splitting tables

[Bengt Richter]

On Sun, 8 Feb 2004 14:41:55 +0000 (UTC), robsom <no.mail at no.mail.it> wrote:

>Il Sat, 07 Feb 2004 22:10:17 +0100, Karl Pflästerer ha scritto:
>
>> What do you want to be done?  To see if an item is missing is trivial:
>> just check the length of the splitted line (a list).  But what the right
>> action in that case is belongs to you; should the user be asked? is
>> always the same column missing? is it possible to distinguish the
>> entries without errors from each other so the programm can decide which
>> column is missing?
>
>Ok, I'll try to give some more detail. These are files with data from
>field measurements and contain information about location, time,
>measurement, measurement flag, error, detection limit, calibration and
>other stuff like that. The problem is that they are made by different
>groups and are not always consistent in their format and I'm trying to
>write a code which is as general as possible.
>When a table has fixed-width columns and each element is correctly aligned
>as in the example I showed you, the problem is solved by slicing the
>string as Skip Montanaro suggested in his answer (thanks!), but this is
>not always the case. For example I can have something like this:
>
>47.455677 456.67
>47.4558 453.8
>47.46789 -9999
>47.4567 456
>
>where -9999 (or somethinbg similar) indicates there is a blank, one space
>divides the columns and the elements can have a different number of
>digits. This is of course a worst-case scenario :)
>That is why I used split in the beginning, but then I fall into the other
>problem, when there is a missing value.
>Any suggestions will be much appreciated, thanks
>
Maybe a modified regex that takes into account particular field formats?
A regex will search for things in order, so you can set one up to match
special things like -9999 while still allowing -9999.9 etc.
What do you know about each field and the separations? Is there always a full
set of fields, even if some are blank?
 
(BTW, my other post misleads in implying that line.rstrip('\n') is necessary
to get the regex to match).

Regards,
Bengt Richter