Add test for python pyx matching

Signed-off-by: Jono Yang <[email protected]>

Author: JonoYang

scanpipe/pipes/d2d.py

@@ -1946,9 +1946,7 @@ def map_elfs_binaries_with_symbols(project, logger=None):
     )
 
     # Collect source symbols from elf related source files
-    elf_from_resources = from_resources.filter(
-        extension__in=[".c", ".cpp", ".h", ".pyx", ".pxd"]
-    )
+    elf_from_resources = from_resources.filter(extension__in=[".c", ".cpp", ".h"])
 
     map_binaries_with_symbols(
         project=project,
@@ -2152,7 +2150,7 @@ def _map_javascript_symbols(to_resource, javascript_from_resources, logger):
 
 
 def map_python_pyx_to_binaries(project, logger=None):
-    """Map ELF binaries to their sources in ``project``."""
+    """Map Cython source to their compiled binaries in ``project``."""
     from_resources = (
         project.codebaseresources.files()
         .from_codebase()

scanpipe/tests/data/d2d-python/from-intbitset.tar.gz

@@ -1630,6 +1630,36 @@ def test_scanpipe_pipes_d2d_map_macho_symbols(self):
             ).count(),
         )
 
+    def test_scanpipe_pipes_d2d_map_python_pyx(self):
+        input_dir = self.project1.input_path
+        input_resources = [
+            self.data / "d2d-python/to-intbitset.whl",
+            self.data / "d2d-python/from-intbitset.tar.gz",
+        ]
+        copy_inputs(input_resources, input_dir)
+        self.from_files, self.to_files = d2d.get_inputs(self.project1)
+        inputs_with_codebase_path_destination = [
+            (self.from_files, self.project1.codebase_path / d2d.FROM),
+            (self.to_files, self.project1.codebase_path / d2d.TO),
+        ]
+        for input_files, codebase_path in inputs_with_codebase_path_destination:
+            for input_file_path in input_files:
+                scancode.extract_archive(input_file_path, codebase_path)
+
+        scancode.extract_archives(
+            self.project1.codebase_path,
+            recurse=True,
+        )
+        pipes.collect_and_create_codebase_resources(self.project1)
+        buffer = io.StringIO()
+        d2d.map_python_pyx_to_binaries(project=self.project1, logger=buffer.write)
+        self.assertEqual(
+            1,
+            CodebaseRelation.objects.filter(
+                project=self.project1, map_type="python_pyx_match"
+            ).count(),
+        )
+
     @skipIf(sys.platform == "darwin", "Test is failing on macOS")
     def test_scanpipe_pipes_d2d_map_winpe_symbols(self):
         input_dir = self.project1.input_path

scanpipe/tests/data/d2d-python/to-intbitset.whl

@@ -86,7 +86,7 @@ install_requires =
     rust-inspector==0.1.0
     binary-inspector==0.1.2
     python-inspector==0.14.0
-    source-inspector @ git+https://github.com/aboutcode-org/source-inspector@990e93b9c6bcb28e105e50af6507a0a2f0515213
+    source-inspector @ git+https://github.com/aboutcode-org/source-inspector@91d02f65e88aa95e72ed2884a47475c434865d1a
     aboutcode-toolkit==11.1.1
     # Utilities
     XlsxWriter==3.2.5