Oppolzer - Informatik / Blog

With my tool, you can choose the delimiter by parameter (some like semicolon
instead of comma, because in some European countries the comma is used instead
of decimal point to separate the decimal fraction - tab is possible too, of
course), and char fields are enclosed into quotes (or apostrophes), if
necessary. ExCel etc. deal with all this, it's tested.

Furthermore, you can choose if you want column headings or not. If you have
column headings (derived from the SQL result), you can later use those headings
to reference your columns, if you use the CSV to do updates on the DB again. If
not, you have to specify the numeric position of the column ... in any case, you
have to tell the tool the data type in this case, because it does no pre scan of
the file, as ExCel does. But this is for the load situation only; for unload,
it's as easy as

unlddn csvhead := unload qualif.db2table;

or

unlddn csv := select * from qualif.db2table where ...;

Kind regards

Bernd



Am 26.02.2015 um 04:11 schrieb P.G.:
> On Wed, 25 Feb 2015 17:42:16 -0600, T. wrote:
>
>> I'd certainly prefer zOS/batch/DFSORT.  Unload the table to FB disk, use
>> DF/SORT to insert x'05'(tab) or comma(CSV) where desired.  I do this
>> quite often to create a flat file that will eventually go back to MS
>> excel/access.  Millions of records?  No big deal.
>>
> And suppose your data contain commas in some column, in some rows
> but not all?  E.g. "Babonas, T."  Well, the designers should have split
> those fields into separate columns, but sometimes they don't.  Choose
> a different separator?  If you don't know a priori which character meets
> that requirement you must do a preliminary scan to find one.  And if
> you find none you must establish a convention for escaping metacharacters.
> I suppose DFSORT can do all this.  (Hmmm...  Has anyone proven that
> DFSORT is Turing-complete?  Or that it isn't?)
>
> Faced routinely with such a problem given a .xlsx file, I open it with
> LibreOffice and export as HTML, which guarantees that its delimiter
> tokens don't appear among the output data.
>
> Bernd probably has an effective solution.  LibreOffice doesn't batch
> at all well.
>
> -- G.
>