@@ -31,12 +31,10 @@ def generate_batch(batch_len):
3131 pa .Array .from_pandas (pd .Categorical ((random .choice (cities ) for i in range (batch_len )), cities ))
3232 ], ['lat' , 'lng' , 'origin' , 'destination' ])
3333
34- def write_record_batches (fd , batch_len , num_batches ):
35- writer = pa .ipc .RecordBatchStreamWriter (fd , generate_batch (1 ).schema )
36- for batch in range (num_batches ):
37- writer .write_batch (generate_batch (batch_len ))
38-
39- writer .close ()
34+ def write_record_batches (filename , batch_len , num_batches ):
35+ with pa .ipc .RecordBatchStreamWriter (filename , generate_batch (1 ).schema ) as writer :
36+ for _ in range (num_batches ):
37+ writer .write_batch (generate_batch (batch_len ))
4038
4139if __name__ == "__main__" :
4240 import argparse
@@ -48,6 +46,5 @@ def write_record_batches(fd, batch_len, num_batches):
4846
4947 args = parser .parse_args ()
5048
51- print "Writing {} {}-element batches to '{}'" .format (args .num_batches , args .batch_size , args .filename )
52- with open (args .filename , 'w' ) as fd :
53- write_record_batches (fd , args .batch_size , args .num_batches )
49+ print ("Writing {} {}-element batches to '{}'" .format (args .num_batches , args .batch_size , args .filename ))
50+ write_record_batches (args .filename , args .batch_size , args .num_batches )
0 commit comments