//! 想法/知识域 Repo:IdeaRepo / KnowledgeRepo / KnowledgeEventsRepo + 向量工具(embedding BLOB 序列化 + 余弦相似度)

use std::sync::Arc;

use rusqlite::{params, Connection, OptionalExtension, Row};
use tokio::sync::Mutex;

use df_types::error::Result;

use crate::db::Database;
use crate::models::{IdeaRecord, KnowledgeEventRecord, KnowledgeRecord};
use df_types::types::IdeaStatus;

use super::impl_repo;
use super::{now_millis_str, storage_err, validate_column_name};

// ============================================================
// 知识库 SELECT 列清单(防 COLS 漂移)
// ============================================================

/// `knowledges` 表对应 `KnowledgeRecord` 15 个字段的列名(顺序与结构体一致)。
///
/// 多处 `search`/`search_vector` 内联 COLS 串的 DRY 收口(CR-260615-03):集中一处定义,
/// 配合下方 `KNOWLEDGE_COL_COUNT` 断言,任一处加列漏改会被测试 `test_knowledge_cols_matches_record`
/// 立即捕获(`knowledge_from_row` 按 name 取列,SELECT 漏列会运行时 rusqlite 报错,故提前断言)。
///
/// V23 新增 embedding_status 列(嵌入失败可补偿重试),已纳入列清单 + 计数。
const KNOWLEDGE_COLS: &str = "id,kind,title,content,tags,status,confidence,reuse_count,verified,source_project,source_ref,reasoning,embedding_status,created_at,updated_at";
/// `KnowledgeRecord` 字段数(与上面列清单的逗号分隔项数一致,被测试断言)。
/// 仅测试期消费(列漂移断言);保留为非 `cfg(test)` 以便测试外的阅读者一眼看到字段数。
#[cfg_attr(not(test), allow(dead_code))]
const KNOWLEDGE_COL_COUNT: usize = 15;
/// `search_vector` 用的列清单:KNOWLEDGE_COLS + embedding(余弦计算用,不入 KnowledgeRecord)。
const KNOWLEDGE_COLS_WITH_EMBEDDING: &str = concat!(
    "id,kind,title,content,tags,status,confidence,reuse_count,verified,",
    "source_project,source_ref,reasoning,embedding_status,created_at,updated_at,embedding"
);

/// `ideas` 表对应 `IdeaRecord` 14 个字段的列名(顺序与结构体一致)。
///
/// 同 KNOWLEDGE_COLS 的列漂移防护(CR-260615-03):idea 表 INSERT/UPDATE/from_row 三处
/// 各写一份列名串,加列须三处同步(如 V24 加 related_ids 即三处齐改),漏一处
/// 只在运行时 rusqlite 报错(INSERT 列数与参数数不匹配 / from_row 取不到列)。集中一处
/// 定义 + 配合 `IDEA_COL_COUNT` 断言 + 测试 `test_idea_cols_matches_record`,加列漏改即捕获。
///
/// V24 新增 related_ids 列(灵感间关联关系持久化打底),已纳入列清单 + 计数。
#[cfg_attr(not(test), allow(dead_code))]
const IDEA_COLS: &str = "id,title,description,status,priority,score,tags,source,promoted_to,ai_analysis,scores,related_ids,created_at,updated_at";
/// `IdeaRecord` 字段数(与上面列清单的逗号分隔项数一致,被测试断言)。
#[cfg_attr(not(test), allow(dead_code))]
const IDEA_COL_COUNT: usize = 14;

// ============================================================
// 向量工具 — embedding BLOB 序列化 + 余弦相似度
// ============================================================

/// Vec<f32> → 小端字节 BLOB
fn f32s_to_blob(v: &[f32]) -> Vec<u8> {
    v.iter().flat_map(|f| f.to_le_bytes()).collect()
}

/// BLOB → Vec<f32>(长度非 4 倍数时截断尾部残字节)
fn blob_to_f32s(blob: &[u8]) -> Vec<f32> {
    blob.chunks_exact(4)
        .map(|c| f32::from_le_bytes([c[0], c[1], c[2], c[3]]))
        .collect()
}

/// 余弦相似度(确定性数学,与任何实现结果一致)
fn cosine_similarity(a: &[f32], b: &[f32]) -> f32 {
    let dot: f32 = a.iter().zip(b).map(|(x, y)| x * y).sum();
    let norm_a: f32 = a.iter().map(|x| x * x).sum::<f32>().sqrt();
    let norm_b: f32 = b.iter().map(|x| x * x).sum::<f32>().sqrt();
    dot / (norm_a * norm_b + 1e-8)
}

/// 解析 JSON 数组字符串为 Vec<String>。NULL / 空 / 非法 → 空 Vec。
fn parse_json_id_array(raw: &Option<String>) -> Vec<String> {
    match raw {
        Some(s) if !s.is_empty() => match serde_json::from_str::<Vec<String>>(s) {
            Ok(v) => v,
            Err(_) => Vec::new(),
        },
        _ => Vec::new(),
    }
}

// ============================================================
// from_row 辅助函数
// ============================================================

fn idea_from_row(row: &Row<'_>) -> std::result::Result<IdeaRecord, rusqlite::Error> {
    Ok(IdeaRecord {
        id: row.get("id")?,
        title: row.get("title")?,
        description: row.get("description")?,
        status: {
            let s: String = row.get("status")?;
            IdeaStatus::from_db_str(&s).unwrap_or_default()
        },
        priority: row.get("priority")?,
        score: row.get("score")?,
        tags: row.get("tags")?,
        source: row.get("source")?,
        promoted_to: row.get("promoted_to")?,
        ai_analysis: row.get("ai_analysis")?,
        scores: row.get("scores")?,
        related_ids: row.get("related_ids")?,
        created_at: row.get("created_at")?,
        updated_at: row.get("updated_at")?,
    })
}

fn knowledge_from_row(row: &Row<'_>) -> std::result::Result<KnowledgeRecord, rusqlite::Error> {
    Ok(KnowledgeRecord {
        id: row.get("id")?,
        kind: row.get("kind")?,
        title: row.get("title")?,
        content: row.get("content")?,
        tags: row.get("tags")?,
        status: row.get("status")?,
        confidence: row.get("confidence")?,
        reuse_count: row.get("reuse_count")?,
        verified: row.get::<_, i32>("verified")? != 0,
        source_project: row.get("source_project")?,
        source_ref: row.get("source_ref")?,
        reasoning: row.get("reasoning")?,
        embedding_status: row.get("embedding_status")?,
        created_at: row.get("created_at")?,
        updated_at: row.get("updated_at")?,
    })
}

fn knowledge_event_from_row(row: &Row<'_>) -> std::result::Result<KnowledgeEventRecord, rusqlite::Error> {
    Ok(KnowledgeEventRecord {
        id: row.get("id")?,
        knowledge_id: row.get("knowledge_id")?,
        event_type: row.get("event_type")?,
        source_ref: row.get("source_ref")?,
        context_json: row.get("context_json")?,
        timestamp: row.get("timestamp")?,
    })
}

// ============================================================
// Repo 实现
// ============================================================

// ============================================================
// IdeaQuery — 多条件查询入参(F-260621-02 status 下沉 + 关键词 + 排序 + 分页)
// ============================================================

