = t
	h.stor = testutil.NewStorage()
	h.stor.OnLog(testingLogger(t))
	h.stor.OnClose(testingPreserveOnFailed(t))
	h.o = o
	h.ro = nil
	h.wo = nil

	if err := h.openDB0(); err != nil {
		// So that it will come after fatal message.
		defer h.stor.Close()
		h.t.Fatal("Open (init): got error: ", err)
	}
}

func (h *dbHarness) openDB0() (err error) {
	h.t.Log("opening DB")
	h.db, err = Open(h.stor, h.o)
	return
}

func (h *dbHarness) openDB() {
	if err := h.openDB0(); err != nil {
		h.t.Fatal("Open: got error: ", err)
	}
}

func (h *dbHarness) closeDB0() error {
	h.t.Log("closing DB")
	return h.db.Close()
}

func (h *dbHarness) closeDB() {
	if h.db != nil {
		if err := h.closeDB0(); err != nil {
			h.t.Error("Close: got error: ", err)
		}
		h.db = nil
	}
	h.stor.CloseCheck()
	runtime.GC()
}

func (h *dbHarness) reopenDB() {
	if h.db != nil {
		h.closeDB()
	}
	h.openDB()
}

func (h *dbHarness) close() {
	if h.db != nil {
		h.closeDB0()
		h.db = nil
	}
	h.stor.Close()
	h.stor = nil
	runtime.GC()
}

func (h *dbHarness) openAssert(want bool) {
	db, err := Open(h.stor, h.o)
	if err != nil {
		if want {
			h.t.Error("Open: assert: got error: ", err)
		} else {
			h.t.Log("Open: assert: got error (expected): ", err)
		}
	} else {
		if !want {
			h.t.Error("Open: assert: expect error")
		}
		db.Close()
	}
}

func (h *dbHarness) write(batch *Batch) {
	if err := h.db.Write(batch, h.wo); err != nil {
		h.t.Error("Write: got error: ", err)
	}
}

func (h *dbHarness) put(key, value string) {
	if err := h.db.Put([]byte(key), []byte(value), h.wo); err != nil {
		h.t.Error("Put: got error: ", err)
	}
}

func (h *dbHarness) putMulti(n int, low, hi string) {
	for i := 0; i < n; i++ {
		h.put(low, "begin")
		h.put(hi, "end")
		h.compactMem()
	}
}

func (h *dbHarness) maxNextLevelOverlappingBytes(want int64) {
	t := h.t
	db := h.db

	var (
		maxOverlaps int64
		maxLevel    int
	)
	v := db.s.version()
	if len(v.levels) > 2 {
		for i, tt := range v.levels[1 : len(v.levels)-1] {
			level := i + 1
			next := v.levels[level+1]
			for _, t := range tt {
				r := next.getOverlaps(nil, db.s.icmp, t.imin.ukey(), t.imax.ukey(), false)
				sum := r.size()
				if sum > maxOverlaps {
					maxOverlaps = sum
					maxLevel = level
				}
			}
		}
	}
	v.release()

	if maxOverlaps > want {
		t.Errorf("next level most overlapping bytes is more than %d, got=%d level=%d", want, maxOverlaps, maxLevel)
	} else {
		t.Logf("next level most overlapping bytes is %d, level=%d want=%d", maxOverlaps, maxLevel, want)
	}
}

func (h *dbHarness) delete(key string) {
	t := h.t
	db := h.db

	err := db.Delete([]byte(key), h.wo)
	if err != nil {
		t.Error("Delete: got error: ", err)
	}
}

func (h *dbHarness) assertNumKeys(want int) {
	iter := h.db.NewIterator(nil, h.ro)
	defer iter.Release()
	got := 0
	for iter.Next() {
		got++
	}
	if err := iter.Error(); err != nil {
		h.t.Error("assertNumKeys: ", err)
	}
	if want != got {
		h.t.Errorf("assertNumKeys: want=%d got=%d", want, got)
	}
}

func (h *dbHarness) getr(db Reader, key string, expectFound bool) (found bool, v []byte) {
	t := h.t
	v, err := db.Get([]byte(key), h.ro)
	switch err {
	case ErrNotFound:
		if expectFound {
			t.Errorf("Get: key '%s' not found, want found", key)
		}
	case nil:
		found = true
		if !expectFound {
			t.Errorf("Get: key '%s' found, want not found", key)
		}
	default:
		t.Error("Get: got error: ", err)
	}
	return
}

func (h *dbHarness) get(key string, expectFound bool) (found bool, v []byte) {
	return h.getr(h.db, key, expectFound)
}

func (h *dbHarness) getValr(db Reader, key, value string) {
	t := h.t
	found, r := h.getr(db, key, true)
	if !found {
		return
	}
	rval := string(r)
	if rval != value {
		t.Errorf("Get: invalid value, got '%s', want '%s'", rval, value)
	}
}

func (h *dbHarness) getVal(key, value string) {
	h.getValr(h.db, key, value)
}

func (h *dbHarness) allEntriesFor(key, want string) {
	t := h.t
	db := h.db
	s := db.s

	ikey := makeIkey(nil, []byte(key), kMaxSeq, ktVal)
	iter := db.newRawIterator(nil, nil, nil, nil)
	if !iter.Seek(ikey) && iter.Error() != nil {
		t.Error("AllEntries: error during seek, err: ", iter.Error())
		return
	}
	res := "[ "
	first := true
	for iter.Valid() {
		if ukey, _, kt, kerr := parseIkey(iter.Key()); kerr == nil {
			if s.icmp.uCompare(ikey.ukey(), ukey) != 0 {
				break
			}
			if !first {
				res += ", "
			}
			first = false
			switch kt {
			case ktVal:
				res += string(iter.Value())
			case ktDel:
				res += "DEL"
			}
		} else {
			if !first {
				res += ", "
			}
			first = false
			res += "CORRUPTED"
		}
		iter.Next()
	}
	if !first {
		res += " "
	}
	res += "]"
	if res != want {
		t.Errorf("AllEntries: assert failed for key %q, got=%q want=%q", key, res, want)
	}
}

// Return a string that contains all key,value pairs in order,
// formatted like "(k1->v1)(k2->v2)".
func (h *dbHarness) getKeyVal(want string) {
	t := h.t
	db := h.db

	s, err := db.GetSnapshot()
	if err != nil {
		t.Fatal("GetSnapshot: got error: ", err)
	}
	res := ""
	iter := s.NewIterator(nil, nil)
	for iter.Next() {
		res += fmt.Sprintf("(%s->%s)", string(iter.Key()), string(iter.Value()))
	}
	iter.Release()

	if res != want {
		t.Errorf("GetKeyVal: invalid key/value pair, got=%q want=%q", res, want)
	}
	s.Release()
}

func (h *dbHarness) waitCompaction() {
	t := h.t
	db := h.db
	if err := db.compTriggerWait(db.tcompCmdC); err != nil {
		t.Error("compaction error: ", err)
	}
}

func (h *dbHarness) waitMemCompaction() {
	t := h.t
	db := h.db

	if err := db.compTriggerWait(db.mcompCmdC); err != nil {
		t.Error("compaction error: ", err)
	}
}

func (h *dbHarness) compactMem() {
	t := h.t
	db := h.db

	t.Log("starting memdb compaction")

	db.writeLockC <- struct{}{}
	defer func() {
		<-db.writeLockC
	}()

	if _, err := db.rotateMem(0, true); err != nil {
		t.Error("compaction error: ", err)
	}

	if h.totalTables() == 0 {
		t.Error("zero tables after mem compaction")
	}

	t.Log("memdb compaction done")
}

func (h *dbHarness) compactRangeAtErr(level int, min, max string, wanterr bool) {
	t := h.t
	db := h.db

	var _min, _max []byte
	if min != "" {
		_min = []byte(min)
	}
	if max != "" {
		_max = []byte(max)
	}

	t.Logf("starting table range compaction: level=%d, min=%q, max=%q", level, min, max)

	if err := db.compTriggerRange(db.tcompCmdC, level, _min, _max); err != nil {
		if wanterr {
			t.Log("CompactRangeAt: got error (expected): ", err)
		} else {
			t.Error("CompactRangeAt: got error: ", err)
		}
	} else if wanterr {
		t.Error("CompactRangeAt: expect error")
	}

	t.Log("table range compaction done")
}

func (h *dbHarness) compactRangeAt(level int, min, max string) {
	h.compactRangeAtErr(level, min, max, false)
}

func (h *dbHarness) compactRange(min, max string) {
	t := h.t
	db := h.db

	t.Logf("starting DB range compaction: min=%q, max=%q", min, max)

	var r util.Range
	if min != "" {
		r.Start = []byte(min)
	}
	if max != "" {
		r.Limit = []byte(max)
	}
	if err := db.CompactRange(r); err != nil {
		t.Error("CompactRange: got error: ", err)
	}

	t.Log("DB range compaction done")
}

func (h *dbHarness) sizeOf(start, limit string) uint64 {
	sz, err := h.db.SizeOf([]util.Range{
		{[]byte(start), []byte(limit)},
	})
	if err != nil {
		h.t.Error("SizeOf: got error: ", err)
	}
	return sz.Sum()
}

func (h *dbHarness) sizeAssert(start, limit string, low, hi uint64) {
	sz := h.sizeOf(start, limit)
	if sz < low || sz > hi {
		h.t.Errorf("sizeOf %q to %q not in range, want %d - %d, got %d",
			shorten(start), shorten(limit), low, hi, sz)
	}
}

func (h *dbHarness) getSnapshot() (s *Snapshot) {
	s, err := h.db.GetSnapshot()
	if err != nil {
		h.t.Fatal("GetSnapshot: got error: ", err)
	}
	return
}

func (h *dbHarness) getTablesPerLevel() string {
	res := ""
	nz := 0
	v := h.db.s.version()
	for level, tables := range v.levels {
		if level > 0 {
			res += ","
		}
		res += fmt.Sprint(len(tables))
		if len(tables) > 0 {
			nz = len(res)
		}
	}
	v.release()
	return res[:nz]
}

