Add support for pip-inspect.deplock files

Add parser for pip-inspect.deplock files generated by deplock
which has all the package metadata, i.e. the resolved versions
and the dependency relationships.

Reference: aboutcode-org/scancode.io#1262
Signed-off-by: Ayan Sinha Mahapatra <ayansmahapatra@gmail.com>

Author: AyanSinhaMahapatra

src/packagedcode/__init__.py

@@ -207,6 +207,9 @@
     windows.MicrosoftUpdateManifestHandler,
 
     win_pe.WindowsExecutableHandler,
+
+    # These are handlers for deplock generated files
+    pypi.PipInspectDeplockHandler,
 ]
 
 if on_linux:

src/packagedcode/pypi.py

@@ -563,6 +563,102 @@ def parse(cls, location, package_only=False):
         yield models.PackageData.from_data(package_data, package_only)
 
 
+class PipInspectDeplockHandler(models.DatafileHandler):
+    datasource_id = 'pypi_inspect_deplock'
+    path_patterns = ('*pip-inspect.deplock',)
+    default_package_type = 'pypi'
+    default_primary_language = 'Python'
+    description = 'Python poetry pyproject.toml'
+    # These are files generated by deplock, see https://github.com/nexB/dependency-inspector
+    documentation_url = 'https://pip.pypa.io/en/stable/cli/pip_inspect/'
+
+    @classmethod
+    def get_resolved_package_from_metadata(cls, metadata, package_only=False):
+
+        requires_dist = metadata.get('requires_dist')
+        dependencies_for_resolved = get_requires_dependencies(
+            requires=requires_dist,
+        )
+        package_data = dict(
+            datasource_id=cls.datasource_id,
+            type=cls.default_package_type,
+            primary_language='Python',
+            name=metadata.get('name'),
+            version=metadata.get('version'),
+            extracted_license_statement=metadata.get('license'),
+            description=metadata.get('description'),
+            keywords=metadata.get('keywords'),
+            is_virtual=True,
+            dependencies=[
+                dep.to_dict()
+                for dep in dependencies_for_resolved
+            ],
+        )
+        return models.PackageData.from_data(package_data, package_only)
+
+    @classmethod
+    def parse(cls, location, package_only=False):
+
+        with open(location) as f:
+            content = f.read()
+
+        data = json.loads(content)
+        installed_packages = data.get('installed')
+        if not installed_packages:
+            return
+
+        main_package_metadata = {}
+        dependencies = []
+
+        for package_metadata in installed_packages:
+            package_metadata_dep = package_metadata.get('metadata')
+
+            # `direct_url` is only present for root package
+            # `requested` is true for root package and direct dependencies only
+            if package_metadata.get('requested') and 'direct_url' in package_metadata:
+                main_package_metadata = package_metadata_dep
+                main_package_requires = main_package_metadata.get('requires_dist')
+                dependencies_for_main = get_requires_dependencies(
+                    requires=main_package_requires,
+                )
+                dependencies.extend([
+                    dep.to_dict()
+                    for dep in dependencies_for_main
+                ])
+                continue
+
+            package_data_dep = cls.get_resolved_package_from_metadata(
+                metadata=package_metadata_dep,
+                package_only=package_only,
+            )
+            dep_purl = package_data_dep.purl
+
+            dependency = models.DependentPackage(
+                purl=dep_purl,
+                extracted_requirement=None,
+                scope=None,
+                is_runtime=True,
+                is_optional=False,
+                is_direct=False,
+                is_resolved=True,
+                resolved_package=package_data_dep.to_dict()
+            )
+            dependencies.append(dependency.to_dict())
+
+        pip_version = data.get('pip_version')
+        inspect_version = data.get('version')
+        extra_data = {
+            "pip_version": pip_version,
+            "inspect_version": inspect_version,
+        }
+
+        package_data_main = cls.get_resolved_package_from_metadata(
+            metadata=main_package_metadata,
+            package_only=package_only,
+        )
+        package_data_main.dependencies = dependencies
+        package_data_main.extra_data = extra_data
+        yield package_data_main
 
 
 META_DIR_SUFFIXES = '.dist-info', '.egg-info', 'EGG-INFO',
@@ -1494,7 +1590,7 @@ def get_dist_dependencies(dist):
     return get_requires_dependencies(requires=dist.requires)
 
 
-def get_requires_dependencies(requires, default_scope='install'):
+def get_requires_dependencies(requires, default_scope='install', is_direct=True):
     """
     Return a list of DependentPackage found in a ``requires`` list of
     requirement strings or an empty list.
@@ -1539,6 +1635,7 @@ def get_requires_dependencies(requires, default_scope='install'):
                 is_runtime=True,
                 is_optional=True if bool(extra) else False,
                 is_resolved=is_resolved,
+                is_direct=is_direct,
                 extracted_requirement=str(req),
         ))
 

tests/packagedcode/data/plugin/help.txt

@@ -713,6 +713,13 @@ Package type:  pypi
   description:       PyPI extracted egg PKG-INFO
   path_patterns:    '*/EGG-INFO/PKG-INFO'
 --------------------------------------------
+Package type:  pypi
+  datasource_id:     pypi_inspect_deplock
+  documentation URL: https://pip.pypa.io/en/stable/cli/pip_inspect/
+  primary language:  Python
+  description:       Python poetry pyproject.toml
+  path_patterns:    '*pip-inspect.deplock'
+--------------------------------------------
 Package type:  pypi
   datasource_id:     pypi_poetry_pyproject_toml
   documentation URL: https://packaging.python.org/en/latest/specifications/pyproject-toml/