Refine the look_for_existing_package to handle more cases #295

Signed-off-by: tdruez <[email protected]>

Author: tdruez

product_portfolio/importers.py

@@ -8,6 +8,7 @@
 
 import json
 from collections import defaultdict
+from contextlib import suppress
 
 from django import forms
 from django.core.exceptions import ValidationError
@@ -768,22 +769,34 @@ def import_dependency(self, dependency_data):
 
     def look_for_existing_package(self, package_data):
         package = None
-        unique_together_lookups = {
+        package_qs = Package.objects.scope(self.user.dataspace)
+
+        # 1. Check if the Package already exists in the local Dataspace
+        # Using exact match first: purl + download_url + filename
+        exact_match_lookups = {
             field: package_data.get(field, "") for field in self.unique_together_fields
         }
-
-        # Check if the Package already exists in the local Dataspace
-        try:
-            package = Package.objects.scope(self.user.dataspace).get(**unique_together_lookups)
+        with suppress(ObjectDoesNotExist):
+            package = package_qs.get(**exact_match_lookups)
             self.existing["package"].append(str(package))
-        except ObjectDoesNotExist:
-            package = None
+            return package
+
+        # 2. If the package data does not include a download_url value:
+        # Attemp to find an existing package using purl-only match.
+        if not package_data.get("download_url"):
+            purl_lookups = {field: package_data.get(field, "") for field in PACKAGE_URL_FIELDS}
+            same_purl_packages = package_qs.filter(**purl_lookups)
+            if len(same_purl_packages) == 1:
+                package = same_purl_packages[0]
+                self.existing["package"].append(str(package))
+                return package
 
         # Check if the Package already exists in the reference Dataspace
+        # Using exact match only: purl + download_url + filename
         reference_dataspace = Dataspace.objects.get_reference()
         user_dataspace = self.user.dataspace
         if not package and user_dataspace != reference_dataspace:
-            qs = Package.objects.scope(reference_dataspace).filter(**unique_together_lookups)
+            qs = Package.objects.scope(reference_dataspace).filter(**exact_match_lookups)
             if qs.exists():
                 reference_object = qs.first()
                 try: