Question

我有以下两个用于存储电影标题的SQL表，以及我见过它们的日期：

CREATE TABLE movies (
  id int(10) unsigned NOT NULL AUTO_INCREMENT,
  title varchar(90) NOT NULL,
  year varchar(7) DEFAULT NULL,
  watchAgain enum('0', '1') NOT NULL DEFAULT '0',
  PRIMARY KEY (id)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8 AUTO_INCREMENT=3 ;

INSERT INTO movies (id, title, year, watchAgain) 
  VALUES (1, 'Wall Street', '1987', '1'), (2, 'Wag the Dog', '1997', '1');

CREATE TABLE movies_seen (
  object_id int(10) NOT NULL DEFAULT '0',
  date varchar(10) CHARACTER SET utf8 COLLATE utf8_unicode_ci 
    NOT NULL DEFAULT '0',
  seen_again enum('0', '1') NOT NULL DEFAULT '0',
  PRIMARY KEY (object_id, date)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;

INSERT INTO `movies_seen` (object_id, date, seen_again)   
  VALUES (1, '1156187700', '0'), (1, '1218395700', '0'), (2, '0', '0');

要查找我应该再次观看的电影，我会在数据库中询问标题为seen_again == 1，或seen_again == 0是否存在movies_seen：

SELECT m.year, m.title, s.date AS last_seen
FROM movies m 
LEFT JOIN movies_seen s ON m.id = s.object_id
WHERE 
     (
     m.watchAgain = '0'
     AND m.id NOT IN (SELECT s.object_id 
                      FROM movies m 
                      INNER JOIN movies_seen s ON m.id = s.object_id)
    OR m.watchAgain = '1'
    )
ORDER BY m.title

这基本上有效，但是当movies_seen表对同一个标题有2个entires时，它会在结果中显示重复：

year    title           last_seen 
1997    Wag the Dog 0
1987    Wall Street 1156187700
1987    Wall Street 1218395700

如果我只获取标题，它适用于DISTINCT，但由于我需要额外的列，这显然会失败。想法？

Answer 1

在MySQL中，GROUP BY子句可以执行与DISTINCT

相同的任务

SELECT m.year
  , m.title
  , s.date AS last_seen 
FROM movies m  
LEFT JOIN movies_seen s ON m.id = s.object_id 
WHERE (m.watchAgain = '0'             
  AND m.id NOT IN (SELECT s.object_id 
                   FROM movies m 
                   INNER JOIN movies_seen s ON m.id = s.object_id)             
  OR m.watchAgain = '1') 
GROUP BY m.title /*<<--- this will do a distinct on title only*/
ORDER BY m.title

AFAIK，这只适用于MySQL。

顺便说一句：由于你在m.title上进行分组，MySQL对标题进行了排序，因此技术上不再需要ORDER BY子句。

Answer 2

SELECT m.year, m.title, s.date AS last_seen
FROM movies m 
WHERE exists  
    ( 
      select 1 from movies_seen s 
          where m.id = s.object_id 
           and ( m.watchAgain = '0'
            AND m.id NOT IN (SELECT s.object_id FROM movies m INNER JOIN movies_seen s ON m.id = s.object_id)
            OR m.watchAgain = '1'
            )
    )
ORDER BY m.title

将LEFT JOIN与子查询组合时的SQL DISTINCT

2 个答案: