OK, just when you thought it was safe... <g>
I have glob_dot_filenames set so that files beginning with "." are included.
Here's one of the things I'm calling zap_word() from:
flist=*
for fname in $flist; do
if [ -d "$fname" ]; then
flist=`zap_word "$fname" "$flist"`
fi
done
echo [flist]=[$flist]
(Actually, I have a faster routine than the above that works fine, but I'm
trying to understand why the above does|doesn't work depending on the
following...)
When zap_word() includes 'echo $2 ...', the above works fine (all
directories including '.' and '..' are removed from flist). When I change
that to 'echo "$2" ...', the result is '[flist]=[]'. Knowing that a .
matches any single character, this makes some sense ('echo * | sed "s/ . /
/' wipes out '*' in the first successful 'if' iteration).
What's driving me nuts is that when I try to echo the 'echo' command in the
function in order to _see_ what's happening, I get the same result with and
without the quotes around $2, i.e. within the function...
#echo test
echo 'echo '$2' | sed '"s/^/ / ; s/\$/ / ; s/ $1 / / ; s/^ // ; s/ \$//"
...and...
#echo test
echo 'echo '"$2"' | sed '"s/^/ / ; s/\$/ / ; s/ $1 / / ; s/^ // ; s/ \$//"
both produce identical results:
echo * | sed s/^/ / ; s/$/ / ; s/ . / / ; s/^ // ; s/ $//
Thus, I can't _see_ what's happening.
If I change the 'flist=*' line to 'flist=". .. the_rest_of_the_filelist"
(i.e the actual expanded file list), then both techiques work, and both
return...
echo . .. the_rest_of_the_filelist | sed s/^/ / ; s/$/ / ; s/ . / / ; s/^ //
; s/ $//
...with the above #echo test commands. Again, I can't _see_ any difference
in what's going on. But, this tells me that expansion of '*' _before_
calling the function is somehow key to what's going on.
Finally, if I set flist=`echo *` (forcing the expansion of '*' before
calling the function), the results are identical to the previous test
(actual expanded file list).
Bottom line... I understand what you're saying about '"$2"' vs. '$2', but I
can't see or figure out the difference between them in this '.' case. On
the up side, the '\$' works fine. :)
As always, thanks for your help.
bd
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of Cameron Simpson
Sent: Sunday, January 07, 2001 6:39 PM
To: [EMAIL PROTECTED]
Cc: Brad Doster
Subject: Re: Whole word substitution with sed
On Sun, Jan 07, 2001 at 05:12:06PM -0800, Brad Doster <[EMAIL PROTECTED]>
wrote:
| zap_word() {
| # Arguments: "word to delete" "word list to delete from"
| echo $2 | sed "s/^/ / ; s/$/ / ; s/ $1 / / ; s/^ // ; s/ $//"
| }
|
| In this case, double quotes are needed because of the $1 usage. I was
| delighted when Matthew suggested the double quotes, as that got it working
| for me. Even better though, is now I understand *why*. :)
Just for the good habit, recode that as:
zap_word() {
# Arguments: "word to delete" "word list to delete from"
echo "$2" | sed "s/^/ / ; s/\$/ / ; s/ $1 / / ; s/^ // ; s/ \$//"
}
That is, make sure $2 is properly quoted. It'll work fine as is until
the day you say:
zap_word "foo bar" # nb: there's 8 chars in that string
Also backslash the $s in your sed which _aren't_ parameter subs.
By adopting these habits you make you code more robust against the
unanticipated needs of the future.
Cheers,
--
Cameron Simpson, DoD#743 [EMAIL PROTECTED] http://www.zip.com.au/~cs/
There are old climbers, and there are bold climbers; but there are no old
bold climbers.
_______________________________________________
Redhat-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/redhat-list
