Output replaces HTML Entities with unicode literals

[elidickinson]

Running transform seems to translate HTML entities in the source into unicode literals. For example:
<p>&copy; &nbsp;&nbsp; 2014</p>
becomes
<p>©    2014</p>

This is causing issues for me and I'm guessing it's just a side effect of the lxml settings and not intentional. My understanding is that "&copy" has better email client compatibility as "©" (If anything I'd prefer an option to go the other way: escape any unicode literals in the source)

[elidickinson]

OK, so it seems etree.tostring looks at the encoding to decide whether it can output something as a literal or has to escape it. So calling .transform(encoding='ascii') gets me:
<p>&#169; &#160;&#160; 2014</p>
Which is probably close enough for me. It'd be cool if there was a way to preserve the named entities though.

[dandv]

Running Copyright © 2015 iDoRecall, Inc. through premailer.io, I saw © converted to just "c", not the Unicode copyright symbol:

Copyright c 2015 iDoRecall, Inc.

Definitely need an option against that transformation.

[peterbe]

If there's a way to tell lxml to stay the heck out of trying to do that trick that would be great. And something we ought to do by default.

[1951FDG]

Noticed that ’ get converted as well as "

[OrangeDog]

a way to tell lxml

Maybe resolve_entities=False?