使用时间戳计算总持续时间

Question

在我的 SQLite 表中，我有以下字段（timestamp、zone_id、object_count 和 object_ids），这些是人们在区域中排队的数据。每当有人加入队列时，他都会被传感器捕获并分配一个object_id。 timestamp 是传感器服务器每秒向数据库发送的数据。

我需要找到队列中每个 object_ids 的持续时间，以使用烧瓶框架计算和显示我的 Web 应用程序中的等待时间。比如object_ids（2166）在2020-10-19 17:03:46.000000进入队列，在2020-10-19 17:03:50.000000离开，他的排队时间是5秒。简单的逻辑是从object_id 最后出现的时间戳中减去object_id 第一次出现时的时间戳。

#       timestamp           zone_id   object_count      object_ids #  

2020-10-19 17:03:43.000000    10           2         ['2140', '2143']
2020-10-19 17:03:44.000000    10           2         ['2140', '2143']
2020-10-19 17:03:45.000000    10           2         ['2140', '2143']
2020-10-19 17:03:46.000000    10           3         ['2140', '2143', '2166']
2020-10-19 17:03:47.000000    10           3         ['2140', '2143', '2166']
2020-10-19 17:03:49.000000    10           3         ['2140', '2143', '2166']
2020-10-19 17:03:50.000000    10           3         ['2140', '2143', '2166']
2020-10-19 17:03:51.000000    10           2         ['2140', '2143']
2020-10-19 17:03:53.000000    10           2         ['2140', '2143']

如何查询和解决？

Answer 1

假设您的表名是 queues 并且定义了 object_ids 的表是 objects（将名称更改为实际名称），所以您有这样的内容：

CREATE table objects (`object_id` TEXT);
INSERT INTO objects (`object_id`) VALUES
  ('2140'), ('2143'), ('2166');

CREATE TABLE queues (`timestamp` TEXT, `zone_id` INTEGER, `object_count` INTEGER, `object_ids` VARCHAR(24));
INSERT INTO queues (`timestamp`, `zone_id`, `object_count`, `object_ids`) VALUES
  ('2020-10-19 17:03:43.000000', '10', '2', '[''2140'', ''2143'']'),
  ('2020-10-19 17:03:44.000000', '10', '2', '[''2140'', ''2143'']'),
  ('2020-10-19 17:03:45.000000', '10', '2', '[''2140'', ''2143'']'),
  ('2020-10-19 17:03:46.000000', '10', '3', '[''2140'', ''2143'', ''2166'']'),
  ('2020-10-19 17:03:47.000000', '10', '3', '[''2140'', ''2143'', ''2166'']'),
  ('2020-10-19 17:03:49.000000', '10', '3', '[''2140'', ''2143'', ''2166'']'),
  ('2020-10-19 17:03:50.000000', '10', '3', '[''2140'', ''2143'', ''2166'']'),
  ('2020-10-19 17:03:51.000000', '10', '2', '[''2140'', ''2143'']'),
  ('2020-10-19 17:03:53.000000', '10', '2', '[''2140'', ''2143'']');

通过使用窗口函数MIN() 和MAX() 以及函数strftime()，您可以使用连接两个表的SQLite 查询获得所需的结果：

SELECT DISTINCT o.object_id,
       strftime('%s', MAX(q.timestamp) OVER (PARTITION BY o.object_id)) - 
       strftime('%s', MIN(q.timestamp) OVER (PARTITION BY o.object_id)) diff
FROM objects o INNER JOIN queues q
ON ',' || REPLACE(REPLACE(REPLACE(REPLACE(q.object_ids, '''', ''), '[', ''), ']', ''), ' ', '') || ','
LIKE '%,' || o.object_id || ',%'

如果您的 SQLite 版本不支持窗口函数，请使用 GROUP BY:

SELECT o.object_id,
       strftime('%s', MAX(q.timestamp)) - strftime('%s', MIN(q.timestamp)) diff
FROM objects o INNER JOIN queues q
ON ',' || REPLACE(REPLACE(REPLACE(REPLACE(q.object_ids, '''', ''), '[', ''), ']', ''), ' ', '') || ','
LIKE '%,' || o.object_id || ',%'
GROUP BY o.object_id

请参阅demo。
结果：

> object_id | diff
> :-------- | ---:
> 2140      |   10
> 2143      |   10
> 2166      |    4

Answer 2

因此，在您的 application.py 中，您应该有此代码以获取正确的数据库，这应该位于文件中的某个位置，以便您的所有函数都可以访问它。

engine = create_engine(os.getenv("DATABASE_URL"))
db = scoped_session(sessionmaker(bind=engine))

然后在你希望有时间拥有的功能中的某个地方，你应该有这样的东西

timestamp = db.execute(text("SELECT timestamp FROM NAME_OF_YOUR_TABLE").fetchall()#this will get you all of the timestamps

现在您在 python 中获得了表格中的时间戳，因此您可以轻松地做任何您想做的事情，您可以轻松地将其转换为 datetime 格式，然后只需使用 datetime.datetime.now() 函数并减去它来自彼此，如果您还有其他需要，请随时在下面发表评论。

编辑：所以为了遍历 IDs 列，你需要这样的东西：

T1 = db.execute(text("SELECT object_ids FROM NAME_OF_YOUR_TABLE")).fetchall()
for i in T1:
   
   timestamp = db.execute(text("SELECT timestamps FROM NAME_OF_YOUR_TABLE WHERE object_ids=:object_ids"), {'object_ids':i}).fetchone()
   print('Timestamp = ' + timestamp 'and id = ' + i)