Refactor package assembly for installed wheels

Reference: #4167
Signed-off-by: Ayan Sinha Mahapatra <[email protected]>

Author: AyanSinhaMahapatra

src/packagedcode/pypi.py

@@ -390,7 +390,11 @@ def assign_package_to_resources(cls, package, resource, codebase, package_adder)
         Assign files to package for an installed wheel. This requires a bit
         of navigation around as the files can be in multiple places.
         """
-        site_packages = resource.parent(codebase).parent(codebase)
+        dist_info_dir = resource.parent(codebase)
+        if not dist_info_dir:
+            return
+
+        site_packages = dist_info_dir.parent(codebase)
         if not site_packages:
             return
         package_data = resource.package_data
@@ -400,9 +404,7 @@ def assign_package_to_resources(cls, package, resource, codebase, package_adder)
         )
 
         package_data = models.PackageData.from_dict(package_data[0])
-
         package_uid = package.package_uid
-
         if package_uid:
             # save thyself!
             package_adder(package_uid, resource, codebase)
@@ -414,76 +416,32 @@ def assign_package_to_resources(cls, package, resource, codebase, package_adder)
                 # relative paths need special treatment
                 # most of thense are references to bin ../../../bin/wheel
                 cannot_resolve = False
-                ref_resource = None
+                ref_resource = site_packages
                 while path_ref.startswith('..'):
                     _, _, path_ref = path_ref.partition('../')
-                    ref_resource = site_packages.parent(codebase)
+                    ref_resource = ref_resource.parent(codebase)
                     if not ref_resource:
                         cannot_resolve = True
                         break
+
                 if cannot_resolve or not ref_resource:
                     # TODO:w e should log these kind of things
                     continue
                 else:
-                    if package_uid:
+                    ref_resource = codebase.get_resource(
+                        path=os.path.join(ref_resource.path, path_ref)
+                    )
+                    if ref_resource and package_uid:
                         package_adder(package_uid, ref_resource, codebase)
             else:
-                ref_resource = get_resource_for_path(
-                    path=path_ref,
-                    root=site_packages,
-                    codebase=codebase,
+                # These are absolute paths from the site-packages directory
+                ref_resource = codebase.get_resource(
+                    path=os.path.join(site_packages.path, path_ref)
                 )
                 if ref_resource and package_uid:
                     package_adder(package_uid, ref_resource, codebase)
 
 
-def get_resource_for_path(path, root, codebase):
-    """
-    Return a resource in ``codebase`` that has a ``path`` relative to the
-    ``root` Resource
-
-    For example, say we start from this:
-        path: this/is/that therefore segments [this, is, that]
-        root: /usr/foo
-
-    We would have these iterations:
-    iteration1
-        root = /usr/foo
-        segments = [this, is, that]
-        seg  this
-        segments = [is, that]
-        children = [/usr/foo/this]
-        root = /usr/foo/this
-
-    iteration2
-        root = /usr/foo/this
-        segments = [is, that]
-        seg  is
-        segments = [that]
-        children = [/usr/foo/this/is]
-        root = /usr/foo/this/is
-
-    iteration3
-        root = /usr/foo/this/is
-        segments = [that]
-        seg  that
-        segments = []
-        children = [/usr/foo/this/is/that]
-        root = /usr/foo/this/is/that
-
-    finally return root as /usr/foo/this/is/that
-    """
-    segments = path.strip('/').split('/')
-    while segments:
-        seg = segments.pop(0)
-        children = [c for c in root.children(codebase) if c.name == seg]
-        if len(children) != 1:
-            return
-        else:
-            root = children[0]
-    return root
-
-
 class PyprojectTomlHandler(BaseExtractedPythonLayout):
     datasource_id = 'pypi_pyproject_toml'
     path_patterns = ('*pyproject.toml',)

tests/packagedcode/data/pypi/site-packages/codebase/bin/pip

@@ -39,3 +39,4 @@ click/termui.py,sha256=mZ12uc3-loFaV__vr8buxn9uIjAhy7QwVuZOQ8jDdjc,28873
 click/testing.py,sha256=ptpMYgRY7dVfE3UDgkgwayu9ePw98sQI3D7zZXiCpj4,16063
 click/types.py,sha256=rj2g3biAXKkNKV8vlwbIKSUlixhXybH84N84fwCYqUU,35092
 click/utils.py,sha256=M8tuplcFFHROha3vQ60ZRSakSB_ng6w9e8Uc1AugPZ0,18934
+../../../bin/pip,,

tests/packagedcode/data/pypi/site-packages/codebase/lib/python3.9/site-packages/click-8.0.4.dist-info/RECORD

@@ -325,6 +325,23 @@
       ],
       "scan_errors": []
     },
+    {
+      "path": "bin",
+      "type": "directory",
+      "package_data": [],
+      "for_packages": [],
+      "scan_errors": []
+    },
+    {
+      "path": "bin/pip",
+      "type": "file",
+      "package_data": [],
+      "for_packages": [
+        "pkg:pypi/[email protected]?uuid=fixed-uid-done-for-testing-5642512d1758",
+        "pkg:pypi/[email protected]?uuid=fixed-uid-done-for-testing-5642512d1758"
+      ],
+      "scan_errors": []
+    },
     {
       "path": "lib",
       "type": "directory",
@@ -849,6 +866,15 @@
               "sha256": "33cb6ea6570514744e85adef43ad194526a4481fe783ac3d7bc51cd40ba03d9d",
               "sha512": null,
               "extra_data": {}
+            },
+            {
+              "path": "../../../bin/pip",
+              "size": null,
+              "sha1": null,
+              "md5": null,
+              "sha256": null,
+              "sha512": null,
+              "extra_data": {}
             }
           ],
           "is_private": false,