Merge pull request #276 from CycloneDX/fix/bom-validation-nested-components-isue-275

fix: BOM validation fails when Components or Services are nested #275 

fix: updated dependencies #271, #270, #269 and #256

Author: madpah

.isort.cfg

@@ -2,7 +2,7 @@
 ## read the docs: https://pycqa.github.io/isort/docs/configuration/options.html
 ## keep in sync with flake8 config - in `tox.ini` file
 known_first_party = cyclonedx
-skip_gitignore = true
+skip_gitignore = false
 skip_glob =
     build/*,dist/*,__pycache__,.eggs,*.egg-info*,
     *_cache,*.cache,
@@ -15,3 +15,6 @@ ensure_newline_before_comments = true
 include_trailing_comma = true
 line_length = 120
 multi_line_output = 3
+src_paths =
+    cyclonedx
+    tests

.pre-commit-config.yaml

@@ -7,13 +7,13 @@ repos:
         entry: poetry run tox -e mypy
         pass_filenames: false
         language: system
-#  - repo: local
-#    hooks:
-#      - id: system
-#        name: isort
-#        entry: poetry run isort
-#        pass_filenames: false
-#        language: system
+  - repo: local
+    hooks:
+      - id: system
+        name: isort
+        entry: poetry run isort -c .
+        pass_filenames: false
+        language: system
   - repo: local
     hooks:
       - id: system

cyclonedx/model/bom.py

@@ -18,7 +18,7 @@
 # Copyright (c) OWASP Foundation. All Rights Reserved.
 import warnings
 from datetime import datetime, timezone
-from typing import Iterable, Optional
+from typing import Iterable, Optional, Set
 from uuid import UUID, uuid4
 
 from sortedcontainers import SortedSet
@@ -356,6 +356,16 @@ def external_references(self) -> "SortedSet[ExternalReference]":
     def external_references(self, external_references: Iterable[ExternalReference]) -> None:
         self._external_references = SortedSet(external_references)
 
+    def _get_all_components(self) -> Set[Component]:
+        components: Set[Component] = set()
+        if self.metadata.component:
+            components.update(self.metadata.component.get_all_nested_components(include_self=True))
+
+        for c in self.components:
+            components.update(c.get_all_nested_components(include_self=True))
+
+        return components
+
     def has_vulnerabilities(self) -> bool:
         """
         Check whether this Bom has any declared vulnerabilities.
@@ -376,8 +386,8 @@ def validate(self) -> bool:
         """
 
         # 1. Make sure dependencies are all in this Bom.
-        all_bom_refs = set([self.metadata.component.bom_ref] if self.metadata.component else []) | set(
-            map(lambda c: c.bom_ref, self.components)) | set(map(lambda s: s.bom_ref, self.services))
+        all_bom_refs = set(map(lambda c: c.bom_ref, self._get_all_components())) | set(
+            map(lambda s: s.bom_ref, self.services))
 
         all_dependency_bom_refs = set().union(*(c.dependencies for c in self.components))
         dependency_diff = all_dependency_bom_refs - all_bom_refs
@@ -389,9 +399,9 @@ def validate(self) -> bool:
         # 2. Dependencies should exist for the Component this BOM is describing, if one is set
         if self.metadata.component and not self.metadata.component.dependencies:
             warnings.warn(
-                f'The Component this BOM is describing {self.metadata.component.purl} has no defined dependencies'
-                f'which means the Dependency Graph is incomplete - you should add direct dependencies to this Component'
-                f'to complete the Dependency Graph data.',
+                f'The Component this BOM is describing (PURL={self.metadata.component.purl}) has no defined '
+                f'dependencies which means the Dependency Graph is incomplete - you should add direct dependencies to '
+                f'this Component to complete the Dependency Graph data.',
                 UserWarning
             )
 

cyclonedx/model/component.py

@@ -20,7 +20,7 @@
 import warnings
 from enum import Enum
 from os.path import exists
-from typing import Any, Iterable, Optional
+from typing import Any, Iterable, Optional, Set
 
 # See https://github.com/package-url/packageurl-python/issues/65
 from packageurl import PackageURL  # type: ignore
@@ -1159,6 +1159,16 @@ def has_vulnerabilities(self) -> bool:
         """
         return bool(self.get_vulnerabilities())
 
+    def get_all_nested_components(self, include_self: bool = False) -> Set["Component"]:
+        components = set()
+        if include_self:
+            components.add(self)
+
+        for c in self.components:
+            components.update(c.get_all_nested_components(include_self=True))
+
+        return components
+
     def get_pypi_url(self) -> str:
         if self.version:
             return f'https://pypi.org/project/{self.name}/{self.version}'

cyclonedx/output/json.py

@@ -115,6 +115,9 @@ def _specialise_output_for_schema_version(self, bom_json: Dict[Any, Any]) -> str
             del bom_json['metadata']['properties']
 
         # Iterate Components
+        if self.get_bom().metadata.component:
+            bom_json['metadata'] = self._recurse_specialise_component(bom_json=bom_json['metadata'],
+                                                                      base_key='component')
         bom_json = self._recurse_specialise_component(bom_json=bom_json)
 
