TsDB2/demo/verify_hash_index/main.go

package main

import (
	"git.ycl.pub/yl32940/TsDB2/DB"
	"git.ycl.pub/yl32940/TsDB2/DB/core"
	"fmt"
	"os"
	"time"
)

// Record 测试表结构 — 通过 vrtx 标签声明索引，UseTable 自动创建
type Record struct {
	DB.BaseModel
	Name  string `vrtx:"maxlen:32"`
	Age   int64  `vrtx:"index:idx_hash_age,type:hash"`
	Score int64  `vrtx:"index:idx_btree_score"`
	Dept  string `vrtx:"maxlen:16"`
}

// PerfRecord 性能测试用 — 同一字段同时建 Hash 和 B+树 做对比
type PerfRecord struct {
	DB.BaseModel
	Name  string `vrtx:"maxlen:32"`
	Age   int64  `vrtx:"index:idx_hash_age,type:hash"`
	Score int64  `vrtx:"index:idx_btree_age"`
	Dept  string `vrtx:"maxlen:16"`
}

var dbPath string

var passed, failed int

func check(msg string, ok bool, detail ...string) {
	if ok {
		passed++
		fmt.Printf("  ✅ %s\n", msg)
	} else {
		failed++
		d := ""
		if len(detail) > 0 {
			d = " — " + detail[0]
		}
		fmt.Printf("  ❌ %s%s\n", msg, d)
	}
}

func section(title string) {
	fmt.Printf("\n%s %s\n", "==========", title)
}

// searchAge 用引擎的键编码查询 Age 索引, 返回查找错误 (nil 表示命中)。
//
// 不能手拼 uint64(age) 大端: int64 列现在落 FieldType_Int64, 用保序编码 (符号位翻转),
// 手拼的键在 v7 之后与盘上键不匹配, SearchByIndex 会一律 not found。
func searchAge(ct *core.Table, idxName string, age int64) error {
	key, err := ct.BuildIndexKey(idxName, age)
	if err != nil {
		return err
	}
	_, err = ct.SearchByIndex(idxName, key)
	return err
}

// countFindable 统计可被 Hash 索引查找到的 Age 值数量
func countFindable(db *DB.DB, tableName string, idxName string, minAge, maxAge int64) int {
	coreTbl, _ := db.CoreDB().FindTable(tableName)
	count := 0
	for age := minAge; age < maxAge; age++ {
		if searchAge(coreTbl, idxName, age) == nil {
			count++
		}
	}
	return count
}

func main() {
	dbPath = fmt.Sprintf("hash_test_%d.db", time.Now().UnixNano())
	os.Remove(dbPath)

	fmt.Println("============================================================")
	fmt.Println("  TsDB2 Hash 桶式索引 — 完整测试")
	fmt.Println("  覆盖: CRUD/分页/桶分裂(Split)/桶合并(Merge)/持久化/性能")
	fmt.Println("============================================================")

	// ============================================================
	// 1. 建表 (索引由 vrtx 标签自动创建)
	// ============================================================
	section("1. 建表 + 自动索引")

	db, err := DB.Open(dbPath)
	if err != nil {
		panic(err)
	}
	db.SetAutoCommitDelay(0)
	db.SetBufferPoolCapacity(4000)

	tbl, err := DB.UseTable[Record](db, "hash_test")
	check("UseTable[Record] 建表 (自动创建索引)", err == nil)

	coreTbl, _ := db.CoreDB().FindTable("hash_test")

	// 验证索引已自动创建
	err = searchAge(coreTbl, "idx_hash_age", 25)
	check("idx_hash_age (Hash) 已自动创建", err == nil || err.Error() == "key not found")

	// ============================================================
	// 2. 大批量插入 — 触发多页存储 (3000行, 每桶~187条目)
	// ============================================================
	section("2. 大批量插入 (3000 行, 触发多页)")

