temp commit

tests/unit/services/domains/openalex/test_client.py

@@ -0,0 +1,64 @@
+import pytest
+from app.services.domains.openalex.types import OpenAlexWork
+
+def test_parse_openalex_work_from_api_dict() -> None:
+    raw_api_response = {
+        "id": "https://openalex.org/W2741809807",
+        "doi": "https://doi.org/10.1038/s41586-020-0315-z",
+        "title": "Machine learning and the physical sciences",
+        "publication_year": 2019,
+        "cited_by_count": 1420,
+        "ids": {
+            "openalex": "https://openalex.org/W2741809807",
+            "doi": "https://doi.org/10.1038/s41586-020-0315-z",
+            "mag": "2741809807",
+            "pmid": "https://pubmed.ncbi.nlm.nih.gov/32040050",
+            "pmcid": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7325852"
+        },
+        "open_access": {
+            "is_oa": True,
+            "oa_url": "https://example.com/pdf"
+        },
+        "authorships": [
+            {
+                "author_position": "first",
+                "author": {
+                    "id": "https://openalex.org/A1969205032",
+                    "display_name": "Giuseppe Carleo"
+                }
+            },
+            {
+                "author_position": "middle",
+                "author": {
+                    "id": "https://openalex.org/A4356881717",
+                    "display_name": "Ignacio Cirac"
+                }
+            }
+        ]
+    }
+
+    work = OpenAlexWork.from_api_dict(raw_api_response)
+
+    assert work.openalex_id == "https://openalex.org/W2741809807"
+    assert work.title == "Machine learning and the physical sciences"
+    assert work.doi == "10.1038/s41586-020-0315-z"
+    assert work.pmid == "32040050"
+    assert work.pmcid == "PMC7325852"
+    assert work.publication_year == 2019
+    assert work.cited_by_count == 1420
+    assert work.is_oa is True
+    assert work.oa_url == "https://example.com/pdf"
+    assert len(work.authors) == 2
+    assert work.authors[0].display_name == "Giuseppe Carleo"
+    assert work.authors[1].display_name == "Ignacio Cirac"
+
+def test_parse_openalex_work_empty() -> None:
+    work = OpenAlexWork.from_api_dict({"id": "W123"})
+    assert work.openalex_id == "W123"
+    assert work.doi is None
+    assert work.pmid is None
+    assert work.title is None
+    assert work.publication_year is None
+    assert work.cited_by_count == 0
+    assert work.is_oa is False
+    assert len(work.authors) == 0

tests/unit/services/domains/openalex/test_matching.py

@@ -0,0 +1,55 @@
+import pytest
+from app.services.domains.openalex.types import OpenAlexWork
+from app.services.domains.openalex.matching import find_best_match
+
+def test_find_best_match_exact_title():
+    cand1 = OpenAlexWork.from_api_dict({"id": "W1", "title": "Exact Title of the Paper"})
+    cand2 = OpenAlexWork.from_api_dict({"id": "W2", "title": "Totally Different Paper"})
+    
+    match = find_best_match("Exact Title of the Paper", 2020, "Author A", [cand1, cand2])
+    assert match is not None
+    assert match.openalex_id == "W1"
+
+def test_find_best_match_fuzzy_title():
+    # Only differences are punctuation or minor phrasing (e.g., matching a preprint title vs published)
+    cand1 = OpenAlexWork.from_api_dict({"id": "W1", "title": "Fuzzier Title: A Study on OpenAlex"})
+    cand2 = OpenAlexWork.from_api_dict({"id": "W2", "title": "Some completely unrelated work"})
+    
+    match = find_best_match("Fuzzier Title A Study on OpenAlex", 2021, "Author B", [cand1, cand2])
+    assert match is not None
+    assert match.openalex_id == "W1"
+
+def test_find_best_match_rejects_low_score():
+    cand1 = OpenAlexWork.from_api_dict({"id": "W1", "title": "Cats in hats"})
+    
+    match = find_best_match("Dogs with logs", 2020, "Author A", [cand1])
+    assert match is None
+
+def test_find_best_match_year_tiebreaker():
+    # Both titles are very similar, one has exact year.
+    cand1 = OpenAlexWork.from_api_dict({"id": "W1", "title": "The exact same title", "publication_year": 2018})
+    cand2 = OpenAlexWork.from_api_dict({"id": "W2", "title": "The exact same title", "publication_year": 2020})
+    
+    match = find_best_match("The exact same title", 2020, "Author A", [cand1, cand2])
+    assert match is not None
+    assert match.openalex_id == "W2"
+
+def test_find_best_match_author_tiebreaker():
+    # Titles and years match exactly. Author overlap decides it.
+    cand1 = OpenAlexWork.from_api_dict({
+        "id": "W1", 
+        "title": "A popular title", 
+        "publication_year": 2020,
+        "authorships": [{"author": {"display_name": "Smith, J"}}]
+    })
+    cand2 = OpenAlexWork.from_api_dict({
+        "id": "W2", 
+        "title": "A popular title", 
+        "publication_year": 2020,
+        "authorships": [{"author": {"display_name": "Doe, J"}}]
+    })
+    
+    # Target authors contains "Doe"
+    match = find_best_match("A popular title", 2020, "A Einstein, J Doe", [cand1, cand2])
+    assert match is not None
+    assert match.openalex_id == "W2"