The India User Group is live! Be a part of our first Indian user event: JOIN THE EVENT

Schema errors while using the Chunked writing built in fucntionality

PapaA
Level 3
Schema errors while using the Chunked writing built in fucntionality

I want to split a dataframe of 4000 columns in 5 diffrent tables to be able to write it back to SQL.

However, our instance is strugling with writing a table of 1000 columns to schema, thus I had to obtain a writer as per DSS docs.

However, while the writer is working for the first round it crushes on the second one raising a schema error which we cannot really crack.

Could you please help on what goes wrong?  Why the schema of the empty table is not been updated based on the input that is about to be writen?

Thanks in advance,
A

 

 

 

 

sql_datasets= [
  "DTO_Risk_Factors_unshifted_1",
"DTO_Risk_Factors_unshifted_2"]

inp = dataiku.Dataset("DLS_TEST.DTO_Risk_Factors")

start = 0
end = 950
step = 950


for out in  sql_datasets:

    print ('starting location', start)
    print ('ending location', end)
    print(out)

    out = dataiku.Dataset(out)
    
    with out.get_writer() as writer:
        inp = dataiku.Dataset("DLS_TEST.DTO_Risk_Factors")
        for df in inp.iter_dataframes( chunksize=10500):
            print (len(df))
            # preprocess\

            df_temp = df.iloc[:, start:end ]
            print(df_temp.shape)
            
            # Write the processed dataframe
            writer.write_dataframe(df_temp)
            
    start +=  step
    end +=  step
    

 

 

 

 

 

 

 

0 Kudos
1 Reply
SarinaS
Dataiker
Dataiker

Hi @papam,

I think it should work to use the following two lines to write the schema + dataset given the example that you outline: 

dataiku.Dataset.write_schema_from_dataframe(out, df_temp)
dataiku.Dataset.write_dataframe(out, df_temp)

And here's an example of what your for loop might look like with this setup:

# iterate over output datasets 
for out in  sql_datasets:
   out = dataiku.Dataset(out)
 
   for df in inp.iter_dataframes( chunksize=10500):
        df_temp = df.iloc[:, start:end ]
        # Write the processed dataframe
        dataiku.Dataset.write_schema_from_dataframe(out, df_temp)
        dataiku.Dataset.write_dataframe(out, df_temp)

There's a little more information on this in the Note in this section.

Hope that helps,

Sarina

A banner prompting to get Dataiku DSS
Public