func (h *dbHarness) tablesPerLevel(want string) {
	res := h.getTablesPerLevel()
	if res != want {
		h.t.Errorf("invalid tables len, want=%s, got=%s", want, res)
	}
}

func (h *dbHarness) totalTables() (n int) {
	v := h.db.s.version()
	for _, tables := range v.levels {
		n += len(tables)
	}
	v.release()
	return
}

type keyValue interface {
	Key() []byte
	Value() []byte
}

func testKeyVal(t *testing.T, kv keyValue, want string) {
	res := string(kv.Key()) + "->" + string(kv.Value())
	if res != want {
		t.Errorf("invalid key/value, want=%q, got=%q", want, res)
	}
}

func numKey(num int) string {
	return fmt.Sprintf("key%06d", num)
}

var _bloom_filter = filter.NewBloomFilter(10)

func truno(t *testing.T, o *opt.Options, f func(h *dbHarness)) {
	for i := 0; i < 4; i++ {
		func() {
			switch i {
			case 0:
			case 1:
				if o == nil {
					o = &opt.Options{Filter: _bloom_filter}
				} else {
					old := o
					o = &opt.Options{}
					*o = *old
					o.Filter = _bloom_filter
				}
			case 2:
				if o == nil {
					o = &opt.Options{Compression: opt.NoCompression}
				} else {
					old := o
					o = &opt.Options{}
					*o = *old
					o.Compression = opt.NoCompression
				}
			}
			h := newDbHarnessWopt(t, o)
			defer h.close()
			switch i {
			case 3:
				h.reopenDB()
			}
			f(h)
		}()
	}
}

func trun(t *testing.T, f func(h *dbHarness)) {
	truno(t, nil, f)
}

func testAligned(t *testing.T, name string, offset uintptr) {
	if offset%8 != 0 {
		t.Errorf("field %s offset is not 64-bit aligned", name)
	}
}

func Test_FieldsAligned(t *testing.T) {
	p1 := new(DB)
	testAligned(t, "DB.seq", unsafe.Offsetof(p1.seq))
	p2 := new(session)
	testAligned(t, "session.stNextFileNum", unsafe.Offsetof(p2.stNextFileNum))
	testAligned(t, "session.stJournalNum", unsafe.Offsetof(p2.stJournalNum))
	testAligned(t, "session.stPrevJournalNum", unsafe.Offsetof(p2.stPrevJournalNum))
	testAligned(t, "session.stSeqNum", unsafe.Offsetof(p2.stSeqNum))
}

func TestDB_Locking(t *testing.T) {
	h := newDbHarness(t)
	defer h.stor.Close()
	h.openAssert(false)
	h.closeDB()
	h.openAssert(true)
}

func TestDB_Empty(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.get("foo", false)

		h.reopenDB()
		h.get("foo", false)
	})
}

func TestDB_ReadWrite(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.put("foo", "v1")
		h.getVal("foo", "v1")
		h.put("bar", "v2")
		h.put("foo", "v3")
		h.getVal("foo", "v3")
		h.getVal("bar", "v2")

		h.reopenDB()
		h.getVal("foo", "v3")
		h.getVal("bar", "v2")
	})
}

func TestDB_PutDeleteGet(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.put("foo", "v1")
		h.getVal("foo", "v1")
		h.put("foo", "v2")
		h.getVal("foo", "v2")
		h.delete("foo")
		h.get("foo", false)

		h.reopenDB()
		h.get("foo", false)
	})
}

func TestDB_EmptyBatch(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.get("foo", false)
	err := h.db.Write(new(Batch), h.wo)
	if err != nil {
		t.Error("writing empty batch yield error: ", err)
	}
	h.get("foo", false)
}

func TestDB_GetFromFrozen(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{WriteBuffer: 100100})
	defer h.close()

	h.put("foo", "v1")
	h.getVal("foo", "v1")

	h.stor.Stall(testutil.ModeSync, storage.TypeTable) // Block sync calls
	h.put("k1", strings.Repeat("x", 100000))           // Fill memtable
	h.put("k2", strings.Repeat("y", 100000))           // Trigger compaction
	for i := 0; h.db.getFrozenMem() == nil && i < 100; i++ {
		time.Sleep(10 * time.Microsecond)
	}
	if h.db.getFrozenMem() == nil {
		h.stor.Release(testutil.ModeSync, storage.TypeTable)
		t.Fatal("No frozen mem")
	}
	h.getVal("foo", "v1")
	h.stor.Release(testutil.ModeSync, storage.TypeTable) // Release sync calls

	h.reopenDB()
	h.getVal("foo", "v1")
	h.get("k1", true)
	h.get("k2", true)
}

func TestDB_GetFromTable(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.put("foo", "v1")
		h.compactMem()
		h.getVal("foo", "v1")
	})
}

func TestDB_GetSnapshot(t *testing.T) {
	trun(t, func(h *dbHarness) {
		bar := strings.Repeat("b", 200)
		h.put("foo", "v1")
		h.put(bar, "v1")

		snap, err := h.db.GetSnapshot()
		if err != nil {
			t.Fatal("GetSnapshot: got error: ", err)
		}

		h.put("foo", "v2")
		h.put(bar, "v2")

		h.getVal("foo", "v2")
		h.getVal(bar, "v2")
		h.getValr(snap, "foo", "v1")
		h.getValr(snap, bar, "v1")

		h.compactMem()

		h.getVal("foo", "v2")
		h.getVal(bar, "v2")
		h.getValr(snap, "foo", "v1")
		h.getValr(snap, bar, "v1")

		snap.Release()

		h.reopenDB()
		h.getVal("foo", "v2")
		h.getVal(bar, "v2")
	})
}

func TestDB_GetLevel0Ordering(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.db.memdbMaxLevel = 2

		for i := 0; i < 4; i++ {
			h.put("bar", fmt.Sprintf("b%d", i))
			h.put("foo", fmt.Sprintf("v%d", i))
			h.compactMem()
		}
		h.getVal("foo", "v3")
		h.getVal("bar", "b3")

		v := h.db.s.version()
		t0len := v.tLen(0)
		v.release()
		if t0len < 2 {
			t.Errorf("level-0 tables is less than 2, got %d", t0len)
		}

		h.reopenDB()
		h.getVal("foo", "v3")
		h.getVal("bar", "b3")
	})
}

func TestDB_GetOrderedByLevels(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.put("foo", "v1")
		h.compactMem()
		h.compactRange("a", "z")
		h.getVal("foo", "v1")
		h.put("foo", "v2")
		h.compactMem()
		h.getVal("foo", "v2")
	})
}

func TestDB_GetPicksCorrectFile(t *testing.T) {
	trun(t, func(h *dbHarness) {
		// Arrange to have multiple files in a non-level-0 level.
		h.put("a", "va")
		h.compactMem()
		h.compactRange("a", "b")
		h.put("x", "vx")
		h.compactMem()
		h.compactRange("x", "y")
		h.put("f", "vf")
		h.compactMem()
		h.compactRange("f", "g")

		h.getVal("a", "va")
		h.getVal("f", "vf")
		h.getVal("x", "vx")

		h.compactRange("", "")
		h.getVal("a", "va")
		h.getVal("f", "vf")
		h.getVal("x", "vx")
	})
}

func TestDB_GetEncountersEmptyLevel(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.db.memdbMaxLevel = 2

		// Arrange for the following to happen:
		//   * sstable A in level 0
		//   * nothing in level 1
		//   * sstable B in level 2
		// Then do enough Get() calls to arrange for an automatic compaction
		// of sstable A.  A bug would cause the compaction to be marked as
		// occuring at level 1 (instead of the correct level 0).

		// Step 1: First place sstables in levels 0 and 2
		for i := 0; ; i++ {
			if i >= 100 {
				t.Fatal("could not fill levels-0 and level-2")
			}
			v := h.db.s.version()
			if v.tLen(0) > 0 && v.tLen(2) > 0 {
				v.release()
				break
			}
			v.release()
			h.put("a", "begin")
			h.put("z", "end")
			h.compactMem()

			h.getVal("a", "begin")
			h.getVal("z", "end")
		}

		// Step 2: clear level 1 if necessary.
		h.compactRangeAt(1, "", "")
		h.tablesPerLevel("1,0,1")

		h.getVal("a", "begin")
		h.getVal("z", "end")

		// Step 3: read a bunch of times
		for i := 0; i < 200; i++ {
			h.get("missing", false)
		}

		// Step 4: Wait for compaction to finish
		h.waitCompaction()

		v := h.db.s.version()
		if v.tLen(0) > 0 {
			t.Errorf("level-0 tables more than 0, got %d", v.tLen(0))
		}
		v.release()

		h.getVal("a", "begin")
		h.getVal("z", "end")
	})
}

func TestDB_IterMultiWithDelete(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.put("a", "va")
		h.put("b", "vb")
		h.put("c", "vc")
		h.delete("b")
		h.get("b", false)

		iter := h.db.NewIterator(nil, nil)
		iter.Seek([]byte("c"))
		testKeyVal(t, iter, "c->vc")
		iter.Prev()
		testKeyVal(t, iter, "a->va")
		iter.Release()

		h.compactMem()

		iter = h.db.NewIterator(nil, nil)
		iter.Seek([]byte("c"))
		testKeyVal(t, iter, "c->vc")
		iter.Prev()
		testKeyVal(t, iter, "a->va")
		iter.Release()
	})
}

func TestDB_IteratorPinsRef(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.put("foo", "hello")

	// Get iterator that will yield the current contents of the DB.
	iter := h.db.NewIterator(nil, nil)

