gen.p6

unit class Text::Generator:ver<0.0.1>:auth<ZzZombo>;

has %!word-chain;
has @!delimeters;
has @!whitespace;

grammar BasicText
{
	token TOP { (<whitespace> || <word> || <symbol>)* }
	token whitespace { \s+ }
	token word { <.alnum>+[<[' & . , -]+:Dash_Punctuation+:Connector_Punctuation> <.alnum>]* <.alnum>* }
	token control { <.cntrl>+ }
	token symbol { \S } #everything else.
	token eof { $ }
}

class BasicText::Actions
{
	method TOP($/)
	{
		my @tokens;
		for $/[0].list -> $m
		{
			my Str:D $type=<whitespace word control symbol eof>.first({$m{$_}:exists});
			@tokens.push: %(value=>~$m{$type},:$type);
		}
		make @tokens.push: %(value=>'',:type('eof'));
		Nil;
	}
}

method feed(Str:D $input,Bool:D :$from-file=False,:$grammar=BasicText,:$actions=BasicText::Actions,:%paired)
{
	%paired ||= %('(' => ')','[' => ']','{' => '}','༺' => '༻','༼' => '༽',
		'᚛' => '᚜','⁅' => '⁆','⁽' => '⁾','₍' => '₎','⌈' => '⌉',
		'⌊' => '⌋','〈' => '〉','❨' => '❩','❪' => '❫','❬' => '❭',
		'❮' => '❯','❰' => '❱','❲' => '❳','❴' => '❵','⟅' => '⟆',
		'⟦' => '⟧','⟨' => '⟩','⟪' => '⟫','⟬' => '⟭','⟮' => '⟯',
		'⦃' => '⦄','⦅' => '⦆','⦇' => '⦈','⦉' => '⦊','⦋' => '⦌',
		'⦍' => '⦐','⦏' => '⦎','⦑' => '⦒','⦓' => '⦔','⦕' => '⦖',
		'⦗' => '⦘','⧘' => '⧙','⧚' => '⧛','⧼' => '⧽','⸢' => '⸣',
		'⸤' => '⸥','⸦' => '⸧','⸨' => '⸩','〈' => '〉','《' => '》',
		'「' => '」','『' => '』','【' => '】','〔' => '〕',
		'〖' => '〗','〘' => '〙','〚' => '〛','﹙' => '﹚',
		'﹛' => '﹜','﹝' => '﹞','（' => '）','［' => '］','｛' => '｝',
		'｟' => '｠','｢' => '｣');
	my @input=$grammar.parse(' '~($from-file ?? $input.IO.slurp !! $input),:$actions).made;
	my @pair-stack;
	my Str ($whitespace,$prefix,$word,$postfix,$prev-word);
	#say "INPUT: ",@input.perl;
	for @input -> $elem
	{
		#say quietly "PROCESSING $elem.perl() ({$prev-word.defined ?? $prev-word !! 'None'})";
		if $elem<type> ~~ <whitespace eof>.any || $word.defined && $elem<type> eq 'word'
		{
			if $prefix.defined || $word.defined
			{
				$word //= '';
				%!word-chain{$word} //= %({$_=>BagHash.new} for <children whitespace prefix postfix>);
				++%!word-chain{$word}<count>;
				++%!word-chain{$word}<whitespace>{$whitespace // ''};
				++%!word-chain{$word}<prefix>{$prefix // ''};
				++%!word-chain{$word}<postfix>{$postfix // ''};
				++%!word-chain{$prev-word}<children>{$word} if $prev-word.defined;
				#say quietly "SAVED $word ({$prev-word.defined ?? $prev-word !! 'None'}): %!word-chain{$word}.perl()";
				$prev-word=$word;
				undefine(($prefix,$word,$postfix));
			}
			else
			{
				$whitespace=$elem<value>;
			}
		}
		elsif $elem<type> eq 'symbol'
		{
			(!$word.defined ?? $prefix !! $postfix) ~= $elem<value>;
		}
		elsif $elem<type> eq 'word'
		{
			$word=$elem<value>;
		}
		#say quietly "INFO: $elem<value>.perl()/$whitespace.defined()|$prefix.defined()|$word.defined()|$postfix.defined()/";
		#say quietly "INFO: $elem<value>.perl()/$whitespace|$prefix|$word|$postfix/";
	}
	#%!word-chain.say;
	Nil;
}
method generate(Num(Cool) $words where * ~~ 0..^Inf=1024,Cool:D :$start='',Cool:D :$end='')
{
	my Str:D @res;
	my $word-bag=BagHash.new-from-pairs: |%!word-chain.map({.key=>.value<count>});
	my Str $word=$word-bag.roll;
	my Str $prev-word;
	my Int:D $word-count=0;
	#for %!chain.pairs
	{
		#say "|$_.key().perl()| ($_.value()<children>.perl()) || ($_.value()<after>.perl())";
	}
	@res.push: $start if $start;
	my @pair-stack;
	repeat
	{
		my $w = %!word-chain{$word}<whitespace>.roll // '';
		#$w.=tc if !$word-count || $prev-word ~~ /<[ . ! ? ]>/;
		$w ~= %!word-chain{$word}<prefix>.roll // '';
		$w ~= $word;
		$w ~= %!word-chain{$word}<postfix>.roll // '';
		@res.push: $w;
		#say "<{$prev-word // 'N0N3'}> <$word> ",%!word-chain{$prev-word}<children>,' ',%!word-chain{$prev-word}<whitespace>;
		$prev-word=$word;
		$word=%!word-chain{$word}<children>.roll;
		if !$word.defined
		{
			#if $word-count > 8 || rand < 0.05
			{
				@res.push: $end if $end;
				@res.push: $start if $start;
				$word-count=0;
			}
			$word=$word-bag.roll;
		}
	}
	until @res.elems>=$words;
	@res.push: $end if $end;
	@res.join.subst(/^\s+/,'');
}

#=====

use Text::Generator;

my $gen=Text::Generator.new;
$gen.feed('test.txt',:from-file);
#$gen.feed('The establishment of dedicated intelligence organizations was directly linked to the colonial rivalries between the major European powers and the accelerating development of military technology.');
$gen.generate(:start(""),:end("\n")).say;