Fix Scala case class mapping to source files #1875

In Scala, case classes and inner classes defined within sealed traits
or objects compile to separate .class files that may not follow the
standard himBHsseparated naming convention. This fix implements a fallback
mechanism to handle these cases.

Changes:
- Added custom get_normalized_path() method to ScalaLanguage class
- Enhanced _map_jvm_to_class_resource() with Scala-specific fallback
  that searches for source files in the same package directory when
  exact match is not found
- Added comprehensive test case for Scala case class mapping

Fixes #1875

Author: pradhyum6144

scanpipe/pipes/d2d.py

@@ -164,22 +164,31 @@ def map_checksum(project, checksum_field, logger=None):
 def _map_jvm_to_class_resource(
     to_resource, from_resources, from_classes_index, jvm_lang: jvm.JvmLanguage
 ):
-    """
-    Map the ``to_resource`` .class file Resource with a Resource in
-    ``from_resources`` source files, using the ``from_classes_index`` index of
-    from/ fully qualified binary files.
-    """
     for extension in jvm_lang.source_extensions:
         normalized_path = jvm_lang.get_normalized_path(
             path=to_resource.path, extension=extension
         )
         match = pathmap.find_paths(path=normalized_path, index=from_classes_index)
         if not match:
+            if jvm_lang.name == "scala":
+                package_path = str(Path(normalized_path).parent)
+                potential_sources = from_resources.filter(
+                    path__startswith=package_path,
+                    extension__in=jvm_lang.source_extensions
+                )
+                for from_resource in potential_sources:
+                    from_source_root_parts = from_resource.path.strip("/").split("/")
+                    from_source_root = "/".join(from_source_root_parts[:-1])
+                    pipes.make_relation(
+                        from_resource=from_resource,
+                        to_resource=to_resource,
+                        map_type=jvm_lang.binary_map_type,
+                        extra_data={"from_source_root": f"{from_source_root}/"},
+                    )
             return
 
         for resource_id in match.resource_ids:
             from_resource = from_resources.get(id=resource_id)
-            # compute the root of the packages on the source side
             from_source_root_parts = from_resource.path.strip("/").split("/")
             from_source_root = "/".join(
                 from_source_root_parts[: -match.matched_path_length]

scanpipe/pipes/jvm.py

@@ -178,6 +178,22 @@ class ScalaLanguage(JvmLanguage):
     package_regex = re.compile(r"^\s*package\s+([\w\.]+)\s*;?")
     binary_map_type = "scala_to_class"
 
+    @classmethod
+    def get_normalized_path(cls, path, extension):
+        if not path.endswith(cls.binary_extensions):
+            raise ValueError(
+                f"Only path ending with {cls.binary_extensions} are supported."
+            )
+        path_obj = Path(path.strip("/"))
+        class_name = path_obj.name
+        
+        if "$" in class_name:
+            class_name, _, _ = class_name.partition("$")
+        else:
+            class_name, _, _ = class_name.partition(".")
+        
+        return str(path_obj.parent / f"{class_name}{extension}")
+
 
 class KotlinLanguage(JvmLanguage):
     name = "kotlin"

scanpipe/tests/pipes/test_d2d.py

@@ -612,6 +612,44 @@ def test_scanpipe_pipes_d2d_map_jar_to_scala_source(self):
         self.assertEqual(from2, relation.from_resource)
         self.assertEqual(to_jar, relation.to_resource)
 
+    def test_scanpipe_pipes_d2d_map_scala_case_classes_to_source(self):
+        from1 = make_resource_file(
+            self.project1,
+            path="from/pekko-cluster-sharding-typed/org/apache/pekko/cluster/sharding/typed/"
+            "ClusterShardingQuery.scala",
+            extra_data={"scala_package": "org.apache.pekko.cluster.sharding.typed"},
+        )
+        to1 = make_resource_file(
+            self.project1,
+            path="to/pekko-cluster-sharding-typed/org/apache/pekko/cluster/sharding/typed/"
+            "GetClusterShardingStats.class",
+        )
+        to2 = make_resource_file(
+            self.project1,
+            path="to/pekko-cluster-sharding-typed/org/apache/pekko/cluster/sharding/typed/"
+            "GetShardRegionState.class",
+        )
+        to3 = make_resource_file(
+            self.project1,
+            path="to/pekko-cluster-sharding-typed/org/apache/pekko/cluster/sharding/typed/"
+            "ClusterShardingQuery.class",
+        )
+
+        buffer = io.StringIO()
+        d2d.map_jvm_to_class(
+            self.project1, logger=buffer.write, jvm_lang=jvm.ScalaLanguage
+        )
+
+        expected = "Mapping 3 .class resources to 1 ('.scala',)"
+        self.assertIn(expected, buffer.getvalue())
+        self.assertEqual(3, self.project1.codebaserelations.count())
+
+        for to_resource in [to1, to2, to3]:
+            relation = self.project1.codebaserelations.get(to_resource=to_resource)
+            self.assertEqual(from1, relation.from_resource)
+            self.assertEqual("scala_to_class", relation.map_type)
+
+
     def test_scanpipe_pipes_d2d_map_jar_to_kotlin_source(self):
         from1 = make_resource_file(
             self.project1,