对日期和时间的数据进行加减

PostgreSQL 中的时间序列分析

Jasmin Ludolf

Content Developer, DataCamp

时间差

AGE()：用第一个参数减去第二个参数，或与当前日期比较
AGE(timestamp, timestamp) 或 AGE(timestamp)

SELECT
AGE('2018-01-02','2017-01-01') AS "2017",
AGE('2021-01-05','2020-01-01') AS "2020";

|2017        |2020         |
|------------|-------------|
|1 year 1 day|1 year 4 days|

误解差异

SELECT 
AGE('2017-12-31','2017-01-01') AS "2017",
AGE('2018-12-31','2018-01-01') AS "2018",
AGE('2019-12-31','2019-01-01') AS "2019",
AGE('2020-12-31','2020-01-01') AS "2020";

|2017           |2018           |2019           |2020           |
|---------------|---------------|---------------|---------------|
|11 mons 30 days|11 mons 30 days|11 mons 30 days|11 mons 30 days|

减号运算符

SELECT
'2018-01-01'::DATE - '2017-01-01'::DATE AS "2017",
'2019-01-01'::DATE - '2018-01-01'::DATE AS "2018",
'2020-01-01'::DATE - '2019-01-01'::DATE AS "2019",
'2021-01-01'::DATE - '2020-01-01'::DATE AS "2020";

|2017|2018|2019|2020|
|----|----|----|----|
| 365| 365| 365| 366|

使用减号运算符

-：减号运算符
返回 INTERVAL 数据类型
INTERVAL 用于存储和操作一段时间

SELECT
    '2021-01-01 00:03:00'::TIMESTAMP - '2021-01-01 00:01:30'::TIMESTAMP 
    AS interval;

|interval|
|--------|
|00:01:30|

时间间隔

WITH line_324 AS (
    SELECT station, arrival_time 
    FROM train_schedule 
    WHERE train_id=324 )
SELECT hillsdale.arrival_time - millbrae.arrival_time AS diff
FROM line_324 AS millbrae, line_324 AS hillsdale
WHERE millbrae.station='Millbrae' 
AND hillsdale.station='Hillsdale';

CTE：公共表表达式，为单个查询定义临时表

¹ https://www.caltrain.com/schedules/weekdaytimetable.html

时间间隔

WITH line_324 AS (
    SELECT station, arrival_time 
    FROM train_schedule 
    WHERE train_id=324 )
SELECT hillsdale.arrival_time - millbrae.arrival_time AS diff
FROM line_324 AS millbrae, line_324 AS hillsdale
WHERE millbrae.station='Millbrae' 
AND hillsdale.station='Hillsdale';

|diff    |
|--------|
|00:08:00|

¹ https://www.caltrain.com/schedules/weekdaytimetable.html

减去一个区间

SELECT 
    '2020-02-01'::DATE - INTERVAL '1 month' AS "1 month sooner";

|1 month sooner     |
|-------------------|
|2020-01-01 00:00:00|

将区间转换为指定时间单位

EXTRACT(epoch FROM start_time - end_time)

SELECT EXTRACT(epoch FROM
    ('2021-01-01 00:03:00'::TIMESTAMP - '2021-01-01 00:01:30'::TIMESTAMP))

    AS seconds;

|seconds|
|-------|
|   90.0|

将区间转换为指定时间单位

SELECT (EXTRACT(epoch FROM
    ('2021-01-01 00:01:00'::TIMESTAMP - '2020-12-31 23:59:30'::TIMESTAMP)))
    / 60 AS minutes;

|minutes|
|-------|
|    1.5|

时间相加

SELECT '2019-02-01'::DATE + INTERVAL '28 days' AS "28 days later";

|28 days later      |
|-------------------|
|2019-03-01 00:00:00|

SELECT '2020-02-01'::DATE + INTERVAL '28 days' AS "28 days later";

|28 days later      |
|-------------------|
|2020-02-29 00:00:00|

给日期加一个月

SELECT '2019-02-01'::DATE + INTERVAL '1 month' AS "1 month later";

|1 month later      |
|-------------------|
|2019-03-01 00:00:00|

SELECT '2020-02-01'::DATE + INTERVAL '1 month' AS "1 month later";

|1 month later      |
|-------------------|
|2020-03-01 00:00:00|

Passons à la pratique !

PostgreSQL 中的时间序列分析