	// Write to force compactions
	h.put("foo", "newvalue1")
	for i := 0; i < 100; i++ {
		h.put(numKey(i), strings.Repeat(fmt.Sprintf("v%09d", i), 100000/10))
	}
	h.put("foo", "newvalue2")

	iter.First()
	testKeyVal(t, iter, "foo->hello")
	if iter.Next() {
		t.Errorf("expect eof")
	}
	iter.Release()
}

func TestDB_Recover(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.put("foo", "v1")
		h.put("baz", "v5")

		h.reopenDB()
		h.getVal("foo", "v1")

		h.getVal("foo", "v1")
		h.getVal("baz", "v5")
		h.put("bar", "v2")
		h.put("foo", "v3")

		h.reopenDB()
		h.getVal("foo", "v3")
		h.put("foo", "v4")
		h.getVal("foo", "v4")
		h.getVal("bar", "v2")
		h.getVal("baz", "v5")
	})
}

func TestDB_RecoverWithEmptyJournal(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.put("foo", "v1")
		h.put("foo", "v2")

		h.reopenDB()
		h.reopenDB()
		h.put("foo", "v3")

		h.reopenDB()
		h.getVal("foo", "v3")
	})
}

func TestDB_RecoverDuringMemtableCompaction(t *testing.T) {
	truno(t, &opt.Options{WriteBuffer: 1000000}, func(h *dbHarness) {

		h.stor.Stall(testutil.ModeSync, storage.TypeTable)
		h.put("big1", strings.Repeat("x", 10000000))
		h.put("big2", strings.Repeat("y", 1000))
		h.put("bar", "v2")
		h.stor.Release(testutil.ModeSync, storage.TypeTable)

		h.reopenDB()
		h.getVal("bar", "v2")
		h.getVal("big1", strings.Repeat("x", 10000000))
		h.getVal("big2", strings.Repeat("y", 1000))
	})
}

func TestDB_MinorCompactionsHappen(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{WriteBuffer: 10000})
	defer h.close()

	n := 500

	key := func(i int) string {
		return fmt.Sprintf("key%06d", i)
	}

	for i := 0; i < n; i++ {
		h.put(key(i), key(i)+strings.Repeat("v", 1000))
	}

	for i := 0; i < n; i++ {
		h.getVal(key(i), key(i)+strings.Repeat("v", 1000))
	}

	h.reopenDB()
	for i := 0; i < n; i++ {
		h.getVal(key(i), key(i)+strings.Repeat("v", 1000))
	}
}

func TestDB_RecoverWithLargeJournal(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.put("big1", strings.Repeat("1", 200000))
	h.put("big2", strings.Repeat("2", 200000))
	h.put("small3", strings.Repeat("3", 10))
	h.put("small4", strings.Repeat("4", 10))
	h.tablesPerLevel("")

	// Make sure that if we re-open with a small write buffer size that
	// we flush table files in the middle of a large journal file.
	h.o.WriteBuffer = 100000
	h.reopenDB()
	h.getVal("big1", strings.Repeat("1", 200000))
	h.getVal("big2", strings.Repeat("2", 200000))
	h.getVal("small3", strings.Repeat("3", 10))
	h.getVal("small4", strings.Repeat("4", 10))
	v := h.db.s.version()
	if v.tLen(0) <= 1 {
		t.Errorf("tables-0 less than one")
	}
	v.release()
}

func TestDB_CompactionsGenerateMultipleFiles(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		WriteBuffer: 10000000,
		Compression: opt.NoCompression,
	})
	defer h.close()

	v := h.db.s.version()
	if v.tLen(0) > 0 {
		t.Errorf("level-0 tables more than 0, got %d", v.tLen(0))
	}
	v.release()

	n := 80

	// Write 8MB (80 values, each 100K)
	for i := 0; i < n; i++ {
		h.put(numKey(i), strings.Repeat(fmt.Sprintf("v%09d", i), 100000/10))
	}

	// Reopening moves updates to level-0
	h.reopenDB()
	h.compactRangeAt(0, "", "")

	v = h.db.s.version()
	if v.tLen(0) > 0 {
		t.Errorf("level-0 tables more than 0, got %d", v.tLen(0))
	}
	if v.tLen(1) <= 1 {
		t.Errorf("level-1 tables less than 1, got %d", v.tLen(1))
	}
	v.release()

	for i := 0; i < n; i++ {
		h.getVal(numKey(i), strings.Repeat(fmt.Sprintf("v%09d", i), 100000/10))
	}
}

func TestDB_RepeatedWritesToSameKey(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{WriteBuffer: 100000})
	defer h.close()

	maxTables := h.o.GetWriteL0PauseTrigger() + 7

	value := strings.Repeat("v", 2*h.o.GetWriteBuffer())
	for i := 0; i < 5*maxTables; i++ {
		h.put("key", value)
		n := h.totalTables()
		if n > maxTables {
			t.Errorf("total tables exceed %d, got=%d, iter=%d", maxTables, n, i)
		}
	}
}

func TestDB_RepeatedWritesToSameKeyAfterReopen(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{WriteBuffer: 100000})
	defer h.close()

	h.reopenDB()

	maxTables := h.o.GetWriteL0PauseTrigger() + 7

	value := strings.Repeat("v", 2*h.o.GetWriteBuffer())
	for i := 0; i < 5*maxTables; i++ {
		h.put("key", value)
		n := h.totalTables()
		if n > maxTables {
			t.Errorf("total tables exceed %d, got=%d, iter=%d", maxTables, n, i)
		}
	}
}

func TestDB_SparseMerge(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{Compression: opt.NoCompression})
	defer h.close()

	h.putMulti(7, "A", "Z")

	// Suppose there is:
	//    small amount of data with prefix A
	//    large amount of data with prefix B
	//    small amount of data with prefix C
	// and that recent updates have made small changes to all three prefixes.
	// Check that we do not do a compaction that merges all of B in one shot.
	h.put("A", "va")
	value := strings.Repeat("x", 1000)
	for i := 0; i < 100000; i++ {
		h.put(fmt.Sprintf("B%010d", i), value)
	}
	h.put("C", "vc")
	h.compactMem()
	h.compactRangeAt(0, "", "")
	h.waitCompaction()

	// Make sparse update
	h.put("A", "va2")
	h.put("B100", "bvalue2")
	h.put("C", "vc2")
	h.compactMem()

	h.waitCompaction()
	h.maxNextLevelOverlappingBytes(20 * 1048576)
	h.compactRangeAt(0, "", "")
	h.waitCompaction()
	h.maxNextLevelOverlappingBytes(20 * 1048576)
	h.compactRangeAt(1, "", "")
	h.waitCompaction()
	h.maxNextLevelOverlappingBytes(20 * 1048576)
}

func TestDB_SizeOf(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		Compression: opt.NoCompression,
		WriteBuffer: 10000000,
	})
	defer h.close()

	h.sizeAssert("", "xyz", 0, 0)
	h.reopenDB()
	h.sizeAssert("", "xyz", 0, 0)

	// Write 8MB (80 values, each 100K)
	n := 80
	s1 := 100000
	s2 := 105000

	for i := 0; i < n; i++ {
		h.put(numKey(i), strings.Repeat(fmt.Sprintf("v%09d", i), s1/10))
	}

	// 0 because SizeOf() does not account for memtable space
	h.sizeAssert("", numKey(50), 0, 0)

	for r := 0; r < 3; r++ {
		h.reopenDB()

		for cs := 0; cs < n; cs += 10 {
			for i := 0; i < n; i += 10 {
				h.sizeAssert("", numKey(i), uint64(s1*i), uint64(s2*i))
				h.sizeAssert("", numKey(i)+".suffix", uint64(s1*(i+1)), uint64(s2*(i+1)))
				h.sizeAssert(numKey(i), numKey(i+10), uint64(s1*10), uint64(s2*10))
			}

			h.sizeAssert("", numKey(50), uint64(s1*50), uint64(s2*50))
			h.sizeAssert("", numKey(50)+".suffix", uint64(s1*50), uint64(s2*50))

			h.compactRangeAt(0, numKey(cs), numKey(cs+9))
		}

		v := h.db.s.version()
		if v.tLen(0) != 0 {
			t.Errorf("level-0 tables was not zero, got %d", v.tLen(0))
		}
		if v.tLen(1) == 0 {
			t.Error("level-1 tables was zero")
		}
		v.release()
	}
}

func TestDB_SizeOf_MixOfSmallAndLarge(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{Compression: opt.NoCompression})
	defer h.close()

	sizes := []uint64{
		10000,
		10000,
		100000,
		10000,
		100000,
		10000,
		300000,
		10000,
	}

	for i, n := range sizes {
		h.put(numKey(i), strings.Repeat(fmt.Sprintf("v%09d", i), int(n)/10))
	}

	for r := 0; r < 3; r++ {
		h.reopenDB()

		var x uint64
		for i, n := range sizes {
			y := x
			if i > 0 {
				y += 1000
			}
			h.sizeAssert("", numKey(i), x, y)
			x += n
		}

		h.sizeAssert(numKey(3), numKey(5), 110000, 111000)

		h.compactRangeAt(0, "", "")
	}
}

func TestDB_Snapshot(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.put("foo", "v1")
		s1 := h.getSnapshot()
		h.put("foo", "v2")
		s2 := h.getSnapshot()
		h.put("foo", "v3")
		s3 := h.getSnapshot()
		h.put("foo", "v4")

		h.getValr(s1, "foo", "v1")
		h.getValr(s2, "foo", "v2")
		h.getValr(s3, "foo", "v3")
		h.getVal("foo", "v4")

		s3.Release()
		h.getValr(s1, "foo", "v1")
		h.getValr(s2, "foo", "v2")
		h.getVal("foo", "v4")

		s1.Release()
		h.getValr(s2, "foo", "v2")
		h.getVal("foo", "v4")

		s2.Release()
		h.getVal("foo", "v4")
	})
}

func TestDB_SnapshotList(t *testing.T) {
	db := &DB{snapsList: list.New()}
	e0a := db.acquireSnapshot()
	e0b := db.acquireSnapshot()
	db.seq = 1
	e1 := db.acquireSnapshot()
	db.seq = 2
	e2 := db.acquireSnapshot()

	if db.minSeq() != 0 {
		t.Fatalf("invalid sequence number, got=%d", db.minSeq())
	}
	db.releaseSnapshot(e0a)
	if db.minSeq() != 0 {
		t.Fatalf("invalid sequence number, got=%d", db.minSeq())
	}
	db.releaseSnapshot(e2)
	if db.minSeq() != 0 {
		t.Fatalf("invalid sequence number, got=%d", db.minSeq())
	}
	db.releaseSnapshot(e0b)
	if db.minSeq() != 1 {
		t.Fatalf("invalid sequence number, got=%d", db.minSeq())
	}
	e2 = db.acquireSnapshot()
	if db.minSeq() != 1 {
		t.Fatalf("invalid sequence number, got=%d", db.minSeq())
	}
	db.releaseSnapshot(e1)
	if db.minSeq() != 2 {
		t.Fatalf("invalid sequence number, got=%d", db.minSeq())
	}
	db.releaseSnapshot(e2)
	if db.minSeq() != 2 {
		t.Fatalf("invalid sequence number, got=%d", db.minSeq())
	}
}

func TestDB_HiddenValuesAreRemoved(t *testing.T) {
	trun(t, func(h *dbHarness) {
		s := h.db.s

		m := 2
		h.db.memdbMaxLevel = m

		h.put("foo", "v1")
		h.compactMem()
		v := s.version()
		num := v.tLen(m)
		v.release()
		if num != 1 {
			t.Errorf("invalid level-%d len, want=1 got=%d", m, num)
		}

