fixed error in summary logging

added a final summary to console output

some code cleanups (deleted commented function)

README.md

@@ -85,18 +85,11 @@ pip install -r requirements.txt
 ```
 
 ## Possible Future Improvements
-- [ ] Improve summary and user interface
-    - [ ] Add a better summary of the actions taken
-    - [ ] Add summary also to the console output
 - [ ] Better handling of folders with saved html files
-  - [ ] Deal with `_files` folders in the source folder
-  - [ ] Move it only if all files are duplicates
+  - [ ] Deal with `_files` folders in the source folder - Move it only if all files are duplicates
 - [ ] More ways to influence how the script works
   - [ ] Add an argument to act only if the entire folder is a subfolder of a target folder, recursively (bottom-up)
-  - [ ] Option to keep the source folder structure in the move_to folder
   - [ ] Option to send duplicates to recycle bin instead of move_to folder
-- [ ] More safeguards
-  - [ ] Check any file operation for errors
 ## Known Issues
 - [ ] Even if argument --copy_to_all is not present, still need to move the duplicates to the move_to folder without copying them to other folders
 - [ ] Issue with files with non-standard characters in the filename - no reproducible yet

df_finder3.py

@@ -26,7 +26,7 @@ def main(args):
 
     deleted_source_folders = fm.delete_empty_folders_in_tree(args.src, True) if args.delete_empty_folders else 0
     hash_manager.save_data()
-    output_results(args, deleted_source_folders, duplicate_source_files_moved, files_created, files_moved, hash_manager)
+    output_results(args, files_moved, files_created, deleted_source_folders, duplicate_source_files_moved)
 
 
 if __name__ == "__main__":

duplicate_files_in_folders/duplicates_finder.py

@@ -146,7 +146,7 @@ def process_duplicates(combined: Dict, args) -> (int, int):
                 copy_or_move_file(tgt['path'], args.move_to, src, args.target, not args.run, move=True)
                 files_moved += 1
 
-    return files_created, files_moved
+    return files_moved, files_created
 
 
 def clean_source_duplications(args, combined):

duplicate_files_in_folders/utils.py

@@ -226,22 +226,39 @@ def print_error(message):
     logger.critical(f"{message}")
     sys.exit(1)
 
-
-def output_results(args, deleted_source_folders, duplicate_source_files_moved, files_created, files_moved, hash_manager):
+def output_results(args, files_moved, files_created, deleted_source_folders, duplicate_source_files_moved):
     summary_header = "Summary (Test Mode):" if not args.run else "Summary:"
     separator = "-" * max(len(summary_header), 40)
-    cache_hits = f"Hash requests: {hash_manager.persistent_cache_requests + hash_manager.temporary_cache_requests}," + \
-                 f" Cache hits: {hash_manager.persistent_cache_hits + hash_manager.temporary_cache_hits}"
-    logger.info(summary_header)
-    logger.info(separator)
+    blank_line = ""
+    fixed_width = 25
 
+    # Header
+    log_and_print("")
+    log_and_print(summary_header)
+    log_and_print(separator)
+    hash_manager = HashManager.get_instance()
+    # Cache hits information
+    cache_hits = f"Hash requests: {hash_manager.persistent_cache_requests + hash_manager.temporary_cache_requests}, " \
+                 f"Cache hits: {hash_manager.persistent_cache_hits + hash_manager.temporary_cache_hits}"
     logger.debug(cache_hits)
-    res_str = f'Move: {files_moved} files, Create: {files_created} copies'
+
+    # Detailed summary
+    summary_lines = {
+        'Files Moved': f"{files_moved}",
+        'Files Created': f"{files_created} copies",
+    }
+
     if duplicate_source_files_moved:
-        res_str += f", Moved {duplicate_source_files_moved} duplicate files from the source folder"
+        summary_lines['Duplicate Files Moved'] = f"{duplicate_source_files_moved} duplicate files from the source folder"
     if deleted_source_folders:
-        res_str += f", Deleted: {deleted_source_folders} empty folders in the source folder"
-    logger.info(res_str)
+        summary_lines['Empty Folders Deleted'] = f"{deleted_source_folders} empty folders in the source folder"
