François, Bernd, et al,
This might be a fix. Or, it might not. It certainly needs more
testing. See attached. The attached is a patch against version
2.30.10.
Even if it is not the correct fix, it does point us at the region of
code where a fix is likely needed.
Part of the problem is that it is difficult to determine in advance
whether a namespace prefix definition needs to be repeated on
enclosed elements. For example, is it possible that:
- ElementA needs namespace prefix definition "http://aaa";
- ElementA contains ElementB which is in a different namespace.
- ElementB contains ElementC which is in namespace "http://aaa";.
Or, is that pathological? Maybe, but isn't all XML pathological?
Dave
On Tue, Dec 11, 2018 at 12:52:20PM +0000, Zimmermann, Bernd wrote:
> Hello Francois,
>
> I just tried with the --no-namespace-defs option:
>
> # Generated Tue Dec 11 13:35:53 2018 by generateDS.py version 2.30.8.
>
> # Python 3.7.0 (v3.7.0:1bf9cc5093, Jun 27 2018, 04:59:51) [MSC v.1914 64
> bit (AMD64)]
>
> #
> # Command line options:
> # ('--no-namespace-defs', '')
> # ('-o', 'NP2V18Test2.py')
>
> with no other result.
>
> The generated .py files are identical except for the header …
>
> So again, the xml file gets the namespace printed in all tags.
>
> Omitting a defaultns in the generatedsnamespaces.py file results in an
> invalid xml
>
> missing the namespace in the root element.
>
> Exporting with parameter namespacedef_=’xmlns: …’ again results in
> printing
>
> the namespaces in all xml tags.
>
>
>
> So as I wrote in my message before, I think the intention of using
> namespaces is,
>
> to declare them just in the root element and not repeating it in every
>
> xml tag. Multiple namespaces are handled by the namespaceprefix_ value.
>
>
>
> I would still suggest not to pass the namespacedef_ to the children.
>
>
>
> Best regards,
>
> Bernd
>
>
>
> Von: François Guimond <[email protected]>
> Gesendet: Dienstag, 11. Dezember 2018 04:18
> An: Zimmermann, Bernd <[email protected]>
> Cc: [email protected]; [email protected]
> Betreff: Re: [Generateds-users] New export bug in 2.30.8 ?
>
>
>
> Bernd,
>
>
>
> Ok, I got something similar. ( it always happens 5 min after the send
> button is pressed... 😊 )
>
>
>
> use the --no-namespace-defs option when you call generateds.py to avoid
> the problem.
>
>
>
> I will investigate later this week.
>
>
>
> Le lun. 10 déc. 2018 à 21:59, François Guimond <[email protected]> a
> écrit :
>
> Hi Bernd, Dave,
>
>
>
> The problem Bernd describe is not related to namespaceprefix_, but
> namespacedef_ instead.
>
>
>
> The change was introduced in revision #278.
>
>
>
> Now I didn't fully analyzed what is going on, but with the model I am
> working with I am using namespacedef_ of my export function, and I don't
> get the same behavior.
>
>
>
> So Bernd could you try to change your export function call to something
> like this:
>
>
>
> output = StringIO()
> output.write('<?xml version="1.0" encoding="UTF-8"?>\n')
> testxml.export(output, 0, namespaceprefix_='',
> namespacedef_='xmlns="http://uri.etsi.org/02657/v1.18.1#/RetainedData";'
> )
>
>
>
> and tell us if it makes any difference? That should guide the the
> investigation if it's related to GenerateDSNamespaceDefs or not.
>
>
>
> Thanks
>
> François Guimond
>
>
>
> NB as for reverting 2.30.8, I agree your current output is not ideal,
> but I believe your XML is still valid, so let us investigate...
>
>
>
> Le lun. 10 déc. 2018 à 13:48, Zimmermann, Bernd
> <[email protected]> a écrit :
>
> Just a follow up to my last message:
>
> Looking via bitbucket I think, that we are talking about this change:
>
> +Version 2.30.4 (11/06/2018)
> +- A fix from François Guimond for passing namespaceprefix_ to
> + ``self.exportChildren``. Thanks François.
>
>
> https://bitbucket.org/dkuhlman/generateds/commits/f75d7ec327a5e9d839cbb73e7995c30461850e72#chg-generateDS.py
>
> which introduced the passing of namespaceprefix_ to childen.
>
> As I showed below, this results in printing the namespace in every xml
> tag
> which is hard to read and in my opinion unnecessary.
>
> So I would vote for the old version, only keeping the namespace in the
> top element.
>
> Another possibility would be a check if there are default namespaces
> for
> different elements.
>
> Regards,
> Bernd
>
> -----Ursprüngliche Nachricht-----
> Von: Zimmermann, Bernd
> Gesendet: Montag, 10. Dezember 2018 10:39
> An: 'Dave Kuhlman' <[email protected]>; generateds-users
> <[email protected]>
> Betreff: New export bug in 2.30.8 ?
>
> Hello,
>
> just updated from 2.29.24 to 2.30.8 and discovered a maybe bug in the
> export function.
