refactor: simplify scancode conversion using new Resource functions

Author: abraemer

src/opossum_lib/scancode/convert_scancode_to_opossum.py

@@ -21,10 +21,7 @@ def convert_scancode_to_opossum(filename: str) -> OpossumFileContent:
 
     scancode_data = load_scancode_json(filename)
 
-    filetree = scancode_to_file_tree(scancode_data)
-    resources = filetree.to_opossum_resources()
-    with open("debug.json", "w") as out:
-        out.write(resources[0].model_dump_json(indent=4, by_alias=True))
+    resources = [scancode_to_file_tree(scancode_data)]
 
     scancode_header = extract_scancode_header(scancode_data, filename)
     metadata = opossum_model.Metadata(

src/opossum_lib/scancode/helpers.py

@@ -5,59 +5,25 @@
 
 from __future__ import annotations
 
-from pydantic import BaseModel
+from pathlib import Path
 
 import opossum_lib.opossum_model as opossum_model
 from opossum_lib.scancode.constants import SCANCODE_SOURCE_NAME
-from opossum_lib.scancode.helpers import check_schema, path_segments
 from opossum_lib.scancode.model import File, FileType, ScanCodeData
 
 
-class ScanCodeFileTree(BaseModel):
-    file: File
-    children: dict[str, ScanCodeFileTree] = {}
-
-    def get_path(self, path: list[str]) -> ScanCodeFileTree:
-        if len(path) == 0:
-            return self
-        next_segment, *rest = path
-        if next_segment not in self.children:
-            self.children[next_segment] = ScanCodeFileTree.model_construct(None)  # type: ignore
-        return self.children[next_segment].get_path(rest)
-
-    def revalidate(self) -> None:
-        check_schema(self)
-        for child in self.children.values():
-            child.revalidate()
-
-    def to_opossum_resources(
-        self,
-    ) -> list[opossum_model.Resource]:
-        def process_node(
-            node: ScanCodeFileTree,
-        ) -> opossum_model.Resource:
-            return opossum_model.Resource(
-                path=node.file.path,
-                attributions=get_attribution_info(node.file),
-                type=convert_resource_type(node.file.type),
-                children={
-                    key: process_node(child) for (key, child) in node.children.items()
-                },
-            )
-
-        return [process_node(self)]
-
-
-def scancode_to_file_tree(scancode_data: ScanCodeData) -> ScanCodeFileTree:
-    temp_root = ScanCodeFileTree.model_construct(file=None)  # type: ignore
+def scancode_to_file_tree(scancode_data: ScanCodeData) -> opossum_model.Resource:
+    temp_root = opossum_model.Resource(path=Path(""))
     for file in scancode_data.files:
-        segments = path_segments(file.path)
-        temp_root.get_path(segments).file = file
+        resource = opossum_model.Resource(
+            path=Path(file.path),
+            attributions=get_attribution_info(file),
+            type=convert_resource_type(file.type),
+        )
+        temp_root.add_resource(resource)
 
     assert len(temp_root.children) == 1
-    root = list(temp_root.children.values())[0]
-    check_schema(root)
-    return root
+    return list(temp_root.children.values())[0]
 
 
 def get_attribution_info(file: File) -> list[opossum_model.OpossumPackage]:

src/opossum_lib/scancode/resource_tree.py

@@ -14,6 +14,16 @@
 )
 
 
+def _create_reference_scancode_files() -> list[File]:
+    return [
+        _create_file("A", FileType.DIRECTORY),
+        _create_file("A/B", FileType.DIRECTORY),
+        _create_file("A/file1", FileType.FILE),
+        _create_file("A/file2.txt", FileType.FILE),
+        _create_file("A/B/file3", FileType.FILE),
+    ]
+
+
 def _create_file(
     path: str,
     type: FileType,

tests/test_scancode/model_helpers.py

@@ -4,83 +4,20 @@
 
 from copy import deepcopy
 
-import pytest
-from pydantic import ValidationError
-
 from opossum_lib.opossum_model import OpossumPackage, SourceInfo
 from opossum_lib.scancode.constants import SCANCODE_SOURCE_NAME
 from opossum_lib.scancode.model import (
     Copyright,
-    File,
     FileBasedLicenseDetection,
     FileType,
     Match,
-    ScanCodeData,
 )
 from opossum_lib.scancode.resource_tree import (
-    ScanCodeFileTree,
     get_attribution_info,
-    scancode_to_file_tree,
 )
 from tests.test_scancode.model_helpers import _create_file
 