		// Place a table at level last-1 to prevent merging with preceding mutation
		h.put("a", "begin")
		h.put("z", "end")
		h.compactMem()
		v = s.version()
		if v.tLen(m) != 1 {
			t.Errorf("invalid level-%d len, want=1 got=%d", m, v.tLen(m))
		}
		if v.tLen(m-1) != 1 {
			t.Errorf("invalid level-%d len, want=1 got=%d", m-1, v.tLen(m-1))
		}
		v.release()

		h.delete("foo")
		h.put("foo", "v2")
		h.allEntriesFor("foo", "[ v2, DEL, v1 ]")
		h.compactMem()
		h.allEntriesFor("foo", "[ v2, DEL, v1 ]")
		h.compactRangeAt(m-2, "", "z")
		// DEL eliminated, but v1 remains because we aren't compacting that level
		// (DEL can be eliminated because v2 hides v1).
		h.allEntriesFor("foo", "[ v2, v1 ]")
		h.compactRangeAt(m-1, "", "")
		// Merging last-1 w/ last, so we are the base level for "foo", so
		// DEL is removed.  (as is v1).
		h.allEntriesFor("foo", "[ v2 ]")
	})
}

func TestDB_DeletionMarkers2(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()
	s := h.db.s

	m := 2
	h.db.memdbMaxLevel = m

	h.put("foo", "v1")
	h.compactMem()
	v := s.version()
	num := v.tLen(m)
	v.release()
	if num != 1 {
		t.Errorf("invalid level-%d len, want=1 got=%d", m, num)
	}

	// Place a table at level last-1 to prevent merging with preceding mutation
	h.put("a", "begin")
	h.put("z", "end")
	h.compactMem()
	v = s.version()
	if v.tLen(m) != 1 {
		t.Errorf("invalid level-%d len, want=1 got=%d", m, v.tLen(m))
	}
	if v.tLen(m-1) != 1 {
		t.Errorf("invalid level-%d len, want=1 got=%d", m-1, v.tLen(m-1))
	}
	v.release()

	h.delete("foo")
	h.allEntriesFor("foo", "[ DEL, v1 ]")
	h.compactMem() // Moves to level last-2
	h.allEntriesFor("foo", "[ DEL, v1 ]")
	h.compactRangeAt(m-2, "", "")
	// DEL kept: "last" file overlaps
	h.allEntriesFor("foo", "[ DEL, v1 ]")
	h.compactRangeAt(m-1, "", "")
	// Merging last-1 w/ last, so we are the base level for "foo", so
	// DEL is removed.  (as is v1).
	h.allEntriesFor("foo", "[ ]")
}

func TestDB_CompactionTableOpenError(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{OpenFilesCacheCapacity: -1})
	defer h.close()

	h.db.memdbMaxLevel = 2

	im := 10
	jm := 10
	for r := 0; r < 2; r++ {
		for i := 0; i < im; i++ {
			for j := 0; j < jm; j++ {
				h.put(fmt.Sprintf("k%d,%d", i, j), fmt.Sprintf("v%d,%d", i, j))
			}
			h.compactMem()
		}
	}

	if n := h.totalTables(); n != im*2 {
		t.Errorf("total tables is %d, want %d", n, im*2)
	}

	h.stor.EmulateError(testutil.ModeOpen, storage.TypeTable, errors.New("open error during table compaction"))
	go h.db.CompactRange(util.Range{})
	if err := h.db.compTriggerWait(h.db.tcompCmdC); err != nil {
		t.Log("compaction error: ", err)
	}
	h.closeDB0()
	h.openDB()
	h.stor.EmulateError(testutil.ModeOpen, storage.TypeTable, nil)

	for i := 0; i < im; i++ {
		for j := 0; j < jm; j++ {
			h.getVal(fmt.Sprintf("k%d,%d", i, j), fmt.Sprintf("v%d,%d", i, j))
		}
	}
}

func TestDB_OverlapInLevel0(t *testing.T) {
	trun(t, func(h *dbHarness) {
		h.db.memdbMaxLevel = 2

		// Fill levels 1 and 2 to disable the pushing of new memtables to levels > 0.
		h.put("100", "v100")
		h.put("999", "v999")
		h.compactMem()
		h.delete("100")
		h.delete("999")
		h.compactMem()
		h.tablesPerLevel("0,1,1")

		// Make files spanning the following ranges in level-0:
		//  files[0]  200 .. 900
		//  files[1]  300 .. 500
		// Note that files are sorted by min key.
		h.put("300", "v300")
		h.put("500", "v500")
		h.compactMem()
		h.put("200", "v200")
		h.put("600", "v600")
		h.put("900", "v900")
		h.compactMem()
		h.tablesPerLevel("2,1,1")

		// Compact away the placeholder files we created initially
		h.compactRangeAt(1, "", "")
		h.compactRangeAt(2, "", "")
		h.tablesPerLevel("2")

		// Do a memtable compaction.  Before bug-fix, the compaction would
		// not detect the overlap with level-0 files and would incorrectly place
		// the deletion in a deeper level.
		h.delete("600")
		h.compactMem()
		h.tablesPerLevel("3")
		h.get("600", false)
	})
}

func TestDB_L0_CompactionBug_Issue44_a(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.reopenDB()
	h.put("b", "v")
	h.reopenDB()
	h.delete("b")
	h.delete("a")
	h.reopenDB()
	h.delete("a")
	h.reopenDB()
	h.put("a", "v")
	h.reopenDB()
	h.reopenDB()
	h.getKeyVal("(a->v)")
	h.waitCompaction()
	h.getKeyVal("(a->v)")
}

func TestDB_L0_CompactionBug_Issue44_b(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.reopenDB()
	h.put("", "")
	h.reopenDB()
	h.delete("e")
	h.put("", "")
	h.reopenDB()
	h.put("c", "cv")
	h.reopenDB()
	h.put("", "")
	h.reopenDB()
	h.put("", "")
	h.waitCompaction()
	h.reopenDB()
	h.put("d", "dv")
	h.reopenDB()
	h.put("", "")
	h.reopenDB()
	h.delete("d")
	h.delete("b")
	h.reopenDB()
	h.getKeyVal("(->)(c->cv)")
	h.waitCompaction()
	h.getKeyVal("(->)(c->cv)")
}

func TestDB_SingleEntryMemCompaction(t *testing.T) {
	trun(t, func(h *dbHarness) {
		for i := 0; i < 10; i++ {
			h.put("big", strings.Repeat("v", opt.DefaultWriteBuffer))
			h.compactMem()
			h.put("key", strings.Repeat("v", opt.DefaultBlockSize))
			h.compactMem()
			h.put("k", "v")
			h.compactMem()
			h.put("", "")
			h.compactMem()
			h.put("verybig", strings.Repeat("v", opt.DefaultWriteBuffer*2))
			h.compactMem()
		}
	})
}

func TestDB_ManifestWriteError(t *testing.T) {
	for i := 0; i < 2; i++ {
		func() {
			h := newDbHarness(t)
			defer h.close()

			h.put("foo", "bar")
			h.getVal("foo", "bar")

			// Mem compaction (will succeed)
			h.compactMem()
			h.getVal("foo", "bar")
			v := h.db.s.version()
			if n := v.tLen(0); n != 1 {
				t.Errorf("invalid total tables, want=1 got=%d", n)
			}
			v.release()

			if i == 0 {
				h.stor.EmulateError(testutil.ModeWrite, storage.TypeManifest, errors.New("manifest write error"))
			} else {
				h.stor.EmulateError(testutil.ModeSync, storage.TypeManifest, errors.New("manifest sync error"))
			}

			// Merging compaction (will fail)
			h.compactRangeAtErr(0, "", "", true)

			h.db.Close()
			h.stor.EmulateError(testutil.ModeWrite, storage.TypeManifest, nil)
			h.stor.EmulateError(testutil.ModeSync, storage.TypeManifest, nil)

			// Should not lose data
			h.openDB()
			h.getVal("foo", "bar")
		}()
	}
}

func assertErr(t *testing.T, err error, wanterr bool) {
	if err != nil {
		if wanterr {
			t.Log("AssertErr: got error (expected): ", err)
		} else {
			t.Error("AssertErr: got error: ", err)
		}
	} else if wanterr {
		t.Error("AssertErr: expect error")
	}
}

func TestDB_ClosedIsClosed(t *testing.T) {
	h := newDbHarness(t)
	db := h.db

	var iter, iter2 iterator.Iterator
	var snap *Snapshot
	func() {
		defer h.close()

		h.put("k", "v")
		h.getVal("k", "v")

		iter = db.NewIterator(nil, h.ro)
		iter.Seek([]byte("k"))
		testKeyVal(t, iter, "k->v")

		var err error
		snap, err = db.GetSnapshot()
		if err != nil {
			t.Fatal("GetSnapshot: got error: ", err)
		}

		h.getValr(snap, "k", "v")

		iter2 = snap.NewIterator(nil, h.ro)
		iter2.Seek([]byte("k"))
		testKeyVal(t, iter2, "k->v")

		h.put("foo", "v2")
		h.delete("foo")

