Allow author records with just 'death_date' to count as "dated" (internetarchive#11150)

* allow author records with just 'death_date' to count as "dated"
fixes internetarchive#10736
* refactor a has_dates() method to simplify logic
* explicitly test death_date only matching
for internetarchive#10736

Author: hornc

openlibrary/catalog/add_book/load_book.py

@@ -142,6 +142,9 @@ def find_author(author: dict[str, Any]) -> list["Author"]:
     Searches OL for an author by a range of queries.
     """
 
+    def has_dates(author: "dict | Author") -> bool:
+        return 'birth_date' in author or 'death_date' in author
+
     def walk_redirects(obj, seen):
         seen.add(obj['key'])
         while obj['type']['key'] == '/type/redirect':
@@ -210,17 +213,16 @@ def get_redirected_authors(authors: list["Author"]):
             break
     match = []
     seen = set()
+    # If author has dates, we only consider dated candidates,
+    # otherwise only include undated candidates.
     for a in things:
-        key = a['key']
-        if key in seen:
+        if key := a['key'] in seen:
             continue
         seen.add(key)
-        assert a.type.key == '/type/author'
-        if 'birth_date' in author and 'birth_date' not in a:
+        if has_dates(author) != has_dates(a):
             continue
-        if 'birth_date' not in author and 'birth_date' in a:
-            continue
-        if not author_dates_match(author, a):
+        assert a.type.key == '/type/author'
+        if has_dates(author) and not author_dates_match(author, a):
             continue
         match.append(a)
     if not match:

openlibrary/catalog/utils/__init__.py

@@ -58,7 +58,7 @@ def author_dates_match(a: dict, b: "dict | Author") -> bool:
     Checks if the years of two authors match. Only compares years,
     not names or keys. Works by returning False if any year specified in one record
     does not match that in the other, otherwise True. If any one author does not have
-    dates, it will return True.
+    dates, it will return True (i.e. "possible match").
 
     :param dict a: Author import dict {"name": "Some One", "birth_date": "1960"}
     :param dict b: Author import dict {"name": "Some One"}

openlibrary/catalog/utils/tests/test_catalog_utils.py

@@ -0,0 +1,34 @@
+import pytest
+
+from openlibrary.catalog.utils import author_dates_match
+
+EXISTING = {'birth_date': '1904', 'death_date': '1996'}
+
+
+MATCH_CASES = [
+    {'birth_date': '1904', 'death_date': '1996'},
+    {'death_date': '1996'},
+    {'birth_date': '1904'},
+]
+
+
+NON_MATCH_CASES = [
+    {'birth_date': '1794', 'death_date': '1823'},
+    {'birth_date': '1904', 'death_date': '2005'},  # one date mismatch
+]
+
+
+@pytest.mark.parametrize('a', MATCH_CASES)
+def test_author_dates_match_true(a):
+    assert author_dates_match(a, EXISTING)
+
+
+@pytest.mark.parametrize('a', NON_MATCH_CASES)
+def test_author_dates_match_false(a):
+    assert not author_dates_match(a, EXISTING)
+
+
+def test_author_dates_match_death_only():
+    a = {'death_date': '1996'}
+    b = {'death_date': '1996'}
+    assert author_dates_match(a, b)