Datasets: ImportError: cannot import name 'extract_zipped_paths' from 'requests.utils'

[louis925]

Short description
ImportError: cannot import name 'extract_zipped_paths' from 'requests.utils'

Environment information

• Operating System: Windows 10 Pro 64bits
• Python version: 3.7.5
• tensorflow-datasets/tfds-nightly version: tensorflow-datasets 3.1.0
• tensorflow/tensorflow-gpu/tf-nightly/tf-nightly-gpu version: tensorflow 2.1.0

Reproduction instructions

pip install tensorflow-datasets

This will also install the latest requests==2.23.0.

However, it is not compatible with 2.23.0.

import tensorflow_datasets as tfds

This give you

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-30-46a8a2031c9c> in <module>()
----> 1 import tensorflow_datasets as tfds

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\tensorflow_datasets\__init__.py in <module>()
     44 # needs to happen before anything else, since the imports below will try to
     45 # import tensorflow, too.
---> 46 from tensorflow_datasets.core import tf_compat
     47 tf_compat.ensure_tf_install()
     48 

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\tensorflow_datasets\core\__init__.py in <module>()
     21 tf_compat.ensure_tf_install()
     22 
---> 23 from tensorflow_datasets.core.dataset_builder import BeamBasedBuilder  # pylint:disable=g-import-not-at-top
     24 from tensorflow_datasets.core.dataset_builder import BuilderConfig
     25 from tensorflow_datasets.core.dataset_builder import DatasetBuilder

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\tensorflow_datasets\core\dataset_builder.py in <module>()
     33 from tensorflow_datasets.core import constants
     34 from tensorflow_datasets.core import dataset_utils
---> 35 from tensorflow_datasets.core import download
     36 from tensorflow_datasets.core import file_format_adapter
     37 from tensorflow_datasets.core import lazy_imports_lib

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\tensorflow_datasets\core\download\__init__.py in <module>()
     17 
     18 from tensorflow_datasets.core.download.checksums import add_checksums_dir
---> 19 from tensorflow_datasets.core.download.download_manager import DownloadConfig
     20 from tensorflow_datasets.core.download.download_manager import DownloadManager
     21 from tensorflow_datasets.core.download.extractor import iter_archive

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\tensorflow_datasets\core\download\download_manager.py in <module>()
     32 from tensorflow_datasets.core import utils
     33 from tensorflow_datasets.core.download import checksums
---> 34 from tensorflow_datasets.core.download import downloader
     35 from tensorflow_datasets.core.download import extractor
     36 from tensorflow_datasets.core.download import resource as resource_lib

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\tensorflow_datasets\core\download\downloader.py in <module>()
     31 import promise
     32 import requests
---> 33 from requests.utils import extract_zipped_paths
     34 from six.moves import urllib
     35 import tensorflow as tf

ImportError: cannot import name 'extract_zipped_paths'

Link to logs
If applicable,

Expected behavior
Limit the request==2.21.0 in the setup.py of tensorflow-datasets pip installation script so it can be properly import.

Additional context
The current tensorflow-datasets package is not compatible with the latest requests==2.23.0 package but only compatible with 2.21.0. However, the tensorflow-datasets didn't specify the upper compatible version, which cause new user not able to import the package properly.

See also: https://stackoverflow.com/questions/57890219/importerror-cannot-import-name-extract-zipped-paths

[Conchylicultor]

Thank you for reporting. This should have been fixed with tfds-nightly. Could you try to install tfds-nightly and check that the issue has been fixed ?

[vijayphoenix]

Closing the issue as it seems to have been fixed in tfds-nightly.
Feel free to open the issue if you are still facing some problems