ElementTree, XML and Unicode -- C0 Controls

[Sébastien Boisgérault]

Hi all,

The unicode code points in the 0000-001F range --
except newline, tab, carriage return -- are not legal
XML 1.0 characters.

Attempts to serialize and deserialize such strings
with ElementTree will fail:

>>elt = Element("root", char=u"\u0000")
xml = tostring(elt)
xml

'<root char="\x00" />'

>>fromstring(xml)

[...]
xml.parsers.expat.ExpatError: not well-formed (invalid token): line 1,
column 12

Good ! But I was expecting a failure *earlier*, in
the "tostring" function -- I basically assumed that
ElementTree would refuse to generate a XML
fragment that is not well-formed.

Could anyone comment on the rationale behind
the current behavior ? Is it a performance issue,
the search for non-valid unicode code points being
too expensive ?

Cheers,

SB

[Fredrik Lundh]

Sébastien Boisgérault wrote:

Could anyone comment on the rationale behind
the current behavior ? Is it a performance issue,
the search for non-valid unicode code points being
too expensive ?

the default serializer doesn't do any validation or well-formedness checks at all; it assumes
that you know what you're doing.

</F>

[Sébastien Boisgérault]

On Dec 11, 4:51 pm, "Fredrik Lundh" <[e-mail]:

Sébastien Boisgérault wrote:

Could anyone comment on the rationale behind
the current behavior ? Is it a performance issue,
the search for non-valid unicode code points being
too expensive ?

the default serializer doesn't do any validation or well-formedness checks at all; it assumes
that you know what you're doing.

</F>

Fair enough !

Thanks Fredrik.

SB