OA: green access parsing fix

cern-sis · Jul 31, 2024 · e9f5b14 · e9f5b14
1 parent 1ba37db
commit e9f5b14
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 32 deletions.
diff --git a/dags/open_access/parsers.py b/dags/open_access/parsers.py
@@ -89,12 +89,11 @@ def parse_subset_green_access(records):
         is_it_wanted_record_by_540_publication = (
             not is_subset_540_publication_golden_access(datafields_540)
         )
+        green_access_by_field = (
+            is_it_wanted_record_by_856 or is_it_wanted_record_by_540_preprint
+        )
 
-        if (
-            is_it_wanted_record_by_856
-            or is_it_wanted_record_by_540_preprint
-            or is_it_wanted_record_by_540_publication
-        ):
+        if green_access_by_field and is_it_wanted_record_by_540_publication:
             filtered_records.append(record)
 
     return filtered_records
@@ -106,11 +105,7 @@ def parse_subset_golden_access(records):
         datafields_540 = record.findall("datafield/[@tag='540']")
         if datafields_540 is None:
             continue
-        is_it_wanted_record_by_540_publication = (
-            is_subset_540_publication_golden_access(datafields_540)
-        )
-
-        if is_it_wanted_record_by_540_publication:
+        if is_subset_540_publication_golden_access(datafields_540):
             filtered_records.append(record)
     return filtered_records
 

diff --git a/tests/open_access/test_parser.py b/tests/open_access/test_parser.py
@@ -25,33 +25,13 @@
     "2882298",
 ]
 
-expected_green = [
-    "2894668",
-    "2891489",
-    "2891488",
-    "2891487",
-    "2888511",
-    "2888151",
-    "2886038",
-    "2884472",
-    "2884471",
-    "2884470",
-    "2884469",
-    "2883672",
-    "2882429",
-    "2882335",
-    "2882328",
-    "2882327",
-    "2882324",
-    "2882322",
-    "2882311",
-    "2882298",
-]
+expected_green = ["2891489", "2891487", "2886038", "2884472", "2884469"]
 
 
 def test_get_golden_access_records_dois(shared_datadir):
     with open(shared_datadir / "search.xml") as file:
         records_ids = get_golden_access_records_ids(file.read())
+        print()
         assert records_ids == expected_golden
 
 
@@ -107,4 +87,5 @@ def test_parse_subset_540_publications(shared_datadir):
 def test_get_green_access_records_dois(shared_datadir):
     with open(shared_datadir / "search.xml") as file:
         records_ids = get_green_access_records_ids(file.read())
+        print(records_ids)
         assert records_ids == expected_green