Initial implementation of --package-only

Signed-off-by: Ayan Sinha Mahapatra <ayansmahapatra@gmail.com>

Author: AyanSinhaMahapatra

src/packagedcode/bower.py

@@ -25,7 +25,7 @@ class BowerJsonHandler(models.DatafileHandler):
     documentation_url = 'https://bower.io'
 
     @classmethod
-    def parse(cls, location):
+    def parse(cls, location, package_only=False):
         with io.open(location, encoding='utf-8') as loc:
             package_data = json.load(loc)
 
@@ -87,7 +87,12 @@ def parse(cls, location):
                 )
             )
 
-        yield models.PackageData(
+        if package_only:
+            package_klass = models.PackageDataOnly
+        else:
+            package_klass = models.PackageData
+
+        yield package_klass(
             datasource_id=cls.datasource_id,
             type=cls.default_package_type,
             name=name,
@@ -98,5 +103,5 @@ def parse(cls, location):
             parties=parties,
             homepage_url=homepage_url,
             vcs_url=vcs_url,
-            dependencies=dependencies
+            dependencies=dependencies,
         )

src/packagedcode/models.py

@@ -905,6 +905,17 @@ def get_license_detections_and_expression(self):
         )
 
 
+class PackageDataOnly(PackageData):
+    """
+    PackageData class which skips the license/copyright detection during instance
+    creation.
+    """
+
+    def __attrs_post_init__(self):
+        if self.extracted_license_statement and not isinstance(self.extracted_license_statement, str):
+            self.extracted_license_statement = saneyaml.dump(self.extracted_license_statement)
+
+
 def get_default_relation_license(datasource_id):
     from packagedcode import HANDLER_BY_DATASOURCE_ID
     handler = HANDLER_BY_DATASOURCE_ID[datasource_id]

src/packagedcode/plugin_package.py

@@ -112,7 +112,6 @@ def get_available_package_parsers(docs=False):
     return all_data_packages
 
 
-
 @scan_impl
 class PackageScanner(ScanPlugin):
     """
@@ -161,7 +160,19 @@ class PackageScanner(ScanPlugin):
             help_group=SCAN_GROUP,
             sort_order=21,
         ),
-
+        PluggableCommandLineOption(
+            (
+                '--package-only',
+            ),
+            is_flag=True,
+            default=False,
+            help=(
+                'Only detect package information and skip license/copyright detection steps, '
+                'in application package and dependency manifests, lockfiles and related data.'
+            ),
+            help_group=SCAN_GROUP,
+            sort_order=22,
+        ),
         PluggableCommandLineOption(
             ('--list-packages',),
             is_flag=True,
@@ -172,10 +183,10 @@ class PackageScanner(ScanPlugin):
         ),
     ]
 
-    def is_enabled(self, package, system_package, **kwargs):
-        return package or system_package
+    def is_enabled(self, package, system_package, package_only, **kwargs):
+        return package or system_package or package_only
 
-    def get_scanner(self, package=True, system_package=False, **kwargs):
+    def get_scanner(self, package=True, system_package=False, package_only=False, **kwargs):
         """
         Return a scanner callable to scan a file for package data.
         """
@@ -185,9 +196,10 @@ def get_scanner(self, package=True, system_package=False, **kwargs):
             get_package_data,
             application=package,
             system=system_package,
+            package_only=package_only,
         )
 
-    def process_codebase(self, codebase, strip_root=False, **kwargs):
+    def process_codebase(self, codebase, strip_root=False, package_only=False, **kwargs):
         """
         Populate the ``codebase`` top level ``packages`` and ``dependencies``
         with package and dependency instances, assembling parsed package data
@@ -196,6 +208,11 @@ def process_codebase(self, codebase, strip_root=False, **kwargs):
         Also perform additional package license detection that depends on either
         file license detection or the package detections.
         """
+        # If we only want purls, we want to skip both the package
+        # assembly and the extra package license detection steps
+        if package_only:
+            return
+
         has_licenses = hasattr(codebase.root, 'license_detections')
 
