Add tfrecord gzip compression

fhoering commented 5 years ago

added in TFRecordFileInputFormat & TFRecordFileOutputFormat
improved unit tests to handle both cases
didn't change TFRecordFileOutputFormatV1 (couldn't test and not used by SparkConnector)
the most risky is the readFully reimplementation, if you think this is too risky we can also remove it, it was essentially done for the unit test, I don't actually use the TFRecordFileInputFormat

Tested in Python with (ZLIB option is not supported)

import tensorflow as tf
specs = {
    'data': tf.VarLenFeature(dtype=tf.int64)
}
dataset = tf.data.TFRecordDataset(["tf-records/part-m-00000"], compression_type="")
dataset = dataset.map(
    lambda x: tf.parse_single_example(x, features=specs)
)
next_tfr = dataset.make_one_shot_iterator().get_next()
with tf.Session() as sess:
    try:
        for i in range(10):
            print(sess.run(next_tfr))
    except tf.errors.OutOfRangeError:
        pass

import tensorflow as tf
specs = {
    'data': tf.VarLenFeature(dtype=tf.int64)
}
dataset = tf.data.TFRecordDataset(["tf-records/part-m-00000.gz"], compression_type="GZIP")
dataset = dataset.map(
    lambda x: tf.parse_single_example(x, features=specs)
)
next_tfr = dataset.make_one_shot_iterator().get_next()
with tf.Session() as sess:
    try:
        for i in range(10):
            print(sess.run(next_tfr))
    except tf.errors.OutOfRangeError:
        pass

googlebot commented 5 years ago

Thanks for your pull request. It looks like this may be your first contribution to a Google open source project (if not, look below for help). Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

:memo: Please visit https://cla.developers.google.com/ to sign.

Once you've signed (or fixed any issues), please reply here (e.g. I signed it!) and we'll verify it.

What to do if you already signed the CLA

Individual signers

It's possible we don't have your GitHub username or you're using a different email address on your commit. Check your existing CLA data and verify that your email is set on your git commits.

Corporate signers

Your company has a Point of Contact who decides which employees are authorized to participate. Ask your POC to be added to the group of authorized contributors. If you don't know who your Point of Contact is, direct the Google project maintainer to go/cla#troubleshoot (Public version).
The email used to register you as an authorized contributor must be the email used for the Git commit. Check your existing CLA data and verify that your email is set on your git commits.
The email used to register you as an authorized contributor must also be attached to your GitHub account.

ℹ️ Googlers: Go here for more info.

jhseu commented 5 years ago

Mind handling the CLA?

googlebot commented 5 years ago

CLAs look good, thanks!

ℹ️ Googlers: Go here for more info.

jhseu commented 5 years ago

Have you tested the contents of the TFRecords files? The test just seems to parse them only.

fhoering commented 5 years ago

I changed the test again to have tests for a fixed dataset. I also compared the tf record content with what is read by the python code.

tensorflow / ecosystem