		// closing DB
		iter.Release()
		iter2.Release()
	}()

	assertErr(t, db.Put([]byte("x"), []byte("y"), h.wo), true)
	_, err := db.Get([]byte("k"), h.ro)
	assertErr(t, err, true)

	if iter.Valid() {
		t.Errorf("iter.Valid should false")
	}
	assertErr(t, iter.Error(), false)
	testKeyVal(t, iter, "->")
	if iter.Seek([]byte("k")) {
		t.Errorf("iter.Seek should false")
	}
	assertErr(t, iter.Error(), true)

	assertErr(t, iter2.Error(), false)

	_, err = snap.Get([]byte("k"), h.ro)
	assertErr(t, err, true)

	_, err = db.GetSnapshot()
	assertErr(t, err, true)

	iter3 := db.NewIterator(nil, h.ro)
	assertErr(t, iter3.Error(), true)

	iter3 = snap.NewIterator(nil, h.ro)
	assertErr(t, iter3.Error(), true)

	assertErr(t, db.Delete([]byte("k"), h.wo), true)

	_, err = db.GetProperty("leveldb.stats")
	assertErr(t, err, true)

	_, err = db.SizeOf([]util.Range{{[]byte("a"), []byte("z")}})
	assertErr(t, err, true)

	assertErr(t, db.CompactRange(util.Range{}), true)

	assertErr(t, db.Close(), true)
}

type numberComparer struct{}

func (numberComparer) num(x []byte) (n int) {
	fmt.Sscan(string(x[1:len(x)-1]), &n)
	return
}

func (numberComparer) Name() string {
	return "test.NumberComparer"
}

func (p numberComparer) Compare(a, b []byte) int {
	return p.num(a) - p.num(b)
}

func (numberComparer) Separator(dst, a, b []byte) []byte { return nil }
func (numberComparer) Successor(dst, b []byte) []byte    { return nil }

func TestDB_CustomComparer(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		Comparer:    numberComparer{},
		WriteBuffer: 1000,
	})
	defer h.close()

	h.put("[10]", "ten")
	h.put("[0x14]", "twenty")
	for i := 0; i < 2; i++ {
		h.getVal("[10]", "ten")
		h.getVal("[0xa]", "ten")
		h.getVal("[20]", "twenty")
		h.getVal("[0x14]", "twenty")
		h.get("[15]", false)
		h.get("[0xf]", false)
		h.compactMem()
		h.compactRange("[0]", "[9999]")
	}

	for n := 0; n < 2; n++ {
		for i := 0; i < 100; i++ {
			v := fmt.Sprintf("[%d]", i*10)
			h.put(v, v)
		}
		h.compactMem()
		h.compactRange("[0]", "[1000000]")
	}
}

func TestDB_ManualCompaction(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.db.memdbMaxLevel = 2

	h.putMulti(3, "p", "q")
	h.tablesPerLevel("1,1,1")

	// Compaction range falls before files
	h.compactRange("", "c")
	h.tablesPerLevel("1,1,1")

	// Compaction range falls after files
	h.compactRange("r", "z")
	h.tablesPerLevel("1,1,1")

	// Compaction range overlaps files
	h.compactRange("p1", "p9")
	h.tablesPerLevel("0,0,1")

	// Populate a different range
	h.putMulti(3, "c", "e")
	h.tablesPerLevel("1,1,2")

	// Compact just the new range
	h.compactRange("b", "f")
	h.tablesPerLevel("0,0,2")

	// Compact all
	h.putMulti(1, "a", "z")
	h.tablesPerLevel("0,1,2")
	h.compactRange("", "")
	h.tablesPerLevel("0,0,1")
}

func TestDB_BloomFilter(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		DisableBlockCache: true,
		Filter:            filter.NewBloomFilter(10),
	})
	defer h.close()

	key := func(i int) string {
		return fmt.Sprintf("key%06d", i)
	}

	const n = 10000

	// Populate multiple layers
	for i := 0; i < n; i++ {
		h.put(key(i), key(i))
	}
	h.compactMem()
	h.compactRange("a", "z")
	for i := 0; i < n; i += 100 {
		h.put(key(i), key(i))
	}
	h.compactMem()

	// Prevent auto compactions triggered by seeks
	h.stor.Stall(testutil.ModeSync, storage.TypeTable)

	// Lookup present keys. Should rarely read from small sstable.
	h.stor.ResetCounter(testutil.ModeRead, storage.TypeTable)
	for i := 0; i < n; i++ {
		h.getVal(key(i), key(i))
	}
	cnt, _ := h.stor.Counter(testutil.ModeRead, storage.TypeTable)
	t.Logf("lookup of %d present keys yield %d sstable I/O reads", n, cnt)
	if min, max := n, n+2*n/100; cnt < min || cnt > max {
		t.Errorf("num of sstable I/O reads of present keys not in range of %d - %d, got %d", min, max, cnt)
	}

	// Lookup missing keys. Should rarely read from either sstable.
	h.stor.ResetCounter(testutil.ModeRead, storage.TypeTable)
	for i := 0; i < n; i++ {
		h.get(key(i)+".missing", false)
	}
	cnt, _ = h.stor.Counter(testutil.ModeRead, storage.TypeTable)
	t.Logf("lookup of %d missing keys yield %d sstable I/O reads", n, cnt)
	if max := 3 * n / 100; cnt > max {
		t.Errorf("num of sstable I/O reads of missing keys was more than %d, got %d", max, cnt)
	}

	h.stor.Release(testutil.ModeSync, storage.TypeTable)
}

func TestDB_Concurrent(t *testing.T) {
	const n, secs, maxkey = 4, 2, 1000

	runtime.GOMAXPROCS(n)
	trun(t, func(h *dbHarness) {
		var closeWg sync.WaitGroup
		var stop uint32
		var cnt [n]uint32

		for i := 0; i < n; i++ {
			closeWg.Add(1)
			go func(i int) {
				var put, get, found uint
				defer func() {
					t.Logf("goroutine %d stopped after %d ops, put=%d get=%d found=%d missing=%d",
						i, cnt[i], put, get, found, get-found)
					closeWg.Done()
				}()

				rnd := rand.New(rand.NewSource(int64(1000 + i)))
				for atomic.LoadUint32(&stop) == 0 {
					x := cnt[i]

					k := rnd.Intn(maxkey)
					kstr := fmt.Sprintf("%016d", k)

					if (rnd.Int() % 2) > 0 {
						put++
						h.put(kstr, fmt.Sprintf("%d.%d.%-1000d", k, i, x))
					} else {
						get++
						v, err := h.db.Get([]byte(kstr), h.ro)
						if err == nil {
							found++
							rk, ri, rx := 0, -1, uint32(0)
							fmt.Sscanf(string(v), "%d.%d.%d", &rk, &ri, &rx)
							if rk != k {
								t.Errorf("invalid key want=%d got=%d", k, rk)
							}
							if ri < 0 || ri >= n {
								t.Error("invalid goroutine number: ", ri)
							} else {
								tx := atomic.LoadUint32(&(cnt[ri]))
								if rx > tx {
									t.Errorf("invalid seq number, %d > %d ", rx, tx)
								}
							}
						} else if err != ErrNotFound {
							t.Error("Get: got error: ", err)
							return
						}
					}
					atomic.AddUint32(&cnt[i], 1)
				}
			}(i)
		}

		time.Sleep(secs * time.Second)
		atomic.StoreUint32(&stop, 1)
		closeWg.Wait()
	})

	runtime.GOMAXPROCS(1)
}

func TestDB_Concurrent2(t *testing.T) {
	const n, n2 = 4, 4000

	runtime.GOMAXPROCS(n*2 + 2)
	truno(t, &opt.Options{WriteBuffer: 30}, func(h *dbHarness) {
		var closeWg sync.WaitGroup
		var stop uint32

		for i := 0; i < n; i++ {
			closeWg.Add(1)
			go func(i int) {
				for k := 0; atomic.LoadUint32(&stop) == 0; k++ {
					h.put(fmt.Sprintf("k%d", k), fmt.Sprintf("%d.%d.", k, i)+strings.Repeat("x", 10))
				}
				closeWg.Done()
			}(i)
		}

		for i := 0; i < n; i++ {
			closeWg.Add(1)
			go func(i int) {
				for k := 1000000; k < 0 || atomic.LoadUint32(&stop) == 0; k-- {
					h.put(fmt.Sprintf("k%d", k), fmt.Sprintf("%d.%d.", k, i)+strings.Repeat("x", 10))
				}
				closeWg.Done()
			}(i)
		}

		cmp := comparer.DefaultComparer
		for i := 0; i < n2; i++ {
			closeWg.Add(1)
			go func(i int) {
				it := h.db.NewIterator(nil, nil)
				var pk []byte
				for it.Next() {
					kk := it.Key()
					if cmp.Compare(kk, pk) <= 0 {
						t.Errorf("iter %d: %q is successor of %q", i, pk, kk)
					}
					pk = append(pk[:0], kk...)
					var k, vk, vi int
					if n, err := fmt.Sscanf(string(it.Key()), "k%d", &k); err != nil {
						t.Errorf("iter %d: Scanf error on key %q: %v", i, it.Key(), err)
					} else if n < 1 {
						t.Errorf("iter %d: Cannot parse key %q", i, it.Key())
					}
					if n, err := fmt.Sscanf(string(it.Value()), "%d.%d", &vk, &vi); err != nil {
						t.Errorf("iter %d: Scanf error on value %q: %v", i, it.Value(), err)
					} else if n < 2 {
						t.Errorf("iter %d: Cannot parse value %q", i, it.Value())
					}

					if vk != k {
						t.Errorf("iter %d: invalid value i=%d, want=%d got=%d", i, vi, k, vk)
					}
				}
				if err := it.Error(); err != nil {
					t.Errorf("iter %d: Got error: %v", i, err)
				}
				it.Release()
				closeWg.Done()
			}(i)
		}

		atomic.StoreUint32(&stop, 1)
		closeWg.Wait()
	})

