Cyrene_v2/internal/runtime/memory/tidy_apply_test.go

package memory

import (
	"context"
	"encoding/json"
	"errors"
	"sync"
	"testing"
	"time"

	"git.yeij.top/AskaEth/Cyrene_v2/internal/runtime/embedding"
)

// 本文件锁定记忆整理**应用路径**（切片 TIDY-1；design/23 §7/§8）的六条判据：
//
//	① merge = 软合并：被吸收行 status='merged' + merged_into，存活行
//	   merged_from/merge_count 更新，**content 逐字不变**，memory_items 总行数不变；
//	② **幂等**：重复应用同一条候选 ⇒ skipped/already_decided，且不产生第二次副作用
//	   （merge_count 不再涨、没有第二行被改）；
//	③ **可 revert**：回滚后字段复原（status/merged_into/merged_from/merge_count/kind），
//	   行数不变，候选退回 pending；
//	④ archive = 软删（status='archived'），reclassify = 改 kind，都不碰 content；
//	⑤ link 不可应用 ⇒ skipped/action_not_applyable，零改动；
//	⑥ 批量（按 run）有上限、逐条结果、超限如实报 Truncated + Total。

// tidyItemRow 是断言用的一行记忆关键字段（含正文，用来钉「绝不改 content」）。
type tidyItemRow struct {
	status     string
	mergedInto string
	mergedFrom string
	mergeCount int
	content    string
	kind       string
}

// loadTidyItemRow 读一行记忆的整理列 + 正文。
func loadTidyItemRow(t *testing.T, store *Store, id string) tidyItemRow {
	t.Helper()
	var r tidyItemRow
	err := store.db.QueryRowContext(context.Background(),
		`SELECT status, merged_into, merged_from, merge_count, content, kind
		 FROM memory_items WHERE id = ?`, id).
		Scan(&r.status, &r.mergedInto, &r.mergedFrom, &r.mergeCount, &r.content, &r.kind)
	if err != nil {
		t.Fatalf("读记忆行 %s: %v", id, err)
	}
	return r
}

// countMemoryItems 数 memory_items 的总行数（软删除 ⇒ 应用前后必须不变）。
func countMemoryItems(t *testing.T, store *Store) int {
	t.Helper()
	var n int
	if err := store.db.QueryRowContext(context.Background(),
		`SELECT COUNT(*) FROM memory_items`).Scan(&n); err != nil {
		t.Fatalf("数 memory_items: %v", err)
	}
	return n
}

// mergeCandidateFromRun 跑一轮整理并返回唯一的 merge 候选（顺带把运行 id 带出来）。
func mergeCandidateFromRun(t *testing.T, store *Store, emb embedding.Embedder) TidyCandidate {
	t.Helper()
	tr := newTestTidier(t, store, tidyTestCfg(0.80), emb)
	stats, err := tr.RunOnce(context.Background())
	if err != nil {
		t.Fatalf("RunOnce: %v", err)
	}
	cands, err := store.ListTidyCandidates(context.Background(), "", 50)
	if err != nil {
		t.Fatalf("ListTidyCandidates: %v", err)
	}
	merges := candidatesByAction(cands, TidyMerge)
	if len(merges) != 1 {
		t.Fatalf("merge 候选 = %d, want 1（本轮 run=%s）", len(merges), stats.RunID)
	}
	if merges[0].RunID != stats.RunID {
		t.Fatalf("候选 run_id = %q, want 本轮 run_id = %q", merges[0].RunID, stats.RunID)
	}
	return merges[0]
}

// TestTidyApplyMergeIsSoftIdempotentAndRevertible 是①②③三条判据的合并用例：
// 软合并 → 重复应用无副作用 → 回滚后字段复原。
func TestTidyApplyMergeIsSoftIdempotentAndRevertible(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	short := seedTidyItem(t, store, habitItem(fixtureHabitShort, base, 0.9))
	long := seedTidyItem(t, store, habitItem(fixtureHabitLong, base.Add(time.Hour), 0.9))

	cand := mergeCandidateFromRun(t, store, stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureHabitShort: {1, 0},
		fixtureHabitLong:  simVector(0.95),
	}})
	if cand.SourceIDs[0] != short.ID || cand.TargetID != long.ID {
		t.Fatalf("候选方向不对：sources=%v target=%q", cand.SourceIDs, cand.TargetID)
	}
	rowsBefore := countMemoryItems(t, store)
	shortBefore := loadTidyItemRow(t, store, short.ID)
	longBefore := loadTidyItemRow(t, store, long.ID)