/// 灵感多条件查询入参。
///
/// 所有字段可选;全 None → 等价 `list_all`(向后兼容旧全量调用)。
/// 设计对齐 `查询能力补全方案-2026-06-21.md` 4.1:可选字段 struct 而非逐个加 IPC 参数,
/// 复用 `KnowledgeRepo::search` 的动态 WHERE 拼接模式(if-let 分支拼 SQL + 分支化参数绑定)。
///
/// - `status`:状态精确匹配(走 `idx_tasks_status` 同类索引语义,后端 WHERE 收口前端 filter)
/// - `keyword`:`title LIKE %kw% OR description LIKE %kw%`(对齐知识库 LIKE 检索,不上 FTS5)
/// - `order_by`:白名单枚举(`created_at`/`updated_at`/`priority`/`status`/`score`,
///   见 `validate_idea_order_by`,防 SQL 注入;不进字符串拼接)
/// - `limit`/`offset`:钳制上限 200(对齐 `KnowledgeEventsRepo::list_recent`)
///
/// `Deserialize`:Tauri IPC 从前端 JSON 反序列化为命令参数。
/// `Default`:命令层兼容旧 `status` 单参数路径(构造 `IdeaQuery { status, ..Default }`)。
#[derive(Debug, Clone, Default, serde::Deserialize)]
pub struct IdeaQuery {
    pub status: Option<String>,
    pub keyword: Option<String>,
    pub order_by: Option<String>,
    pub limit: Option<u32>,
    pub offset: Option<u32>,
}

/// `order_by` 白名单校验(防 SQL 注入,对齐 `impl_repo!` 宏 `validate_column_name` 思路)。
///
/// 列名直进 SQL 字符串(`ORDER BY {col} DESC`),故必须白名单枚举校验,不接受任意字符串。
/// 允许的排序列:created_at(默认/新在前)、updated_at(最近更新)、priority(优先级)、
/// status(状态聚合)、score(评分,后端 NULL 视为 0)。
fn validate_idea_order_by(col: &str) -> df_types::error::Result<&'static str> {
    Ok(match col {
        "created_at" => "created_at",
        "updated_at" => "updated_at",
        "priority" => "priority",
        "status" => "status",
        "score" => "score",
        _ => {
            return Err(df_types::error::Error::Storage(format!(
                "ideas order_by 不允许的字段名: {col}"
            )))
        }
    })
}

impl_repo!(
    /// 想法表 CRUD
    IdeaRepo,
    IdeaRecord,
    "ideas",
    from_row => |row| idea_from_row(row),
    insert  => |conn, rec| {
        conn.execute(
            "INSERT INTO ideas (id, title, description, status, priority, score, tags, source, promoted_to, ai_analysis, scores, related_ids, created_at, updated_at)
             VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9, ?10, ?11, ?12, ?13, ?14)",
            params![
                rec.id, rec.title, rec.description, rec.status.as_str(), rec.priority,
                rec.score, rec.tags, rec.source, rec.promoted_to, rec.ai_analysis,
                rec.scores, rec.related_ids, rec.created_at, rec.updated_at
            ],
        )
    },
    update  => |conn, rec| {
        conn.execute(
            "UPDATE ideas SET title = ?1, description = ?2, status = ?3, priority = ?4, score = ?5, tags = ?6, source = ?7, promoted_to = ?8, ai_analysis = ?9, scores = ?10, related_ids = ?11, updated_at = ?12 WHERE id = ?13",
            params![
                rec.title, rec.description, rec.status.as_str(), rec.priority,
                rec.score, rec.tags, rec.source, rec.promoted_to, rec.ai_analysis,
                rec.scores, rec.related_ids, rec.updated_at, rec.id
            ],
        )
    }
);

impl_repo!(
    /// 知识库表 CRUD
    KnowledgeRepo,
    KnowledgeRecord,
    "knowledges",
    from_row => |row| knowledge_from_row(row),
    insert  => |conn, rec| {
        let verified = if rec.verified { 1i32 } else { 0i32 };
        conn.execute(
            "INSERT INTO knowledges (id, kind, title, content, tags, status, confidence, reuse_count, verified, source_project, source_ref, reasoning, embedding_status, created_at, updated_at)
             VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9, ?10, ?11, ?12, ?13, ?14, ?15)",
            params![
                rec.id, rec.kind, rec.title, rec.content, rec.tags, rec.status, rec.confidence,
                rec.reuse_count, verified, rec.source_project, rec.source_ref, rec.reasoning,
                rec.embedding_status, rec.created_at, rec.updated_at
            ],
        )
    },
    update  => |conn, rec| {
        let verified = if rec.verified { 1i32 } else { 0i32 };
        conn.execute(
            "UPDATE knowledges SET kind = ?1, title = ?2, content = ?3, tags = ?4, status = ?5, confidence = ?6, reuse_count = ?7, verified = ?8, source_project = ?9, source_ref = ?10, reasoning = ?11, embedding_status = ?12, updated_at = ?13 WHERE id = ?14",
            params![
                rec.kind, rec.title, rec.content, rec.tags, rec.status, rec.confidence,
                rec.reuse_count, verified, rec.source_project, rec.source_ref, rec.reasoning,
                rec.embedding_status, rec.updated_at, rec.id
            ],
        )
    }
);

// KnowledgeRepo 的整体更新已由 impl_repo! 宏统一生成的 update_full 提供。

impl IdeaRepo {
    /// 多条件查询:动态 WHERE 拼接(status / keyword) + 白名单排序 + 分页(F-260621-02)。
    ///
    /// 复用 `KnowledgeRepo::search` 的动态 WHERE 模式:if-let 分支按可选条件拼 SQL 片段,
    /// 各分支化参数绑定到 `?N` 占位符。`order_by` 经 `validate_idea_order_by` 白名单校验后
    /// 拼入(防 SQL 注入),`limit` 钳制上限 200(对齐 `KnowledgeEventsRepo::list_recent`)。
    ///
    /// **向后兼容**:空 query(全 None)→ `WHERE` 子句空 + 默认 `created_at DESC`,等价 `list_all`。
    ///
    /// `score DESC NULLS LAST`:SQLite NULL 在 ASC 升序最前、DESC 降序最后,但为与前端原
    /// computed 语义一致(`(b.score ?? 0) - (a.score ?? 0)`,NULL 当 0),显式 `COALESCE(score,0)`
    /// 把 NULL 当 0 排序,避免 NULL 意外下沉到结果末尾。
    pub async fn list_by_query(&self, q: &IdeaQuery) -> Result<Vec<IdeaRecord>> {
        // 解析 + 校验 order_by(默认 created_at DESC)
        let order_col = match &q.order_by {
            Some(o) => validate_idea_order_by(o)?,
            None => "created_at",
        };

        let conn = self.conn.clone();

        // 所有按值移动进闭包的量,提前 clone 避免 move 借用问题
        let status = q.status.clone();
        let keyword = q.keyword.clone();
        // 钳制 limit 上限 200(对齐 list_recent),默认不限制分页(None → 不拼 LIMIT)
        let limit_i: Option<i64> = q.limit.map(|l| (l.min(200)) as i64);
        let offset_i: i64 = q.offset.unwrap_or(0) as i64;

        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();

            // 累积 WHERE 子句 + 收集参数(按出现顺序绑定占位符)。
            // deleted_at IS NULL 恒带(常量条件无占位符),回收站任务不进结果
            // (对标 task_repo list_by_query 同款模式,防回收站泄漏,不可被 query 关闭)。
            let mut where_clauses: Vec<String> = vec!["deleted_at IS NULL".to_string()];
            let mut params_vec: Vec<Box<dyn rusqlite::ToSql>> = Vec::new();

            if let Some(s) = &status {
                // 占位符编号用 params_vec.len()+1(参数实际位置),非 where_clauses.len()+1
                // (where_clauses 含 deleted_at IS NULL 常量无占位符子句,len() 会偏移致 ?N 与参数错位
                //  — 父⑤⑤.1 加 deleted_at 恒带引入的潜伏 bug,非空 status/keyword 查询 rusqlite 报
                //  "needed N, got M"。对标 task_repo list_by_query ②.2 同款修复)
                where_clauses.push(format!("status = ?{}", params_vec.len() + 1));
                params_vec.push(Box::new(s.clone()));
            }
            if let Some(kw) = &keyword {
                let escaped = kw.replace('%', "\\%").replace('_', "\\_");
                let pat = format!("%{escaped}%");
                let p1 = params_vec.len() + 1;
                let p2 = p1 + 1;
                where_clauses.push(format!("(title LIKE ?{p1} OR description LIKE ?{p2}) ESCAPE '\\'"));
                params_vec.push(Box::new(pat.clone()));
                params_vec.push(Box::new(pat));
            }

