Make xmltest.sh script work for UTF-16 builds

expat/tests/udiffer.py

@@ -0,0 +1,26 @@
+#! /usr/bin/env python3
+
+import sys
+import difflib
+
+if len(sys.argv) != 3:
+    sys.exit(2)
+
+try:
+    with open(sys.argv[1]) as infile1:
+        first = infile1.readlines()
+except UnicodeDecodeError:
+    with open(sys.argv[1], encoding="utf_16") as infile1:
+        first = infile1.readlines()
+try:
+    with open(sys.argv[2]) as infile2:
+        second = infile2.readlines()
+except UnicodeDecodeError:
+    with open(sys.argv[2], encoding="utf_16") as infile2:
+        second = infile2.readlines()
+
+diffs = list(difflib.unified_diff(first, second,
+                                  fromfile=sys.argv[1], tofile=sys.argv[2]))
+if diffs:
+    sys.stdout.writelines(diffs)
+    sys.exit(1)

expat/tests/xmltest.sh

@@ -31,6 +31,8 @@ TS="$MYDIR"
 # OUTPUT must terminate with the directory separator.
 OUTPUT="$TS/out/"
 # OUTPUT=/home/tmp/xml-testsuite-out/
+# Unicode-aware diff utility
+DIFF="$TS/udiffer.py"
 
 
 # RunXmlwfNotWF file reldir
@@ -57,7 +59,7 @@ RunXmlwfWF() {
   read outdata < outfile 
   if test "$outdata" = "" ; then 
       if [ -f "out/$file" ] ; then 
-          diff -u "$OUTPUT$reldir$file" "out/$file" > outfile 
+          $DIFF "$OUTPUT$reldir$file" "out/$file" > outfile 
           if [ -s outfile ] ; then 
               cp outfile "$OUTPUT$reldir$file.diff"
               echo "Output differs: $reldir$file"