Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!mailrus!cwjcc!hal!ncoast!nc386!allbery
From: allbery@nc386.UUCP (Brandon S. Allbery)
Newsgroups: news.software.b
Subject: Re: bdf for use in C news (was Re: C news vs B news speed)
Message-ID: <1989Jul23.182920.15519@nc386.uucp>
Date: 23 Jul 89 18:29:20 GMT
References: <1989Jul12.210254.8631@utzoo.uucp> <1018@cniysc.cinnet.COM> <297@ecijmm.UUCP> <411@lxn.eds.com>
Sender: allbery@nc386.uucp (Brandon S. Allbery)
Reply-To: allbery@nc386.UUCP (Brandon S. Allbery)
Organization: North Coast Computer Resources, Cleveland, OH
Lines: 33
In-reply-to: chris@lxn.eds.com (Christopher D. Orr)

There is a very simple way to extract the number of free blocks from a
System V "df"; it's currently working on nc386 (SCO Xenix), telotech (Altos
System V) and ncoast (AT&T System III(!)), each of which has a slightly
different "df" output (in fact, I had to rewrite "bdf"-for-Perl to work on
ncoast).

Extracted from our "spacefor":

## this is set up for the stupid System V df
# which only proves that Collyer and Spencer don't understand System V df....
df $arg | sed 's/^[^ ]*  *([^)][^)]*):  *\([0-9][0-9]*\) .*$/\1/' | awk "{
		nb = (\$1 - $desire) * $dfunit / $1
		if (nb > 10000)
			nb = 10000	# ensure representable as integer
		nb = int(nb)
		if (nb <= 0)
			print 0
		else
			print nb
		exit
	}"

Note the "sed" command in the pipeline; it extracts the "blocks" part of the
"df" output, and does so quite portably.

I left the "awk" mostly as is, but could have written the sed script as part
of the awk script instead or performed other optimizations.

++Brandon
-- 
Brandon S. Allbery, moderator of comp.sources.misc	     allbery@NCoast.ORG
uunet!hal.cwru.edu!ncoast!allbery		    ncoast!allbery@hal.cwru.edu
   * This message brought to you courtesy the "Watcher" for the 4th NCoast *