Added link to the THREDDS Catalog to the Collection object. #23
Conversation
|
@huard |
|
|
||
| link = pystac.Link(rel="source", | ||
| target=url, | ||
| media_type="text/html", |
There was a problem hiding this comment.
In the case of the STAC Item that represents a NetCDF, source should probably refer to application/x-netcdf with an URL of the actual file? This URL should be indicated at the top level of the NCML file as @location using Dataset.to_cf_dict.
If this change is done, instead of looking for /catalog/ in the URL, the /dodsC/ must be considered.
This code can be reused though, to inject source in the STAC Collection, which would be a XML THREDDS catalog URL.
There was a problem hiding this comment.
I also found that the NCML has a id attribute in THREDDSMetadata with value birdhouse/testdata/xclim/cmip6/sic_SImon_CCCma-CanESM5_ssp245_r13i1p2f1_2020.nc.
This seems more reliable than parsing the URL ourselves.
There was a problem hiding this comment.
- The href is not the URL of the actual netCDF file, but of its catalog entry, so not sure the mime should be application/netcdf. The netcdf link is listed among the Assets.
- Yes, I'm using that ID already.
There was a problem hiding this comment.
When loading in xncml.Dataset, the "@location" in returned attributes directly at the root is the NetCDF file URL.
This is the same location attribute directly at the top of:
https://pavics.ouranos.ca/twitcher/ows/proxy/thredds/ncml/birdhouse/testdata/xclim/cmip6/sic_SImon_CCCma-CanESM5_ssp245_r13i1p2f1_2020.nc?catalog=https%3A%2F%2Fpavics.ouranos.ca%2Ftwitcher%2Fows%2Fproxy%2Fthredds%2Fcatalog%2Fbirdhouse%2Ftestdata%2Fxclim%2Fcmip6%2Fcatalog.html&dataset=birdhouse%2Ftestdata%2Fxclim%2Fcmip6%2Fsic_SImon_CCCma-CanESM5_ssp245_r13i1p2f1_2020.nc
The link actually returned uses dodsC service, so the response from THREDDS (https://pavics.ouranos.ca/twitcher/ows/proxy/thredds/fileServer/birdhouse/testdata/xclim/cmip6/sic_SImon_CCCma-CanESM5_ssp245_r13i1p2f1_2020.nc) that I get is in fact a 400 HTML 😞
What we could do is employ the fileServer endpoint instead, where that would be the real source in application/x-netcdf (https://pavics.ouranos.ca/twitcher/ows/proxy/thredds/fileServer/birdhouse/testdata/xclim/cmip6/sic_SImon_CCCma-CanESM5_ssp245_r13i1p2f1_2020.nc).
I think the NCML XML href and the THREDDS Catalog HTML href would be relevant in links as rel=describedby and rel=alternate respectively. The NCML href could also use convertedfrom actually!
The rel=collection href could also be inserted since you have it.
There was a problem hiding this comment.
https://github.com/crim-ca/ncml2stac/blob/main/notebooks/ncml2stac.ipynb contains the updated item_link implementation using @location attribute
STACpopulator/input.py
Outdated
| attrs["access_urls"] = ds.access_urls | ||
|
|
||
| attrs["catalog_url"] = self.catalog_head.catalog_url |
There was a problem hiding this comment.
For some reason, ds.access_urls causes an AttributeError on my side. Do you have an idea why that could be?
When I skip this step to avoid the error, my resulting STAC Item JSON (see bottom of output in https://github.com/crim-ca/ncml2stac/pull/3/files#diff-09442021b34927015b7a3703f6ed65611059a807c957feb4dfe2cd9955ccd68e) has the assets references, but the names are different:
"assets": {
"httpserver_service": { ... },
"opendap_service": { ... },
...
}Seems to be caused by this logic: https://github.com/crim-ca/stac-populator/pull/23/files#diff-ea68b4da6eb5897aa6022852045603d912245daf3f0b044cb9abbd176e758455R196-R199
|
Should (?) be fixed via #25. |
No description provided.