#50 regsplit primitive

[Natasha O'Brien]

There should be a "regsplit" or "regexplode" primitive
that splits a string on matches of a regular
expression, like Perl's "split" function. e.g.:

"potassium" "[aeiou]+" regsplit

would return:

{"p", "t", "ss", "m"}

[Revar Desmera]

Logged In: YES

user_id=6331

regsplit[ str:str str:pat -- list:matches ]
"\r#&$%^\r" var! marker
str @ pat @ marker @ REG_ALL
marker @ explode_array
;

( or, arguably more pedanticly correct: )

regsplit[ str:str str:pat -- list:matches ]
pat @ "^\(.*\)%s\(.*\)$" fmtstring pat !
{ }list var! out
begin
str @ pat @ 0 regexp if
dup 1 [] out @ []<- out !
2 [] dup strlen str @ strlen = if
1 strcut swap pop
then
str !
else
str @ out @ []<- exit
then
repeat
;

[Revar Desmera]

Logged In: YES
user_id=6331

Is possible the marker string in my previous comment should be
different, so as not to have any &s or stuff that regsub may
expand. I just wrote that off the top of my head. Idea is to use a
merker string that is highly unlikely to occur in the string you
expect to split.

[Kyle Hamilton]

Logged In: YES
user_id=7718

Still, it'd be quicker to implement it in the server.

Natty, do you consider this one done enough? Or should it
stay open and in the feature requests?

[Wyld]

Added REGSPLIT and REGSPLIT_NOEMPTY in FB7.