            // where_clauses 恒含 deleted_at IS NULL(初始项),永非空,直接 join 拼 WHERE。
            let where_sql = format!(" WHERE {}", where_clauses.join(" AND "));

            // 排序:score 走 COALESCE(NULL 当 0,对齐前端 computed 语义);其余直接列名。
            // order_col 来自白名单 &'static str,format! 出来是 String,生命周期随 sql 一起 OK。
            let order_expr = if order_col == "score" {
                "COALESCE(score, 0) DESC".to_string()
            } else {
                format!("{order_col} DESC")
            };

            // LIMIT/OFFSET:limit 为 None → 不拼(全量),offset 仅在 limit 存在时有意义。
            let where_param_count = params_vec.len();
            let limit_sql_bound = match limit_i {
                Some(_) => format!(
                    " LIMIT ?{} OFFSET ?{}",
                    where_param_count + 1,
                    where_param_count + 2
                ),
                None => String::new(),
            };

            let sql = format!(
                "SELECT {IDEA_COLS} FROM ideas{where_sql} ORDER BY {order_expr}{limit_sql_bound}"
            );

            let mut stmt = guard.prepare(&sql).map_err(storage_err)?;

            // 组装参数引用数组(where 参数 + 可选 limit/offset)。
            // limit/offset 也压入 Box<Vec> 收口:借用引用需指向同一生命周期存活处,
            // 收口到 params_vec 后再统一取引用,避免局部 l 生命周期不足(E0597)。
            if let Some(l) = limit_i {
                params_vec.push(Box::new(l));
                params_vec.push(Box::new(offset_i));
            }
            let param_refs: Vec<&dyn rusqlite::ToSql> =
                params_vec.iter().map(|p| p.as_ref()).collect();

            let rows = stmt
                .query_map(param_refs.as_slice(), |row| idea_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }

    /// 软删:标记 deleted_at(进回收站,可恢复)。仅作用于未删灵感,返回是否命中。
    /// 对标 TaskRepo::soft_delete / ProjectRepo::soft_delete。
    pub async fn soft_delete(&self, id: &str) -> Result<bool> {
        let conn = self.conn.clone();
        let id = id.to_owned();
        let now = now_millis_str();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let affected = guard
                .execute(
                    "UPDATE ideas SET deleted_at = ?1, updated_at = ?1 WHERE id = ?2 AND deleted_at IS NULL",
                    params![now, id],
                )
                .map_err(storage_err)?;
            Ok(affected > 0)
        })
        .await
        .map_err(storage_err)?
    }

    /// 恢复:清 deleted_at(从回收站还原)。仅作用于已删灵感,返回是否命中。
    /// 对标 TaskRepo::restore / ProjectRepo::restore。
    pub async fn restore(&self, id: &str) -> Result<bool> {
        let conn = self.conn.clone();
        let id = id.to_owned();
        let now = now_millis_str();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let affected = guard
                .execute(
                    "UPDATE ideas SET deleted_at = NULL, updated_at = ?1 WHERE id = ?2 AND deleted_at IS NOT NULL",
                    params![now, id],
                )
                .map_err(storage_err)?;
            Ok(affected > 0)
        })
        .await
        .map_err(storage_err)?
    }

    /// 双向同步关联关系:原子地更新主体灵感及其所有关联目标的 `related_ids`。
    ///
    /// `subject_id` 的 `related_ids` 被设为 `new_target_ids`(全量替换);
    /// 新增的关联目标追加 `subject_id` 到其 `related_ids`;
    /// 移除的关联目标从中删除 `subject_id`。
    /// 全部操作在同一 SQLite 事务中完成,保证原子性。
    pub async fn sync_related_ids(
        &self,
        subject_id: &str,
        new_target_ids: &[String],
    ) -> Result<()> {
        use rusqlite::Transaction;
        let conn = self.conn.clone();
        let subject_id = subject_id.to_owned();
        let new_target_ids = new_target_ids.to_vec();
        let now = now_millis_str();
        tokio::task::spawn_blocking(move || {
            let mut guard = conn.blocking_lock();
            let tx: Transaction = guard
                .transaction()
                .map_err(storage_err)?;

            // 1. 读主体当前 related_ids
            let old_raw: Option<String> = tx
                .query_row(
                    "SELECT related_ids FROM ideas WHERE id = ?1",
                    params![subject_id],
                    |row| row.get(0),
                )
                .optional()
                .map_err(storage_err)?
                .flatten();

            // 2. 解析新旧集合
            let old_set: std::collections::HashSet<String> =
                parse_json_id_array(&old_raw).into_iter().collect();
            let new_set: std::collections::HashSet<String> =
                new_target_ids.iter().cloned().collect();

            let added: Vec<&str> = new_set
                .difference(&old_set)
                .map(|s| s.as_str())
                .filter(|id| *id != subject_id) // 不自关联
                .collect();
            let removed: Vec<&str> = old_set
                .difference(&new_set)
                .map(|s| s.as_str())
                .filter(|id| *id != subject_id)
                .collect();

            // 3. 更新 added 目标:追加 subject_id
            for target_id in &added {
                let cur: Option<String> = tx
                    .query_row(
                        "SELECT related_ids FROM ideas WHERE id = ?1",
                        params![target_id],
                        |row| row.get(0),
                    )
                    .optional()
                    .map_err(storage_err)?
                    .flatten();
                let mut ids: Vec<String> = parse_json_id_array(&cur);
                if !ids.iter().any(|i| i == &subject_id) {
                    ids.push(subject_id.clone());
                }
                let json = serde_json::to_string(&ids).map_err(|e| {
                    storage_err::<df_types::error::Error>(e.into())
                })?;
                tx.execute(
                    "UPDATE ideas SET related_ids = ?1, updated_at = ?2 WHERE id = ?3",
                    params![json, &now, target_id],
                )
                .map_err(storage_err)?;
            }

            // 4. 更新 removed 目标:移除 subject_id
            for target_id in &removed {
                let cur: Option<String> = tx
                    .query_row(
                        "SELECT related_ids FROM ideas WHERE id = ?1",
                        params![target_id],
                        |row| row.get(0),
                    )
                    .optional()
                    .map_err(storage_err)?
                    .flatten();
                let mut ids: Vec<String> = parse_json_id_array(&cur);
                ids.retain(|i| i != &subject_id);
                let json = serde_json::to_string(&ids).map_err(|e| {
                    storage_err::<df_types::error::Error>(e.into())
                })?;
                tx.execute(
                    "UPDATE ideas SET related_ids = ?1, updated_at = ?2 WHERE id = ?3",
                    params![json, &now, target_id],
                )
                .map_err(storage_err)?;
            }

            // 5. 更新主体
            let new_json = serde_json::to_string(&new_target_ids)
                .map_err(|e| storage_err::<df_types::error::Error>(e.into()))?;
            tx.execute(
                "UPDATE ideas SET related_ids = ?1, updated_at = ?2 WHERE id = ?3",
                params![new_json, &now, &subject_id],
            )
            .map_err(storage_err)?;

            tx.commit().map_err(storage_err)?;
            Ok(())
        })
        .await
        .map_err(storage_err)?
    }

