TIL from reading an early Jan post on the Django mailing list. Turns out that PyPI is making a whole bunch of data available on Google BigQuery.
In a subsequent post, Alex Gaynor gave us a query that extracts Python versions used to download a package on PyPI in the previous two weeks:
SELECT
REGEXP_EXTRACT(details.python, r"^([^\.]+\.[^\.]+\.[^\.]+)") as python_version,
COUNT(*) as download_count,
FROM
TABLE_DATE_RANGE(
[the-psf:pypi.downloads],
DATE_ADD(CURRENT_TIMESTAMP(), -2, "week"),
CURRENT_TIMESTAMP()
)
WHERE
LOWER(file.project) = 'django'
GROUP BY
python_version,
ORDER BY
download_count DESC
LIMIT 100
Alex's post contains a table for download of Django (Python 3.5.2 is the most popular, quite a relief there!)
Here's the top 10 for piprot (not as popular as I'd hoped):
| Row | python_version | download_count |
|---|---|---|
| 1 | null | 826 |
| 2 | 3.4.5 | 572 |
| 3 | 2.7.9 | 220 |
| 4 | 3.5.2 | 179 |
| 5 | 2.7.6 | 170 |
| 6 | 2.7.13 | 163 |
| 7 | 2.7.12 | 87 |
| 8 | 3.5.1 | 57 |
| 9 | 3.3.6 | 34 |
| 10 | 3.4.4 | 34 |
Alex's post: https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/django-developers/Qab-hRG-SKs/0u1J-fNQFAAJ Original PyPI announcement from May 2016: https://mail.python.org/pipermail/distutils-sig/2016-May/028986.html