如何使用别名联接到表

Question

Thanks in advance for any help.

I'm wondering how to join two SQL statements I have, based on the fact that I've given the columns aliases.

The source table looks like this (please bear in mind that changing the layout of this table is enormously expensive due to other requirements):

ColumnID    RowIdx    Val
1           0         "Mr A"
1           1         "Mr B"
1           2         "Mr C"
2           0         "M40 2TB"
2           1         "G23 XYN"
2           2         "HJ 23N"

I use two sql statements to create views on the table, which I think join. This looks something like this:

-- Statement #1
CREATE TEMP TABLE nameTable AS (SELECT 
    max(CASE WHEN ColumnID=1 THEN Val ELSE null END) AS Name, 
    row_number() over (ORDER BY RowIdx) AS rowindex 

FROM myTable 

WHERE (ColumnID=1) 

GROUP BY RowIdx
ORDER BY RowIdx);

-- Statement #2
CREATE TEMP TABLE postcodeTable AS (SELECT 
    max(CASE WHEN ColumnID=2 THEN Val ELSE null END) AS PostCode, 
    row_number() over (ORDER BY RowIdx) AS rowindex 

FROM myTable 

WHERE (ColumnID=2) 

GROUP BY RowIdx
ORDER BY RowIdx);

-- Statement #3
SELECT * FROM nameTable
INNER JOIN postcodeTable ON nameTable.rowIndex=postcodeTable.rowIndex

The result is the following:

Name    PostCode    RowIndex
Mr A    M40 2TB     0
Mr B    G23 XYN     1
Mr C    HJ 23N      2

I would like to combine these into a single statement (as it makes generating the statement simpler for my program.

I have chosen to omit some logic from statements 1 and 2 as it is needlessly complicated for my example, but what I have left out prohibits me doing this:

SELECT 
    max(CASE WHEN ColumnID=1 THEN Val ELSE null END) AS Name, 
    max(CASE WHEN ColumnID=2 THEN Val ELSE null END) AS PostCode, 
    row_number() over (ORDER BY RowIdx) AS rowindex 

FROM myTable 

WHERE (ColumnID=1 OR ColumnID=2) 

GROUP BY RowIdx
ORDER BY RowIdx

I'm now considering doing something like this, but I'm not sure if it's possible:

SELECT 
    max(CASE WHEN ColumnID=1 THEN Val ELSE null END) AS Name, 
    row_number() over (ORDER BY RowIdx) AS rowindex 

FROM myTable 

INNER JOIN (SELECT 
        max(CASE WHEN ColumnID=2 THEN Val ELSE null END) AS PostCode, 
        row_number() over (ORDER BY RowIdx) AS rowindex 

    FROM myTable 

    WHERE (ColumnID=2) 

    GROUP BY RowIdx
    ORDER BY RowIdx) AS postcodeTable ON rowIndex=postcodeTable.rowIndex

WHERE (ColumnID=1) 

GROUP BY RowIdx
ORDER BY RowIdx);

Is this something I should even be considering doing? Or should I just stick with temporary tables? If it is, how do I get this to work? The compile issue in on the rowIndex=postcodeTable.rowIndex rowIndex doesn't exist.

I know the original schema is formatted weirdly, but there are all sorts of external reasons for that. Please let me know if you need any more info.

Answer 1

I may be missing something, but I don't understand the need for the complexity in your example queries. Would something like the following (untested) query work?

SELECT Name, PostCode, A.RowIdx AS RowIdx 
FROM
    (SELECT Val AS Name, RowIdx FROM myTable WHERE ColumnID=1) A
    INNER JOIN
    (SELECT Val AS PostCode, RowIdx FROM myTable WHERE ColumnID=2) B
ON A.RowIdx = B.RowIdx;