Cyrene_v2/internal/runtime/memory/preinsert_test.go

package memory

import (
	"context"
	"testing"
	"time"

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

// 本文件是 D3「入库前审查」（design/23 §15.1 D3）的**四态判定与 fail-safe** 测试：
// 完全等 ⇒ skip、真包含 ⇒ merge/modify、不相似/并列 ⇒ create、审查不可用 ⇒
// 一行都不改写。隔离、异步与开关口径在 preinsert_async_test.go。

// reviewRowState 是测试直接读回的一行记忆的整理列。
type reviewRowState struct {
	status     string
	mergedInto string
	mergeCount int
	confidence float64
	content    string
}

// readReviewRow 直读一行记忆的整理列（status 等列不在 Store.Get 的读侧面上）。
func readReviewRow(t *testing.T, store *Store, id string) reviewRowState {
	t.Helper()
	var st reviewRowState
	if err := store.db.QueryRowContext(context.Background(),
		`SELECT status, merged_into, merge_count, confidence, content FROM memory_items WHERE id = ?`,
		id).Scan(&st.status, &st.mergedInto, &st.mergeCount, &st.confidence, &st.content); err != nil {
		t.Fatalf("read row %s: %v", id, err)
	}
	return st
}

// countActiveRows 数一个分区里还在册（status='active'）的条目数。
func countActiveRows(t *testing.T, store *Store, personaID string) int {
	t.Helper()
	var n int
	if err := store.db.QueryRowContext(context.Background(),
		`SELECT COUNT(*) FROM memory_items WHERE persona_id = ? AND status = ?`,
		personaID, string(tidyStatusActive)).Scan(&n); err != nil {
		t.Fatalf("count active rows: %v", err)
	}
	return n
}

// newTestReviewer 构建固定时钟的审查服务。阈值显式写死，其余走设计稿默认值。
func newTestReviewer(t *testing.T, store *Store, emb embedding.Embedder, threshold float64) *PreInsertReviewer {
	t.Helper()
	r, err := NewPreInsertReviewer(ReviewConfig{
		Enabled:   true,
		QueueSize: 8,
		ScanLimit: 50,
		Timeout:   5 * time.Second,
		Threshold: threshold,
	}, ReviewDeps{
		Store:    store,
		Embedder: emb,
		Now:      func() time.Time { return tidyTestTime(t) },
		Log:      testLogger(),
	})
	if err != nil {
		t.Fatalf("NewPreInsertReviewer: %v", err)
	}
	return r
}

// reviewAt 构造与既有条目同分区、可指定内容与溯源的新条目。
func reviewAt(content string, at time.Time, confidence float64) MemoryItem {
	item := habitItem(content, at, confidence)
	item.SourceConversation = FormatSourceConversation("web", "msg-review")
	return item
}

// TestReviewExactDuplicateDoesNotCreateASecondActiveRow 是 design/23 §15 D3-9①：
// **完全等价 ⇒ 不新建**（这正是主人报的「反复新建」）。活跃行数必须仍是 1。
func TestReviewExactDuplicateDoesNotCreateASecondActiveRow(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	existing := seedTidyItem(t, store, reviewAt(fixtureHabitShort, base, 0.9))
	dup := seedTidyItem(t, store, reviewAt(fixtureHabitShort, base.Add(time.Minute), 0.9))

	r := newTestReviewer(t, store, stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureHabitShort: {1, 0},
	}}, 0.80)
	rev, err := r.Review(ctx, dup)
	if err != nil {
		t.Fatalf("Review: %v", err)
	}
	if rev.Decision != IngestSkip {
		t.Fatalf("decision = %q, want %q", rev.Decision, IngestSkip)
	}
	if rev.TargetID != existing.ID {
		t.Fatalf("命中的既有 id = %q, want %q", rev.TargetID, existing.ID)
	}
	got := readReviewRow(t, store, dup.ID)
	if got.status != ingestStatusMerged || got.mergedInto != existing.ID {
		t.Fatalf("重复行未折回既有行：status=%q merged_into=%q", got.status, got.mergedInto)
	}
	survivor := readReviewRow(t, store, existing.ID)
	if survivor.status != tidyStatusActive || survivor.mergeCount != 2 {
		t.Fatalf("既有行 survivor = %+v, want active 且 merge_count=2", survivor)
	}
	if n := countActiveRows(t, store, "persona.a"); n != 1 {
		t.Fatalf("活跃行数 = %d, want 1（反复新建没有被治住）", n)
	}
	reviews, err := store.ListIngestReviews(ctx, 10)
	if err != nil {
		t.Fatalf("ListIngestReviews: %v", err)
	}
	if len(reviews) != 1 || reviews[0].Decision != IngestSkip || reviews[0].RuleVer != IngestRuleVersion {
		t.Fatalf("审查留痕 = %+v, want 恰为一条 skip 且带规则版本", reviews)
	}
	if reviews[0].SessionID != TidySessionID {
		t.Fatalf("留痕会话 = %q, want 复用 %q（记忆管理会话）", reviews[0].SessionID, TidySessionID)
	}
}

