Set the license expression from the detected license value #284

Signed-off-by: Thomas Druez <[email protected]>

Author: tdruez

scanpipe/pipelines/inspect_manifest.py

@@ -70,4 +70,5 @@ def create_packages_from_manifest(self):
                 raise Exception(f"No packages could be resolved for {input_location}")
 
             for package_data in resolved_packages:
+                package_data = resolve.set_license_expression(package_data)
                 update_or_create_package(self.project, package_data)

scanpipe/pipes/resolve.py

@@ -24,6 +24,7 @@
 
 from attributecode.model import About
 from packagedcode import APPLICATION_PACKAGE_DATAFILE_HANDLERS
+from packagedcode.licensing import get_normalized_expression
 from packageurl import PackageURL
 from python_inspector.resolve_cli import resolver_api
 
@@ -83,3 +84,27 @@ def get_default_package_type(input_location):
     "about": resolve_about_packages,
     "pypi": resolve_pypi_packages,
 }
+
+
+def set_license_expression(package_data):
+    """
+    Sets the license expression from a detected license dict/str in provided
+    `package_data`.
+    """
+    declared_license = package_data.get("declared_license")
+    license_expression = package_data.get("license_expression")
+
+    if declared_license and not license_expression:
+        license_str = ""
+
+        if isinstance(declared_license, dict):
+            license_str = declared_license.get("license")
+
+        if not license_str:
+            license_str = repr(declared_license)
+
+        license_expression = get_normalized_expression(query_string=license_str)
+        if license_expression:
+            package_data["license_expression"] = license_expression
+
+    return package_data

scanpipe/tests/test_pipes.py

@@ -45,6 +45,7 @@
 from scanpipe.pipes import fetch
 from scanpipe.pipes import filename_now
 from scanpipe.pipes import make_codebase_resource
+from scanpipe.pipes import resolve
 from scanpipe.pipes import rootfs
 from scanpipe.pipes import scancode
 from scanpipe.pipes import strip_root
@@ -735,6 +736,23 @@ def test_scanpipe_pipes_rootfs_has_hash_diff(self):
         codebase_resource = CodebaseResource(sha256="sha256", md5="md5")
         self.assertFalse(rootfs.has_hash_diff(install_file, codebase_resource))
 
+    def test_scanpipe_pipes_resolve_set_license_expression(self):
+        declared_license = {"license": "MIT"}
+        data = resolve.set_license_expression({"declared_license": declared_license})
+        self.assertEqual("mit", data.get("license_expression"))
+
+        declared_license = {
+            "classifiers": [
+                "License :: OSI Approved :: Python Software Foundation License"
+            ]
+        }
+        data = resolve.set_license_expression({"declared_license": declared_license})
+        self.assertEqual("python", data.get("license_expression"))
+
+        declared_license = "GPL 2.0"
+        data = resolve.set_license_expression({"declared_license": declared_license})
+        self.assertEqual("gpl-2.0", data.get("license_expression"))
+
     def test_scanpipe_pipes_windows_tag_uninteresting_windows_codebase_resources(self):
         p1 = Project.objects.create(name="Analysis")
         resource1 = CodebaseResource.objects.create(