BF: avoid clobbering of dict(..., **rec) type and always add a record for the author for new dandisets

.github/workflows/cli-integration.yml

@@ -78,9 +78,43 @@ jobs:
         if: matrix.dandi-version == 'release'
         run: pip install "dandi[test]"
 
-      - name: Install dev dandi
-        if: matrix.dandi-version == 'master'
-        run: pip install "dandi[test] @ git+https://github.com/dandi/dandi-cli"
+      - name: Extract CLI PR reference from description
+        if: matrix.dandi-version == 'master' && github.event_name == 'pull_request'
+        id: cli-pr
+        run: |
+          if echo "${{ github.event.pull_request.body }}" | grep -i "Requires_CLI_PR:"; then
+            CLI_PR_URL=$(echo "${{ github.event.pull_request.body }}" | grep -i "Requires_CLI_PR:" | sed 's/.*Requires_CLI_PR:[[:space:]]*//' | head -1 | tr -d '\r\n')
+            echo "CLI PR URL found: $CLI_PR_URL"
+            # Extract PR number from URL (e.g., https://github.com/dandi/dandi-cli/pull/123)
+            CLI_PR_NUM=$(echo "$CLI_PR_URL" | grep -o '[0-9]\+' | tail -1)
+            echo "CLI PR number: $CLI_PR_NUM"
+            echo "cli_pr_num=$CLI_PR_NUM" >> $GITHUB_OUTPUT
+          else
+            echo "No CLI PR reference found in description"
+            echo "cli_pr_num=" >> $GITHUB_OUTPUT
+          fi
+
+      - name: Get CLI PR branch name
+        if: matrix.dandi-version == 'master' && steps.cli-pr.outputs.cli_pr_num != ''
+        id: cli-branch
+        run: |
+          CLI_PR_NUM="${{ steps.cli-pr.outputs.cli_pr_num }}"
+          echo "Getting branch name for CLI PR #$CLI_PR_NUM"
+          BRANCH_NAME=$(curl -s "https://api.github.com/repos/dandi/dandi-cli/pulls/$CLI_PR_NUM" | jq -r '.head.ref')
+          echo "Branch name: $BRANCH_NAME"
+          echo "branch_name=$BRANCH_NAME" >> $GITHUB_OUTPUT
+
+      - name: Install dev dandi from specific PR branch
+        if: matrix.dandi-version == 'master' && steps.cli-branch.outputs.branch_name != '' && steps.cli-branch.outputs.branch_name != 'null'
+        run: |
+          echo "Installing dandi from PR branch: ${{ steps.cli-branch.outputs.branch_name }}"
+          pip install "dandi[test] @ git+https://github.com/dandi/dandi-cli@${{ steps.cli-branch.outputs.branch_name }}"
+
+      - name: Install dev dandi from master
+        if: matrix.dandi-version == 'master' && (steps.cli-branch.outputs.branch_name == '' || steps.cli-branch.outputs.branch_name == 'null')
+        run: |
+          echo "Installing dandi from master branch"
+          pip install "dandi[test] @ git+https://github.com/dandi/dandi-cli"
 
       - name: Run dandi-api tests in dandi-cli
         run: |

dandiapi/api/management/commands/createsuperuser.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import os
 from typing import TYPE_CHECKING
 
 from composed_configuration._allauth_support.management.commands import createsuperuser
@@ -11,8 +12,12 @@
     from composed_configuration._allauth_support.createsuperuser import EmailAsUsernameProxyUser
 
 
-def create_usermetadata(instance: EmailAsUsernameProxyUser, *args, **kwargs):
-    UserMetadata.objects.create(user=instance, status=UserMetadata.Status.APPROVED)
+def create_usermetadata(instance: EmailAsUsernameProxyUser, created: bool, **kwargs):  # noqa: FBT001
+    if created and not hasattr(instance, '_usermetadata_created'):
+        UserMetadata.objects.get_or_create(
+            user=instance, defaults={'status': UserMetadata.Status.APPROVED}
+        )
+        instance._usermetadata_created = True  # noqa: SLF001
 
 
 class Command(createsuperuser.Command):
@@ -25,6 +30,24 @@ def handle(self, *args, **kwargs) -> str | None:
         # Save the return value of the parent class function so we can return it later
         return_value = super().handle(*args, **kwargs)
 
