DRAFT: Implement metadata schema in Django & PostgreSQL

backend/README.md

@@ -42,7 +42,7 @@ This folder contains a Django + PostgreSQL backend supporting the EMBER Archive
 1. Enter postgres shell 
     ```
     psql -U postgres
-    ```
+    ````
 1. Inside postgres shell run:
     ```
     CREATE USER ember WITH PASSWORD 'test123'; # password is up to you

backend/apps/core/models/__init__.py

@@ -1,2 +1 @@
-# TODO: Import all models here
-# from .project import EmberProject
+# Import all models here

backend/apps/projects/admin.py

@@ -1,3 +1,10 @@
 from django.contrib import admin
 
+from apps.projects.models import Contributor, EmberProject, Funding, Publication
+
 # Register your models here.
+
+admin.site.register(EmberProject)
+admin.site.register(Contributor)
+admin.site.register(Funding)
+admin.site.register(Publication)

backend/apps/projects/migrations/0001_initial.py

@@ -0,0 +1,91 @@
+# Generated by Django 6.0 on 2026-01-08 16:27
+
+import django.core.validators
+import django.db.models.deletion
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    initial = True
+
+    dependencies = [
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='Contributor',
+            fields=[
+                ('id', models.BigAutoField(primary_key=True, serialize=False)),
+                ('external_identifiers', models.JSONField(blank=True, default=dict, help_text='External identifiers such as ORCID')),
+                ('name', models.CharField(help_text='Full name of the contributor')),
+                ('email', models.EmailField(blank=True, help_text='Contact email address', max_length=254)),
+                ('institution', models.CharField(blank=True, help_text='Institutional affiliation')),
+            ],
+        ),
+        migrations.CreateModel(
+            name='Funding',
+            fields=[
+                ('id', models.BigAutoField(primary_key=True, serialize=False)),
+                ('funding_institute', models.CharField(help_text='Name of the funding institute (e.g., NIH, NSF, DARPA)')),
+                ('award_number', models.CharField(help_text='Funding award or grant number')),
+                ('award_title', models.CharField(blank=True, help_text='Title of the funded award')),
+                ('funding_url', models.URLField(blank=True, help_text='URL to the funding announcement or award page')),
+            ],
+        ),
+        migrations.CreateModel(
+            name='Publication',
+            fields=[
+                ('id', models.BigAutoField(primary_key=True, serialize=False)),
+                ('publication_doi', models.CharField(blank=True, help_text='DOI for the publication', validators=[django.core.validators.RegexValidator(message='This field must be formatted as DOI', regex='^10\\\\.\\\\d{4,9}/[-._;()/:A-Z0-9]+$')])),
+                ('title', models.CharField(help_text='Title of the publication')),
+                ('journal', models.CharField(blank=True, help_text='Journal or venue where the work was published')),
+                ('year', models.IntegerField(blank=True, help_text='Year of publication', validators=[django.core.validators.RegexValidator(message='This field must be a valid year', regex='\\d{4}')])),
+                ('publication_url', models.URLField(blank=True, help_text='URL to the publication landing page or preprint')),
+            ],
+        ),
+        migrations.CreateModel(
+            name='EmberProject',
+            fields=[
+                ('id', models.BigAutoField(primary_key=True, serialize=False)),
+                ('project_id', models.CharField(help_text='Project identifier')),
+                ('project_title', models.CharField(help_text='Title of the project')),
+                ('project_description', models.TextField(help_text='Text description of the project')),
+                ('data_use_agreement', models.TextField(blank=True, help_text='Description or reference to the data use agreement governing access to restricted or controlled data')),
+                ('data_use_agreement_required', models.BooleanField(default=False, help_text='Indicates whether a data use agreement is required to access any project data')),
+                ('data_availability_emberdandi', models.BooleanField(default=False, help_text='Indicates if data is available in EMBER-DANDI', verbose_name='Data in EMBER-DANDI?')),
+                ('data_availability_emberrestricted', models.BooleanField(default=False, help_text='Indicates if data is available in EMBERrestricted', verbose_name='Data in EMBERrestrictred?')),
+                ('data_availability_embervault', models.BooleanField(default=False, help_text='Indicates if data is available in EMBERvault', verbose_name='Data in EMBERvault?')),
+                ('access_tier_summary', models.TextField(blank=True, help_text='Text description of different data tiers')),
+                ('last_metadata_update', models.DateTimeField(help_text='Date and time of last metadata update')),
+                ('metadata_version', models.CharField(default='1', help_text='Version of the metadata')),
+                ('ember_doi', models.CharField(blank=True, help_text='DOI in EMBER system (empty string allowed until assigned)', validators=[django.core.validators.RegexValidator(message='This field must be formatted as DOI', regex='^10\\\\.\\\\d{4,9}/[-._;()/:A-Z0-9]+$')])),
+                ('access_level_emberdandisets', models.JSONField(blank=True, default=list, help_text='EMBER-DANDI dandiset ids', validators=[django.core.validators.RegexValidator(message='This field must be a valid EMBER-DANDI id', regex='EMBER-DANDI:\\\\d{6}$')])),
+                ('access_level_restricted_datasets', models.JSONField(blank=True, default=list, help_text='Restricted dataset ids')),
+                ('access_level_access_vault_ids', models.JSONField(blank=True, default=list, help_text='EMBERvault dataset ids')),
+                ('related_repositories', models.JSONField(blank=True, default=list, help_text='Identifiers for related repositories')),
+                ('related_dandisets', models.JSONField(blank=True, default=list, help_text='DOIs for related dandisets')),
+                ('related_data', models.JSONField(blank=True, default=list, help_text='Identifiers for other related datasets')),
+                ('data_administrator', models.ForeignKey(help_text='Primary point of contact for data management', on_delete=django.db.models.deletion.RESTRICT, to='projects.contributor')),
+                ('funding', models.ManyToManyField(blank=True, help_text='Funding sources supporting this project', related_name='projects', to='projects.funding')),
+                ('related_publications', models.ManyToManyField(blank=True, help_text='Publications associated with this project', to='projects.publication')),
+            ],
+            options={
+                'verbose_name': 'EMBER Project',
+                'verbose_name_plural': 'EMBER Projects',
+            },
+        ),
+        migrations.CreateModel(
+            name='PublicationContributor',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('contributor', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, to='projects.contributor')),
+                ('publication', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, to='projects.publication')),
+            ],
+        ),
+        migrations.AddField(
+            model_name='publication',
+            name='authors',
+            field=models.ManyToManyField(blank=True, help_text='Authors of the publication', through='projects.PublicationContributor', to='projects.contributor'),
+        ),
+    ]

