Add DataCite support for peer review DOIs

[taslangraham]

Important

Datacite does not currently support assigning unique DOIs to article versions. When support is added, the peer review export will need to be updated accordingly.

Related to:

• Add support for review-related DOIs #11332

With the addition of internal support for peer review DOIs, support for including these in Crossref metadata exports and deposits is necessary. It will need to be added only for public peer reviews with proper reviewer identity anonymity taken into consideration

DataCite schema: https://schema.datacite.org/meta/kernel-4/metadata.xsd
DataCite schema documentation: https://datacite-metadata-schema.readthedocs.io/en/4.7/
Guide on XML validation: https://support.datacite.org/docs/how-do-i-validate-doi-metadata

Steps to test the changes:
Prerequisites:

• Ensure Datacite plugin is enabled, selected as DOI registration agency, and configured with test credentials

---

1. Take a submission through the editorial process and produce a published article with at least one public peer review.
2. Ensure peer review DOIs are enabled.
3. Export Datacite article metadata (meant for external deposit). This will produce files depending on the following scenarios:
   • If DOIs for multiple submissions are selected for export, a zip file will be generated containing an XML file for each selected submission. However, if only a single submission's DOI is to be exported, then a single XML file will be generated and downloaded. This behaviour predates this ticket.
   • If there are multiple public peer reviews to be exported (i.e a submission has multiple public peer reviews or multiple submissions with public peer reviews were selected), then a zip file will be generated containing an XML for each of the peer reviews. However, if only a single peer review is to be exported, then a single XML file will be generated and downloaded.
4. Confirm peer review XML validity by following the guide at https://support.datacite.org/docs/how-do-i-validate-doi-metadata.
5. Deposit Datacite article metadata. Observe the change in status for the article DOI and Peer Review DOI(s).

What application are you using?
OJS 3.6

Additional information
N/A

Pull Requests
ojs: pkp/ojs#5715
ops (submodule update): pkp/ops#1363
omp (submodule update): pkp/omp#2422
pkp-lib: #13119

[taslangraham]

@kaitlinnewson Could you review the PRs listed in the issue description, please? I've included an example of the exported peer review XML below.

Example Datacite Peer Review XML

<?xml version="1.0" encoding="utf-8"?>
<resource xmlns="http://datacite.org/schema/kernel-4" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4/metadata.xsd">
  <identifier identifierType="DOI">10.82589/9ga1ap89</identifier>
  <creators>
    <creator>
      <creatorName>:unkn</creatorName>
    </creator>
  </creators>
  <titles>
    <title>Review: Towards Designing an Intercultural Curriculum: A Case Study from the Atlantic Coast of Nicaragua (Round1/Review1)</title>
  </titles>
  <publisher>Test Journal 1</publisher>
  <publicationYear>2026</publicationYear>
  <subjects>
    <subject>21st Century</subject>
    <subject>Diversity</subject>
    <subject>Multilingual</subject>
    <subject>Multiethnic</subject>
    <subject>Participatory Pedagogy</subject>
    <subject>Language</subject>
    <subject>Culture</subject>
    <subject>Gender</subject>
    <subject>Egalitarianism</subject>
    <subject>Social Transformation</subject>
  </subjects>
  <dates>
    <date dateType="Submitted">2026-07-23</date>
    <date dateType="Accepted">2026-07-23</date>
    <date dateType="Updated">2026-07-23</date>
    <date dateType="Issued">2026-07-23</date>
  </dates>
  <language>en</language>
  <resourceType resourceTypeGeneral="PeerReview">Peer Review</resourceType>
  <alternateIdentifiers>
    <alternateIdentifier alternateIdentifierType="publisherId">1-1-14-r47</alternateIdentifier>
  </alternateIdentifiers>
  <relatedIdentifiers>
    <relatedIdentifier relatedIdentifierType="DOI" relationType="Reviews">10.82589/vv72k817</relatedIdentifier>
  </relatedIdentifiers>
  <descriptions>
    <description descriptionType="Abstract">One of the challenges still to be met in the 21st century is that of genuinely embracing diversity. How can education help to overcome the barriers that continue to exist between people on the basis of language, culture and gender? This case study takes the Atlantic Coast of Nicaragua as an example of a multilingual/multiethnic region and examines how the community university URACCAN is contributing to the development of interculturality. It describes participatory research that was carried out with university staff and students with the intention of defining an intercultural curriculum and appropriate strategies for delivering such. One model used as a basis for discussions was the Model for Community Understanding from the Wales Curriculum Council, which emphasises the belonging of the individual to different communities or cultures at the same time. Factors supporting the development of an intercultural curriculum include the university’s close involvement with the ethnic communities it serves. However, ethno-linguistic power relations within the region and the country as a whole, still militate against egalitarianism within the university. The research highlights the importance of participatory pedagogy as the basis for promoting interculturality and achieving lasting social transformation.</description>
  </descriptions>
  <relatedItems>
    <relatedItem relationType="IsPublishedIn" relatedItemType="Journal">
      <relatedItemIdentifier relatedItemIdentifierType="EISSN">1438-5627</relatedItemIdentifier>
      <titles>
        <title>Test Journal 1</title>
      </titles>
      <volume>1</volume>
      <issue>2</issue>
    </relatedItem>
  </relatedItems>
</resource>

