SQL-SAP HANA-控制台内容太长-将一些代码传递给过程？

Question

I have an SQL Script in my SAP HANA SQL Console with ~1700 rows. I have so many rows, as I need to change alot of entries in one column which is imported from one "raw table". More detailed description about this "change row entries" problem The repetitive WHEN THEN Code looks like: `

Create Column Table xxx
Select Distinct 
   aaa AS "aaa",
   bbb AS "bbb",
   ccc AS "ccc",
   CASE 
      WHEN importedColumn like 'xx123x' THEN REPLACE_REGEXPR ('xx123x' FLAG 'i' IN importedColumn WITH 'xxx')
      WHEN importedColumn like 'yy345y' THEN REPLACE_REGEXPR ('yy345y' FLAG 'i' IN importedColumn WITH 'yyy') 
      WHEN importedColumn like 'zzz345z' THEN REPLACE_REGEXPR ('zzz345z'FLAG 'i' IN importedColumn WITH 'zzzz') 
      etc.
      etc. 
      ELSE xxx 
   END AS replace_regexpr 
FROM...
WHERE...

This (generally) works fine. The problem is, I have so many different WHEN THEN Statements (Always checking the entries of this one column and maybe change them) that I cannot execute the Code anymore -> SQL console Content is too large. Can I extract this case Statement somehow and store it in an procedure or something like this? I really have no clue how I can handle this, so I am happy for every advise!

• Note: The column entries that need to be changed do not have any common pattern, so I cannot group them or something like this. This means that I need that many WHEN THEN Statements.

Answer 1

From this and the linked question I take it that you have a single REGEX rule that should be applied to the input data if and only if a certain pattern is found (compared via LIKE operator).

I would put all matching pattern and replacement rules into a separate table, making maintenance, debugging and the SELECT statement a lot easier:

create column table xxx (id int primary key
                       , "impColumn" nvarchar(4000)
);

create column table rules (id int primary key
                         , matcher nvarchar(4000)
                         , regex nvarchar(4000));       

insert into rules values (1, 'xx123x', 'xxx');
insert into rules values (2, 'yy345y', 'yyy');
insert into rules values (3, 'zzz345z', 'zzz');


insert into xxx values (1, 'xx123x');
insert into xxx values (2, 'yy345y');
insert into xxx values (3, 'zzz345z');

insert into xxx values (4, 'xx123xyy345y');
insert into xxx values (5, 'xx1zzz345zzzz345z23x');
insert into xxx values (6, 'xx123xyy345yzzz345z');

select
    x.*,
    r.*,
    REPLACE_REGEXPR (r.matcher
                    FLAG 'i' 
                    IN x."impColumn"  
                    WITH r.regex) as output
from 
    xxx x
    left outer join
    rules r
    on x."impColumn" like   r.matcher  ;

You could change the join condition to on x."impColumn" like '%'|| r.matcher ||'%'

to apply all matching rules to the input rows, but be aware that this produces one output row per applied rule, which is probably not what you want.

The output without the placeholder looks like this:

Which is pretty much the same what your example SQL produces.

As for the problem in SAP HANA Studio, I guess that this is due to the sheer size of the script file you try to run. By editing the JRE VM memory for HANA Studio you may be able to avoid this problem.

Please check the hdbstudio.ini and set the VM parameter to

• -Xmx4096m
• -Xms512m

and restart HANA Studio.