Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Unioning two tables with different number of columns

Tags:

sql

mysql

People also ask

Can you UNION tables with different number of columns?

The columns of joining tables may be different in JOIN but in UNION the number of columns and order of columns of all queries must be same.

Can you join two tables with different columns?

Merging tables by columns. Multiple tables can be merged by columns in SQL using joins. Joins merge two tables based on the specified columns (generally, the primary key of one table and a foreign key of the other).

Does UNION require same number of columns?

JOIN operations do NOT require the same number of columns be selected in both tables. UNION operations are different that joins. Think of it as two separate lists of data that can be "pasted" together in one big block. You can't have columns that don't match.


Add extra columns as null for the table having less columns like

Select Col1, Col2, Col3, Col4, Col5 from Table1
Union
Select Col1, Col2, Col3, Null as Col4, Null as Col5 from Table2

I came here and followed above answer. But mismatch in the Order of data type caused an error. The below description from another answer will come handy.

Are the results above the same as the sequence of columns in your table? because oracle is strict in column orders. this example below produces an error:

create table test1_1790 (
col_a varchar2(30),
col_b number,
col_c date);

create table test2_1790 (
col_a varchar2(30),
col_c date,
col_b number);

select * from test1_1790
union all
select * from test2_1790;

ORA-01790: expression must have same datatype as corresponding expression

As you see the root cause of the error is in the mismatching column ordering that is implied by the use of * as column list specifier. This type of errors can be easily avoided by entering the column list explicitly:

select col_a, col_b, col_c from test1_1790 union all select col_a, col_b, col_c from test2_1790; A more frequent scenario for this error is when you inadvertently swap (or shift) two or more columns in the SELECT list:

select col_a, col_b, col_c from test1_1790
union all
select col_a, col_c, col_b from test2_1790;

OR if the above does not solve your problem, how about creating an ALIAS in the columns like this: (the query is not the same as yours but the point here is how to add alias in the column.)

SELECT id_table_a, 
       desc_table_a, 
       table_b.id_user as iUserID, 
       table_c.field as iField
UNION
SELECT id_table_a, 
       desc_table_a, 
       table_c.id_user as iUserID, 
       table_c.field as iField

for any extra column if there is no mapping then map it to null like the following SQL query

Select Col1, Col2, Col3, Col4, Col5 from Table1
Union
Select Col1, Col2, Col3, Null as Col4, Null as Col5 from Table2````

Normally you need to have the same number of columns when you're using set based operators so Kangkan's answer is correct.

SAS SQL has specific operator to handle that scenario:

SAS(R) 9.3 SQL Procedure User's Guide

CORRESPONDING (CORR) Keyword

The CORRESPONDING keyword is used only when a set operator is specified. CORR causes PROC SQL to match the columns in table expressions by name and not by ordinal position. Columns that do not match by name are excluded from the result table, except for the OUTER UNION operator.

SELECT * FROM tabA
OUTER UNION CORR
SELECT * FROM tabB;

For:

+---+---+
| a | b |
+---+---+
| 1 | X |
| 2 | Y |
+---+---+

OUTER UNION CORR

+---+---+
| b | d |
+---+---+
| U | 1 |
+---+---+

<=>

+----+----+---+
| a  | b  | d |
+----+----+---+
|  1 | X  |   |
|  2 | Y  |   |
|    | U  | 1 |
+----+----+---+

U-SQL supports similar concept:

OUTER UNION BY NAME ON (*)

OUTER

requires the BY NAME clause and the ON list. As opposed to the other set expressions, the output schema of the OUTER UNION includes both the matching columns and the non-matching columns from both sides. This creates a situation where each row coming from one of the sides has "missing columns" that are present only on the other side. For such columns, default values are supplied for the "missing cells". The default values are null for nullable types and the .Net default value for the non-nullable types (e.g., 0 for int).

BY NAME

is required when used with OUTER. The clause indicates that the union is matching up values not based on position but by name of the columns. If the BY NAME clause is not specified, the matching is done positionally.

If the ON clause includes the “*” symbol (it may be specified as the last or the only member of the list), then extra name matches beyond those in the ON clause are allowed, and the result’s columns include all matching columns in the order they are present in the left argument.

And code:

@result =    
    SELECT * FROM @left
    OUTER UNION BY NAME ON (*) 
    SELECT * FROM @right;

EDIT:

The concept of outer union is supported by KQL:

kind:

inner - The result has the subset of columns that are common to all of the input tables.

outer - The result has all the columns that occur in any of the inputs. Cells that were not defined by an input row are set to null.

Example:

let t1 = datatable(col1:long, col2:string)  
[1, "a",  
2, "b",
3, "c"];
let t2 = datatable(col3:long)
[1,3];
t1 | union kind=outer t2;

Output:

+------+------+------+
| col1 | col2 | col3 |
+------+------+------+
|    1 | a    |      |
|    2 | b    |      |
|    3 | c    |      |
|      |      |    1 |
|      |      |    3 |
+------+------+------+

demo