[GeoNetwork-devel] More (better?) XSD validation errors and XSLT 2.0 support

[Simon P. <sp...@ut...>]

Hi,

I've been doing a bit of investigation into JAXP in GN with a view to 
getting XSLT 2.0 support and improvements to the XSD validation error 
messages - turns out that the current approach where Jeeves throws an 
exception on the first validation error returned from the JAXP validator 
class, loses additional messages that can help provide really useful 
context for some error messages. For example, leaving a gco:Distance 
element blank causes a fairly cryptic error to be returned by the 
current validator in GN/Jeeves:

org.xml.sax.SAXParseException: cvc-datatype-valid.1.2.1: '' is not a 
valid value for 'double'.

Not terribly helpful because it doesn't tell you which element has the 
invalid value. A fairly simple mod is to insert an error handler and 
collect all the errors and then throw an exception afterwards. With this 
in place the validation messages returned from the above become:

ERROR(1) org.xml.sax.SAXParseException: cvc-datatype-valid.1.2.1: '' is 
not a valid value for 'double'.
ERROR(2) org.xml.sax.SAXParseException: cvc-complex-type.2.2: Element 
'gco:Distance' must have no element [children], and the value must be valid.

Ok - still a bit cryptic - but at least we can guess that the element 
that doesn't have a valid value is gco:Distance and that makes it a lot 
easier to find the problem area in the XML view. Possibly useful to is 
that we can also get more validation errors if there are other problems 
in the document, like this one which has a blank gco:Distance and also 
doesn't have a topicCategory code:

ERROR(1) org.xml.sax.SAXParseException: cvc-datatype-valid.1.2.1: '' is 
not a valid value for 'double'.
ERROR(2) org.xml.sax.SAXParseException: cvc-complex-type.2.2: Element 
'gco:Distance' must have no element [children], and the value must be valid.
ERROR(3) org.xml.sax.SAXParseException: cvc-enumeration-valid: Value '' 
is not facet-valid with respect to enumeration '[farming, biota, 
boundaries, climatologyMeteorologyAtmosphere, economy, elevation, 
environment, geoscientificInformation, health, 
imageryBaseMapsEarthCover, intelligenceMilitary, inlandWaters, location, 
oceans, planningCadastre, society, structure, transportation, 
utilitiesCommunication]'. It must be a value from the enumeration.
ERROR(4) org.xml.sax.SAXParseException: cvc-type.3.1.3: The value '' of 
element 'gmd:MD_TopicCategoryCode' is not valid.

and to make things just a little bit nicer we could let the user select 
the number of validation messages they want to see as part of their 
session. Might speed up the process of fixing these problems a bit...

It's still a pretty cryptic interface but at least there is a bit more 
info to help with context and maybe it begins to open the door for 
connecting up these error messages with the XML/advanced editor views 
(It's a shame that these errors return line number and column number of 
-1 - more investigation required - but it might be related to the in 
memory source we're using....)

XSLT 2.0 support - we have a few quite large XSLT convertors being 
written (eg. for GCMD DIF to ISO 19115/19139) that we'd like to use in 
GN. The writers have chosen to use XSLT 2.0 and it does kind of work 
within GN now so long as the writers use xalan constructs, however it 
seems nice to have XSLT 2.0 support (I think?) and most of the converter 
writers seem to be using SAXON, so I've done a trial by switching over 
our GN 2.2 working copy to use SAXON in JAXP - GN and Intermap seem to 
work ok once you convert the xalan functions in a few of the xslts 
(easy) - but is this something people want? And, I'm sure someone has 
probably tried this before, so maybe there are reasons why we stick with 
xalan and XSLT 1.0?

Cheers,
Simon