         # Iterate Services
@@ -155,60 +158,67 @@ def _get_schema_uri(self) -> Optional[str]:
 
     def _recurse_specialise_component(self, bom_json: Dict[Any, Any], base_key: str = 'components') -> Dict[Any, Any]:
         if base_key in bom_json.keys():
-            for i in range(len(bom_json[base_key])):
-                if not self.component_supports_mime_type_attribute() \
-                        and 'mime-type' in bom_json[base_key][i].keys():
-                    del bom_json[base_key][i]['mime-type']
-
-                if not self.component_supports_supplier() and 'supplier' in bom_json[base_key][i].keys():
-                    del bom_json[base_key][i]['supplier']
-
-                if not self.component_supports_author() and 'author' in bom_json[base_key][i].keys():
-                    del bom_json[base_key][i]['author']
-
-                if self.component_version_optional() and 'version' in bom_json[base_key][i] \
-                        and bom_json[base_key][i].get('version', '') == "":
-                    del bom_json[base_key][i]['version']
-
-                if not self.component_supports_pedigree() and 'pedigree' in bom_json[base_key][i].keys():
-                    del bom_json[base_key][i]['pedigree']
-                elif 'pedigree' in bom_json[base_key][i].keys():
-                    if 'ancestors' in bom_json[base_key][i]['pedigree'].keys():
-                        # recurse into ancestors
-                        bom_json[base_key][i]['pedigree'] = self._recurse_specialise_component(
-                            bom_json=bom_json[base_key][i]['pedigree'], base_key='ancestors'
-                        )
-                    if 'descendants' in bom_json[base_key][i]['pedigree'].keys():
-                        # recurse into descendants
-                        bom_json[base_key][i]['pedigree'] = self._recurse_specialise_component(
-                            bom_json=bom_json[base_key][i]['pedigree'], base_key='descendants'
-                        )
-                    if 'variants' in bom_json[base_key][i]['pedigree'].keys():
-                        # recurse into variants
-                        bom_json[base_key][i]['pedigree'] = self._recurse_specialise_component(
-                            bom_json=bom_json[base_key][i]['pedigree'], base_key='variants'
-                        )
-
-                if not self.external_references_supports_hashes() and 'externalReferences' \
-                        in bom_json[base_key][i].keys():
-                    for j in range(len(bom_json[base_key][i]['externalReferences'])):
-                        del bom_json[base_key][i]['externalReferences'][j]['hashes']
-
-                if not self.component_supports_properties() and 'properties' in bom_json[base_key][i].keys():
-                    del bom_json[base_key][i]['properties']
-
-                # recurse
-                if 'components' in bom_json[base_key][i].keys():
-                    bom_json[base_key][i] = self._recurse_specialise_component(bom_json=bom_json[base_key][i])
-
-                if not self.component_supports_evidence() and 'evidence' in bom_json[base_key][i].keys():
-                    del bom_json[base_key][i]['evidence']
-
-                if not self.component_supports_release_notes() and 'releaseNotes' in bom_json[base_key][i].keys():
-                    del bom_json[base_key][i]['releaseNotes']
+            if isinstance(bom_json[base_key], dict):
+                bom_json[base_key] = self._specialise_component_data(component_json=bom_json[base_key])
+            else:
+                for i in range(len(bom_json[base_key])):
+                    bom_json[base_key][i] = self._specialise_component_data(component_json=bom_json[base_key][i])
 
         return bom_json
 
+    def _specialise_component_data(self, component_json: Dict[Any, Any]) -> Dict[Any, Any]:
+        if not self.component_supports_mime_type_attribute() and 'mime-type' in component_json.keys():
+            del component_json['mime-type']
+
+        if not self.component_supports_supplier() and 'supplier' in component_json.keys():
+            del component_json['supplier']
+
+        if not self.component_supports_author() and 'author' in component_json.keys():
+            del component_json['author']
+
+        if self.component_version_optional() and 'version' in component_json \
+                and component_json.get('version', '') == "":
+            del component_json['version']
+
+        if not self.component_supports_pedigree() and 'pedigree' in component_json.keys():
+            del component_json['pedigree']
+        elif 'pedigree' in component_json.keys():
+            if 'ancestors' in component_json['pedigree'].keys():
+                # recurse into ancestors
+                component_json['pedigree'] = self._recurse_specialise_component(
+                    bom_json=component_json['pedigree'], base_key='ancestors'
+                )
+            if 'descendants' in component_json['pedigree'].keys():
+                # recurse into descendants
+                component_json['pedigree'] = self._recurse_specialise_component(
+                    bom_json=component_json['pedigree'], base_key='descendants'
+                )
+            if 'variants' in component_json['pedigree'].keys():
+                # recurse into variants
+                component_json['pedigree'] = self._recurse_specialise_component(
+                    bom_json=component_json['pedigree'], base_key='variants'
+                )
+
+        if not self.external_references_supports_hashes() and 'externalReferences' \
+                in component_json.keys():
+            for j in range(len(component_json['externalReferences'])):
+                del component_json['externalReferences'][j]['hashes']
+
+        if not self.component_supports_properties() and 'properties' in component_json.keys():
+            del component_json['properties']
+
+        # recurse
+        if 'components' in component_json.keys():
+            component_json = self._recurse_specialise_component(bom_json=component_json)
+
+        if not self.component_supports_evidence() and 'evidence' in component_json.keys():
+            del component_json['evidence']
+
+        if not self.component_supports_release_notes() and 'releaseNotes' in component_json.keys():
+            del component_json['releaseNotes']
+
+        return component_json
+
 
 class JsonV1Dot0(Json, SchemaVersion1Dot0):