Question

更新：我正在尝试将mysql代码翻译成postgres代码，我想在where子句中的两个日期之间进行过滤

原始Mysql

SELECT concat(month(DD1.open_date),'/', DATE_FORMAT(DD1.open_date, '%y')) AS month_year,
           venues.name AS venue_name,
           DD1.`server_name` AS SERVER_NAME,
           DD1.`server_id` AS `Server ID`,
           AVG(lr.`rating`) AS server_overall_rating,
           all_overall_rating,
           AVG(lr.`service`) AS server_service_rating,
           all_service_rating,
           avg(spend) AS server_check_avg,
           all_check_avg,
           avg(cover_avg) AS server_cover_avg,
           all_cover_avg
    FROM loyalty_visits DD1
    JOIN
      (SELECT avg(lr.rating) AS all_overall_rating,
              avg(lr.service) AS all_service_rating,
              avg(cover_avg) AS all_cover_avg,
              avg(spend) AS all_check_avg,
              open_date
       FROM loyalty_visits lv
       LEFT OUTER JOIN loyalty_ratings lr ON lr.loyalty_visit_id = lv.id
       WHERE lv.venue_id = 4046
       AND open_date BETWEEN date_sub(now(), interval 12 MONTH) AND now()
       GROUP BY lv.venue_ID,
                MONTH(open_date), year(open_date)) AS overall_cover_avg ON MONTH(overall_cover_avg.open_date) = MONTH(DD1.open_date)
    AND year(overall_cover_avg.open_date) = year(DD1.open_date)
    LEFT OUTER JOIN loyalty_member_visits lmv ON lmv.loyalty_visit_id = DD1.id
    LEFT OUTER JOIN loyalty_ratings lr ON lmv.loyalty_visit_id = lr.loyalty_visit_id
    JOIN venues ON venues.id = DD1.venue_id
    WHERE DD1.parent_venue_id = 4041
      AND DD1.server_id = 6991
      AND DD1.venue_id = 4046
      AND DD1.open_date BETWEEN date('2015-01-29 14:15:55') AND date('2013-09-29 10:15:55')
    GROUP BY DD1.venue_ID,
             MONTH(DD1.open_date), year(DD1.open_date)
    ORDER BY DD1.open_date LIMIT 6

的Postgres 注意：我知道我的投影与原始的mysql相比并不是全部，因为我实际上是在尝试理解subquerying（我不知道它是如何工作的）并且还试图了解一些select语句的来源来自mysql版本，所以我可以充分转换为Postgres。

同时

忠诚度调用因为Mysql中的DD1被改为member_visits，因为在Postgres中的mv和loyalty_ratings，因为lr被改为在Postgres中仅被评为r。其余的应该是一样的。

SELECT mv.check_number,
mv.open_date,
mv.code,
mv.cover_avg,
reservation_loyalty_user_id,
v.restaurant_id,
table_name,
terminal_id
FROM member_visits as mv
FROM member_visits as mv
JOIN
      (SELECT avg(lr.rating) AS all_overall_rating,
              avg(lr.service) AS all_service_rating,
              avg(cover_avg) AS all_cover_avg,
              avg(spend) AS all_check_avg,
              open_date
       FROM loyalty_visits lv
       LEFT OUTER JOIN ratings lr ON lr.loyalty_visit_id = lv.id
       WHERE lv.venue_id = 4046
         AND open_date BETWEEN date_sub(now(), interval 3 MONTH) AND now()
       GROUP BY lv.venue_ID,
                   MONTH(open_date), year(open_date)) AS overall_cover_avg  ON     MONTH(overall_cover_avg.open_date) = MONTH(DD1.open_date)
    AND year(overall_cover_avg.open_date) = year(DD1.open_date)
LEFT JOIN ratings r 
    ON mv.loyalty_visit_id = r.loyalty_visit_id
JOIN venues v
    ON v.id = mv.venue_id
WHERE mv.parent_venue_id = 4041
AND mv.server_id = 6991
AND mv.venue_id = 4046
AND mv.open_date BETWEEN date_trunc('2015/01/20', day) AND date_trunc('2013/09/29', day)
GROUP BY mv.venue_ID,
            MONTH(mv.open_date), YEAR(mv.open_date)
ORDER BY mv.open_date
limit 10

我遇到两个错误：

1）在这一行

AND open_date BETWEEN date_sub(now(), interval 12 MONTH) AND now()，

从现在到12个月之间是否有可能没有数据？我不确定如何调试它。

2）我也得到一个错误 invalid input syntax for integer: "2015-01-2013 14:15:55"

我看过这篇文章Filter by Dates in SQL，但我不确定我喜欢这个解决方案，而且它是一个SQL解决方案。然后我尝试使用date_trunc，但我不确定这是否正确。

我应该使用的任何东西？

Answer 1

Postgres中的

There is no date_sub()函数。只需从时间戳（或日期）中减去间隔。您的间隔文字也是错误的。

SQL标准要求将值放在单引号中，例如interval '12' month或interval '5' minute。 Postgres还允许使用语法，其中值和单位采用单引号interval '1 month'

AND open_date BETWEEN now() - interval '12' MONTH AND now()

您的date_trunc使用情况is also wrong。字符串文字'2015/01/20'只会转换为当前日期格式与此匹配的日期。一般来说，使用ANSI日期文字来确保您的SQL在每个环境中都运行，无论日期格式设置如何都更好

date_trunc('day', DATE '2015-01-20')

Postgres也（总是）接受使用ISO格式的字符串文字，因此'2015-01-20'的工作方式与Postgres中DATE '2015-01-20'的工作方式相同

Postgres中也没有date()功能。我建议使用ANSI日期和时间戳文字。因此，date('2015-01-29 14:15:55')使用符合SQL标准的timestamp '2015-01-29 14:15:55'而不是to_date()。如果您不想使用ISO日期格式，则应使用to_timestamp()或date_format functions with an appropriate format mask。

Postgres中也没有to_char()功能。使用year()。再次is documented in the manual

Postgres中也没有month()或extract(year from open_date)功能。使用SQL标准：extract(month from open_date)或date_part()或Postgres特定的{{1}}函数。这也是documented in the manual

如何按postgres中的日期过滤 - 从mysql转换为postgres

1 个答案: