Refine the map_java_to_class implementation #1552 (#1554)

Signed-off-by: tdruez <tdruez@nexb.com>

Author: tdruez

scanpipe/pipes/d2d.py

@@ -192,15 +192,25 @@ def map_java_to_class(project, logger=None):
     to_resources = project_files.to_codebase().has_no_relation()
 
     to_resources_dot_class = to_resources.filter(extension=".class")
-    resource_count = to_resources_dot_class.count()
-    if logger:
-        logger(f"Mapping {resource_count:,d} .class resources to .java")
+    from_resources_dot_java = (
+        from_resources.filter(extension=".java")
+        # The "java_package" extra_data value is set during the `find_java_packages`,
+        # it is required to build the index.
+        .filter(extra_data__java_package__isnull=False)
+    )
+    to_resource_count = to_resources_dot_class.count()
+    from_resource_count = from_resources_dot_java.count()
 
-    from_resources_dot_java = from_resources.filter(extension=".java")
-    if not from_resources_dot_java.exists():
+    if not from_resource_count:
         logger("No .java resources to map.")
         return
 
+    if logger:
+        logger(
+            f"Mapping {to_resource_count:,d} .class resources to "
+            f"{from_resource_count:,d} .java"
+        )
+
     # build an index using from-side Java fully qualified class file names
     # built from the "java_package" and file name
     indexables = get_indexable_qualified_java_paths(from_resources_dot_java)
@@ -209,7 +219,7 @@ def map_java_to_class(project, logger=None):
     from_classes_index = pathmap.build_index(indexables, with_subpaths=False)
 
     resource_iterator = to_resources_dot_class.iterator(chunk_size=2000)
-    progress = LoopProgress(resource_count, logger)
+    progress = LoopProgress(to_resource_count, logger)
 
     for to_resource in progress.iter(resource_iterator):
         _map_java_to_class_resource(to_resource, from_resources, from_classes_index)
@@ -228,11 +238,8 @@ def get_indexable_qualified_java_paths_from_values(resource_values):
         (123, "org/apache/commons/LoggerImpl.java")
     """
     for resource_id, resource_name, resource_extra_data in resource_values:
-        java_package = resource_extra_data and resource_extra_data.get("java_package")
-        if not java_package:
-            continue
         fully_qualified = jvm.get_fully_qualified_java_path(
-            java_package,
+            java_package=resource_extra_data.get("java_package"),
             filename=resource_name,
         )
         yield resource_id, fully_qualified

scanpipe/tests/pipes/test_d2d.py

@@ -365,7 +365,7 @@ def test_scanpipe_pipes_d2d_map_java_to_class(self):
         buffer = io.StringIO()
         d2d.map_java_to_class(self.project1, logger=buffer.write)
 
-        expected = "Mapping 3 .class resources to .java"
+        expected = "Mapping 3 .class resources to 2 .java"
         self.assertIn(expected, buffer.getvalue())
 
         self.assertEqual(2, self.project1.codebaserelations.count())
@@ -390,7 +390,7 @@ def test_scanpipe_pipes_d2d_map_java_to_class_no_java(self):
         make_resource_file(self.project1, path="to/Abstract.class")
         buffer = io.StringIO()
         d2d.map_java_to_class(self.project1, logger=buffer.write)
-        expected = "Mapping 1 .class resources to .java" "No .java resources to map."
+        expected = "No .java resources to map."
         self.assertIn(expected, buffer.getvalue())
 
     def test_scanpipe_pipes_d2d_map_jar_to_source(self):