	res, err := store.ApplyTidyCandidate(ctx, cand.ID, TidyApplyOptions{Actor: "owner:test"})
	if err != nil {
		t.Fatalf("ApplyTidyCandidate: %v", err)
	}
	if res.Status != TidyApplyApplied {
		t.Fatalf("应用结果 = %+v, want applied", res)
	}

	absorbed := loadTidyItemRow(t, store, short.ID)
	if absorbed.status != ingestStatusMerged || absorbed.mergedInto != long.ID {
		t.Fatalf("被吸收行 = %+v, want status=merged merged_into=%s", absorbed, long.ID)
	}
	if absorbed.content != shortBefore.content {
		t.Fatalf("被吸收行 content 被改了：%q → %q", shortBefore.content, absorbed.content)
	}
	survivor := loadTidyItemRow(t, store, long.ID)
	if survivor.content != longBefore.content {
		t.Fatalf("存活行 content 被改了：%q → %q", longBefore.content, survivor.content)
	}
	if survivor.mergeCount != longBefore.mergeCount+1 {
		t.Fatalf("存活行 merge_count = %d, want %d", survivor.mergeCount, longBefore.mergeCount+1)
	}
	if !jsonHasID(t, survivor.mergedFrom, short.ID) {
		t.Fatalf("存活行 merged_from = %s, want 含 %s", survivor.mergedFrom, short.ID)
	}
	if got := countMemoryItems(t, store); got != rowsBefore {
		t.Fatalf("软合并改变了行数：%d → %d（行永不物理删除）", rowsBefore, got)
	}
	applied, err := store.GetTidyCandidate(ctx, cand.ID)
	if err != nil {
		t.Fatalf("GetTidyCandidate: %v", err)
	}
	if applied.Status != tidyCandidateApplied {
		t.Fatalf("候选状态 = %q, want %q", applied.Status, tidyCandidateApplied)
	}

	// ② 重复应用：skipped/already_decided，且**零副作用**。
	again, err := store.ApplyTidyCandidate(ctx, cand.ID, TidyApplyOptions{Actor: "owner:test"})
	if err != nil {
		t.Fatalf("第二次 ApplyTidyCandidate 不该报错（幂等命中不是错误）：%v", err)
	}
	if again.Status != TidyApplySkipped || again.Reason != tidySkipAlreadyDecided {
		t.Fatalf("第二次应用 = %+v, want skipped/%s", again, tidySkipAlreadyDecided)
	}
	if len(again.AffectedIDs) != 0 {
		t.Fatalf("跳过的结果不得声称改过行：%v", again.AffectedIDs)
	}
	if got := loadTidyItemRow(t, store, long.ID).mergeCount; got != survivor.mergeCount {
		t.Fatalf("重复应用把 merge_count 又加了一次：%d → %d", survivor.mergeCount, got)
	}
	if got := countMemoryItems(t, store); got != rowsBefore {
		t.Fatalf("重复应用改变了行数：%d → %d", rowsBefore, got)
	}

	// ③ 回滚：字段逐字复原（updated_at 除外 —— 它是「什么时候动过」的审计面）。
	rev, err := store.RevertTidyCandidate(ctx, cand.ID, TidyApplyOptions{Actor: "owner:test"})
	if err != nil {
		t.Fatalf("RevertTidyCandidate: %v", err)
	}
	if rev.Status != TidyApplyApplied {
		t.Fatalf("回滚结果 = %+v, want applied", rev)
	}
	if got := loadTidyItemRow(t, store, short.ID); got != shortBefore {
		t.Fatalf("回滚后被吸收行未复原：\n got=%+v\nwant=%+v", got, shortBefore)
	}
	if got := loadTidyItemRow(t, store, long.ID); got.mergeCount != longBefore.mergeCount || got.mergedFrom != "[]" {
		t.Fatalf("回滚后存活行合并元信息未复原：%+v（want merge_count=%d merged_from=[]）", got, longBefore.mergeCount)
	}
	if got := countMemoryItems(t, store); got != rowsBefore {
		t.Fatalf("回滚改变了行数：%d → %d", rowsBefore, got)
	}
	if back, err := store.GetTidyCandidate(ctx, cand.ID); err != nil || back.Status != tidyCandidatePending {
		t.Fatalf("回滚后候选状态 = %q (err=%v), want pending", back.Status, err)
	}
	// 回滚是幂等的：再回一次 ⇒ skipped/not_applied。
	if againRev, err := store.RevertTidyCandidate(ctx, cand.ID, TidyApplyOptions{Actor: "owner:test"}); err != nil {
		t.Fatalf("第二次回滚不该报错：%v", err)
	} else if againRev.Status != TidyApplySkipped || againRev.Reason != tidySkipNotApplied {
		t.Fatalf("第二次回滚 = %+v, want skipped/%s", againRev, tidySkipNotApplied)
	}
}

