Fix Tag Integer Enums (#1388)

backend/alembic/versions/fad14119fb92_delete_tags_with_wrong_enum.py

@@ -0,0 +1,39 @@
+"""Delete Tags with wrong Enum
+
+Revision ID: fad14119fb92
+Revises: 72bdc9929a46
+Create Date: 2024-04-25 17:05:09.695703
+
+"""
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "fad14119fb92"
+down_revision = "72bdc9929a46"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # Some documents may lose their tags but this is the only way as the enum
+    # mapping may have changed since tag switched to string (it will be reindexed anyway)
+    op.execute(
+        """
+        DELETE FROM document__tag
+        WHERE tag_id IN (
+            SELECT id FROM tag
+            WHERE source ~ '^[0-9]+$'
+        )
+        """
+    )
+
+    op.execute(
+        """
+        DELETE FROM tag
+        WHERE source ~ '^[0-9]+$'
+        """
+    )
+
+
+def downgrade() -> None:
+    pass

backend/danswer/db/models.py

@@ -1063,7 +1063,9 @@ class PermissionSyncRun(Base):
 
     id: Mapped[int] = mapped_column(Integer, primary_key=True)
     # Not strictly needed but makes it easy to use without fetching from cc_pair
-    source_type: Mapped[DocumentSource] = mapped_column(Enum(DocumentSource))
+    source_type: Mapped[DocumentSource] = mapped_column(
+        Enum(DocumentSource, native_enum=False)
+    )
     # Currently all sync jobs are handled as a group permission sync or a user permission sync
     update_type: Mapped[PermissionSyncJobType] = mapped_column(
         Enum(PermissionSyncJobType)
@@ -1093,7 +1095,9 @@ class ExternalPermission(Base):
     # Email is needed because we want to keep track of users not in Danswer to simplify process
     # when the user joins
     user_email: Mapped[str] = mapped_column(String)
-    source_type: Mapped[DocumentSource] = mapped_column(Enum(DocumentSource))
+    source_type: Mapped[DocumentSource] = mapped_column(
+        Enum(DocumentSource, native_enum=False)
+    )
     external_permission_group: Mapped[str] = mapped_column(String)
     user = relationship("User")
 
@@ -1120,6 +1124,8 @@ class EmailToExternalUserCache(Base):
     # Email is needed because we want to keep track of users not in Danswer to simplify process
     # when the user joins
     user_email: Mapped[str] = mapped_column(String)
-    source_type: Mapped[DocumentSource] = mapped_column(Enum(DocumentSource))
+    source_type: Mapped[DocumentSource] = mapped_column(
+        Enum(DocumentSource, native_enum=False)
+    )
 
     user = relationship("User")