RegEx: text not inside

[[e-mail] [sed-users]]

Hello,

I've read a difficult problem:
a RegEx, that matches a string, which doesn't include a 2nd string.
foo-xxx--bar -> no match
foo-xyx--bar -> match
It should match, if "foo.+?bar" doesn't contain "xx", something like
"foo.+(?!xx)bar", but that matches always.

Any idea?

-davidoff12001

[Tim Chase [e-mail] [sed-users]]

Post by ***@yahoo.de [sed-users]
a RegEx, that matches a string, which doesn't include a 2nd string.
foo-xxx--bar -> no match
foo-xyx--bar -> match
It should match, if "foo.+?bar" doesn't contain "xx", something like
"foo.+(?!xx)bar", but that matches always.

You mention wanting to match, but not what you want to do with the
match. Just print it? Extract portions? Find lines that match and
then perform some other action?

You also write to the sed-users mailing list but your syntax at the
bottom looks like a Perlish regex, so I want to confirm that you're
using sed, not some other program.

In a Perlish, you'd do something like

foo(?:(?!xx).)*bar

which works at least here in Python (which uses mostly-PCREs)

-tim

[[e-mail] [sed-users]]

Post by Tim Chase ***@thechases.com [sed-users]
You also write to the sed-users mailing list but your syntax at the
bottom looks like a Perlish regex, so I want to confirm that you're
using sed, not some other program.

You are right, it's a 3rd party program, that uses PCRE. I thought,
they are compatible and asked here, because here write very clever
people.

Post by Tim Chase ***@thechases.com [sed-users]
foo(?:(?!xx).)*bar

And my last thought was right. :-) :-) I don't understand it but this
works, great! Thank you very much!!

-davidoff12001

[Tim Chase [e-mail] [sed-users]]

Post by ***@yahoo.de [sed-users]

Post by Tim Chase ***@thechases.com [sed-users]
foo(?:(?!xx).)*bar

And my last thought was right. :-) :-) I don't understand it but
this works, great!

It translates as

foo # the literal "foo"
(?: # a non-collecting group
(?!xx) # assert that "xx" doesn't match here
. # any single character
) # the end of the non-collecting group
* # that "any character where 'xx' doesn't match"
# zero or more times
bar # the literal "bar"

-tim

[[e-mail] [sed-users]]

Post by Tim Chase ***@thechases.com [sed-users]
(?: # a non-collecting group

Thanks for the explanation. These groups I've used rarely and forgot
them.

Post by Tim Chase ***@thechases.com [sed-users]
. # any single character
) # the end of the non-collecting group
* # that "any character where 'xx' doesn't match"

Nice idea.

-davidoff12001

[Davide Brini [e-mail] [sed-users]]

Post by ***@yahoo.de [sed-users]
Hello,
a RegEx, that matches a string, which doesn't include a 2nd string.
foo-xxx--bar -> no match
foo-xyx--bar -> match
It should match, if "foo.+?bar" doesn't contain "xx", something like
"foo.+(?!xx)bar", but that matches always.

The best you can do in sed is to check that there is a match for (as an
example) foo-...-bar and if so, check that there is no match for whatever
you want to exclude.
ISTR that some there are plans to add PCRE support to GNU sed, so at some
point you may be able to do it with a single regex in sed too.

But for the time being, use perl. Seriously.

[Jim Hill [e-mail] [sed-users]]

That's pushing regex limits, but it's easily within sed's:

/foo.*bar/!d
/foo.*xx.*bar/d


[Non-text portions of this message have been removed]

[Daniel Goldman [e-mail] [sed-users]]

Not quite sure what you want to match (not match). The regex also seems
a bit odd to me. To simplify things, it would help if you would give
concrete test cases, like:


Match:
........
........
........
........


No Match:
........
........
........
........


Then someone can make a test file, propose a solution.


Daniel

Post by ***@yahoo.de [sed-users]
Hello,
a RegEx, that matches a string, which doesn't include a 2nd string.
foo-xxx--bar -> no match
foo-xyx--bar -> match
It should match, if "foo.+?bar" doesn't contain "xx", something like
"foo.+(?!xx)bar", but that matches always.
Any idea?
-davidoff12001
------------------------------------
------------------------------------