>
> 2.29.24
> was: def export(self, outfile, level, namespaceprefix_='',
> name_='Target', namespacedef_='', pretty_print=True):
>
> 2.30.8
> is: def export(self, outfile, level, namespaceprefix_='',
> namespacedef_='', name_='Target', pretty_print=True):
>
> well Ok, change of the parameter order, but there must be something
> wrong:
>
> test code:
>
> requestID = Natparas2.RequestID('DE','045','1234567')
> dt = datetime.datetime.now().replace(microsecond=0)
> localtime = dt.astimezone(pytz.timezone("Europe/Berlin"))
> timestamp = str(localtime).replace("
> ","").replace("-","").replace(":","")
> cSPID = '49045'
> retainedDataHeader = Natparas2.RetainedDataHeader(requestID,
> cSPID, timestamp)
> requestAcknowledgement = Natparas2.RequestAcknowledgement()
> retainedDataPayload = Natparas2.RetainedDataPayload(None,
> requestAcknowledgement)
> testxml =
> Natparas2.RetainedDataMessage('0.4.0.2.3.0.14',retainedDataHeader,
> retainedDataPayload)
>
> output = StringIO()
> output.write('<?xml version="1.0" encoding="UTF-8"?>\n')
> testxml.export(output, 0, '', 'retainedDataMessage')
>
> From 2.29.24:
> <?xml version="1.0" encoding="UTF-8"?>
> <retainedDataMessage
> xmlns="http://uri.etsi.org/02657/v1.18.1#/RetainedData";>
> <rdHeaderId>0.4.0.2.3.0.14</rdHeaderId>
> <retainedDataHeader>
> <requestID>
> <countryCode>DE</countryCode>
> <authorisedOrganisationID>045</authorisedOrganisationID>
> <requestNumber>1234567</requestNumber>
> </requestID>
> <cSPID>49045</cSPID>
> <timeStamp>20181210101011+0100</timeStamp>
> </retainedDataHeader>
> <retainedDataPayload>
> <requestAcknowledgement/>
> </retainedDataPayload>
> </retainedDataMessage>
>
> as it sould be.
> Note: using default ns for RetainedDataMessage:
>
> GenerateDSNamespaceDefs = {
> "RetainedDataMessage":
> 'xmlns="http://uri.etsi.org/02657/v1.18.1#/RetainedData";',
> }
>
> defined only for the xml tag < RetainedDataMessage> and as the output
> shows, it is only printed in line 2 of the output.
> All other xml tags are without NS.
>
> From 2.30.8:
> <?xml version="1.0" encoding="UTF-8"?>
> <RetainedDataMessage
> xmlns="http://uri.etsi.org/02657/v1.18.1#/RetainedData";>
> <rdHeaderId>0.4.0.2.3.0.14</rdHeaderId>
> <retainedDataHeader
> xmlns="http://uri.etsi.org/02657/v1.18.1#/RetainedData";>
> <requestID
> xmlns="http://uri.etsi.org/02657/v1.18.1#/RetainedData";>
> <countryCode>DE</countryCode>
> <authorisedOrganisationID>045</authorisedOrganisationID>
> <requestNumber>1234567</requestNumber>
> </requestID>
> <cSPID>49045</cSPID>
> <timeStamp>20181210102515+0100</timeStamp>
> </retainedDataHeader>
> <retainedDataPayload
> xmlns="http://uri.etsi.org/02657/v1.18.1#/RetainedData";>
> <requestAcknowledgement
> xmlns="http://uri.etsi.org/02657/v1.18.1#/RetainedData"/>
> </retainedDataPayload>
> </RetainedDataMessage>
>
> Difference:
>
> ALL other xml tags get the defaultNS which is wrong here, because only
> <RetainedDataMessage> shall have it.
> All other xml tags should go without defaultNS.
>
> Code compare shows:
>
> in 2.29.24:
> self.exportChildren(outfile, level + 1, namespaceprefix_='',
> name_='Target', pretty_print=pretty_print)
>
> in 2.30.8:
> self.exportChildren(outfile, level + 1, namespaceprefix_,
> namespacedef_, name_='Target', pretty_print=pretty_print)
>
> here the namespacedef_ is added, but if the children is another xml
> tag with another name like <retainedDataHeader>, and the defaultNS ist
> wrong here because it is not valid and only valid for
> <RetainedDataMessage>.
> So it must be checked in advance if it is valid here or not and if is
> defined for the given tag name.
>
> Please fix it 😉
> Perhaps either roll back to the old behaviour or please implement a
> name check in the children print function to check if the children xml
> tag name hat got a defaultNS defined for its tag name.
>
> Regards,
> Bernd
>
> _______________________________________________
> generateds-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/generateds-users
--
Dave Kuhlman
http://www.davekuhlman.org
diff -r 3dfb7c08d525 generateDS.py
--- a/generateDS.py Fri Nov 30 13:20:23 2018 -0800
+++ b/generateDS.py Tue Dec 11 13:05:06 2018 -0800
@@ -2146,13 +2146,14 @@
"namespaceprefix_, pretty_print=pretty_print)\n" % \
(fill, mappedName)
else:
- namespaceprefix = 'namespaceprefix_'
+ namespaceprefix = "namespaceprefix_"
+ namespacedef = "namespacedef_=''"
if child.prefix and 'ref' in child.attrs:
namespaceprefix += "='%s:'" % child.prefix
+ namespacedef += "=namespacedef_"
s1 = "%s self.%s.export(outfile, level, %s, " \
- "namespacedef_, " \
- "name_='%s', pretty_print=pretty_print)\n" % \
- (fill, mappedName, namespaceprefix, name)
+ "%s, name_='%s', pretty_print=pretty_print)\n" % \
+ (fill, mappedName, namespaceprefix, namespacedef, name)
wrt(s1)
# end generateExportFn_1
_______________________________________________
generateds-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/generateds-users
