Re: UNSTRINGing a CSV line (was: If you were inventing CoBOL...)

On 23 Sep 2004 11:00:00 GMT, l.willms@jpberlin.de (Lueko Willms)
wrote:

>.    Am  22.09.04
>schrieb  robert@wagner.net.yourmammaharvests (Robert Wagner)
>    auf  /COMP/LANG/COBOL
>     in  vst1l01fre1nu9dq6tvkh7pbsshpebjlem@4ax.com
>  ueber  Re: If you were inventing CoBOL...
> 
>
>RW> Parse-CSV-Row.
>RW> move 1 to CSV-Pointer
>RW> perform until CSV-Pointer greater than length(CSV-Row)
>RW>     move spaces to Next-Token
>
>        move zero to item-number
>
>RW>     if  CSV-Row (CSV-Pointer:1) = '"' or "'"
>RW>         add 1 to CSV-Pointer
>RW>         unstring CSV-Row
>RW>             delimited by CSV-Row(CSV-Pointer-1:1)
>
>*>   I would change "Next-Token" to a table item:
>
>RW>             into Next-Token
>                into CSV-item(item-number)
>
>*>   and add TALLYING:
>                   TALLYING IN item-number

Your two suggestions are in conflict. Occurs Depending On cannot be
used on table items.

>RW>             with pointer CSV-Pointer
>RW>         unstring CSV-Row delimited by ', ' or ','
>RW>             into trash with pointer CSV-Pointer
>RW>    else
>RW>         unstring CSV-Row
>RW>             delimited by ', ' or ','
>                into CSV-item(item-number)
>                   tallying in item-number
>RW>             with pointer CSV-Pointer
>RW>    end-if
>RW> end-perform.
>
>   Also, the delimiter should be made into a variable "CSV-delimiter"
>e.g., so that it can be set to comma, semikolon, TAB, US, or whatever.
>
>   I also think one should use someting like
>
>    INSPECT CSV-item[CSV-Pointer]
>       TALLYING CSV-Pointer FOR LEADING SPACES CSV-delimiter
>
>   to skip over the leading spaces of each item.

Good for leading spaces, but I don't want to discard delimiters.

>   Wouldn't it be possible to implement this as an object CLASS?

Yes. In that case, we're discussing code inside the unpack method.