--- a/bin/sort_date.py	Thu Aug 24 18:21:41 2023 +0100
+++ b/bin/sort_date.py	Mon Aug 28 21:07:43 2023 +0100
@@ -20,10 +20,24 @@
 
 codecs.register_error('percent',percent_encode)
 
+# From RFC-3986:
+# gen-delims    = ":" / "/" / "?" / "#" / "[" / "]" / "@"
+# sub-delims    = "!" / "$" / "&" / "'" / "(" / ")"
+#                / "*" / "+" / "," / ";" / "="
+# But # _is_ escaped in Java surt results
+#  and additionally " \ : < = > ? \ ^  _ ` { | } are not
+
+# Also, Java surt strips _all_ leading 'www.',
+#  where python3 surt only strips the first one.
+
 def cdx_key(uristring):
-  return quote(unquote(surt(uristring),
-                       errors='percent'),
-               safe='/,:)?=').lower()
+  _surt = quote(unquote(surt(uristring),
+                        errors='percent'),
+                safe='!"$&\'()*+,/:;<=>?@[\\]^_`{|}').lower()
+  while _surt.endswith(",www"):
+    _surt = _surt[:-4]
+
+  return _surt
 
 def keyed(l):
   uri, cc_stamp, dateTime = l.split('\t',2)