	runtime.GOMAXPROCS(1)
}

func TestDB_CreateReopenDbOnFile(t *testing.T) {
	dbpath := filepath.Join(os.TempDir(), fmt.Sprintf("goleveldbtestCreateReopenDbOnFile-%d", os.Getuid()))
	if err := os.RemoveAll(dbpath); err != nil {
		t.Fatal("cannot remove old db: ", err)
	}
	defer os.RemoveAll(dbpath)

	for i := 0; i < 3; i++ {
		stor, err := storage.OpenFile(dbpath)
		if err != nil {
			t.Fatalf("(%d) cannot open storage: %s", i, err)
		}
		db, err := Open(stor, nil)
		if err != nil {
			t.Fatalf("(%d) cannot open db: %s", i, err)
		}
		if err := db.Put([]byte("foo"), []byte("bar"), nil); err != nil {
			t.Fatalf("(%d) cannot write to db: %s", i, err)
		}
		if err := db.Close(); err != nil {
			t.Fatalf("(%d) cannot close db: %s", i, err)
		}
		if err := stor.Close(); err != nil {
			t.Fatalf("(%d) cannot close storage: %s", i, err)
		}
	}
}

func TestDB_CreateReopenDbOnFile2(t *testing.T) {
	dbpath := filepath.Join(os.TempDir(), fmt.Sprintf("goleveldbtestCreateReopenDbOnFile2-%d", os.Getuid()))
	if err := os.RemoveAll(dbpath); err != nil {
		t.Fatal("cannot remove old db: ", err)
	}
	defer os.RemoveAll(dbpath)

	for i := 0; i < 3; i++ {
		db, err := OpenFile(dbpath, nil)
		if err != nil {
			t.Fatalf("(%d) cannot open db: %s", i, err)
		}
		if err := db.Put([]byte("foo"), []byte("bar"), nil); err != nil {
			t.Fatalf("(%d) cannot write to db: %s", i, err)
		}
		if err := db.Close(); err != nil {
			t.Fatalf("(%d) cannot close db: %s", i, err)
		}
	}
}

func TestDB_DeletionMarkersOnMemdb(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.put("foo", "v1")
	h.compactMem()
	h.delete("foo")
	h.get("foo", false)
	h.getKeyVal("")
}

func TestDB_LeveldbIssue178(t *testing.T) {
	nKeys := (opt.DefaultCompactionTableSize / 30) * 5
	key1 := func(i int) string {
		return fmt.Sprintf("my_key_%d", i)
	}
	key2 := func(i int) string {
		return fmt.Sprintf("my_key_%d_xxx", i)
	}

	// Disable compression since it affects the creation of layers and the
	// code below is trying to test against a very specific scenario.
	h := newDbHarnessWopt(t, &opt.Options{Compression: opt.NoCompression})
	defer h.close()

	// Create first key range.
	batch := new(Batch)
	for i := 0; i < nKeys; i++ {
		batch.Put([]byte(key1(i)), []byte("value for range 1 key"))
	}
	h.write(batch)

	// Create second key range.
	batch.Reset()
	for i := 0; i < nKeys; i++ {
		batch.Put([]byte(key2(i)), []byte("value for range 2 key"))
	}
	h.write(batch)

	// Delete second key range.
	batch.Reset()
	for i := 0; i < nKeys; i++ {
		batch.Delete([]byte(key2(i)))
	}
	h.write(batch)
	h.waitMemCompaction()

	// Run manual compaction.
	h.compactRange(key1(0), key1(nKeys-1))

	// Checking the keys.
	h.assertNumKeys(nKeys)
}

func TestDB_LeveldbIssue200(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.put("1", "b")
	h.put("2", "c")
	h.put("3", "d")
	h.put("4", "e")
	h.put("5", "f")

	iter := h.db.NewIterator(nil, h.ro)

	// Add an element that should not be reflected in the iterator.
	h.put("25", "cd")

	iter.Seek([]byte("5"))
	assertBytes(t, []byte("5"), iter.Key())
	iter.Prev()
	assertBytes(t, []byte("4"), iter.Key())
	iter.Prev()
	assertBytes(t, []byte("3"), iter.Key())
	iter.Next()
	assertBytes(t, []byte("4"), iter.Key())
	iter.Next()
	assertBytes(t, []byte("5"), iter.Key())
}

func TestDB_GoleveldbIssue74(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		WriteBuffer: 1 * opt.MiB,
	})
	defer h.close()

	const n, dur = 10000, 5 * time.Second

	runtime.GOMAXPROCS(runtime.NumCPU())

	until := time.Now().Add(dur)
	wg := new(sync.WaitGroup)
	wg.Add(2)
	var done uint32
	go func() {
		var i int
		defer func() {
			t.Logf("WRITER DONE #%d", i)
			atomic.StoreUint32(&done, 1)
			wg.Done()
		}()

		b := new(Batch)
		for ; time.Now().Before(until) && atomic.LoadUint32(&done) == 0; i++ {
			iv := fmt.Sprintf("VAL%010d", i)
			for k := 0; k < n; k++ {
				key := fmt.Sprintf("KEY%06d", k)
				b.Put([]byte(key), []byte(key+iv))
				b.Put([]byte(fmt.Sprintf("PTR%06d", k)), []byte(key))
			}
			h.write(b)

			b.Reset()
			snap := h.getSnapshot()
			iter := snap.NewIterator(util.BytesPrefix([]byte("PTR")), nil)
			var k int
			for ; iter.Next(); k++ {
				ptrKey := iter.Key()
				key := iter.Value()

				if _, err := snap.Get(ptrKey, nil); err != nil {
					t.Fatalf("WRITER #%d snapshot.Get %q: %v", i, ptrKey, err)
				}
				if value, err := snap.Get(key, nil); err != nil {
					t.Fatalf("WRITER #%d snapshot.Get %q: %v", i, key, err)
				} else if string(value) != string(key)+iv {
					t.Fatalf("WRITER #%d snapshot.Get %q got invalid value, want %q got %q", i, key, string(key)+iv, value)
				}

				b.Delete(key)
				b.Delete(ptrKey)
			}
			h.write(b)
			iter.Release()
			snap.Release()
			if k != n {
				t.Fatalf("#%d %d != %d", i, k, n)
			}
		}
	}()
	go func() {
		var i int
		defer func() {
			t.Logf("READER DONE #%d", i)
			atomic.StoreUint32(&done, 1)
			wg.Done()
		}()
		for ; time.Now().Before(until) && atomic.LoadUint32(&done) == 0; i++ {
			snap := h.getSnapshot()
			iter := snap.NewIterator(util.BytesPrefix([]byte("PTR")), nil)
			var prevValue string
			var k int
			for ; iter.Next(); k++ {
				ptrKey := iter.Key()
				key := iter.Value()

				if _, err := snap.Get(ptrKey, nil); err != nil {
					t.Fatalf("READER #%d snapshot.Get %q: %v", i, ptrKey, err)
				}

				if value, err := snap.Get(key, nil); err != nil {
					t.Fatalf("READER #%d snapshot.Get %q: %v", i, key, err)
				} else if prevValue != "" && string(value) != string(key)+prevValue {
					t.Fatalf("READER #%d snapshot.Get %q got invalid value, want %q got %q", i, key, string(key)+prevValue, value)
				} else {
					prevValue = string(value[len(key):])
				}
			}
			iter.Release()
			snap.Release()
			if k > 0 && k != n {
				t.Fatalf("#%d %d != %d", i, k, n)
			}
		}
	}()
	wg.Wait()
}

func TestDB_GetProperties(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	_, err := h.db.GetProperty("leveldb.num-files-at-level")
	if err == nil {
		t.Error("GetProperty() failed to detect missing level")
	}

	_, err = h.db.GetProperty("leveldb.num-files-at-level0")
	if err != nil {
		t.Error("got unexpected error", err)
	}

	_, err = h.db.GetProperty("leveldb.num-files-at-level0x")
	if err == nil {
		t.Error("GetProperty() failed to detect invalid level")
	}
}

func TestDB_GoleveldbIssue72and83(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		WriteBuffer:            1 * opt.MiB,
		OpenFilesCacheCapacity: 3,
	})
	defer h.close()

	const n, wn, dur = 10000, 100, 30 * time.Second

	runtime.GOMAXPROCS(runtime.NumCPU())

	randomData := func(prefix byte, i int) []byte {
		data := make([]byte, 1+4+32+64+32)
		_, err := crand.Reader.Read(data[1 : len(data)-8])
		if err != nil {
			panic(err)
		}
		data[0] = prefix
		binary.LittleEndian.PutUint32(data[len(data)-8:], uint32(i))
		binary.LittleEndian.PutUint32(data[len(data)-4:], util.NewCRC(data[:len(data)-4]).Value())
		return data
	}

	keys := make([][]byte, n)
	for i := range keys {
		keys[i] = randomData(1, 0)
	}

	until := time.Now().Add(dur)
	wg := new(sync.WaitGroup)
	wg.Add(3)
	var done uint32
	go func() {
		i := 0
		defer func() {
			t.Logf("WRITER DONE #%d", i)
			wg.Done()
		}()

