Ampersands should always be escaped, even if they look like entities

[Leonard Richardson]

This test made sense for Beautiful Soup 3, but not 4:

----
        self.assertEqual(
            self.sub.substitute_xml("ÁT&T"),
             "ÁT&T")
----

Depending on how BS3 parsed a document, entities like "Á" might or might not be turned into the corresponding Unicode characters. If you saw "Á" in a document, there was no way to tell whether the original document said "Á" or "&Aacute". So we had code that only turned "&" into "&" if it looked like the "&" was not the beginning of an entity.

In Beautiful Soup 4, entities are always turned into the corresponding Unicode characters. So there's no reason not to turn "&" into "&".

The one wrinkle is entities that aren't HTML entities, like '&foo;'. You could argue that '&foo;' should come out the same way it went in, instead of being turned into "&foo;". But that's not the way it works now, and no one has complained.

[Leonard Richardson (leonardr)]

Fixed in revision 301.