SQL Server - Is it possible to define a table column as a table? - sql-server

I know that this is possible in Oracle and I wonder if SQL Server also supports it (searched for answer without success).
It would greatly simplify my life in the current project if I could define a column of a table to be a table itself, something like:
Table A:
Column_1 Column_2
+----------+----------------------------------------+
| 1 | Columns_2_1 Column_2_2 |
| | +-------------+------------------+ |
| | | 'A' | 12345 | |
| | +-------------+------------------+ |
| | | 'B' | 777777 | |
| | +-------------+------------------+ |
| | | 'C' | 888888 | |
| | +-------------+------------------+ |
+----------+----------------------------------------+
| 2 | Columns_2_1 Column_2_2 |
| | +-------------+------------------+ |
| | | 'X' | 555555 | |
| | +-------------+------------------+ |
| | | 'Y' | 666666 | |
| | +-------------+------------------+ |
| | | 'Z' | 000001 | |
| | +-------------+------------------+ |
+----------+----------------------------------------+
Thanks in advance.

There is one option where you can store data as XML
Declare #YourTable table (ID int,XMLData xml)
Insert Into #YourTable values
(1,'<root><ID>1</ID><Active>1</Active><First_Name>John</First_Name><Last_Name>Smith</Last_Name><EMail>john.smith#email.com</EMail></root>')
,(2,'<root><ID>2</ID><Active>0</Active><First_Name>Jane</First_Name><Last_Name>Doe</Last_Name><EMail>jane.doe#email.com</EMail></root>')
Select ID
,Last_Name = XMLData.value('(root/Last_Name)[1]' ,'nvarchar(50)')
,First_Name = XMLData.value('(root/First_Name)[1]' ,'nvarchar(50)')
From #YourTable
Returns
ID Last_Name First_Name
1 Smith John
2 Doe Jane

Actually, for a normalized database we do not require such functionality.
Because if we need to insert a table within a column than we can create a child table and reference it as a foreign key in the parent table.
In spite, if you still insist to such functionality than you can use SQL Server 2016 to support JSON data where you can store any associative list in JSON format.
Like:
DECLARE #json NVARCHAR(4000)
SET #json =
N'{
"info":{
"type":1,
"address":{
"town":"Bristol",
"county":"Avon",
"country":"England"
},
"tags":["Sport", "Water polo"]
},
"type":"Basic"
}'
SELECT
JSON_VALUE(#json, '$.type') as type,
JSON_VALUE(#json, '$.info.address.town') as town,
JSON_QUERY(#json, '$.info.tags') as tags
SELECT value
FROM OPENJSON(#json, '$.info.tags')
In older versions, this can be achieved through xml as shown in previous answer.
Your can also make use of "sql_variant" datatype to map your table.
Previously, I was also in search of such features as available in Oracle. But after reading various articles and blogs from experts, I was convinced, such features will make the things more complex beside helping.
Only storing the data in required format is not important, It is worthy when it is also efficiently available (readable).
Hope this will help you to take your decision.

Related

Mariadb Parititioning

There is table named history in Zabbix database, I have created partitions on this table.
And the partition type is range and column type is UNIX_TYPESTAMP.
After the date is changed zabbix service does not insert data to the related partition.
What is the problem?
And how do I display all partitions?
Could you please help how do I write data to the related partitions?
Sample of Partition creation statement;
.
.
.
ALTER TABLE zabbix.history_test PARTITION BY RANGE(clock)(PARTITION
p28082021 VALUES LESS THAN(UNIX_TIMESTAMP("2021-08-28 00:00:00"
))ENGINE=InnoDB);
Server version: 10.1.31-MariaDB MariaDB Server
EXPLAIN PARTITIONS SELECT * FROM zabbix.history;
+------+-------------+---------+------------+------+---------------+------
| id | select_type | table | partitions | type | possible_keys | key |
key_len | ref | rows | Extra |
| 1 | SIMPLE | history | p28082021 | ALL | NULL | NULL
| NULL | NULL | 18956757 | |
SELECT DISTINCT PARTITION_EXPRESSION FROM
INFORMATION_SCHEMA.PARTITIONS WHERE TABLE_NAME='history' AND
TABLE_SCHEMA='zabbix';
+----------------------+
| PARTITION_EXPRESSION |
+----------------------+
| clock |
+----------------------+
MariaDB [(none)]> SELECT PARTITION_ORDINAL_POSITION, TABLE_ROWS, PARTITION_METHOD
FROM information_schema.PARTITIONS
WHERE TABLE_SCHEMA = 'zabbix' AND TABLE_NAME = 'history';
+----------------------------+------------+------------------+
| PARTITION_ORDINAL_POSITION | TABLE_ROWS | PARTITION_METHOD |
+----------------------------+------------+------------------+
| 1 | 18851132 | RANGE |
+----------------------------+------------+------------------+
SELECT FROM_UNIXTIME(MAX(clock)) FROM zabbix.history;
+---------------------------+
| FROM_UNIXTIME(MAX(clock)) |
+---------------------------+
| 2018-04-07 23:06:06 |
+---------------------------+
SELECT FROM_UNIXTIME(MIN(clock)) FROM zabbix.history;
+---------------------------+
| FROM_UNIXTIME(MIN(clock)) |
+---------------------------+
| 2018-04-06 01:06:23 |
+---------------------------+
This document help me to create partition on clock column.
There are stored procedures, that create partitions,you can check it.
https://www.zabbix.org/wiki/Docs/howto/mysql_partition

