fix: Implement dictionary merging in merge_property to correctly combine remote_ids during author merges. (#11747)

Author: RayBB

openlibrary/plugins/upstream/merge_authors.py

@@ -115,6 +115,12 @@ def merge_docs(self, master, dup):
     def merge_property(self, a, b):
         if isinstance(a, list) and isinstance(b, list):
             return uniq(a + b, key=dicthash)
+        elif isinstance(a, dict) and isinstance(b, dict):
+            # Merge dictionaries (e.g., remote_ids)
+            # Master (a) values take preference over duplicate (b) values
+            result = b.copy()
+            result.update(a)
+            return result
         elif not a:
             return b
         else:

openlibrary/plugins/upstream/tests/test_merge_authors.py

@@ -126,6 +126,16 @@ def test_merge_property(self):
         assert engine.merge_property("foo", "bar") == "foo"
         assert engine.merge_property(["foo"], ["bar"]) == ["foo", "bar"]
         assert engine.merge_property(None, ["bar"]) == ["bar"]
+        # Test dict merging (for remote_ids)
+        assert engine.merge_property({}, {"wikidata": "Q123"}) == {"wikidata": "Q123"}
+        assert engine.merge_property({"viaf": "123"}, {"wikidata": "Q456"}) == {
+            "viaf": "123",
+            "wikidata": "Q456",
+        }
+        # When both have the same key, master's value should take preference
+        assert engine.merge_property({"wikidata": "Q111"}, {"wikidata": "Q222"}) == {
+            "wikidata": "Q111"
+        }
 
 
 def test_get_many():
@@ -297,6 +307,30 @@ def test_work_authors(self):
             "authors": [{"type": "/type/author_role", "author": {"key": "/authors/a"}}],
         }
 
+    def test_remote_ids_merge_wikidata_from_duplicate(self):
+        """Test that Wikidata ID from duplicate author is merged into master.
+
+        This reproduces the bug where remote_ids.wikidata was lost during merge.
+        See: https://github.com/internetarchive/openlibrary/issues/11698
+        """
+        # Master author has no remote_ids (or empty dict)
+        a = dict(TEST_AUTHORS.a)
+
+        # Duplicate author has a Wikidata ID
+        b = dict(TEST_AUTHORS.b, remote_ids={"wikidata": "Q12345"})
+
+        c = dict(TEST_AUTHORS.c, remote_ids={"wikidata": "Q12346", "viaf": "123456"})
+
+        web.ctx.site.add([a, b, c])
+        self.engine.merge("/authors/a", ["/authors/b", "/authors/c"])
+
+        # The Wikidata ID should be merged into the master
+        master_remote_ids = dict(web.ctx.site.get("/authors/a").get('remote_ids'))
+        assert master_remote_ids == {
+            "wikidata": "Q12345",
+            "viaf": "123456",
+        }, "remote_ids from duplicate should be merged into master"
+
 
 def test_dicthash():
     assert dicthash({}) == dicthash({})