#!/lang/perl -n
# this converts the text of Great Expectations into a play, approximately!
# should run pg2line and splitsentences2 first
chomp;
$orig = $_;
if (/^CHAPTER /) {
	# get chapter headings too
	$copy = 3;
	print "\n\n";
	$non_speech = 0;
	$break = 1;
}
if ($copy > 0) {
	print "$_\n";
	--$copy;
} else {
	$line = "";
	while (s/(^|\s|-)'(.*?[^\w\s])'/$1\@/) {
		my $q = $2;
		if (length $q > 12) {
			$q =~ s/,$/./; # not always right, but mostly
		}
		if ($q =~ /,$/) {
			$q .= " ";
		} else {
			$q .= "  ";
		}
		$line .= $q;
	}
	if ($line) {
		if ($break) { print "\n"; $break = 0; }
		$non_speech = 0;
		$line =~ s/ +$//;
		$line =~ s/,$/./;
		# try to work out who the speaker was... hoho!
		$speaker = $_;
		$verb = "";
		$speaker =~ s/^[^@]*[.!?]   *//;  # get rid of any sentences before the speech
		$speaker =~ s/[.!?]  .*//;  # get rid of any sentences at the end
		if ($speaker eq "\@") {
			# FIXME it's probably the last but one who spoke... or perhaps someone who hasn't been introduced yet?
			$speaker = "";
		} elsif ($speaker =~ /^\@/) {
			$speaker =~ s/^\@[\s\W]*//;
			if ($speaker =~ s/^([a-z]+[ei]d) //) {$verb = $1;}
			$speaker =~ s/([^\s\w.'-]|( - )).*//;
		} else {
			$speaker =~ s/[\s\W]*\@.*//;
			if ($speaker =~ s/ ([a-z]+[ei]d)$//) {$verb = $1;}
			$speaker =~ s/.*([^\s\w'-]|( - ))//;
		}
		if ($speaker =~ s/.*?([A-Z])/$1/) {
			$speaker =~ s/ [^A-Z].*//;
		} else {
			$speaker =~ s/.*?(^| )the /the /;
			$speaker =~ s/\b(with|to|at|in (a|an))\b.*//;
		}
		$speaker =~ s/\b[a-z]+ly\b//g;  # no adverbs!
		$speaker =~ s/\bthe\b//g;
		$speaker =~ s/\W+$//;
		$speaker =~ s/^ *//; $speaker =~ s/ *$//; $speaker =~ s/ +/ /g;
		$speaker =~ s/^Jew/the Jew/;  # special treatment for the villain
			# I was going to change to Fagin, but it might be Barney.
			# I think Dickens was a bit anti-semetic.
		if (!$verb && $speaker =~ s/ ([a-z]+[ei]d)$//) {$verb = $1;}
		if ($speaker eq "") { $speaker = "?"; }
		print "$speaker:  $line\n";
	} else {
		# put a break if we've had 700 characters of non-speech text...
		$non_speech+=length($_);
		if ($non_speech >= 700) {
			$break = 1;
			$non_speech = -1000000000; # hehe
		}
	}
}