Cannot install the dependency: lshash3

[avidale]

My package depend on pqkmeans. Some time ago, it started producing errors during installation, because pqkmeans requires lshash3, and for some reason, lshash3 doesn't install anymore.

When I run pip install pqkmeans (Ubuntu, Python 3.10, reproducible in this notebook), I get the following error:

trace

Collecting pqkmeans==1.0.4
  Downloading pqkmeans-1.0.4.tar.gz (158 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 158.5/158.5 kB 3.3 MB/s eta 0:00:00
  Preparing metadata (setup.py) ... done
Requirement already satisfied: numpy in /usr/local/lib/python3.10/dist-packages (from pqkmeans==1.0.4) (1.23.5)
Requirement already satisfied: scikit-learn in /usr/local/lib/python3.10/dist-packages (from pqkmeans==1.0.4) (1.2.2)
Collecting pipe (from pqkmeans==1.0.4)
  Using cached pipe-2.0-py3-none-any.whl (8.8 kB)
Requirement already satisfied: scipy in /usr/local/lib/python3.10/dist-packages (from pqkmeans==1.0.4) (1.11.3)
Collecting typing (from pqkmeans==1.0.4)
  Downloading typing-3.7.4.3.tar.gz (78 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 78.6/78.6 kB 8.3 MB/s eta 0:00:00
  Preparing metadata (setup.py) ... done
Requirement already satisfied: six in /usr/local/lib/python3.10/dist-packages (from pqkmeans==1.0.4) (1.16.0)
Collecting texmex_python==1.0.0 (from pqkmeans==1.0.4)
  Using cached texmex_python-1.0.0-py3-none-any.whl
Collecting lshash3 (from texmex_python==1.0.0->pqkmeans==1.0.4)
  Using cached lshash3-0.0.8.tar.gz (9.4 kB)
  Preparing metadata (setup.py) ... done
Requirement already satisfied: joblib>=1.1.1 in /usr/local/lib/python3.10/dist-packages (from scikit-learn->pqkmeans==1.0.4) (1.3.2)
Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.10/dist-packages (from scikit-learn->pqkmeans==1.0.4) (3.2.0)
Collecting bitarray (from lshash3->texmex_python==1.0.0->pqkmeans==1.0.4)
  Using cached bitarray-2.8.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (286 kB)
Building wheels for collected packages: pqkmeans, typing, lshash3
  Building wheel for pqkmeans (setup.py) ... done
  Created wheel for pqkmeans: filename=pqkmeans-1.0.4-cp310-cp310-linux_x86_64.whl size=236311 sha256=940d560d65bd2e8ecec652ff7756502f85e47a99a9101b0b500afac859cb8948
  Stored in directory: /root/.cache/pip/wheels/84/fa/fe/7cf22e3d319bb9c9652768e4761b2b219541ad0334b127f340
  Building wheel for typing (setup.py) ... done
  Created wheel for typing: filename=typing-3.7.4.3-py3-none-any.whl size=26305 sha256=a64e1e1c1f212dc750d7dffea39c73c41cb5d2e96e166917de49adeefd665c72
  Stored in directory: /root/.cache/pip/wheels/7c/d0/9e/1f26ebb66d9e1732e4098bc5a6c2d91f6c9a529838f0284890
  error: subprocess-exited-with-error
  × python setup.py bdist_wheel did not run successfully.
  │ exit code: 1
  ╰─> See above for output.
  note: This error originates from a subprocess, and is likely not a problem with pip.
  Building wheel for lshash3 (setup.py) ... error
  ERROR: Failed building wheel for lshash3
  Running setup.py clean for lshash3
Successfully built pqkmeans typing
Failed to build lshash3
ERROR: Could not build wheels for lshash3, which is required to install pyproject.toml-based projects

Here is where we have discovered the problem: avidale/compress-fasttext#19

Can you please fix this error, e.g. by switching to a well-maintained implementation of lshash, such as https://github.com/guofei9987/pyLSHash?

[gojomo]

Happened to be poking around after the SO question, & so also wanted to point out:

• lshash3 (not updated since 2017) is only brought in via the mysterious texmex-python package, which has no description on either PyPI or Github, and is maybe just an optional demo dataset that'd be easy to drop from requirements unless needed?
• package lshashpy3 may be another suitable replacement for a texmex-python maintainer to drop-in - seems based on same original Kay Zhu code, so might be API-compatible

[Hi-king]

@avidale @gojomo
Thanks for pointing out the issue with the current installation.
I have created a PR (#47) to move texmex-python to an additional dependency :)

Unfortunately, however, another issue (numpy compatibility) is preventing installation with python 3.10 at the moment: please use Python 3.9 🙇

[Hi-king]

@avidale @gojomo Update: We've released recent v1.0.6, which supports python 3.10 & 3.11. Plz try pip install -U "pqkmeans>=1.0.6"