Fixes case switching keybinds gu and gU so they dont chew up newlines

[Nomarian]

Fixes #1238 #1320

| tr does the appropriate behavior but gawk supports other languages. so gawk is used

This makes it unportable, but if portability is desired, then | tr should be used and if which gawk returns true then gawk is used.

[rnpnr]

I don't think you need something so complicated:

awk '{print tolower($0)}'

should do the trick. You don't need to mess with the field separator or force gawk.

Personally I think we should just axe this whole passing off text editing functionality to external tools but that's a different story.

[Nomarian]

It is necessary because print adds a newline. (so selections are not usable) So you must use printf or ORS="". But then you have a problem with RS="", which is that awk has this weird thing that eats away the last newline before EOF. This is an all AWKs thing

Now that I think about it, gawk can use \0 in RS, so if its that.

[rnpnr]

Frankly I think it's a problem with the way vis handles the result. But we can't really fix here without breaking every other use of vis_pipe (including from the user perspective).

Like I said, this whole passing off text editing tasks to external programs needs to be thrown away. It's not bloat for a text editor to perform text edits.

[Nomarian]

No no, its fine, its awk, awk is the problem. Both sed and tr work perfectly well to transform text, take a bunch of text, transform it, print it out, its a simple gsub() call.

; echo this | tr -s '[:lower:]' '[:upper:]' | wc
      1       1       5

; echo -n this | tr -s '[:lower:]' '[:upper:]' | wc
      0       1       4

; echo -n this | sed 's/this/that/' | wc
      0       1       4

; echo this | sed 's/this/that/' | wc
      1       1       5

But awk does not work this way, awk reads a chunk of text until it hits the the record separator (RS), and it throws away the separator itself, you don't know if there is a RS, which means this is tricky (and wrong)

; echo this | awk 'BEGIN {ORS=""} {print toupper($0)}' | wc
      0       1       4
; echo -n this | awk 'BEGIN {ORS=""} {print toupper($0)}' | wc
      0       1       4

gawk is the only one that tells if there is an RS by saving it in RT (record terminator), but this is gawk specific.

; echo -n this | gawk 'BEGIN {ORS=""} {print toupper($0) RT}' | wc
      0       1       4
; echo this | gawk 'BEGIN {ORS=""} {print toupper($0) RT}' | wc
      1       1       5

if RS is multiple characters, it can be a regex, which means this is also possible

; echo -n '^$hi^$' | awk 'BEGIN {RS="^$";ORS=""} {print toupper($0) RT}' | wc
      0       1       6
; echo '^$hi^$' | awk 'BEGIN {RS="^$";ORS=""} {print toupper($0) RT}' | wc
      1       1       7

This is by far the best, most portable solution, there is one awk that does not support RS to be a regex (goawk), but that's also not the default anywhere.

I will keep using this last one for a while and see if there's a pitfall.