I have a table with one column of type text which contains a json string. I would like to do a query where I select a bunch of rows (around 50) and for these rows update a single value in the json that is saved in the text field. So lets say it currently looks like this
{"amount":"45","level":1}
I want to update the amount value for every one of these rows, to for example "level" * 5.
I can't figure out a way to do this with one query since it does not seem possible to alter a single value for a text type field like this. Or am I missing something? Otherwise i will just have to alter it manually for every single row I need to change which would be a pain.
You need to first cast the value to a proper jsonb value, then you can manipulate it using JSON functions.
update the_table
set the_column = jsonb_set(the_column::jsonb, '{level}', to_jsonb((the_column::jsonb ->> 'level')::int * 5))::text
where ....
The expression (the_column::jsonb ->> 'level')::int * 5 extracts the current value of level converts it into an integer and multiplies it with 5. The to_jsonb() around it is necessary because jsonb_set() requires a jsonb value as the last parameter
The '{level}' parameter tells jsonb_set() to put the new value (see above) into the (top level) key level
And finally the whole jsonb value is converted back to a text value.
If you really store JSON in that column, you should think about converting that column to the jsonb data type to avoid all that casting back and forth.
Or maybe think about a properly normalized model where this would be as simple as set level = level * 5
Related
I have a table mytable that stores columns in the form of JSON strings, which contain multiple key-value pairs. Now, I want to extract only a particular value corresponding to one key.
The column that stores these strings is of varchar datatype, and is created as:
insert into mytable(empid, json_column) values (1,'{"FIRST_NAME":"TOM","LAST_NAME" :"JENKINS", "DATE_OF_JOINING" :"2021-06-10", "SALARY" :"1000" }').
As you can see, json_column is created by inserting only a string. Now, I want to do something like:
select json_column.FIRST_NAME from mytable
I just want to extract the value corresponding to key FIRST_NAME.
Though my actual table is far more complex than this example, and I cannot convert these JSON keys into different columns themselves. But, this example clearly illustrates my issue.
This needs to be done over Redshift, please help me out with any valuable suggestions.
using function json_extract_path_text of Redshift can solve this problem easily, as follows:
select json_extract_path_text(json_column, 'FIRST_NAME') from mytable;
I have a table which has two varchar columns where the raw data is stored. This raw data contains value and the field name. Field name is like my_cursor.attribute1. This cursor is of the data type (my_custompackage.custom_data_type).
Now, I need to get the data type of the column present in the custom_data_type.
This is wrong, but to give an idea it's something like
my_custompackage.custom_data_type.attribute13
But so far, I couldn't achieve anything. I have tried taking the value into a separate variable. Like `
select field_name into temp_variable from dual;
Then
select dump(temp_variable) into my_data_type
but it didn't work and I was getting the string value. So, could you please tell me how to proceed with this?
I've got an Sqlite DB where the data looks purely numeric (integers) and the column is typed TEXT. I would like to type it as INTEGER if possible.
What query can check if every cell of a certain column can be successfully casted to INT?
SELECT * FROM table WHERE INT(column) != NULL
Alternatively I would like to check if the cells are numeric (don't have any letters/symbols)
SELECT * FROM table WHERE column NOT LIKE "%a-z%"
As a side note, I wanted to do this to reduce the size of the DB, but since Sqlite uses dynamic typing (per cell typing) would this have ANY effect on the size of the DB?
You have to check whether all values can be converted into an integer:
SELECT *
FROM MyTable
WHERE CAST(MyColumn AS INTEGER) IS NOT MyColumn;
the following function deletes all blanks in a text or varchar column and returns the modified text/varchar as an int:
select condense_and_change_to_int(number_as_text_column) from mytable;
This exact query does work.
Though my goal is to apply this function to all rows of a column in order to consistently change its values. How would I do this? Is it possible with the UPDATE-clause, or do i need to do this within a function itself? I tried the following:
UPDATE mytable
SET column_to_be_modiefied = condense_and_change_to_int(column_to_be_modiefied);
Basically i wanted to input the value of the current row, modify it and save it to the column permanantly.
I'd welcome all ideas regarding how to solve scenarios like these. I'm working with postgresql (but welcome also more general solutions).
Is it possible with an update? Well, yes and sort-of.
From your description, the input to the function is a string of some sort. The output is a number. In general, numbers should be assigned to columns with a number type. The assumption is that the column in question is a number.
However, your update should work. The result will be a string representation of the number.
After running the update, you can change the column type, with something like:
alter table mytable alter column column_to_be_modiefied int;
After running my query I get 1 column result as
5
6
98
101
Is there a way to store this result as array so that I can use it later
in queries like
WHERE NOT IN ('5','6','98','101')
I am aware of storing single variable results but is this possible?
I can not use #Table variable as I will be rerunning the query again in the future and it goes out of scope
There are multiple way of storing those column data like using Temporary Tables or View or Table valued function but IMO there is no need of storing that column data anywhere. You can directly use that column in any query saying below (or) perform a JOIN which would be much better option than NOT IN
select * from
table2
where some_column not in (select column1 from this_table);
While this method is not recommended, storing an array in a single column can be done using CSV's(Comma Separated Values). Simply create a VARCHAR array and store it by storing a string containing the values in a specific order. Basically store all of your values into a string with each value being separated by a comma in that string. Store that into a column of your choice. You can later fetch the string and parse it with a string parser i.e using the .split() function in python. AGAIN I do not recommend doing this, I would instead use multiple columns, one referring to each value and access them that way instead
Using separate columns would make it easy to use in a Stored Procedure.