Some changes to support SBOM generation for b build unbundled APEXs.

1) Use output file path of installed files in build system since there is no PRODUCT_OUT in Bazel
2) Use CONTAINS to describe the relationship between a APEX and files it contains
3) Generate SBOM of APEXs, which is similar to SBOM of products

Bug: 275472038
Test: CIs
(cherry picked from https://android-review.googlesource.com/q/commit:fd7e6517d345d3b8d4af12dae345434d968b83b9)
Merged-In: I41622366e5e6ed9dc78cca7bc7bb69a1f8f9bd9f
Change-Id: I41622366e5e6ed9dc78cca7bc7bb69a1f8f9bd9f

4 files changed, 63 insertions, 39 deletions

diff --git a/core/main.mk b/core/main.mk
index 6a24bd3050..cb4dca6b41 100644
--- a/core/main.mk
+++ b/core/main.mk
@@ -2163,10 +2163,11 @@ endif  # TARGET_BUILD_APPS
 $(shell rm $(PRODUCT_OUT)/sbom-metadata.csv >/dev/null 2>&1)
 $(PRODUCT_OUT)/sbom-metadata.csv: $(installed_files)
 	rm -f $@
-	@echo installed_file$(comma)module_path$(comma)soong_module_type$(comma)is_prebuilt_make_module$(comma)product_copy_files$(comma)kernel_module_copy_files$(comma)is_platform_generated >> $@
+	@echo installed_file$(comma)module_path$(comma)soong_module_type$(comma)is_prebuilt_make_module$(comma)product_copy_files$(comma)kernel_module_copy_files$(comma)is_platform_generated,build_output_path >> $@
 	$(foreach f,$(installed_files),\
 	  $(eval _module_name := $(ALL_INSTALLED_FILES.$f)) \
 	  $(eval _path_on_device := $(patsubst $(PRODUCT_OUT)/%,%,$f)) \
+	  $(eval _build_output_path := $(PRODUCT_OUT)/$(_path_on_device)) \
 	  $(eval _module_path := $(strip $(sort $(ALL_MODULES.$(_module_name).PATH)))) \
 	  $(eval _soong_module_type := $(strip $(sort $(ALL_MODULES.$(_module_name).SOONG_MODULE_TYPE)))) \
 	  $(eval _is_prebuilt_make_module := $(ALL_MODULES.$(_module_name).IS_PREBUILT_MAKE_MODULE)) \
@@ -2184,9 +2185,9 @@ $(PRODUCT_OUT)/sbom-metadata.csv: $(installed_files)
 	  $(eval _is_linker_config := $(if $(findstring $f,$(SYSTEM_LINKER_CONFIG) $(vendor_linker_config_file)),Y)) \
 	  $(eval _is_partition_compat_symlink := $(if $(findstring $f,$(PARTITION_COMPAT_SYMLINKS)),Y)) \
 	  $(eval _is_platform_generated := $(_is_build_prop)$(_is_notice_file)$(_is_dexpreopt_image_profile)$(_is_product_system_other_avbkey)$(_is_event_log_tags_file)$(_is_system_other_odex_marker)$(_is_kernel_modules_blocklist)$(_is_fsverity_build_manifest_apk)$(_is_linker_config)$(_is_partition_compat_symlink)) \
-	  @echo /$(_path_on_device)$(comma)$(_module_path)$(comma)$(_soong_module_type)$(comma)$(_is_prebuilt_make_module)$(comma)$(_product_copy_files)$(comma)$(_kernel_module_copy_files)$(comma)$(_is_platform_generated) >> $@ $(newline) \
+	  @echo /$(_path_on_device)$(comma)$(_module_path)$(comma)$(_soong_module_type)$(comma)$(_is_prebuilt_make_module)$(comma)$(_product_copy_files)$(comma)$(_kernel_module_copy_files)$(comma)$(_is_platform_generated)$(comma)$(_build_output_path) >> $@ $(newline) \
 	  $(if $(_post_installed_dexpreopt_zip), \
-	  for i in $$(zipinfo -1 $(_post_installed_dexpreopt_zip)); do echo /$$i$(comma)$(_module_path)$(comma)$(_soong_module_type)$(comma)$(_is_prebuilt_make_module)$(comma)$(_product_copy_files)$(comma)$(_kernel_module_copy_files)$(comma)$(_is_platform_generated) >> $@ ; done $(newline) \
+	  for i in $$(zipinfo -1 $(_post_installed_dexpreopt_zip)); do echo /$$i$(comma)$(_module_path)$(comma)$(_soong_module_type)$(comma)$(_is_prebuilt_make_module)$(comma)$(_product_copy_files)$(comma)$(_kernel_module_copy_files)$(comma)$(_is_platform_generated)$(comma)$(PRODUCT_OUT)/$$i >> $@ ; done $(newline) \
 	  ) \
 	)
 
