Re: [PyWrapper-devel] Valid response model

[Roger H. <ro...@hy...>]

Thanks Markus,

I'll have a look at it over the next few days - still have shopping =20
to do myself.

I was presuming that the schemas would have to be modified in some =20
way to produce usable output models. I'll get the code and have a =20
hack at the schemas to see what I can get going.

Many thanks and Merry Christmas,

Roger


On 18 Dec 2006, at 16:04, Markus D=F6ring wrote:

> Roger,
> I've updated pywrapper in the SVN trunk. Your example =20
> test1_broken.xsd works fine now. I had to change some methods in =20
> the structure_handler.py and I hope I didnt break anything. I dont =20
> think so, but as you might know we are still missing comfortable =20
> unit tests, so all I can do is run through some schemas/models I =20
> know they should work (like abcd and tcs). and that test is =20
> successfully acomplished.
>
> But if I run your previous complex real world example I still got =20
> an error. I found that you are using a recursive construct =20
> TaxonName that can be a link or another TaxonName. This is not =20
> supported by pywrapper at the moment. When encountering such =20
> constructs the software should give out a warning and ignore the =20
> recursion (just work with it as if there was no further reference). =20=

> But right now it raises an error and refuses to work with that =20
> model. I still need some christmas presents so I cant continue on =20
> that, maybe something for the coding fest - together with the yet =20
> unsopported "group" tag you mentioned earlier (shouldnt be too hard).
>
> Anyway you should be able to continue with your RDF test. Just =20
> avoid recursions and also the xml schema "group" tag.
> And please keep us informed about the results!
>
> Markus
>
>
>
>
> On Dec 15, 2006, at 5:32 PM, Roger Hyam wrote:
>
>> Thanks for the kind words Markus. I'm glad some one is glad I am =20
>> trying to do this ;) I hope it doesn't end up with too much work =20
>> for you.
>>
>> I have narrowed down this problem to a simple test case.
>>
>> This is a more simple XSD with a single import and it works fine:
>>
>> http://localhost:8080/pywrapper/pywrapper?=20
>> op=3Dschema&location=3Dhttp://rs.tdwg.org/ontology/voc/xsd/test1.xsd
>>
>> The second version is just the same but includes the use of an =20
>> attribute reference in the imported file:
>>
>> http://localhost:8080/pywrapper/pywrapper?=20
>> op=3Dschema&location=3Dhttp://rs.tdwg.org/ontology/voc/xsd/=20
>> test1_broken.xsd
>>
>> and it doesn't work!
>>
>> Looking at the logs
>>
>> The error message is
>>
>> "PyWrapperError: Global element reference 'http://rs.tdwg.org/=20
>> ontology/voc/Test2#/rdf:about' missing"
>>
>> But we have a slightly earlier log message saying
>>
>> 2006-12-15 16:14:14,789 - DEBUG from pywrapper.model: =20
>> self._globalRefs=3D[u'http://www.w3.org/1999/02/22-rdf-syntax-ns#/=20
>> @about', u'http://rs.tdwg.org/ontology/voc/Test2#/MyElement', =20
>> u'http://www.w3.org/1999/02/22-rdf-syntax-ns#/RDF', u'http://=20
>> www.w3.org/1999/02/22-rdf-syntax-ns#/MyRdfString']
>>
>> So it looks to me like it is just getting the namespace of the =20
>> attribute reference wrong and so not finding it in the globalRefs =20
>> look up. It does it right for the element but not for the attribute.
>>
>> Looking at the code for _buildAttribute(self, attrs) on line 269 =20
>> in structure_handler.py  it  doesn't mention namespace (unlike the =20=

>> _buildElement attribute) and it confuses me as it seems to pass =20
>> the type as the namespace to the Attribute object it creates.
>>
>> Now this code all looks like clever stuff and I don't want to go =20
>> trying to fix anything when I don't really know what I am doing - =20
>> though I will have a go if no one else wants to. You may be able =20
>> to fix this in minutes rather than my hours and would probably =20
>> create fewer bugs .
>>
>> It may be that I am doing something bad in the XSD although I am =20
>> pretty confident I am OK on that.
>>
>> Would once again be grateful for you comments,
>>
>> Roger
>>
>>
>>
>>
>> On 14 Dec 2006, at 10:48, Markus D=F6ring wrote:
>>
>>> Hi Roger,
>>> that's exactly what we were all thinking of doing for some time =20
>>> already. Good finally someone DOES it!
>>> The xml schema parser inside pywrapper is a custom =20
>>> implementation. I couldnt find any existing schema parser out =20
>>> there, honestly. So as you can imagine this is a pretty tough job =20=

>>> and I have tried it with many different schemas and progressed in =20=

>>> implementing xml schema, but still dont support all of it (for =20
>>> example references and patterns).
>>>
>>> Well, and your RDF schema is honestly the most complex I've seen =20
>>> that was fed to pywrapper.
>>> So I tried to trace the problem and as you see pywrapper =20
>>> complaints that the attribute  http://rs.tdwg.org/ontology/voc/=20
>>> TaxonName#/rdf:about  is being referenced, but is nowhere defined.
>>> pywrapper expands all namespace prefixes into the full namespace. =20=