		b := new(Batch)
		for ; i < wn && atomic.LoadUint32(&done) == 0; i++ {
			b.Reset()
			for _, k1 := range keys {
				k2 := randomData(2, i)
				b.Put(k2, randomData(42, i))
				b.Put(k1, k2)
			}
			if err := h.db.Write(b, h.wo); err != nil {
				atomic.StoreUint32(&done, 1)
				t.Fatalf("WRITER #%d db.Write: %v", i, err)
			}
		}
	}()
	go func() {
		var i int
		defer func() {
			t.Logf("READER0 DONE #%d", i)
			atomic.StoreUint32(&done, 1)
			wg.Done()
		}()
		for ; time.Now().Before(until) && atomic.LoadUint32(&done) == 0; i++ {
			snap := h.getSnapshot()
			seq := snap.elem.seq
			if seq == 0 {
				snap.Release()
				continue
			}
			iter := snap.NewIterator(util.BytesPrefix([]byte{1}), nil)
			writei := int(seq/(n*2) - 1)
			var k int
			for ; iter.Next(); k++ {
				k1 := iter.Key()
				k2 := iter.Value()
				k1checksum0 := binary.LittleEndian.Uint32(k1[len(k1)-4:])
				k1checksum1 := util.NewCRC(k1[:len(k1)-4]).Value()
				if k1checksum0 != k1checksum1 {
					t.Fatalf("READER0 #%d.%d W#%d invalid K1 checksum: %#x != %#x", i, k, k1checksum0, k1checksum0)
				}
				k2checksum0 := binary.LittleEndian.Uint32(k2[len(k2)-4:])
				k2checksum1 := util.NewCRC(k2[:len(k2)-4]).Value()
				if k2checksum0 != k2checksum1 {
					t.Fatalf("READER0 #%d.%d W#%d invalid K2 checksum: %#x != %#x", i, k, k2checksum0, k2checksum1)
				}
				kwritei := int(binary.LittleEndian.Uint32(k2[len(k2)-8:]))
				if writei != kwritei {
					t.Fatalf("READER0 #%d.%d W#%d invalid write iteration num: %d", i, k, writei, kwritei)
				}
				if _, err := snap.Get(k2, nil); err != nil {
					t.Fatalf("READER0 #%d.%d W#%d snap.Get: %v\nk1: %x\n -> k2: %x", i, k, writei, err, k1, k2)
				}
			}
			if err := iter.Error(); err != nil {
				t.Fatalf("READER0 #%d.%d W#%d snap.Iterator: %v", i, k, writei, err)
			}
			iter.Release()
			snap.Release()
			if k > 0 && k != n {
				t.Fatalf("READER0 #%d W#%d short read, got=%d want=%d", i, writei, k, n)
			}
		}
	}()
	go func() {
		var i int
		defer func() {
			t.Logf("READER1 DONE #%d", i)
			atomic.StoreUint32(&done, 1)
			wg.Done()
		}()
		for ; time.Now().Before(until) && atomic.LoadUint32(&done) == 0; i++ {
			iter := h.db.NewIterator(nil, nil)
			seq := iter.(*dbIter).seq
			if seq == 0 {
				iter.Release()
				continue
			}
			writei := int(seq/(n*2) - 1)
			var k int
			for ok := iter.Last(); ok; ok = iter.Prev() {
				k++
			}
			if err := iter.Error(); err != nil {
				t.Fatalf("READER1 #%d.%d W#%d db.Iterator: %v", i, k, writei, err)
			}
			iter.Release()
			if m := (writei+1)*n + n; k != m {
				t.Fatalf("READER1 #%d W#%d short read, got=%d want=%d", i, writei, k, m)
			}
		}
	}()

	wg.Wait()
}

func TestDB_TransientError(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		WriteBuffer:              128 * opt.KiB,
		OpenFilesCacheCapacity:   3,
		DisableCompactionBackoff: true,
	})
	defer h.close()

	const (
		nSnap = 20
		nKey  = 10000
	)

	var (
		snaps [nSnap]*Snapshot
		b     = &Batch{}
	)
	for i := range snaps {
		vtail := fmt.Sprintf("VAL%030d", i)
		b.Reset()
		for k := 0; k < nKey; k++ {
			key := fmt.Sprintf("KEY%8d", k)
			b.Put([]byte(key), []byte(key+vtail))
		}
		h.stor.EmulateError(testutil.ModeOpen|testutil.ModeRead, storage.TypeTable, errors.New("table transient read error"))
		if err := h.db.Write(b, nil); err != nil {
			t.Logf("WRITE #%d error: %v", i, err)
			h.stor.EmulateError(testutil.ModeOpen|testutil.ModeRead, storage.TypeTable, nil)
			for {
				if err := h.db.Write(b, nil); err == nil {
					break
				} else if errors.IsCorrupted(err) {
					t.Fatalf("WRITE #%d corrupted: %v", i, err)
				}
			}
		}

		snaps[i] = h.db.newSnapshot()
		b.Reset()
		for k := 0; k < nKey; k++ {
			key := fmt.Sprintf("KEY%8d", k)
			b.Delete([]byte(key))
		}
		h.stor.EmulateError(testutil.ModeOpen|testutil.ModeRead, storage.TypeTable, errors.New("table transient read error"))
		if err := h.db.Write(b, nil); err != nil {
			t.Logf("WRITE #%d  error: %v", i, err)
			h.stor.EmulateError(testutil.ModeOpen|testutil.ModeRead, storage.TypeTable, nil)
			for {
				if err := h.db.Write(b, nil); err == nil {
					break
				} else if errors.IsCorrupted(err) {
					t.Fatalf("WRITE #%d corrupted: %v", i, err)
				}
			}
		}
	}
	h.stor.EmulateError(testutil.ModeOpen|testutil.ModeRead, storage.TypeTable, nil)

	runtime.GOMAXPROCS(runtime.NumCPU())

	rnd := rand.New(rand.NewSource(0xecafdaed))
	wg := &sync.WaitGroup{}
	for i, snap := range snaps {
		wg.Add(2)

		go func(i int, snap *Snapshot, sk []int) {
			defer wg.Done()

			vtail := fmt.Sprintf("VAL%030d", i)
			for _, k := range sk {
				key := fmt.Sprintf("KEY%8d", k)
				xvalue, err := snap.Get([]byte(key), nil)
				if err != nil {
					t.Fatalf("READER_GET #%d SEQ=%d K%d error: %v", i, snap.elem.seq, k, err)
				}
				value := key + vtail
				if !bytes.Equal([]byte(value), xvalue) {
					t.Fatalf("READER_GET #%d SEQ=%d K%d invalid value: want %q, got %q", i, snap.elem.seq, k, value, xvalue)
				}
			}
		}(i, snap, rnd.Perm(nKey))

		go func(i int, snap *Snapshot) {
			defer wg.Done()

			vtail := fmt.Sprintf("VAL%030d", i)
			iter := snap.NewIterator(nil, nil)
			defer iter.Release()
			for k := 0; k < nKey; k++ {
				if !iter.Next() {
					if err := iter.Error(); err != nil {
						t.Fatalf("READER_ITER #%d K%d error: %v", i, k, err)
					} else {
						t.Fatalf("READER_ITER #%d K%d eoi", i, k)
					}
				}
				key := fmt.Sprintf("KEY%8d", k)
				xkey := iter.Key()
				if !bytes.Equal([]byte(key), xkey) {
					t.Fatalf("READER_ITER #%d K%d invalid key: want %q, got %q", i, k, key, xkey)
				}
				value := key + vtail
				xvalue := iter.Value()
				if !bytes.Equal([]byte(value), xvalue) {
					t.Fatalf("READER_ITER #%d K%d invalid value: want %q, got %q", i, k, value, xvalue)
				}
			}
		}(i, snap)
	}

	wg.Wait()
}

func TestDB_UkeyShouldntHopAcrossTable(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		WriteBuffer:                 112 * opt.KiB,
		CompactionTableSize:         90 * opt.KiB,
		CompactionExpandLimitFactor: 1,
	})
	defer h.close()

	const (
		nSnap = 190
		nKey  = 140
	)

	var (
		snaps [nSnap]*Snapshot
		b     = &Batch{}
	)
	for i := range snaps {
		vtail := fmt.Sprintf("VAL%030d", i)
		b.Reset()
		for k := 0; k < nKey; k++ {
			key := fmt.Sprintf("KEY%08d", k)
			b.Put([]byte(key), []byte(key+vtail))
		}
		if err := h.db.Write(b, nil); err != nil {
			t.Fatalf("WRITE #%d error: %v", i, err)
		}

		snaps[i] = h.db.newSnapshot()
		b.Reset()
		for k := 0; k < nKey; k++ {
			key := fmt.Sprintf("KEY%08d", k)
			b.Delete([]byte(key))
		}
		if err := h.db.Write(b, nil); err != nil {
			t.Fatalf("WRITE #%d  error: %v", i, err)
		}
	}

	h.compactMem()

	h.waitCompaction()
	for level, tables := range h.db.s.stVersion.levels {
		for _, table := range tables {
			t.Logf("L%d@%d %q:%q", level, table.fd.Num, table.imin, table.imax)
		}
	}

	h.compactRangeAt(0, "", "")
	h.waitCompaction()
	for level, tables := range h.db.s.stVersion.levels {
		for _, table := range tables {
			t.Logf("L%d@%d %q:%q", level, table.fd.Num, table.imin, table.imax)
		}
	}
	h.compactRangeAt(1, "", "")
	h.waitCompaction()
	for level, tables := range h.db.s.stVersion.levels {
		for _, table := range tables {
			t.Logf("L%d@%d %q:%q", level, table.fd.Num, table.imin, table.imax)
		}
	}
	runtime.GOMAXPROCS(runtime.NumCPU())

	wg := &sync.WaitGroup{}
	for i, snap := range snaps {
		wg.Add(1)

		go func(i int, snap *Snapshot) {
			defer wg.Done()

			vtail := fmt.Sprintf("VAL%030d", i)
			for k := 0; k < nKey; k++ {
				key := fmt.Sprintf("KEY%08d", k)
				xvalue, err := snap.Get([]byte(key), nil)
				if err != nil {
					t.Fatalf("READER_GET #%d SEQ=%d K%d error: %v", i, snap.elem.seq, k, err)
				}
				value := key + vtail
				if !bytes.Equal([]byte(value), xvalue) {
					t.Fatalf("READER_GET #%d SEQ=%d K%d invalid value: want %q, got %q", i, snap.elem.seq, k, value, xvalue)
				}
			}
		}(i, snap)
	}

	wg.Wait()
}

func TestDB_TableCompactionBuilder(t *testing.T) {
	gomega.RegisterTestingT(t)
	stor := testutil.NewStorage()
	stor.OnLog(testingLogger(t))
	stor.OnClose(testingPreserveOnFailed(t))
	defer stor.Close()

	const nSeq = 99

	o := &opt.Options{
		WriteBuffer:                 112 * opt.KiB,
		CompactionTableSize:         43 * opt.KiB,
		CompactionExpandLimitFactor: 1,
		CompactionGPOverlapsFactor:  1,
		DisableBlockCache:           true,
	}
	s, err := newSession(stor, o)
	if err != nil {
		t.Fatal(err)
	}
	if err := s.create(); err != nil {
		t.Fatal(err)
	}
	defer s.close()
	var (
		seq        uint64
		targetSize = 5 * o.CompactionTableSize
		value      = bytes.Repeat([]byte{'0'}, 100)
	)
	for i := 0; i < 2; i++ {
		tw, err := s.tops.create()
		if err != nil {
			t.Fatal(err)
		}
		for k := 0; tw.tw.BytesLen() < targetSize; k++ {
			key := []byte(fmt.Sprintf("%09d", k))
			seq += nSeq - 1
			for x := uint64(0); x < nSeq; x++ {
				if err := tw.append(makeIkey(nil, key, seq-x, ktVal), value); err != nil {
					t.Fatal(err)
				}
			}
		}
		tf, err := tw.finish()
		if err != nil {
			t.Fatal(err)
		}
		rec := &sessionRecord{}
		rec.addTableFile(i, tf)
		if err := s.commit(rec); err != nil {
			t.Fatal(err)
		}
	}

