Identify DOI in RERO DOC field 775 $o

* When MARC field 775 $o contains a DOI in RERO DOC, its value is imported in the `identifiedBy` field instead of `otherEdition`. * Closes #659. Co-Authored-by: Miguel Moreira <[email protected]>
rero · Nov 5, 2021 · d2789b6 · d2789b6
1 parent dc2d35c
commit d2789b6
Show file tree

Hide file tree

Showing 3 changed files with 143 additions and 8 deletions.
diff --git a/sonar/modules/documents/dojson/rerodoc/model.py b/sonar/modules/documents/dojson/rerodoc/model.py
@@ -638,12 +638,23 @@ def marc21_to_other_edition(self, key, value):
     if not electronic_locator or not public_note:
         return None
 
-    return {
-        'document': {
-            'electronicLocator': electronic_locator
-        },
-        'publicNote': public_note
-    }
+    # if the value matches a DOI, apply `identifiedBy[type:bf:Doi]`
+    matches = re.search(r'(?P<doi>10\.\d{4,9}/[-._;()/:a-zA-Z0-9]+)', value.get('o'))
+    if matches and matches.group('doi'):
+        identified_by = self.get('identifiedBy', [])
+        identified_by.append({
+            'type': 'bf:Doi',
+            'value': matches.group('doi')
+        })
+        self['identifiedBy'] = identified_by
+        return None
+    else:
+        return {
+            'document': {
+                'electronicLocator': electronic_locator
+            },
+            'publicNote': public_note
+        }
 
 
 @overdo.over('collections', '^982..')

diff --git a/tests/ui/documents/dojson/rerodoc/test_rerodoc_model.py b/tests/ui/documents/dojson/rerodoc/test_rerodoc_model.py
@@ -1614,6 +1614,121 @@ def test_marc21_to_other_edition(app):
     data = overdo.do(marc21json)
     assert not data.get('otherEdition')
 
+    # Incorrect DOIs - they do get collected in `otherEdition`
+    # instead of `identifiedBy`
+    marc21xml = """
+    <record>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">http://dx.doi.org/10.1130%2F0091-7613(2002)030%3C0655:CWCIAP%3E2.0.CO%3B2</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">http://dx.doi.org/0.1021/jp0558775</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">http://dx.doi.org/1017/S0031182010000296</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">http://dx.doi.org/1039/B926873A</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">http://dx.doi.org/0.1016/j.str.2012.09.019</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">https://doi.org/10.1111%2Fj.1467-9280.2009.02364.x</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+    </record>
+    """
+    marc21json = create_record(marc21xml)
+    data = overdo.do(marc21json)
+    assert data.get('otherEdition') == [{
+        'document': {
+            'electronicLocator': 'http://dx.doi.org/10.1130%2F0091-7613(2002)030%3C0655:CWCIAP%3E2.0.CO%3B2'
+        },
+        'publicNote': 'version publiée'
+    }, {
+        'document': {
+            'electronicLocator': 'http://dx.doi.org/0.1021/jp0558775'
+        },
+        'publicNote': 'version publiée'
+    }, {
+        'document': {
+            'electronicLocator': 'http://dx.doi.org/1017/S0031182010000296'
+        },
+        'publicNote': 'version publiée'
+    }, {
+        'document': {
+            'electronicLocator': 'http://dx.doi.org/1039/B926873A'
+        },
+        'publicNote': 'version publiée'
+    }, {
+        'document': {
+            'electronicLocator': 'http://dx.doi.org/0.1016/j.str.2012.09.019'
+        },
+        'publicNote': 'version publiée'
+    }, {
+        'document': {
+            'electronicLocator': 'https://doi.org/10.1111%2Fj.1467-9280.2009.02364.x'
+        },
+        'publicNote': 'version publiée'
+    }]
+    assert not data.get('identifiedBy')
+
+    # Well-formed DOIs - they get collected in `identifiedBy`
+    # instead of`otherEdition`
+    marc21xml = """
+    <record>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">http://dx.doi.org/10.1002/1521-3773(20020104)41:1</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">10.1016/j.apergo.2008.03.002</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">: https://doi.pangaea.de/10.1594/PANGAEA.914883</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">https://www.brepolsonline.net/doi/abs/10.1484/J.BPM.5.110808</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+        <datafield tag="775" ind1=" " ind2=" ">
+            <subfield code="o">https://doi.org710.35662/unine-thesis-2747</subfield>
+            <subfield code="g">version publiée</subfield>
+        </datafield>
+    </record>
+    """
+    marc21json = create_record(marc21xml)
+    data = overdo.do(marc21json)
+    assert not data.get('otherEdition')
+    assert data.get('identifiedBy') == [{
+        'type': 'bf:Doi',
+        'value': '10.1002/1521-3773(20020104)41:1'
+    },
+    {
+        'type': 'bf:Doi',
+        'value': '10.1016/j.apergo.2008.03.002'
+    },
+    {
+        'type': 'bf:Doi',
+        'value': '10.1594/PANGAEA.914883'
+    },
+    {
+        'type': 'bf:Doi',
+        'value': '10.1484/J.BPM.5.110808'
+    },
+    {
+        'type': 'bf:Doi',
+        'value': '10.35662/unine-thesis-2747'
+    }]
+
 
 def test_marc21_to_specific_collection(app, bucket_location,
                                        without_oaiset_signals):

diff --git a/tests/unit/documents/loaders/test_rerodoc_loader.py b/tests/unit/documents/loaders/test_rerodoc_loader.py
@@ -105,6 +105,10 @@ def test_rerodoc_loader(app, organisation):
                     <marc:subfield code="g">Publisher's version</marc:subfield>
                     <marc:subfield code="o">https://doi.org/10.1093/mnras/stu2500</marc:subfield>
                 </marc:datafield>
+                <marc:datafield tag="775" ind1=" " ind2=" ">
+                    <marc:subfield code="g">Alternative version</marc:subfield>
+                    <marc:subfield code="o">http://www-geol.unine.ch/GEOMAGNETISME/tract.html</marc:subfield>
+                </marc:datafield>
                 <marc:datafield tag="919" ind1=" " ind2=" ">
                     <marc:subfield code="a">Consortium of Swiss Academic Libraries</marc:subfield>
                     <marc:subfield code="b">Zurich</marc:subfield>
@@ -134,6 +138,11 @@ def test_rerodoc_loader(app, organisation):
             'Swissbib',
             'value':
             '(NATIONALLICENCE)oxford-10.1093/mnras/stu2500'
+        }, {
+            'type':
+            'bf:Doi',
+            'value':
+            '10.1093/mnras/stu2500'
         }],
         'collections': [{
             '$ref': 'https://sonar.ch/api/collections/1'
@@ -167,9 +176,9 @@ def test_rerodoc_loader(app, organisation):
         }],
         'otherEdition': [{
             'document': {
-                'electronicLocator': 'https://doi.org/10.1093/mnras/stu2500'
+                'electronicLocator': 'http://www-geol.unine.ch/GEOMAGNETISME/tract.html'
             },
-            'publicNote': "Publisher's version"
+            'publicNote': "Alternative version"
         }],
         'language': [{
             'type': 'bf:Language',