Question

我编写了一个查询，其中包含一组子查询：

SELECT
    `board_id`,
    `post_count`,
    ROUND(`post_age_avg`, 3) as `post_age_avg`,
    ROUND(`post_rating_avg`, 3) as `post_rating_avg`,
    ROUND(`board_age`, 3) as `board_age`,
    ROUND(1 - (`post_age_avg` / `board_age`), 3) AS `board_usage`,
    ROUND((`post_count` / `board_age`) * `post_rating_avg`, 3) AS `board_rating`
FROM
    (SELECT
        `board_id`,
        (SELECT COUNT(*) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`) AS `post_count`,
        (SELECT AVG(TIME_TO_SEC(TIMEDIFF(NOW(), `post`.`created_on`)) / 3600) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`) as `post_age_avg`,
        (SELECT AVG(`rating`) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`) AS `post_rating_avg`,
        TIME_TO_SEC(TIMEDIFF(NOW(), `board`.`created_on`)) / 3600 AS `board_age`
    FROM `board`) AS `board_stats`

现在，虽然我对优化或改进此查询的建议持开放态度，但我的问题是以何种方式存储此查询以供将来使用。

视图会很棒，但是根据CREATE VIEW上的MySQL手册页：

SELECT语句不能在FROM子句中包含子查询。

所以我在程序中封装了查询。工作正常，致电：

CALL get_board_stats();

然而，我很快发现，使用过程作为子查询时，使用过程的灵活性有限（读零）。像其他有类似问题的SO一样，我发现：

SELECT * FROM (CALL get_board_stats()) AS `board_stats`;

其中的任何排列在语法上都是无效的。

所以我的问题是;我怎样才能实现（，如果可能的话）一个场景，在这个场景中可以存储这个查询，以便以后在后续查询中用作“虚拟表”，允许人们做类似的事情：

SELECT * FROM /* give_me_board_stats_somehow() */ WHERE ...

Alrighty @OMG Ponies，这是我要使用的最终版本（ for now ），因为它生成相同的结果集，具有相同的数值精度。它与你的非常相似，除了我省略了JOIN支持（再次）子查询，尽管这次不是派生表中的列。有些东西告诉我这样效率较低（比较JOIN解决方案）或许你可以对此有所了解：

SELECT
    `board`.`board_id`,
    (SELECT COUNT(*) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`) AS `post_count`,
    ROUND((SELECT AVG(TIME_TO_SEC(TIMEDIFF(NOW(), `post`.`created_on`)) / 3600) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`), 3) AS `post_age_avg`,
    ROUND((SELECT AVG(`rating`) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`), 3) AS `post_rating_avg`,
    ROUND(TIME_TO_SEC(TIMEDIFF(NOW(), `board`.`created_on`)) / 3600, 3) AS `board_age`,
    ROUND(1 - ((SELECT AVG(TIME_TO_SEC(TIMEDIFF(NOW(), `post`.`created_on`)) / 3600) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`) / (TIME_TO_SEC(TIMEDIFF(NOW(), `board`.`created_on`)) / 3600)), 3) AS `board_usage`,
    ROUND(((SELECT COUNT(*) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`) / (TIME_TO_SEC(TIMEDIFF(NOW(), `board`.`created_on`)) / 3600)) * (SELECT AVG(`rating`) FROM `post` WHERE `post`.`board_id` = `board`.`board_id`), 3) AS `board_rating`
FROM `board`

^{（我发布的格式比这个格式更好，但Workbench美化很糟糕，查询较多，我没有努力：P ）}

出于某种原因，您的解决方案继续为board_usage和board_rating提供不正确的结果。

Answer 1

您不需要派生表 - 使用：

CREATE VIEW your_view AS
   SELECT b.board_id
          COUNT(p.post_id) AS post_count,
          ROUND(AVG(TIME_TO_SEC(TIMEDIFF(NOW(), p.created_on)) / 3600), 3) AS post_age_avg,
          ROUND(AVG(p.rating), 3) AS post_rating_avg,
          ROUND(TIME_TO_SEC(TIMEDIFF(NOW(), b.created_on)) / 3600, 3) AS board_age,
          ROUND(1 - (AVG(TIME_TO_SEC(TIMEDIFF(NOW(), p.created_on)) / 3600) / TIME_TO_SEC(TIMEDIFF(NOW(), b.created_on)) / 3600), 3) AS board_usage,
          ROUND(COUNT(p.*) / (TIME_TO_SEC(TIMEDIFF(NOW(), b.created_on)) / 3600), 3) AS board_rating
     FROM BOARD b
LEFT JOIN POST p ON p.board_id = b.board_id
 GROUP BY b.board_id

......虽然使用了ROUND我不推荐，直到经过任何计算后你都需要这些值。

MySQL子查询，视图和过程;哪个（如果有的话）是正确的？

1 个答案: