数据增强 json文件处理

曾经终败给现在 2023-01-22 06:51 11阅读 0赞

# 111.json文件格式 
    # { "path": ['n1', 'n2', 'n3', 'n1'],
    # "classes": ['c1', 'c2', 'c3', 'c1'],
    # "display_names": ['d1', 'd2', 'd3', 'd1']
    # }
    
    
    import json
    from collections import Counter
    
    path = '/home/name/Documents/111.json'
    
    with open(path, "r") as file:
        jf = json.load(file)
    
    # jf = json.load(open(path, "r")) # jf is a dict
    # print(jf)
    
    cnt = dict(Counter(jf['classes']))
    # print(cnt)
    
    new_jf = { }   # build a new dict
    for k in jf:
        new_jf[k] = []
    # print(new_jf)
    
    times =1
    for i, c in enumerate(jf["classes"]):  # now is 298
        # print(i, c )
        if cnt[c] < 5:
            times = 10
        elif cnt[c] < 10:
            times = 3
        elif cnt[c] < 20:
            times = 2
        for k in new_jf:
            new_jf[k].extend([jf[k][i]] * times)
    # print(new_jf)
    # for i ,c in enumerate(new_jf["classes"]): # after aug is 1451
    # print(i, c)
    
    with open(path, "w", encoding='utf-8') as f:
        json.dump(new_jf, f, indent=4, ensure_ascii=False) # json.dump()传递两个参数 第一个要写入的，第二个要存储的位置
    # indent表示间隔的长度
    # ensure_ascii=False 是否要转为ASCII码
    # 我要把new_jf写入path的路径中
    
    with open(path,"r") as fm:  # 载入json,查看是否已经改好
        data = json.load(fm)
    # print(data)
    cnt = dict(Counter(data["classes"]))
    print(cnt)

#################################################### demo_test
    import json
    from collections import Counter
    
    dic = { "path": ['n1', 'n2', 'n3', 'n1'],
           "classes": ['c1', 'c2', 'c3', 'c1'],
           "display_names": ['d1', 'd2', 'd3', 'd1']
           }
    
    new_dic = { }  # 新建一个dict
    for k in dic:  # 建立dict里的结构
        new_dic[k] = []
    print(new_dic)
    
    
    cnt = dict(Counter(dic['classes']))
    print(cnt)
    # for i in cnt.keys(): # 此法无法由'c1'去确定'n1','d1', --> 给他们编号,通过位置编号确定各个对应的元素
    # for k in dic:
    # print(k)
    # new_dic[k].extend()
    
    print(list(enumerate(dic["classes"])))
    for i, c in enumerate(dic["classes"]):
        print(i,c)
        if cnt[c] == 2:
            times = 3
        else:
            times = 5
        for k in new_dic:
            new_dic[k].extend([dic[k][i]] * times)
    print(new_dic)