@@ -2196,14 +2197,14 @@ sbom: $(PRODUCT_OUT)/sbom.spdx.json
 $(PRODUCT_OUT)/sbom.spdx.json: $(PRODUCT_OUT)/sbom.spdx
 $(PRODUCT_OUT)/sbom.spdx: $(PRODUCT_OUT)/sbom-metadata.csv $(GEN_SBOM)
 	rm -rf $@
-	$(GEN_SBOM) --output_file $@ --metadata $(PRODUCT_OUT)/sbom-metadata.csv --product_out_dir=$(PRODUCT_OUT) --build_version $(BUILD_FINGERPRINT_FROM_FILE) --product_mfr="$(PRODUCT_MANUFACTURER)" --json
+	$(GEN_SBOM) --output_file $@ --metadata $(PRODUCT_OUT)/sbom-metadata.csv --build_version $(BUILD_FINGERPRINT_FROM_FILE) --product_mfr "$(PRODUCT_MANUFACTURER)" --json
 
 $(call dist-for-goals,droid,$(PRODUCT_OUT)/sbom.spdx.json:sbom/sbom.spdx.json)
 else
 apps_only_sbom_files := $(sort $(patsubst %,%.spdx.json,$(filter %.apk,$(apps_only_installed_files))))
 $(apps_only_sbom_files): $(PRODUCT_OUT)/sbom-metadata.csv $(GEN_SBOM)
 	rm -rf $@
-	$(GEN_SBOM) --output_file $@ --metadata $(PRODUCT_OUT)/sbom-metadata.csv --product_out_dir=$(PRODUCT_OUT) --build_version $(BUILD_FINGERPRINT_FROM_FILE) --product_mfr="$(PRODUCT_MANUFACTURER)" --unbundled
+	$(GEN_SBOM) --output_file $@ --metadata $(PRODUCT_OUT)/sbom-metadata.csv --build_version $(BUILD_FINGERPRINT_FROM_FILE) --product_mfr "$(PRODUCT_MANUFACTURER)" --unbundled_apk
 
 sbom: $(apps_only_sbom_files)
 
diff --git a/tools/sbom/generate-sbom.py b/tools/sbom/generate-sbom.py
index 56509c9a11..1e98699a47 100755
--- a/tools/sbom/generate-sbom.py
+++ b/tools/sbom/generate-sbom.py
@@ -19,7 +19,6 @@ Generate the SBOM of the current target product in SPDX format.
 Usage example:
   generate-sbom.py --output_file out/target/product/vsoc_x86_64/sbom.spdx \
                    --metadata out/target/product/vsoc_x86_64/sbom-metadata.csv \
-                   --product_out_dir=out/target/product/vsoc_x86_64 \
                    --build_version $(cat out/target/product/vsoc_x86_64/build_fingerprint.txt) \
                    --product_mfr=Google
 """
@@ -89,11 +88,11 @@ def get_args():
   parser.add_argument('-v', '--verbose', action='store_true', default=False, help='Print more information.')
   parser.add_argument('--output_file', required=True, help='The generated SBOM file in SPDX format.')
   parser.add_argument('--metadata', required=True, help='The SBOM metadata file path.')
-  parser.add_argument('--product_out_dir', required=True, help='The parent directory of all the installed files.')
   parser.add_argument('--build_version', required=True, help='The build version.')
   parser.add_argument('--product_mfr', required=True, help='The product manufacturer.')
   parser.add_argument('--json', action='store_true', default=False, help='Generated SBOM file in SPDX JSON format')
-  parser.add_argument('--unbundled', action='store_true', default=False, help='Generate SBOM file for unbundled module')
+  parser.add_argument('--unbundled_apk', action='store_true', default=False, help='Generate SBOM for unbundled APKs')
+  parser.add_argument('--unbundled_apex', action='store_true', default=False, help='Generate SBOM for unbundled APEXs')
 
   return parser.parse_args()
 
@@ -127,7 +126,6 @@ def new_file_id(file_path):
 
 
 def checksum(file_path):
-  file_path = args.product_out_dir + '/' + file_path
   h = hashlib.sha1()
   if os.path.islink(file_path):
     h.update(os.readlink(file_path).encode('utf-8'))
@@ -334,9 +332,8 @@ def generate_package_verification_code(files):
   return h.hexdigest()
 
