BookDataModel: add dry_run argument to merge_into

bookwyrm/management/commands/deduplicate_book_data.py

@@ -6,7 +6,7 @@ from django.db.models import Count
 from bookwyrm import models
 
 
-def dedupe_model(model):
+def dedupe_model(model, dry_run=False):
     """combine duplicate editions and update related models"""
     print(f"deduplicating {model.__name__}:")
     fields = model._meta.get_fields()
@@ -27,10 +27,13 @@ def dedupe_model(model):
             print("----------")
             objs = model.objects.filter(**{field.name: value}).order_by("id")
             canonical = objs.first()
-            print(f"merging into {canonical.remote_id} based on {field.name} {value}:")
+            action = "would merge" if dry_run else "merging"
+            print(
+                f"{action} into {model.__name__} {canonical.remote_id} based on {field.name} {value}:"
+            )
             for obj in objs[1:]:
                 print(f"- {obj.remote_id}")
-                absorbed_fields = obj.merge_into(canonical)
+                absorbed_fields = obj.merge_into(canonical, dry_run=dry_run)
                 print(f"  absorbed fields: {absorbed_fields}")
 
 
@@ -39,9 +42,17 @@ class Command(BaseCommand):
 
     help = "merges duplicate book data"
 
+    def add_arguments(self, parser):
+        """add the arguments for this command"""
+        parser.add_argument(
+            "--dry_run",
+            action="store_true",
+            help="don't actually merge, only print what would happen",
+        )
+
     # pylint: disable=no-self-use,unused-argument
     def handle(self, *args, **options):
         """run deduplications"""
-        dedupe_model(models.Edition)
-        dedupe_model(models.Work)
-        dedupe_model(models.Author)
+        dedupe_model(models.Edition, dry_run=options["dry_run"])
+        dedupe_model(models.Work, dry_run=options["dry_run"])
+        dedupe_model(models.Author, dry_run=options["dry_run"])

bookwyrm/management/merge_command.py

@@ -8,6 +8,11 @@ class MergeCommand(BaseCommand):
         """add the arguments for this command"""
         parser.add_argument("--canonical", type=int, required=True)
         parser.add_argument("--other", type=int, required=True)
+        parser.add_argument(
+            "--dry_run",
+            action="store_true",
+            help="don't actually merge, only print what would happen",
+        )
 
     # pylint: disable=no-self-use,unused-argument
     def handle(self, *args, **options):
@@ -25,6 +30,8 @@ class MergeCommand(BaseCommand):
             print("other book doesn’t exist!")
             return
 
-        absorbed_fields = other.merge_into(canonical)
-        print(f"{other.remote_id} has been merged into {canonical.remote_id}")
+        absorbed_fields = other.merge_into(canonical, dry_run=options["dry_run"])
+
+        action = "would be" if options["dry_run"] else "has been"
+        print(f"{other.remote_id} {action} merged into {canonical.remote_id}")
         print(f"absorbed fields: {absorbed_fields}")

bookwyrm/models/book.py

@@ -110,12 +110,16 @@ class BookDataModel(ObjectMixin, BookWyrmModel):
         """only send book data updates to other bookwyrm instances"""
         super().broadcast(activity, sender, software=software, **kwargs)
 
-    def merge_into(self, canonical: Self) -> Dict[str, Any]:
+    def merge_into(self, canonical: Self, dry_run=False) -> Dict[str, Any]:
         """merge this entity into another entity"""
         if canonical.id == self.id:
             raise ValueError(f"Cannot merge {self} into itself")
 
-        absorbed_fields = canonical.absorb_data_from(self)
+        absorbed_fields = canonical.absorb_data_from(self, dry_run=dry_run)
+
+        if dry_run:
+            return absorbed_fields
+
         canonical.save()
 
         self.merged_model.objects.create(deleted_id=self.id, merged_into=canonical)
@@ -149,7 +153,7 @@ class BookDataModel(ObjectMixin, BookWyrmModel):
         self.delete()
         return absorbed_fields
 
-    def absorb_data_from(self, other: Self) -> Dict[str, Any]:
+    def absorb_data_from(self, other: Self, dry_run=False) -> Dict[str, Any]:
         """fill empty fields with values from another entity"""
         absorbed_fields = {}
         for data_field in self._meta.get_fields():
@@ -162,7 +166,8 @@ class BookDataModel(ObjectMixin, BookWyrmModel):
             if isinstance(data_field, fields.ArrayField):
                 if new_values := list(set(other_value) - set(canonical_value)):
                     # append at the end (in no particular order)
-                    setattr(self, data_field.name, canonical_value + new_values)
+                    if not dry_run:
+                        setattr(self, data_field.name, canonical_value + new_values)
                     absorbed_fields[data_field.name] = new_values
             elif isinstance(data_field, fields.PartialDateField):
                 if (
@@ -170,11 +175,13 @@ class BookDataModel(ObjectMixin, BookWyrmModel):
                     or (other_value.has_day and not canonical_value.has_day)
                     or (other_value.has_month and not canonical_value.has_month)
                 ):
-                    setattr(self, data_field.name, other_value)
+                    if not dry_run:
+                        setattr(self, data_field.name, other_value)
                     absorbed_fields[data_field.name] = other_value
             else:
                 if not canonical_value:
-                    setattr(self, data_field.name, other_value)
+                    if not dry_run:
+                        setattr(self, data_field.name, other_value)
                     absorbed_fields[data_field.name] = other_value
         return absorbed_fields