Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[HOPSWORKS-2338] handle empty tfrecord files in TFDataEngine #230

Merged
merged 4 commits into from
Jan 28, 2021
Merged

[HOPSWORKS-2338] handle empty tfrecord files in TFDataEngine #230

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
1bec37c
filter empty files for tf data
Jan 26, 2021
bca4f84
filter empty files for tf data
Jan 26, 2021
a985d1c
filter empty files for tf data
Jan 26, 2021
2c817a7
filter empty files for tf data
Jan 26, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 5 additions & 16 deletions python/hsfs/core/tfdata_engine.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,9 +76,7 @@ def __init__(
self._training_dataset_format = self._training_dataset.data_format

self._input_files = self._get_training_dataset_files(
self._training_dataset.location,
self._split,
filter_empty=True if self._training_dataset_format == "csv" else False,
self._training_dataset.location, self._split
)

if self._feature_names is None:
Expand Down Expand Up @@ -380,9 +378,7 @@ def _get_tf_dataset(self, input_files, cycle_length):

return dataset, tfrecord_feature_description

def _get_training_dataset_files(
self, training_dataset_location, split, filter_empty=False
):
def _get_training_dataset_files(self, training_dataset_location, split):
"""
returns list of absolute path of training input files
:param training_dataset_location: training_dataset_location
Expand All @@ -395,7 +391,7 @@ def _get_training_dataset_files(

if training_dataset_location.startswith("hopsfs"):
input_files = self._get_hopsfs_dataset_files(
training_dataset_location, split, filter_empty
training_dataset_location, split
)
elif training_dataset_location.startswith("s3"):
input_files = self._get_s3_dataset_files(training_dataset_location, split)
Expand All @@ -405,7 +401,7 @@ def _get_training_dataset_files(
return input_files

@staticmethod
def _get_hopsfs_dataset_files(training_dataset_location, split, filter_empty):
def _get_hopsfs_dataset_files(training_dataset_location, split):
path = training_dataset_location.replace("hopsfs", "hdfs")
if split is None:
path = hdfs.path.abspath(path)
Expand All @@ -417,19 +413,12 @@ def _get_hopsfs_dataset_files(training_dataset_location, split, filter_empty):
all_list = hdfs.ls(path, recursive=True)

# Remove directories and spark '_SUCCESS'
include_file = True
for file in all_list:
# remove empty file if any
if filter_empty:
_file_size = hdfs.path.getsize(file)
if _file_size == 0:
include_file = False
else:
include_file = True
if (
not hdfs.path.isdir(file)
and not file.endswith("_SUCCESS")
and include_file
and hdfs.path.getsize(file) >= 1
):
input_files.append(file)

Expand Down