Zugg Software :: View topic

Posted: Tue Mar 18, 2008 3:02 pm

I am sometimes in the situation where I want to either match a blank line or a line that starts with several possibilities. I've never been able to capture a completely blank line with anything except for:

^$

If I try to combine it with other things such as:

^(?:First|Second)?$

or

^(?:First|Second|$)

It never works for capturing a blank line. Anyone know why?

Posted: Tue Mar 18, 2008 3:09 pm

thats regex right? you got the regex box checked i assume

Posted: Tue Mar 18, 2008 11:03 pm

yes, of course.

Posted: Tue Mar 18, 2008 11:42 pm

Both ^ and $ are anchors. They don't match any characters, they match positions. So (?:First|Second|$) will fail if "First" or "Second" is not found. The $ in the expression is simply ignored (from your POV). You could try ^(?:First|Second|\s+)$. It might do what you want.

Adept Joined: 21 Sep 2005 Posts: 250 Location: Austin, TX

This looks like a bug, and it is present in 2.20 as well, so it should get reported. Zugg has actually done some recent work on the regex code

Anaristos' comment is correct regarding the anchor characters. However, the simple pattern '^ ?$' should match (a) a blank line, or (b) a line with a single space on it, and it does not appear to do so. It will match (b), but not (a), according to my tests.

e.g. try this at the command-line
#REGEX {^$} {#say blank 1}
#REGEX {^ ?$} {#say blank 2}
#show ""
#show " "

Blank 2 should fire on both #shows, but it will only fire on the second.

Posted: Tue Mar 18, 2008 11:59 pm

The problem with ^(?:First|Second)?$ is that it will only succeed if you have "First" or "Second" once in the line starting at the beginning of the line and nothing else, everything else is a failure so the trigger won't fire.
Both ^ and $ are anchors. They don't match any characters, they match positions. So (?:First|Second|$) will fail if "First" or "Second" is not found anywhere in the line. The $ in the expression is simply ignored (from your POV). You could try ^(?:First|Second|\s+)$. It might do what you want.

Posted: Wed Mar 19, 2008 12:08 am

Sorry, JQuilici, I was editing my post. You are correct, of course. (^$) is the standard way to match a blank line. Regarding ^ ?$, it does work for me on tests outside of cMUD. So maybe there is a problem.

Posted: Wed Mar 19, 2008 3:20 am

maybe if you removed the space from between ^ and ?

Adept Joined: 04 Feb 2005 Posts: 246

All your regexps are OK. These should all match a blank line, as you expected:

^$
^(?:First|Second)?$
^(?:First|Second|$)

Testing using perl on UNIX:

Adept Joined: 21 Sep 2005 Posts: 250 Location: Austin, TX

Shalimar, the point is that the '?' character in a regular expression means 'match the previous thing 0 or 1 times'. Or, equivalently, 'match whether or not the previous thing appears'. So, '^ ?$' means 'match <start of line><optional space><end of line>', and should thus match a blank line, or a line with a single space on it. '^?$' would mean something entirely different (and would be almost worthless - it would match any line that had an ending!)

The pattern in the original post ('^(?:First|Second)?$') should have matched a blank line for the same reason. It should also have matched lines that contained, in their entirety, "First" or "Second". I gather that "First" and "Second" are really placeholders for something else...what he's trying to achieve is a pattern that will match one of a handful of things, or a blank line, but nothing else.

Posted: Wed Mar 19, 2008 4:19 am

^(?:First|Second|$) will return true for a null line, but not for a blank line. ^(?:First|Second)?$ will not return true for a blank line, either. $ will always be taken by the regex engine to mean "anchor this to the end of the line".

Posted: Wed Mar 19, 2008 4:36 am

So is this a bug? The absolute only way I can get it to match on a blank line is with:

^$

Nothing else works, and I've tried a lot of stuff including:

^\s*$

Which should work and I can't explain why it doesn't. If it did work I bet the other more complicated ones involving multiple items discussed above would work too.

Posted: Wed Mar 19, 2008 4:40 am

^\s*$ works for me outside of CMUD. I haven't tested it with the client, but it is a perfectly valid way to test for a blank line. Remember, though. ^ and $ are not part of the line, they just tell the engine the starting and ending positions for the search.

SubAdmin Joined: 18 Nov 2001 Posts: 5182

I will do some testing of it and see if I can figure out why it isn't matching. I just have to adjust my test app to give me more information, but I am guessing it is because of the options used.

Adept Joined: 04 Feb 2005 Posts: 246

SubAdmin Joined: 18 Nov 2001 Posts: 5182

Zugg answered in another topic in the Beta forum stating that CMud bypasses the regex engine to match a blank line. He stated that the only pattern supported currently is "^$", and come to think of it I will suggest a small change to him that might make this better. Twisted Evil

Posted: Thu Mar 20, 2008 2:37 am

i thought a pattern of just '$' would also get a blank line. least i remember it used to

Adept Joined: 04 Feb 2005 Posts: 246

No, "$" should match every line, since every line has an end. Just "^" does as well.

Posted: Thu Mar 20, 2008 2:37 pm

then explain why

#GAG $

Doesnt eat up every line :)

I knew it worked!

Posted: Fri Mar 21, 2008 12:01 am

You are simply anchoring the (blank) line to the end.

Adept Joined: 04 Feb 2005 Posts: 246