regexp help: lookahead and lookbehind with spaces

patrickdrd

@alan-kilborn thanks but I would like f1fantasy matched as well

Alan Kilborn

@patrickdrd said in regexp help: lookahead and lookbehind with spaces:

but I would like f1fantasy matched as well

Hmm, the target moves…

I think you can figure out how to change my regex do that, given your first guess at it. :-)

patrickdrd

@alan-kilborn \s* is the 0 or more spaces but I can’t figure out where to attach it to, I tried all places and it doesn’t work

I also got a couple more conditions to check for like mls and ucl

patrickdrd

ok got it,

(?:(f1|ucl|mls)|(fantasy))\s*(?(1)(?2)|(?1))

guy038

Hello, @patrickdrd, @alan-kilborn and All,

Actually, two solutions are possibles :

Regex A : SEARCH (?-i:(f1|ucl|mls)|(fantasy))\s*(?(1)(?2)|(?1))

Regex B : SEARCH (?-i:(f1|ucl|mls)|(fantasy))\x20*(?(1)(?2)|(?1))

It does not match exactly the same occurrences !

Paste the text below in a new tab :

f1fantasy
f1 fantasy
f1      fantasy
uclfantasy
ucl fantasy
ucl            fantasy
mlsfantasy
mls fantasy
mls    fantasy

fantasyf1
fantasy f1
fantasy                f1
fantasyucl
fantasy ucl
fantasy          ucl
fantasymls
fantasy mls
fantasy                       mls
============================================================
Match with \x{000D}\x{000A} (CRLF) in between :
f1
fantasy

Match with \x{000A} (LF) in between :
f1
fantasy

Match with \x{000D} (CR) in between :
f1
fantasy  

Match with \x{0009} (TABULATION) in between :
fantasy	f1

Match with \x{0011} ( VERTICAL TABULATION ) in between :
f1fantasy

Match with \x{0085} ( NO-BREAK SPACE ) in between :
fantasy f1

with a Multi-lignes MIX of these SPECIFIC chars in between :
fantasy			
		            f1

Open the Mark dialog ( Ctrl + M )
Tick only the Purge for each search and Wrap around options
Click on the Mark All button

As you can see, the regex A, in addition to match usual space chars, also matches a lot a “SPACE” characters !

=> After the line of equal signs, some matches, possibly multi lines, occurred !

So the regex B, more rigorous, just matches zero or more \x20 chars ;-))

=> The part, after the equal signs is not detected at all !

If your different words may be present in any case, simply replace the -i in-line modifier, within the non-capturing group, by the i modifier !

Best regards,

guy038

patrickdrd

@guy038 I prefere the first approach,
because I want it to match tabs as well,
also \n shouldn’t be a problem because I’m matching single line items

Neil Schipper

@patrickdrd and @guy038,

In much the same way that Goldilocks found Papa Bear’s bed too hard, and Mama Bear’s bed too soft, but Baby Bear’s bed just right …

in the present context, surely \s is too promiscuous, and \x20 is too brittle, while \h is just right!

guy038

Hi, @patrickdrd, @alan-kilborn, @neil-schipper and All,

@neil-schipper is perfectly right ! This third solution (?-i:(f1|ucl|mls)|(fantasy))\h*(?(1)(?2)|(?1)) is just what you need as \h* will match any combination, possibly null, of Space, Tabulation and/or No-Break Space char(s) !

BR

guy038

patrickdrd

@neil-schipper I still tend to think that \s is slightly better because why not match “accidental” carriage returns/ line feeds?

which is higher? the possibility of an “accidental” carriage return/ line feed or a “next line mismatch”?
I don’t know if you understand, but I think the first is higher

Neil Schipper

@patrickdrd said in regexp help: lookahead and lookbehind with spaces:

why not match “accidental” carriage returns/ line feeds?

I don’t know enough about your data or the intended purpose of the pair matching to say either way, but I will say that, since you are gathering up either ab or ba pairs, then, a small error can make every subsequent pair a different pair from what would have matched without the error.

It’s up to you to think through whether this harms what you’re trying to accomplish, and if it does, to try to devise a strategy that could detect a malformed pair, and maybe skip it and then resync and pick up all subsequent pairs with desired grouping.