Add sha256 uniqueness to CollectionVersion

This should be ZDU compatible. The transition will be completed with the
next y-release.

fixes: #1052

Co-authored-by: Matthias Dellweg <[email protected]>

CHANGES/1052.feature

@@ -0,0 +1,2 @@
+CollectionVersion global uniqueness constraint is now its sha256 digest. Repository level uniqueness
+is still (namespace, name, version).

pulp_ansible/app/management/commands/datarepair-ansible-collection-sha256.py

@@ -0,0 +1,56 @@
+from gettext import gettext as _
+
+from django.core.management import BaseCommand
+from django.db import transaction
+
+from pulp_ansible.app.models import CollectionVersion
+
+
+class Command(BaseCommand):
+    """
+    Django management command to repair ansible collection versions without sha256.
+    """
+
+    help = (
+        "This script repairs ansible collection versions without sha256 if artifacts are available."
+    )
+
+    def add_arguments(self, parser):
+        """Set up arguments."""
+        parser.add_argument(
+            "--dry-run",
+            action="store_true",
+            help=_("Don't modify anything, just collect results."),
+        )
+
+    def handle(self, *args, **options):
+        dry_run = options["dry_run"]
+        failed_units = 0
+        repaired_units = 0
+
+        unit_qs = CollectionVersion.objects.filter(sha256__isnull=True)
+        count = unit_qs.count()
+        print(f"CollectionVersions to repair: {count}")
+        if count == 0:
+            return
+
+        for unit in unit_qs.iterator():
+            try:
+                content_artifact = unit.contentartifact_set.get()
+                artifact = content_artifact.artifact
+                unit.sha256 = artifact.sha256
+
+                if not dry_run:
+                    with transaction.atomic():
+                        unit.save(update_fields=["sha256"])
+            except Exception as e:
+                failed_units += 1
+                print(
+                    f"Failed to migrate collection version '{unit.namespace}.{unit.name}' "
+                    f"'{unit.version}': {e}"
+                )
+            else:
+                repaired_units += 1
+
+        print(f"Successfully repaired collection versions: {repaired_units}")
+        print(f"Collection versions failed to repair: {failed_units}")

pulp_ansible/app/migrations/0056_collectionversion_sha256.py

@@ -0,0 +1,23 @@
+# Generated by Django 3.2.14 on 2022-07-15 22:51
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('ansible', '0055_alter_collectionversion_version_alter_role_version'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='collectionversion',
+            name='sha256',
+            field=models.CharField(default='', max_length=64, null=True),
+            preserve_default=False,
+        ),
+        migrations.AlterUniqueTogether(
+            name="collectionversion",
+            unique_together={("sha256",), ("namespace", "name", "version")},
+        ),
+    ]

pulp_ansible/app/migrations/0057_collectionversion_sha256_migrate.py

@@ -0,0 +1,57 @@
+# Generated by Django 3.2.14 on 2022-07-21 22:35
+from django.db import migrations
+
+
+def add_sha256_to_current_models(apps, schema_editor):
+    """Adds the sha256 to current CollectionVersion models."""
+    CollectionVersion = apps.get_model("ansible", "CollectionVersion")
+    collection_versions_to_update = []
+    collection_versions_on_demand = []
+
+    for collection_version in (
+        CollectionVersion.objects.prefetch_related(
+            "content_artifacts", "content_artifacts__artifact"
+        )
+        .filter(sha256="")
+        .only("pk", "sha256")
+        .iterator()
+    ):
+        content_artifact = collection_version.contentartifact_set.get()
+        if content_artifact.artifact:
+            collection_version.sha256 = content_artifact.artifact.sha256
+            collection_versions_to_update.append(collection_version)
+        else:
+            collection_versions_on_demand.append(collection_version)
+        if len(collection_versions_to_update) >= 1024:
+            CollectionVersion.objects.bulk_update(
+                collection_versions_to_update,
+                [
+                    "sha256",
+                ],
+            )
+            collection_versions_to_update.clear()
+    # Update remaining collection versions
+    if len(collection_versions_to_update) > 0:
+        CollectionVersion.objects.bulk_update(
+            collection_versions_to_update,
+            [
+                "sha256",
+            ],
+        )
+
+    # If there are on-demand collections then the next migration will fail, so error here with
+    # helpful message on how to fix. No work will be performed by this migration on a second-run.
+    if len(collection_versions_on_demand) > 0:
+        raise Exception(
+            f"On demand collections found. Please remove or upload/sync their data: "
+            f"{[c.pk for c in collection_versions_on_demand]}"
+        )
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("ansible", "0056_collectionversion_sha256"),
+    ]
+
+    operations = [migrations.RunPython(add_sha256_to_current_models, migrations.RunPython.noop)]

