countries: adds to country list and update enhancer for country

dags/common/constants.py

@@ -205,6 +205,7 @@
         ("United States", "USA"),
         ("USA", "USA"),
         ("U.S.A", "USA"),
+        ("U.S.A.", "USA"),
         ("America", "USA"),
         ("Uruguay", "Uruguay"),
         ("Uzbekistan", "Uzbekistan"),

dags/common/enhancer.py

@@ -60,9 +60,12 @@ def __construct_authors(self, item):
                     continue
 
                 if not affiliation.get("country"):
-                    affiliation["country"] = parse_country_from_value(affiliation.get("value"))
+                    _parsed_country = parse_country_from_value(affiliation.get("value"))
+                    if _parsed_country:
+                        affiliation["country"] = _parsed_country
 
-                affiliation["country"] = get_country_ISO_name(affiliation["country"])
+                if affiliation.get("country"):
+                    affiliation["country"] = get_country_ISO_name(affiliation["country"])
 
         return item
 

dags/common/utils.py

@@ -290,10 +290,7 @@ def find_country_match_from_mapping(affiliation_value):
             return COUNTRIES_DEFAULT_MAPPING[key]
 
 def get_country_ISO_name(country):
-    if COUNTRIES_DEFAULT_MAPPING[country]:
+    if COUNTRIES_DEFAULT_MAPPING.get(country):
         return COUNTRIES_DEFAULT_MAPPING[country]
-    countries = pycountry.countries.search_fuzzy(country)
-    if len(countries) > 1 or len(countries) == 0:
-        return country
     else:
-        return countries[0].name
+        return country

tests/units/aps/data/json_response_content.json

@@ -27,12 +27,23 @@
           "firstname": "Hailin",
           "surname": "Wang",
           "affiliationIds": ["a1", "a1"]
+        },
+        {
+          "type": "Person",
+          "name": "Tim R. Borel",
+          "firstname": "Tim R.",
+          "surname": "Borel",
+          "affiliationIds": ["a2"]
         }
       ],
       "affiliations": [
         {
           "name": "Department of Physics, University of Oregon, Eugene, Oregon 97403, USA",
           "id": "a1"
+        },
+        {
+          "name": "Department of Physics, Tsinghua University, Beijing 100084",
+          "id": "a2"
         }
       ],
       "date": "2021-04-12",

tests/units/aps/test_aps_parser.py

@@ -91,6 +91,18 @@ def parsed_articles(parser, articles):
                             }
                         ],
                     },
+                    {
+                        "full_name": "Tim R. Borel",
+                        "given_names": "Tim R.",
+                        "surname": "Borel",
+                        "affiliations": [
+                            {
+                                "value": "Department of Physics, Tsinghua University, Beijing 100084",
+                                "organization": "Department of Physics, Tsinghua University",
+                                # "country": "China",
+                            }
+                        ],
+                    },
                 ],
                 [
                     {

tests/units/common/test_enhancer.py

@@ -207,6 +207,66 @@
     "record_creation_date": "2022-05-20T00:00:00",
     "titles": [{"title": "title", "subtitle": "subtitle", "source": publisher}],
 }
+input_with_affiliation_value_only = {
+    "abstract": "this is abstracts",
+    "authors": [
+        {
+            "affiliations": [
+                {
+                    "value": "Department of Physics, Tsinghua University, Beijing 100084",
+                    "organization": "Department of Physics, Tsinghua University",
+                }
+            ],
+            "email": "[email protected]",
+            "full_name": "Test Surname, Test names",
+            "given_names": "Test names",
+            "surname": "Test Surname",
+            "orcid": "Test Id",
+        }
+    ],
+    "copyright_holder": "copyright_holder",
+    "copyright_year": "2020",
+    "copyright_statement": "copyright_statement",
+    "copyright_material": "copyright_material",
+    "date_published": "2022-05-20",
+    "title": "title",
+    "subtitle": "subtitle",
+}
+expected_output_with_affiliation_value_only = {
+    "abstracts": [{"value": "this is abstracts", "source": publisher}],
+    "authors": [
+        {
+            "affiliations": [
+                {
+                    "value": "Department of Physics, Tsinghua University, Beijing 100084",
+                    "organization": "Department of Physics, Tsinghua University",
+                    # "country": "China"
+                }
+            ],
+            "email": "[email protected]",
+            "full_name": "Test Surname, Test names",
+            "given_names": "Test names",
+            "surname": "Test Surname",
+            "orcid": "Test Id",
+        }
+    ],
+    "acquisition_source": {
+        "source": publisher,
+        "method": publisher,
+        "date": "2022-05-20T00:00:00",
+    },
+    "copyright": [
+        {
+            "holder": "copyright_holder",
+            "year": "2020",
+            "statement": "copyright_statement",
+            "material": "copyright_material",
+        }
+    ],
+    "imprints": [{"date": "2022-05-20", "publisher": publisher}],
+    "record_creation_date": "2022-05-20T00:00:00",
+    "titles": [{"title": "title", "subtitle": "subtitle", "source": publisher}],
+}
 
 
 @pytest.mark.parametrize(
@@ -231,6 +291,11 @@
             expected_output_with_affiliation_of_cooperation_agreement_with_CERN,
             publisher,
         ),
+        pytest.param(
+            input_with_affiliation_value_only,
+            expected_output_with_affiliation_value_only,
+            publisher,
+        ),
     ],
 )
 @freeze_time("2022-05-20")