Merge pull request #6 from niradar/aider

close: #3

.gitignore

@@ -6,3 +6,4 @@
 .coverage
 conda.bat
 profile.*
+.aider*

duplicate_files_in_folders/duplicates_finder.py

@@ -196,16 +196,19 @@ def process_duplicates(combined: Dict, args: Namespace) -> (int, int):
 
         # Copy or move files to reference locations
         if not args.copy_to_all:
-            copy_or_move_file(src_filepath, args.move_to, ref_files[0]['path'], args.reference_dir, move=True)
+            copy_or_move_file(src_filepath, args.move_to, ref_files[0]['path'], args.reference_dir, move=True,
+                              keep_structure=args.keep_structure, scan_base_path=args.scan_dir)
             files_moved += 1
         else:
             num_to_copy = max(0, len(ref_files) - len(srcs_to_move))
             for i in range(num_to_copy):
-                copy_or_move_file(src_filepath, args.move_to, ref_files[i]['path'], args.reference_dir, False)
+                copy_or_move_file(src_filepath, args.move_to, ref_files[i]['path'], args.reference_dir, move=False,
+                                  keep_structure=args.keep_structure, scan_base_path=args.scan_dir)
                 files_created += 1
 
             for (src, _), tgt in zip(srcs_to_move, ref_files[num_to_copy:]):
-                copy_or_move_file(src, args.move_to, tgt['path'], args.reference_dir, move=True)
+                copy_or_move_file(src, args.move_to, tgt['path'], args.reference_dir, move=True,
+                                  keep_structure=args.keep_structure, scan_base_path=args.scan_dir)
                 files_moved += 1
 
     return files_moved, files_created
@@ -254,5 +257,6 @@ def clean_scan_dir_duplications(args: Namespace, combined: Dict) -> int:
                   locations['scan'] if os.path.exists(file_info['path'])]
     scan_dups_move_to: str = str(os.path.join(args.move_to, os.path.basename(args.scan_dir) + "_dups"))
     for src_path in scan_paths:
-        copy_or_move_file(src_path, scan_dups_move_to, src_path, args.scan_dir, move=True)
+        copy_or_move_file(src_path, scan_dups_move_to, src_path, args.scan_dir, move=True,
+                          keep_structure=args.keep_structure, scan_base_path=args.scan_dir)
     return len(scan_paths)

duplicate_files_in_folders/utils.py

@@ -67,6 +67,8 @@ def initialize_arguments():
                         help='Do not delete empty folders in the scan_dir folder. Default is to delete.')
     parser.add_argument('--full_hash', action='store_true',
                         help='Use full file hash for comparison. Default is partial.')
+    parser.add_argument('--keep_structure', action='store_true',
+                        help='Keep the original scan folder structure in the destination folder.')
     parser.set_defaults(delete_empty_folders=True)
     parser.add_argument('--clear_cache', action='store_true', help=argparse.SUPPRESS)  # for testing
     parser.add_argument('--extra_logging', action='store_true', help=argparse.SUPPRESS)  # for testing
@@ -155,19 +157,28 @@ def parse_arguments(cust_args=None, check_folders=True):
 
 
 def copy_or_move_file(scan_file_path: str, destination_base_path: str, ref_file_path: str, base_ref_path: str,
-                      move: bool = True) -> str:
+                      move: bool = True, keep_structure: bool = False, scan_base_path: str = None) -> str:
     """
     Copy or move a file from the scan directory to the destination directory based on the reference file path.
+    :param scan_base_path: The base path of the scan directory. Required if keep_structure is True.
     :param scan_file_path: Full path of the file we want to copy/move
     :param ref_file_path: The full path to the reference file within the base reference directory.
                           This path is used to determine the relative path for the destination.
     :param destination_base_path: The base path where the file should be copied or moved to.
     :param base_ref_path: The base directory path of the reference files.
                           This is used to calculate the relative path of the ref_file_path.
     :param move: True to move the file, False to copy it
+    :param keep_structure: True to keep the original scan folder structure in the destination folder, False otherwise
     :return: the final destination path
     """
-    destination_path = os.path.join(destination_base_path, os.path.relpath(ref_file_path, base_ref_path))
+    if keep_structure:
+        if scan_base_path is None:
+            raise ValueError("scan_base_path must be provided if keep_structure is True.")
+        sub_path = os.path.relpath(scan_file_path, scan_base_path)
+        destination_path = os.path.join(destination_base_path, sub_path)
+    else:
+        destination_path = os.path.join(destination_base_path, os.path.relpath(ref_file_path, base_ref_path))
+    logger.debug(f"Copying {scan_file_path} to {destination_path}")
     destination_dir = os.path.dirname(destination_path)
     file_manager = FileManager.get_instance()
     if not os.path.exists(destination_dir):

tests/helpers_testing.py

@@ -79,9 +79,16 @@ def setup_teardown():
     shutil.rmtree(TEMP_DIR)
 
 
-def setup_test_files(scan_files, ref_files):
-    copy_files(scan_files, os.path.join(TEMP_DIR, SCAN_DIR_NAME))
-    copy_files(ref_files, os.path.join(TEMP_DIR, REF_DIR_NAME))
+def setup_test_files(scan_files, ref_files, subfolder=None):
+    scan_dir = os.path.join(TEMP_DIR, SCAN_DIR_NAME)
+    ref_dir = os.path.join(TEMP_DIR, REF_DIR_NAME)
+    if subfolder:
+        scan_dir = os.path.join(scan_dir, subfolder)
+        ref_dir = os.path.join(ref_dir, subfolder)
+        os.makedirs(scan_dir, exist_ok=True)
+        os.makedirs(ref_dir, exist_ok=True)
+    copy_files(scan_files, scan_dir)
+    copy_files(ref_files, ref_dir)
 
 
 def get_folder_structure_include_subfolders(folder):

tests/test_duplicates_finder.py

@@ -244,3 +244,28 @@ def test_process_duplicates(setup_teardown):
     files_moved, files_created = process_duplicates(duplicates, args)
     assert files_created == 0
     assert files_moved == 5
+
+
+def test_process_duplicates_keep_structure(setup_teardown):
+    scan_dir, reference_dir, move_to_dir, common_args = setup_teardown
+    os.makedirs(os.path.join(scan_dir, "subfolder"))
+    setup_test_files(range(1, 6), [])
+    setup_test_files(range(6, 11), [], subfolder="subfolder")
+    time.sleep(0.1)  # sleep to make sure the modified date is different
+    setup_test_files([], range(1, 11))
+
+    common_args = ["--scan", scan_dir, "--reference_dir", reference_dir, "--move_to", move_to_dir, "--run",
+                   "--keep_structure"]
+    args = parse_arguments(common_args)
+    duplicates, scan_stats, ref_stats = find_duplicates_files_v3(args, scan_dir, reference_dir)
+    files_moved, files_created = process_duplicates(duplicates, args)
+
+    assert files_created == 0
+    assert files_moved == 10
+
+    print_all_folders(scan_dir, reference_dir, move_to_dir)
+    assert os.path.exists(os.path.join(move_to_dir, "subfolder", "6.jpg"))
+    assert os.path.exists(os.path.join(move_to_dir, "subfolder", "7.jpg"))
+    assert os.path.exists(os.path.join(move_to_dir, "subfolder", "8.jpg"))
+    assert os.path.exists(os.path.join(move_to_dir, "subfolder", "9.jpg"))
+    assert os.path.exists(os.path.join(move_to_dir, "subfolder", "10.jpg"))