SQL 包含多个连接的属性值，即使缺少属性类型

Question

此问题基于“SQL for ignoring rows that have a particular property from a joined table”。我得到了有用的答案和 cmets，但我决定与我最初的请求不同。

我有两张桌子：drinks 和 properties。他们可以通过drink_id 加入。属性有几种可能的类型。我想创建一个报告饮料的查询，每个属性都有一个列，即使它为空。如果同一种饮料有多次出现相同的属性类型，我想我会想要所有笛卡尔组合，但那不是

Oracle 11，如果这有所作为。

+----------+--------------+-------------+
| drink_id |  drink_name  | drink_brand |
+----------+--------------+-------------+
|        1 | orange juice | tropicana   |
|        2 | seltzer      | schweppes   |
|        3 | cola         | pepsi       |
|        4 | diet cola    | pepsi       |
+----------+--------------+-------------+

+----------+-----------+-----------+
| drink_id | prop_type | prop_val  |
+----------+-----------+-----------+
|        1 | color     | orange    |
|        2 | color     | clear     |
|        3 | color     | brown     |
|        4 | sweetener | aspartame |
+----------+-----------+-----------+

期望的输出：

+--------------+-------------+-------------+-----------+
|  drink_name  | drink_brand | drink_color | sweetener |
+--------------+-------------+-------------+-----------+
| orange juice | tropicana   | orange      | <null>    |
| seltzer      | schweppes   | clear       | <null>    |
| cola         | pepsi       | brown       | <null>    |
| diet cola    | pepsi       | <null>      | aspartame |
+--------------+-------------+-------------+-----------+

我正在考虑这样的事情，但是如果给定饮料的一个或两个属性都不可用，我不知道如何包含行。

select drink_name, drink_brand, colorprop.prop_val as drink_color 
from drinks
left join properties colorprop
on drinks.drink_id = colorprop.drink_id
where colorprop.prop_type = 'color'
left join properties sweetprop
on drinks.drink_id = sweetprop.drink_id
where sweetprop.prop_type = 'sweetener'

更新： 如评论和答案中所示，我的问题主要是仓促的语法错误。我可以使用以下 R 代码对此进行测试：

library(sqldf)
drinks <- data.frame(drink_id =  c(1,2,3,4), 
  drink_name =  c("orangejuice", "seltzer", "cola", "dietcola"), 
  drink_brand = c("tropicana", "schweppes", "pepsi", "pepsi"))
names(drinks) <- c("drink_id", "drink_name", "drink_brand")

properties <- data.frame(drink_id =  c(1,2,3,4), 
  prop_type = c("color", "color", "color", "sweetener"), 
  prop_val = c("orange", "clear", "brown", "aspartame"))
names(properties) <- c("drink_id", "prop_type", "prop_val")

drinkquery <- 
"select drink_name, drink_brand, 
  colorprop.prop_val as drink_color, sweetprop.prop_val as sweetener
  from drinks
  left join properties colorprop
  on drinks.drink_id = colorprop.drink_id AND colorprop.prop_type = 'color'
  left join properties sweetprop
  on drinks.drink_id = sweetprop.drink_id AND sweetprop.prop_type = 'sweetener'"

sqldf(drinkquery)

Answer 1

我从来没有像你那样在JOINs 上看到过WHERE 子句，但你已经很接近了。我认为，将prop_type 作为JOIN 条件的一部分应该可以满足您的需求。

select drink_name, drink_brand, colorprop.prop_val as drink_color,
sweetprop.prop_val as sweetener
from drinks
left join properties colorprop
on drinks.drink_id = colorprop.drink_id AND colorprop.prop_type = 'color'
left join properties sweetprop
on drinks.drink_id = sweetprop.drink_id AND sweetprop.prop_type = 'sweetener'