Merge pull request #327 from scieloorg/copilot/fix-preprint-data-crossref

Crossref export: emit hasPreprint relation from related-article preprint entries

Author: robertatakenaka

articlemeta/export_crossref.py

@@ -1134,6 +1134,36 @@ def transform(self, data):
         data = self._transform_translations(data)
         return data
 
+    @staticmethod
+    def _get_preprint_relations(raw):
+        """Return the list of related-article entries marked as preprint.
+
+        SciELO stores related-article info in ISIS field ``v241``, with
+        subfields ``i`` (identifier/href), ``t`` (related-article-type) and
+        ``n`` (ext-link-type). Only entries whose type is ``preprint`` and
+        whose link type is ``doi`` (or unspecified) carry a usable DOI for
+        the Crossref ``hasPreprint`` relation.
+        """
+        try:
+            related = raw.data['article'].get('v241') or []
+        except (AttributeError, KeyError, TypeError):
+            return []
+
+        preprints = []
+        for item in related:
+            if not isinstance(item, dict):
+                continue
+            if item.get('t') != 'preprint':
+                continue
+            identifier = item.get('i') or item.get('_')
+            if not identifier:
+                continue
+            ext_link_type = item.get('n')
+            if ext_link_type and ext_link_type != 'doi':
+                continue
+            preprints.append(identifier)
+        return preprints
+
     def _transform_original(self, data):
         raw, xml = data
 
@@ -1168,6 +1198,19 @@ def _transform_original(self, data):
 
             program_node.append(related_item_node)
 
+        # program/related_item (hasPreprint)
+        for preprint_doi in self._get_preprint_relations(raw):
+            related_item_node = ET.Element('related_item')
+
+            intra_work_relation_node = ET.Element('intra_work_relation')
+            intra_work_relation_node.set(
+                'relationship-type', 'hasPreprint')
+            intra_work_relation_node.set('identifier-type', 'doi')
+            intra_work_relation_node.text = preprint_doi
+            related_item_node.append(intra_work_relation_node)
+
+            program_node.append(related_item_node)
+
         journal_article_node.append(program_node)
 
         return data

tests/test_export_crossref.py

@@ -1342,6 +1342,83 @@ def test_related_item_for_multilingue_document(self):
                     content[3],
                     intra_work_relation.attrib.get('relationship-type'))
 
+    def test_related_item_includes_has_preprint_relation(self):
+        self._article.data['article']['v241'] = [
+            {
+                'i': '10.1590/SciELOPreprints.9348',
+                't': 'preprint',
+                'n': 'doi',
+            }
+        ]
+        xmlcrossref = create_xmlcrossref_with_n_journal_article_element(
+            ['pt', 'en', 'es'])
+
+        data = [self._article, xmlcrossref]
+        xmlcrossref = export_crossref.XMLProgramRelatedItemPipe()
+        raw, xml = xmlcrossref.transform(data)
+
+        # main journal_article keeps its translations and gains the preprint
+        main_program = xml.findall('.//journal_article')[0].find('program')
+        relations = main_program.findall('related_item/intra_work_relation')
+        relation_types = [r.attrib.get('relationship-type') for r in relations]
+
+        self.assertIn('hasPreprint', relation_types)
+        preprint_node = next(
+            r for r in relations
+            if r.attrib.get('relationship-type') == 'hasPreprint'
+        )
+        self.assertEqual('doi', preprint_node.attrib.get('identifier-type'))
+        self.assertEqual('10.1590/SciELOPreprints.9348', preprint_node.text)
+
+        # translations should not get the hasPreprint relation
+        for journal_article in xml.findall('.//journal_article')[1:]:
+            translation_relations = journal_article.findall(
+                'program/related_item/intra_work_relation')
+            translation_types = [
+                r.attrib.get('relationship-type')
+                for r in translation_relations
+            ]
+            self.assertNotIn('hasPreprint', translation_types)
+
+    def test_related_item_without_preprint_does_not_emit_has_preprint(self):
+        # ensure no v241 entries -> no hasPreprint relation is emitted
+        self._article.data['article'].pop('v241', None)
+        xmlcrossref = create_xmlcrossref_with_n_journal_article_element(
+            ['pt', 'en', 'es'])
+
+        data = [self._article, xmlcrossref]
+        xmlcrossref = export_crossref.XMLProgramRelatedItemPipe()
+        raw, xml = xmlcrossref.transform(data)
+
+        relation_types = [
+            r.attrib.get('relationship-type')
+            for r in xml.findall(
+                './/program/related_item/intra_work_relation')
+        ]
+        self.assertNotIn('hasPreprint', relation_types)
+
+    def test_related_item_ignores_non_preprint_related_articles(self):
+        self._article.data['article']['v241'] = [
+            {
+                'i': '10.1590/some-other.1234',
+                't': 'commentary',
+                'n': 'doi',
+            }
+        ]
+        xmlcrossref = create_xmlcrossref_with_n_journal_article_element(
+            ['pt', 'en', 'es'])
+
+        data = [self._article, xmlcrossref]
+        xmlcrossref = export_crossref.XMLProgramRelatedItemPipe()
+        raw, xml = xmlcrossref.transform(data)
+
+        relation_types = [
+            r.attrib.get('relationship-type')
+            for r in xml.findall(
+                './/program/related_item/intra_work_relation')
+        ]
+        self.assertNotIn('hasPreprint', relation_types)
+
     def test_collection_for_multilingue_document(self):
         xmlcrossref = create_xmlcrossref_with_n_journal_article_element(
             ['pt', 'en', 'es'], 'doi_data')