gmlas.d.11: Validate geometries (2) fails for Building 3D GML

[JohannaOtt]

I tested an INSPIRE Buildings 3D GML dataset and got this error on gmlas.d.11: "System error in the Executable Test Suite. Please contact a system administrator. Error information:
[err:XPTY0004] Cannot return item()* as node(): ((db:open-pre("etf-tdb-89ce6cd7-....
(67/33)"
The link to the failed test is: http://inspire-sandbox.jrc.ec.europa.eu/etf-webapp//v2/TestRuns/EID0416b1cb-e2d0-4f62-ba48-31fc83890b55.html?lang=en#EIDd414305c-a7be-4ffe-947d-3ea845a4276d
In case you need the dataset to investigate, please send me an email address where I can send the GML file to.

[robsgnao]

Hello,
thanks for sharing the link, will have a look at it asap; current deployment at JRC has a limitation on the size of uploaded files, I have the impression from the error code that this may be the reason.
Can you please share with us the GML file, or the link to it?
Thanks.

[JohannaOtt]

Hello,
the test is executed when using a smaller GML file (containing 3 buildings, I sent it to [email protected] and hope it is going to reach you this way) - thank you for the advice.

Unfortunately the validator now returns an error concerning the srsDimension on the LinearRings ("XML document 'testExport_small.gml', Building 'Building_DETHL56P00003jxG': The feature geometry is not a valid GML geometry: java.lang.IllegalArgumentException: Points of LinearRing do not form a closed linestring
XML document 'testExport_small.gml', Building 'Building_DETHL56P00003k6U': The feature geometry is not a valid GML geometry: Error in XML document (line: 6, column: 248, character offset: 1300): Cannot parse 'gml:posList': contains 15 values, but coordinate dimension is 2. This does not match.
XML document 'testExport_small.gml', Building 'Building_DENW19AL0000dlkx': The feature geometry is not a valid GML geometry: Error in XML document (line: 4, column: 1641, character offset: 2646): Cannot parse 'gml:posList': contains 111 values, but coordinate dimension is 2. This does not match.").

In the GML file, there is a srsDimension of 3 at the level of the MultiSurface that contains the concerned LinearRings, but this information seems not to be taken into account when validating the LinearRings. Is that expected?

[cportele]

@JohannaOtt - to make the small test file available, either attach it to a comment (just drag an drop the file to the comment) or send it to a non-github email (e.g. [email protected]). Thanks!

[cportele]

Thanks for the sample document. A general issue is that the srsName attribute states that the coordinate reference system is ETRS89 lat/long (a Geographic 2D CRS) while the (not normative) srsDimension attribute says the dimension is 3. The coordinates in the geometries mostly seem to be 3D (but see below). I.e., the CRS declaration seems to be incorrect.

On the three reported errors:

XML document 'testExport_small.gml', Building 'Building_DETHL56P00003jxG': The feature geometry is not a valid GML geometry: java.lang.IllegalArgumentException: Points of LinearRing do not form a closed linestring

This error is not clear to me. It may be the case that the underlying geometry library is not able to handle 3D rings.

XML document 'testExport_small.gml', Building 'Building_DETHL56P00003k6U': The feature geometry is not a valid GML geometry: Error in XML document (line: 6, column: 248, character offset: 1300): Cannot parse 'gml:posList': contains 15 values, but coordinate dimension is 2. This does not match.

Since the CRS is declared as a 2D CRS this message is correct.

However, even if I replace the CRS declaration and use http://www.opengis.net/def/crs/EPSG/0/7409 (which may be the correct CRS assuming the elevation is in EVRF), the error message is the same. This may be related to #51 and https://github.com/etf-validator/etf-gmlgeox/issues/19. I will check this.

XML document 'testExport_small.gml', Building 'Building_DENW19AL0000dlkx': The feature geometry is not a valid GML geometry: Error in XML document (line: 4, column: 1641, character offset: 2646): Cannot parse 'gml:posList': contains 111 values, but coordinate dimension is 2. This does not match.

Some of the geometries contain strange position lists that seem to be neither 2D or 3D coordinates: 51.69294046740865 8.879992259177829 4.434772336235498 4.499922260151334 0.0015476999193630699 51.39410408540394 51.69293247009222 8.879997213831695 4.434774835456967 4.499922415729554 0.0015475999384063058 .... Maybe an issue in the generation of the coordinates?

[jonherrmann]

@robsgnao could you please check if the file 'testquery.xq' changed in commit 643d300 is up to date?

[robsgnao]

Yes, 'testquery.xq' is up to date on JRC deployment.

[cportele]

We have created two issues in the GmlGeoX repository to address this:

• One issue is that the CRS definition file in the deegree library does not contain all allowed CRSs in INSPIRE (e.g. 7409 is not included). This would fix the issue reported here including the first message about the unclosed ring (once the GML has been corrected). See https://github.com/etf-validator/etf-gmlgeox/issues/21
• Another issue is that GmlGeoX silently ignores CRS that are not supported (and as a fallback assumes a 2D CRS). An unknown CRS should be reported. See https://github.com/etf-validator/etf-gmlgeox/issues/22

[JohannaOtt]

@cportele thanks for investigating and creating the new tickets - and also for the hint concerning the geometries with the strange position lists. I will have a look into how they are generated.

[MalcolmCoull]

Hi everybody
Not sure if it's allowed to add to this rather than create a new ticket (and I appreciate that this is still flagged as 'Open') but I've been revisiting some INSPIRE conversions this week and have been getting the same/similar error to @JohannaOtt. I'm working with the Soils Theme, trying to convert from OSGB using FME. I have a workbench which reprojects the data, does some attributed management, converts polygons to multisurfaces and writes out to GML.
About a week ago I 'successfully' wrote out a file which passed all of the validations at https://inspire.ec.europa.eu/validator/ but which was in completely the wrong place. After adjusting a few parameters it is now correctly located but when I try and validate I get a combination of ''gml:posList': contains X values, but coordinate dimension is 2. This does not match.' and The feature geometry is not a valid GML geometry: java.lang.IllegalArgumentException: Points of LinearRing do not form a closed linestring'. I have GeometryValidators (for those who use FME) after every stage of the process and none of them report any errors during translation, and I have tried every combination of projection options available in FME and they all give an output with these issues.