From b01eeb03ebd34a737f698b647976f8f6dc1b3775 Mon Sep 17 00:00:00 2001
From: redpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Thu, 28 Oct 2010 22:19:32 +0000
Subject: small fixes

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@700 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 word-aligner/support/extract_vocab.pl                  | 5 ++++-
 word-aligner/support/generate_per_sentence_grammars.pl | 1 +
 word-aligner/support/make_lex_grammar.pl               | 9 ++++++---
 3 files changed, 11 insertions(+), 4 deletions(-)

(limited to 'word-aligner')

diff --git a/word-aligner/support/extract_vocab.pl b/word-aligner/support/extract_vocab.pl
index 070d4202..ef94b7d6 100755
--- a/word-aligner/support/extract_vocab.pl
+++ b/word-aligner/support/extract_vocab.pl
@@ -7,7 +7,10 @@ my $wc = 0;
 while(<>) {
   chomp;
   my @words = split /\s+/;
-  for my $word (@words) { $wc++; $dict{$word}++; }
+  for my $word (@words) {
+    die if $word eq '';
+    $wc++; $dict{$word}++;
+  }
 }
 
 my $tc = 0;
diff --git a/word-aligner/support/generate_per_sentence_grammars.pl b/word-aligner/support/generate_per_sentence_grammars.pl
index 8779ac9c..6a02bb0a 100755
--- a/word-aligner/support/generate_per_sentence_grammars.pl
+++ b/word-aligner/support/generate_per_sentence_grammars.pl
@@ -14,6 +14,7 @@ open FILT,"|gzip -c > $ARGV[3]" or die "Can't write $ARGV[3]: $!";
 open PSG,">$ARGV[4]" or die "Can't write $ARGV[4]: $!";
 open OTRAIN,">$ARGV[5]" or die "Can't write $ARGV[5]: $!";
 
+binmode OTRAIN, ":utf8";
 binmode FILT, ":utf8";
 binmode PSG, ":utf8";
 binmode STDOUT, ":utf8";
diff --git a/word-aligner/support/make_lex_grammar.pl b/word-aligner/support/make_lex_grammar.pl
index 3e243125..e4cbf7ba 100755
--- a/word-aligner/support/make_lex_grammar.pl
+++ b/word-aligner/support/make_lex_grammar.pl
@@ -35,7 +35,7 @@ my %sizes = ();
 while(<M1>) {
   chomp;
   my ($f, $e, $lp) = split /\s+/;
-  $model1{$f}->{$e} = 1;
+  $model1{$f}->{$e} = 1e-12 + exp($lp);
   $sizes{$f}++;
 }
 close M1;
@@ -50,7 +50,7 @@ while(<IM1>) {
   $invm1{$e}->{$f} = 1;
   $esizes{$e}++;
   if (($sizes{$f} or 0) < $LIMIT_SIZE && !(defined $model1{$f}->{$e})) {
-    $model1{$f}->{$e} = 1;
+    $model1{$f}->{$e} = 1e-12;
     $sizes{$f}++;
     $inv_add++;
   }
@@ -66,7 +66,7 @@ while(<M1>) {
   chomp;
   my ($f, $e, $lp) = split /\s+/;
   if (($esizes{$e} or 0) < $LIMIT_SIZE && !(defined $invm1{$e}->{$f})) {
-    $invm1{$e}->{$f} = 1;
+    $invm1{$e}->{$f} = 1e-12;
     $esizes{$e}++;
     $dir_add++;
   }
@@ -106,6 +106,7 @@ my $ADD_111 = 1;
 my $ADD_ID = 1;
 my $ADD_PUNC = 1;
 my $ADD_NULL = 1;
+my $ADD_MODEL1 = 1;
 my $ADD_STEM_ID = 0;
 my $ADD_SYM = 0;
 my $BEAM_RATIO = 50;
@@ -184,6 +185,8 @@ for my $f (sort keys %fdict) {
     my $total_eandf = $ecounts{$e} + $fcounts{$f};
     my $dice = 2 * $efcount / $total_eandf;
     my @feats;
+    if (defined $m1 && $ADD_MODEL1) { push @feats, "Model1=$m1"; my $m1d = $m1 * $dice; push @feats, "M1Dice=$m1d"; }
+    if ($ADD_MODEL1 && !defined $m1) { push @feats, "NoModel1=1"; }
     if ($ADD_FIDENT && $efcount > $MIN_FEATURE_COUNT) {
       $fc++;
       push @feats, "F$fc=1";
-- 
cgit v1.2.3