locale/diagnostics.sh emits CSV format; detects even one omission

locale/diagnostics.sh

@@ -5,22 +5,25 @@
 # How many messages in total?
 total=`grep '^msgid' audacity.pot | wc -l`
 
+#CSV header
+echo "Name,Completed,Pct. completed,Bad lines"
+
+declare -i badlines
+
 for filename in `ls *.po`; do
     # If there are errors from msgcmp, then the last line on standard error
     # contains a count of problematic messages; else it won't match the
     # pattern in awk, so assume no errors
     errors=`msgcmp $filename audacity.pot 2>&1 | \
-	awk '/msgcmp: found [0-9]* fatal errors/ { nn = $3 } END {print 0+nn}'`
+	awk '/msgcmp: found [0-9]* fatal error/ { nn = $3 } END {print 0+nn}'`
     complete=$((total-errors))
-    # A few spaces after the filename makes the columns mostly aligned
-    echo "${filename}   " $'\t' $complete $'\t' "$((complete*100/total)) %"
-done | sort -n -k3
 
-echo
+    # detect whether this sequence occurs in any .po file.  It will break
+    # msgfmt on Windows.
+    badlines=`fgrep '#~|' ${filename} | wc -l`
 
-# detect whether this sequence occurs in any .po file.  It will break
+    # produce comma-separated values
-# msgfmt on Windows.
+    echo "$filename,$complete,$((complete*100/total)),$badlines"
-echo "Files with illegal comment sequence:"
+done | sort -n -t , -k3
-fgrep -l '#~|' *.po
 
 exit 0