How to mark selected columns of a table for display

I use PostgreSQL 10.1 and:
CREATE TABLE human
(
id ... NOT NULL,
gender ...,
height ...,
weight ...,
eye ...,
hair ...,
...
);
I have an input form through which I insert the data. I wish an elegant and proper way by which I can SELECT which columns required to be DISPLAYED in that form, something like weight ... DISPLAYED, or eye ... NOT DISPLAYED, .
One way is to correspond NULL with DISPLAYED (when NOT NULL then display it, or when NULL then do not display it) and use information_schema which (corresponding) makes me no so happy:
Another way is to:
CREATE TABLE human_column
(
id ... NOT NULL,
characteristic character varying(...),
is_displayed boolean
);
where characteristic data are the names of the columns of human table.
Is there a better way to add a direct foreign attribute to the columns of a table? (In 51.7. pg_attribute there is a column named attoptions. Would it be used?)
specifying "options" for columns to define if they will be "displayed" or not seems a little overhead. Imagine you keep such list in human_column. To modify it you would need to update it with new is_displayed values. Then you would need to build column list to be selected in query.
When you create a view, you do the same (specify a list of columns to be displayed) and then you can just query the view, without need to dynamically build the query. Also you can always check the current list of included columns from catalog or information_schema.
The only "not cosy" feature of a view - you can't change columns in it, thus you have to drop and create it again.
drop/create view on demand looks cheaper to me then dynamically building query with list of columns on each select still.
demo:
db=# create view v as select oid,datname from pg_database;
CREATE VIEW
db=# select * from v;
oid | datname
-------+-----------
13505 | postgres
16384 | t
1 | template1
13504 | template0
16419 | o
(5 rows)
checking list of columns:
db=# select column_name,ordinal_position,column_default,is_nullable,data_type,character_maximum_length from information_schema.columns where table_name = 'v';
column_name | ordinal_position | column_default | is_nullable | data_type | character_maximum_length
-------------+------------------+----------------+-------------+-----------+--------------------------
oid | 1 | | YES | oid |
datname | 2 | | YES | name |
(2 rows)
same for original table:
db=# select column_name,ordinal_position,column_default,is_nullable,data_type,character_maximum_length from information_schema.columns where table_name = 'pg_database';
column_name | ordinal_position | column_default | is_nullable | data_type | character_maximum_length
---------------+------------------+----------------+-------------+-----------+--------------------------
datname | 1 | | NO | name |
datdba | 2 | | NO | oid |
encoding | 3 | | NO | integer |
datcollate | 4 | | NO | name |
datctype | 5 | | NO | name |
datistemplate | 6 | | NO | boolean |
datallowconn | 7 | | NO | boolean |
datconnlimit | 8 | | NO | integer |
datlastsysoid | 9 | | NO | oid |
datfrozenxid | 10 | | NO | xid |
datminmxid | 11 | | NO | xid |
dattablespace | 12 | | NO | oid |
datacl | 13 | | YES | ARRAY |
(13 rows)

T-SQL Merging data

I've imported data from an XML file by using SSIS to SQL Server.
The result what I got in the database is similar to this:
+-------+---------+---------+-------+
| ID | Name | Brand | Price |
+-------+---------+---------+-------+
| 2 | NULL | NULL | 100 |
| NULL | SLX | NULL | NULL |
| NULL | NULL | Blah | NULL |
| NULL | NULL | NULL | 100 |
+-------+---------+---------+-------+
My desired result would be:
+-------+---------+---------+-------+
| ID | Name | Brand | Price |
+-------+---------+---------+-------+
| 2 | SLX | Blah | 100 |
+-------+---------+---------+-------+
Is there a pretty solution to solve this in T-SQL?
I've already tried it with a SELECT MAX(ID) and then a GROUP BY ID, but I'm still stuck with the NULL values. Also I've tried it with MERGE, but also a failure.
Could someone give me a direction where to search further?
You can select MAX on all columns....
SELECT MAX(ID), MAX(NAME), MAX(BRAND), MAX(PRICE)
FROM [TABLE]
Click here for a fiddley fidd fiddle...

