ez_transfer.datasets

csv_reader

class easytransfer.datasets.csv_reader.CSVReader(input_glob, batch_size, is_training, thread_num=1, input_queue=None, output_queue=None, job_name='DISTCSVReader', **kwargs)[source]

Read csv format

Parameters:
  • input_glob -- input file fp
  • batch_size -- input batch size
  • is_training -- True or False
  • thread_num -- thread number
process(input_data)[source]
close()[source]
class easytransfer.datasets.csv_reader.BundleCSVReader(input_glob, batch_size, worker_hosts, task_index, is_training=False, **kwargs)[source]

Read group of csv formats

Parameters:
  • input_glob -- input file fp
  • batch_size -- input batch size
  • worker_hosts -- worker hosts
  • task_index -- task index
  • is_training -- True or False

csv_writer

class easytransfer.datasets.csv_writer.CSVWriter(output_glob, output_schema, input_queue=None, **kwargs)[source]

Writer csv format

Parameters:
  • output_glob -- output file fp
  • output_schema -- output_schema

odps_table_reader

class easytransfer.datasets.odps_table_reader.OdpsTableReader(input_glob, batch_size, is_training, thread_num=1, input_queue=None, output_queue=None, slice_id=0, slice_count=1, job_name='DISTOdpsTableReader', **kwargs)[source]

Read odps table

Parameters:
  • input_glob -- input file fp
  • batch_size -- input batch size
  • is_training -- True or False
process(input_data)[source]

odps_table_writer

class easytransfer.datasets.odps_table_writer.OdpsTableWriter(output_glob, output_schema, slice_id, input_queue, job_name='DistOdpsTableWriter', **kwargs)[source]

Writer odps table

Parameters:
  • output_glob -- output file fp
  • output_schema -- output_schema

tfrecord_reader

class easytransfer.datasets.tfrecord_reader.TFRecordReader(input_glob, batch_size, is_training, thread_num=1, input_queue=None, output_queue=None, job_name='DISTTFRecordReader', **kwargs)[source]

Read tfrecords

Parameters:
  • input_glob -- input file fp
  • batch_size -- input batch size
  • is_training -- True or False
class easytransfer.datasets.tfrecord_reader.BundleTFRecordReader(input_glob, batch_size, worker_hosts, task_index, is_training=False, **kwargs)[source]

tfrecord_writer

class easytransfer.datasets.tfrecord_writer.TFRecordWriter(output_glob, output_schema, input_queue)[source]

Writer tfrecords

Parameters:
  • output_glob -- output file fp
  • output_schema -- output_schema