reading data from csv files

[Jakson A. Aquino]

Hi Andreas,

On Thu, Sep 07, 2006 at 09:47:07AM -0700, Andreas Beyer wrote:

> Fine-tuning via command line options is good, but eventually you may
> want to allow for defining the details when opening the file (in
> interactive mode). 

This isn't difficult to implement. We have just to define the
better way of doing it.

> This would allow to define details for every file separately.  For
> instance there could be a sub-menu DATA MANAGEMENT/FILE FORMAT
> OPTIONS

I can add an option to statistrc defining whether statist will try to
auto detect the file format or ask the user. In auto detect mode (and
not in silent mode), if some error happens (different number of
columns for different lines), the user would be asked to set the
options.  Before asking the questions, statist could show the first
few lines (truncated after about 40 characters), just as it shows the
content of the current directory when asking for a file name.

The sub-menu DATA MANAGEMENT/FILE FORMAT is useful too. In auto detect
mode, statist will read:

     1,1 2,2        1 1 2 2
     1,1 2,2   as   1 1 2 2
     1,1 2,2        1 1 2 2

Because statist would find the same number of columns in all lines,
there would be no error message. Only the presence of column names
would avoid that statist wrongly read this file. The sub-menu item
would make statist re-read the file with the correct options. Would
this statist behavior be OK in your opinion? Or were you think in
something different?

Jakson

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://www.intevation.de/pipermail/statist-list/attachments/20060907/831bbfab/attachment.bin