[kaitlinnewson]

Hi @taslangraham, I have not had a chance yet to give it a look, but ran an initial review through Claude which found a few items worth addressing - here's the report:

Review: DataCite support for peer review DOIs (#13062)

PRs reviewed: ojs#5715, pkp-lib#13119, omp#2422, ops#1363.

The approach is sound and closely mirrors the already-merged Crossref peer review work (#12509): same anonymity rule (reviewer named only for SUBMISSION_REVIEW_METHOD_OPEN), same round/review numbering, same landing-page URL shape. Comments below, ordered by severity.

---

pkp/ojs#5715

1. Fatal error for journals whose locales lack the new string

plugins/generic/datacite/filter/DataciteXmlFilter.php:504-551

The title branch only populates $titles when Locale::getBundle($locale)->translateSingular('plugins.generic.dataCite.reviewTitle', …) returns non-null. getBundle() loads only that locale's files, with no fallback (lib/pkp/classes/i18n/Locale.php:319). Until the string is translated, a journal whose supported locales are all non-English will continue on every iteration, leaving $titles undefined; line 568 then calls getTranslationsByPrecedence(array $localizedData, …) with an undefined variable, which is a TypeError rather than the intended "titles are required" exception.

Crossref sidesteps this by using __($key, $params, $locale), which falls back to the default locale. At minimum, initialise $titles = [] and fall back for the primary locale.

2. Locale key casing

plugins/generic/datacite/locale/en/locale.po

plugins.generic.dataCite.reviewTitle uses camelCase dataCite; every other key in the file is plugins.generic.datacite.* / plugins.importexport.datacite.*. Worth correcting before it reaches Weblate.

3. Deposited landing-page URL includes a locale segment

plugins/generic/datacite/DataciteExportPlugin.php:458-471

The new ReviewAssignment case stops at $params, so $escape and $urlLocaleForPage fall back to their defaults (false, null). Every other case in this method passes null, null, true, '' to strip the locale, and Crossref's peer review filter passes null, true, ''. As written, the URL registered with DataCite pins a language.

4. Accepted date can be deposited as 1970-01-01

plugins/generic/datacite/filter/DataciteXmlFilter.php:605-607

$dateConsidered = $reviewAssignment->getDateConsidered();
$dates[DATACITE_DATE_ACCEPTED] = $dateConsidered;

This is unguarded, unlike every other branch in the method. unconsiderReview() (lib/pkp/classes/controllers/grid/users/reviewer/PKPReviewerGridHandler.php:724-727) sets dateConsidered to null while leaving dateAcknowledged intact, so null is reachable — date('Y-m-d', strtotime(null)) yields 1970-01-01, plus a PHP 8.1 deprecation notice. Wrap in !empty() like the neighbouring cases.

5. Queries repeated per locale

plugins/generic/datacite/filter/DataciteXmlFilter.php:504-550

The review-round lookup and the "all reviews in round" query plus usort are locale-invariant, but they run once per locale, per review. They should be hoisted above the foreach; Crossref batches the equivalent work once in loadAllReviewsPerRound().

Since the numbering block is otherwise a verbatim copy of PeerReviewCrossrefXmlFilter::createTitlesNode(), a shared helper in pkp-lib would keep the two agencies from drifting apart.

