From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <fcc66408f625921462ad0a49197ed66a@coraid.com>
From: erik quanstrom <quanstro@coraid.com>
Date: Mon, 18 Dec 2006 07:05:00 -0500
To: steve@quintile.net, 9fans@cse.psu.edu
Subject: Re: [9fans] silly rc quoting question
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Cc: 
Topicbox-Message-UUID: f5c7d9de-ead1-11e9-9d60-3106f5b1d025

how about
	; eval 'a = (' `{cat junk} ')'
	; whatis a
	a=(f1 f2 'free text' f4)
	; echo $#a
	4

the reason that your original failed is that /bin/read returns one
line at a time.  the rc backquote operator then tokenizes not
according to rc's syntax rules but by $ifs.  the default for $ifs is
space, newline, tab.  thus the string
	f1 f2 'free text' f4newline
is parsed as
	"f1", "f2", "'free", "text'", f4

- erik

On Mon Dec 18 06:34:41 EST 2006, steve@quintile.net wrote:
> I have realised I don't understand rc's quoting
> rules as well as I thought I did. I want to parse
> the output of a program which has one record per line,
> and white space delimited fields in each line.
> 
> Some fields have whitespace in them but if they do then
> these fields are surrounded in quotes; here I will use cat
> and a file as an example, in "real life" its all a
> bit more complex.
> 
> 	term% cat junk
> 	f1 f2 'free text' f4
> 
> I expect when I parse this using the backquote operator then
> the quoting would be respected, this appears not to be
> the case:
> 
> 	term% cat junk | while(a=`{read}){
> 		echo $#a
> 		echo $a(3)
> 	}
> 	5
> 	'free
> 
> Where as I would expect:
> 
> 	4
> 	free text
> 
> I know I cannot do this with awk (unless I build the parser in awk
> (ugh)). Having said this I have been tempted to produce a non-standard,
> awk9 which understands plan9's unified quoting rules, even things like
> 
> 	ls -l | awk '{print $6, $10}'
> 
> are not reliable given whitespace in file names.
> 
> Is it possible to use rc to parse data using rc's rules?
> 
> -Steve