SQL query question

[Centrally]

I'm trying to use autoIT script to populate database but I bumped on one problem.

What I need is query that would check if value in specific field exist (like serial name of product), if it exist than update selected fields or if not insert new value.

This is initial query I have made:

"INSERT INTO `Baza1` (`Datum Objave`,`Serija`,`Ime predmeta`,`Broj pregleda`,`cijena`,`Link do oglasa`,`Koeficijent` ) VALUES (" & $datum1 & "," & $serija & ", " & $ime1 & ", " & $pregledBroj1 & ", " & $cijena1 & ", " & $pathlink & ", " & $koeficient & ")"

This query would only insert new data without checking if it already exist.

I have tried to use replace query but I simply cannot understand it's format.

I don't know how to select what field to check for duplicate and how to update other fields if criteria is matched.

Any explanation or brief part of code example is highly appreciated. Thanks

 

[Zedna]

Instead of

INSERT INTO table (columns) VALUES (values)

use

INSERT INTO table (columns) SELECT values WHERE NOT EXISTS(SELECT 1 FROM table WHERE id = your_id)

Then use your UPDATE

[MHz]

REPLACE is an alias for INSERT for compatibility with other databases so I would not be concerned about using it.

This is how to use UPDATE to change the values.

Here is an example using the SQLite3.exe.

Create a table

CREATE TABLE "tab" VALUES ('col1', 'col2');

Insert values

INSERT INTO "tab" VALUES ('cat', 'dog');

Show output using select

SELECT * FROM "tab";
cat|dog

Update if "col1" = 'pig'. (pig does not exist in col1 so value in col2 is not set)

UPDATE "tab" SET "col2" = 'mouse' WHERE "col1" = 'pig';

Show output using select (no change)

SELECT * FROM "tab";
cat|dog

Update if "col1" = 'cat'

UPDATE "tab" SET "col2" = 'mouse' WHERE "col1" = 'cat';

Show output using select (changed)

SELECT * FROM "tab";
cat|mouse

This is probably a slow method as you are using UPDATE no matter if needs updating or not. You could use SELECT to get the column of interest with all of the values. Find in the values which exist that you have, and then UPDATE the values that are missing which is probably what Zedna is doing with his nested SELECT. Although if you knew the value did not exist in the column when using SELECT, then you could just use INSERT for the new row.

UPDATE with comments

UPDATE "tab" -- update this table
SET "col2" = 'dog' -- set value dog in this column
WHERE "col1" = 'cat' --only where cat is in this column
;

[Centrally]

Thank you for your answers I will respond back to this topic when I test your examples.

[Xenobiologist]

Using a "real" database like DB2 then the command Merge is what you are looking for.

(null)

[jchd]

BTW I suspect you use MySQL. While schema names (tables, columns, ...) may be enclosed in backquotes, literal strings should use single quotes and even then, should be escaped (have significant single quotes doubled).

Your question is perhaps more a DB design question.

BTW, there is no such thing as a "real" DB engine, just like for programmers.

[Centrally]

 

REPLACE is an alias for INSERT for compatibility with other databases so I would not be concerned about using it.

This is how to use UPDATE to change the values.

Here is an example using the SQLite3.exe.

Create a table

CREATE TABLE "tab" VALUES ('col1', 'col2');

Insert values

INSERT INTO "tab" VALUES ('cat', 'dog');

Show output using select

SELECT * FROM "tab";
cat|dog

Update if "col1" = 'pig'. (pig does not exist in col1 so value in col2 is not set)

UPDATE "tab" SET "col2" = 'mouse' WHERE "col1" = 'pig';

Show output using select (no change)

SELECT * FROM "tab";
cat|dog

Update if "col1" = 'cat'

UPDATE "tab" SET "col2" = 'mouse' WHERE "col1" = 'cat';

Show output using select (changed)

SELECT * FROM "tab";
cat|mouse

This is probably a slow method as you are using UPDATE no matter if needs updating or not. You could use SELECT to get the column of interest with all of the values. Find in the values which exist that you have, and then UPDATE the values that are missing which is probably what Zedna is doing with his nested SELECT. Although if you knew the value did not exist in the column when using SELECT, then you could just use INSERT for the new row.

UPDATE with comments

UPDATE "tab" -- update this table
SET "col2" = 'dog' -- set value dog in this column
WHERE "col1" = 'cat' --only where cat is in this column
;

 

First off all I am using MySQL and UDF >EzMySql. 

Now from your set of examples I have learned how to check if specific value in one field exist but I still don't know how to do next thing.

So if I use 

UPDATE "tab" SET "col2" = 'mouse' WHERE "col1" = 'pig';

It does check table for pig and if found is set to mouse, but now if it wasn't found how to add it in a new row. Something like ELSE statement is what I'm looking for 

that would add new row if existing value is not found.

@Zedna

I don't understand this part of query:

WHERE NOT EXISTS(SELECT 1 FROM table WHERE id = your_id)

What this 1 stand for and what id/your_id should be placed here. 

@jchd

I don't care about SQLi yet because this part of db should be manually uploaded to display website with charts few times a day, meaning no actual interaction with website itself.

EDIT:

What do you think about this approach?

INSERT INTO table1
  (`col1`, `col2`, `col3`, `col4`)
VALUES
  ('val1', 'val2', 'val3', 'val4')
ON DUPLICATE KEY UPDATE
  `col2`='val2',
  `col3`='val3', [...]

Edited November 2, 2013 by Centrally

[Centrally]

UPDATE:

This is my end working query if someone is on same road in future:

"INSERT INTO `baza1` (`Datum Objave`,`Serija`,`Ime predmeta`,`Broj pregleda`,`cijena`,`Link do oglasa`,`Koeficijent`) VALUES (" & $datum1 & "," & $serija & ", " & $ime1 & ", " & $pregledBroj1 & ", " & $cijena1 & ", " & $pathlink & ", " & $koeficient & ") ON DUPLICATE KEY UPDATE  `Broj pregleda`=" & $pregledBroj1 & ",  `cijena`=" & $cijena1 & ", `Koeficijent`=" & $koeficient & ""

I removed primary key and set unique ones to id and `Ime predmeta` which is checked for duplicate and decided upon for either updating or inserting.

I would like to give big thanks for all of you here trying to help me find solution.

[MHz]

The id column mentioned is a custom column that many seem to use to show the row number. In SQLite. this is created regardless and can be accessed as rowid, _rowid_ or oid. I cannot speak for all databases. I may consider they do something similar perhaps.

To access the rowid in SQLite is similar to

SELECT "rowid", * FROM "table";

which would output the rowid and the rest of the columns.
So you do not need to create the custom column id though someone may correct me based on better knowledge. Creating a integer primary key can be an alias for the rowid so the rowid is shown to you as say the custom column id.

@Centrally

If that statement solves your initial problem then you have done well.

I am not MySQL tested nor heavily experienced with databases yet.

Edited November 2, 2013 by MHz