	N1 := 3000
	t0 := time.Now()
	for i := 1; i <= N1; i++ {
		err := tbl.Insert(&Record{
			Name:  fmt.Sprintf("User%05d", i),
			Age:   int64(20 + i%100),
			Score: int64(i % 500),
			Dept:  fmt.Sprintf("D%d", i%20),
		})
		if err != nil {
			check(fmt.Sprintf("Insert ID=%d", i), false, err.Error())
			break
		}
	}
	insertTime := time.Since(t0)
	check(fmt.Sprintf("插入 %d 行 耗时 %.2fs (%.0f rows/s)", N1, insertTime.Seconds(), float64(N1)/insertTime.Seconds()), true)

	if err := db.Commit(); err != nil {
		panic(err)
	}
	check("Commit 持久化", true)

	// 验证数据可查
	found1 := countFindable(db, "hash_test", "idx_hash_age", 20, 120)
	check(fmt.Sprintf("Hash 查找: %d/100 个 Age 值可查", found1), found1 > 50)

	u100, err := tbl.Find(100)
	check("PK 查找 ID=100", err == nil && u100 != nil && u100.Name == "User00100")

	// 验证行数
	cnt, _ := tbl.Count()
	check(fmt.Sprintf("行数: %d (应=%d)", cnt, N1), cnt == int64(N1))

	// ============================================================
	// 3. 桶分裂测试 — 集中插入触发 Split (阈值=800)
	// ============================================================
	section("3. 桶分裂测试 (Split)")

	splitN := 1000
	fmt.Printf("  集中插入 %d 行 Age=25 (全部 hash 到同一桶, 阈值=800)...\n", splitN)

	t0 = time.Now()
	for i := N1 + 1; i <= N1+splitN; i++ {
		if err := tbl.Insert(&Record{
			Name:  fmt.Sprintf("Split%05d", i),
			Age:   25,
			Score: int64(i % 300),
			Dept:  fmt.Sprintf("S%d", i%5),
		}); err != nil {
			check(fmt.Sprintf("Split Insert ID=%d", i), false, err.Error())
			break
		}
	}
	check(fmt.Sprintf("集中插入 %d 行 耗时 %.2fs", splitN, time.Since(t0).Seconds()), true)

	if err := db.Commit(); err != nil {
		panic(err)
	}

	// 验证分裂后数据完整性
	err25 := searchAge(coreTbl, "idx_hash_age", 25)
	check("分裂后 Hash 查找 Age=25", err25 == nil)

	splitFound := countFindable(db, "hash_test", "idx_hash_age", 20, 120)
	check(fmt.Sprintf("分裂后 Hash 查找: %d/100 个 Age 值可查", splitFound), splitFound > 50)

	uSplit, err := tbl.Find(uint64(N1 + 1))
	check("分裂后 PK 查找", err == nil && uSplit != nil)

	cnt2, _ := tbl.Count()
	check(fmt.Sprintf("分裂后行数: %d (应=%d)", cnt2, N1+splitN), cnt2 == int64(N1+splitN))

	// ============================================================
	// 4. 删除测试
	// ============================================================
	section("4. Hash 索引删除测试")

	delN := 500
	for i := 1; i <= delN; i++ {
		if err := tbl.Delete(uint64(i)); err != nil {
			check(fmt.Sprintf("Delete ID=%d", i), false, err.Error())
			break
		}
	}
	check(fmt.Sprintf("删除 %d 行 (ID 1~%d)", delN, delN), true)

	if err := db.Commit(); err != nil {
		panic(err)
	}

	_, err = tbl.Find(1)
	check("删除后 PK 查找 ID=1 (应不存在)", err != nil)

	_, err = tbl.Find(uint64(delN + 1))
	check(fmt.Sprintf("删除后 PK 查找 ID=%d (应存在)", delN+1), err == nil)

	err = searchAge(coreTbl, "idx_hash_age", 25)
	check("删除后 Hash 查找 Age=25", err == nil)

	// ============================================================
	// 5. 桶合并测试 — 大量删除触发 Merge
	// ============================================================
	section("5. 桶合并测试 (Merge)")

	deleteCount := 0
	for i := uint64(delN + 1); i <= uint64(N1+splitN); i++ {
		if err := tbl.Delete(i); err == nil {
			deleteCount++
		}
	}
	check(fmt.Sprintf("删除 %d 行", deleteCount), deleteCount > 0)

	if err := db.Commit(); err != nil {
		panic(err)
	}

	remaining, _ := tbl.Count()
	check(fmt.Sprintf("剩余行数: %d (已删除全部 %d 行)", remaining, N1+splitN), remaining == 0)