 
-class TestRevalidate:
-    def test_successfully_revalidate_valid_file_tree(self) -> None:
-        dummy_file = _create_file("A", FileType.FILE)
-        valid_structure = ScanCodeFileTree(
-            file=dummy_file,
-            children={
-                "A": ScanCodeFileTree(file=dummy_file),
-                "B": ScanCodeFileTree(
-                    file=dummy_file, children={"C": ScanCodeFileTree(file=dummy_file)}
-                ),
-            },
-        )
-        valid_structure.revalidate()
-
-    def test_fail_to_revalidate_file_tree_invalid_at_toplevel(self) -> None:
-        dummy_file = _create_file("A", FileType.FILE)
-        invalid_structure = ScanCodeFileTree.model_construct(
-            children={
-                "A": ScanCodeFileTree(file=dummy_file),
-                "B": ScanCodeFileTree(
-                    file=dummy_file, children={"C": ScanCodeFileTree(file=dummy_file)}
-                ),
-            },
-            file=None,  # type: ignore
-        )
-        with pytest.raises(ValidationError):
-            invalid_structure.revalidate()
-
-    def test_fail_to_revalidate_file_tree_invalid_only_at_lower_level(self) -> None:
-        dummy_file = _create_file("A", FileType.FILE)
-        invalid_structure = ScanCodeFileTree(
-            file=dummy_file,
-            children={
-                "A": ScanCodeFileTree(file=dummy_file),
-                "B": ScanCodeFileTree(
-                    file=dummy_file,
-                    children={"C": ScanCodeFileTree.model_construct(None)},  # type: ignore
-                ),
-            },
-        )
-        with pytest.raises(ValidationError):
-            invalid_structure.revalidate()
-
-
-def test_scancode_to_resource_tree_produces_expected_result() -> None:
-    files = _create_reference_scancode_files()
-    scancode_data = ScanCodeData(
-        headers=[], packages=[], dependencies=[], license_detections=[], files=files
-    )
-
-    tree = scancode_to_file_tree(scancode_data)
-    reference = _create_reference_node_structure()
-
-    assert tree == reference
-
-
 def test_get_attribution_info_directory() -> None:
     folder = _create_file("A", FileType.DIRECTORY)
     assert get_attribution_info(folder) == []
@@ -159,29 +96,3 @@ def test_get_attribution_info_file_multiple() -> None:
         attribution_confidence=50,
     )
     assert set(attributions) == {expected1, expected2}
-
-
-def _create_reference_scancode_files() -> list[File]:
-    return [
-        _create_file("A", FileType.DIRECTORY),
-        _create_file("A/B", FileType.DIRECTORY),
-        _create_file("A/file1", FileType.FILE),
-        _create_file("A/file2.txt", FileType.FILE),
-        _create_file("A/B/file3", FileType.FILE),
-    ]
-
-
-def _create_reference_node_structure() -> ScanCodeFileTree:
-    folder, subfolder, file1, file2, file3 = _create_reference_scancode_files()
-    inner = ScanCodeFileTree(
-        file=subfolder, children={"file3": ScanCodeFileTree(file=file3)}
-    )
-    reference = ScanCodeFileTree(
-        file=folder,
-        children={
-            "B": inner,
-            "file1": ScanCodeFileTree(file=file1),
-            "file2.txt": ScanCodeFileTree(file=file2),
-        },
-    )
-    return reference