pulp_ansible/app/models.py

@@ -161,6 +161,7 @@ class CollectionVersion(Content):
     """
 
     TYPE = "collection_version"
+    repo_key_fields = ("name", "namespace", "version")
 
     # Data Fields
     authors = psql_fields.ArrayField(models.CharField(max_length=64), default=list, editable=False)
@@ -178,6 +179,7 @@ class CollectionVersion(Content):
     namespace = models.CharField(max_length=64, editable=False)
     repository = models.CharField(default="", blank=True, max_length=2000, editable=False)
     requires_ansible = models.CharField(null=True, max_length=255)
+    sha256 = models.CharField(max_length=64, null=True, blank=False)
 
     version = models.CharField(max_length=128, db_collation="pulp_ansible_semver")
     version_major = models.IntegerField()
@@ -227,7 +229,7 @@ def __str__(self):
 
     class Meta:
         default_related_name = "%(app_label)s_%(model_name)s"
-        unique_together = ("namespace", "name", "version")
+        unique_together = (("namespace", "name", "version"), ("sha256",))
         constraints = [
             UniqueConstraint(
                 fields=("collection", "is_highest"),
@@ -533,6 +535,7 @@ class Meta:
 
     def finalize_new_version(self, new_version):
         """Finalize repo version."""
+        remove_duplicates(new_version)
         removed_collection_versions = new_version.removed(
             base_version=new_version.base_version
         ).filter(pulp_type=CollectionVersion.get_pulp_type())

pulp_ansible/app/serializers.py

@@ -490,8 +490,10 @@ def deferred_validate(self, data):
         # Call super to ensure that data contains artifact
         data = super().deferred_validate(data)
         artifact = data.get("artifact")
-        if (sha256 := data.pop("sha256", None)) and sha256 != artifact.sha256:
+        if (sha256 := data.get("sha256")) and sha256 != artifact.sha256:
             raise ValidationError(_("Expected sha256 did not match uploaded artifact's sha256"))
+        else:
+            data["sha256"] = artifact.sha256
 
         collection_info = process_collection_artifact(
             artifact=artifact,
@@ -509,23 +511,7 @@ def deferred_validate(self, data):
 
     def retrieve(self, validated_data):
         """Reuse existing CollectionVersion if provided artifact matches."""
-        namespace = validated_data["namespace"]
-        name = validated_data["name"]
-        version = validated_data["version"]
-        artifact = self.context["artifact"]
-        # TODO switch this check to use digest when ColVersion uniqueness constraint is changed
-        col = CollectionVersion.objects.filter(
-            namespace=namespace, name=name, version=version
-        ).first()
-        if col:
-            if col._artifacts.get() != artifact:
-                raise ValidationError(
-                    _("Collection {}.{}-{} already exists with a different artifact").format(
-                        namespace, name, version
-                    )
-                )
-
-        return col
+        return CollectionVersion.objects.filter(sha256=validated_data["sha256"]).first()
 
     def create(self, validated_data):
         """Final step in creating the CollectionVersion."""
@@ -549,6 +535,8 @@ class Meta:
             "expected_version",
         )
         model = CollectionVersion
+        # There was an autogenerated validator rendering sha256 required.
+        validators = []
 
 
 class CollectionVersionSerializer(ContentChecksumSerializer, CollectionVersionUploadSerializer):