Merge pull request #2273 from aboutcode-org/make_packages_unqiue

make package_url field unique for PackageV2

Author: TG1999

CHANGELOG.rst

@@ -1,6 +1,11 @@
 Release notes
 =============
 
+Version v38.5.0
+---------------------
+
+- fix: Make package_url field unique for PackageV2
+
 Version v38.4.0
 ---------------------
 

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = vulnerablecode
-version = 38.4.0
+version = 38.5.0
 license = Apache-2.0 AND CC-BY-SA-4.0
 
 # description must be on ONE line https://github.com/pypa/setuptools/issues/1390

vulnerabilities/migrations/0122_auto_20260415_1155.py

@@ -0,0 +1,36 @@
+from django.db import migrations
+from django.db.models import F, Window
+from django.db.models.functions import RowNumber
+
+
+def remove_duplicate_package_urls(apps, schema_editor):
+    PackageV2 = apps.get_model("vulnerabilities", "PackageV2")
+
+    duplicates = (
+        PackageV2.objects
+        .annotate(
+            rn=Window(
+                expression=RowNumber(),
+                partition_by=[F("package_url")],
+                order_by=F("id").desc(),
+            )
+        )
+        .filter(rn__gt=1)
+    )
+
+    BATCH_SIZE = 1000
+    ids = list(duplicates.values_list("id", flat=True))
+
+    for i in range(0, len(ids), BATCH_SIZE):
+        PackageV2.objects.filter(id__in=ids[i:i+BATCH_SIZE]).delete()
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("vulnerabilities", "0121_advisoryv2_is_latest_alter_advisoryv2_advisory_id_and_more"),
+    ]
+
+    operations = [
+        migrations.RunPython(remove_duplicate_package_urls, migrations.RunPython.noop),
+    ]

vulnerabilities/migrations/0123_alter_packagev2_options_alter_packagev2_package_url_and_more.py

@@ -0,0 +1,60 @@
+# Generated by Django 5.2.11 on 2026-04-15 11:59
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("vulnerabilities", "0122_auto_20260415_1155"),
+    ]
+
+    operations = [
+        migrations.AlterModelOptions(
+            name="packagev2",
+            options={
+                "ordering": [
+                    "type",
+                    "namespace",
+                    "name",
+                    "version_rank",
+                    "version",
+                    "qualifiers",
+                    "subpath",
+                ]
+            },
+        ),
+        migrations.AlterField(
+            model_name="packagev2",
+            name="package_url",
+            field=models.CharField(
+                db_index=True,
+                help_text="The Package URL for this package.",
+                max_length=1000,
+                unique=True,
+            ),
+        ),
+        migrations.AlterUniqueTogether(
+            name="packagev2",
+            unique_together={("type", "namespace", "name", "version", "qualifiers", "subpath")},
+        ),
+        migrations.AddIndex(
+            model_name="packagev2",
+            index=models.Index(
+                fields=["type", "namespace", "name"], name="vulnerabili_type_ca0efc_idx"
+            ),
+        ),
+        migrations.AddIndex(
+            model_name="packagev2",
+            index=models.Index(
+                fields=["type", "namespace", "name", "qualifiers", "subpath"],
+                name="vulnerabili_type_c98c98_idx",
+            ),
+        ),
+        migrations.AddIndex(
+            model_name="packagev2",
+            index=models.Index(
+                fields=["type", "namespace", "name", "version"], name="vulnerabili_type_1af1cc_idx"
+            ),
+        ),
+    ]

vulnerabilities/models.py

@@ -3477,7 +3477,8 @@ def from_purl(self, purl: Union[PackageURL, str]):
         """
         Return a new Package given a ``purl`` PackageURL object or PURL string.
         """
-        return PackageV2.objects.create(**purl_to_dict(purl=purl))
+        package, _ = PackageV2.objects.get_or_create(**purl_to_dict(purl=purl))
+        return package
 
 
 class PackageV2(PackageURLMixin):
@@ -3490,6 +3491,7 @@ class PackageV2(PackageURLMixin):
         null=False,
         help_text="The Package URL for this package.",
         db_index=True,
+        unique=True,
     )
 
     plain_package_url = models.CharField(
@@ -3520,6 +3522,24 @@ class PackageV2(PackageURLMixin):
         db_index=True,
     )
 
+    class Meta:
+        unique_together = ["type", "namespace", "name", "version", "qualifiers", "subpath"]
+        ordering = ["type", "namespace", "name", "version_rank", "version", "qualifiers", "subpath"]
+        indexes = [
+            # Index for getting al versions of a package
+            models.Index(fields=["type", "namespace", "name"]),
+            models.Index(fields=["type", "namespace", "name", "qualifiers", "subpath"]),
+            # Index for getting a specific version of a package
+            models.Index(
+                fields=[
+                    "type",
+                    "namespace",
+                    "name",
+                    "version",
+                ]
+            ),
+        ]
+
     def __str__(self):
         return self.package_url
 

vulnerablecode/__init__.py

@@ -14,7 +14,7 @@
 
 import git
 
-__version__ = "38.4.0"
+__version__ = "38.5.0"
 
 
 PROJECT_DIR = Path(__file__).resolve().parent