// TestReviewSubsetMergesAndIsRevertable 是 design/23 §15 D3-9②：真包含 ⇒ 合并，
// 且**旧行可 revert**（软删除绝不物理删除，§8）。
func TestReviewSubsetMergesAndIsRevertable(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	existing := seedTidyItem(t, store, reviewAt(fixtureHabitLong, base, 0.9))
	short := seedTidyItem(t, store, reviewAt(fixtureHabitShort, base.Add(time.Minute), 0.9))

	r := newTestReviewer(t, store, stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureHabitLong:  {1, 0},
		fixtureHabitShort: simVector(0.95),
	}}, 0.80)
	rev, err := r.Review(ctx, short)
	if err != nil {
		t.Fatalf("Review: %v", err)
	}
	if rev.Decision != IngestMerge || rev.TargetID != existing.ID {
		t.Fatalf("verdict = %+v, want merge 并入超集 %s", rev, existing.ID)
	}
	absorbed := readReviewRow(t, store, short.ID)
	if absorbed.status != ingestStatusMerged || absorbed.content != fixtureHabitShort {
		t.Fatalf("被并入行 = %+v, want merged 且原文逐字保留", absorbed)
	}
	survivor := readReviewRow(t, store, existing.ID)
	if survivor.mergeCount != 2 || survivor.content != fixtureHabitLong {
		t.Fatalf("存活行 = %+v, want merge_count=2 且 content 不改", survivor)
	}

	if err := store.RevertMerged(ctx, short.ID); err != nil {
		t.Fatalf("RevertMerged: %v", err)
	}
	restored := readReviewRow(t, store, short.ID)
	if restored.status != tidyStatusActive || restored.mergedInto != "" || restored.content != fixtureHabitShort {
		t.Fatalf("revert 后 = %+v, want active 且原文无损", restored)
	}
	afterRevert := readReviewRow(t, store, existing.ID)
	if afterRevert.mergeCount != 1 {
		t.Fatalf("revert 后存活行 merge_count = %d, want 1（登记也一并摘掉）", afterRevert.mergeCount)
	}
}

// TestReviewSupersetModifiesExisting 是 design/23 §15 D3-6 的「改写」路径：
// 既有条目是新条目的子集 ⇒ **新行承载更全的内容**，旧行置为被合并。
func TestReviewSupersetModifiesExisting(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	existing := seedTidyItem(t, store, reviewAt(fixtureHabitShort, base, 0.9))
	longer := seedTidyItem(t, store, reviewAt(fixtureHabitLong, base.Add(time.Minute), 0.9))

	r := newTestReviewer(t, store, stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureHabitShort: {1, 0},
		fixtureHabitLong:  simVector(0.95),
	}}, 0.80)
	rev, err := r.Review(ctx, longer)
	if err != nil {
		t.Fatalf("Review: %v", err)
	}
	if rev.Decision != IngestModify || rev.TargetID != existing.ID {
		t.Fatalf("verdict = %+v, want modify（既有行被改写）", rev)
	}
	if got := readReviewRow(t, store, longer.ID); got.status != tidyStatusActive || got.content != fixtureHabitLong {
		t.Fatalf("新行 = %+v, want 存活且承载更全内容", got)
	}
	old := readReviewRow(t, store, existing.ID)
	if old.status != ingestStatusMerged || old.mergedInto != longer.ID {
		t.Fatalf("旧行 = %+v, want merged 进新行", old)
	}
	if n := countActiveRows(t, store, "persona.a"); n != 1 {
		t.Fatalf("活跃行数 = %d, want 1", n)
	}
}

