Minor, mostly-cosmetic fixes to filter_scps.pl and split_data.sh (should not affect any recipes)

egs/wsj/s5/utils/filter_scps.pl

@@ -25,9 +25,27 @@
 # the -f <n> switch
 
 
+$field = 1;
+$shifted = 0;
+$print_warnings = 1;
+do {
+  $shifted=0;
+  if ($ARGV[0] eq "-f") {
+    $field = $ARGV[1];
+    shift @ARGV; shift @ARGV;
+    $shifted = 1;
+  }
+  if (@ARGV[0] eq "--no-warn") {
+    $print_warnings = 0;
+    shift @ARGV;
+    $shifted = 1;
+  }
+} while ($shifted);
+
+
 if(@ARGV != 4) {
-  die "Usage: utils/filter_scps.pl  <job-range-specifier> <filter-pattern> <input-scp> <output-scp-pattern>\n" .
-       "e.g.:  utils/filter_scps.pl [-f <field-to-filter-on>] JOB=1:10 data/train/split10/JOB/spk2utt data/train/feats.scp data/train/split10/JOB/feats.scp\n" .
+  die "Usage: utils/filter_scps.pl [-f <field-to-filter-on>] <job-range-specifier> <filter-pattern> <input-scp> <output-scp-pattern>\n" .
+       "e.g.:  utils/filter_scps.pl  JOB=1:10 data/train/split10/JOB/spk2utt data/train/feats.scp data/train/split10/JOB/feats.scp\n" .
        "similar to utils/filter_scp.pl, but it uses multiple filters and output multiple filtered files.\n".
        "The -f option specifies the field in <input-scp> that we filter on (default: 1)." .
        "See also: utils/filter_scp.pl\n";
@@ -45,17 +63,6 @@ if ($ARGV[0] =~ m/^([\w_][\w\d_]*)+=(\d+):(\d+)$/) { # e.g. JOB=1:10
   die "filter_scps.pl: bad job-range specifier $ARGV[0]: expected e.g. JOB=1:10";
 }
 
-$field = 1;
-$shifted = 0;
-do {
-  $shifted=0;
-  if ($ARGV[0] eq "-f") {
-    $field = $ARGV[1];
-    shift @ARGV; shift @ARGV;
-    $shifted=1
-  }
-} while ($shifted);
-
 $idlist = shift @ARGV;
 
 if ($idlist !~ m/$jobname/ &&
@@ -154,9 +161,10 @@ for ($jobid = $jobstart; $jobid <= $jobend; $jobid++) {
   close(FW);
 }
 
-if ($warn_uncovered) {
+if ($warn_uncovered && $print_warnings) {
   print STDERR "filter_scps.pl: warning: some input lines did not get output\n";
 }
-if ($warn_multiply_covered) {
+if ($warn_multiply_covered && $print_warnings) {
   print STDERR "filter_scps.pl: warning: some input lines were output to multiple files\n";
 }
+

egs/wsj/s5/utils/split_data.sh

@@ -32,7 +32,7 @@ fi
 data=$1
 numsplit=$2
 
-if [ $numsplit -le 0 ]; then
+if ! [ "$numsplit" -gt 0 ]; then
   echo "Invalid num-split argument $numsplit";
   exit 1;
 fi
@@ -117,7 +117,10 @@ done
 # split some things that are indexed by speaker
 for f in spk2gender spk2warp cmvn.scp; do
   if [ -f $data/$f ]; then
-    utils/filter_scps.pl JOB=1:$numsplit \
+    ! $split_per_spk && warning_opt="--no-warn"
+    # suppress warnings from filter_scps.pl about 'some input lines were output
+    # to multiple files', which is expected in this case.
+    utils/filter_scps.pl $warning_opt JOB=1:$numsplit \
       $data/split$numsplit/JOB/spk2utt $data/$f $data/split$numsplit/JOB/$f || exit 1;
   fi
 done