Improve 3dtiles loading from CSW search

[landryb]

Description

the support added in #9368 (and #9585) could be improved, because from what i understand as of now (and i think this is undocumented ?), to successfully load a 3dtiles dataset from a CSW search, one needs:

• a single entry with <dc:format>3D Tiles</dc:format>
• a single <dc:URI> pointing at the JSON dataset url

any other case doesn't work. We have MDs in our geonetwork catalog with several <dc:format> and several<dc:URI>, and only some of them are pointing to 3dtiles. They are properly labeled with protocol="OGC:3DTILES" (which was added in GN upstream in geonetwork/core-geonetwork#6950, which is used in #9021) so i would have expected those URIs to be selected. Afaict with the current code, only the first/only dc:URI is used/tried ?

I've tried fixing our MDs in geonetwork to add the format tag, but i cant remove the other links. So my idea to fix that, avoiding breaking other existing setups:

• if there's several <dc:format>, select the MD only if 3D Tiles is present in the list
• if there's only one <dc:URI> use it regardless of the protocol (present or not, whatever the value)
• if there's several <dc:URI>, filter only the ones with protocol="OGC:3DTILES" (there might be several, i have to look at how the CSW search results allow to select within several OGC MD links)

sample datasets, for example:

• in https://ids.craig.fr/mapstore/#/ select 'local catalog', search for '3dtiles', the 2 records have links to 3dtiles (they dont have 3dtiles in the format list now, but that's the same if present, ive tested it).
• https://ids.craig.fr/geocat/srv/fre/catalog.search#/metadata/cd983dca-c7bd-4bf9-bb48-ab8ddf7af8a4 is a sample MD with several links to 3dtiles.

What kind of improvement you want to add?

• Minor changes to existing features

[landryb]

an example of the dcat output for a record in our catalog, with several formats/URI:

      <dc:rights>otherRestrictions</dc:rights>
      <dc:rights>license</dc:rights>
      <dc:rights>otherRestrictions</dc:rights>
      <dc:language>fre</dc:language>
      <dc:source>Produit par ATGT pour le Centre Régional Auvergne-Rhône-Alpes de l'Information Géographique (CRAIG) dans le cadre d'un projet de modélisation 3D des territoires d'Auvergne-Rhône-Alpes, avec la participation financière du Département de l'Isère (CD 38).

