一个可寻找目录中重复图片文件的脚本，用于找到android项目中重复的图片资源

png_finder.py

import os
import os.path
import hashlib
import sys

hash_array = {}

def file_md5(file_path):
    md5 = hashlib.md5()
    with file(file_path, "rb") as f:
        for block in iter(lambda: f.read(128), ""):
            md5.update(block)
    return md5.hexdigest()

def find_png(target_dir):
    for root, dirs, files in os.walk(target_dir):
        #png_files = [item for item in files if item.endswith('.png')]
        for item in files:
            if item.endswith('.png') or item.endswith('.webp') or item.endswith('.jpg'):
                file_path = os.path.join(root, item)
                file_hash = file_md5(file_path)
                #print file_path, file_hash

                if file_hash in hash_array and not (hash_array[file_hash] is None):
                    hash_array[file_hash].append(file_path)
                else:
                    hash_array[file_hash] = []
                    hash_array[file_hash].append(file_path)

if __name__ == '__main__':
    if len(sys.argv) < 2:
        print 'Usage: png_finder <dir>'
    else:
        find_png(sys.argv[1])

        file_count = 0
        file_size = 0
        for (k, v) in hash_array.iteritems():
            v_len = len(v)
            if v_len > 1:
                file_count = file_count + v_len - 1
                for i in range(v_len - 1):
                    file_size = file_size + os.path.getsize(v[i])
                print k, v, v_len

        print 'duplicated file count', file_count
        print 'duplicated file size', file_size, 'B'