	// Build grandparent.
	v := s.version()
	c := newCompaction(s, v, 1, append(tFiles{}, v.levels[1]...))
	rec := &sessionRecord{}
	b := &tableCompactionBuilder{
		s:         s,
		c:         c,
		rec:       rec,
		stat1:     new(cStatStaging),
		minSeq:    0,
		strict:    true,
		tableSize: o.CompactionTableSize/3 + 961,
	}
	if err := b.run(new(compactionTransactCounter)); err != nil {
		t.Fatal(err)
	}
	for _, t := range c.levels[0] {
		rec.delTable(c.sourceLevel, t.fd.Num)
	}
	if err := s.commit(rec); err != nil {
		t.Fatal(err)
	}
	c.release()

	// Build level-1.
	v = s.version()
	c = newCompaction(s, v, 0, append(tFiles{}, v.levels[0]...))
	rec = &sessionRecord{}
	b = &tableCompactionBuilder{
		s:         s,
		c:         c,
		rec:       rec,
		stat1:     new(cStatStaging),
		minSeq:    0,
		strict:    true,
		tableSize: o.CompactionTableSize,
	}
	if err := b.run(new(compactionTransactCounter)); err != nil {
		t.Fatal(err)
	}
	for _, t := range c.levels[0] {
		rec.delTable(c.sourceLevel, t.fd.Num)
	}
	// Move grandparent to level-3
	for _, t := range v.levels[2] {
		rec.delTable(2, t.fd.Num)
		rec.addTableFile(3, t)
	}
	if err := s.commit(rec); err != nil {
		t.Fatal(err)
	}
	c.release()

	v = s.version()
	for level, want := range []bool{false, true, false, true} {
		got := len(v.levels[level]) > 0
		if want != got {
			t.Fatalf("invalid level-%d tables len: want %v, got %v", level, want, got)
		}
	}
	for i, f := range v.levels[1][:len(v.levels[1])-1] {
		nf := v.levels[1][i+1]
		if bytes.Equal(f.imax.ukey(), nf.imin.ukey()) {
			t.Fatalf("KEY %q hop across table %d .. %d", f.imax.ukey(), f.fd.Num, nf.fd.Num)
		}
	}
	v.release()

	// Compaction with transient error.
	v = s.version()
	c = newCompaction(s, v, 1, append(tFiles{}, v.levels[1]...))
	rec = &sessionRecord{}
	b = &tableCompactionBuilder{
		s:         s,
		c:         c,
		rec:       rec,
		stat1:     new(cStatStaging),
		minSeq:    0,
		strict:    true,
		tableSize: o.CompactionTableSize,
	}
	stor.EmulateErrorOnce(testutil.ModeSync, storage.TypeTable, errors.New("table sync error (once)"))
	stor.EmulateRandomError(testutil.ModeRead|testutil.ModeWrite, storage.TypeTable, 0.01, errors.New("table random IO error"))
	for {
		if err := b.run(new(compactionTransactCounter)); err != nil {
			t.Logf("(expected) b.run: %v", err)
		} else {
			break
		}
	}
	if err := s.commit(rec); err != nil {
		t.Fatal(err)
	}
	c.release()

	stor.EmulateErrorOnce(testutil.ModeSync, storage.TypeTable, nil)
	stor.EmulateRandomError(testutil.ModeRead|testutil.ModeWrite, storage.TypeTable, 0, nil)

	v = s.version()
	if len(v.levels[1]) != len(v.levels[2]) {
		t.Fatalf("invalid tables length, want %d, got %d", len(v.levels[1]), len(v.levels[2]))
	}
	for i, f0 := range v.levels[1] {
		f1 := v.levels[2][i]
		iter0 := s.tops.newIterator(f0, nil, nil)
		iter1 := s.tops.newIterator(f1, nil, nil)
		for j := 0; true; j++ {
			next0 := iter0.Next()
			next1 := iter1.Next()
			if next0 != next1 {
				t.Fatalf("#%d.%d invalid eoi: want %v, got %v", i, j, next0, next1)
			}
			key0 := iter0.Key()
			key1 := iter1.Key()
			if !bytes.Equal(key0, key1) {
				t.Fatalf("#%d.%d invalid key: want %q, got %q", i, j, key0, key1)
			}
			if next0 == false {
				break
			}
		}
		iter0.Release()
		iter1.Release()
	}
	v.release()
}

func testDB_IterTriggeredCompaction(t *testing.T, limitDiv int) {
	const (
		vSize = 200 * opt.KiB
		tSize = 100 * opt.MiB
		mIter = 100
		n     = tSize / vSize
	)

	h := newDbHarnessWopt(t, &opt.Options{
		Compression:       opt.NoCompression,
		DisableBlockCache: true,
	})
	defer h.close()

	h.db.memdbMaxLevel = 2

	key := func(x int) string {
		return fmt.Sprintf("v%06d", x)
	}

	// Fill.
	value := strings.Repeat("x", vSize)
	for i := 0; i < n; i++ {
		h.put(key(i), value)
	}
	h.compactMem()

	// Delete all.
	for i := 0; i < n; i++ {
		h.delete(key(i))
	}
	h.compactMem()

	var (
		limit = n / limitDiv

		startKey = key(0)
		limitKey = key(limit)
		maxKey   = key(n)
		slice    = &util.Range{Limit: []byte(limitKey)}

		initialSize0 = h.sizeOf(startKey, limitKey)
		initialSize1 = h.sizeOf(limitKey, maxKey)
	)

	t.Logf("inital size %s [rest %s]", shortenb(int(initialSize0)), shortenb(int(initialSize1)))

	for r := 0; true; r++ {
		if r >= mIter {
			t.Fatal("taking too long to compact")
		}

		// Iterates.
		iter := h.db.NewIterator(slice, h.ro)
		for iter.Next() {
		}
		if err := iter.Error(); err != nil {
			t.Fatalf("Iter err: %v", err)
		}
		iter.Release()

		// Wait compaction.
		h.waitCompaction()

		// Check size.
		size0 := h.sizeOf(startKey, limitKey)
		size1 := h.sizeOf(limitKey, maxKey)
		t.Logf("#%03d size %s [rest %s]", r, shortenb(int(size0)), shortenb(int(size1)))
		if size0 < initialSize0/10 {
			break
		}
	}

	if initialSize1 > 0 {
		h.sizeAssert(limitKey, maxKey, initialSize1/4-opt.MiB, initialSize1+opt.MiB)
	}
}

func TestDB_IterTriggeredCompaction(t *testing.T) {
	testDB_IterTriggeredCompaction(t, 1)
}

func TestDB_IterTriggeredCompactionHalf(t *testing.T) {
	testDB_IterTriggeredCompaction(t, 2)
}

func TestDB_ReadOnly(t *testing.T) {
	h := newDbHarness(t)
	defer h.close()

	h.put("foo", "v1")
	h.put("bar", "v2")
	h.compactMem()

	h.put("xfoo", "v1")
	h.put("xbar", "v2")

	t.Log("Trigger read-only")
	if err := h.db.SetReadOnly(); err != nil {
		h.close()
		t.Fatalf("SetReadOnly error: %v", err)
	}

	mode := testutil.ModeCreate | testutil.ModeRemove | testutil.ModeRename | testutil.ModeWrite | testutil.ModeSync
	h.stor.EmulateError(mode, storage.TypeAll, errors.New("read-only DB shouldn't writes"))

	ro := func(key, value, wantValue string) {
		if err := h.db.Put([]byte(key), []byte(value), h.wo); err != ErrReadOnly {
			t.Fatalf("unexpected error: %v", err)
		}
		h.getVal(key, wantValue)
	}

	ro("foo", "vx", "v1")

	h.o.ReadOnly = true
	h.reopenDB()

	ro("foo", "vx", "v1")
	ro("bar", "vx", "v2")
	h.assertNumKeys(4)
}

func TestDB_BulkInsertDelete(t *testing.T) {
	h := newDbHarnessWopt(t, &opt.Options{
		Compression:         opt.NoCompression,
		CompactionTableSize: 128 * opt.KiB,
		CompactionTotalSize: 1 * opt.MiB,
		WriteBuffer:         256 * opt.KiB,
	})
	defer h.close()

	const R = 100
	const N = 2500
	key := make([]byte, 4)
	value := make([]byte, 256)
	for i := 0; i < R; i++ {
		offset := N * i
		for j := 0; j < N; j++ {
			binary.BigEndian.PutUint32(key, uint32(offset+j))
			h.db.Put(key, value, nil)
		}
		for j := 0; j < N; j++ {
			binary.BigEndian.PutUint32(key, uint32(offset+j))
			h.db.Delete(key, nil)
		}
	}

	if tot := h.totalTables(); tot > 10 {
		t.Fatalf("too many uncompacted tables: %d (%s)", tot, h.getTablesPerLevel())
	}
}