add --parallel to sorting options

1 files changed, 10 insertions, 6 deletions

diff --git a/scripts/training/train-model.perl.missing_bin_dir b/scripts/training/train-model.perl.missing_bin_dir
index f168f38cb..9d04cc1a1 100755
--- a/scripts/training/train-model.perl.missing_bin_dir
+++ b/scripts/training/train-model.perl.missing_bin_dir
@@ -19,7 +19,7 @@ if ($SCRIPTS_ROOTDIR eq '') {
 $SCRIPTS_ROOTDIR =~ s/\/training$//;
 $SCRIPTS_ROOTDIR = $ENV{"SCRIPTS_ROOTDIR"} if defined($ENV{"SCRIPTS_ROOTDIR"});
 
-my($_ROOT_DIR, $_CORPUS_DIR, $_GIZA_E2F, $_GIZA_F2E, $_MODEL_DIR, $_TEMP_DIR, $_SORT_BUFFER_SIZE, $_SORT_BATCH_SIZE, $_SORT_COMPRESS, $_CORPUS,
+my($_ROOT_DIR, $_CORPUS_DIR, $_GIZA_E2F, $_GIZA_F2E, $_MODEL_DIR, $_TEMP_DIR, $_SORT_BUFFER_SIZE, $_SORT_BATCH_SIZE,  $_SORT_COMPRESS, $_SORT_PARALLEL, $_CORPUS,
    $_CORPUS_COMPRESSION, $_FIRST_STEP, $_LAST_STEP, $_F, $_E, $_MAX_PHRASE_LENGTH,
    $_LEXICAL_FILE, $_NO_LEXICAL_WEIGHTING, $_VERBOSE, $_ALIGNMENT,
    $_ALIGNMENT_FILE, $_ALIGNMENT_STEM, @_LM, $_EXTRACT_FILE, $_GIZA_OPTION, $_HELP, $_PARTS,
@@ -58,8 +58,9 @@ $_HELP = 1
 		       'model-dir=s' => \$_MODEL_DIR,
 		       'temp-dir=s' => \$_TEMP_DIR,
            'sort-buffer-size=s' => \$_SORT_BUFFER_SIZE,
-           'sort-batch-size=s' => \$_SORT_BATCH_SIZE,
+           'sort-batch-size=i' => \$_SORT_BATCH_SIZE,
            'sort-compress=s' => \$_SORT_COMPRESS,
+           'sort-parallel=i' => \$_SORT_PARALLEL,
 		       'extract-file=s' => \$_EXTRACT_FILE,
 		       'alignment=s' => \$_ALIGNMENT,
 		       'alignment-file=s' => \$_ALIGNMENT_FILE,
@@ -233,11 +234,14 @@ my $__SORT_BATCH_SIZE = "";
 $__SORT_BATCH_SIZE = "--batch-size $_SORT_BATCH_SIZE" if $_SORT_BATCH_SIZE;
 
 my $__SORT_COMPRESS = "";
-$__SORT_COMPRESS = "--compress-program=$_SORT_COMPRESS" if $_SORT_COMPRESS;
+$__SORT_COMPRESS = "--compress-program $_SORT_COMPRESS" if $_SORT_COMPRESS;
+
+my $__SORT_PARALLEL = "";
+$__SORT_PARALLEL = "--parallel $_SORT_PARALLEL" if $_SORT_PARALLEL;
 
 # supporting scripts/binaries from this package
 my $PHRASE_EXTRACT = "$SCRIPTS_ROOTDIR/training/phrase-extract/extract";
-$PHRASE_EXTRACT = "$SCRIPTS_ROOTDIR/generic/extract-parallel.perl $_CORES $SPLIT_EXEC \"$SORT_EXEC $__SORT_BUFFER_SIZE $__SORT_BATCH_SIZE $__SORT_COMPRESS\" $PHRASE_EXTRACT";
+$PHRASE_EXTRACT = "$SCRIPTS_ROOTDIR/generic/extract-parallel.perl $_CORES $SPLIT_EXEC \"$SORT_EXEC $__SORT_BUFFER_SIZE $__SORT_BATCH_SIZE $__SORT_COMPRESS $__SORT_PARALLEL\" $PHRASE_EXTRACT";
 
 my $RULE_EXTRACT;
 if (defined($_GHKM)) {
@@ -246,7 +250,7 @@ if (defined($_GHKM)) {
 else {
   $RULE_EXTRACT = "$SCRIPTS_ROOTDIR/training/phrase-extract/extract-rules";
 }
-$RULE_EXTRACT = "$SCRIPTS_ROOTDIR/generic/extract-parallel.perl $_CORES $SPLIT_EXEC \"$SORT_EXEC $__SORT_BUFFER_SIZE $__SORT_BATCH_SIZE $__SORT_COMPRESS\" $RULE_EXTRACT";
+$RULE_EXTRACT = "$SCRIPTS_ROOTDIR/generic/extract-parallel.perl $_CORES $SPLIT_EXEC \"$SORT_EXEC $__SORT_BUFFER_SIZE $__SORT_BATCH_SIZE $__SORT_COMPRESS $__SORT_PARALLEL\" $RULE_EXTRACT";
 
 my $LEXICAL_REO_SCORER = "$SCRIPTS_ROOTDIR/training/lexical-reordering/score";
 my $MEMSCORE = "$SCRIPTS_ROOTDIR/training/memscore/memscore";
@@ -255,7 +259,7 @@ my $SYMAL = "$SCRIPTS_ROOTDIR/training/symal/symal";
 my $GIZA2BAL = "$SCRIPTS_ROOTDIR/training/symal/giza2bal.pl";
 
 my $PHRASE_SCORE = "$SCRIPTS_ROOTDIR/training/phrase-extract/score";
-$PHRASE_SCORE = "$SCRIPTS_ROOTDIR/generic/score-parallel.perl $_CORES \"$SORT_EXEC $__SORT_BUFFER_SIZE $__SORT_BATCH_SIZE $__SORT_COMPRESS\" $PHRASE_SCORE";
+$PHRASE_SCORE = "$SCRIPTS_ROOTDIR/generic/score-parallel.perl $_CORES \"$SORT_EXEC $__SORT_BUFFER_SIZE $__SORT_BATCH_SIZE $__SORT_COMPRESS $__SORT_PARALLEL\" $PHRASE_SCORE";
 
 my $PHRASE_CONSOLIDATE = "$SCRIPTS_ROOTDIR/training/phrase-extract/consolidate";