You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
In DataCite/Datacite export, when the user adds a DAI author identifier, include in creator nameIdentifier property #12296
When the user adds a DAI author identifier type, the DataCite/Datacite export should include the DAI author identifier in the creator nameIdentifier property, like the OpenAIRE/oai_datacite export does. Currently, the DataCite/Datacite export omits the DAI identifier entirely.
This issue is part of a larger effort to align both metadata exports, described in #11916 and #12281.
What kind of user is the feature intended for?
API User, Superuser, and Sysadmin
What inspired the request?
This is part of a larger effort to align the OpenAIRE/oai_datacite and DataCite/Datacite metadata exports. There is more context at #11916 and #12281.
What existing behavior do you want changed?
Currently, when the user adds a DAI author identifier type and a DAI identifier, the DataCite/Datacite export does not include the identifier in the nameIdentifier property.
The OpenAIRE/oai_datacite export does include it. For example:
The DataCite/Datacite export should be changed to include the DAI identifier in the nameIdentifier property in the same way that the OpenAIRE/oai_datacite export does.
Any open or closed issues related to this feature request?
changed the title [-]In DataCite/Datacite export, when the user adds a DAI author identifier, include in creator nameIdentifier element[/-][+]In DataCite/Datacite export, when the user adds a DAI author identifier, include in creator nameIdentifier property[/+]on Apr 18, 2026
. Just returning a uri regardless of whether it's http may not be possible since this method is used elsewhere (maybe some refactoring to have a getIdentifierasUri() method that does everything except check for http as we want for getIdentifierAsUrl ()).
and doesn't seem to use our tests that the input matches the regex for that id type. This code is why DAI doesn't have schemeUri for DAI - the code to get it only works on identifiers starting with http. Relates to #12995.
@jggautier - did you test with non-URL identifier forms or invalid identifiers? I think the DataCite export should handle these, whereas the OPenAire one may allow invalid ones and drop the schemeUri entry if the identifier doesn't start with http
Unless I'm missing something, those checks are applied in the DataCite xml export and not in the OpenAire export so an ORCID with too many digits or any other typo should result in no entry in DataCite and incorrect info passed on in the OpenAire export.
(For example, the ORCID one says an ORCID may have an https://orcid.org/ in front of a required 3 sets of four digits separated by dashes and a final set of either four digits or three digits and an 'X' - those are ORCID's rules for how an ORCID can be written.)
Neither actually have checks to see if the identifier is valid in the sense of referring to a real/live entry - the check is just on the syntax.
Ah okay. I didn't realize that how Dataverse checks for invalid identifiers when creating the DataCite export is different from how it checks for invalid identifiers when creating the OpenAIRE export.
Thanks for pointing that out. I think Dataverse should check for invalid identifiers in the same way when creating both exports.
In my testing I used the DAI info:eu-repo/dai/nl/072728442, which as far as I can tell is valid when considering the regex in that code you pointed to. At least that's what an LLM tells me.
I think I'll do more testing to see what Dataverse does with invalid identifiers when creating the DataCite and OpenAIRE exports. And open another GitHub issue about changing how Dataverse creates these two exports so that it checks for invalid identifiers in the same way.
Of course I think ideally we would tell Dataverse to prevent depositors from entering invalid identifiers as they're describing their deposits, like on the deposit form. The community's probably talked about this from time to time, too, right? But I think that's out of scope of this GitHub issue and the larger parent GitHub issue.
and doesn't seem to use our tests that the input matches the regex for that id type. This code is why DAI doesn't have schemeUri for DAI - the code to get it only works on identifiers starting with http. Relates to #12995.
Earlier this week I wrote that I'd do more testing to see what Dataverse does with invalid identifiers when creating the DataCite and OpenAIRE exports and open another GitHub issue about changing how Dataverse creates these two exports so that it checks for invalid identifiers in the same way.
@sekmiller, I'm wondering if it might be more efficient instead to tell Dataverse that when it's creating the OpenAIRE/oai_datacite export, use the same code that Dataverse uses for the DataCite/Datacite export to figure out if the identifier that a user enters in the Author Identifier field is valid, and exclude invalid identifiers from the exports.
