BBEdit-Talk Digest #1722
1) Re: Grep Help
by David Kurtz <[EMAIL PROTECTED]>
2) Re: Grep help
by Alan Storm <[EMAIL PROTECTED]>
3) Re: Grep help
by Ronald J Kimball <[EMAIL PROTECTED]>
4) Re: Grep help
by Charlie Garrison <[EMAIL PROTECTED]>
5) Re: Grep Help .. works with Replace and Find again
by Alan Storm <[EMAIL PROTECTED]>
6) Re: Grep help
by Seth Dillingham <[EMAIL PROTECTED]>
7) Re: Grep help
by Ronald J Kimball <[EMAIL PROTECTED]>
8) Re: Grep help
by Joao Carlos de Pinho <[EMAIL PROTECTED]>
9) Re: Grep help
by Joao Carlos de Pinho <[EMAIL PROTECTED]>
10) Re: Grep help
by Seth Dillingham <[EMAIL PROTECTED]>
11) Re: Grep help
by Ronald J Kimball <[EMAIL PROTECTED]>
12) Re: Grep help
by "Jared Earle" <[EMAIL PROTECTED]>
From: David Kurtz <[EMAIL PROTECTED]>
Date: August 8, 2006 2:12:12 PM EDT
Subject: Re: Grep Help
On Aug 8, 2006, at 9:09 AM, Ted Burger <[EMAIL PROTECTED]> wrote:
I have a LONG list of words:
order_id
order_num
category
listingsize
boldtype
I want to change it to:
fieldname="order_id" postargumentname="order_id"
[...]
I tried:
GREP is checked
Search: (.*)
Replace: fieldname="&" postargumentname="&"
But I get:
fieldname="order_id" postargumentname="order_id"fieldname=""
postargumentname=""
(.*) will match the end of line. In other words, it will match
anything, including nothing. That's why you're seeing what appears to
be a doubling up of keys with empty values.
What you want is either of the following:
(.+) # this matches *at least* one character
^(.*)$ # or here, grep knows you don't want the NULL
From: Alan Storm <[EMAIL PROTECTED]>
Date: August 8, 2006 2:12:14 PM EDT
Subject: Re: Grep help
On Tue, 8 Aug 2006, Ted Burger wrote:
Search: (.*)
Replace: fieldname="&" postargumentname="&"
But I get:
fieldname="order_id" postargumentname="order_id"fieldname=""
postargumentname=""
I'm not sure why your search is failing with replace all, but if you
add
the starts with/ends with meta characters to your regular
expressions it fixes things up.
Search: (^.*$)
--
Alan Storm
http://alanstorm.com/
From: Ronald J Kimball <[EMAIL PROTECTED]>
Date: August 8, 2006 2:38:37 PM EDT
Subject: Re: Grep help
For the record, this is not a bug. Here's what happens:
Text
example
Find
(.*)
The regex engine starts out trying to match at the beginning of the
string,
just before the first e. .* matches the e, then the x, and so on
until it
has matched 'example'. . can't match newline, and as that's the end of
the regex, a successful match has been found.
Then the regex engine tries to match from where it left off, between
the
second e and the newline. .* matches zero characters, then can't
match the
newline. As that's the end of the regex, another successful match has
been
found.
Then the regex engine tries to match from where it left off, still
between
the second e and the newline. .* matches zero characters, then can't
match
the newline. That would be a successful match, but the engine has
already
found a zero-length match at this position. To avoid an infinite
loop, the
regex engine skips this match, advances one character, and tries again.
Ronald
From: Charlie Garrison <[EMAIL PROTECTED]>
Date: August 8, 2006 2:38:44 PM EDT
Subject: Re: Grep help
Good morning,
On 8/8/06 at 12:09 PM -0400, Ted Burger <[EMAIL PROTECTED]> wrote:
I am getting doubles of the 'Replace'.
Any thoughts?
Instead of using this replace pattern:
Replace: fieldname="&" postargumentname="&"
Use this:
fieldname="\1" postargumentname="\1"
Charlie
--
Charlie Garrison <[EMAIL PROTECTED]>
PO Box 141, Windsor, NSW 2756, Australia
From: Alan Storm <[EMAIL PROTECTED]>
Date: August 8, 2006 2:39:04 PM EDT
Subject: Re: Grep Help .. works with Replace and Find again
On Tue, 8 Aug 2006, David Kurtz wrote:
(.*) will match the end of line. In other words, it will match
anything, including nothing. That's why you're seeing what appears to
be a doubling up of keys with empty values.
I thought the same thing at first. However, if that's the case, why
doesn't the repeating happen if you go through each line individually
using the Replace & Find again command?
--
Alan Storm
http://alanstorm.com/
From: Seth Dillingham <[EMAIL PROTECTED]>
Date: August 8, 2006 3:24:58 PM EDT
Subject: Re: Grep help
On 8/8/06, Ronald J Kimball said:
For the record, this is not a bug. Here's what happens:
Text
example
Find
(.*)
The regex engine starts out trying to match at the beginning of the
string,
just before the first e. .* matches the e, then the x, and so on
until it
has matched 'example'. . can't match newline, and as that's the end
of
the regex, a successful match has been found.
[snip]
I'm not trying to make a big issue out of this — obviously, it's up to
BareBones to say if this is actually a bug or not — but just because
you can
explain what's happening doesn't mean it's not a bug.
I think it's a bug.
.* is supposed to match everything except newlines (unless you use the
flag
that says it matches newlines also).