backend/apps/projects/models/__init__.py

@@ -1,2 +1,6 @@
-# TODO: Import all models here
-# from .project import EmberProject
+from .contributor import Contributor, PublicationContributor
+from .funding import Funding
+from .project import EmberProject
+from .publication import Publication
+
+__all__ = ["EmberProject", "Contributor", "PublicationContributor", "Funding", "Publication"]

backend/apps/projects/models/contributor.py

@@ -0,0 +1,26 @@
+from django.db import models
+
+
+class Contributor(models.Model):
+    """
+    Person or Organization metadata.
+    """
+
+    id = models.BigAutoField(primary_key=True)
+    external_identifiers = models.JSONField(
+        blank=True, default=dict, help_text="External identifiers such as ORCID"
+    )
+    name = models.CharField(help_text="Full name of the contributor")
+    email = models.EmailField(blank=True, help_text="Contact email address")
+    institution = models.CharField(blank=True, help_text="Institutional affiliation")
+
+
+class PublicationContributor(models.Model):
+    """
+    Person or organization contributing to a project.
+    """
+
+    publication = models.ForeignKey("Publication", on_delete=models.CASCADE)
+    contributor = models.ForeignKey(Contributor, on_delete=models.CASCADE)
+    # TODO: Add role (?)
+    # role = models.CharField(max_length=100, blank=True, help_text="Role of the contributor in this project")

