Tag: sql

Oracle Collection Instead of Dual

I’m still retrofitting a bunch of SQL queries to use bind_by_name and came across a strange scenario. I created a recursive query (STARTS WITH / CONNECT BY PRIOR) but I needed to grab the original value too. The quickest way to accomplish this was to union in something like “select 12345CDE as equipment_id from dual”. But the only way to get a bunch of these original values grafted onto the result set is to iterate through the array once to build my :placeholder1, :placeholder2, …, placeholderN placeholders and then iterate through the array again to bind each placeholder to its proper value.

I’ve been working with Oracle collections for LIKE and IN queries, and thought I could use a table that only exists within the query to glom the entire array into a single placeholder. It works! A query like

select column_value equipment_id from TABLE(sys.ODCIVARCHAR2LIST('12345CDE', '23456BCD', '34567ABC') );

Adds each of the values to my result set.

Which means I can use a query like “select column_value as equipment_id from TABLE(:myIDs)” and bind the collection to :myIDs.

PostgreSQL Sequences

I’m having a problem with a database refusing to change INSERTS to UPDATES on the ON CONFLICT condition — it insists that a ‘duplicate key value violates unique constraint’. A little time with a search engine tells me that sequences can get out of sync, and then you cannot insert items into the table. How do you know your sequence is out of sync?

SELECT NEXTVAL((SELECT PG_GET_SERIAL_SEQUENCE('"table_name"', 'name_of_column_with_sequence'))) as "NextValue", MAX("name_of_column_with_sequence") AS "Max Value" FROM "table_name";

So null seems like it would be a problem!

For future reference, when the next value is smaller than the max value in the table, the solution is to set the series value based on the max value

SELECT SETVAL((SELECT PG_GET_SERIAL_SEQUENCE('"table_name"', 'name_of_column_with_sequence')), (SELECT (MAX("name_of_column_with_sequence") + 1) FROM "table_name"), FALSE);

Oracle: Query Timestamp Today

I wanted a quick way to query a logging table for today’s records. I figured there had to be some way to put a variable “today” into the query rather than put in the numbers for the current date. Voila — a query that retrieves records where the timestamp is greater than or equal to today:

select *
from ossa_central_logging
where action_ts >= cast(trunc(current_timestamp) as timestamp)
order by action_ts desc;

Oracle – Select Top

I discovered the “rownum” trick early in my usage of Oracle databases — especially useful for sampling data to see what’s in there, something like “select * from dataTable where rownum < 6” gets you the first five records. But that’s not suitable if you want to sort the records. In this particular case, I have a series of names. I want to find the highest number value in the series so I can name my object with the next sequential name.

Enter “fetch first”  … this appears to be available since 12c (so older database installations may still require a more convoluted solution):

SELECT set_name from set_data
WHERE set_name LIKE 'Something-With-A-Series-%'
ORDER BY set_name DESC
fetch first 1 row only;

Which returns the last name in the series.

Oracle – Group By Having

I needed a query to find records where duplicate name values exist. I know how to group by and count, but the table has millions of records. I don’t want the 99% of the data where no duplication occurs. By using “having” in conjunction with “group by”, I am able to restrict the output to the groups that match my criterion.

select display_name, count(display_name) from circuit
group by display_name
having count(display_name) > 1;

My result set is the display name & occurrence count for that display name without all of the ‘good’ records where there’s a unique display name. (Yes, I know uniqueness could be enforced. The real scenario isn’t this straight-forward. There are times where the display name should be the same and I’ve got additional filters that drop out those cases).

 

MySQL: Moving Data From One Table To Another

Our OpenHAB persistence data is stored in MySQL. There’s an “items” table which correlates each ItemName string to an ItemID integer. There are then Item#### tables that store persistence data for each item. If you rename an item, this means a new table is created and previous persistence data is no longer associated with the item. For some items, that’s fine — I don’t really care when the office light was on last month. But there’s persistence data that we use over a long term — outdoor temperature, luminance, electrical usage. In these cases, we want to pull the old data into the new table. There’s a quick one-liner SQL command that accomplishes this:

INSERT INTO NewTable SELECT * from OldTable;
e.g. INSERT INTO Item3857 SELECT * FROM Item3854;

You can drop the old table too:

DROP OldTable;

But I run a cleanup script against the item list so often don’t bother to remove tables one-off.