    /// 列出回收站(deleted_at IS NOT NULL),按更新时间(≈删除时间)降序。
    /// 对标 TaskRepo::list_deleted / ProjectRepo::list_deleted。
    pub async fn list_deleted(&self) -> Result<Vec<IdeaRecord>> {
        let conn = self.conn.clone();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut stmt = guard
                .prepare(&format!(
                    "SELECT {IDEA_COLS} FROM ideas WHERE deleted_at IS NOT NULL ORDER BY updated_at DESC"
                ))
                .map_err(storage_err)?;
            let rows = stmt
                .query_map([], |row| idea_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }
}

impl KnowledgeRepo {
    /// 检索知识: title/content LIKE 匹配,可选 kind 过滤,按 reuse_count 降序,top-N
    ///
    /// 克制检索: top-N≤3(由调用方 limit 控制),精确匹配优先(语义模糊后做)。
    pub async fn search(&self, query: &str, kind: Option<&str>, limit: usize) -> Result<Vec<KnowledgeRecord>> {
        let conn = self.conn.clone();
        let escaped = query.replace('%', "\\%").replace('_', "\\_");
        let pattern = format!("%{escaped}%");
        let kind = kind.map(|s| s.to_owned());
        let limit_i = limit as i64;
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut results = Vec::new();
            if let Some(k) = &kind {
                let mut stmt = guard
                    .prepare(&format!("SELECT {KNOWLEDGE_COLS} FROM knowledges WHERE status = 'published' AND (title LIKE ?1 ESCAPE '\\' OR content LIKE ?2 ESCAPE '\\') AND kind = ?3 ORDER BY reuse_count DESC LIMIT ?4"))
                    .map_err(storage_err)?;
                let rows = stmt
                    .query_map(params![pattern, pattern, k, limit_i], |row| knowledge_from_row(row))
                    .map_err(storage_err)?;
                for r in rows {
                    results.push(r.map_err(storage_err)?);
                }
            } else {
                let mut stmt = guard
                    .prepare(&format!("SELECT {KNOWLEDGE_COLS} FROM knowledges WHERE status = 'published' AND (title LIKE ?1 ESCAPE '\\' OR content LIKE ?2 ESCAPE '\\') ORDER BY reuse_count DESC LIMIT ?3"))
                    .map_err(storage_err)?;
                let rows = stmt
                    .query_map(params![pattern, pattern, limit_i], |row| knowledge_from_row(row))
                    .map_err(storage_err)?;
                for r in rows {
                    results.push(r.map_err(storage_err)?);
                }
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }

    /// 按状态列出(审核收件箱用): 按 confidence 语义排序(high>medium>low),次按 created_at
    ///
    /// 用 CASE WHEN 替代纯 TEXT 排序(字典序 high>low>medium 非预期语义)。
    pub async fn list_by_status(&self, status: &str) -> Result<Vec<KnowledgeRecord>> {
        let conn = self.conn.clone();
        let status = status.to_owned();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut stmt = guard
                .prepare(&format!(
                    "SELECT {KNOWLEDGE_COLS} FROM knowledges WHERE status = ?1
                     ORDER BY CASE confidence
                         WHEN 'high' THEN 3
                         WHEN 'medium' THEN 2
                         WHEN 'low' THEN 1
                         ELSE 0
                     END DESC, created_at DESC",
                ))
                .map_err(storage_err)?;
            let rows = stmt
                .query_map(params![status], |row| knowledge_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }

    /// 复用计数 +1(SQL 行级原子操作,并发安全)
    pub async fn increment_reuse_count(&self, id: &str) -> Result<bool> {
        let conn = self.conn.clone();
        let id = id.to_owned();
        let now = now_millis_str();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let affected = guard
                .execute(
                    "UPDATE knowledges SET reuse_count = reuse_count + 1, updated_at = ?1 WHERE id = ?2",
                    params![now, id],
                )
                .map_err(storage_err)?;
            Ok(affected > 0)
        })
        .await
        .map_err(storage_err)?
    }

    /// 写入向量嵌入(BLOB = Vec<f32> 小端字节序列化)
    ///
    /// embedding 列不进 KnowledgeRecord(IPC 不需要传向量给前端),专用方法读写。
    /// V23:同时把 embedding_status 置 'done'(成功标记),供补偿重试逻辑判别。
    pub async fn set_embedding(&self, id: &str, embedding: &[f32]) -> Result<bool> {
        let conn = self.conn.clone();
        let id = id.to_owned();
        let blob = f32s_to_blob(embedding);
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let affected = guard
                .execute(
                    "UPDATE knowledges SET embedding = ?1, embedding_status = 'done' WHERE id = ?2",
                    params![blob, id],
                )
                .map_err(storage_err)?;
            Ok(affected > 0)
        })
        .await
        .map_err(storage_err)?
    }

    /// 标记嵌入生成失败(embedding_status = 'failed'),供补偿重试逻辑定位。
    ///
    /// 失败时不写 embedding(保持 NULL,检索侧 `embedding IS NOT NULL` 自然跳过该条走 LIKE)。
    /// 幂等:重复标记 failed 无副作用(同值覆写)。
    pub async fn mark_embedding_failed(&self, id: &str) -> Result<bool> {
        let conn = self.conn.clone();
        let id = id.to_owned();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let affected = guard
                .execute(
                    "UPDATE knowledges SET embedding_status = 'failed' WHERE id = ?1",
                    params![id],
                )
                .map_err(storage_err)?;
            Ok(affected > 0)
        })
        .await
        .map_err(storage_err)?
    }

    /// 列出 embedding_status = 'failed' 的已发布知识(补偿重试入口用)。
    ///
    /// 仅返回 published(候选/归档不参与检索,重试无意义),按 created_at 升序(老条目优先补)。
    pub async fn list_failed_embeddings(&self) -> Result<Vec<KnowledgeRecord>> {
        let conn = self.conn.clone();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut stmt = guard
                .prepare(&format!(
                    "SELECT {KNOWLEDGE_COLS} FROM knowledges \
                     WHERE status = 'published' AND embedding_status = 'failed' \
                     ORDER BY created_at ASC"
                ))
                .map_err(storage_err)?;
            let rows = stmt
                .query_map([], |row| knowledge_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }

    /// 向量检索: 加载全部 published 且有 embedding 的记录,纯 Rust 余弦相似度取 top-N
    ///
    /// 返回 (记录, 相似度分数)。数据量 <50k 时暴力遍历 <50ms,够用;
    /// 更大规模再升 sqlite-vec HNSW(结果不变,只提速)。
    ///
    /// 列限定: 显式列出所需列(与 search/list_by_status/top_used 一致),避免 SELECT *
    /// 拉到未知新增列;embedding 单独取(不入 KnowledgeRecord)。
    ///
    /// TODO(性能,低优先): 调用方(hybrid_search→merge_hybrid_results→build_knowledge_context)
    /// 实际只消费 id/kind/title/content/reuse_count;reasoning(AI 生成大文本)、tags、
    /// source_project/source_ref 等元字段未被使用却仍随每行读出。真正省 IO 需返回精简结构
    /// (如 KnowledgeVectorHit { id, kind, title, content, reuse_count })替换返回类型,
    /// 但这会改变 search_vector 签名与 merge_hybrid_results 调用契约——当前保守不动,
    /// 待向量检索量级或 reasoning 文本体积成为瓶颈再单独立项。SELECT 列化本身不省字段,
    /// 仅消除 SELECT * 的隐式依赖与未知列风险。
    pub async fn search_vector(&self, query_vec: &[f32], limit: usize) -> Result<Vec<(KnowledgeRecord, f32)>> {
        let conn = self.conn.clone();
        let query_vec = query_vec.to_vec();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            // 显式列: 15 个 KnowledgeRecord 字段(含 embedding_status)+ embedding(余弦计算用,不入 KnowledgeRecord)
            let mut stmt = guard
                .prepare(&format!(
                    "SELECT {KNOWLEDGE_COLS_WITH_EMBEDDING} FROM knowledges WHERE status = 'published' AND embedding IS NOT NULL"
                ))
                .map_err(storage_err)?;
            let rows = stmt
                .query_map([], |row| {
                    let rec = knowledge_from_row(row)?;
                    let blob: Vec<u8> = row.get("embedding")?;
                    Ok((rec, blob))
                })
                .map_err(storage_err)?;

