Skip to main content
 首页 » 编程设计

python之如何使用 Python 和 Pandas 在时间翻转时添加 24 小时

2024年11月24日8dyllove98

我必须分析一些日志,并基于此,我必须计算一些东西,我被一件事困住了。 在这里,我试图以一种简单的形式重新创建我的问题。 假设我在“stackoverflow.txt”文件中有以下日志

23:58:03.458 
23:58:13.446 
23:58:23.447 
23:58:33.440 
23:58:43.440 
23:58:53.440 
23:59:03.434 
23:59:13.435 
23:59:23.428 
23:59:33.428 
23:59:43.429 
23:59:53.435 
00:00:03.429 
00:00:13.423 
00:00:23.417 
00:00:33.411 
00:00:43.418 
00:00:53.411 
00:01:03.405 
00:01:13.406 
00:01:23.400 
00:01:33.406 
00:01:43.400 
00:01:53.411 
00:02:03.400 
00:02:13.406 
00:02:23.394 
00:02:33.400 
00:02:43.394 

我使用了以下 Python 程序,将这个时间转换为毫秒。

import pandas as pd 
df = pd.read_csv("stackoverflow.txt", header=None) 
# Split Time String into Hour Minutes Seconds and Milliseconds 
new_df = df[0].str.split(":", n=-1, expand=True) 
df['Hours'] = new_df[0] 
df['Minutes'] = new_df[1] 
# Split Seconds.Milliseconds information into Seconds and Milliseconds separately 
new_df = new_df[2].str.split(".", n=-1, expand=True) 
df['Seconds'] = new_df[0] 
df['Milliseconds'] = new_df[1] 
# These generated data frames are string, convert them into Integers 
# df['Hours'] = df['Hours'].apply(lambda x: int(x,10))  
# Another way of doing, good thing is that both are consuming same amount of time, checked using %time 
df['Hours'] = pd.to_numeric(df['Hours'], errors='coerce') 
df['Minutes'] = pd.to_numeric(df['Minutes'], errors='coerce') 
df['Seconds'] = pd.to_numeric(df['Seconds'], errors='coerce') 
df['Milliseconds'] = pd.to_numeric(df['Milliseconds'], errors='coerce') 
# Calculate Total Time 
df['Total Time(ms)'] = df['Hours']*3600000 + df['Minutes']*60000 + df['Seconds']*1000 + df['Milliseconds'] 
df 

输出如下:

0   Hours   Minutes Seconds Milliseconds    Total Time(ms) 
0   23:58:03.458    23  58  3   458 86283458 
1   23:58:13.446    23  58  13  446 86293446 
2   23:58:23.447    23  58  23  447 86303447 
3   23:58:33.440    23  58  33  440 86313440 
4   23:58:43.440    23  58  43  440 86323440 
5   23:58:53.440    23  58  53  440 86333440 
6   23:59:03.434    23  59  3   434 86343434 
7   23:59:13.435    23  59  13  435 86353435 
8   23:59:23.428    23  59  23  428 86363428 
9   23:59:33.428    23  59  33  428 86373428 
10  23:59:43.429    23  59  43  429 86383429 
11  23:59:53.435    23  59  53  435 86393435 
12  00:00:03.429    0   0   3   429 3429 
13  00:00:13.423    0   0   13  423 13423 
14  00:00:23.417    0   0   23  417 23417 
15  00:00:33.411    0   0   33  411 33411 
16  00:00:43.418    0   0   43  418 43418 
17  00:00:53.411    0   0   53  411 53411 
18  00:01:03.405    0   1   3   405 63405 
19  00:01:13.406    0   1   13  406 73406 
20  00:01:23.400    0   1   23  400 83400 
21  00:01:33.406    0   1   33  406 93406 
22  00:01:43.400    0   1   43  400 103400 
23  00:01:53.411    0   1   53  411 113411 
24  00:02:03.400    0   2   3   400 123400 
25  00:02:13.406    0   2   13  406 133406 
26  00:02:23.394    0   2   23  394 143394 
27  00:02:33.400    0   2   33  400 153400 
28  00:02:43.394    0   2   43  394 163394 

但是我想在一天从 23:59 到 00:00 发生变化时添加 24 小时。 我无法理解,我将如何做到这一点。 有人可以帮我实现这个目标吗?

