Workaround a loading issue with cyclonedx-python-lib #1185 (#1186)

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

CHANGELOG.rst

@@ -15,6 +15,10 @@ v34.5.0 (unreleased)
   symbol, string and comments using Pygments.
   https://github.com/nexB/scancode.io/pull/1179
 
+- Workaround an issue with the cyclonedx-python-lib that does not allow to load
+  SBOMs that contains properties with no values.
+  https://github.com/nexB/scancode.io/issues/1185
+
 v34.4.0 (2024-04-22)
 --------------------
 

scanpipe/pipes/cyclonedx.py

@@ -189,6 +189,7 @@ def delete_tools(cyclonedx_document_json):
 
     The new structure is not yet supported by the cyclonedx-python-lib, neither for
     serialization (output) nor deserialization (input).
+    https://github.com/CycloneDX/cyclonedx-python-lib/issues/578
 
     The tools are not used anyway in the context of loading the SBOM component data as
     packages.
@@ -199,6 +200,30 @@ def delete_tools(cyclonedx_document_json):
     return cyclonedx_document_json
 
 
+def delete_empty_dict_property(cyclonedx_document_json):
+    """
+    Remove dict entry where keys are defined but no values are set, such as
+    ``{"name": ""}``.
+
+    Class like cyclonedx.model.contact.OrganizationalEntity raise a
+    NoPropertiesProvidedException while it is not enforced in the spec.
+
+    See https://github.com/CycloneDX/cyclonedx-python-lib/issues/600
+    """
+    entries_to_delete = []
+
+    for component in cyclonedx_document_json["components"]:
+        for property_name, property_value in component.items():
+            if isinstance(property_value, dict) and not any(property_value.values()):
+                entries_to_delete.append((component, property_name))
+
+    # Now delete the keys outside the loop
+    for component, property_name in entries_to_delete:
+        del component[property_name]
+
+    return cyclonedx_document_json
+
+
 def resolve_cyclonedx_packages(input_location):
     """Resolve the packages from the `input_location` CycloneDX document file."""
     input_path = Path(input_location)
@@ -215,7 +240,9 @@ def resolve_cyclonedx_packages(input_location):
                 f'CycloneDX document "{input_path.name}" is not valid:\n{errors}'
             )
             raise ValueError(error_msg)
+
         cyclonedx_document = delete_tools(cyclonedx_document)
+        cyclonedx_document = delete_empty_dict_property(cyclonedx_document)
         cyclonedx_bom = Bom.from_json(data=cyclonedx_document)
 
     else:

scanpipe/pipes/resolve.py

@@ -42,6 +42,16 @@
 """
 
 
+def resolve_manifest_resources(resource, package_registry):
+    """Get package data from resource."""
+    packages = get_packages_from_manifest(resource.location, package_registry) or []
+
+    for package_data in packages:
+        package_data["codebase_resources"] = [resource]
+
+    return packages
+
+
 def get_packages(project, package_registry, manifest_resources, model=None):
     """
     Get package data from package manifests/lockfiles/SBOMs or
@@ -51,15 +61,13 @@ def get_packages(project, package_registry, manifest_resources, model=None):
 
     if not manifest_resources.exists():
         project.add_warning(
-            description="No resources found with package data",
+            description="No resources containing package data found in codebase.",
             model=model,
         )
-        return
+        return []
 
     for resource in manifest_resources:
-        if packages := get_packages_from_manifest(resource.location, package_registry):
-            for package_data in packages:
-                package_data["codebase_resources"] = [resource]
+        if packages := resolve_manifest_resources(resource, package_registry):
             resolved_packages.extend(packages)
         else:
             project.add_error(

scanpipe/tests/test_pipelines.py

@@ -900,7 +900,7 @@ def test_scanpipe_resolve_dependencies_pipeline_integration(self):
         self.assertEqual(1, project1.projectmessages.count())
         message = project1.projectmessages.get()
         self.assertEqual("get_packages_from_manifest", message.model)
-        expected = "No resources found with package data"
+        expected = "No resources containing package data found in codebase."
         self.assertIn(expected, message.description)
 
     def test_scanpipe_resolve_dependencies_pipeline_integration_empty_manifest(self):