Unexpected geometryType for a feature with geometry

[maelle]

👋 @eblondel we're also getting new failures in tests that are related to $getGeometryType() now returning NULL for a feature that has a geometry. We use $getGeometryType() to guess whether a feature will be sf.

Is this a bug or new expected behavior? In the latter case, what would you recommend we use instead?

wfs <- suppressWarnings(
    ows4R::WFSClient$new(
        "https://ows.emodnet-seabedhabitats.eu/geoserver/emodnet_open_maplibrary/wfs",
        serviceVersion = "2.0.0",
        headers = c("User-Agent" = "emodnet.wfs"),
        logger = NULL
    )
)
#> Loading ISO 19139 XML schemas...
#> Loading ISO 19115 codelists...
capabilities <- wfs$getCapabilities()
layer_names <- purrr::map_chr(capabilities$getFeatureTypes(), ~ .x$getName())
types <- capabilities$getFeatureTypes()
type <- types[layer_names == "emodnet_open_maplibrary:dk003069"] |> unlist()
type[[1]]$getGeometryType()
#> NULL
type[[1]]$getBoundingBox()
#>         min      max
#> x  9.575308 10.24418
#> y 54.773777 55.12132
type[[1]]$hasGeometry()
#> [1] FALSE
layer <- wfs$getFeatures("emodnet_open_maplibrary:dk003069")
layer
#> Simple feature collection with 82 features and 8 fields
#> Geometry type: MULTISURFACE
#> Dimension:     XY
#> Bounding box:  xmin: 1065918 ymin: 7318084 xmax: 1140377 ymax: 7385447
#> Projected CRS: WGS 84 / Pseudo-Mercator
#> First 10 features:
#>            gml_id   gid      gui polygon annexi         subtype confidence
#> 1  dk003069.39844 39844 DK003069      61   1170 Geogenic origin       High
#> 2  dk003069.39855 39855 DK003069      72   1110            <NA>       High
#> 3  dk003069.39860 39860 DK003069      77   1110            <NA>       High
#> 4  dk003069.39861 39861 DK003069      78   1110            <NA>       High
#> 5  dk003069.39785 39785 DK003069       2   1170 Geogenic origin       High
#> 6  dk003069.39814 39814 DK003069      31   1170 Geogenic origin       High
#> 7  dk003069.39791 39791 DK003069       8   1170 Geogenic origin       High
#> 8  dk003069.39836 39836 DK003069      53   1170 Geogenic origin       High
#> 9  dk003069.39829 39829 DK003069      46   1170 Geogenic origin       High
#> 10 dk003069.39850 39850 DK003069      67   1110            <NA>       High
#>    val_comm                           geom
#> 1      <NA> MULTISURFACE (POLYGON ((111...
#> 2      <NA> MULTISURFACE (POLYGON ((108...
#> 3      <NA> MULTISURFACE (POLYGON ((111...
#> 4      <NA> MULTISURFACE (POLYGON ((108...
#> 5      <NA> MULTISURFACE (POLYGON ((111...
#> 6      <NA> MULTISURFACE (POLYGON ((111...
#> 7      <NA> MULTISURFACE (POLYGON ((109...
#> 8      <NA> MULTISURFACE (POLYGON ((108...
#> 9      <NA> MULTISURFACE (POLYGON ((108...
#> 10     <NA> MULTISURFACE (POLYGON ((111...

^{Created on 2025-01-24 with reprex v2.1.1}

@salvafern

[eblondel]

@maelle You are interrogating the capabilities fo the WFS and list the feature types (modeled ad WFSFeatureType), but when you list them from the capabilities, you didn't get their full description so for a given feature type, you first need to fetch its description (performing a DescribeFeatureType request behind) with type[[1]]$getDescription(). After what you will see type[[1]]$getGeometryType() returns a geometry type.

For bounding box you get something, because that's a property of the featureType that is available through WFS capabilities document. It's not the case of the full list of feature type elements (~ column definitions).

[maelle]

Thank you! I'm just surprised it used to work.

[eblondel]

That´s weird, WFS describeFeatureType are not called in series when grabing the capabilities, so it´s something that you have to explicitely invoke.

[maelle]

All I can say is that the same tests used to pass and no longer do. I'll fix it, thank you.

[eblondel]

There was a regression (introduced in f76da0a with #131), fixed in 54b7c37

[maelle]

Thank you 🙏