请您参考如下方法:

我建议使用 Timedelta 小号:

df = pd.read_csv("stackoverflow.txt", header=None) 

首先通过 to_timedelta 转换列,然后求差,与Timedelta(0)比较并为下一行添加 pd.Timedelta(24, 'h') .

td = pd.to_timedelta(df[0]) 
df['new'] = td.mask(td.diff().lt(pd.Timedelta(0)).cumsum().gt(0), td + pd.Timedelta(1, 'days')) 
df['newint'] = (df['new'].dt.total_seconds() * 1000).astype(int) 

print (df) 
               0                    new    newint 
0   23:58:03.458 0 days 23:58:03.458000  86283458 
1   23:58:13.446 0 days 23:58:13.446000  86293446 
2   23:58:23.447 0 days 23:58:23.447000  86303447 
3   23:58:33.440 0 days 23:58:33.440000  86313440 
4   23:58:43.440 0 days 23:58:43.440000  86323440 
5   23:58:53.440 0 days 23:58:53.440000  86333440 
6   23:59:03.434 0 days 23:59:03.434000  86343434 
7   23:59:13.435 0 days 23:59:13.435000  86353435 
8   23:59:23.428 0 days 23:59:23.428000  86363428 
9   23:59:33.428 0 days 23:59:33.428000  86373428 
10  23:59:43.429 0 days 23:59:43.429000  86383429 
11  23:59:53.435 0 days 23:59:53.435000  86393435 
12  00:00:03.429 1 days 00:00:03.429000  86403429 
13  00:00:13.423 1 days 00:00:13.423000  86413423 
14  00:00:23.417 1 days 00:00:23.417000  86423417 
15  00:00:33.411 1 days 00:00:33.411000  86433411 
16  00:00:43.418 1 days 00:00:43.418000  86443418 
17  00:00:53.411 1 days 00:00:53.411000  86453411 
18  00:01:03.405 1 days 00:01:03.405000  86463405 
19  00:01:13.406 1 days 00:01:13.406000  86473406 
20  00:01:23.400 1 days 00:01:23.400000  86483400 
21  00:01:33.406 1 days 00:01:33.406000  86493406 
22  00:01:43.400 1 days 00:01:43.400000  86503400 
23  00:01:53.411 1 days 00:01:53.411000  86513411 
24  00:02:03.400 1 days 00:02:03.400000  86523400 
25  00:02:13.406 1 days 00:02:13.406000  86533406 
26  00:02:23.394 1 days 00:02:23.394000  86543394 
27  00:02:33.400 1 days 00:02:33.400000  86553400 
28  00:02:43.394 1 days 00:02:43.394000  86563394 

解决方案是多天的数据 - 所以对于第一次更改添加 1 天,接下来的 2 天...

创建差异,添加累加和并将输出转换为天时间增量,添加到原始数据中的内容:

print (df) 
               0 
0   23:59:23.428 
1   23:59:33.428 
2   23:59:43.429 
3   23:59:53.435 
4   00:00:03.429 
5   00:00:13.423 
6   00:00:23.417 
7   00:00:33.411 
8   23:59:23.428 
9   23:59:33.428 
10  23:59:43.429 
11  23:59:53.435 
12  00:00:03.429 
13  00:00:13.423 
14  00:00:23.417 
15  00:00:33.411 

td = pd.to_timedelta(df[0]) 
days = pd.to_timedelta(td.diff().lt(pd.Timedelta(0)).cumsum(), unit='d') 
 
