delete_json_files.py

import os

from send2trash import send2trash

# get origin dir
photos_path = r"/Volumes/Leslie/G_processing/alltakeouts"

# index all origin files to a dict[filename] = path
log_filename = "json-files-to-delete.log"
log_file = open(log_filename, "w")
filename_to_path = {}
files_to_delete = []
for dirpath, dirnames, filenames in os.walk(photos_path):
    for filename in filenames:
        if not filename.startswith(".") and filename[-4:] == "json":
            files_to_delete.append(os.path.join(dirpath, filename))
            log_file.write(f"{dirpath}/{filename}\n")

print(f"wrote to {log_filename}")

print("lines count:", len(files_to_delete))
#for filename in filename_to_path:
#    print(f"{filename}: {filename_to_path[filename]}")

# Delete files
print("If you want to delete all the files please enter 'DELETE ALL'")
if input() == "DELETE ALL":
    log_file = open("delete-record.log", "w")
    for file in files_to_delete:
        print(f"now deleting {file}")
        #os.remove(file)    # permenantly deletes files
        send2trash(file)    # sends files to trash
        log_file.write(f"Deleted {file}\n")
    print("wrote delete-record.log")
    log_file.write(f"\nAll {len(filename_to_path)} files deleted")
    log_file.close()
    print(f"\n\nAll {len(filename_to_path)} files deleted")
else:
    print("No files deleted")

delete_photos.py

import hashlib
import os

from send2trash import send2trash

# get origin dir
origin_path = r"/Volumes/Leslie/G_processing/alltakeouts/albums"

# get desitnation dir
destination_path = r"/Volumes/Leslie/G_processing/alltakeouts/dates"

origin_log = open("origin.log", "w")
origin_files_dict = {}
origin_count = 0
for dirpath, dirnames, filenames in os.walk(origin_path):
    for filename in filenames:
        if filename.startswith("."):
            continue

        fullpath = os.path.join(dirpath, filename)
        print(f"{origin_count}: reading {fullpath}")
        filedato = open(fullpath, "rb")
        data = filedato.read()
        hashed = hashlib.sha256(data).digest()
        origin_files_dict[hashed] = fullpath
        origin_log.write(f"{dirpath}/{filename}\n")
        origin_count += 1

print("origin files count:", len(origin_files_dict))
print("origin files set count:", len(set(origin_files_dict)))

## crawl all destination dir files, if filename in dict add to list to delete
dest_log = open("destination.log", "w")
dest_files_dict = {}
dest_count = 0
to_delete_log = open("to-delete.log", "w")
files_to_delete = []
for dirpath, dirnames, filenames in os.walk(destination_path):
    for filename in filenames:
        if filename.startswith("."):
            continue

        fullpath = os.path.join(dirpath, filename)
        print(f"{dest_count}: reading {fullpath}")
        filedato = open(fullpath, "rb")
        data = filedato.read()
        hashed = hashlib.sha256(data).digest()
        dest_files_dict[hashed] = fullpath
        dest_log.write(f"{dirpath}/{filename}\n")
        dest_count += 1

        if hashed in origin_files_dict:
            files_to_delete.append(fullpath)
            to_delete_log.write(fullpath + "\n")

print(f"Found {len(files_to_delete)} files")
print(f"Found {len(set(files_to_delete))} (set) files")

# Delete files
print("If you want to delete all the files please enter 'DELETE ALL'")
if input() == "DELETE ALL":
    actually_deleted_log = open("actually-deleted.log", "w")
    for file in files_to_delete:
        # os.remove(file)    # permenantly deletes files
        send2trash(file)    # sends files to trash
        print(f"Deleted {file}")
        actually_deleted_log.write(f"Deleted {file}\n")
    actually_deleted_log.write(f"\nAll {len(files_to_delete)} files deleted")
    actually_deleted_log.close()
    print(f"\nAll {len(files_to_delete)} files deleted")
else:
    print("No files deleted")