posix regex to find A but no B in the same line

Question

I know that perl compatible regex can do this easily by using the "lookaround" technology, but I only have posix regex on my hand which does not support the "lookaround", so my question is that can I find A but not B on a line only using posix regex.

Here is a simple example:

I want to find a "foo" but no "bar" on a line,

this has a foo           # match
i can haz foo            # match
but i haz foo and bar    # NOT match

and here is a similar question

I think the solution can only be tailored for a particular case. — Wiktor Stribiżew, Jan 13 '16 at 23:59
There seems to be this delusion that a sufficiently complex single regex match is somehow more efficient than two simple regex matches — Matt, Jan 14 '16 at 00:00
or in any language something like `yourInput.contains("foo") && !yourInput.contains("bar")` — Jorge Campos, Jan 14 '16 at 00:02
@Matt: Believing that two simple regex matches are more efficient is wrong too. There're no rules, only different situations. — Casimir et Hippolyte, Jan 14 '16 at 00:33
@Casimir: True, but many regex newcomers don't even *think* of using separate regexes until someone suggests it. I know *I* didn't. Matt's comment may have been overly dramatic, but it was worth making. — Alan Moore, Jan 14 '16 at 01:18

Casimir et Hippolyte · Answer 1 · 2016-01-14T00:24:29.260

0

The best way (whatever the language or tool you use) is to test the two patterns (or the two strings) separately. Example with sed:

sed -n '/foo/{/bar/!p}'

It's short, simple and efficient.

To do the same with one ERE pattern, you must write:

^([^b]|b([^a]|$|a([^r]|$)))*foo([^b]|b([^a]|$|a([^r]|$)))*$

Do you think it's a better choice? (and it is a three letters word)

edited Jan 14 '16 at 00:24

answered Jan 14 '16 at 00:13

Casimir et Hippolyte

88,009
5
94
125

posix regex to find A but no B in the same line

1 Answers1