[CentralOH] parsing Windows command line output

[William McVey]

On Wed, 2008-07-23 at 09:30 -0400, Mark Peters wrote:
> Use the default option to split:

The problem is that some of his fields (e.g. first and second) have
embedded whitespace in the actual field values. If his second column
didn't fluctuate between spaces and non-spaces, he would have been able
to do something like:

>>> line= "Quest SQL Watch Agent ORIONMFG1 APPS       MFGSQL-S1A      Online"
>>> line.rsplit(None, 4)
['Quest SQL Watch Agent', 'ORIONMFG1', 'APPS', 'MFGSQL-S1A', 'Online']

However, that breaks on lines like:

>>> line= "Cluster IP Address   ORIONMFG-SAN1        MFGSQL-S1A      Online"
>>> line.rsplit(None, 4)
['Cluster IP', 'Address', 'ORIONMFG-SAN1', 'MFGSQL-S1A', 'Online']

Perhaps ORIONMFG-SAN1 is a special case and could be mapped to "ORIONMFG SAN1":

>>> line.replace("ORIONMFG-SAN1", "ORIONMFG SAN1").rsplit(None, 4)
['Cluster IP Address', 'ORIONMFG', 'SAN1', 'MFGSQL-S1A', 'Online']

Ordinarily, parsing columnar output screams for string slices; however
your format is particularly nasty in that the columns are *mostly* fixed
width, but will (on a case by case basis) overflow the fixed width when
a field is too long and turn into a delimited output format with an
ambiguous field delimiter. A good example of this is that line that
starts with "Quest SQL Watch Agent". If it weren't for the possibility
of field overflow, you could do something like:
        
        for line in file("qry1.txt").readlines():
                fields = line[:21], line[21:42], line[42:58], line[58:]
                resource, group, node, status = [x.strip() for x in fields]

As it happens, this routine *will* parse the output you provided, but be
aware that any resources that are longer than 21 characters will likely
cause the routine to choke. Unless I knew that there was no possibility
of field overflow, I'd try to stick to the rsplit() option, even if it
requires a hack to map particular values into a more accommodating
format.

  -- William

P.S. I'd be really surprised if there is something that the cluster
command is doing that couldn't be done directly within python using the
pywin32 api.