+        # Set first_name and last_name from environment variables if provided
+        first_name = os.environ.get('DJANGO_SUPERUSER_FIRST_NAME')
+        last_name = os.environ.get('DJANGO_SUPERUSER_LAST_NAME')
+
+        if first_name or last_name:
+            # Find the user that was just created
+            email = kwargs.get('email') or os.environ.get('DJANGO_SUPERUSER_EMAIL')
+            if email:
+                try:
+                    user = createsuperuser.user_model.objects.get(email=email)
+                    if first_name:
+                        user.first_name = first_name
+                    if last_name:
+                        user.last_name = last_name
+                    user.save()
+                except createsuperuser.user_model.DoesNotExist:
+                    pass
+
         # Disconnect the signal handler. This isn't strictly necessary, but this avoids any
         # unexpected behavior if, for example, someone extends this command and doesn't
         # realize there's a signal handler attached dynamically.

dandiapi/api/services/version/metadata.py

@@ -22,7 +22,14 @@ def _normalize_version_metadata(
     version_metadata = {
         'schemaKey': 'Dandiset',
         'schemaVersion': settings.DANDI_SCHEMA_VERSION,
-        'contributor': [
+        'contributor': [],
+        **version_metadata,
+    }
+    # if function did not receive already a record for that name
+    # we create one:
+    if not any(r.get('name') == name for r in version_metadata['contributor']):
+        version_metadata['contributor'].insert(
+            0,
             {
                 'name': name,
                 'email': email,
@@ -31,9 +38,7 @@ def _normalize_version_metadata(
                 'affiliation': [],
                 'includeInCitation': True,
             },
-        ],
-        **version_metadata,
-    }
+        )
     # Run the version_metadata through the pydantic model to automatically include any boilerplate
     # like the access or repository fields
     return PydanticDandiset.model_construct(**version_metadata).model_dump(

dandiapi/api/tests/test_dandiset.py

@@ -541,13 +541,13 @@ def test_dandiset_rest_create_with_contributor(api_client, admin_user):
         # This contributor is different from the admin_user
         'contributor': [
             {
-                'name': 'Jane Doe',
+                'name': 'Doe, Jane',
                 'email': '[email protected]',
                 'roleName': ['dcite:ContactPerson'],
                 'schemaKey': 'Person',
                 'affiliation': [],
                 'includeInCitation': True,
-            }
+            },
         ],
     }
 
@@ -571,7 +571,7 @@ def test_dandiset_rest_create_with_contributor(api_client, admin_user):
             'created': TIMESTAMP_RE,
             'modified': TIMESTAMP_RE,
         },
-        'contact_person': 'Jane Doe',
+        'contact_person': 'Doe, John',
         'embargo_status': 'OPEN',
         'star_count': 0,
         'is_starred': False,
@@ -603,21 +603,32 @@ def test_dandiset_rest_create_with_contributor(api_client, admin_user):
         'version': 'draft',
         'url': url,
         'dateCreated': UTC_ISO_TIMESTAMP_RE,
-        'citation': (f'Jane Doe ({year}) {name} (Version draft) [Data set]. DANDI Archive. {url}'),
+        'citation': (
+            f'Doe, John; Doe, Jane ({year}) {name} '
+            f'(Version draft) [Data set]. DANDI Archive. {url}'
+        ),
         '@context': f'https://raw.githubusercontent.com/dandi/schema/master/releases/{settings.DANDI_SCHEMA_VERSION}/context.json',
         'schemaVersion': settings.DANDI_SCHEMA_VERSION,
         'schemaKey': 'Dandiset',
         'access': [{'schemaKey': 'AccessRequirements', 'status': 'dandi:OpenAccess'}],
         'repository': settings.DANDI_WEB_APP_URL,
         'contributor': [
             {
-                'name': 'Jane Doe',
+                'affiliation': [],
+                'email': '[email protected]',
+                'includeInCitation': True,
+                'name': 'Doe, John',
+                'roleName': ['dcite:ContactPerson'],
+                'schemaKey': 'Person',
+            },
+            {
+                'name': 'Doe, Jane',
                 'email': '[email protected]',
                 'roleName': ['dcite:ContactPerson'],
                 'schemaKey': 'Person',
                 'affiliation': [],
                 'includeInCitation': True,
-            }
+            },
         ],
         'assetsSummary': {
             'schemaKey': 'AssetsSummary',