geopython
ogc api records 404 if uuid contains '/' or '%2F'
Description
Some communities tend to place a doi in metadata identifier e.g. 10.5281/zenodo.4088113
If you navigate to this item using /collections/metadata:main/items/10.5281/zenodo.4088113, then a 404 is returned, same error occurs with /collections/metadata:main/items/10.5281%2Fzenodo.4088113 (urlencoded).
I kind of expected option 2 to work fine, in which case we would have to make sure we always encode the '/' to %2F, alternatives could be to prevent '/' in identifiers by substituting to '-' or throw an error on insert.
this problem actually exists on the demo server https://demo.pycsw.org/gisdata/collections/metadata:main/items/http%3A%2F%2Fcapita.wustl.edu%2FDataspaceMetadata_ISO%2FCIRA.VIEWS.BRf.xml
I was curious if the problem exists also on pygeoapi, but there it seems covered, maybe it is considered by the flask api?
seems the problem still exists:
- https://demo.pycsw.org/gisdata/collections/metadata:main/items/http://capita.wustl.edu/DataspaceMetadata_ISO/CIRA.VIEWS.MF.xml
- https://demo.pycsw.org/gisdata/collections/metadata:main/items/http://capita.wustl.edu/DataspaceMetadata_ISO/FZ%20Juelich.MACC.vmr_no.xml
@pvgenuchten I cannot reproduce this issue locally. I've tried inspecting on demo.pycsw.org directly, and found the following.
Given a URL like:
https://demo.pycsw.org/gisdata/collections//metadata:main/items/http://capita.wustl.edu/DataspaceMetadata_ISO/CIRA.VIEWS.MF.xml
And the below pycsw container logs on demo.pycsw.org:
[2024-08-03T19:16:39Z] {/home/pycsw/pycsw/pycsw/ogc/api/records.py:837} DEBUG - Querying repository for item http:/capita.wustl.edu/DataspaceMetadata_ISO/CIRA.VIEWS.MF.xml
Here, we see that http://capita.wustl.edu/DataspaceMetadata_ISO/CIRA.VIEWS.MF.xml is getting converted to http:/capita.wustl.edu/DataspaceMetadata_ISO/CIRA.VIEWS.MF.xml.
I tried adjusting the nginx setup on demo.pycsw.org withmerge_slashes: off; but no luck.
In this case I would say things are working at the application level as expected.
cc @kalxas @ricardogsilva
should we close it, or wait for re-configuration on demo server? @kalxas