utils.py

import os
from datetime import datetime
import filecmp
import exifread


_size_cache = {}

def get_file_size(f):
	if f in _size_cache:
		return _size_cache[f]
	else:
		sz = os.stat(f).st_size 
		_size_cache[f] = sz
		return sz

def is_skipped(path, skip_path):
    if skip_path:
        for sp in skip_path:
            if sp in path:
                return True
    return False

def _get_recursive(parent_dir, skip_path=None):    
    result = [] 
    print(parent_dir)
    for dirpath, dirnames, filenames in os.walk(parent_dir):        
        if not is_skipped(dirpath, skip_path):
            for f in filenames:
                full_path = os.path.join(dirpath, f)
                result.append(full_path)                
        else:
            print('Skipped:', dirpath)
    return result

def get_files_recursive(parent_dir, skip_path=None):
	if isinstance(parent_dir, str):
		return _get_recursive(parent_dir, skip_path)
	else:
		result = []
		for d in parent_dir:
			files = _get_recursive(d, skip_path)
			result.extend(files)
		return result

def get_exif_date(filename):
    with open(filename, 'rb') as f:
	    tags = exifread.process_file(f)
	    date_tag = 'EXIF DateTimeOriginal'
	    if date_tag in tags:
	        date = str(tags[date_tag])
	        return datetime.strptime(date, '%Y:%m:%d %H:%M:%S')

def myfilecmp(f1, f2, shallow=True):
	size1 = get_file_size(f1)
	size2 = get_file_size(f2)
	if size1 and size1 == size2:		
		d1 = get_exif_date(f1)
		d2 = get_exif_date(f2)
		is_date_equal = (d1 and d2 and d1 == d2)
		if not is_date_equal:
			mtime1 = os.stat(f1).st_mtime
			mtime2 = os.stat(f2).st_mtime
			is_date_equal = (mtime1 == mtime2)
		if not shallow and is_date_equal:
			return filecmp.cmp(f1, f2, shallow=False)
		else:
			return is_date_equal
	else:
		return False


def build_file_size_map(files):
	size_map = {}
	for f in files:
		size = get_file_size(f)
		if not size in size_map:
			size_map[size] = []
		size_map[size].append(f)
	return size_map


def get_file_size_map(parent_dir):
	all_files = get_files_recursive(parent_dir)
	return build_file_size_map(all_files)


def list_chunks(lst, n):
    """Yield successive n-sized chunks from lst."""
    for i in range(0, len(lst), n):
        yield lst[i:i + n]