Merge branch 'python-dev' into build-pipeline-with-new-library

python/acl_anthology/people/name.py

@@ -83,14 +83,14 @@ def as_bibtex(self) -> str:
         """
         return latex_encode(self.as_last_first())
 
-    def score(self) -> int:
+    def score(self) -> float:
         """
         Returns:
             A score for this name that is intended for comparing different names that generate the same ID.  Names that are more likely to be the correct canonical variant should return higher scores via this function.
         """
         name = self.as_first_last()
         # Prefer longer variants
-        score = len(name)
+        score = float(len(name))
         # Prefer variants with non-ASCII characters
         score += sum((ord(c) > 127) for c in name)
         # Penalize upper-case characters after word boundaries
@@ -103,7 +103,7 @@ def score(self) -> int:
         # intended to make a difference when a person has both "C, A B" and "B
         # C, A" as names)
         if self.first is not None and len(self.first) > len(self.last):
-            score -= 1
+            score += 0.5
         return score
 
     def slugify(self) -> str:

python/tests/people/name_test.py

@@ -185,7 +185,7 @@ def test_name_scoring():
 def test_name_scoring_first_vs_last_name():
     n1 = Name("Chan Tai", "Man")
     n2 = Name("Chan", "Tai Man")
-    assert n1.score() < n2.score()
+    assert n1.score() > n2.score()
 
 
 def test_name_from_string():

python/tests/people/personindex_test.py

@@ -193,6 +193,12 @@ def test_build_personindex_automatically(index_with_full_anthology):
     assert len(persons) == 1
 
 
+def test_canonical_name_is_never_a_variant(index_with_full_anthology):
+    index = index_with_full_anthology
+    for person in index.values():
+        assert person.canonical_name.script is None
+
+
 def test_get_person_coauthors(index_with_full_anthology):
     index = index_with_full_anthology
     person = index.get_by_name(Name("Kathleen", "Dahlgren"))[0]