tf.raw_ops.StringSplitV2 does not limit a number of splits by max_split attribute

[rkazants]

Issue type

Bug

Have you reproduced the bug with TensorFlow Nightly?

Yes

Source

source

TensorFlow version

2.16.1

Custom code

Yes

OS platform and distribution

No response

Mobile device

No response

Python version

No response

Bazel version

No response

GCC/compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current behavior?

tf.raw_ops.StringSplitV2 does not limit a number of splits by max_split attribute.

Standalone code to reproduce the issue

import tensorflow as tf

tf.raw_ops.StringSplitV2(input=['one<>'], sep='<>', maxsplit=1) # outputs two splits 'one' and '' instead of just one split

tf.raw_ops.StringSplitV2(input=['one<>two'], sep='<>', maxsplit=1) # outputs two splits 'one' and 'two' instead of just one split

Relevant log output

>>> tf.raw_ops.StringSplitV2(input=['one<>'], sep='<>', maxsplit=1)
StringSplitV2(indices=<tf.Tensor: shape=(2, 2), dtype=int64, numpy=
array([[0, 0],
       [0, 1]])>, values=<tf.Tensor: shape=(2,), dtype=string, numpy=array([b'one', b''], dtype=object)>, shape=<tf.Tensor: shape=(2,), dtype=int64, numpy=array([1, 2])>)

>>> tf.raw_ops.StringSplitV2(input=['one<>two'], sep='<>', maxsplit=1)
StringSplitV2(indices=<tf.Tensor: shape=(2, 2), dtype=int64, numpy=
array([[0, 0],
       [0, 1]])>, values=<tf.Tensor: shape=(2,), dtype=string, numpy=array([b'one', b'two'], dtype=object)>, shape=<tf.Tensor: shape=(2,), dtype=int64, numpy=array([1, 2])>)

[sushreebarsa]

@rkazants I was able to replicate the issue reported here. Could you consider using tf.strings.split instead. It offers an optional max_split parameter to limit the number of splits: https://www.tensorflow.org/api_docs/python/tf/strings/split
Thank you!

[github-actions]

This issue is stale because it has been open for 7 days with no activity. It will be closed if no further activity occurs. Thank you.

[rkazants]

Hi @sushreebarsa,

please remove stale label because the issue is not fixed.

Best regards,
Roman