faster consolidating - redirect to stdout then zip

1 files changed, 4 insertions, 1 deletions

diff --git a/scripts/training/train-model.perl.missing_bin_dir b/scripts/training/train-model.perl.missing_bin_dir
index 61e49970f..aac6cef96 100755
--- a/scripts/training/train-model.perl.missing_bin_dir
+++ b/scripts/training/train-model.perl.missing_bin_dir
@@ -1543,7 +1543,7 @@ sub score_phrase_phrase_extract {
     # merging the two halves
     print STDERR "(6.6) consolidating the two halves @ ".`date`;
     return if $___CONTINUE && -e "$ttable_file.gz";
-    my $cmd = "$PHRASE_CONSOLIDATE $ttable_file.half.f2e.gz $ttable_file.half.e2f.gz $ttable_file.gz";
+    my $cmd = "$PHRASE_CONSOLIDATE $ttable_file.half.f2e.gz $ttable_file.half.e2f.gz /dev/stdout";
     $cmd .= " --Hierarchical" if $_HIERARCHICAL;
     $cmd .= " --LogProb" if $LOG_PROB;
     $cmd .= " --NegLogProb" if $NEG_LOG_PROB;
@@ -1552,6 +1552,9 @@ sub score_phrase_phrase_extract {
     $cmd .= " --LowCountFeature" if $LOW_COUNT;
     $cmd .= " --GoodTuring $ttable_file.half.f2e.gz.coc" if $GOOD_TURING;
     $cmd .= " --KneserNey $ttable_file.half.f2e.gz.coc" if $KNESER_NEY;
+    
+    $cmd .= " | gzip -c > $ttable_file.gz";
+    
     safesystem($cmd) or die "ERROR: Consolidating the two phrase table halves failed";
     if (! $debug) { safesystem("rm -f $ttable_file.half.*") or die("ERROR"); }
 }