Create DSpace metadata from Crossref data

* Add function to create DSpace-compliant metadata from Crossref data
* Add metadata mapping file to config folder
* Add corresponding unit test

* https://mitlibraries.atlassian.net/browse/DLSPP-112

awd/crossref.py

@@ -1,3 +1,5 @@
+import json
+
 import requests
 import smart_open
 
@@ -40,7 +42,7 @@ def get_metadata_dict_from_crossref_work(work):
         if key == "author":
             authors = []
             for author in work["author"]:
-                name = f'{author["family"]}, {author["given"]}'
+                name = f'{author.get("family")}, {author.get("given")}'
                 authors.append(name)
             value_dict[key] = authors
         elif key == "title":
@@ -51,3 +53,19 @@ def get_metadata_dict_from_crossref_work(work):
         else:
             value_dict[key] = work[key]
     return value_dict
+
+
+def create_dspace_metadata_from_dict(value_dict, metadata_mapping_path):
+    """Create DSpace JSON metadata from metadata dict and a JSON metadata mapping file."""
+    with open(metadata_mapping_path, "r") as metadata_mapping:
+        metadata_mapping = json.load(metadata_mapping)
+        metadata = []
+        for key in [k for k in metadata_mapping if k in value_dict.keys()]:
+            if isinstance(value_dict[key], list):
+                for list_item in value_dict[key]:
+                    metadata.append({"key": metadata_mapping[key], "value": list_item})
+            else:
+                metadata.append(
+                    {"key": metadata_mapping[key], "value": value_dict[key]}
+                )
+        return {"metadata": metadata}

config/metadata_mapping.json

@@ -0,0 +1,15 @@
+{
+  "author": "dc.contributor.author",
+  "container-title": "dc.relation.journal",
+  "ISSN": "dc.identifier.issn",
+  "issue": "mit.journal.issue",
+  "issued": "dc.date.issued",
+  "language": "dc.langauge",
+  "original-title": "dc.title.alternative",
+  "publisher": "dc.publisher",
+  "short-title": "dc.title.alternative",
+  "subtitle": "dc.title.alternative",
+  "title": "dc.title",
+  "URL": "dc.relation.isversionof",
+  "volume": "mit.journal.volume"
+}

tests/fixtures/dspace_metadata.json

@@ -0,0 +1,91 @@
+{
+  "metadata": [{
+      "key": "dc.contributor.author",
+      "value": "Eivazzadeh‐Keihan, Reza"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Bahojb Noruzi, Ehsan"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Khanmohammadi Chenab, Karim"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Jafari, Amir"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Radinekiyan, Fateme"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Hashemi, Seyed Masoud"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Ahmadpour, Farnoush"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Behboudi, Ali"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Mosafer, Jafar"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Mokhtarzadeh, Ahad"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Maleki, Ali"
+    },
+    {
+      "key": "dc.contributor.author",
+      "value": "Hamblin, Michael R."
+    },
+    {
+      "key": "dc.relation.journal",
+      "value": "Journal of Tissue Engineering and Regenerative Medicine"
+    },
+    {
+      "key": "dc.identifier.issn",
+      "value": "1932-6254"
+    },
+    {
+      "key": "dc.identifier.issn",
+      "value": "1932-7005"
+    },
+    {
+      "key": "mit.journal.issue",
+      "value": "12"
+    },
+    {
+      "key": "dc.date.issued",
+      "value": "2020-09-30"
+    },
+    {
+      "key": "dc.langauge",
+      "value": "en"
+    },
+    {
+      "key": "dc.publisher",
+      "value": "Wiley"
+    },
+    {
+      "key": "dc.title",
+      "value": "Metal‐based nanoparticles for bone tissue engineering"
+    },
+    {
+      "key": "dc.relation.isversionof",
+      "value": "http://dx.doi.org/10.1002/term.3131"
+    },
+    {
+      "key": "mit.journal.volume",
+      "value": "14"
+    }
+  ]
+}

tests/test_crossref.py

@@ -2,7 +2,7 @@
 
 
 def test_get_dois_from_spreadsheet():
-    dois = crossref.get_dois_from_spreadsheet("fixtures/test.csv")
+    dois = crossref.get_dois_from_spreadsheet("tests/fixtures/test.csv")
     for doi in dois:
         assert doi == "10.1002/term.3131"
 
@@ -30,23 +30,22 @@ def test_create_dspace_metadata_from_dict_minimum_metadata():
     metadata = crossref.create_dspace_metadata_from_dict(
         value_dict, "config/metadata_mapping.json"
     )
-    assert value_dict["publisher"] == "Wiley"
-    assert value_dict["author"] == [
-        "Eivazzadeh‐Keihan, Reza",
-        "Bahojb Noruzi, Ehsan",
-        "Khanmohammadi Chenab, Karim",
-        "Jafari, Amir",
-        "Radinekiyan, Fateme",
-        "Hashemi, Seyed Masoud",
-        "Ahmadpour, Farnoush",
-        "Behboudi, Ali",
-        "Mosafer, Jafar",
-        "Mokhtarzadeh, Ahad",
-        "Maleki, Ali",
-        "Hamblin, Michael R.",
+    assert metadata["metadata"] == [
+        {
+            "key": "dc.title",
+            "value": "Metal‐based nanoparticles for bone tissue engineering",
+        },
+        {
+            "key": "dc.relation.isversionof",
+            "value": "http://dx.doi.org/10.1002/term.3131",
+        },
     ]
-    assert value_dict["URL"] == "http://dx.doi.org/10.1002/term.3131"
-    assert value_dict["container-title"] == [
-        "Journal of Tissue Engineering and Regenerative Medicine"
-    ]
-    assert value_dict["issued"] == "2020-09-30"
+
+
+def test_transform_dict_with_metadata_mapping_full_metadata(
+    crossref_value_dict, dspace_metadata
+):
+    metadata = crossref.create_dspace_metadata_from_dict(
+        crossref_value_dict, "config/metadata_mapping.json"
+    )
+    assert metadata == dspace_metadata