3

Good afternoon everyone, first of all I am new to python, so please, bear with me.

I am trying to read and manipulate a .parquet file, so I looked up on the internet what should I do and I found that I should use pyarrow or fastparquet.

So I tried pip install pyarrow on my jupyter botebook and it won't stop running (there is an * on the left side of the cell). So I tried on my command prompt and I got the following error:

Error: Failed building wheel for pyarrow (in red)

Failed to build pyarrow (in white)

Error: Could not build wheels for pyarrow which use PEP 517 and cannot be installed directly (in red)

Of course there is more red text but I was not sure that it was worth copying all and pasting it here. Perhaps these last 3 lines will give you guys a hint on what should I do.

I tried the same for fastparquet, and got the same response on jupyter botebook a mesma situação (there is an * on the left side of the cell) and on the command prompt there is a different response, a lot of red text and it seems here that I need Microsoft Visual C++ 14.0 (Microsoft Visual C++ 14.0 is required.).

Could anyone be kind enough to help me understand why I can't install pyarrow? Or if there is a simpler way to read and manipulate parquet files that I am not aware of I would appreciate.

My Python version is 3.7.4.

feetwet
  • 3,248
  • 7
  • 46
  • 84
Beatriz Campos
  • 31
  • 2
  • hmm how did u try to download it ?did you use in cmd ```py -m pip install pyarrow``` ? – Shad0w Feb 05 '21 at 13:27
  • Hi @Shad0w, I tried "pip install pyarrow" (according to: [https://pypi.org/project/pyarrow/]), I did not know that it had to be "py -m pip install pyarrow". – Beatriz Campos Feb 05 '21 at 13:35
  • did it work for you? cause I just tried it and it worked for me. – Shad0w Feb 05 '21 at 13:37
  • 1
    `pip install pyarrow` and `python -m pip install pyarrow` shouldn't make a big difference. The preferred way to install `pyarrow` is to use `conda` instead of `pip` as this will always install a fitting binary. If you need to stay with `pip`, I would though recommend to update `pip` itself first by running `python -m pip install -U pip` as you might need a newer `pip` version to detect the compiled `pyarrow` package on PyPI. If this still doesn't work, feel free to open an issue at https://issues.apache.org/jira/projects/ARROW/issues with the Apache Arrow project itself. – Uwe L. Korn Feb 05 '21 at 13:55
  • @Shad0w, did not work, I get the same error as listed above. – Beatriz Campos Feb 05 '21 at 14:59
  • @UweL.Korn, I have recently updated my pip, so I tried the command you suggested and: **Requirement already satisfied**, tried again the `pip install` and also the `python -m pip install pyarrow` version and they still don't work. I will try the Apache Arrow Project Itself. Thank you very much, both you guys. – Beatriz Campos Feb 05 '21 at 15:02
  • Hey guys, just noticed here that it says that **RuntimeError: Not supported on 32-bit Windows**, but my system says it is a 64-bit operating system. When I type python on my command prompt this is what I get **Python 3.4.2 (v3.4.2:ab2c023a9432, Oct 6 2014, 22:16:31) [MSC v.1600 64 bit (AMD64)] on win32**. – Beatriz Campos Feb 05 '21 at 15:25
  • 1
    Python 3.4 is definitely a too old Python version, you will need at least 3.6. – Uwe L. Korn Feb 05 '21 at 19:02
  • It could also be that you have installed a 32bit Python on a 64bit machines, then you will also get the above error. But you should definitely install a more recent Python like 3.9. – Uwe L. Korn Feb 05 '21 at 19:03
  • Have you managed to solve that problem? I'm facing the same issue. – IdoS Jun 07 '21 at 11:19
  • Yes @IdoS, I have, take a look on what has been discussed. Were the comments not helpful to you? – Beatriz Campos Jun 07 '21 at 12:26

0 Answers0