df['new'] = td + days 
df['newint'] = (df['new'].dt.total_seconds() * 1000).astype(int) 
print (df) 
               0                    new     newint 
0   23:59:23.428 0 days 23:59:23.428000   86363428 
1   23:59:33.428 0 days 23:59:33.428000   86373428 
2   23:59:43.429 0 days 23:59:43.429000   86383429 
3   23:59:53.435 0 days 23:59:53.435000   86393435 
4   00:00:03.429 1 days 00:00:03.429000   86403429 
5   00:00:13.423 1 days 00:00:13.423000   86413423 
6   00:00:23.417 1 days 00:00:23.417000   86423417 
7   00:00:33.411 1 days 00:00:33.411000   86433411 
8   23:59:23.428 1 days 23:59:23.428000  172763428 
9   23:59:33.428 1 days 23:59:33.428000  172773428 
10  23:59:43.429 1 days 23:59:43.429000  172783429 
11  23:59:53.435 1 days 23:59:53.435000  172793435 
12  00:00:03.429 2 days 00:00:03.429000  172803429 
13  00:00:13.423 2 days 00:00:13.423000  172813423 
14  00:00:23.417 2 days 00:00:23.417000  172823417 
15  00:00:33.411 2 days 00:00:33.411000  172833411 

编辑:

天的解释:

先通过 diff 求差:

print (td.diff()) 
0                         NaT 
1                    00:00:10 
2             00:00:10.001000 
3             00:00:10.006000 
4    -1 days +00:00:09.994000 
5             00:00:09.994000 
6             00:00:09.994000 
7             00:00:09.994000 
8             23:58:50.017000 
9                    00:00:10 
10            00:00:10.001000 
11            00:00:10.006000 
12   -1 days +00:00:09.994000 
13            00:00:09.994000 
14            00:00:09.994000 
15            00:00:09.994000 
Name: 0, dtype: timedelta64[ns] 

然后用 lt 比较( < ) 对于负时间增量:

print (td.diff().lt(pd.Timedelta(0))) 
0     False 
1     False 
2     False 
3     False 
4      True 
5     False 
6     False 
7     False 
8     False 
9     False 
10    False 
11    False 
12     True 
13    False 
14    False 
15    False 
Name: 0, dtype: bool 

通过 cumsum 获取累计和:

print (td.diff().lt(pd.Timedelta(0)).cumsum()) 
0     0 
1     0 
2     0 
3     0 
4     1 
5     1 
6     1 
7     1 
8     1 
9     1 
10    1 
11    1 
12    2 
13    2 
14    2 
15    2 
Name: 0, dtype: int32 

最后转换为天时间增量:

days = pd.to_timedelta(td.diff().lt(pd.Timedelta(0)).cumsum(), unit='d') 
 
print (days) 
0    0 days 
1    0 days 
2    0 days 
3    0 days 
4    1 days 
5    1 days 
6    1 days 
7    1 days 
8    1 days 
9    1 days 
10   1 days 
11   1 days 
12   2 days 
13   2 days 
14   2 days 
15   2 days 
Name: 0, dtype: timedelta64[ns] 

编辑:

在您的解决方案中可以使用相同的想法:

... 
df['Total Time(ms)'] = df['Hours']*3600000 + df['Minutes']*60000 +  
                       df['Seconds']*1000 + df['Milliseconds'] 

s = df['Total Time(ms)'].diff().lt(0).cumsum() * 24 * 60 * 60 * 1000 
df['newint'] = s + df['Total Time(ms)'] 

print (df) 
               0  Hours  Minutes  Seconds  Milliseconds  Total Time(ms)  \ 
0   23:59:23.428     23       59       23           428        86363428    
1   23:59:33.428     23       59       33           428        86373428    
2   23:59:43.429     23       59       43           429        86383429    
3   23:59:53.435     23       59       53           435        86393435    
4   00:00:03.429      0        0        3           429            3429    
5   00:00:13.423      0        0       13           423           13423    
6   00:00:23.417      0        0       23           417           23417    
7   00:00:33.411      0        0       33           411           33411    
8   23:59:23.428     23       59       23           428        86363428    
9   23:59:33.428     23       59       33           428        86373428    
10  23:59:43.429     23       59       43           429        86383429    
11  23:59:53.435     23       59       53           435        86393435    
12  00:00:03.429      0        0        3           429            3429    
13  00:00:13.423      0        0       13           423           13423    
14  00:00:23.417      0        0       23           417           23417    
15  00:00:33.411      0        0       33           411           33411    
 
       newint   
0    86363428   
1    86373428   
2    86383429   
3    86393435   
4    86403429   
5    86413423   
6    86423417   
7    86433411   
8   172763428   
9   172773428   
10  172783429   
11  172793435   
12  172803429   
13  172813423   
14  172823417   
15  172833411