 
-def save_report(report):
-  prefix, _ = os.path.splitext(args.output_file)
-  with open(prefix + '-gen-report.txt', 'w', encoding='utf-8') as report_file:
+def save_report(report_file_path, report):
+  with open(report_file_path, 'w', encoding='utf-8') as report_file:
     for type, issues in report.items():
       report_file.write(type + '\n')
       for issue in issues:
@@ -394,7 +391,7 @@ def report_metadata_file(metadata_file_path, installed_file_metadata, report):
             installed_file_metadata['installed_file'], installed_file_metadata['module_path']))
 
 
-def generate_sbom_for_unbundled():
+def generate_sbom_for_unbundled_apk():
   with open(args.metadata, newline='') as sbom_metadata_file:
     reader = csv.DictReader(sbom_metadata_file)
     doc = sbom_data.Document(name=args.build_version,
@@ -402,7 +399,7 @@ def generate_sbom_for_unbundled():
                              creators=['Organization: ' + args.product_mfr])
     for installed_file_metadata in reader:
       installed_file = installed_file_metadata['installed_file']
-      if args.output_file != args.product_out_dir + installed_file + '.spdx.json':
+      if args.output_file != installed_file_metadata['build_output_path'] + '.spdx.json':
         continue
 
       module_path = installed_file_metadata['module_path']
@@ -412,7 +409,9 @@ def generate_sbom_for_unbundled():
                                   version=args.build_version,
                                   supplier='Organization: ' + args.product_mfr)
       file_id = new_file_id(installed_file)
-      file = sbom_data.File(id=file_id, name=installed_file, checksum=checksum(installed_file))
+      file = sbom_data.File(id=file_id,
+                            name=installed_file,
+                            checksum=checksum(installed_file_metadata['build_output_path']))
       relationship = sbom_data.Relationship(id1=file_id,
                                             relationship=sbom_data.RelationshipType.GENERATED_FROM,
                                             id2=package_id)
@@ -435,24 +434,25 @@ def main():
   args = get_args()
   log('Args:', vars(args))
 
-  if args.unbundled:
-    generate_sbom_for_unbundled()
+  if args.unbundled_apk:
+    generate_sbom_for_unbundled_apk()
     return
 
   global metadata_file_protos
   metadata_file_protos = {}
 
-  doc = sbom_data.Document(name=args.build_version,
-                           namespace=f'https://www.google.com/sbom/spdx/android/{args.build_version}',
-                           creators=['Organization: ' + args.product_mfr])
-
   product_package = sbom_data.Package(id=sbom_data.SPDXID_PRODUCT,
                                       name=sbom_data.PACKAGE_NAME_PRODUCT,
                                       download_location=sbom_data.VALUE_NONE,
                                       version=args.build_version,
                                       supplier='Organization: ' + args.product_mfr,
                                       files_analyzed=True)
-  doc.packages.append(product_package)
+
+  doc = sbom_data.Document(name=args.build_version,
+                           namespace=f'https://www.google.com/sbom/spdx/android/{args.build_version}',
+                           creators=['Organization: ' + args.product_mfr])
+  if not args.unbundled_apex:
+    doc.packages.append(product_package)
 
   doc.packages.append(sbom_data.Package(id=sbom_data.SPDXID_PLATFORM,
                                         name=sbom_data.PACKAGE_NAME_PLATFORM,
@@ -478,18 +478,21 @@ def main():
       module_path = installed_file_metadata['module_path']
       product_copy_files = installed_file_metadata['product_copy_files']
       kernel_module_copy_files = installed_file_metadata['kernel_module_copy_files']
+      build_output_path = installed_file_metadata['build_output_path']
 
       if not installed_file_has_metadata(installed_file_metadata, report):
         continue
-      file_path = args.product_out_dir + '/' + installed_file
-      if not (os.path.islink(file_path) or os.path.isfile(file_path)):
+      if not (os.path.islink(build_output_path) or os.path.isfile(build_output_path)):
         report[ISSUE_INSTALLED_FILE_NOT_EXIST].append(installed_file)
         continue
 
       file_id = new_file_id(installed_file)
       doc.files.append(
-        sbom_data.File(id=file_id, name=installed_file, checksum=checksum(installed_file)))
-      product_package.file_ids.append(file_id)
+        sbom_data.File(id=file_id, name=installed_file, checksum=checksum(build_output_path)))
+      if not args.unbundled_apex:
+        product_package.file_ids.append(file_id)
+      elif len(doc.files) > 1:
+          doc.add_relationship(sbom_data.Relationship(doc.files[0].id, sbom_data.RelationshipType.CONTAINS, file_id))
 
       if is_source_package(installed_file_metadata) or is_prebuilt_package(installed_file_metadata):
         metadata_file_path = get_metadata_file_path(installed_file_metadata)
@@ -533,16 +536,31 @@ def main():
                                                     relationship=sbom_data.RelationshipType.GENERATED_FROM,
                                                     id2=sbom_data.SPDXID_PLATFORM))
 
