datacite: mark additional files as stub

4 files changed, 77 insertions, 1 deletions

diff --git a/python/fatcat_tools/importers/datacite.py b/python/fatcat_tools/importers/datacite.py
index 587a65aa..90bc3db7 100644
--- a/python/fatcat_tools/importers/datacite.py
+++ b/python/fatcat_tools/importers/datacite.py
@@ -547,6 +547,10 @@ class DataciteImporter(EntityImporter):
         if publisher == 'Cambridge Crystallographic Data Centre':
             release_type = 'entry'
 
+        # Supplement files, e.g. "Additional file 1: ASE constructs in questionnaire."
+        if title.lower().startswith('additional file'):
+            release_type = 'stub'
+
         # Language values are varied ("ger", "es", "English", "ENG", "en-us",
         # "other", ...). Try to crush it with langcodes: "It may sound to you
         # like langcodes solves a pretty boring problem. At one level, that's
diff --git a/python/tests/files/datacite/datacite_doc_25.json b/python/tests/files/datacite/datacite_doc_25.json
new file mode 100644
index 00000000..60cd0ab7
--- /dev/null
+++ b/python/tests/files/datacite/datacite_doc_25.json
@@ -0,0 +1,47 @@
+{
+  "attributes": {
+    "doi": "10.7916/d86x0cg1",
+    "creators": [
+      {
+        "name": "Anton Welch",
+        "affiliation": [
+          "Department of pataphysics"
+        ],
+        "nameIdentifiers": []
+      }
+    ],
+    "titles": [
+      {
+        "title": "Additional file 123: ABC"
+      },
+      {
+        "title": "DEF",
+        "titleType": "Subtitle"
+      }
+    ],
+    "publicationYear": 2016,
+    "language": "DE-CH",
+    "types": {
+      "ris": "GEN",
+      "bibtex": "misc",
+      "citeproc": "article",
+      "schemaOrg": "CreativeWork"
+    },
+    "dates": [
+      {
+        "date": "2017-08-24",
+        "dateType": "Created"
+      },
+      {
+        "date": "2019-08-04",
+        "dateType": "Updated"
+      },
+      {
+        "date": "2017",
+        "dateType": "Issued"
+      }
+    ],
+    "isActive": true,
+    "state": "findable"
+  }
+}
diff --git a/python/tests/files/datacite/datacite_result_25.json b/python/tests/files/datacite/datacite_result_25.json
new file mode 100644
index 00000000..8a370bbb
--- /dev/null
+++ b/python/tests/files/datacite/datacite_result_25.json
@@ -0,0 +1,25 @@
+{
+  "extra": {
+    "datacite": {},
+    "month": 8
+  },
+  "title": "Additional file 123: ABC",
+  "subtitle": "DEF",
+  "release_type": "stub",
+  "release_stage": "published",
+  "release_date": "2017-08-24",
+  "release_year": 2017,
+  "ext_ids": {
+    "doi": "10.7916/d86x0cg1"
+  },
+  "contribs": [
+    {
+      "index": 0,
+      "raw_name": "Anton Welch",
+      "role": "author",
+      "raw_affiliation": "Department of pataphysics"
+    }
+  ],
+  "refs": [],
+  "abstracts": []
+}
diff --git a/python/tests/import_datacite.py b/python/tests/import_datacite.py
index 9ee479e8..7293ecac 100644
--- a/python/tests/import_datacite.py
+++ b/python/tests/import_datacite.py
@@ -287,7 +287,7 @@ def test_datacite_conversions(datacite_importer):
     for now.
     """
     datacite_importer.debug = True
-    for i in range(25):
+    for i in range(26):
         src = 'tests/files/datacite/datacite_doc_{0:02d}.json'.format(i)
         dst = 'tests/files/datacite/datacite_result_{0:02d}.json'.format(i)
         print('testing mapping from {} => {}'.format(src, dst))