backend/apps/projects/models/funding.py

@@ -0,0 +1,17 @@
+from django.db import models
+
+
+class Funding(models.Model):
+    """
+    Funding source metadata.
+    """
+
+    id = models.BigAutoField(primary_key=True)
+    funding_institute = models.CharField(
+        help_text="Name of the funding institute (e.g., NIH, NSF, DARPA)"
+    )
+    award_number = models.CharField(help_text="Funding award or grant number")
+    award_title = models.CharField(blank=True, help_text="Title of the funded award")
+    funding_url = models.URLField(
+        blank=True, help_text="URL to the funding announcement or award page"
+    )

backend/apps/projects/models/project.py

@@ -1,3 +1,88 @@
 from django.db import models
 
-# TODO: Define EmberProject
+from apps.projects.validators import doi_validator, ember_dandi_id_validator
+
+
+class EmberProject(models.Model):
+    """
+    EMBER Project metadata.
+    """
+
+    id = models.BigAutoField(primary_key=True)
+    project_id = models.CharField(help_text="Project identifier")
+    project_title = models.CharField(help_text="Title of the project")
+    project_description = models.TextField(help_text="Text description of the project")
+    data_use_agreement = models.TextField(
+        blank=True,
+        help_text="Description or reference to the data use agreement governing access to restricted or controlled data",
+    )
+    data_use_agreement_required = models.BooleanField(
+        default=False,
+        help_text="Indicates whether a data use agreement is required to access any project data",
+    )
+    data_availability_emberdandi = models.BooleanField(
+        default=False,
+        verbose_name="Data in EMBER-DANDI?",
+        help_text="Indicates if data is available in EMBER-DANDI",
+    )
+    data_availability_emberrestricted = models.BooleanField(
+        default=False,
+        verbose_name="Data in EMBERrestrictred?",
+        help_text="Indicates if data is available in EMBERrestricted",
+    )
+    data_availability_embervault = models.BooleanField(
+        default=False,
+        verbose_name="Data in EMBERvault?",
+        help_text="Indicates if data is available in EMBERvault",
+    )
+    access_tier_summary = models.TextField(
+        blank=True, help_text="Text description of different data tiers"
+    )
+    data_administrator = models.ForeignKey(
+        "Contributor",
+        on_delete=models.RESTRICT,
+        help_text="Primary point of contact for data management",
+    )
+    last_metadata_update = models.DateTimeField(
+        help_text="Date and time of last metadata update"
+    )  # default=timezone.now
+    metadata_version = models.CharField(default="1", help_text="Version of the metadata")
+    ember_doi = models.CharField(
+        blank=True,
+        help_text="DOI in EMBER system (empty string allowed until assigned)",
+        validators=[doi_validator],
+    )
+    access_level_emberdandisets = models.JSONField(
+        default=list,
+        blank=True,
+        help_text="EMBER-DANDI dandiset ids",
+        validators=[ember_dandi_id_validator],
+    )
+    access_level_restricted_datasets = models.JSONField(
+        default=list, blank=True, help_text="Restricted dataset ids"
+    )
+    access_level_access_vault_ids = models.JSONField(
+        default=list, blank=True, help_text="EMBERvault dataset ids"
+    )
+    related_publications = models.ManyToManyField(
+        "Publication", blank=True, help_text="Publications associated with this project"
+    )
+    related_repositories = models.JSONField(
+        default=list, blank=True, help_text="Identifiers for related repositories"
+    )
+    related_dandisets = models.JSONField(
+        default=list, blank=True, help_text="DOIs for related dandisets"
+    )
+    related_data = models.JSONField(
+        default=list, blank=True, help_text="Identifiers for other related datasets"
+    )
+    funding = models.ManyToManyField(
+        "Funding",
+        related_name="projects",
+        blank=True,
+        help_text="Funding sources supporting this project",
+    )
+
+    class Meta:
+        verbose_name = "EMBER Project"
+        verbose_name_plural = "EMBER Projects"