Then we could adjust that code to make sure that when the user adds a valid DAI, the OpenAIRE/oai_datacite and DataCite/Datacite exports includes that, too.
Activity
[-]In DataCite/Datacite export, when the user adds a DAI author identifier, include in creator nameIdentifier element[/-][+]In DataCite/Datacite export, when the user adds a DAI author identifier, include in creator nameIdentifier property[/+]qqmyers commentedon Jul 2, 2026
DAI is different in that is the only identifier type whose uri form doesn't start with http. So these lines exclude it from the DataCite export:
dataverse/src/main/java/edu/harvard/iq/dataverse/DatasetAuthor.java
Lines 105 to 107 in 8f52fcc
i() method that does everything except check for http as we want for getIdentifierAsUrl()).FWIW: OpenAire has custom processing here:
dataverse/src/main/java/edu/harvard/iq/dataverse/export/openaire/OpenAireExportUtil.java
Lines 289 to 296 in 8f52fcc
@jggautier - did you test with non-URL identifier forms or invalid identifiers? I think the DataCite export should handle these, whereas the OPenAire one may allow invalid ones and drop the schemeUri entry if the identifier doesn't start with http
jggautier commentedon Jul 2, 2026
Hey @qqmyers.
What do you mean by testing?
Would a non-URL identifier form be something like "info:eu-repo/dai/nl/072728442"?
What do you mean by invalid identifiers? Like whether or not info:eu-repo/dai/nl/072728442 actually identifies a person?
qqmyers commentedon Jul 3, 2026
We have code that defines what a valid identifier looks like for the author identifiers at
dataverse/src/main/java/edu/harvard/iq/dataverse/ExternalIdentifier.java
Lines 7 to 18 in 0df5da7
(For example, the ORCID one says an ORCID may have an https://orcid.org/ in front of a required 3 sets of four digits separated by dashes and a final set of either four digits or three digits and an 'X' - those are ORCID's rules for how an ORCID can be written.)
Neither actually have checks to see if the identifier is valid in the sense of referring to a real/live entry - the check is just on the syntax.
jggautier commentedon Jul 13, 2026
Ah okay. I didn't realize that how Dataverse checks for invalid identifiers when creating the DataCite export is different from how it checks for invalid identifiers when creating the OpenAIRE export.
Thanks for pointing that out. I think Dataverse should check for invalid identifiers in the same way when creating both exports.
In my testing I used the DAI info:eu-repo/dai/nl/072728442, which as far as I can tell is valid when considering the regex in that code you pointed to. At least that's what an LLM tells me.
I think I'll do more testing to see what Dataverse does with invalid identifiers when creating the DataCite and OpenAIRE exports. And open another GitHub issue about changing how Dataverse creates these two exports so that it checks for invalid identifiers in the same way.
Of course I think ideally we would tell Dataverse to prevent depositors from entering invalid identifiers as they're describing their deposits, like on the deposit form. The community's probably talked about this from time to time, too, right? But I think that's out of scope of this GitHub issue and the larger parent GitHub issue.
jggautier commentedon Jul 13, 2026
Hey again @qqmyers. You wrote:
But I couldn't find that GitHub issue or PR with the number 12995. I tried https://github.com/IQSS/dataverse/issues/12995 and
https://github.com/IQSS/dataverse-pm/issues/12995, but they lead to page not found pages. And I tried #1299, but that leads to a GitHub issue that seems unrelated.
Could you post the URL to that #12995 issue?
qqmyers commentedon Jul 13, 2026
It was a a typo: #12295
jggautier commentedon Jul 16, 2026
Thanks!
Earlier this week I wrote that I'd do more testing to see what Dataverse does with invalid identifiers when creating the DataCite and OpenAIRE exports and open another GitHub issue about changing how Dataverse creates these two exports so that it checks for invalid identifiers in the same way.
@sekmiller, I'm wondering if it might be more efficient instead to tell Dataverse that when it's creating the OpenAIRE/oai_datacite export, use the same code that Dataverse uses for the DataCite/Datacite export to figure out if the identifier that a user enters in the Author Identifier field is valid, and exclude invalid identifiers from the exports.
Then we could adjust that code to make sure that when the user adds a valid DAI, the OpenAIRE/oai_datacite and DataCite/Datacite exports includes that, too.
What do you think?