熊猫越界时间戳解决方法

Question

I am trying to write a table from one sql server to another. There are some date values that are very far in the future (year 9000+). Pandas throws the error:

OutOfBoundsDatetime: Out of bounds nanosecond timestamp: 9017-11-10 00:00:00

when trying to write or perform the pd.to_datetime() funciton on the column. I would rather not convert the column to a string before writing and was hoping there is some way to handle dates so far away?

From the googling I've done, it does seem that pandas does do this by design. I have to believe there is an alternative though as the limitation of no far distant dates (especially when the database can handle it fine) seems pretty significant.

Answer 1

The above function treats the out of bounds by setting it to its maximum supported date

def date_include(dataframe):
    dataframe1=dataframe.str.split(" ", n = 1, expand = True)[0]
    dataframe2=dataframe.str.split(" ", n = 1, expand = True)[1]
    dataframe1=pd.to_numeric(dataframe1.str.replace('-',''),errors='coerce')
    dataframe1.loc[dataframe1>22620411]=22620411
    dataframe1=pd.to_datetime(dataframe1,format="%Y%m%d")
    date=dataframe1.astype(str)+" "+dataframe2
    return date