+
+    for key, value in summary_lines.items():
+        log_and_print(f"{key.ljust(fixed_width)}: {value}")
+
+    # Footer
+    log_and_print(separator)
+    log_and_print("")
 
 
 def setup_hash_manager(args):
@@ -252,19 +269,6 @@ def setup_hash_manager(args):
     return hash_manager
 
 
-# def copy_or_move_file(tgt_filepath: str, move_to: str, src_filepath: str, target: str, test_mode, move=True):
-#     new_src_path = os.path.join(move_to, os.path.relpath(tgt_filepath, target))
-#     new_src_dir = os.path.dirname(new_src_path)
-#     fm = FileManager(not test_mode)
-#     if not os.path.exists(new_src_dir):
-#         fm.make_dirs(new_src_dir)
-#     new_filename = check_and_update_filename(new_src_path)
-#     if move:
-#         fm.move_file(src_filepath, new_filename)
-#     else:
-#         fm.copy_file(src_filepath, new_filename)
-#     return new_filename
-
 def copy_or_move_file(target_file_path: str, destination_base_path: str, source_file_path: str, base_target_path: str, is_test_mode: bool, move: bool = True) -> str:
     destination_path = os.path.join(destination_base_path, os.path.relpath(target_file_path, base_target_path))
     destination_dir = os.path.dirname(destination_path)

tests/test_duplicates_finder.py

@@ -181,7 +181,7 @@ def test_process_duplicates(setup_teardown):
 
     args = parse_arguments(common_args)
     duplicates, source_stats, target_stats = find_duplicates_files_v3(args, source_dir, target_dir)
-    files_created, files_moved = process_duplicates(duplicates, args)
+    files_moved, files_created = process_duplicates(duplicates, args)
     assert files_created == 0
     assert files_moved == 5
 
@@ -195,7 +195,7 @@ def test_process_duplicates(setup_teardown):
                    "--ignore_diff", "filename"]
     args = parse_arguments(common_args)
     duplicates, source_stats, target_stats = find_duplicates_files_v3(args, source_dir, target_dir)
-    files_created, files_moved = process_duplicates(duplicates, args)
+    files_moved, files_created = process_duplicates(duplicates, args)
     assert files_created == 0
     assert files_moved == 0
 
@@ -205,7 +205,7 @@ def test_process_duplicates(setup_teardown):
                    "--ignore_diff", "checkall"]
     args = parse_arguments(common_args)
     duplicates, source_stats, target_stats = find_duplicates_files_v3(args, source_dir, target_dir)
-    files_created, files_moved = process_duplicates(duplicates, args)
+    files_moved, files_created = process_duplicates(duplicates, args)
     assert files_created == 0
     assert files_moved == 0
 
@@ -215,7 +215,7 @@ def test_process_duplicates(setup_teardown):
                    "--ignore_diff", "filename,mdate"]
     args = parse_arguments(common_args)
     duplicates, source_stats, target_stats = find_duplicates_files_v3(args, source_dir, target_dir)
-    files_created, files_moved = process_duplicates(duplicates, args)
+    files_moved, files_created = process_duplicates(duplicates, args)
     assert files_created == 0
     assert files_moved == 5
 
@@ -229,7 +229,7 @@ def test_process_duplicates(setup_teardown):
     common_args = ["--src", source_dir, "--target", target_dir, "--move_to", move_to_dir, "--run", "--copy_to_all"]
     args = parse_arguments(common_args)
     duplicates, source_stats, target_stats = find_duplicates_files_v3(args, source_dir, target_dir)
-    files_created, files_moved = process_duplicates(duplicates, args)
+    files_moved, files_created = process_duplicates(duplicates, args)
     assert files_created == 5
     assert files_moved == 5
     assert os.path.exists(os.path.join(move_to_dir, "subfolder"))
@@ -244,7 +244,7 @@ def test_process_duplicates(setup_teardown):
     common_args = ["--src", source_dir, "--target", target_dir, "--move_to", move_to_dir, "--run"]
     args = parse_arguments(common_args)
     duplicates, source_stats, target_stats = find_duplicates_files_v3(args, source_dir, target_dir)
-    files_created, files_moved = process_duplicates(duplicates, args)
+    files_moved, files_created = process_duplicates(duplicates, args)
     assert files_created == 0
     assert files_moved == 5