            let mut scored: Vec<(KnowledgeRecord, f32)> = Vec::new();
            for r in rows {
                let (rec, blob) = r.map_err(storage_err)?;
                let emb = blob_to_f32s(&blob);
                // 维度不匹配(换过 embedding 模型的旧向量)跳过
                if emb.len() != query_vec.len() {
                    continue;
                }
                let score = cosine_similarity(&query_vec, &emb);
                scored.push((rec, score));
            }
            scored.sort_by(|a, b| b.1.partial_cmp(&a.1).unwrap_or(std::cmp::Ordering::Equal));
            scored.truncate(limit);
            Ok(scored)
        })
        .await
        .map_err(storage_err)?
    }

    /// 列出非归档知识(全部 status != 'archived'),按 confidence 语义排序
    pub async fn list_non_archived(&self) -> Result<Vec<KnowledgeRecord>> {
        let conn = self.conn.clone();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut stmt = guard
                .prepare(&format!(
                    "SELECT {KNOWLEDGE_COLS} FROM knowledges WHERE status != 'archived'
                     ORDER BY CASE confidence
                         WHEN 'high' THEN 3
                         WHEN 'medium' THEN 2
                         WHEN 'low' THEN 1
                         ELSE 0
                     END DESC, created_at DESC",
                ))
                .map_err(storage_err)?;
            let rows = stmt
                .query_map([], |row| knowledge_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }

    /// 热门知识(已发布,按复用次数降序)
    pub async fn top_used(&self, limit: usize) -> Result<Vec<KnowledgeRecord>> {
        let conn = self.conn.clone();
        let limit_i = limit as i64;
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut stmt = guard
                .prepare(&format!(
                    "SELECT {KNOWLEDGE_COLS} FROM knowledges WHERE status = 'published' ORDER BY reuse_count DESC LIMIT ?1"
                ))
                .map_err(storage_err)?;
            let rows = stmt
                .query_map(params![limit_i], |row| knowledge_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }
}

impl_repo!(
    /// 知识生命线事件表 CRUD(追加型审计表)
    KnowledgeEventsRepo,
    KnowledgeEventRecord,
    "knowledge_events",
    from_row => |row| knowledge_event_from_row(row),
    insert  => |conn, rec| {
        conn.execute(
            "INSERT INTO knowledge_events (id, knowledge_id, event_type, source_ref, context_json, timestamp)
             VALUES (?1, ?2, ?3, ?4, ?5, ?6)",
            params![rec.id, rec.knowledge_id, rec.event_type, rec.source_ref, rec.context_json, rec.timestamp],
        )
    },
    update  => |conn, rec| {
        conn.execute(
            "UPDATE knowledge_events SET knowledge_id = ?1, event_type = ?2, source_ref = ?3, context_json = ?4, timestamp = ?5 WHERE id = ?6",
            params![rec.knowledge_id, rec.event_type, rec.source_ref, rec.context_json, rec.timestamp, rec.id],
        )
    }
);

impl KnowledgeEventsRepo {
    /// 按知识 ID 查询全部事件(时间正序,构建生命线视图用)
    pub async fn list_by_knowledge(&self, knowledge_id: &str) -> Result<Vec<KnowledgeEventRecord>> {
        let conn = self.conn.clone();
        let knowledge_id = knowledge_id.to_owned();
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut stmt = guard
                .prepare("SELECT id,knowledge_id,event_type,source_ref,context_json,timestamp FROM knowledge_events WHERE knowledge_id = ?1 ORDER BY timestamp ASC")
                .map_err(storage_err)?;
            let rows = stmt
                .query_map(params![knowledge_id], |row| knowledge_event_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }

    /// 按知识 ID + event_type 查询最近 N 条(如引用记录翻页: type=referenced)
    pub async fn list_by_knowledge_type(
        &self,
        knowledge_id: &str,
        event_type: &str,
        limit: usize,
    ) -> Result<Vec<KnowledgeEventRecord>> {
        let conn = self.conn.clone();
        let knowledge_id = knowledge_id.to_owned();
        let event_type = event_type.to_owned();
        let limit_i = limit as i64;
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut stmt = guard
                .prepare("SELECT id,knowledge_id,event_type,source_ref,context_json,timestamp FROM knowledge_events WHERE knowledge_id = ?1 AND event_type = ?2 ORDER BY timestamp DESC LIMIT ?3")
                .map_err(storage_err)?;
            let rows = stmt
                .query_map(params![knowledge_id, event_type, limit_i], |row| knowledge_event_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }

    /// 跨知识列最近 N 条事件(全表 timestamp DESC,top-N)。
    ///
    /// **专用兜底方法**:本表时间列名是 `timestamp` 而非 `created_at`,但
    /// `impl_repo!` 宏生成的 `query()` / `list_all()` 硬编码 `ORDER BY created_at`
    /// (见宏内 `ORDER BY created_at DESC` 字面量),误调 `state.knowledge_events.query(...)`
    /// 会触发 SQLite "no such column: created_at"。本方法走专用 SELECT 绕过宏硬编码,
    /// 供需要跨知识按时间倒序浏览事件的调用方使用(对标 AiToolExecutionRepo::list_recent
    /// 对 ai_tool_executions 表的同款兜底处理——那张表同样无 created_at 列)。
    /// limit 上限钳制 200,防前端恶意/失误传超大值。
    pub async fn list_recent(&self, limit: u32) -> Result<Vec<KnowledgeEventRecord>> {
        let conn = self.conn.clone();
        // 钳制 limit 防滥用(最大 200)
        let safe_limit = limit.min(200) as i64;
        tokio::task::spawn_blocking(move || {
            let guard = conn.blocking_lock();
            let mut stmt = guard
                .prepare(
                    "SELECT id,knowledge_id,event_type,source_ref,context_json,timestamp \
                     FROM knowledge_events ORDER BY timestamp DESC LIMIT ?1",
                )
                .map_err(storage_err)?;
            let rows = stmt
                .query_map(params![safe_limit], |row| knowledge_event_from_row(row))
                .map_err(storage_err)?;
            let mut results = Vec::new();
            for r in rows {
                results.push(r.map_err(storage_err)?);
            }
            Ok(results)
        })
        .await
        .map_err(storage_err)?
    }
}

// ============================================================
// 单元测试 — 向量纯函数 + KnowledgeRepo 内存 DB + COLS 漂移防护
// ============================================================

#[cfg(test)]
mod tests {
    use super::*;
    use crate::db::Database;