         # These steps add proper license detections to package_data and hence

src/packagedcode/recognize.py

@@ -44,6 +44,7 @@ def recognize_package_data(
     location,
     application=True,
     system=False,
+    package_only=False,
 ):
     """
     Return a list of Package objects if any package_data were recognized for
@@ -55,19 +56,24 @@ def recognize_package_data(
     if not filetype.is_file(location):
         return []
 
-    assert application or system
-    if application and system:
+    assert application or system or package_only
+    if package_only or (application and system):
         datafile_handlers = ALL_DATAFILE_HANDLERS
     elif application:
         datafile_handlers = APPLICATION_PACKAGE_DATAFILE_HANDLERS
     elif system:
         datafile_handlers = SYSTEM_PACKAGE_DATAFILE_HANDLERS
 
-    return list(_parse(location, datafile_handlers=datafile_handlers))
+    return list(_parse(
+        location=location,
+        package_only=package_only,
+        datafile_handlers=datafile_handlers,
+    ))
 
 
 def _parse(
     location,
+    package_only=False,
     datafile_handlers=APPLICATION_PACKAGE_DATAFILE_HANDLERS,
 ):
     """
@@ -85,7 +91,7 @@ def _parse(
             logger_debug(f'_parse:.is_datafile: {location}')
 
         try:
-            for parsed in handler.parse(location):
+            for parsed in handler.parse(location=location, package_only=package_only):
                 if TRACE:
                     logger_debug(f' _parse: parsed: {parsed!r}')
                 yield parsed

src/scancode/api.py

@@ -248,20 +248,21 @@ def get_licenses(
 SCANCODE_DEBUG_PACKAGE_API = os.environ.get('SCANCODE_DEBUG_PACKAGE_API', False)
 
 
-def _get_package_data(location, application=True, system=False, **kwargs):
+def _get_package_data(location, application=True, system=False, package_only=False, **kwargs):
     """
     Return a mapping of package manifest information detected in the file at ``location``.
     Include ``application`` packages (such as pypi) and/or ``system`` packages.
     Note that all exceptions are caught if there are any errors while parsing a
     package manifest.
     """
-    assert application or system
+    assert application or system or package_only
     from packagedcode.recognize import recognize_package_data
     try:
         return recognize_package_data(
             location=location,
             application=application,
-            system=system
+            system=system,
+            package_only=package_only,
         ) or []
 
     except Exception as e:
@@ -291,7 +292,7 @@ def get_package_info(location, **kwargs):
     return dict(packages=[p.to_dict() for p in packages])
 
 
-def get_package_data(location, application=True, system=False, **kwargs):
+def get_package_data(location, application=True, system=False, package_only=False, **kwargs):
     """
     Return a mapping of package manifest information detected in the file at
     `location`.
@@ -304,6 +305,7 @@ def get_package_data(location, application=True, system=False, **kwargs):
         location=location,
         application=application,
         system=system,
+        package_only=package_only,
         **kwargs,
     ) or []
 

