Merge pull request #99 from Nautilus-Cyberneering/issue-98-error-vali…

…dating-paths-for-renaming

Fix error validating path for renamed files

src/nautilus_librarian/mods/dvc/domain/utils.py

@@ -28,16 +28,6 @@ def extract_all_added_and_modified_and_renamed_files_from_dvc_diff(
     return all_files
 
 
-def extract_added_and_modified_and_renamed_files_from_dvc_diff(
-    dvc_diff_json, only_basename=True
-):
-    all_files = extract_all_added_and_modified_and_renamed_files_from_dvc_diff(
-        dvc_diff_json, only_basename
-    )
-    files = filter_media_library_files(all_files)
-    return files
-
-
 def extract_all_changed_files_from_dvc_diff(dvc_diff_json, only_basename=True):
     """
     It gets a plain string list with the added, modified, deleted or renamed files from the dvc diff json.
@@ -166,13 +156,40 @@ def extract_renamed_files_from_dvc_diff(dvc_diff_json, only_basename=True):
     return media_files
 
 
-def extract_list_of_media_file_changes_from_dvc_diff_output(
-    dvc_diff, only_basename=True
-):
-    return extract_added_and_modified_and_renamed_files_from_dvc_diff(
-        dvc_diff, only_basename
+def extract_list_of_new_or_renamed_files_from_dvc_diff_output(dvc_diff_json):
+    dvc_diff = DvcDiffParser.from_json(dvc_diff_json)
+
+    added_files = dvc_diff.filter(
+        exclude_added=False,
+        exclude_modified=True,
+        exclude_deleted=True,
+        exclude_renamed=True,
+        only_basename=False,
     )
 
+    renamed_files = dvc_diff.filter(
+        exclude_added=True,
+        exclude_modified=True,
+        exclude_deleted=True,
+        exclude_renamed=False,
+        only_basename=False,
+    )
+
+    # Renamed files are not plain filename list. They are a dict like this:
+    # {
+    #     "new": "000002-32.600.2.tif",
+    #     "old": "000001-32.600.2.tif",
+    # }
+    # That means the image was renamed from "000001-32.600.2.tif" to "000002-32.600.2.tif"
+
+    flat_renamed_files = map(lambda file: file["new"], renamed_files)
+
+    all_files = added_files + list(flat_renamed_files)
+
+    files = filter_media_library_files(all_files)
+
+    return files
+
 
 def extract_added_files_from_dvc_diff(dvc_diff_json):
     """

src/nautilus_librarian/typer/commands/workflows/actions/validate_filepaths_action.py

@@ -1,6 +1,6 @@
 from nautilus_librarian.domain.validate_filepaths import validate_filepath
 from nautilus_librarian.mods.dvc.domain.utils import (
-    extract_list_of_media_file_changes_from_dvc_diff_output,
+    extract_list_of_new_or_renamed_files_from_dvc_diff_output,
     get_new_filepath_if_is_a_renaming_dict,
 )
 from nautilus_librarian.typer.commands.workflows.actions.action_result import (
@@ -20,9 +20,7 @@ def validate_filepaths_action(dvc_diff):
             ResultCode.EXIT, [Message("No media library file changes found")]
         )
 
-    filepaths = extract_list_of_media_file_changes_from_dvc_diff_output(
-        dvc_diff, only_basename=False
-    )
+    filepaths = extract_list_of_new_or_renamed_files_from_dvc_diff_output(dvc_diff)
 
     messages = []
 

tests/test_nautilus_librarian/test_mods/test_dvc/test_domain/test_utils.py

@@ -3,28 +3,38 @@
 from nautilus_librarian.mods.dvc.domain.utils import (
     extract_added_files_from_dvc_diff,
     extract_deleted_files_from_dvc_diff,
-    extract_list_of_media_file_changes_from_dvc_diff_output,
+    extract_list_of_new_or_renamed_files_from_dvc_diff_output,
     extract_modified_files_from_dvc_diff,
     extract_renamed_files_from_dvc_diff,
     get_new_filepath_if_is_a_renaming_dict,
 )
 
 
-def test_extract_list_of_media_file_changes_from_dvc_diff_output():
+def test_extract_list_of_new_or_renamed_files_from_dvc_diff_output():
     dvc_diff = {
         "added": [
             {"path": "data/000001/32/000001-32.600.2.tif"},
         ],
         "deleted": [],
         "modified": [],
-        "renamed": [],
+        "renamed": [
+            {
+                "path": {
+                    "old": "data/000003/32/000003-32.600.2.tif",
+                    "new": "data/000002/32/000002-32.600.2.tif",
+                }
+            }
+        ],
     }
 
-    filenames = extract_list_of_media_file_changes_from_dvc_diff_output(
+    filenames = extract_list_of_new_or_renamed_files_from_dvc_diff_output(
         compact_json(dvc_diff)
     )
 
-    assert filenames == ["000001-32.600.2.tif"]
+    assert filenames == [
+        "data/000001/32/000001-32.600.2.tif",
+        "data/000002/32/000002-32.600.2.tif",  # We include only the new name of the file for renamed ones.
+    ]
 
 
 def test_extract_added_files_from_dvc_diff():

tests/test_nautilus_librarian/test_typer/test_commands/test_workflows/test_actions/test_validate_filepaths_action.py

@@ -23,7 +23,7 @@ def given_a_dvc_diff_object_it_should_validate_the_filepath_of_the_new_media_fil
     assert result.contains_text("000001-32.600.2.tif ✓")
 
 
-def given_a_dvc_diff_object_it_should_validate_the_filepath_of_the_modified_media_files():
+def given_a_dvc_diff_object_it_should_not_validate_the_filepath_of_the_modified_media_files():
 
     dvc_diff_with_modified_image = {
         "added": [],
@@ -37,7 +37,7 @@ def given_a_dvc_diff_object_it_should_validate_the_filepath_of_the_modified_medi
     result = validate_filepaths_action(compact_json(dvc_diff_with_modified_image))
 
     assert result.code == ResultCode.CONTINUE
-    assert result.contains_text("000002-32.600.2.tif ✓")
+    assert not result.contains_text("000002-32.600.2.tif ✓")
 
 
 def given_a_dvc_diff_object_it_should_validate_the_filepath_of_the_renamed_media_files():
@@ -47,14 +47,20 @@ def given_a_dvc_diff_object_it_should_validate_the_filepath_of_the_renamed_media
         "deleted": [],
         "modified": [],
         "renamed": [
-            {"path": "data/000003/32/000003-32.600.2.tif"},
+            {
+                "path": {
+                    "old": "data/000001/32/000001-32.600.2.tif",
+                    "new": "data/000002/32/000002-32.600.2.tif",
+                }
+            }
         ],
     }
 
     result = validate_filepaths_action(compact_json(dvc_diff_with_renamed_image))
 
     assert result.code == ResultCode.CONTINUE
-    assert result.contains_text("000003-32.600.2.tif ✓")
+    assert result.contains_text("000002-32.600.2.tif ✓")
+    assert not result.contains_text("000001-32.600.2.tif ✓")
 
 
 def given_a_wrong_media_filepath_it_should_show_an_error():