- Regular expressions are greedy by default, so the first thing
that can
be matched after a greedy, no-line-feed .* is the next line.
- If you've matched everything except the carriage return, it
doesn't
make any sense to then match the "nothing" between the last
character
in the line and the carriage return.
- The "everything" you matched already was supposed to include that
"nothing" after the last character.
Here's a perl example that shows how it should work.
Create a test file called test.txt. Put a few lines in it, just one
word per
line (to keep it short and simple). Save it at ~/Desktop/test.txt
Open terminal, and run these two lines:
cd ~/Desktop
perl -pi -e 's/(.*)/($1)/' test.txt
The second line tells perl to run a search for every .*, and replacing
each
with the same text but wrapped in parentheses.
Open the file in BBEdit again, and you'll see that each line is
wrapped in
parens, and that's the only change.
So if a line contains just "foo", it will be replaced with "(foo)".
Now change the file back to the way it was (without the parens), and
run the
same find-and-replace in BBEdit:
- turn on grep
- search for (.*)
- replace with (\1)
and click "replace all". You get an extra set of empty parens at the
end of
every line.
This is a bug. (Never would have believed it if I hadn't seen it for
myself! ;-)
(Not that this is a big deal, since there seems to be a pretty easy
workaround.)
Seth
From: Ronald J Kimball <[EMAIL PROTECTED]>
Date: August 8, 2006 4:21:36 PM EDT
Subject: Re: Grep help
On Tue, Aug 08, 2006 at 03:24:58PM -0400, Seth Dillingham wrote:
Here's a perl example that shows how it should work.
Create a test file called test.txt. Put a few lines in it, just one
word per
line (to keep it short and simple). Save it at ~/Desktop/test.txt
Open terminal, and run these two lines:
cd ~/Desktop
perl -pi -e 's/(.*)/($1)/' test.txt
That's only doing a single replacement per line, because you left off
the
/g modifier.
Try this instead:
perl -pi -e 's/(.*)/($1)/g' test.txt
It's not a bug.
Ronald
From: Joao Carlos de Pinho <[EMAIL PROTECTED]>
Date: August 8, 2006 4:22:01 PM EDT
Subject: Re: Grep help
Em 8/8/06 3:38 PM, "Ronald J Kimball" <[EMAIL PROTECTED]>
escreveu:
For the record, this is not a bug.
To answer the original poster, I ran a test in Affrus using that
pattern
over a dummy text, and the results were exactly what I expected, ie, no
duplicate pattern at the end. (And Seth Dillingham achieved the same
results
using Perl in Terminal.)
That's why I said "It seems that you found a bug".
But I'm not a grep expert. Far from it.
João Carlos de Pinho
São Paulo, Brazil
From: Joao Carlos de Pinho <[EMAIL PROTECTED]>
Date: August 8, 2006 5:16:02 PM EDT
Subject: Re: Grep help
Em 8/8/06 5:21 PM, "Ronald J Kimball" <[EMAIL PROTECTED]>
escreveu:
That's only doing a single replacement per line, because you left off
the
/g modifier.
Yes, your're right: it's not a bug.
Believe it or not, when I made the test I even considered to use the /g
modifier, but I eventually said to myself: "For what? In this case, it
would
not make any difference". Well, it makes all the difference. Lesson
learned.
Thanks for the explanation, Ronald.
João Carlos de Pinho
São Paulo, Brazil
From: Seth Dillingham <[EMAIL PROTECTED]>
Date: August 8, 2006 5:16:18 PM EDT
Subject: Re: Grep help
On 8/8/06, Ronald J Kimball said:
perl -pi -e 's/(.*)/($1)/g' test.txt
OK, that one's even weirder than what BBEdit produces. ;-)
Seth
From: Ronald J Kimball <[EMAIL PROTECTED]>
Date: August 9, 2006 9:21:03 AM EDT
Subject: Re: Grep help
On Tue, Aug 08, 2006 at 05:16:18PM -0400, Seth Dillingham wrote:
On 8/8/06, Ronald J Kimball said:
perl -pi -e 's/(.*)/($1)/g' test.txt
OK, that one's even weirder than what BBEdit produces. ;-)
Hehe! That's true.
Given this input:
foo
bar
It produces this output:
(foo)()
()(bar)()
()
This happens because it's reading one line at a time. The regex
matches
the word, then the empty string before the newline, then the empty
string
*after* the newline.
If you have it read in the whole input as one string you'll get the
same behavior that occurs in BBedit:
perl -pi -e 'BEGIN { $/ = undef } s/(.*)/($1)/g' test.txt
Ronald
From: "Jared Earle" <[EMAIL PROTECTED]>
Date: August 9, 2006 11:21:24 AM EDT
Subject: Re: Grep help
On 8/8/06, Ted Burger <[EMAIL PROTECTED]> wrote:
Folks,
I appear to be missing something.
I have a LONG list of words:
order_id
order_num
category
listingsize
boldtype
Does BBEdit do Awk? :D
awk '{print "fieldname="\"$0"\" postargumentname=\""$0"\""}'
I'll grep my coat.
--
Jared Earle :: There is no SPORK
[EMAIL PROTECTED] :: http://www.23x.net
The Spodcast :: http://spodcast.org
--
------------------------------------------------------------------
Please send bug reports to <[EMAIL PROTECTED]>, not to the list.
List FAQ: <http://www.barebones.com/support/lists/bbedit_talk.shtml>
List archives: <http://www.listsearch.com/BBEditTalk.lasso>
To unsubscribe, send mail to: <[EMAIL PROTECTED]>