tests/packagedcode/data/bower/scan-package-only-expected.json

@@ -0,0 +1,113 @@
+{
+  "packages": [],
+  "dependencies": [],
+  "files": [
+    {
+      "path": "scan",
+      "type": "directory",
+      "package_data": [],
+      "for_packages": [],
+      "scan_errors": []
+    },
+    {
+      "path": "scan/bower.json",
+      "type": "file",
+      "package_data": [
+        {
+          "type": "bower",
+          "namespace": null,
+          "name": "John Doe",
+          "version": null,
+          "qualifiers": {},
+          "subpath": null,
+          "primary_language": null,
+          "description": "Physics-like animations for pretty particles",
+          "release_date": null,
+          "parties": [
+            {
+              "type": null,
+              "role": "author",
+              "name": "Betty Beta <bbeta@example.com>",
+              "email": null,
+              "url": null
+            },
+            {
+              "type": null,
+              "role": "author",
+              "name": "John Doe",
+              "email": "john@doe.com",
+              "url": "http://johndoe.com"
+            }
+          ],
+          "keywords": [
+            "motion",
+            "physics",
+            "particles"
+          ],
+          "homepage_url": null,
+          "download_url": null,
+          "size": null,
+          "sha1": null,
+          "md5": null,
+          "sha256": null,
+          "sha512": null,
+          "bug_tracking_url": null,
+          "code_view_url": null,
+          "vcs_url": null,
+          "copyright": null,
+          "holder": null,
+          "declared_license_expression": null,
+          "declared_license_expression_spdx": null,
+          "license_detections": [],
+          "other_license_expression": null,
+          "other_license_expression_spdx": null,
+          "other_license_detections": [],
+          "extracted_license_statement": "- MIT\n- Apache 2.0\n- BSD-3-Clause\n",
+          "notice_text": null,
+          "source_packages": [],
+          "file_references": [],
+          "extra_data": {},
+          "dependencies": [
+            {
+              "purl": "pkg:bower/get-size",
+              "extracted_requirement": "~1.2.2",
+              "scope": "dependencies",
+              "is_runtime": true,
+              "is_optional": false,
+              "is_resolved": false,
+              "resolved_package": {},
+              "extra_data": {}
+            },
+            {
+              "purl": "pkg:bower/eventEmitter",
+              "extracted_requirement": "~4.2.11",
+              "scope": "dependencies",
+              "is_runtime": true,
+              "is_optional": false,
+              "is_resolved": false,
+              "resolved_package": {},
+              "extra_data": {}
+            },
+            {
+              "purl": "pkg:bower/qunit",
+              "extracted_requirement": "~1.16.0",
+              "scope": "devDependencies",
+              "is_runtime": false,
+              "is_optional": true,
+              "is_resolved": false,
+              "resolved_package": {},
+              "extra_data": {}
+            }
+          ],
+          "repository_homepage_url": null,
+          "repository_download_url": null,
+          "api_data_url": null,
+          "datasource_id": "bower_json",
+          "purl": "pkg:bower/John%20Doe"
+        }
+      ],
+      "for_packages": [],
+      "scan_errors": []
+    }
+  ]
+}

tests/packagedcode/test_bower.py

@@ -12,6 +12,8 @@
 from packagedcode import bower
 from packages_test_utils import PackageTester
 from scancode_config import REGEN_TEST_FIXTURES
+from scancode.cli_test_utils import check_json_scan
+from scancode.cli_test_utils import run_scan_click
 
 
 class TestBower(PackageTester):
@@ -40,11 +42,15 @@ def test_parse_bower_json_author_objects(self):
         self.check_packages_data(package, expected_loc, regen=REGEN_TEST_FIXTURES)
 
     def test_end2end_bower_scan_is_moved_to_parent(self):
-        from scancode.cli_test_utils import check_json_scan
-        from scancode.cli_test_utils import run_scan_click
-
         test_file = self.get_test_loc('bower/scan')
         expected_file = self.get_test_loc('bower/scan-expected.json')
         result_file = self.get_temp_file('results.json')
         run_scan_click(['--package', test_file, '--json-pp', result_file])
         check_json_scan(expected_file, result_file, regen=REGEN_TEST_FIXTURES)
+
+    def test_end2end_bower_scan_is_moved_to_parent_package_only(self):
+        test_file = self.get_test_loc('bower/scan')
+        expected_file = self.get_test_loc('bower/scan-package-only-expected.json')
+        result_file = self.get_temp_file('results.json')
+        run_scan_click(['--package-only', test_file, '--json-pp', result_file])
+        check_json_scan(expected_file, result_file, regen=REGEN_TEST_FIXTURES)