// TestTidyApplyMergeUnderConcurrencyHasSingleWinner 钉住执行闸（候选状态的条件
// 更新）在**并发**下也只有一次生效：16 路同时应用同一条候选，恰好一条 applied，
// merge_count 只涨一次。少了这道闸，多路会各自把源行当作 active 再合并一次。
func TestTidyApplyMergeUnderConcurrencyHasSingleWinner(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	short := seedTidyItem(t, store, habitItem(fixtureHabitShort, base, 0.9))
	long := seedTidyItem(t, store, habitItem(fixtureHabitLong, base.Add(time.Hour), 0.9))
	cand := mergeCandidateFromRun(t, store, stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureHabitShort: {1, 0},
		fixtureHabitLong:  simVector(0.95),
	}})

	const racers = 16
	var (
		mu      sync.Mutex
		applied int
		skipped int
		start   = make(chan struct{})
		wg      sync.WaitGroup
	)
	for i := 0; i < racers; i++ {
		wg.Add(1)
		go func() {
			defer wg.Done()
			<-start
			res, err := store.ApplyTidyCandidate(ctx, cand.ID, TidyApplyOptions{Actor: "owner:race"})
			mu.Lock()
			defer mu.Unlock()
			switch {
			case err != nil:
				t.Errorf("并发应用报错：%v", err)
			case res.Status == TidyApplyApplied:
				applied++
			default:
				skipped++
			}
		}()
	}
	close(start)
	wg.Wait()

	if applied != 1 || skipped != racers-1 {
		t.Fatalf("并发应用 applied=%d skipped=%d, want 1/%d", applied, skipped, racers-1)
	}
	if got := loadTidyItemRow(t, store, long.ID).mergeCount; got != 2 {
		t.Fatalf("存活行 merge_count = %d, want 2（只允许一次吸收）", got)
	}
	if got := loadTidyItemRow(t, store, short.ID); got.status != ingestStatusMerged || got.mergedInto != long.ID {
		t.Fatalf("被吸收行 = %+v, want merged into %s", got, long.ID)
	}
}

// TestTidyApplyArchiveAndRevert 钉住④：archive 是软删（status='archived'，内容不变、
// 行数不变），revert 复原为 active。
func TestTidyApplyArchiveAndRevert(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	oneOff := seedTidyItem(t, store, habitItem(fixtureOneOffEvent, base, 0.9))

	tr := newTestTidier(t, store, tidyTestCfg(0.80), nil)
	if _, err := tr.RunOnce(ctx); err != nil {
		t.Fatalf("RunOnce: %v", err)
	}
	cands, err := store.ListTidyCandidates(ctx, "", 50)
	if err != nil {
		t.Fatalf("ListTidyCandidates: %v", err)
	}
	archives := candidatesByAction(cands, TidyArchive)
	if len(archives) != 1 {
		t.Fatalf("archive 候选 = %d, want 1（%+v）", len(archives), cands)
	}
	rowsBefore := countMemoryItems(t, store)
	before := loadTidyItemRow(t, store, oneOff.ID)

	if res, err := store.ApplyTidyCandidate(ctx, archives[0].ID, TidyApplyOptions{Actor: "owner:test"}); err != nil {
		t.Fatalf("ApplyTidyCandidate: %v", err)
	} else if res.Status != TidyApplyApplied {
		t.Fatalf("archive 应用 = %+v, want applied", res)
	}
	archived := loadTidyItemRow(t, store, oneOff.ID)
	if archived.status != tidyStatusArchived {
		t.Fatalf("归档后 status = %q, want %q", archived.status, tidyStatusArchived)
	}
	if archived.content != before.content {
		t.Fatalf("软删改了 content：%q → %q", before.content, archived.content)
	}
	if got := countMemoryItems(t, store); got != rowsBefore {
		t.Fatalf("软删改变了行数：%d → %d（design/23 §8：行保留）", rowsBefore, got)
	}

	if res, err := store.RevertTidyCandidate(ctx, archives[0].ID, TidyApplyOptions{Actor: "owner:test"}); err != nil {
		t.Fatalf("RevertTidyCandidate: %v", err)
	} else if res.Status != TidyApplyApplied {
		t.Fatalf("archive 回滚 = %+v, want applied", res)
	}
	if got := loadTidyItemRow(t, store, oneOff.ID); got != before {
		t.Fatalf("回滚后未复原：\n got=%+v\nwant=%+v", got, before)
	}
	if got := countMemoryItems(t, store); got != rowsBefore {
		t.Fatalf("回滚改变了行数：%d → %d", rowsBefore, got)
	}
}

