In this section, we present how to connect PyPI to MindsDB.

PyPI is a host for maintaining and storing Python packages. It’s a good place for publishing your Python packages in different versions and releases.

Data from PyPI can be utilized within MindsDB to train models and make predictions about your Python packages.

​
Connection

This handler is implemented using the standard Python requests library. It is used to connect to the RESTful service that pypistats.org is serving.

There are no connection arguments required to initialize the handler.

To connect to PyPI using MindsDB, the following CREATE DATABASE statement can be used:

CREATE DATABASE pypi_datasource
WITH ENGINE = 'pypi'

​
Usage

Now, you can use the following queries to view the statistics for Python packages (MindsDB, for example):

Overall downloads, including mirrors:

SELECT *
FROM pypi_datasource.overall WHERE package="mindsdb" AND mirrors=true;

Overall downloads on CPython==2.7:

SELECT *
FROM pypi_datasource.python_minor WHERE package="mindsdb" AND version="2.7";

Recent downloads:

SELECT *
FROM pypi_datasource.recent WHERE package="mindsdb";

Recent downloads in the last day:

SELECT *
FROM pypi_datasource.recent WHERE package="mindsdb" AND period="day";

All downloads on Linux-based distributions:

SELECT date, downloads
FROM pypi_datasource.system WHERE package="mindsdb" AND os="Linux";

Each table takes a required package argument in the WHERE clause, which is the name of the package you want to query.

​
Supported Tables

The following tables are supported by the PyPI handler:

  • overall: daily download quantities for packages.
  • recent: recent download quantities for packages.
  • python_major: daily download quantities for packages, grouped by Python major version.
  • python_minor: daily download quantities for packages, grouped by Python minor version.
  • system: daily download quantities for packages, grouped by operating system.