It can be one line using brackets.

Agreed, it's simple enough for that.

Sorry, don't really how gsub works, but isn't this now looking for the combination of the three characters?

Also, comments of the first two didn't make it to the revised one-liner

isn't this now looking for the combination of the three characters?

No, [] and () are special. The [] means any of these characters, but (in this case) only once.

what do you mean by "only once"?

I phrased that badly. I was trying to explain that the pattern itself matches a single character. In this case the gsub function means that pattern (i.e., the single character) will be matched any time it occurs.

Whereas something like [.?][.?] would match all of .?, ?., ?? and ... No more, no less.

No, [] and () are special.

What do () do?

Seeing an example probably explains best. It's mostly used for getting specific parts out of a larger string. The ? means the preceding is optional.

Edit: http://lua-users.org/wiki/PatternsTutorial

I thought you were suggesting there was a problem, unless you are ;)

Also, comments of the first two didn't make it to the revised one-liner

This one is still to be solved :) (and/or the current splitting on 2 lines is odd, first line with a full comment ends with a period. 2nd line tail of the code starts with a lowercase).

Works fine with interpuncts and pipes, i don’t have any dicts with arrows though. But you also sound as if you're suggesting something is wrong, are you?

No no, but with regexps, best to triple check when making changes.
(The pipe in the middle of [·|↑] made it look like an alternative when in parens (·|↑), but if it's ok for @Frenzie , fine.)

oops, can I add this one as well?

text = text:gsub("%d+(%a+)", "%1"):gsub("(%a+)%d+", "%1")

What's it for though? :-D

these pesky little guys: ¹word or word¹ --> word

Is this for number of syllables?
Or "Note 1"?

it removes numbers from either left or right of a selection. that happens when words have superscripts, for example to say meaning 2 of word X, or footnotes (although crengine is pretty good at not getting those, MuPDF is not)

I suppose exceptions like A1 Sauce are rare enough for it to be an improvement most of the time.

text = text:gsub("%d+(%a+)", "%1"):gsub("(%a+)%d+", "%1")

I guess %a may not match Chinese or Arabic letters.
May be %D (not a digit) could be used instead of %a ?
Is it possible to have numbers in dictionary entries ? Couldn't we just remove any digit ? (Just asking, I guess we could have a wikipedictionary with entries for years.
Is it possible also that some dictionary use superscript letters ? ◌ͣfall, ◌ͨfall ?

I guess %a may not match Chinese or Arabic letters. May be %D (not a digit) could be used instead of %a ?

if you think that is safe, ... I don't mind.

Is it possible to have numbers in dictionary entries ?

apparently, but that wouldn't be problematic here, what would be is as @Frenzie mentioned, things like A1 (the sauce and the paper size), 3D, MI6

Couldn't we just remove any digit ? (Just asking, I guess we could have a wikipedictionary with entries for years.

what do you mean by "remove any digit"?

Is it possible also that some dictionary use superscript letters ? ◌ͣfall, ◌ͨfall ?

that seems highly unlikely but I suppose not impossible, I think is safe to assume it does not happen.

I guess %a may not match Chinese or Arabic letters. May be %D (not a digit) could be used instead of %a ?

if you think that is safe, ... I don't mind.

You should mind testing :) (I just gave a thought, no idea if it is safe and working).

what do you mean by "remove any digit"?

text = text:gsub("%d", ""):
But indeed, it would screw up 1080 and search for MI.

I remember that recently, you proposed to lookup an additional transformed word in some case (dunno if this was merged or not). Again, just asking: what if a word contains digits, lookup first a word with the digits removed, then the original word ?

#13884, but i am not sure that is a better approach, suddenly we'd be searching for {MI6, mi6, MI, mi}