-  product_package.verification_code = generate_package_verification_code(doc.files)
+  if not args.unbundled_apex:
+    product_package.verification_code = generate_package_verification_code(doc.files)
+
+  if args.unbundled_apex:
+    doc.describes = doc.files[0].id
 
   # Save SBOM records to output file
   doc.created = datetime.datetime.now(tz=datetime.timezone.utc).strftime('%Y-%m-%dT%H:%M:%SZ')
-  with open(args.output_file, 'w', encoding="utf-8") as file:
-    sbom_writers.TagValueWriter.write(doc, file)
+  prefix = args.output_file
+  if prefix.endswith('.spdx'):
+    prefix = prefix.removesuffix('.spdx')
+  elif prefix.endswith('.spdx.json'):
+    prefix = prefix.removesuffix('.spdx.json')
+
+  output_file = prefix + '.spdx'
+  if args.unbundled_apex:
+    output_file = prefix + '-fragment.spdx'
+  with open(output_file, 'w', encoding="utf-8") as file:
+    sbom_writers.TagValueWriter.write(doc, file, fragment=args.unbundled_apex)
   if args.json:
-    with open(args.output_file+'.json', 'w', encoding="utf-8") as file:
+    with open(prefix + '.spdx.json', 'w', encoding="utf-8") as file:
       sbom_writers.JSONWriter.write(doc, file)
 
+  save_report(prefix + '-gen-report.txt', report)
+
 
 if __name__ == '__main__':
   main()
diff --git a/tools/sbom/sbom_data.py b/tools/sbom/sbom_data.py
index d2ef48d52c..14c4eb231d 100644
--- a/tools/sbom/sbom_data.py
+++ b/tools/sbom/sbom_data.py
@@ -80,6 +80,7 @@ class RelationshipType:
   DESCRIBES = 'DESCRIBES'
   VARIANT_OF = 'VARIANT_OF'
   GENERATED_FROM = 'GENERATED_FROM'
+  CONTAINS = 'CONTAINS'
 
 
 @dataclass
diff --git a/tools/sbom/sbom_writers.py b/tools/sbom/sbom_writers.py
index b1c66c5f80..85dee9d1a5 100644
--- a/tools/sbom/sbom_writers.py
+++ b/tools/sbom/sbom_writers.py
@@ -110,24 +110,26 @@ class TagValueWriter:
     return tagvalues
 
   @staticmethod
-  def marshal_described_element(sbom_doc):
+  def marshal_described_element(sbom_doc, fragment):
     if not sbom_doc.describes:
       return None
 
     product_package = [p for p in sbom_doc.packages if p.id == sbom_doc.describes]
     if product_package:
       tagvalues = TagValueWriter.marshal_package(product_package[0])
-      tagvalues.append(
-        f'{Tags.RELATIONSHIP}: {sbom_doc.id} {sbom_data.RelationshipType.DESCRIBES} {sbom_doc.describes}')
+      if not fragment:
+        tagvalues.append(
+            f'{Tags.RELATIONSHIP}: {sbom_doc.id} {sbom_data.RelationshipType.DESCRIBES} {sbom_doc.describes}')
 
       tagvalues.append('')
       return tagvalues
 
     file = [f for f in sbom_doc.files if f.id == sbom_doc.describes]
     if file:
-      tagvalues = [
-        f'{Tags.RELATIONSHIP}: {sbom_doc.id} {sbom_data.RelationshipType.DESCRIBES} {sbom_doc.describes}'
-      ]
+      tagvalues = TagValueWriter.marshal_file(file[0])
+      if not fragment:
+        tagvalues.append(
+            f'{Tags.RELATIONSHIP}: {sbom_doc.id} {sbom_data.RelationshipType.DESCRIBES} {sbom_doc.describes}')
 
       return tagvalues
 
@@ -180,6 +182,8 @@ class TagValueWriter:
   def marshal_files(sbom_doc):
     tagvalues = []
     for file in sbom_doc.files:
+      if file.id == sbom_doc.describes:
+        continue
       tagvalues += TagValueWriter.marshal_file(file)
     return tagvalues
 
@@ -204,9 +208,9 @@ class TagValueWriter:
     content = []
     if not fragment:
       content += TagValueWriter.marshal_doc_headers(sbom_doc)
-      described_element = TagValueWriter.marshal_described_element(sbom_doc)
-      if described_element:
-        content += described_element
+    described_element = TagValueWriter.marshal_described_element(sbom_doc, fragment)
+    if described_element:
+      content += described_element
     content += TagValueWriter.marshal_files(sbom_doc)
     tagvalues, marshaled_relationships = TagValueWriter.marshal_packages(sbom_doc)
     content += tagvalues