1 use Getopt::Long; 2 3 $rc = GetOptions("pht=s" => \$phtfile, 4 "ok=s" => \$okfile, 5 "i=s" => \$okfile, 6 "otxt=s" => \$otxt, 7 "o=s" => \$otxt, 8 "showerrs" => \$showerrs); 9 10 if(defined $phtfile) { 11 load_phtfile( $phtfile); 12 } 13 14 sub load_phtfile 15 { 16 my $phtfile = shift(@_); 17 $lphhash{"&"}++; 18 $lph_for_sph{"&"} = "&"; 19 open(PHT, "<$phtfile") || die "error opening phtfile $phtfile\n"; 20 print STDERR "using phtfile $phtfile\n"; 21 <PHT>; # header 22 while(<PHT>) { 23 s/\s+$//g; 24 ($trash,$lph,$sph,$num_states) = split(/\s+/); 25 $lph_for_sph{$sph} = $lph; 26 $lphhash{$lph}++; 27 } 28 close(PHT); 29 } 30 31 open(HH, "<$okfile") || die "error opening okfile $okfile\n"; 32 open(OO, ">$otxt") || die "error opening output dict $otxt\n"; 33 while(<HH>) { 34 s/\s+$//; 35 if(/^LANG\s*=\s*(\S+)/) { # LANG = EN-US 36 my $language = lc($1); 37 my $language_header_line = $_; 38 $language =~ s/\-/\./g; 39 if(!defined $phtfile) { 40 die "Error: ESRSDK is not defined\n" if(!defined $ENV{ESRSDK}); 41 $phtfile = "$ENV{ESRSDK}/config/$language/models/generic.pht"; 42 load_phtfile( $phtfile); 43 } 44 print OO "$language_header_line\n"; 45 next; 46 } 47 ($word, $pron) = split(/\s+/); 48 @sphlist = split(/ */, $pron); 49 @lphlist = (); 50 $nerrs = 0; 51 foreach $sph (@sphlist) { 52 $lph = $lph_for_sph{$sph}; 53 if(!defined $lph) { 54 warn "error: unknown sph $sph in $word $pron\n" ; 55 $lph = "($sph)"; 56 $nerrs++; 57 } 58 push(@lphlist, $lph); 59 } 60 next if($nerrs && !$showerrs) ; 61 print OO "$word \t @lphlist\n"; 62 } 63 close(HH); 64 close(OO); 65