    // ---------- COLS 漂移防护(CR-260615-03) ----------

    /// KNOWLEDGE_COLS 列数须等于 KNOWLEDGE_COL_COUNT(任一处漂移:加列漏改 / 串错位 → 立即失败)。
    /// `knowledge_from_row` 按 name 取列,SELECT 漏列会在运行时被 rusqlite 报错;此断言提前到测试期捕获。
    #[test]
    fn test_knowledge_cols_matches_record() {
        let count = KNOWLEDGE_COLS.split(',').count();
        assert_eq!(
            count, KNOWLEDGE_COL_COUNT,
            "KNOWLEDGE_COLS({count}列) ≠ KNOWLEDGE_COL_COUNT({KNOWLEDGE_COL_COUNT}); \
             修改一处须同步另一处"
        );
        // search_vector 多一列 embedding
        let count_with_emb = KNOWLEDGE_COLS_WITH_EMBEDDING.split(',').count();
        assert_eq!(
            count_with_emb,
            KNOWLEDGE_COL_COUNT + 1,
            "KNOWLEDGE_COLS_WITH_EMBEDDING({count_with_emb}列) ≠ KNOWLEDGE_COL_COUNT+1({}); \
             embedding 列应单独追加",
            KNOWLEDGE_COL_COUNT + 1
        );
        // 每个列名须能被 split 出来(防末尾多逗号 / 空段)
        for col in KNOWLEDGE_COLS.split(',') {
            assert!(!col.is_empty(), "KNOWLEDGE_COLS 含空列名段");
        }
    }

    /// IDEA_COLS 列数须等于 IDEA_COL_COUNT(任一处漂移:加列漏改 INSERT/UPDATE/from_row
    /// 三处之一 → 立即失败)。`idea_from_row` 按 name 取列,SELECT/INSERT 漏列会在运行时被
    /// rusqlite 报错;此断言提前到测试期捕获(本次 V24 加 related_ids 已改 3 处的回归保险)。
    #[test]
    fn test_idea_cols_matches_record() {
        let count = IDEA_COLS.split(',').count();
        assert_eq!(
            count, IDEA_COL_COUNT,
            "IDEA_COLS({count}列) ≠ IDEA_COL_COUNT({IDEA_COL_COUNT}); \
             修改一处须同步另一处(INSERT/UPDATE/from_row/IDEA_COLS/IDEA_COL_COUNT 五处)"
        );
        // 每个列名须能被 split 出来(防末尾多逗号 / 空段)
        for col in IDEA_COLS.split(',') {
            assert!(!col.is_empty(), "IDEA_COLS 含空列名段");
        }
    }

    // ---------- 向量纯函数 ----------

    #[test]
    fn f32s_blob_roundtrip_nonempty() {
        let v = vec![0.0, 1.5, -2.25, 3.14159, -0.0001];
        let blob = f32s_to_blob(&v);
        assert_eq!(blob.len(), v.len() * 4);
        let back = blob_to_f32s(&blob);
        assert_eq!(back, v);
    }

    #[test]
    fn f32s_blob_roundtrip_empty() {
        let v: Vec<f32> = vec![];
        let blob = f32s_to_blob(&v);
        assert!(blob.is_empty());
        assert!(blob_to_f32s(&blob).is_empty());
    }

    #[test]
    fn blob_to_f32s_drops_trailing_partial_bytes() {
        // 1 个完整 f32 + 3 残字节 → chunks_exact 截断尾部
        let blob = f32s_to_blob(&[42.0]);
        let with_garbage: Vec<u8> = blob.into_iter().chain([0xff, 0xff, 0xff]).collect();
        assert_eq!(blob_to_f32s(&with_garbage), vec![42.0]);
    }

    #[test]
    fn cosine_identical_vectors_near_one() {
        let a = vec![1.0, 2.0, 3.0];
        let sim = cosine_similarity(&a, &a);
        assert!((sim - 1.0).abs() < 1e-5, "identical ≈ 1.0, got {}", sim);
    }

    #[test]
    fn cosine_orthogonal_vectors_near_zero() {
        let a = vec![1.0, 0.0];
        let b = vec![0.0, 1.0];
        let sim = cosine_similarity(&a, &b);
        assert!(sim.abs() < 1e-5, "orthogonal ≈ 0.0, got {}", sim);
    }

    #[test]
    fn cosine_opposite_vectors_near_neg_one() {
        let a = vec![1.0, 2.0, 3.0];
        let b = vec![-1.0, -2.0, -3.0];
        let sim = cosine_similarity(&a, &b);
        assert!((sim + 1.0).abs() < 1e-5, "opposite ≈ -1.0, got {}", sim);
    }

    #[test]
    fn cosine_dimension_mismatch_zips_to_shortest() {
        // 实现用 zip(a,b),维度不匹配按较短的那个对齐,不 panic
        let a = vec![1.0, 0.0, 0.0]; // 3 维
        let b = vec![1.0, 0.0]; // 2 维 → zip 取前 2 个
        let sim = cosine_similarity(&a, &b);
        assert!((sim - 1.0).abs() < 1e-5, "zip 截断后应 ≈ 1.0, got {}", sim);
    }

    #[test]
    fn cosine_zero_vector_does_not_panic() {
        // 分母有 +1e-8 保护,零向量返回有限值而非 NaN/inf
        let z = vec![0.0, 0.0, 0.0];
        let sim = cosine_similarity(&z, &z);
        assert!(sim.is_finite(), "零向量相似度应有限, got {}", sim);
    }

    // ---------- KnowledgeRepo 内存 DB ----------

    /// 构造一条 KnowledgeRecord fixture
    fn krec(
        id: &str,
        title: &str,
        content: &str,
        kind: &str,
        status: &str,
        confidence: Option<&str>,
        reuse: i32,
    ) -> KnowledgeRecord {
        KnowledgeRecord {
            id: id.to_string(),
            kind: kind.to_string(),
            title: title.to_string(),
            content: content.to_string(),
            tags: Some("[]".to_string()),
            status: status.to_string(),
            confidence: confidence.map(|s| s.to_string()),
            reuse_count: reuse,
            verified: false,
            source_project: Some("proj-1".to_string()),
            source_ref: Some("conv:c1".to_string()),
            reasoning: None,
            embedding_status: None,
            created_at: "1700000000000".to_string(),
            updated_at: "1700000000000".to_string(),
        }
    }

    async fn setup_repo() -> KnowledgeRepo {
        let db = Database::open_in_memory().await.expect("open_in_memory");
        KnowledgeRepo::new(&db)
    }

    #[tokio::test]
    async fn search_matches_title_and_filters_non_published() {
        let repo = setup_repo().await;
        // 一条命中(title 含关键词)、一条 published 不命中、一条 status 非 published 但命中
        repo.insert(krec("k1", "Rust 异步并发模型", "tokio 运行时", "lesson", "published", Some("high"), 5))
            .await
            .unwrap();
        repo.insert(krec("k2", "无关标题", "无关内容", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.insert(krec("k3", "Rust 异步进阶", "...", "lesson", "candidate", Some("low"), 9))
            .await
            .unwrap();

        let hits = repo.search("异步", None, 10).await.unwrap();
        // 只有 k1 是 published 且命中;k3 命中但非 published 被过滤
        assert_eq!(hits.len(), 1);
        assert_eq!(hits[0].id, "k1");
    }

    #[tokio::test]
    async fn search_matches_content_branch() {
        let repo = setup_repo().await;
        // title 不含关键词,content 含 → 命中 content LIKE 分支
        repo.insert(krec("k1", "标题", "深入理解 Rust 所有权与借用", "lesson", "published", None, 1))
            .await
            .unwrap();

        let hits = repo.search("所有权", None, 10).await.unwrap();
        assert_eq!(hits.len(), 1);
        assert_eq!(hits[0].id, "k1");
    }

    #[tokio::test]
    async fn search_no_keyword_match_returns_empty() {
        let repo = setup_repo().await;
        repo.insert(krec("k1", "Rust", "tokio", "lesson", "published", None, 1))
            .await
            .unwrap();

        let hits = repo.search("不存在的关键词xyz", None, 10).await.unwrap();
        assert!(hits.is_empty());
    }