Les bâtiments ont été tracés en photogrammétrie sur des prises de vues aériennes nadirales à 5 cm de résolution.</dc:source>
      <dc:relation>a3355b46-be00-4851-bafe-41f7d351fc95</dc:relation>
      <dc:format>CityGML</dc:format>
      <dc:format>CityJSON</dc:format>
      <dc:format>COLLADA</dc:format>
      <dc:format>ESRI Geodatabase</dc:format>
      <dc:format>OBJ</dc:format>
      <ows:BoundingBox crs="urn:ogc:def:crs:EPSG:6.6:4326">
        <ows:LowerCorner>44.6959 4.6486</ows:LowerCorner>
        <ows:UpperCorner>45.8139 6.3593</ows:UpperCorner>
      </ows:BoundingBox>
      <dc:URI protocol="" name="02_Bati3D/CityGML" description="Lien de téléchargement de la maquette 3D par zone au format CityGML.">https://drive.opendata.craig.fr/s/opendata?path=%2F3d%2F2021_aura_3d%2F02_Bati3D%2FCityGML</dc:URI>
      <dc:URI protocol="" name="02_Bati3D/CityJSON" description="Lien de téléchargement de la maquette 3D par zone au format CityJSON.">https://drive.opendata.craig.fr/s/opendata?path=%2F3d%2F2021_aura_3d%2F02_Bati3D%2FCityJSON</dc:URI>
      <dc:URI protocol="" name="02_Bati3D/COLLADA" description="Lien de téléchargement de la maquette 3D par zone au format COLLADA.">https://drive.opendata.craig.fr/s/opendata?path=%2F3d%2F2021_aura_3d%2F02_Bati3D%2FCOLLADA</dc:URI>
      <dc:URI protocol="" name="02_Bati3D/ESRI_Geodatabase" description="Lien de téléchargement de la maquette 3D par zone au format ESRI Geodatabase.">https://drive.opendata.craig.fr/s/opendata?path=%2F3d%2F2021_aura_3d%2F02_Bati3D%2FESRI_Geodatabase</dc:URI>
      <dc:URI protocol="" name="02_Bati3D/OBJ" description="Lien de téléchargement de la maquette 3D par zone au format OBJ.">https://drive.opendata.craig.fr/s/opendata?path=%2F3d%2F2021_aura_3d%2F02_Bati3D%2FOBJ</dc:URI>
      <dc:URI protocol="" name="CC des Balcons du Dauphiné - bâti 3D" description="Visualisation de la donnée dans MapStore.">https://3d.craig.fr/datasets/2024_bati3d_isere/cc_balcons_dauphine/solar/tileset.json</dc:URI>
      <dc:URI protocol="OGC:3DTILES" name="CC Entre Bièvre et Rhône - bâti 3D" description="Visualisation de la donnée dans MapStore.">https://3d.craig.fr/datasets/2024_bati3d_isere/cc_eber/solar/tileset.json</dc:URI>
      <dc:URI protocol="" name="CC du Grésivaudan - bâti 3D" description="Visualisation de la donnée dans MapStore.">https://3d.craig.fr/datasets/2024_bati3d_isere/cc_gresivaudan/solar/tileset.json</dc:URI>
      <dc:URI protocol="OGC:3DTILES" name="CA du Pays Voironnais (1) - bâti 3D" description="Visualisation des données dans MapStore.">https://3d.craig.fr/datasets/2024_bati3d_isere/ca_pays_voironnais_1/solar/tileset.json</dc:URI>
      <dc:URI protocol="" name="CA du Pays Voironnais (2) - bâti 3D" description="Visualisation de la donnée dans MapStore.">https://3d.craig.fr/datasets/2024_bati3d_isere/ca_pays_voironnais_2/solar/tileset.json</dc:URI>
    </csw:Record>

[landryb]

i think the testdata that was added in 936c030 looks wrong as URI is a single entry, and not an array like other entries in the file. same thing for the format attribute, which should/could also be an array ?

i see in various places in CSW.js that castArray is used (i guess to transform a single value in an array), and i suppose it should also be used here to lookup the 'right' format ?
right now, for the 3dtiles case, the code assumes that URI and format are single entries, and i dont think they handle properly the fact that there can be an array.

should a filter() function should be added to find the right entries in dc:URI array ?

[tdipisa]

i think the testdata that was added in 936c030 looks wrong as URI is a single entry, and not an array like other entries in the file. same thing for the format attribute, which should/could also be an array ?

We found the same recently and we are looking at this these days. An issue will be opened soon for this. Thank you anyway for reporting that.

[landryb]

We found the same recently and we are looking at this these days. An issue will be opened soon for this. Thank you anyway for reporting that.

great! i've started a wip PR (#10927) to try to fix my usecase for uri.protocol === "OGC:3DTILES".

[landryb]

looking a bit more at the code history, maybe 3dtiles/ogc service type support should be added in the same code that was done for #10272, which would allow to 'select' one of several URI/services available in a single CSW record.

[landryb]

i think the testdata that was added in 936c030 looks wrong as URI is a single entry, and not an array like other entries in the file. same thing for the format attribute, which should/could also be an array ?

We found the same recently and we are looking at this these days. An issue will be opened soon for this. Thank you anyway for reporting that.

i suppose that's #10937 ? if so i'll see how intermixed it is with the 3dtiles loading from CSW issue.

[tdipisa]

i suppose that's #10937 ? if so i'll see how intermixed it is with the 3dtiles loading from CSW issue.

yes @landryb