--- trunk/mbrola.pl	2006/08/05 15:35:19	4
+++ trunk/mbrola.pl	2006/08/08 15:06:33	5
@@ -7,11 +7,7 @@
 use strict;
 use Data::Dump qw/dump/;
 use Term::ReadLine;
-
-my $term = new Term::ReadLine 'Simple Perl calc';
-my $prompt = "Pričaj: ";
-my $OUT = $term->OUT || \*STDOUT;
-
+use File::Slurp;
 
 my $letters = {
 	'a' => [['a'],
@@ -194,7 +190,7 @@
 };
 
 my $recovery;
-foreach my $df (qw/ bp oo uks/) {
+foreach my $df (qw/bp oo uks/) {
 	$recovery->{$df}++;
 }
 
@@ -216,18 +212,34 @@
 
 	my $last_c = '';
 
-	foreach my $i ( 0 .. $#chars ) {
-		my $c = $chars[$i];
+	my $i = 0;
+	while (@chars) {
+		my $c = shift @chars;
 
 		$g .= $c;
 
-		if (defined( $phonemes->{$c} )) {
-			$c = $phonemes->{$c};
-		} elsif (defined( $token_to_grapheme->{$c} )) {
-			$c = $token_to_grapheme->{$c};
+		if (defined( $phonemes->{ lc($c) } )) {
+			$c = $phonemes->{ lc($c) };
+		} elsif (defined( $token_to_grapheme->{ lc($c) } )) {
+			my $tmp = $token_to_grapheme->{ lc($c) };
+			if (length($tmp) > 1) {
+				my @tmp_c = split(//, $tmp);
+				warn "### $c --> $tmp\n";
+				$c = shift @tmp_c;
+				unshift @chars, ( @tmp_c );
+			} else {
+				$c = $tmp;
+			}
 		}
 
-		my $d = $durations->{$c} || $silence->{word};
+		my $d = $durations->{$c} || $durations->{ lc($c) };
+
+		if (! $d) {
+			next if ($c =~ m/\s/);
+			warn "### skipped: $c\n";
+			$c = '_';
+			$d = $silence->{word};
+		}
 
 		my @dur = ( $d );
 
@@ -255,7 +267,7 @@
 			$pho[ $#pho ]->{dur}->[0] += $silence->{word};
 			push @pho, {
 				char => '_',
-				dur => [ $silence->{word} ],
+				dur => [ $silence->{sent} ],
 			};
 			$last_c = $c;
 			next;
@@ -301,11 +313,13 @@
 
 my $mbrola = './bin/mbrola-linux-i386 ./cr1/cr1';
 
-while ( defined ($_ = $term->readline($prompt)) ) {
+my $term = new Term::ReadLine 'Mbrola croatian speaker';
+my $OUT = $term->OUT || \*STDOUT;
 
-	$term->addhistory($_);
+sub play_speak_hr {
+	my $text = shift || return;
 
-	my ($out,$g,$f) = speak_hr( $_ );
+	my ($out,$g,$f) = speak_hr( $text );
 
 	open(my $fh, "| $mbrola - tmp.wav") || die "can't open $mbrola: $!";
 
@@ -318,7 +332,28 @@
 	print $OUT "# $out\n";
 
 	system 'play tmp.wav';
+}
 
+
+if (my $path = shift @ARGV) {
+	my $text = read_file($path) || die "can't read $path: $!";
+
+	# strip html
+	$text =~ s!</?[^>]+>! !gs;
+
+	$text =~ s!\s+! !gs;
+
+	print "-- $text --";
+
+	play_speak_hr( $text );
+	exit;
+}
+
+my $prompt = "Pričaj: ";
+
+while ( defined ($_ = $term->readline($prompt)) ) {
+	$term->addhistory( $_ );
+	play_speak_hr( $_ );
 }