	// 验证全删后哈希索引仍可安全查询
	err = searchAge(coreTbl, "idx_hash_age", 25)
	check("合并后 Hash 索引不崩溃", true)

	// ============================================================
	// 6. 持久化测试 (关闭重开)
	// ============================================================
	section("6. 持久化 + 重开验证")

	countBefore := remaining
	db.Close()
	check("关闭数据库", true)

	db, err = DB.Open(dbPath)
	if err != nil {
		panic(err)
	}
	db.SetBufferPoolCapacity(4000)
	tbl, err = DB.UseTable[Record](db, "hash_test")
	check("重开 UseTable[Record]", err == nil)

	coreTbl, _ = db.CoreDB().FindTable("hash_test")

	countAfter, _ := tbl.Count()
	check(fmt.Sprintf("重开后行数: %d (应=%d)", countAfter, countBefore),
		countAfter == countBefore,
		fmt.Sprintf("expected=%d actual=%d", countBefore, countAfter))

	db.Close()

	// ============================================================
	// 7. 性能对比: Hash vs B+树 (用 PerfRecord 同一字段两种索引)
	// ============================================================
	section("7. 性能对比: Hash vs B+树")

	os.Remove(dbPath)
	db, _ = DB.Open(dbPath)
	db.SetAutoCommitDelay(0)
	db.SetBufferPoolCapacity(4000)

	perfTbl, _ := DB.UseTable[PerfRecord](db, "perf_test")
	perfCoreTbl, _ := db.CoreDB().FindTable("perf_test")

	N := 5000
	t0 = time.Now()
	for i := 1; i <= N; i++ {
		perfTbl.Insert(&PerfRecord{
			Name:  fmt.Sprintf("Perf%06d", i),
			Age:   int64(20 + i%200),
			Score: int64(i % 1000),
			Dept:  fmt.Sprintf("D%d", i%30),
		})
	}
	db.Commit()
	insertT := time.Since(t0)
	fmt.Printf("  插入 %d 行: %.2fs (%.0f rows/s)\n", N, insertT.Seconds(), float64(N)/insertT.Seconds())

	lookupN := 1000
	t0 = time.Now()
	for i := 0; i < lookupN; i++ {
		age := int64(20 + i%200)
		searchAge(perfCoreTbl, "idx_hash_age", age)
	}
	hashTime := time.Since(t0)
	hashQPS := float64(lookupN) / hashTime.Seconds()
	fmt.Printf("  Hash  查找 %d 次: %.3fs (%.0f qps)\n", lookupN, hashTime.Seconds(), hashQPS)

	t0 = time.Now()
	for i := 0; i < lookupN; i++ {
		age := int64(20 + i%200)
		searchAge(perfCoreTbl, "idx_btree_age", age)
	}
	btreeTime := time.Since(t0)
	btreeQPS := float64(lookupN) / btreeTime.Seconds()
	fmt.Printf("  B+树 查找 %d 次: %.3fs (%.0f qps)\n", lookupN, btreeTime.Seconds(), btreeQPS)

	if hashQPS > 0 && btreeQPS > 0 {
		ratio := hashQPS / btreeQPS
		check(fmt.Sprintf("Hash/B+树 性能比: %.2fx", ratio), true)
	}

	db.Close()

	// ============================================================
	// 总结
	// ============================================================
	section("测试总结")
	total := passed + failed
	fmt.Printf("  总计: %d 项测试, ✅ 通过: %d, ❌ 失败: %d\n", total, passed, failed)
	if failed == 0 {
		fmt.Println("  🎉 全部通过！Hash 桶式索引工作正常！")
	} else {
		fmt.Printf("  ⚠️  有 %d 项测试失败，请检查\n", failed)
	}

	// [KEEP-DB] 保留数据库文件供 Analyzer 分析, 屏蔽运行结束后的清理删除
	// os.Remove(dbPath)
}