6. $cache->get() called without isCached()

plugins/generic/datacite/filter/DataciteXmlFilter.php:508-509

PubObjectCache::get() opens with assert($this->isCached(...)) and its docblock explicitly requires callers to check first. This works today only because process() always caches the round beforehand — which also makes the whole block redundant.

7. array_search() can return false

plugins/generic/datacite/filter/DataciteXmlFilter.php:534

false + 1 silently produces "Review1". This is reachable: getExportableDOIsPeerReviewIds() does not require editor confirmation, but the numbering query does (filterByIsConfirmedByEditor(true)).

More broadly, the review number derives from mutable state (public visibility, editor confirmation, completion order), so a later change can renumber already-registered review titles on re-deposit. Crossref has the same characteristic, so this may be accepted behaviour — worth confirming.

8. The review record inherits the article's abstract, keywords, and licence

plugins/generic/datacite/filter/DataciteXmlFilter.php:256-274 and :331

The subjects block and createDescriptionsNode() never receive $reviewAssignment, so the article's abstract is deposited as the review's <description descriptionType="Abstract"> and the article's keywords as its <subjects> — both visible in the sample XML posted on the issue. Crossref sends neither. Worth an explicit decision here, along with whether relatedItems (IsPublishedIn) and rightsList belong on a PeerReview record.

9. Versioning behaviour differs from Crossref

plugins/generic/datacite/filter/DataciteXmlFilter.php:175-176 vs :203

The publication fetched from the review round is immediately overwritten by $article->getCurrentPublication(), so the Reviews related identifier and the review title always reflect the current version. Crossref branches explicitly on SETTING_DOI_VERSIONING and uses the reviewed version's DOI when versioning is enabled.

Given the note at the top of the issue that DataCite doesn't yet support version DOIs, this is probably intentional — if so, drop the discarded fetch and add a comment, as Crossref does.

Minor

• The code emits <resourceType resourceTypeGeneral="PeerReview"/> with no text content, but the sample XML on the issue shows Peer Review as content. Likely a stale sample — worth confirming which version was validated.
• Stray blank line at DataciteXmlFilter.php:728.
• New blocks in filter/filterConfig.xml use 4 spaces; the rest of the file uses tabs.
• :unkn is correct as written — DataCite's Appendix 3 example is literally <creatorName>:unkn</creatorName>, unparenthesised.

---

#13119

10. The filterByCompleted → filterByIsConfirmedByEditor swap is a genuine fix, but the method name no longer matches

classes/publication/Repository.php:1217-1223

filterByCompleted(true) also requires s.status <> STATUS_PUBLISHED and stage_id <> SUBMISSION (classes/submission/reviewAssignment/Collector.php:470-493), so at the default CREATION_TIME_PUBLICATION it returned nothing and peer review DOIs were never minted. Good catch.

Two follow-ups:

• getCompletedReviewAssignments() and its docblock ("Retrieve completed review assignments", "@return Enumerable Completed Review assignments") should be renamed to match what it now returns.
• filterByIsConfirmedByEditor does not require date_completed, unlike getExportableDOIsPeerReviewIds(). Consider retaining a completion condition so the two stay aligned.

Edge case: unconsidering a review that was never acknowledged nulls dateConsidered, which drops its already-minted DOI out of getDoisForSubmission() / getDoisForPublication().

11. The DepositSubmission workaround is effective, but the comment misdescribes it

jobs/doi/DepositSubmission.php:78-84

getConfiguredDoiAgency() returns the PluginRegistry singleton, not "a fresh instance". It's clean only because the job's own $this->agency is a deserialised copy whose DatacitePlugin::$_exportPlugin (holding the PubObjectCache) got populated during depositSubmissions(), while the registry copy did not. The same serialisation failure returns if the registry instance happens to have been used earlier in the process. The call is also inside the loop, so it runs once per review.

A cleaner fix would be to keep the agency out of the job payload entirely: DepositPeerReview already loads the context by ID and could resolve the agency in handle().

---

Other

• No test coverage is added, and there is no DoiDatacite Cypress spec at all — the Crossref work at least updated DoiCrossref.cy.js. Manual test steps are documented on the issue.
• All three submodule PRs point at ab400ac (the pkp-lib PR head) and will need re-pointing after pkp-lib merges.
• OMP and OPS carry no behavioural change: getCompletedReviewAssignments() has no callers outside OJS.