Currently I have a spark job that reads the file, creates a dataframe, does some transformations and then move those records in "year/month/date" format. I am achieving this by:
df.write.option("delimiter", "\t").option("header", False).mode(
"append"
).partitionBy("year", "month", "day").option("compression", "gzip").csv(
config["destination"]
)
I want to achieve the same by pythonic way. So, in the end it should look like:
data/2022/04/14
data/2022/04/15
CodePudding user response:
Based on your question , instead of using partitionBy
you can also modify your config['destination']
, as s3 will take care of the necessary folder creations underneath the s3 path
s3_dump_path = config["destination"] ### 's3:/test-path/'
>>> curr_date = datetime.now().date()
>>> year,month,day = curr_date.strftime('%Y'),curr_date.strftime('%m'),curr_date.strftime('%d')
>>> s3_new_path = '/'.join([s3_dump_path,year,month,day])
>>> s3_new_path
's3:/test-path//2022/04/14'
>>> config["destination"] = s3_new_path
df.write.option("delimiter", "\t").option("header", False).mode(
"append"
).option("compression", "gzip").csv(
config["destination"]
)