Date is converting to Variant in snowflake while loading from pandas-CodePudding

I am loading data from an excel file to snowflake using pandas using the below code

import pandas as pd
import snowflake.connector as snowCtx
from snowflake.connector.pandas_tools import pd_writer,write_pandas

df=pd.read_excel(open(r'SampleSuperstore.xlsx', 'rb'),sheet_name='Orders')
df.columns = df.columns.str.upper()
print(df)

conn = snowCtx.connect(
    user='username',
    password='password',
    account='account',
    database='superstore',
    schema='PUBLIC'
) 

table_name='ORDERS'
success, nchunks, nrows, _ = write_pandas(conn,df,table_name,
                                          chunk_size = 300,
                                          schema = 'PUBLIC')
print(success, nchunks, nrows)

The date column in excel is as below

Order_Date  Ship_Date
08-11-2016  11-11-2016

Even in pandas dataframe it is displaying the same

Order_Date  Ship_Date
08-11-2016  11-11-2016

When I query the result in snowflake, I get the result as below

ORDER_DATE          SHIP_DATE
1478563200000000    1478822400000000

If I change the datatype to date in snowflake it throws me an error stating variant datatype error.

CodePudding user response：

I'm not sure if there's a way to fix that in the chain from Excel to Snowflake, but it's a Unix epoch in microseconds and this can convert it:

set orderdate = 1478563200000000;
set shipdate  = 1478822400000000;

select   dateadd('us',$orderdate,'1970-01-01')::date ORDERDATE
        ,dateadd('us',$shipdate, '1970-01-01')::date SHIPDATE
;

CodePudding user response：

As a workaround, you may convert the dates to strings. Snowflake will convert them to date implicitly:

df['ORDER_DATE'] = df['ORDER_DATE'].astype(str)
df['SHIP_DATE'] = df['SHIP_DATE'].astype(str)

table_name='ORDERS'
success, nchunks, nrows, _ = write_pandas(conn,df,table_name,
                                          chunk_size = 300,
                                          schema = 'PUBLIC')