TCL(Tool Command Language) Scripting :: How to do a NOT in TCL?


	How to do a NOT in TCL? Hi, I have a program that uses tcl7.5 regexp to find the following unix:.* (WARN\|NOTICE) I would like to alter this to something like. unix:.* (WARN\|NOTICE) && ![^core] I have been told that TCL cannot handle logical NOTs, but I thought I'd just check with the experts in this group for confirmation. I tested the second expression above using awk and of course it worked. I have also tried unix:.* (WARN\|NOTICE).[^core]{0} I'm not sure if I have the syntax correct above, but what I'm trying to say is look for unix: then any characters followed by a space, then EITHER WARN or NOTICE, then any characters, but only when there a '0' (zero) occurances of core. Hmm, hope someone understands what I'm trying to do. ;-) Cheers Craig. On May 31, 5:12 pm, cabkiz_fam@hotmail.com wrote: - Hide quoted text - > Hi, > I have a program that uses tcl7.5 regexp to find the following > unix:. (WARN\|NOTICE) > I would like to alter this to something like. > unix:.* (WARN\|NOTICE) && ![^core] > I have been told that TCL cannot handle logical NOTs, but I thought > I'd just > check with the experts in this group for confirmation. > I tested the second expression above using awk and of course it > worked. > I have also tried > unix:.* (WARN\|NOTICE).[^core]{0} > I'm not sure if I have the syntax correct above, but what I'm trying > to say is > look for unix: then any characters followed by a space, then EITHER > WARN or NOTICE, then > any characters, but only when there a '0' (zero) occurances of core. > Hmm, hope someone understands what I'm trying to do. ;-) Logical not is not part of any standard regular expression notation. The regular languages are closed under complementation, which one can see by taking a deterministic machine and swapping the accepting and non-accepting states, but there isn't any straightforward corresponding regular expression notation. However, Tcl does support logical not at the expression level, so you can do things like: [regexp {unix:. (WARN\|NOTICE).} $s] && ![regexp {unix:. (WARN\| NOTICE).core} $s] where the first regular expression is more general and the second one matches a subset of the strings matched by the first. In the specific case you mention, though, a "regular expression" with a negative forward assertion will work: ^unix:. (WARN\|NOTICE(?!core)$ The part (?!core) blocks matches ending in "core". On May 31, 5:12 pm, cabkiz_fam@hotmail.com wrote: - Hide quoted text - > Hi, > I have a program that uses tcl7.5 regexp to find the following > unix:.* (WARN\|NOTICE) > I would like to alter this to something like. > unix:.* (WARN\|NOTICE) && ![^core] > I have been told that TCL cannot handle logical NOTs, but I thought > I'd just > check with the experts in this group for confirmation. > I tested the second expression above using awk and of course it > worked. > I have also tried > unix:.* (WARN\|NOTICE).[^core]{0} > I'm not sure if I have the syntax correct above, but what I'm trying > to say is > look for unix: then any characters followed by a space, then EITHER > WARN or NOTICE, then > any characters, but only when there a '0' (zero) occurances of core. > Hmm, hope someone understands what I'm trying to do. ;-) > Cheers > Craig. Oh, I should say that Tcl 7.5 is very old. I think that it probably doesn't support assertions since the new regular expression package came in with Tcl 8.0. Why on earth are you using such an ancient Tcl? cabkiz_fam@hotmail.com schrieb: > Hi, > I have a program that uses tcl7.5 regexp to find the following > unix:. (WARN\|NOTICE) If you wanted to match either WARN or NOTICE the expression ought to be ((WARN)\|(NOTICE)). Otherwise the only thing it matches is WARNOTICE. > I would like to alter this to something like. > unix:.* (WARN\|NOTICE) && ![^core] I guess you do not mean "any character apart from c,e,o and r", right? Because that's what [^core] means in a regular expression. > I have been told that TCL cannot handle logical NOTs, but I thought > I'd just > check with the experts in this group for confirmation. It can, just not in a single expression. You do if {[regexp {.* ((WARN)\|(NOTICE))} $String] && ![regexp core $String]} { ... } > I tested the second expression above using awk and of course it > worked. Not sure how you managed that. None of the expressions you posted worked for the grep on /my/ machine. > I have also tried > unix:.* (WARN\|NOTICE).[^core]{0} > I'm not sure if I have the syntax correct above, but what I'm trying > to say is > look for unix: then any characters followed by a space, then EITHER > WARN or NOTICE, then > any characters, but only when there a '0' (zero) occurances of core. > Hmm, hope someone understands what I'm trying to do. ;-) Lots of Greetings! Volker -- For email replies, please substitute the obvious. En Fri, 01 Jun 2007 18:27:43 +0200, Volker Hetzer <firstname.lastn@ieee.org> escribi: > (WARN\|NOTICE) > I have a program that uses tcl7.5 regexp to find the following > unix:. (WARN\|NOTICE) > If you wanted to match either WARN or NOTICE the expression > ought to be ((WARN)\|(NOTICE)). Otherwise the only thing it > matches is WARNOTICE. Did you try it before answering? (Documents) 1 % regexp {^(WARN\|NOTICE)$} NOTICE 1 (Documents) 2 % Ramon Rib - Hide quoted text - cabkiz_fam@hotmail.com wrote: > I have a program that uses tcl7.5 regexp to find the following > unix:.* (WARN\|NOTICE) > I would like to alter this to something like. > unix:.* (WARN\|NOTICE) && ![^core] > I have been told that TCL cannot handle logical NOTs, but I > thought I'd just > check with the experts in this group for confirmation. > I tested the second expression above using awk and of course it > worked. > I have also tried > unix:.* (WARN\|NOTICE).[^core]{0} > I'm not sure if I have the syntax correct above, but what I'm > trying to say is > look for unix: then any characters followed by a space, then > EITHER WARN or NOTICE, then > any characters, but only when there a '0' (zero) occurances of > core. I assume you are updating an old script that used to run on Tcl7.5, but you will be using a more recent version of Tcl. In that case you may be able to use the negative lookahead feature for regular expressions. I have to admit I am usually surprised by the results when I try to use this feature, but I think for your case this should work: regexp {unix:. (WARN\|NOTICE)(?!.core)} $str Schelte. -- set Reply-To [string map {nospam schelte} $header(From)] cabkiz_fam@hotmail.com wrote: > I would like to alter this to something like. > unix:. (WARN\|NOTICE) && ![^core] > I have been told that TCL cannot handle logical NOTs, but I thought > I'd just check with the experts in this group for confirmation. It doesn't handle logical negation, since that's a surprisingly non- trivial concept with REs for fairly deep reasons. However, it does provide some techniques that can help. Most important among these is the negative lookahead assertion, which is almost negation. But with some important restrictions, the most important of which are that they cannot have backreferences in (not that that matters this time) and that they are always non-greedy. They're also quite slow (though we mitigate that by putting them at the end of the RE). The RE you are looking for is probably something like this: unix:.\m(WARN\|NOTICE)\M(?!.\mcore\M.$). You'll want to make sure you put that in {braces}, of course. (The \m and \M ensure that you don't get tripped up by words like "scores"...) Donal (alas, no time to explain properly what this does today...)