// TestTidyApplyReclassifyAndRevert 钉住④的另一半：reclassify 只改 kind 一列，
// 内容不变；revert 把 kind 还原成候选证据里的旧值。
func TestTidyApplyReclassifyAndRevert(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	misclassified := MemoryItem{
		PersonaID: "persona.a", Scope: ScopePersona, Kind: KindWorldEvent,
		Content: fixtureHabitShort, Confidence: 0.9, CreatedAt: base,
	}
	superset := habitItem(fixtureHabitLong, base.Add(time.Hour), 0.9)
	bad := seedTidyItem(t, store, misclassified)
	seedTidyItem(t, store, superset)

	tr := newTestTidier(t, store, tidyTestCfg(0.80), stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureHabitShort: {1, 0},
		fixtureHabitLong:  simVector(0.95),
	}})
	if _, err := tr.RunOnce(ctx); err != nil {
		t.Fatalf("RunOnce: %v", err)
	}
	cands, err := store.ListTidyCandidates(ctx, "", 50)
	if err != nil {
		t.Fatalf("ListTidyCandidates: %v", err)
	}
	res := candidatesByAction(cands, TidyReclassify)
	if len(res) != 1 {
		t.Fatalf("reclassify 候选 = %d, want 1（%+v）", len(res), cands)
	}
	before := loadTidyItemRow(t, store, bad.ID)

	if out, err := store.ApplyTidyCandidate(ctx, res[0].ID, TidyApplyOptions{Actor: "owner:test"}); err != nil {
		t.Fatalf("ApplyTidyCandidate: %v", err)
	} else if out.Status != TidyApplyApplied {
		t.Fatalf("reclassify 应用 = %+v, want applied", out)
	}
	after := loadTidyItemRow(t, store, bad.ID)
	if after.kind != string(KindRelationship) {
		t.Fatalf("reclassify 后 kind = %q, want %q", after.kind, KindRelationship)
	}
	if after.content != before.content || after.status != before.status {
		t.Fatalf("reclassify 改了不该改的列：%+v（before=%+v）", after, before)
	}

	if _, err := store.RevertTidyCandidate(ctx, res[0].ID, TidyApplyOptions{Actor: "owner:test"}); err != nil {
		t.Fatalf("RevertTidyCandidate: %v", err)
	}
	if got := loadTidyItemRow(t, store, bad.ID); got.kind != before.kind {
		t.Fatalf("回滚后 kind = %q, want %q（还原成候选证据里的旧值）", got.kind, before.kind)
	}
}

// TestTidyApplyLinkIsNotApplyable 钉住⑤：link 没有执行语义（design/23 §14：
// 链接的存储形态未定），应用它只能得到 skipped，且**零改动**。
func TestTidyApplyLinkIsNotApplyable(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	seedTidyItem(t, store, habitItem(fixtureEnergyRedBull, base, 0.95))
	seedTidyItem(t, store, habitItem(fixtureEnergyIced, base.Add(time.Minute), 0.95))

	tr := newTestTidier(t, store, tidyTestCfg(0.80), stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureEnergyRedBull: {1, 0},
		fixtureEnergyIced:    simVector(0.99),
	}})
	if _, err := tr.RunOnce(ctx); err != nil {
		t.Fatalf("RunOnce: %v", err)
	}
	cands, err := store.ListTidyCandidates(ctx, "", 50)
	if err != nil {
		t.Fatalf("ListTidyCandidates: %v", err)
	}
	links := candidatesByAction(cands, TidyLink)
	if len(links) != 1 {
		t.Fatalf("link 候选 = %d, want 1", len(links))
	}
	before := dumpTidyTable(t, store, "memory_items", memoryItemsColumns)
	out, err := store.ApplyTidyCandidate(ctx, links[0].ID, TidyApplyOptions{Actor: "owner:test"})
	if err != nil {
		t.Fatalf("ApplyTidyCandidate(link): %v", err)
	}
	if out.Status != TidyApplySkipped || out.Reason != tidySkipActionNotApplyable {
		t.Fatalf("link 应用结果 = %+v, want skipped/%s", out, tidySkipActionNotApplyable)
	}
	if len(out.AffectedIDs) != 0 {
		t.Fatalf("不可应用的动作不得声称改过行：%v", out.AffectedIDs)
	}
	if got := dumpTidyTable(t, store, "memory_items", memoryItemsColumns); len(got) != len(before) {
		t.Fatalf("link 应用改动了记忆：%d → %d 行", len(before), len(got))
	}
	if cand, err := store.GetTidyCandidate(ctx, links[0].ID); err != nil || cand.Status != tidyCandidatePending {
		t.Fatalf("不可应用的候选应保持 pending，实际 %q (err=%v)", cand.Status, err)
	}
}

