为Android sqlite优化SQL语句

Question

我正在开发一个跟踪用户当前位置并将其存储到SQLite数据库的应用程序。一切工作正常，但现在我有问题时查询数据库超过1000条记录的轨道，它需要近1.5分钟。在我的桌面上，它只需要1秒。

我知道这是一个查询与许多子选择，但我不能以另一种方式得到正确的结果。在我看来，这属于像avg（）和sum（）这样的集合函数。

这里是我的查询：

Cursor c = readableDB 
    .rawQuery(
      "SELECT DISTINCT t._id , title , strftime('%Y-%m-%d' , starttime , 'unixepoch' , 'localtime') as date , description, " 
      + "round((SELECT sum(distToPrev)/1000 FROM positions p WHERE p.trackid=t._id) , 2) as distance , " 
      + "(SELECT count(latitude) FROM positions p WHERE p.trackid=t._id) as waypoints, " 
      + "(SELECT (avg(speed)*3.6) FROM positions p WHERE p.trackid=t._id) as avgspeed, " 
      + "(SELECT (max(speed)*3.6) FROM positions p WHERE p.trackid=t._id) as maxspeed, " 
      + "(SELECT sum(altitudeUP) FROM positions p WHERE p.trackid=t._id) as climb , " 
      + "(SELECT avg(heartbeat) FROM heartbeats h WHERE h.trackid=t._id) as avgheartbeat , " 
      + "(SELECT max(heartbeat) FROM heartbeats h WHERE h.trackid=t._id) as maxheartbeat , " 
      + "(SELECT avg(cadence) FROM heartbeats h WHERE h.trackid=t._id) as avgcadence " 
      + "FROM tracks t LEFT JOIN heartbeats h ON t._id = h.trackid WHERE t._id = ?", 
      new String[]{String.valueOf(trackId)}); 
    c.moveToFirst(); 

我怎样才能优化这个查询？我已经以这种方式尝试过了，但结果是错误的，花费了相同的时间。

SELECT t._id , title , strftime('%Y-%m-%d' , starttime , 'unixepoch' , 'localtime') as date , description, 
sum(distToPrev)/1000 as distance , 
count(latitude) as waypoints, 
(avg(speed)*3.6) as avgspeed, 
(max(speed)*3.6) as maxspeed, 
sum(altitudeUP) as climb , 
avg(heartbeat) as avgheartbeat , 
max(heartbeat) as maxheartbeat , 
avg(cadence) as avgcadence 
FROM tracks t 
LEFT JOIN heartbeats h ON t._id = h.trackid 
INNER JOIN positions p ON t._id = p.trackid 
    WHERE t._id = ? 

由于2个小时我正在寻找解决方案，我不知道我在做什么错。也许我必须休息一下。

编辑：

在这里，我创建语句：

CREATE TABLE heartbeats(_id INTEGER PRIMARY KEY AUTOINCREMENT, trackid INTEGER NOT NULL, heartbeat INTEGER NOT NULL, cadence INTEGER, timestamp TIMESTAMP); 


CREATE TABLE positions(_id INTEGER PRIMARY KEY AUTOINCREMENT, trackid INTEGER NOT NULL, longitude REAL NOT NULL, latitude REAL NOT NULL, altitude REAL, altitudeUP REAL, speed REAL, accuracy REAL, distToPrev REAL, timestamp TIMESTAMP); 


CREATE TABLE tracks(_id INTEGER PRIMARY KEY AUTOINCREMENT, title TEXT NOT NULL, description TEXT, starttime DATETIME NOT NULL, endtime DATETIME); 

Answer 1

这将做的工作：

SELECT tid, date, waypoints, avg(heartbeat) AS avgheartbeat 
FROM (SELECT t._id AS tid, strftime('%Y-%m-%d' , starttime , 'unixepoch' , 'localtime') AS date, count(latitude) AS waypoints 
     FROM tracks t LEFT JOIN positions p ON t._id = p.trackid 
     WHERE t._id = ? 
     GROUP BY t._id , strftime('%Y-%m-%d' , starttime , 'unixepoch' , 'localtime')) 
    LEFT JOIN heartbeats h ON tid = h.trackid 
GROUP BY tid, date, waypoints 

根据需要添加字段。

Answer 2

像这样的东西可能会关闭。我已经包括了参数中多次查询通知：

SELECT 
    T._id, 
    T.title, 
    STRFTIME('%Y-%m-%d' , T.starttime , 'unixepoch' , 'localtime') as date, 
    T.description, 
    P_SQ.distance, 
    P_SQ.waypoints, 
    P_SQ.avg_speed, 
    P_SQ.max_speed, 
    P_SQ.climb 
FROM 
    Tracks T 
LEFT OUTER JOIN 
(
    SELECT 
     P.trackid, 
     SUM(P.distToPrev) AS distance, 
     COUNT(P.latitude) AS waypoints, 
     AVG(P.speed) * 3.6 AS avg_speed, 
     MAX(P.speed) * 3.6 AS max_speed, 
     SUM(altitudeUp) AS climb 
    FROM 
     Positions P 
    WHERE 
     P.trackid = ? 
    GROUP BY 
     P.trackid 
) P_SQ ON P_SQ.trackid = T._id 
LEFT OUTER JOIN 
(
    SELECT 
     H.trackid, 
     AVG(heartbeat) AS avg_heartbeat, 
     MAX(heartbeat) AS max_heartbeat, 
     AVG(cadence) AS avg_cadence 
    FROM 
     Heartbeats 
    WHERE 
     H.trackid = ? 
    GROUP BY 
     H.trackid 
) H_SQ ON H_SQ.trackid = T._id 
WHERE 
    T._id = ? 

Answer 3

不要过于简单化，但我认为你可以通过查询过程简单的索引，你是扫描领域剃了大量的时间把你的查询。

通过索引字段，您可以减少所有表扫描，随着记录数量的增长，查询速度会变慢。

索引的缺点是空间需求，但根据我在Android上的经验，它很少是一个因素。