>>> So in this case http://rs.tdwg.org/ontology/voc/TaxonName# is the =20=

>>> targetnamespace of the imported lib/TaxonName.xsd schema file. So =20=

>>> in there I suspect is a reference to rdf:about, which pywrapper =20
>>> thinks is an attribute name. I think this is because I cant find =20
>>> anywhere the prefix rdf: being defined properly. I suspect oXygen =20=

>>> and alike know what rdf: means. So try to add a namespace =20
>>> declaration to the schemas (for owl etc too). Then pywrapper =20
>>> should expand rdf:about into the full qualified name http://=20
>>> www.w3.org/1999/02/22-rdf-syntax-ns#about and hopefully find the =20
>>> global declaration of rdf:about in the lib/rdf.xsd schema.
>>>
>>> But again, I think I've never tried a schema with that many =20
>>> namespaces and imports so I keep my fingers crossed...
>>>
>>> Please let me know what happens!
>>> Markus
>>>
>>>
>>>
>>> On Dec 13, 2006, at 11:29 PM, Roger Hyam wrote:
>>>
>>>> Hi All,
>>>>
>>>> I am looking at what TAPIR output models I could use with =20
>>>> PyWrapper. I have developed XML Schemas that will help produce =20
>>>> valid RDF instance data for a simple OWL ontology. Here is an =20
>>>> example:
>>>>
>>>> http://rs.tdwg.org/ontology/voc/xsd/TaxonName/schema.xsd
>>>>
>>>> If you open this link in OxygenXML or Spy you can expand all the =20=

>>>> nodes and create instance documents that look like valid RDF. =20
>>>> Here is an example instance document:
>>>>
>>>> http://rs.tdwg.org/ontology/voc/xsd/TaxonName/template.xml
>>>>
>>>> So the plan would be to use the schema above (or something =20
>>>> similar) as the structure in an output model so that a TAPIR =20
>>>> response could be a valid semantic web document and a LSID =20
>>>> metadata response. A complete LSID authority would need very =20
>>>> little code on top of this.
>>>>
>>>> The downside is that these schemas are complex because of the =20
>>>> number of imports required to support the different namespaces =20
>>>> involved.
>>>>
>>>> Anyhow. The first step was to check that PyWrapper would parse =20
>>>> schemas like the one above so I did this:
>>>>
>>>> http://localhost:8080/pywrapper/pywrapper?=20
>>>> op=3Dschema&location=3Dhttp://rs.tdwg.org/ontology/voc/xsd/TaxonName/=
=20
>>>> schema.xsd
>>>>
>>>> on my local machine and got
>>>>
>>>> "The models response structure is invalid: Global element =20
>>>> reference 'http://rs.tdwg.org/ontology/voc/TaxonName#/rdf:about' =20=

>>>> missing"
>>>>
>>>> I am not sure why I get this as OxygenXML doesn't have a problem =20=

>>>> with expanding the links and validating the rdf:about attribute =20
>>>> so I *think* the schema is OK. The parser is handling the import =20=

>>>> of TaxonName.xsd OK. Is it a problem with attributes? Is it a =20
>>>> bug or a known limitation or an error in my schema?
>>>>
>>>> The are also problems with the group element but I can always =20
>>>> work round this by just repeating the element.
>>>>
>>>> Any ideas on the rdf:about problem would be most welcome.
>>>>
>>>> Many thanks,
>>>>
>>>> Roger
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> ****************************************
>>>>   Roger Hyam <ro...@td...>
>>>>   Biodiversity Information Standards
>>>> ****************************************
>>>>
>>>>
>>>> -------------------------------------------------------------------=20=

>>>> ------
>>>> Take Surveys. Earn Cash. Influence the Future of IT
>>>> Join SourceForge.net's Techsay panel and you'll get the chance =20
>>>> to share your
>>>> opinions on IT & business topics through brief surveys - and =20
>>>> earn cash
>>>> http://www.techsay.com/default.php?=20
>>>> page=3Djoin.php&p=3Dsourceforge&CID=3DDEVDEV_________________________=
____=20
>>>> __________________
>>>> PyWrapper-devel mailing list
>>>> PyW...@li...
>>>> https://lists.sourceforge.net/lists/listinfo/pywrapper-devel
>>>
>>
>> ---------------------------------------------------------------------=20=

>> ----
>> Take Surveys. Earn Cash. Influence the Future of IT
>> Join SourceForge.net's Techsay panel and you'll get the chance to =20
>> share your
>> opinions on IT & business topics through brief surveys - and earn =20
>> cash
>> http://www.techsay.com/default.php?=20
>> page=3Djoin.php&p=3Dsourceforge&CID=3DDEVDEV___________________________=
____=20
>> ________________
>> PyWrapper-devel mailing list
>> PyW...@li...
>> https://lists.sourceforge.net/lists/listinfo/pywrapper-devel
>