Add management commands to merge a pair of editions or authors

bookwyrm/management/commands/deduplicate_book_data.py

@@ -3,38 +3,7 @@
 from django.core.management.base import BaseCommand
 from django.db.models import Count
 from bookwyrm import models
-
-
-def update_related(canonical, obj):
-    """update all the models with fk to the object being removed"""
-    # move related models to canonical
-    related_models = [
-        (r.remote_field.name, r.related_model) for r in canonical._meta.related_objects
-    ]
-    for (related_field, related_model) in related_models:
-        related_objs = related_model.objects.filter(**{related_field: obj})
-        for related_obj in related_objs:
-            print("replacing in", related_model.__name__, related_field, related_obj.id)
-            try:
-                setattr(related_obj, related_field, canonical)
-                related_obj.save()
-            except TypeError:
-                getattr(related_obj, related_field).add(canonical)
-                getattr(related_obj, related_field).remove(obj)
-
-
-def copy_data(canonical, obj):
-    """try to get the most data possible"""
-    for data_field in obj._meta.get_fields():
-        if not hasattr(data_field, "activitypub_field"):
-            continue
-        data_value = getattr(obj, data_field.name)
-        if not data_value:
-            continue
-        if not getattr(canonical, data_field.name):
-            print("setting data field", data_field.name, data_value)
-            setattr(canonical, data_field.name, data_value)
-    canonical.save()
+from bookwyrm.management.merge import merge_objects
 
 
 def dedupe_model(model):
@@ -61,10 +30,7 @@ def dedupe_model(model):
             print("keeping", canonical.remote_id)
             for obj in objs[1:]:
                 print(obj.remote_id)
-                copy_data(canonical, obj)
-                update_related(canonical, obj)
-                # remove the outdated entry
-                obj.delete()
+                merge_objects(canonical, obj)
 
 
 class Command(BaseCommand):

bookwyrm/management/commands/merge_authors.py

@@ -0,0 +1,12 @@
+""" PROCEED WITH CAUTION: uses deduplication fields to permanently
+merge author data objects """
+from bookwyrm import models
+from bookwyrm.management.merge_command import MergeCommand
+
+
+class Command(MergeCommand):
+    """merges two authors by ID"""
+
+    help = "merges specified authors into one"
+
+    MODEL = models.Author

bookwyrm/management/commands/merge_editions.py

@@ -0,0 +1,12 @@
+""" PROCEED WITH CAUTION: uses deduplication fields to permanently
+merge edition data objects """
+from bookwyrm import models
+from bookwyrm.management.merge_command import MergeCommand
+
+
+class Command(MergeCommand):
+    """merges two editions by ID"""
+
+    help = "merges specified editions into one"
+
+    MODEL = models.Edition

bookwyrm/management/merge.py

@@ -0,0 +1,50 @@
+from django.db.models import ManyToManyField
+
+
+def update_related(canonical, obj):
+    """update all the models with fk to the object being removed"""
+    # move related models to canonical
+    related_models = [
+        (r.remote_field.name, r.related_model) for r in canonical._meta.related_objects
+    ]
+    for (related_field, related_model) in related_models:
+        # Skip the ManyToMany fields that aren’t auto-created. These
+        # should have a corresponding OneToMany field in the model for
+        # the linking table anyway. If we update it through that model
+        # instead then we won’t lose the extra fields in the linking
+        # table.
+        related_field_obj = related_model._meta.get_field(related_field)
+        if isinstance(related_field_obj, ManyToManyField):
+            through = related_field_obj.remote_field.through
+            if not through._meta.auto_created:
+                continue
+        related_objs = related_model.objects.filter(**{related_field: obj})
+        for related_obj in related_objs:
+            print("replacing in", related_model.__name__, related_field, related_obj.id)
+            try:
+                setattr(related_obj, related_field, canonical)
+                related_obj.save()
+            except TypeError:
+                getattr(related_obj, related_field).add(canonical)
+                getattr(related_obj, related_field).remove(obj)
+
+
+def copy_data(canonical, obj):
+    """try to get the most data possible"""
+    for data_field in obj._meta.get_fields():
+        if not hasattr(data_field, "activitypub_field"):
+            continue
+        data_value = getattr(obj, data_field.name)
+        if not data_value:
+            continue
+        if not getattr(canonical, data_field.name):
+            print("setting data field", data_field.name, data_value)
+            setattr(canonical, data_field.name, data_value)
+    canonical.save()
+
+
+def merge_objects(canonical, obj):
+    copy_data(canonical, obj)
+    update_related(canonical, obj)
+    # remove the outdated entry
+    obj.delete()

bookwyrm/management/merge_command.py

@@ -0,0 +1,29 @@
+from bookwyrm.management.merge import merge_objects
+from django.core.management.base import BaseCommand
+
+
+class MergeCommand(BaseCommand):
+    """base class for merge commands"""
+
+    def add_arguments(self, parser):
+        """add the arguments for this command"""
+        parser.add_argument("--canonical", type=int, required=True)
+        parser.add_argument("--other", type=int, required=True)
+
+    # pylint: disable=no-self-use,unused-argument
+    def handle(self, *args, **options):
+        """merge the two objects"""
+        model = self.MODEL
+
+        try:
+            canonical = model.objects.get(id=options["canonical"])
+        except model.DoesNotExist:
+            print("canonical book doesn’t exist!")
+            return
+        try:
+            other = model.objects.get(id=options["other"])
+        except model.DoesNotExist:
+            print("other book doesn’t exist!")
+            return
+
+        merge_objects(canonical, other)