// TestReviewDissimilarAndParallelFactsCreate 是 design/23 §15 D3-9③ 与主人红线
// 「不要什么内容都合并」：不相似、以及**相似但并列**的事实都正常新建。
func TestReviewDissimilarAndParallelFactsCreate(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	seedTidyItem(t, store, reviewAt(fixtureEnergyRedBull, base, 0.95))
	parallel := seedTidyItem(t, store, reviewAt(fixtureEnergyIced, base.Add(time.Minute), 0.95))

	r := newTestReviewer(t, store, stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureEnergyRedBull: {1, 0},
		fixtureEnergyIced:    simVector(0.99),
	}}, 0.80)
	rev, err := r.Review(ctx, parallel)
	if err != nil {
		t.Fatalf("Review: %v", err)
	}
	if rev.Decision != IngestCreate {
		t.Fatalf("并列事实被判成 %q，主人红线是「不要什么内容都合并」", rev.Decision)
	}
	if n := countActiveRows(t, store, "persona.a"); n != 2 {
		t.Fatalf("活跃行数 = %d, want 2", n)
	}
}

// TestReviewTemporalPollutionIsNeverMerged 是 design/23 §5.2-1：含时效词
// （假期/期间…）的记忆不参与合并 —— 有期限的状态不是长期事实。
func TestReviewTemporalPollutionIsNeverMerged(t *testing.T) {
	t.Parallel()
	ctx := context.Background()
	store := newTestStore(t)
	base := tidyTestTime(t)
	seedTidyItem(t, store, reviewAt(fixtureHabitPlain, base, 0.9))
	vacation := seedTidyItem(t, store, reviewAt(fixtureHabitVacation, base.Add(time.Minute), 0.9))

	r := newTestReviewer(t, store, stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureHabitPlain:    {1, 0},
		fixtureHabitVacation: simVector(0.97),
	}}, 0.80)
	rev, err := r.Review(ctx, vacation)
	if err != nil {
		t.Fatalf("Review: %v", err)
	}
	if rev.Decision != IngestCreate || rev.Veto != VetoTemporal {
		t.Fatalf("verdict = %+v, want create + temporal 否决", rev)
	}
	if q := readReviewRow(t, store, vacation.ID); q.status != tidyStatusActive {
		t.Fatalf("时效条目被改写了：%+v", q)
	}
}

// TestReviewFailureNeverTouchesMemory 是 design/23 §15 D3-9④ 的 fail-safe：
// 审查不可用（超时/取消）⇒ **照旧保留**（一行都不改写）+ 交给调用方记 WARN。
func TestReviewFailureNeverTouchesMemory(t *testing.T) {
	t.Parallel()
	store := newTestStore(t)
	base := tidyTestTime(t)
	existing := seedTidyItem(t, store, reviewAt(fixtureHabitLong, base, 0.9))
	short := seedTidyItem(t, store, reviewAt(fixtureHabitShort, base.Add(time.Minute), 0.9))

	r := newTestReviewer(t, store, stubEmbedder{vecs: map[string]embedding.Vector{
		fixtureHabitLong:  {1, 0},
		fixtureHabitShort: simVector(0.95),
	}}, 0.80)
	// 已经取消的 context：审查必须如实失败，而不是「猜一个结论」。
	dead, cancel := context.WithCancel(context.Background())
	cancel()
	if _, err := r.Review(dead, short); err == nil {
		t.Fatal("审查在不可用时返回了成功，必须如实报错")
	}
	if got := readReviewRow(t, store, short.ID); got.status != tidyStatusActive {
		t.Fatalf("审查失败却改写了记忆：%+v", got)
	}
	if got := readReviewRow(t, store, existing.ID); got.mergeCount != 1 {
		t.Fatalf("审查失败却改了 survivor：%+v", got)
	}
	if reviews, err := store.ListIngestReviews(context.Background(), 10); err != nil || len(reviews) != 0 {
		t.Fatalf("审查失败不该留决定留痕：reviews=%+v err=%v", reviews, err)
	}
}