需要使用 SQLite 查询进行一些奇怪的时间转换

Question

我有一个 SQLite 数据库，包含列

time

、

name

、

value

。

time

列存储以秒为单位的纪元，例如“1695765399”，并且每隔一秒记录一次。

将有一个预定的触发器在 NZST 12:05（午夜）+5 分钟运行此查询，以便为 SQLite 数据库提供一点空间来记录数据。对于每个唯一的

name

，我需要在

time

列中获取 5 分钟前的 12:00:00 NZST 时间（以秒为单位的纪元）（在 SQLite 数据库中，不会丢失任何秒）。并以 UTC '%Y-%m-%dT%H:%M:%SZ' 格式输出。

创建数据：

import sqlite3
import random
from datetime import datetime, timedelta

conn = sqlite3.connect('your_database.db')
c = conn.cursor()

# Create the table
c.execute('''
    CREATE TABLE your_table (
        name TEXT,
        value REAL,
        time INTEGER
    )
''')

# Start time at 11:55pm
start_time = datetime.now().replace(hour=23, minute=55, second=0)

# Insert data from 11:55pm to 12:10am
for i in range((10 + 5) * 60):  # 10 minutes before and 5 minutes after midnight
    current_time = start_time + timedelta(seconds=i)
    epoch_time = int(current_time.timestamp())
    for name in ['1', '2', '3']:
        value = random.random()
        c.execute('INSERT INTO your_table VALUES (?, ?, ?)', (name, value, epoch_time))

# Commit the changes and close the connection
conn.commit()
conn.close()

我期望的结果：

utc_time             name           value
------------------------------------------
xxxxx11:00:00         1              100
xxxxx11:00:00         2              200 

Answer 1

也许以下可以作为基础。

• 您的问题中有些内容似乎不相符，例如%H 是 24 小时制的小时，午夜实际是第二天的 00:00:00。

无论如何，也许以下内容可能是您想要的基础。

• 请注意，结果输出中已包含其他数据，希望有助于理解各个方面。

它自己的查询：-

WITH local_utc_modifier(lum) AS 
    /* CTE (Common Table Expression a temp table) with the difference between local and utc (not sure - is correct but principle IF device is setup accordingly)*/
    (SELECT (strftime('%s','now','localtime')) - (strftime('%s','now','UTC')))
SELECT
    name,
    (SELECT lum FROM local_utc_modifier) AS lum , /* the modifier value: include to demonstrate  <<<<<<<<<<EXTRA>>>>>>>>>>*/ 
    strftime('%Y-%m-%dT%H:%M:%SZ',time,'unixepoch','-'||(SELECT * FROM local_utc_modifier)||' seconds') /* output value */ AS time,
    value, /* NOTE ARBRITRAY */
    max(value) AS maxv, /* include to demonstrate arbrtrary value <<<<<<<<<<EXTRA>>>>>>>>>>*/
    min(value) AS minv, /* as prev <<<<<<<<<<EXTRA>>>>>>>>>>*/
    avg(value) AS avgv, /* as prev <<<<<<<<<<EXTRA>>>>>>>>>>*/
    count() AS number_found /* more than 1!!!! SEE ABOVE RESULTS <<<<<<<<<<EXTRA>>>>>>>>>> */
    
FROM your_table
/* Only midnight rows to a second */
WHERE strftime('%H:%M:%S',time,'unixepoch','-'||(SELECT * FROM local_utc_modifier)||' seconds') = '00:00:00'
    /* MIGHT WANT ADDITIONAL CONDITION JUST FOR CURRENT DATE */
/* Note if only 1 row per name then no need for GROUP BY HOWEVER MUST NOT THEN HAVE aggregate functions (count,max,min,avg) */
GROUP BY name
;

• 请参阅评论，因为查询的某些部分您可能不想要/不需要。

上述演示：-

DROP TABLE IF EXISTS your_table;
CREATE TABLE IF NOT EXISTS your_table (
        name TEXT,
        value REAL,
        time INTEGER
);

/* Insert some testing data */
INSERT INTO your_table (name,value,time) 
    VALUES
    (1,100,1695765399),(2,200,1695765405),(3,300,1695765410),(3,10,1695765415),(2,20,1695765420),
    /* these ones are at midnight*/
    (1,100,1695765420 + 180),(1,15,1695765600),(2,222,1695765600),(3,333,1695765600),
    /* 1 second past midnight */
    (5,555,1695765601)
    ;
SELECT name, value, datetime(time,'unixepoch') AS easyreadtime, time FROM your_table;
WITH local_utc_modifier(lum) AS 
    /* CTE (Common Table Expression a temp table) with the difference between local and utc (not sure - is correct but principle IF device is setup accordingly)*/
    (SELECT (strftime('%s','now','localtime')) - (strftime('%s','now','UTC')))
SELECT
    name,
    (SELECT lum FROM local_utc_modifier) AS lum , /* the modifier value: include to demonstrate  <<<<<<<<<<EXTRA>>>>>>>>>>*/ 
    strftime('%Y-%m-%dT%H:%M:%SZ',time,'unixepoch','-'||(SELECT * FROM local_utc_modifier)||' seconds') /* output value */ AS time,
    value, /* NOTE ARBRITRAY */
    max(value) AS maxv, /* include to demonstrate arbrtrary value <<<<<<<<<<EXTRA>>>>>>>>>>*/
    min(value) AS minv, /* as prev <<<<<<<<<<EXTRA>>>>>>>>>>*/
    avg(value) AS avgv, /* as prev <<<<<<<<<<EXTRA>>>>>>>>>>*/
    count() AS number_found /* more than 1!!!! SEE ABOVE RESULTS <<<<<<<<<<EXTRA>>>>>>>>>> */
    
FROM your_table
/* Only midnight rows to a second */
WHERE strftime('%H:%M:%S',time,'unixepoch','-'||(SELECT * FROM local_utc_modifier)||' seconds') = '00:00:00'
    /* MIGHT WANT ADDITIONAL CONDITION JUST FOR CURRENT DATE */
/* Note if only 1 row per name then no need for GROUP BY HOWEVER MUST NOT THEN HAVE aggregate functions (count,max,min,avg) */
GROUP BY name
;

DROP TABLE IF EXISTS your_table; /* CLEANUP DEMO ENVIRONMENT */

• 重要以上均使用AEST时间，因此测试数据需要考虑时区差异。 运行上述命令后，结果为：-

• 第一次加载的数据SELECT:-

- note the 2 highlighted name **1** rows both are at midnight (to demonstrate GROUP BY and aggregate functions)

• 结果，按照对你的问题的解释：-

• 第二列是额外的，只是为了根据设备向您显示本地和 utc 之间的时差（AEST 时间）。
• 最后四列演示了聚合函数分组。

聚合函数和 GROUP BY

对于名称 1，number_found（来自

count

函数）显示有 2 行与 WHERE 条件匹配（可能注意大小写）。 GROUP BY 子句定义了将被视为聚合函数所应用的组的行集，请注意，对于组，将输出 1 行。

省略 GROUP BY 子句，如果使用任何聚合函数，则所有输出都被视为 1 组。

如果选择输出一列并且它不受聚合表达式的影响，则提取的值将是组中某一行的值。它是一个任意值（例如，对于提取的值列

15

，它可能是

100

）（显示最大值、最小值和平均值）