[infra] AttributeError: np.unicode_ was removed in the NumPy 2.0 release. Use np.str_ instead.

[aspeddro]

Descrição
Erro ao fazer o import quando o pacote python é instalado com todos os grupos [all] ou [avro]

Reproduzir

cd /tmp
mkdir beta-21
cd beta-21

python --version

Python 3.9.17

python -m venv .venv
source .venv/bin/activate
python -m pip install "basedosdados[all]==2.0.0b21"

python -m pip show basedosdados

Name: basedosdados
Version: 2.0.0b21
Summary: Organizar e facilitar o acesso a dados brasileiros através de tabelas públicas no BigQuery.
Home-page: https://github.com/basedosdados/mais
Author: Joao Carabetta
Author-email: [email protected]
License: MIT
Location: /tmp/beta-21/.venv/lib/python3.9/site-packages
Requires: google-api-python-client, google-cloud-bigquery, google-cloud-bigquery-connection, google-cloud-bigquery-storage, google-cloud-storage, loguru, pandas, pandas-gbq, pydata-google-auth, tomlkit, tqdm
Required-by:

Python 3.9.17 (main, Jun 30 2023, 18:04:45)
[GCC 13.1.1 20230429] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import basedosdados
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/tmp/beta-21/.venv/lib/python3.9/site-packages/basedosdados/__init__.py", line 27, in <module>
    from basedosdados.upload.table import Table
  File "/tmp/beta-21/.venv/lib/python3.9/site-packages/basedosdados/upload/table.py", line 23, in <module>
    from basedosdados.upload.datatypes import Datatype
  File "/tmp/beta-21/.venv/lib/python3.9/site-packages/basedosdados/upload/datatypes.py", line 11, in <module>
    import pandavro
  File "/tmp/beta-21/.venv/lib/python3.9/site-packages/pandavro/__init__.py", line 22, in <module>
    np.unicode_: 'string',
  File "/tmp/beta-21/.venv/lib/python3.9/site-packages/numpy/__init__.py", line 397, in __getattr__
    raise AttributeError(
AttributeError: `np.unicode_` was removed in the NumPy 2.0 release. Use `np.str_` instead.
>>>

[aspeddro]

Não apareceu só em 2.0.0b21

[aspeddro]

Relacionado ynqa/pandavro#56

[aspeddro]

Acho que podemos adicionar uma restrição para o numpy, <2.0