[Qgis-user] The XML of getcapabilities should not have any ascii code over 128

[Andreas Neumann]

Hi Andrea/Matthias,

XML can clearly host UTF-8 or UTF-16 - the whole character range - 
depending on what you use in your text editor and specify in your XML 
header.

However, there are some reserved characters, like:

* < (lesser sign)
* > (greater sign)
* " (double quotes)
* & (ampersand)

Did I miss one?

Normally, the XML parser/writer automatically takes care about properly 
escaping/unescaping these characters - if not, I would consider this a 
bug in QGIS. Not only would the OWS outputs be non-wellformed, but also 
QGIS desktop would have a problem with non-wellformed .qgs files.

I never experienced such an issue.

Andrea: can you be more precise with the offending characters and a 
small demo project?

Thanks,
Andreas

On 06.10.2015 22:36, Andrea Peri wrote:
> HI Matthias,
>
> I'm not able to reproduce the issue.
> My try to produce a small project.
> I'm sure why it is not reproducable.
> Perhaps beyond the original issue there was some else trouble.
> Perhaps a bad trasmission ?
> I don't know at now.
>
> So I can only suspect that the issue was another and when I resend the
> new project with my changes, I unconsciously resolve also the real
> issue.
>
> I return on this issue if it will reappear.
>
> Thx for all,
>
> Andrea.
>
>
> 2015-10-06 18:17 GMT+02:00 Matthias Kuhn <matthias at opengis.ch>:
>> Hi Andrea
>>
>> I would suggest to open an issue report with a description of the problem
>> and attach a link to an affected service or attach an affected project file.
>>
>> In the details of the report you can outline the workaround which you are
>> currently taking (i.e. only preserve standard 7-bit ascii characters) as
>> well as a short summary of and a link to this discussion here.
>>
>> Regards,
>> Matthias
>>
>>
>> On 10/04/2015 07:13 PM, Andrea Peri wrote:
>>
>> In the xml response.
>> When it happened I go into the metadata abstracts using the qgis gui, remove
>> any character over 7bit code and after this the response became well formed.
>>
>> Actually we use qgis-server 2.6.
>>
>> A.
>>
>> Il 04/ott/2015 18:31, "Matthias Kuhn" <matthias at opengis.ch> ha scritto:
>>> Hi Andrea,
>>>
>>> On 10/04/2015 05:55 PM, aperi2007 wrote:
>>>
>>> Hi Mattias,
>>>
>>> Sorry but this question is hard to understand for me .
>>> Perhaps I miss to analyze another step:
>>>
>>> If the operating system of the user is with a character set different from
>>> UTF8.
>>>
>>> And the user is editing a qgis project.
>>> The qgis project what character set is using ?
>>> AFAIK the qgis carset is set in the properties page of qgis.
>>>
>>>
>>> Where exactly do you start experiencing strange behavior?
>>> In the text box in the properties, in the project file or only in the XML
>>> response from the server?
>>>
>>> -- Matthias
>>>
>>>
>>> If this is true,
>>> How can the clipboard of the operating system to know what is the charset
>>> set in the properties page of qgis ?
>>>
>>> AFAIK the clipboard change always and only in the charset of the OS.
>>>
>>> A.
>>>
>>> Il 04/10/2015 15:30, Matthias Kuhn ha scritto:
>>>
>>> Hi Andrea,
>>>
>>> Is it really that there should only be 7bit characters or is it that the
>>> encoding is not properly set in the answer?
>>> XML fully supports utf-8 (it is even the default character encoding) and
>>> I assume that in 2015 every modern XML client should support this as well.
>>>
>>> Can you give an example server address and ideas of affected clients?
>>>
>>> I have the feeling that there's something else fishy and if there is
>>> really a need to replace characters above 127 then it should be an
>>> opt-in configuration option.
>>>
>>> Matthias
>>>
>>> On 10/04/2015 11:59 AM, Andrea Peri wrote:
>>>
>>> Hi,
>>>
>>> we experienced this user case problem.
>>>
>>> In the qgis-server getcapabilities response the information of the
>>> "Abstract" is taken from the field:
>>> layer-properties -> metadata -> Abstract.
>>>
>>> But ofet the user fill this field with some cut/paste from other
>>> documents.
>>> So doing, if fill this field with many unacceptable character.
>>> Like these:
>>>
>>> “ ” ’
>>>
>>> or the very hard to see "no blank space".
>>>
>>> The qgis accept all of this without any problem,
>>> but when return the repsonse from a GetCapabiities request.
>>>
>>> The response is not valid for the xml.
>>> So many standard xml clients report a not well formed xml.
>>>
>>> We don't have any good ideas to how resolve this issue.
>>>
>>> My question is if could be possible in the qgis-server site to do a
>>> xml conversion of all this character.
>>> Instead of send them directly.
>>>
>>> AFAIK all of this could be coded to be corected coded in a XML response.
>>>
>>> Any thought on this ?
>>>
>>> Thx.
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Qgis-user mailing list
>>> Qgis-user at lists.osgeo.org
>>> http://lists.osgeo.org/mailman/listinfo/qgis-user
>>>
>>>
>>>
>
>