BookDataModel.merge_into: return and log absorbed fields

bookwyrm/management/commands/deduplicate_book_data.py

@@ -30,7 +30,8 @@ def dedupe_model(model):
             print(f"merging into {canonical.remote_id} based on {field.name} {value}:")
             for obj in objs[1:]:
                 print(f"- {obj.remote_id}")
-                obj.merge_into(canonical)
+                absorbed_fields = obj.merge_into(canonical)
+                print(f"  absorbed fields: {absorbed_fields}")
 
 
 class Command(BaseCommand):

bookwyrm/management/merge_command.py

@@ -25,4 +25,6 @@ def handle(self, *args, **options):
             print("other book doesn’t exist!")
             return
 
-        other.merge_into(canonical)
+        absorbed_fields = other.merge_into(canonical)
+        print(f"{other.remote_id} has been merged into {canonical.remote_id}")
+        print(f"absorbed fields: {absorbed_fields}")

bookwyrm/models/book.py

@@ -2,7 +2,7 @@
 
 from itertools import chain
 import re
-from typing import Any
+from typing import Any, Dict
 from typing_extensions import Self
 
 from django.contrib.postgres.search import SearchVectorField
@@ -108,12 +108,12 @@ def broadcast(self, activity, sender, software="bookwyrm", **kwargs):
         """only send book data updates to other bookwyrm instances"""
         super().broadcast(activity, sender, software=software, **kwargs)
 
-    def merge_into(self, canonical: Self) -> None:
+    def merge_into(self, canonical: Self) -> Dict[str, Any]:
         """merge this entity into another entity"""
         if canonical.id == self.id:
             raise ValueError(f"Cannot merge {self} into itself")
 
-        canonical.absorb_data_from(self)
+        absorbed_fields = canonical.absorb_data_from(self)
         canonical.save()
 
         self.merged_model.objects.create(deleted_id=self.id, merged_into=canonical)
@@ -145,9 +145,11 @@ def merge_into(self, canonical: Self) -> None:
                     getattr(related_obj, related_field).remove(self)
 
         self.delete()
+        return absorbed_fields
 
-    def absorb_data_from(self, other: Self) -> None:
+    def absorb_data_from(self, other: Self) -> Dict[str, Any]:
         """fill empty fields with values from another entity"""
+        absorbed_fields = {}
         for data_field in self._meta.get_fields():
             if not hasattr(data_field, "activitypub_field"):
                 continue
@@ -156,6 +158,8 @@ def absorb_data_from(self, other: Self) -> None:
                 continue
             if not getattr(self, data_field.name):
                 setattr(self, data_field.name, data_value)
+                absorbed_fields[data_field.name] = data_value
+        return absorbed_fields
 
 
 class MergedBookDataModel(models.Model):