    #[tokio::test]
    async fn search_kind_filter_narrows_results() {
        let repo = setup_repo().await;
        repo.insert(krec("k1", "Rust 规范", "...", "lesson", "published", None, 1))
            .await
            .unwrap();
        repo.insert(krec("k2", "Rust 决策", "...", "decision", "published", None, 1))
            .await
            .unwrap();

        // 不过滤 kind → 两条都命中
        let all = repo.search("Rust", None, 10).await.unwrap();
        assert_eq!(all.len(), 2);
        // 只取 lesson → 仅 k1
        let only_lesson = repo.search("Rust", Some("lesson"), 10).await.unwrap();
        assert_eq!(only_lesson.len(), 1);
        assert_eq!(only_lesson[0].id, "k1");
    }

    #[tokio::test]
    async fn search_orders_by_reuse_count_desc() {
        let repo = setup_repo().await;
        repo.insert(krec("low", "Rust A", "...", "lesson", "published", None, 1))
            .await
            .unwrap();
        repo.insert(krec("high", "Rust B", "...", "lesson", "published", None, 50))
            .await
            .unwrap();
        repo.insert(krec("mid", "Rust C", "...", "lesson", "published", None, 10))
            .await
            .unwrap();

        let hits = repo.search("Rust", None, 10).await.unwrap();
        let ids: Vec<_> = hits.iter().map(|h| h.id.as_str()).collect();
        assert_eq!(ids, vec!["high", "mid", "low"]);
    }

    #[tokio::test]
    async fn search_truncates_to_limit() {
        let repo = setup_repo().await;
        for i in 0..5 {
            repo.insert(krec(&format!("k{i}"), &format!("Rust-{i}"), "...", "lesson", "published", None, i))
                .await
                .unwrap();
        }
        let hits = repo.search("Rust", None, 2).await.unwrap();
        assert_eq!(hits.len(), 2);
        // reuse_count 最高的两条(k4=4, k3=3)
        assert_eq!(hits[0].id, "k4");
        assert_eq!(hits[1].id, "k3");
    }

    #[tokio::test]
    async fn list_by_status_orders_by_confidence_semantics() {
        // confidence 字典序 high>low>medium 非预期,实现用 CASE 强制 high>medium>low
        let repo = setup_repo().await;
        repo.insert(krec("low", "t", "c", "lesson", "pending_review", Some("low"), 0))
            .await
            .unwrap();
        repo.insert(krec("high", "t", "c", "lesson", "pending_review", Some("high"), 0))
            .await
            .unwrap();
        repo.insert(krec("medium", "t", "c", "lesson", "pending_review", Some("medium"), 0))
            .await
            .unwrap();
        // created_at 相同,纯靠 confidence 排序

        let list = repo.list_by_status("pending_review").await.unwrap();
        let confidences: Vec<_> = list.iter().map(|r| r.confidence.as_deref().unwrap_or("")).collect();
        assert_eq!(confidences, vec!["high", "medium", "low"]);
    }

    #[tokio::test]
    async fn list_by_status_filters_other_status() {
        let repo = setup_repo().await;
        repo.insert(krec("a", "t", "c", "lesson", "pending_review", Some("high"), 0))
            .await
            .unwrap();
        repo.insert(krec("b", "t", "c", "lesson", "published", Some("high"), 0))
            .await
            .unwrap();

        let list = repo.list_by_status("pending_review").await.unwrap();
        assert_eq!(list.len(), 1);
        assert_eq!(list[0].id, "a");
    }

    #[tokio::test]
    async fn increment_reuse_count_is_atomic_plus_one() {
        let repo = setup_repo().await;
        repo.insert(krec("k1", "Rust", "...", "lesson", "published", None, 0))
            .await
            .unwrap();

        // 连续 +1 两次
        assert!(repo.increment_reuse_count("k1").await.unwrap());
        assert!(repo.increment_reuse_count("k1").await.unwrap());

        let rec = repo.get_by_id("k1").await.unwrap().expect("记录存在");
        assert_eq!(rec.reuse_count, 2);

        // 不存在的 id → false
        assert!(!repo.increment_reuse_count("nope").await.unwrap());
    }

    #[tokio::test]
    async fn search_vector_orders_by_cosine_and_truncates() {
        let repo = setup_repo().await;
        // 三条 published 记录,embedding 维度均为 2
        repo.insert(krec("exact", "e", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.insert(krec("orth", "e", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.insert(krec("neg", "e", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        // 一条非 published(应被过滤)
        repo.insert(krec("hidden", "e", "c", "lesson", "candidate", None, 0))
            .await
            .unwrap();

        repo.set_embedding("exact", &[1.0, 0.0]).await.unwrap(); // 与 query 完全同向
        repo.set_embedding("orth", &[0.0, 1.0]).await.unwrap(); // 正交 ≈ 0
        repo.set_embedding("neg", &[-1.0, 0.0]).await.unwrap(); // 反向 ≈ -1
        repo.set_embedding("hidden", &[1.0, 0.0]).await.unwrap(); // 非 published 过滤掉

        let results = repo.search_vector(&[1.0, 0.0], 10).await.unwrap();
        // hidden 被 status 过滤,剩 3 条
        assert_eq!(results.len(), 3);
        // 按相似度降序: exact(≈1) > orth(≈0) > neg(≈-1)
        assert_eq!(results[0].0.id, "exact");
        assert_eq!(results[1].0.id, "orth");
        assert_eq!(results[2].0.id, "neg");
        assert!((results[0].1 - 1.0).abs() < 1e-5);
        assert!(results[1].1.abs() < 1e-5);
        assert!((results[2].1 + 1.0).abs() < 1e-5);
    }

    #[tokio::test]
    async fn search_vector_skips_dimension_mismatch() {
        let repo = setup_repo().await;
        repo.insert(krec("dim2", "e", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.insert(krec("dim3", "e", "c", "lesson", "published", None, 0))
            .await
            .unwrap();

        repo.set_embedding("dim2", &[1.0, 0.0]).await.unwrap();
        repo.set_embedding("dim3", &[1.0, 0.0, 0.0]).await.unwrap(); // 维度不匹配

        // query 是 2 维,dim3 被跳过
        let results = repo.search_vector(&[1.0, 0.0], 10).await.unwrap();
        assert_eq!(results.len(), 1);
        assert_eq!(results[0].0.id, "dim2");
    }

    #[tokio::test]
    async fn search_vector_truncates_to_limit() {
        let repo = setup_repo().await;
        for i in 0..4 {
            repo.insert(krec(&format!("k{i}"), "e", "c", "lesson", "published", None, 0))
                .await
                .unwrap();
            // 全部与 query 同向,相似度相同 → 仅验证 truncate 生效
            repo.set_embedding(&format!("k{i}"), &[1.0, 0.0]).await.unwrap();
        }
        let results = repo.search_vector(&[1.0, 0.0], 2).await.unwrap();
        assert_eq!(results.len(), 2);
    }

    #[tokio::test]
    async fn search_vector_ignores_records_without_embedding() {
        let repo = setup_repo().await;
        // published 但未写 embedding
        repo.insert(krec("noemb", "e", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        let results = repo.search_vector(&[1.0, 0.0], 10).await.unwrap();
        assert!(results.is_empty());
    }

    // ---------- V23 embedding_status 补偿重试 ----------

    #[tokio::test]
    async fn set_embedding_marks_status_done() {
        // V23:set_embedding 成功写入时应同步置 embedding_status='done'
        let repo = setup_repo().await;
        repo.insert(krec("k1", "t", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.set_embedding("k1", &[1.0, 0.0]).await.unwrap();
        let rec = repo.get_by_id("k1").await.unwrap().expect("记录存在");
        assert_eq!(rec.embedding_status.as_deref(), Some("done"));
    }