Is this a good design for a table?

Each user/person could know one or more languages.
All I can think is a table like
+----------+------+-----+------------+-----+-----+-----+-------+
| PersonID | Java | PHP | Javascript | C++ | C | CSS | HTML |
+----------+------+-----+------------+-----+-----+-----+-------+
| 1 | Yes | Yes | No | Yes | No | Yes | No |
| 2 | No | Yes | Yes | No | Yes | No | No |
| 3 | Yes | No | Yes | Yes | Yes | Yes | No |
+----------+------+-----+------------+-----+-----+-----+-------+
Considering I'm going to need at least 100 columns for all the languages, is it normal to have that many columns? Something tells me this is the wrong approach.
Thank you very much and sorry about my english!
I would suggest you to create three tables.
One table contains the information of the Person like his Name etc.
Second table contains two columns LanguageId and Language name.
+------------+-----------+
| LanguageID | Name |
+------------+-----------+
| 1 | Javascript|
| 2 | C |
| 3 | C++ |
+------------+-----------+
Third table contains the Id, PersonId, LanguageID. In this table you can join the above two tables record.
+---+----------+------------+
|ID | PersonID | LanguageID |
+---+----------+------------+
|1 | 1 | 1 |
|2 | 2 | 2 |
|3 | 3 | 3 |
+---+----------+------------+
Reasons to support my answer:
In future if you want to add any new language in your table then it
would be easier to add that in the main table.
You can join the two tables easily and get the result
A little improvement we can do over Rahul Tripathi response is to remove the "Known" column. You need only two tables for this case. One containing PersonId and LanguageId the person knows. The second table is for the languages only.
You know what languages one person knows by joining both tables. For example if you need to know the list of known languages you can do:
SELECT p.PersonId, l.Name
FROM Person p INNER JOIN Language l ON (p.LanguageId = l.LanguageId)
WHERE (p.PersonId = theIdYouNeedToKnow)

Rearranging and deduplicating SQL columns based on column data

Sorry I know that's a rubbish Title but I couldn't think of a more concise way of describing the issue.
I have a (MSSQL 2008) table that contains telephone numbers:
| CustomerID | Tel1 | Tel2 | Tel3 | Tel4 | Tel5 | Tel6 |
| Cust001 | 01222222 | 012333333 | 07111111 | 07222222 | 01222222 | NULL |
| Cust002 | 07444444 | 015333333 | 07555555 | 07555555 | NULL | NULL |
| Cust003 | 01333333 | 017777777 | 07888888 | 07011111 | 016666666 | 013333 |
I'd like to:
Remove any duplicate phone numbers
Rearrange the telephone numbers so that anything beginning with "07" is the first phone number. If there are multiple 07's, they should be in the first fields. The order of the numbers apart from that doesn't really matter.
So, for example, after processing, the table would look like:
| CustomerID | Tel1 | Tel2 | Tel3 | Tel4 | Tel5 | Tel6 |
| Cust001 | 07111111 | 07222222 | 01222222 | 012333333 | NULL | NULL |
| Cust002 | 07444444 | 07555555 | 015333333 | NULL | NULL | NULL |
| Cust003 | 07888888 | 07011111 | 016666666 | 013333 | 01333333 | 017777777 |
I'm struggling to figure out how to efficiently achieve my goal (there are 600,000+ records in the table). Can anyone help?
I've created a fiddle if it'll help anyone play around with the scenario.
You can break up the numbers into individual rows using UNPIVOT, then reorder them based on the occurence of the '07' prefix using ROW_NUMBER(), and finally recombine it using PIVOT to end up with the 6 Tel columns again.
select *
FROM
(
select CustomerID, Col, Tel
FROM
(
select *, Col='Tel' + RIGHT(
row_number() over (partition by CustomerID
order by case
when Tel like '07%' then 1
else 2
end),10)
from phonenumbers
UNPIVOT (Tel for Seq in (Tel1,Tel2,Tel3,Tel4,Tel5,Tel6)) seqs
) U
) P
PIVOT (MAX(TEL) for Col IN (Tel1,Tel2,Tel3,Tel4,Tel5,Tel6)) V;
SQL Fiddle
Perhaps using cursor to collect all customer id and sorting the fields...traditional sorting technique as we used to do in school c++ ..lolz...like to know if any other method possible.
If you dont get any then it is the last way . It will take a long time for sure to execute.

Resources