生成年度数据编码的诡异Bug

问题是酱紫的
现在有一个如下包含1900-2100这201年间每天的日期信息字典列表:

[{'year': 1901, 'month': 1, 'day': 1, 'lunar_year': 1900, 'lunar_month': 11, 'lunar_leap': False, 'lunar_day': 11}, {'year': 1901, 'month': 1, 'day': 2, 'lunar_year': 1900, 'lunar_month': 11, 'lunar_leap': False, 'lunar_day': 12},...........{'year': 2100, 'month': 12, 'day': 31, 'lunar_year': 2100, 'lunar_leap': False, 'lunar_month': 12, 'lunar_day': 1}]

然后我打算按照如下规则转换成年度数据编码表:

一年的数据可用20位的二进制数表示

0001 1000 1000 1000 1000

• 中间的12位代表当年12个月的天数,0代表29天,1代表30天.
• 最右边的4位转换成十进制的数表示闰几月.
• 最左边的4位代表闰月的天数,0001代表30天,0000代表29天.

我参考的帖子的作者是用pandas进行编码的(不过他没有给出对应的代码),作者得到的结果如下:

CHINESEYEARCODE = [
19416,
19168, 42352, 21717, 53856, 55632, 91476, 22176, 39632,
21970, 19168, 42422, 42192, 53840, 119381, 46400, 54944,
44450, 38320, 84343, 18800, 42160, 46261, 27216, 27968,
109396, 11104, 38256, 21234, 18800, 25958, 54432, 59984,
92821, 23248, 11104, 100067, 37600, 116951, 51536, 54432,
120998, 46416, 22176, 107956, 9680, 37584, 53938, 43344,
46423, 27808, 46416, 86869, 19872, 42416, 83315, 21168,
43432, 59728, 27296, 44710, 43856, 19296, 43748, 42352,
21088, 62051, 55632, 23383, 22176, 38608, 19925, 19152,
42192, 54484, 53840, 54616, 46400, 46752, 103846, 38320,
18864, 43380, 42160, 45690, 27216, 27968, 44870, 43872,
38256, 19189, 18800, 25776, 29859, 59984, 27480, 23232,
43872, 38613, 37600, 51552, 55636, 54432, 55888, 30034,
22176, 43959, 9680, 37584, 51893, 43344, 46240, 47780,
44368, 21977, 19360, 42416, 86390, 21168, 43312, 31060,
27296, 44368, 23378, 19296, 42726, 42208, 53856, 60005,
54576, 23200, 30371, 38608, 19195, 19152, 42192, 118966,
53840, 54560, 56645, 46496, 22224, 21938, 18864, 42359,
42160, 43600, 111189, 27936, 44448, 84835, 37744, 18936,
18800, 25776, 92326, 59984, 27296, 108228, 43744, 37600,
53987, 51552, 54615, 54432, 55888, 23893, 22176, 42704,
21972, 21200, 43448, 43344, 46240, 46758, 44368, 21920,
43940, 42416, 21168, 45683, 26928, 29495, 27296, 44368,
84821, 19296, 42352, 21732, 53600, 59752, 54560, 55968,
92838, 22224, 19168, 43476, 41680, 53584, 62034, 54560
]

由于没找到合适的pandas教程,我直接用基础的python来编码,结果如下:

[19168, 42352, 21717, 53856, 55632, 91476, 22176, 39632, 21970, 19168, 42422, 42192, 53840, 119381, 46400, 54944, 44450, 38320, 84343, 18800, 42160, 46261, 27216, 27968, 109396, 11104, 38256, 21234, 18800, 25958, 54432, 59984, 92821, 23248, 11104, 100067, 37600, 116951, 51536, 54432, 120998, 46416, 22176, 107956, 9680, 37584, 53938, 43344, 46423, 27808, 46416, 86869, 19872, 42416, 83315, 21168, 43432, 59728, 27296, 44710, 43856, 19296, 43748, 42352, 21088, 62051, 55632, 23383, 22176, 38608, 19925, 19152, 42192, 54484, 53840, 54616, 46400, 46752, 103846, 38320, 18864, 43380, 42160, 45690, 27216, 27968, 44870, 43872, 38256, 19189, 18800, 25776, 29859, 59984, 27480, 23232, 43872, 38613, 37600, 51552, 55636, 54432, 55888, 30034, 22176, 43959, 9680, 37584, 51893, 43344, 46240, 47780, 44368, 21977, 19360, 42416, 86390, 21168, 43312, 31060, 27296, 44368, 23378, 19296, 42726, 42208, 53856, 60005, 54576, 23200, 30371, 38608, 19195, 19152, 42192, 118966, 53840, 54560, 56645, 46496, 22224, 21938, 18864, 42359, 42160, 43600, 111189, 27936, 44448, 84835, 37744, 18936, 18800, 25776, 92326, 59984, 27308, 108228, 43744, 37600, 53987, 51552, 55127, 54432, 55888, 23893, 22176, 42704, 21972, 21200, 43448, 43344, 46240, 111270, 44368, 21920, 48036, 42416, 21168, 45683, 26928, 29495, 27296, 44368, 84821, 19296, 42352, 21732, 53600, 125032, 54560, 55968, 92838, 22224, 19168, 47572, 41680, 53584]

除了作者自己加上去的第一个编码以及最后两个由于不好处理暂时被我忽略掉的编码以外,有六个编码我的结果和作者不同，并且经过核对作者是对的,情况如下:

年份	正确编码	我的编码	差值	正确编码二进制	错误编码二进制
2057	27296	27308	12	0b 110 1010 1010 0000	0b 110 1010 1010 1100
2063	54615	55127	512	0b 1101 0101 0101 0111	0b 1101 0111 0101 0111
2074	46758	111270	64512	0b 1011 0110 1010 0110	0b 1 1011 0010 1010 0110
2077	43940	48036	4096	0b 1010 1011 1010 0100	0b 1011 1011 1010 0100
2090	59752	125032	65280	0b 1110 1001 0110 1000	0b 1 1110 1000 0110 1000
2096	43476	47572	4096	0b 1010 1001 1101 0100	0b 1011 1001 1101 0100

这个Bug有点迷~

相关链接

参考文章
项目地址
ps:可以先看看项目的readme的关于开发这段