trunk: Fixed inconsistency in validating reco2file_and_channel by...

trunk: Fixed inconsistency in validating reco2file_and_channel by validate_data_dir.sh with vs without segments

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@5013 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

egs/wsj/s5/utils/validate_data_dir.sh

@@ -188,8 +188,19 @@ if [ -f $data/wav.scp ]; then
       # this file is needed only for ctm scoring; it's indexed by recording-id.
       check_sorted_and_uniq $data/reco2file_and_channel
       ! cat $data/reco2file_and_channel | \
-        awk '{if (NF != 3 || ($3 != "A" && $3 != "B")) { print "Bad line ", $0; exit 1; }}' && \
-        echo "$0: badly formatted reco2file_and_channel file" && exit 1;
+        awk '{if (NF != 3 || ($3 != "A" && $3 != "B" )) { 
+                if ( NF == 3 && $3 == "1" ) {
+                  warning_issued = 1;
+                } else {
+                  print "Bad line ", $0; exit 1; 
+                }
+              }
+            } 
+            END {
+              if (warning_issued == 1) {
+                print "The channel should be marked as A or B, not 1! You should change it ASAP! "
+              }
+            }' && echo "$0: badly formatted reco2file_and_channel file" && exit 1;
       cat $data/reco2file_and_channel | awk '{print $1}' > $tmpdir/utts.r2fc
       if ! cmp -s $tmpdir/utts{,.r2fc}; then
         echo "$0: Error: in $data, utterance-ids extracted from segments and reco2file_and_channel"