如何在PostgreSQL中對某一列進行分組？

Question

我已在PostgreSQL中下表tableA：

+-------------+-------------------------+ 
| OperationId |   Error   | 
+-------------+-------------------------+ 
|   1 | MajorCategoryX:DetailsP | 
|   2 | MajorCategoryX:DetailsQ | 
|   3 | MajorCategoryY:DetailsR | 
+-------------+-------------------------+ 

如何組MajorErrorCategory這樣，我得到以下？

+----------------+------------+ 
| Category | ErrorCount | 
+----------------+------------+ 
| MajorCategoryX |   2 | 
| MajorCategoryY |   1 | 
+----------------+------------+ 

Category是Error分裂後的第一部分 '：'。

Answer 1

假設長度之前:可以改變，你可以結合使用substring與strpos實現結果：

SELECT 
    SUBSTRING(error, 0, STRPOS(error, ':')) AS Category,  
    COUNT(*) AS ErrorCount 
FROM t 
GROUP BY SUBSTRING(error, 0, STRPOS(error, ':')) 

Sample SQL Fiddle

如果你不想重複你可以的函數調用f課程將這部分包裝在一個suquery或common table表達式中。

Answer 2

考慮substring()功能：

SELECT substring(TableName.Error,1,14) AS Category, 
     Count(*) As ErrorCount 
FROM TableName 
GROUP BY substring(TableName.Error,1,14) 

Answer 3

這就是我想出了使用子查詢和split_part功能：

SELECT *, COUNT(ErrorSplit) 
FROM (
    SELECT split_part(Error, ':', 1) AS ErrorSplit 
    FROM tableA 
) AS tableSplit 
GROUP BY ErrorSplit; 

輸出：

errorsplit | count 
---------------------- 
MajorCategoryX | 2 
MajorCategoryY | 1 

SQL Fiddle

Answer 4

split_part()似乎最簡單的（as @ub3rst4r mentioned）：

• Cut string after first occurrence of a character

但你並不需要一個子查詢：

SELECT split_part(error, ':', 1) AS category, count(*) AS errorcount 
FROM tbl 
GROUP BY 1; 

而且count(*)比count(<expression>)稍快。

GROUP BY 1是對第一個SELECT項目的位置參考，並且是對較長表達式的方便的簡寫。例如：

• Select first row in each GROUP BY group?