// TestTidyApplyBatchByRunHonoursLimitAndReportsTruncated 钉住⑥：按 run 批量应用
// 逐条给结果、上限生效、超限如实报 Truncated + Total，并把该轮留痕的 applied 累加。
func TestTidyApplyBatchByRunHonoursLimitAndReportsTruncated(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	const oneOffs = 5
	ids := make([]string, 0, oneOffs)
	for i := 0; i < oneOffs; i++ {
		item := habitItem(fixtureOneOffEvent, base.Add(time.Duration(i)*time.Minute), 0.9)
		ids = append(ids, seedTidyItem(t, store, item).ID)
	}

	tr := newTestTidier(t, store, tidyTestCfg(0.80), nil)
	stats, err := tr.RunOnce(ctx)
	if err != nil {
		t.Fatalf("RunOnce: %v", err)
	}
	if stats.ByAction[string(TidyArchive)] != oneOffs {
		t.Fatalf("ByAction[archive] = %d, want %d（%+v）", stats.ByAction[string(TidyArchive)], oneOffs, stats.ByAction)
	}

	batch, err := store.ApplyTidyCandidates(ctx, stats.RunID, TidyApplyOptions{Actor: "owner:test", Limit: 2})
	if err != nil {
		t.Fatalf("ApplyTidyCandidates: %v", err)
	}
	if batch.Total != oneOffs || len(batch.Results) != 2 || !batch.Truncated {
		t.Fatalf("批量回执 = %+v, want total=%d results=2 truncated=true", batch, oneOffs)
	}
	if batch.Applied != 2 || batch.Archived != 2 || batch.Skipped != 0 || batch.Failed != 0 {
		t.Fatalf("批量计数 = %+v, want applied=2 archived=2", batch)
	}
	for _, res := range batch.Results {
		if res.Status != TidyApplyApplied || res.Action != TidyArchive || len(res.AffectedIDs) != 1 {
			t.Fatalf("逐条结果 = %+v, want 每条 applied/archive 带 1 个 affected id", res)
		}
	}
	archived := 0
	for _, id := range ids {
		if loadTidyItemRow(t, store, id).status == tidyStatusArchived {
			archived++
		}
	}
	if archived != 2 {
		t.Fatalf("实际归档 = %d, want 2", archived)
	}
	runs, err := store.ListTidyRuns(ctx, 10)
	if err != nil {
		t.Fatalf("ListTidyRuns: %v", err)
	}
	if len(runs) != 1 || runs[0].Applied != 2 {
		t.Fatalf("运行留痕 applied = %d (runs=%d), want 2", runs[0].Applied, len(runs))
	}

	// 剩余 3 条：第二批应用，超限标记消失（Total 已 ≤ 上限）。
	batch2, err := store.ApplyTidyCandidates(ctx, stats.RunID, TidyApplyOptions{Actor: "owner:test", Limit: 2})
	if err != nil {
		t.Fatalf("第二批: %v", err)
	}
	if batch2.Total != oneOffs-2 || !batch2.Truncated || batch2.Applied != 2 {
		t.Fatalf("第二批回执 = %+v, want total=3 truncated=true applied=2", batch2)
	}

	// runId 拼错必须是「查不到这一轮」，不是「这一轮没有待定候选」。
	if _, err := store.ApplyTidyCandidates(ctx, "no-such-run", TidyApplyOptions{Actor: "owner:test"}); !errors.Is(err, ErrTidyRunRecordNotFound) {
		t.Fatalf("未知 runId 的 err = %v, want ErrTidyRunRecordNotFound", err)
	}
	// runId 为空 ⇒ 「全部待定应用」这一档（没有轮次筛）。
	rest, err := store.ApplyTidyCandidates(ctx, "", TidyApplyOptions{Actor: "owner:test"})
	if err != nil {
		t.Fatalf("全部待定应用: %v", err)
	}
	if rest.Total != 1 || rest.Applied != 1 || rest.Archived != 1 {
		t.Fatalf("全部待定回执 = %+v, want total=1 applied=1 archived=1", rest)
	}
}

// jsonHasID 报告一个 JSON 数组文本里是否含 id。
func jsonHasID(t *testing.T, jsonText, id string) bool {
	t.Helper()
	var ids []string
	if err := json.Unmarshal([]byte(jsonText), &ids); err != nil {
		t.Fatalf("解析 %q: %v", jsonText, err)
	}
	return containsID(ids, id)
}