SQLite: result of mathematical operation always is a text - sqlite

I created the following table:
CREATE TABLE test (a INT,b INT);
After I inserted some data:
INSERT INTO test VALUES(1,2);
When I execute this SELECT:
SELECT cast(b as real) as x, a * b as y FROM teste
the fields "x" and "y" return with datatype TEXT. I'm using Delphi and SQLiteStudio 2.1.5, and both return the same datatype.
I need that field x being real and y being int.
Someone could help me?


Sqlite does not have column data type. It is just text. Based on your expression, it will try to do the conversion for you. For your math expression, you need to do a cast to give it is a hint to delphi to create your desired column type.


For me it's strange that in Sqlite, column types are "recommendations". But analyzing this I changed my program. I created the fields manually with correct type. So the program start to work.
The problem is when you open a query and the Delphi create the columns automatically. In this case, the columns with cast or with mathematical operation are created like text field.

Related

Octave: Read BLOB data from sqlite db file

I try to read BLOB data from a database file with Octave 6.2.0 and the mex-sqlite3-master package.
I am able to select and read any other data from my database file. For the column containing BLOB data it gives me the following:
octave> x=sqlite3('file.db', 'SELECT column FROM list');
error: sqlite3: unsupported column type
octave> x=sqlite3('file.db', 'SELECT column FROM list WHERE column=CAST(column AS TEXT)');
gives no error, however x with a dimension of 1x0.
The BLOB-data contains hexadecimal numbers. I am fine having them as string (and work my way further, no probs).
What can I do to extract the BLOB-data in a processable format?
Thanks for any hint!

thanks a lot for your answers! I tried the "select CAST(column AS TEXT) from bl;" hint. It created an array of the expected size, however with empty cells.

I think you might want:
SELECT hex(column) FROM list
instead of doing a CAST to TEXT.
See Sqlite: How to cast(data as TEXT) for BLOB.

Date parameter mis-read in Delphi SQLite Query

What is wrong with my code:
ExecSql('DELETE FROM STLac WHERE RegN=99 AND BegDate>= 2016-12-14');
This runs, but deletes ALL the rows in STLac for RegN, not just the rows with BegDate on or after 2016-12-14.
Originally I had:
ExecSql('DELETE FROM STLac WHERE RegN=99 AND BegDate>= :myDdate,[myDate]);
which has the advantage I hoped of not being particular to the date format. So I tried the literal date should in the format SQLite likes. Either way I get all rows deleted, not just those on or after the specified date.
Scott S.

Try double quote while putting date. As any value must be provided in between quotes until and unless that column is not int
ExecSql('DELETE FROM STLac WHERE RegN=99 AND BegDate>= "2016-12-14"');

SQLite does not have datetime format as such, so you have to figure out how date is actually represented in the table and change your query to provide the same format. First execute the "select" statement in some kind of management tool,
select * from STLac where RegN = 99 and BegDate >= '2016-12-14' --(or '2016.12.04' or something else)
which displays the result in the grid; when you see the expected rows, change it to "delete" query and copy into your Delphi program.

Handling dimension with empty names in icCube

How can I get icCube to handle a dimension where there are empty values in the backing table? I get the following error message from icCube when I try to scan our tables:
Data table 'public.accounting_area_dim', line '22' : The member key:'22' has no name (nameColumn:area) (level:[Area].[Area].[Area])
This is absolutely correct, there are empty strings in the dimension table - I'd like to treat it as "Unknown" or something similar, is this possible in icCube?

You can transform the empty string into something else within the table definition as following:
Hope that helps.

teradata : to calulate cast as length of column

I need to use cast function with length of column in teradata.
say I have a table with following data ,
id | name
1|dhawal
2|bhaskar
I need to use cast operation something like
select cast(name as CHAR(<length of column>) from table
how can i do that?
thanks
Dhawal

You have to find the length by looking at the table definition - either manually (show table) or by writing dynamic SQL that queries dbc.ColumnsV.
update
You can find the maximum length of the actual data using
select max(length(cast(... as varchar(<large enough value>))) from TABLE
But if this is for FastExport I think casting as varchar(large-enough-value) and postprocessing to remove the 2-byte length info FastExport includes is a better solution (since exporting a CHAR() will results in a fixed-length output file with lots of spaces in it).
You may know this already, but just in case: Teradata usually recommends switching to TPT instead of the legacy fexp.

pl/pgsql merging or combining arrays

How to merge/combine arrays in pl/pgsql?
For example I have an 3 arrays: {1,2,3}, {"a","b","c"}, and {32,43,23}
After merging I need to get:
{{1,"a",32}, {2,"b",43}, {3,"c",23}}
My version of PostgreSQL is 9.0

It sounds like you want an n-argument zip function, as found in some functional languages and languages with functional extensions.
In this case you can't do exactly what yu want, because those arrays are of hetrogeneous types. PostgreSQL arrays must be of homogenous type, so this won't work. The desired result you show is an invalid array.
You could create an array of ROWs (anonymous records), or cast all the values to text.
For example:
SELECT array_agg(ROW(a,b,c))
FROM (
SELECT
unnest('{1,2,3}'::integer[]),
unnest('{"a","b","c"}'::text[]),
unnest('{32,43,23}'::integer[])
)
x(a,b,c);
will produce:
{"(1,a,32)","(2,b,43)","(3,c,23)"}
which is an array of three rowtypes cast to text. It will be awkward to work with because Pg has only very limited support for anonymous records; most importantly in this case you cannot cast a text value to RECORD(integer,text,integer), you must actually CREATE TYPE and cast to the defined type.
Because of that limitation, you may instead want to cast all the values to text and use a two-dimensional array of text. You'd expect to be able to do that with a simple array_agg, but frustratingly this fails:
SELECT array_agg(ARRAY[a::text,b,c::text])
FROM (
SELECT
unnest('{1,2,3}'::integer[]),
unnest('{"a","b","c"}'::text[]),
unnest('{32,43,23}'::integer[])
)
x(a,b,c);
producing:
ERROR: could not find array type for data type text[]
because array_agg doesn't support arrays as input. You need to define another variant of array_agg that takes a text[] input. I wrote one a while ago but can't find it now; I'll try to locate it and update if I find it. In the mean time you can work around it by casting the inner array to text:
SELECT array_agg(ARRAY[a::text,b,c::text]::text)
FROM (
SELECT
unnest('{1,2,3}'::integer[]),
unnest('{"a","b","c"}'::text[]),
unnest('{32,43,23}'::integer[])
)
x(a,b,c);
producing output like:
{"{1,a,32}","{2,b,43}","{3,c,23}"}
... OK, I haven't found the one I wrote, but here's an example from Erwin that does the job fine. Try this:
CREATE AGGREGATE array_agg_mult (anyarray) (
SFUNC = array_cat
,STYPE = anyarray
,INITCOND = '{}'
);
SELECT array_agg_mult(ARRAY[ARRAY[a::text,b,c::text]])
FROM (
SELECT
unnest('{1,2,3}'::integer[]),
unnest('{"a","b","c"}'::text[]),
unnest('{32,43,23}'::integer[])
)
x(a,b,c);
Output:
{{1,a,32},{2,b,43},{3,c,23}}

Resources