从一个表中选择所有记录并从其他表中匹配记录

Question

我工作了几个小时，只是为了得到我想要的SQL代码的结果，但没有运气。我正在尝试从一个表中选择所有记录并匹配其他表中的记录。

请参阅实时样本https://ucwa.skype.com/websdk。

url_id  short_url   campaign    Clicks
1       test1       campaign1   2
2       test2       campaign2   2

上面示例中的SQL代码：

SELECT u.url_id, u.short_url, c.campaign, Count(*) AS Clicks
FROM (urls u LEFT JOIN log l ON l.url_id = u.url_id) LEFT JOIN campaign c ON u.campaign_id = c.campaign_id
WHERE (((DATE_FORMAT(l.date_time, '%Y-%m-%d')) Between '2017-03-14' And '2017-03-17'))
GROUP BY u.url_id, u.short_url, c.campaign; 

我想要的结果就是这个。包括来自网址表的所有记录。

url_id  short_url   campaign    Clicks
1       test1       campaign1   2
2       test2       campaign2   2
3       test3       campaign1   0
4       test4                   0

我尝试在WHERE子句中添加l.url_id is null，但点击的值为1而不是零（0）

url_id  short_url   campaign    Clicks
1       test1       campaign1   2
2       test2       campaign2   2
3       test3       campaign1   1
4       test4                   1

上面示例中的SQL代码：

SELECT u.url_id, u.short_url, c.campaign, Count(*) AS Clicks
FROM (urls u LEFT JOIN log l ON l.url_id = u.url_id) LEFT JOIN campaign c ON u.campaign_id = c.campaign_id
WHERE (((DATE_FORMAT(l.date_time, '%Y-%m-%d')) Between '2017-03-14' And '2017-03-17')) or l.url_id is null
GROUP BY u.url_id, u.short_url, c.campaign; 

Answer 1

使用COUNT(l.id)代替COUNT(*)。

在l.date_time IS NULL子句中添加WHERE。

检查http://rextester.com/YOX73980

SELECT u.url_id, u.short_url, c.campaign, Count(l.id) AS Clicks
FROM urls u 
LEFT JOIN log l ON l.url_id = u.url_id 
LEFT JOIN campaign c ON u.campaign_id = c.campaign_id
WHERE l.date_time IS NULL OR (DATE_FORMAT(l.date_time, '%Y-%m-%d')) Between '2017-03-14' And '2017-03-17'
GROUP BY u.url_id, u.short_url, c.campaign; 

Answer 2

在log表上聚合可能在概念上更容易，因为这是生成点击次数的原因，而不是其他表。沿着这些方向，我们可以尝试以下查询：

SELECT t1.url_id,
       t1.short_url,
       COALESCE(t2.campaign, 'NA') AS campaign,
       COALESCE(t3.num_clicks, 0) AS Clicks
FROM urls t1
LEFT JOIN campaign t2
    ON t1.campaign_id = t2.campaign_id
LEFT JOIN
(
    SELECT url_id, COUNT(*) AS num_clicks
    FROM log
    WHERE DATE_FORMAT(date_time, '%Y-%m-%d') BETWEEN '2017-03-14' AND '2017-03-17'
    GROUP BY url_id
) t3
    ON t1.url_id = t3.url_id;

<强>输出：

在这里演示：

Rextester

Answer 3

SELECT u.url_id, u.short_url, c.campaign, Count(c.campaign_id) AS Clicks
FROM urls u LEFT JOIN log l ON l.url_id = u.url_id 
LEFT JOIN campaign c ON u.campaign_id = c.campaign_id 
AND (((DATE_FORMAT(l.date_time, '%Y-%m-%d')) Between '2017-03-14' And '2017-03-17')) 
GROUP BY u.url_id, u.short_url, c.campaign; 

检查Demo