    #[tokio::test]
    async fn mark_embedding_failed_sets_status() {
        let repo = setup_repo().await;
        repo.insert(krec("k1", "t", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        assert!(repo.mark_embedding_failed("k1").await.unwrap());
        let rec = repo.get_by_id("k1").await.unwrap().expect("记录存在");
        assert_eq!(rec.embedding_status.as_deref(), Some("failed"));
    }

    #[tokio::test]
    async fn mark_embedding_failed_idempotent() {
        // 重复标记 failed 无副作用
        let repo = setup_repo().await;
        repo.insert(krec("k1", "t", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.mark_embedding_failed("k1").await.unwrap();
        assert!(repo.mark_embedding_failed("k1").await.unwrap());
        let rec = repo.get_by_id("k1").await.unwrap().unwrap();
        assert_eq!(rec.embedding_status.as_deref(), Some("failed"));
    }

    #[tokio::test]
    async fn mark_embedding_failed_missing_row_returns_false() {
        let repo = setup_repo().await;
        // 不存在的 id → affected=0
        let ok = repo.mark_embedding_failed("ghost").await.unwrap();
        assert!(!ok);
    }

    #[tokio::test]
    async fn list_failed_embeddings_only_returns_published_failed() {
        let repo = setup_repo().await;
        // k1:published + failed → 应列出
        repo.insert(krec("k1", "t", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.mark_embedding_failed("k1").await.unwrap();
        // k2:candidate + failed → 不应列出(候选不参与检索,重试无意义)
        repo.insert(krec("k2", "t", "c", "lesson", "candidate", None, 0))
            .await
            .unwrap();
        repo.mark_embedding_failed("k2").await.unwrap();
        // k3:published + done → 不应列出
        repo.insert(krec("k3", "t", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.set_embedding("k3", &[1.0, 0.0]).await.unwrap();
        // k4:published + 未标记(NULL)→ 不应列出
        repo.insert(krec("k4", "t", "c", "lesson", "published", None, 0))
            .await
            .unwrap();

        let failed = repo.list_failed_embeddings().await.unwrap();
        let ids: Vec<_> = failed.iter().map(|r| r.id.as_str()).collect();
        assert_eq!(ids, vec!["k1"], "只应返回 published + failed 的条目");
    }

    #[tokio::test]
    async fn list_failed_embeddings_empty_when_none_failed() {
        let repo = setup_repo().await;
        repo.insert(krec("k1", "t", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.set_embedding("k1", &[1.0, 0.0]).await.unwrap();
        let failed = repo.list_failed_embeddings().await.unwrap();
        assert!(failed.is_empty());
    }

    #[tokio::test]
    async fn failed_retry_flow_done_after_set_embedding() {
        // 补偿重试完整流程:failed → set_embedding 成功 → done(不再出现在 list_failed)
        let repo = setup_repo().await;
        repo.insert(krec("k1", "t", "c", "lesson", "published", None, 0))
            .await
            .unwrap();
        repo.mark_embedding_failed("k1").await.unwrap();
        assert_eq!(repo.list_failed_embeddings().await.unwrap().len(), 1);
        // 重试成功
        repo.set_embedding("k1", &[0.5, 0.5]).await.unwrap();
        assert!(repo.list_failed_embeddings().await.unwrap().is_empty());
        let rec = repo.get_by_id("k1").await.unwrap().unwrap();
        assert_eq!(rec.embedding_status.as_deref(), Some("done"));
    }

    // ---------- IdeaRepo 软删回收站(对标 task_repo 软删测试,V28) ----------

    /// 构造一条 IdeaRecord fixture(14 字段全填,IdeaRecord 不含 deleted_at —— 纯 SQL 过滤)。
    fn irec(id: &str, title: &str) -> IdeaRecord {
        IdeaRecord {
            id: id.to_string(),
            title: title.to_string(),
            description: String::new(),
            status: IdeaStatus::Draft,
            priority: 1,
            score: None,
            tags: None,
            source: None,
            promoted_to: None,
            ai_analysis: None,
            scores: None,
            related_ids: None,
            created_at: "1700000000000".to_string(),
            updated_at: "1700000000000".to_string(),
        }
    }

    async fn setup_idea_repo() -> IdeaRepo {
        let db = Database::open_in_memory().await.expect("open_in_memory");
        IdeaRepo::new(&db)
    }

    #[tokio::test]
    async fn idea_soft_delete_marks_and_filters_from_list_by_query() {
        let repo = setup_idea_repo().await;
        repo.insert(irec("i1", "保留")).await.unwrap();
        repo.insert(irec("i2", "软删")).await.unwrap();

        // 软删 i2:返回 true(命中)
        assert!(repo.soft_delete("i2").await.unwrap());

        // list_by_query(空 query = 等价全量未删)应只返回 i1,i2 进回收站被过滤
        let active = repo.list_by_query(&IdeaQuery::default()).await.unwrap();
        let ids: Vec<_> = active.iter().map(|r| r.id.as_str()).collect();
        assert_eq!(ids, vec!["i1"], "soft_delete 后 list_by_query 应过滤回收站");
    }

    #[tokio::test]
    async fn idea_soft_delete_idempotent_on_deleted() {
        let repo = setup_idea_repo().await;
        repo.insert(irec("i1", "t")).await.unwrap();
        assert!(repo.soft_delete("i1").await.unwrap());
        // 已删再删:WHERE deleted_at IS NULL 不命中 → false
        assert!(!repo.soft_delete("i1").await.unwrap());
    }

    #[tokio::test]
    async fn idea_soft_delete_missing_returns_false() {
        let repo = setup_idea_repo().await;
        // 不存在的 id → affected=0
        assert!(!repo.soft_delete("ghost").await.unwrap());
    }

    #[tokio::test]
    async fn idea_restore_brings_back_to_active() {
        let repo = setup_idea_repo().await;
        repo.insert(irec("i1", "t")).await.unwrap();
        repo.soft_delete("i1").await.unwrap();
        // 恢复:返回 true,回到 list_by_query
        assert!(repo.restore("i1").await.unwrap());
        let active = repo.list_by_query(&IdeaQuery::default()).await.unwrap();
        let ids: Vec<_> = active.iter().map(|r| r.id.as_str()).collect();
        assert_eq!(ids, vec!["i1"], "restore 后灵感应回到活跃列表");
    }

    #[tokio::test]
    async fn idea_restore_only_affects_deleted() {
        let repo = setup_idea_repo().await;
        repo.insert(irec("i1", "t")).await.unwrap();
        // 未删灵感 restore:WHERE deleted_at IS NOT NULL 不命中 → false
        assert!(!repo.restore("i1").await.unwrap());
    }

    #[tokio::test]
    async fn idea_list_deleted_returns_only_trash_ordered_by_updated_desc() {
        let repo = setup_idea_repo().await;
        repo.insert(irec("i1", "活跃")).await.unwrap();
        repo.insert(irec("i2", "回收1")).await.unwrap();
        repo.insert(irec("i3", "回收2")).await.unwrap();

        // 软删 i2 / i3(updated_at 在 soft_delete 时被刷新为 now)
        repo.soft_delete("i2").await.unwrap();
        // 让 i3 的删除时间晚于 i2,保证 updated_at DESC 顺序确定(i3 在前)
        tokio::time::sleep(std::time::Duration::from_millis(5)).await;
        repo.soft_delete("i3").await.unwrap();

        let deleted = repo.list_deleted().await.unwrap();
        let ids: Vec<_> = deleted.iter().map(|r| r.id.as_str()).collect();
        // i1 活跃不出现;i3 删除最晚在前
        assert_eq!(ids, vec!["i3", "i2"], "list_deleted 应只含回收站灵感,按 updated_at DESC");
    }
}