LEFT加入PSQL，其中AND參數在另一個表上

Question

我有一個PSQL查詢，其中左側的AND參數加入到另一個表格下面的檢查表示例中。

發票

id | account |  invoice_date  | reference | total_amount | status 
-----+---------+-------------------------+-----------+--------------+-------- 
164 |  100 | 2016-08-03 03:05:08.996 |  161 |  2000.00 |  
165 |  100 | 2016-08-03 21:42:07.865 |  164 |   0 |  
167 |  100 | 2016-08-03 22:56:41.731 |  166 |  100.00 |  
168 |  100 | 1970-01-01 00:33:20  |  161 |   200 |  
169 |  100 | 2016-08-08 00:00:00  |  161 |   200 | 

Invoice_items

id | invoice | invoice_item_type | product | quantity | unit_price | reference | amount 
-----+---------+-------------------+---------+----------+------------+-----------+--------- 
143 |  164 |     1 |  6 |   |   |  161 | 2000.00 
144 |  165 |     1 |  11 |   |   |  164 |  0 
145 |  167 |     1 |  8 |   |   |  166 | 100.00 

也有另一種表的產品，但只有相關領域有ID

這是我的查詢

select products.id, sum(invoice_items.amount) as total_revenue 
from products 
    left join invoice_items on invoice_items.product = products.id 
    left join invoices on invoice_items.invoice = invoices.id 
         and invoices.invoice_date= current_date 
group by products.id; 

我需要查詢的是列出所有Product ID並在total_revenue列中放置產品的銷售總額（在'product'字段類似的invoice_items表中添加'amount'）當前日期（找到在INVOICES表上）。但是當我運行這個查詢時，它列出了產品的所有total_amounts。我錯過了什麼？

樣品輸出。 8，6和11必須爲空

id | total_revenue 
-----+--------------- 
125 |    
154 |    
119 |    
129 |    
    8 |  100.00 
112 |    
    5 |    
132 |    
104 |    
113 |    
143 |    
152 |    
121 |    
127 |    
165 |    
139 |    
146 |    
    15 |    
    2 |    
147 |    
149 |    
166 |    
169 |    
    13 |    
106 |    
122 |    
    9 |    
    11 |    0 
110 |    
120 |    
130 |    
155 |    
134 |    
136 |    
101 |    
168 |    
131 |    
157 |    
161 |    
103 |    
150 |    
159 |    
107 |    
108 |    
145 |    
    4 |    
    12 |    
158 |    
167 |    
138 |    
162 |    
100 |    
156 |    
163 |    
124 |    
123 |    
109 |    
153 |    
102 |    
105 |    
151 |    
116 |    
133 |    
140 |    
160 |    
148 |    
126 |    
141 |    
    7 |    
118 |    
    10 |    
164 |    
128 |    
    14 |    
144 |    
135 |    
    1 |    
    6 |  2000.00 
    3 |    
137 |    
117 |    
142 |    
111 |   

Answer 1

日期約束進行篩選在發票表中輸出記錄，而您也需要它過濾掉invoice_items表中的記錄 - 但它不這樣做，因爲兩者都是左連接。派生表將很容易解決這個問題，並給出你想要的結果。爲了簡潔和可讀性，我還添加了一些表別名。

像這樣：

SELECT 
    p.id, SUM(inv.amount) AS total_revenue 

FROM 
    products p LEFT JOIN 

    (SELECT 
     ii.product, i.invoice_date, ii.amount 
    FROM 
     invoice_items ii JOIN 
     invoices i ON 
      ii.invoice = i.id) inv ON 
      inv.product = p.id AND 
      inv.invoice_date= current_date 

GROUP BY p.id; 

Answer 2

您會得到大量的NULL值，這似乎是基於您的查詢預期的。

這看起來像是一個「INNER」連接類型的問題，而不是「左」連接。左連接將保留結果集中沒有產品發票的所有實例。

您還可以將整個查詢有關「DATE」到having子句（個人而言，我更喜歡這裏的/有子查詢，因爲我覺得邏輯更清晰）：

SELECT products.id, SUM(invoice_items.amount) AS total_revenue 
FROM products 
    INNER JOIN invoice_items ON invoice_items.product = products.id 
HAVING EXISTS (SELECT 1 FROM invoices WHERE invoice_items.invoice = invoices.id 
         AND invoices.invoice_date= current_date) 
GROUP BY products.id; 

Answer 3

我明白了。具有預期結果的較小樣本數據將會有所幫助。

這個問題有點複雜的解釋。 left join保留所有產品和發票，無論日期如何。 invoices的最終加入將帶來僅在當前日期匹配的發票。但是，由於您正在第二張表上進行彙總，因此每個匹配的行都會出現一次（即使當天沒有發票項目），它也會顯示在您的結果中。

解決辦法：使用case聲明，以確定是否有匹配的invoice項目表：

select p.id, 
     sum(ii.amount * (case when i.id is not null then 1 end)) as total_revenue 
from products p left join 
    invoice_items ii 
    on ii.product = p.id left join 
    invoices i 
    on ii.invoice = i.id and i.invoice_date = current_date 
group by p.id; 

我還懷疑的日期權條件爲：

 on ii.invoice = i.id and i.invoice_date >= current_date and 
     i.invoice_date < current_date + interval '1 day' 

此外，將其作爲子查詢寫入可節省外部聚合，並應解決問題：

select p.*, 
     (select sum(ii.amount) 
     from invoice_items ii join 
      invoices i 
      on ii.invoice = i.id and i.invoice_date >= current_date and 
       i.invoice_date < current_date + interval '1 day' 
     where ii.product = p.id 
     ) as total_revenue 
from products p;