From c70c8f5c24b76ccd9f08374e3fe0d8287ca3fb9d Mon Sep 17 00:00:00 2001
From: Nicola <ntarocco@gmail.com>
Date: Thu, 14 Nov 2024 18:58:05 +0100
Subject: [PATCH] deposit: really remove old user email when present

* It is not enough to remove the `current_user_mail` legacy field from
  the merge comparison, but it must be removed also by the dump when
  merging the deposit and the record.
* Given that the `current_user_mail` field might still be present in
  some records, it should be ignored/removed when loading the record via
  marshmallow.
---
 cds/modules/invenio_deposit/api.py                 | 4 +++-
 cds/modules/records/serializers/schemas/project.py | 8 +++++++-
 cds/modules/records/serializers/schemas/video.py   | 8 +++++++-
 3 files changed, 17 insertions(+), 3 deletions(-)

diff --git a/cds/modules/invenio_deposit/api.py b/cds/modules/invenio_deposit/api.py
index 82fafe42b..b4a1982a1 100644
--- a/cds/modules/invenio_deposit/api.py
+++ b/cds/modules/invenio_deposit/api.py
@@ -188,7 +188,7 @@ def merge_with_published(self):
         args = [lca.dumps(), first.dumps(), self.dumps()]
         for arg in args:
             del arg["$schema"], arg["_deposit"]
-            # pop optional removed key `current_user_mail` when present
+            # skip legacy `current_user_mail` when comparing for merging
             arg.get("_cds", {}).pop("current_user_mail", None)
         args.append({})
         m = Merger(*args)
@@ -196,6 +196,8 @@ def merge_with_published(self):
             m.run()
         except UnresolvedConflictsException:
             raise MergeConflict()
+        # remove legacy `current_user_mail` when merging
+        lca.get("_cds", {}).pop("current_user_mail", None)
         return patch(m.unified_patches, lca)
 
     @index
diff --git a/cds/modules/records/serializers/schemas/project.py b/cds/modules/records/serializers/schemas/project.py
index 06b13ccb1..9cabff3b8 100644
--- a/cds/modules/records/serializers/schemas/project.py
+++ b/cds/modules/records/serializers/schemas/project.py
@@ -19,7 +19,7 @@
 """Project JSON schema."""
 
 from invenio_jsonschemas import current_jsonschemas
-from marshmallow import Schema, fields, post_load
+from marshmallow import Schema, fields, pre_load, post_load
 
 from ....deposit.api import Project, deposit_video_resolver
 from .common import (
@@ -44,6 +44,12 @@ class _CDSSSchema(Schema):
     state = fields.Raw()
     modified_by = fields.Int()
 
+    @pre_load
+    def remove_legacy_fields(self, data, **kwargs):
+        """Remove legacy fields."""
+        data.pop("current_user_mail", None)
+        return data
+
 
 class ProjectDepositSchema(DepositSchema):
     """Project Deposit Schema."""
diff --git a/cds/modules/records/serializers/schemas/video.py b/cds/modules/records/serializers/schemas/video.py
index 1a0597670..c83e2e83a 100644
--- a/cds/modules/records/serializers/schemas/video.py
+++ b/cds/modules/records/serializers/schemas/video.py
@@ -19,7 +19,7 @@
 """Video JSON schema."""
 
 from invenio_jsonschemas import current_jsonschemas
-from marshmallow import Schema, fields, post_load
+from marshmallow import Schema, fields, pre_load, post_load
 
 from ....deposit.api import Video
 from ..fields.datetime import DateString
@@ -47,6 +47,12 @@ class _CDSSSchema(Schema):
     extracted_metadata = fields.Raw()
     modified_by = fields.Int()
 
+    @pre_load
+    def remove_legacy_fields(self, data, **kwargs):
+        """Remove legacy fields."""
+        data.pop("current_user_mail", None)
+        return data
+
 
 class VideoDepositSchema(DepositSchema):
     """Project Deposit Schema."""