Mailing-List: contact bloodhound-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: bloodhound-dev@incubator.apache.org
Received-SPF: pass (athena.apache.org: domain of matevzb@gmail.com designates
 209.85.219.47 as permitted sender)
MIME-Version: 1.0
In-Reply-To: 
 <CAGMZAuOaDBHox89GoNjnpQtcXeH6io_nFOMUHc2GmEfzeM6oJw@mail.gmail.com>
References: 
 <CA+DjKZc4rh9Fd6Th=kvVQMD3qpnZ+cyjh=K-EHwjo0zcv_P3TQ@mail.gmail.com>
 <CAJSKAsjgAQMwMNFwO2CNyTR71mLpk0rVmR44OdVcfCa8aBztWg@mail.gmail.com>
 <2BA32F18-E010-433A-8E9F-359192ABA8F8@wandisco.com>
 <CA+DjKZcBwnEJfn5VLdET79WF2buQu1r_xcHVcJp1h=2KFefFgA@mail.gmail.com>
 <CA+DjKZce2r0Ke7BRutgtebAT-sNj=9XMAiPnbmTX8fvAaMDyoQ@mail.gmail.com>
 <CAGMZAuPd49K=QRfb=PbEHLO_rTY=ab85AcuGjnS_ubxx6xKa2A@mail.gmail.com>
 <CAGMZAuPsme2g_WPJ_Rn_kyh1p35T_+Zx5Y2mFnVz70L293Hmcg@mail.gmail.com>
 <CAGMZAuPkL3PKLLY1TiKR22uYK+Jwp4zrECHM8zgUnEDyBXoL-A@mail.gmail.com>
 <CA+DjKZeQEj+x9usbu0iSCcTPUqfJ81xRKBjaMgr9fzsxcEKidw@mail.gmail.com>
 <CAGMZAuOaDBHox89GoNjnpQtcXeH6io_nFOMUHc2GmEfzeM6oJw@mail.gmail.com>
From: Matevz Bradac <matevzb@gmail.com>
Date: Sun, 2 Dec 2012 18:45:14 +0100
Message-ID: 
 <CA+DjKZcMyg0g4bjJ+FVt94i_H49_VoR7NpwtMCxqYid3=VoALg@mail.gmail.com>
Subject: Re: Genshi XHTML + HTML 5
To: bloodhound-dev@incubator.apache.org
Content-Type: text/plain; charset=UTF-8

I tried solving this on the stream level, but no luck so far. First I
implemented a test wrapper for the stream:
class BloodhoundStream(Stream):
    # Stream is using __slots__, so the subclass should as well
    __slots__ = []

    def render(self, method=None, encoding='utf-8', out=None, **kwargs):
        kwargs['doctype'] = DocType.HTML5
        super(BloodhoundStream, self).render(method, encoding, out, **kwargs)

and in the filter stream I changed the stream's type using:
  stream.__class__ = BloodhoundStream

The Stream however defines an __or__ operator (for chaining purposes)
and that one returns back
a Stream object. So I defined and __or__ in the BloodhoundStream as well:
    def __or__(self, function):
        return BloodhoundStream(_ensure(function(self)),
serializer=self.serializer)

This still doesn't work, because Chrome's render_template() uses
Template.generate() for the initial
stream, and that one returns Stream as well. Since this is the first
Stream in line, all the subsequent
BloodhoundStream objects also become Stream (Stream | BloodhoundStream
| BloodhoundStream
produces a Stream in the end).

Due to the above complications (and those not yet discovered), I'd say
it would be easier and less error-prone
to just override the genshi.output.get_serializer() with BH's wrapper.
The function is pretty trivial and unlikely
to change much, so it looks like a safe alternative.

--
matevz

On Sat, Dec 1, 2012 at 6:13 PM, Olemis Lang <olemis@gmail.com> wrote:
> On 12/1/12, Matevz Bradac <matevzb@gmail.com> wrote:
>> Thanks for the suggestion, but unfortunately it won't work on the
>> instance level. =/
>
> :'(
>
>> The Stream class is using __slots__, so most instance attributes are
>> read-only, e.g. this:
>>     import types
>>     stream.render = types.MethodType(new_render, stream)
>> doesn't work - AttributeError exception.
>>
>
> I see .
>
>> Brane suggested monkeypatching on a class/module level, so maybe
>> genshi.core.Stream.serialize(),
>> or even genshi.output.get_serializer() would be good candidates.
>>
>
> that's risky . I'd really suggest to try to solve this at stream level
> since monkey patching at module level may be a serious matter .
>
> FWIW ... most of the time I advocate for wrapping rather than monkey
> patching , because it's easier to deal with the impact and scope of
> the changes .
>
> --
> Regards,
>
> Olemis.
>
> Blog ES: http://simelo-es.blogspot.com/
> Blog EN: http://simelo-en.blogspot.com/
>
> Featured article: