improve utf-8 issue detection

scripts/check_series.sh

@@ -30,6 +30,22 @@ do
       ERRORS=`expr $ERRORS + 1`
       echo $i appears $COUNT times in series
     fi
+#    NUM=3
+    NUM=`grep -n '^diff' $i | head -1 | sed -e 's@\:.*$@@'`
+    if [ "$NUM" = "" ]
+    then
+      cp $i /tmp/cs$$
+    else
+      NUM=`expr $NUM - 1`
+      head -$NUM $i > /tmp/cs$$
+    fi
+    if file /tmp/cs$$ | grep 'ISO-' >/dev/null
+    then
+      ERRORS=`expr $ERRORS + 1`
+      echo non-ascii characters in $i
+      continue
+    fi
+    rm /tmp/cs$$
     if head -1 $i | grep -v '^# HG changeset patch$' >/dev/null
     then
       ERRORS=`expr $ERRORS + 1`