Hi Brad,
Thanks for your response! I'll study your examples more carefully in the
upcoming days,
The grammar system seems to explicitly "enforcing" me to separate the
parsing logic and the handling logic, which is a switch that is doable for
me... unless both logics get 'inter-mingled'? The latter is the 'risk' that
someone takes by getting used to the /g and /capproach of Perl 5 I suspect
(it enabled me to — at least for my mind — quickly cobble together
convoluted parsing schemes, schemes that I suspect that I am not able to
come up with by expressing a complete singular grammar. whether that 'fear'
is justified is something that I must discover in the upcoming times...)
I also suspect that somehow implementing the functionality of Perl 5's /c —
adding it to the syntax I mean, perhaps as a variant of m:pos like
m:stickypos — is possible, but I have no clue yet how to do that myself.
Enough to explore in the upcoming times, that's for sure! ;)
Thanks,
Regards,
Raymond.
On Sun, 18 Aug 2019 at 17:55, Brad Gilbert <b2gi...@gmail.com> wrote:
> The Perl6 regex system was simplified.
> Instead you use the rest of Perl6 to implement those features.
> Both inside and outside of the regexes.
> (Which means there are fewer esoteric features that are rarely used, and
> often forgotten or never learned.)
>
> It might be best to just store the position at the end of the regex.
>
> use v6
> my $test = " foo bar";
> my $pos = 0;
>
> $test ~~ m /^\s+ {$pos = $/.pos}/;
> $test ~~ m :pos($pos) /foo\s+ {$pos = $/.pos}/;
> $test ~~ m :pos($pos) /willnotmatch {$pos = $/.pos}/;
> $test ~~ m :pos($pos) /bar {$pos = $/.pos}/;
>
> say $/.pos; # yields "13"
>
> If you want something a bit closer:
>
> In here `\G` translates to `:pos($/.pos // $pos)` and `/c` translates to
> `{let $pos = $/.pos}`.
>
> use v6
> my $test = " foo bar";
> my $pos = 0;
>
>
> $test ~~ m /^\s+/;
> $test ~~ m :pos($/.pos // $pos) /foo\s+/;
> $test ~~ m :pos($/.pos // $pos) /{let $pos = $/.pos} willnotmatch/;
> $test ~~ m :pos($/.pos // $pos) /bar/;
>
> say $pos; # yields "13"
>
> You could combine ($/.pos // $pos) into a function:
>
> use v6
> my $test = " foo bar";
> my $pos = 0;
>
> sub pos () { OUTERS::<$/>.pos // $pos } # may defeat optimizations
>
> $test ~~ m /^\s+/;
> $test ~~ m :pos(pos) /foo\s+/;
> $test ~~ m :pos(pos) /{let $pos = $/.pos} willnotmatch/;
> $test ~~ m :pos(pos) /bar/;
>
> say $pos; # yields "13"
>
> Note that in Perl6 regexes are code.
> (They can have parameters and lexical variables.)
>
> Personally I would just combine them into a grammar.
>
> grammar FooBar {
> token TOP {
> $<space> = \s+
> <foo>
> [
> || <willnotmatch>
> || <bar>
> ]
> }
>
> token foo { foo \s+ }
> token willnotmatch { willnotmatch }
> token bar { bar }
> }
>
> The || is to get the Perl5 semantics of trying them in order. It would
> actually be better to use |.
>
> Anyway the result is structured.
>
> say FooBar.parse($test);
> ｢ foo bar｣
> space => ｢ ｣
> foo => ｢foo ｣
> bar => ｢bar｣
>
> You could get the same structure out of a single regex as well.
>
> $test ~~ m {
> ^
> $<space> = \s+
> $<foo> = [foo \s+]
> [
> | $<willnotmatch> = willnotmatch
> | $<bar> = bar
> ]
> }
>
> I'm not sure if you want to include the space(s) after `foo` into
> `$/<foo>`.
> But to keep it the same as the original I used `[]` to group `foo` and
> `\s+`.
>
> On Sun, Aug 18, 2019 at 9:48 AM Raymond Dresens <raymond.dres...@gmail.com>
> wrote:
>
>> Hello,
>>
>> In the past few days I've been converting some "incremental
>> parsing"-regex code from perl 5 to perl 6 (I haven't not touched grammars,
>> yet...)
>>
>> In Perl 5 I often used the /g and /c modifiers so that the following
>> snippet of code doesn't die:
>>
>> # perl5
>>
>> my $test = " foo bar";
>>
>> die unless $test =~ /^\s+/g;
>> die unless $test =~ /\Gfoo\s+/g;
>> die if $test =~ /\Gwillnotmatch/gc; # ...what is the equivalent
>> of the 'c' modifier in Perl 6?
>> die unless $test =~ /\Gbar/g;
>>
>> say pos $test; # yields "13"
>>
>> I managed to translate such code in Perl 6, by using the "m:p/.../"
>> regex,
>>
>> When I anticipate an unsuccessful match then I must temporarily store
>> $/.pos and provide it to the next regex (e.g. "m:p($P)//"), so it seems,
>>
>> That works.... but it riddles my current solutions with "$P = $/.pos;"
>> assignments.
>>
>> Is there a way to retain this match like in Perl 5?
>>
>> Is there a better way in general?
>>
>> ... perhaps it's time to look into grammars? ;)
>>
>> Thanks!
>>
>> Regards,
>>
>> Raymond Dresens.
>>
>>
>>
>>
>>
>>
