feat: Management commands to clean tags and nan values

denig/management/commands/cleannan.py

@@ -0,0 +1,19 @@
+from django.core.management.base import BaseCommand
+from django.db.models import Q
+
+from denig.models import Document, Fragment
+
+
+class Command(BaseCommand):
+    help = "Clean up nan values from the database."
+
+    def handle(self, *args, **kwargs):
+        fields_to_clean = ["transcription", "notes"]
+
+        for field in fields_to_clean:
+            nan_filter = Q(**{f"{field}__iexact": "nan"})
+            records = Fragment.objects.filter(nan_filter)
+
+            for record in records:
+                setattr(record, field, "")
+                record.save()

denig/management/commands/fixtags.py

@@ -0,0 +1,29 @@
+from django.core.management.base import BaseCommand
+
+from denig.models import Document
+
+
+class Command(BaseCommand):
+    help = "Split concatenated tags into separate tags"
+
+    def handle(self, *args, **kwargs):
+        objects = Document.objects.all()
+
+        for obj in objects:
+            tags = obj.tags.names()
+            updated = False
+
+            for tag in tags:
+                if " ; " in tag:
+                    split_tags = tag.split(" ; ")
+                    obj.tags.remove(tag)
+
+                    for split_tag in split_tags:
+                        obj.tags.add(split_tag.strip())
+
+                    updated = True
+
+            if updated:
+                obj.save()
+
+        self.stdout.write(self.style.SUCCESS("Successfully split concatenated tags"))