backend/apps/projects/models/publication.py

@@ -0,0 +1,30 @@
+from django.db import models
+
+from apps.projects.validators import doi_validator, year_validator
+
+
+class Publication(models.Model):
+    """
+    Publication metadata.
+    """
+
+    id = models.BigAutoField(primary_key=True)
+    publication_doi = models.CharField(
+        blank=True, help_text="DOI for the publication", validators=[doi_validator]
+    )
+    title = models.CharField(help_text="Title of the publication")
+    authors = models.ManyToManyField(
+        "Contributor",
+        through="PublicationContributor",
+        blank=True,
+        help_text="Authors of the publication",
+    )
+    journal = models.CharField(
+        blank=True, help_text="Journal or venue where the work was published"
+    )
+    year = models.IntegerField(
+        blank=True, help_text="Year of publication", validators=[year_validator]
+    )
+    publication_url = models.URLField(
+        blank=True, help_text="URL to the publication landing page or preprint"
+    )

backend/apps/projects/serializers.py

@@ -1,11 +1,31 @@
-# from rest_framework import serializers
-
-# TODO: Project Serializer
-# class ProjectSerializer(serializers.ModelSerializer):
-#     class Meta:
-#         model = EmberProject
-#         fields = [
-#             "project_id",
-#             "project_title",
-#             ...
-#         ]
+from rest_framework import serializers
+
+from apps.projects.models import Contributor, EmberProject, Funding, Publication
+
+
+class ContributorSerializer(serializers.ModelSerializer):
+    class Meta:
+        model = Contributor
+        fields = "__all__"
+
+
+class FundingSerializer(serializers.ModelSerializer):
+    class Meta:
+        model = Funding
+        fields = "__all__"
+
+
+class PublicationSerializer(serializers.ModelSerializer):
+    class Meta:
+        model = Publication
+        fields = "__all__"
+
+
+class ProjectSerializer(serializers.ModelSerializer):
+    data_administrator = ContributorSerializer(read_only=True)
+    related_publications = PublicationSerializer(many=True, read_only=True)
+    funding = FundingSerializer(many=True, read_only=True)
+
+    class Meta:
+        model = EmberProject
+        fields = "__all__"

backend/apps/projects/validators.py

@@ -0,0 +1,11 @@
+from django.core.validators import RegexValidator
+
+doi_validator = RegexValidator(
+    regex=r"^10\\.\\d{4,9}/[-._;()/:A-Z0-9]+$", message="This field must be formatted as DOI"
+)
+
+ember_dandi_id_validator = RegexValidator(
+    regex=r"EMBER-DANDI:\\d{6}$", message="This field must be a valid EMBER-DANDI id"
+)
+
+year_validator = RegexValidator(regex=r"\d{4}", message="This field must be a valid year")

backend/apps/projects/views.py

@@ -1,28 +1,31 @@
+from django.shortcuts import get_object_or_404
 from drf_spectacular.utils import OpenApiParameter, extend_schema
 from rest_framework.response import Response
 from rest_framework.views import APIView
 
 from apps.core.serializers import MessageSerializer
+from apps.projects.models.project import EmberProject
+from apps.projects.serializers import ProjectSerializer
 
 
 @extend_schema(tags=["projects"])
 class ProjectsView(APIView):
     @extend_schema(
         operation_id="projects_list",
-        responses={200: MessageSerializer},  # TODO: ProjectSerializer(many=True)
+        responses={200: ProjectSerializer(many=True), 404: MessageSerializer},
         description="Get all EMBER Projects",
     )
     def get(self, request):
-        return Response(
-            {"message": "Placeholder: List all Projects"},
-        )
+        projects_list = EmberProject.objects
+        serializer = ProjectSerializer(projects_list, many=True)
+        return Response(serializer.data)
 
 
 @extend_schema(tags=["projects"])
 class ProjectView(APIView):
     @extend_schema(
         operation_id="projects_retrieve",
-        responses={200: MessageSerializer},  # TODO: ProjectSerializer
+        responses={200: ProjectSerializer, 404: MessageSerializer},
         description="Get an EMBER Project by its project id",
         parameters=[
             OpenApiParameter(
@@ -34,6 +37,6 @@ class ProjectView(APIView):
         ],
     )
     def get(self, request, project_id: str):
-        return Response(
-            {"message": f"Placeholder: Get Project with id={project_id}"},
-        )
+        project = get_object_or_404(EmberProject, project_id=project_id)
+        serializer = ProjectSerializer(project)
+        return Response(serializer.data)