sab/pypinfo

Easily view PyPI download statistics via Google's BigQuery.

pypinfo: View PyPI download statistics with ease.

---

pypinfo is a simple CLI to access PyPI download statistics via Google's BigQuery.

Installation

pypinfo is distributed on PyPI as a universal wheel and is available on Linux/macOS and Windows and supports Python 3.5+ and PyPy.

This is relatively painless, I swear.

1. Go to https://bigquery.cloud.google.com.
2. Sign up if you haven't already. The first TB of queried data each month is free. Each additional TB is $5.
3. Go to https://console.developers.google.com/cloud-resource-manager and create a new project if you don't already have one. Any name is fine, but I recommend you choose something to do with PyPI like pypinfo. This way you know what the project is designated for.
4. Go to https://console.cloud.google.com/apis/api/bigquery-json.googleapis.com/overview and make sure the correct project is chosen using the drop-down on top. Click the button on top to enable.
5. Follow https://cloud.google.com/storage/docs/authentication#generating-a-private-key to create credentials in JSON format. During creation, choose BigQuery User as role. (If BigQuery is not an option in the list, wait 15-20 minutes and try creating the credentials again.) After creation, note the download location. Move the file wherever you want.
6. pip install pypinfo
7. pypinfo --auth path/to/your_credentials.json, or set an environment variable GOOGLE_APPLICATION_CREDENTIALS that points to the file.

Usage

$ pypinfo
Usage: pypinfo [OPTIONS] [PROJECT] [FIELDS]... COMMAND [ARGS]...

  Valid fields are:

  project | version | file | pyversion | percent3 | percent2 | impl | impl-version |

  openssl | date | month | year | country | installer | installer-version |

  setuptools-version | system | system-release | distro | distro-version | cpu

Options:
  -a, --auth TEXT         Path to Google credentials JSON file.
  --run / --test          --test simply prints the query.
  -j, --json              Print data as JSON, with keys `rows` and `query`.
  -i, --indent INTEGER    JSON indentation level.
  -t, --timeout INTEGER   Milliseconds. Default: 120000 (2 minutes)
  -l, --limit TEXT        Maximum number of query results. Default: 10
  -d, --days TEXT         Number of days in the past to include. Default: 30
  -sd, --start-date TEXT  Must be negative. Default: -31
  -ed, --end-date TEXT    Must be negative. Default: -1
  -w, --where TEXT        WHERE conditional. Default: file.project = "project"
  -o, --order TEXT        Field to order by. Default: download_count
  -p, --pip               Only show installs by pip.
  -pc, --percent          Print percentages.
  -md, --markdown         Output as Markdown.
  --version               Show the version and exit.
  --help                  Show this message and exit.

pypinfo accepts 0 or more options, followed by exactly 1 project, followed by 0 or more fields. By default only the last 30 days are queried. Let's take a look at some examples!

Tip: If queries are resulting in NoneType errors, increase timeout.

Downloads for a project

$ pypinfo requests
Served from cache: False
Data processed: 6.87 GiB
Data billed: 6.87 GiB
Estimated cost: $0.04

| download_count |
| -------------- |
|      9,316,415 |

All downloads

$ pypinfo ""
Served from cache: False
Data processed: 0.00 B
Data billed: 0.00 B
Estimated cost: $0.00

| download_count |
| -------------- |
|    661,224,259 |

Downloads for a project by Python version

$ pypinfo django pyversion
Served from cache: False
Data processed: 10.81 GiB
Data billed: 10.81 GiB
Estimated cost: $0.06

| python_version | download_count |
| -------------- | -------------- |
| 3.5            |        539,194 |
| 2.7            |        495,207 |
| 3.6            |        310,750 |
| None           |         84,524 |
| 3.4            |         64,621 |
| 3.7            |          3,022 |
| 2.6            |          2,966 |
| 3.3            |          1,638 |
| 1.17           |            285 |
| 3.2            |            188 |
| 3.1            |              4 |
| 2.5            |              3 |

All downloads by country code

$ pypinfo "" country
Served from cache: False
Data processed: 2.40 GiB
Data billed: 2.40 GiB
Estimated cost: $0.02

| country | download_count |
| ------- | -------------- |
| US      |    420,722,571 |
| CN      |     27,235,750 |
| IE      |     24,011,857 |
| DE      |     19,112,463 |
| GB      |     18,485,428 |
| FR      |     17,394,541 |
| None    |     15,867,055 |
| JP      |     12,381,087 |
| CA      |     11,666,733 |
| KR      |     10,239,761 |
| AU      |      9,573,248 |
| SG      |      8,500,881 |
| IN      |      8,467,755 |
| RU      |      6,243,255 |
| NL      |      6,096,337 |
| BR      |      5,992,892 |
| IL      |      4,924,533 |
| PL      |      2,902,368 |
| HK      |      2,873,318 |
| SE      |      2,604,146 |

Downloads for a project by system and distribution

$ pypinfo cryptography system distro
Served from cache: False
Data processed: 14.75 GiB
Data billed: 14.75 GiB
Estimated cost: $0.08

| system_name | distro_name                     | download_count |
| ----------- | ------------------------------- | -------------- |
| Linux       | Ubuntu                          |      1,314,938 |
| Linux       | Debian GNU/Linux                |        381,857 |
| Linux       | None                            |        359,993 |
| Linux       | CentOS Linux                    |        210,950 |
| Linux       | Amazon Linux AMI                |        198,807 |
| None        | None                            |        179,950 |
| Windows     | None                            |        176,495 |
| Darwin      | macOS                           |         75,030 |
| Linux       | Alpine Linux                    |         66,296 |
| Linux       | CentOS                          |         62,812 |
| Linux       | Red Hat Enterprise Linux Server |         47,030 |
| Linux       | debian                          |         33,601 |
| Linux       | Raspbian GNU/Linux              |         29,467 |
| Linux       | Fedora                          |         20,112 |
| Linux       | openSUSE Leap                   |         11,549 |
| Darwin      | OS X                            |          6,970 |
| Linux       | Linux                           |          6,894 |
| Linux       | Virtuozzo                       |          6,611 |
| FreeBSD     | None                            |          5,898 |
| Linux       | RedHatEnterpriseServer          |          4,415 |

Most popular projects in the past year

$ pypinfo --days 365 "" project
Served from cache: False
Data processed: 87.84 GiB
Data billed: 87.84 GiB
Estimated cost: $0.43

| project         | download_count |
| --------------- | -------------- |
| simplejson      |    267,459,163 |
| six             |    213,697,561 |
| setuptools      |    164,144,954 |
| botocore        |    162,843,025 |
| python-dateutil |    159,786,908 |
| pip             |    155,164,096 |
| pyasn1          |    142,647,378 |
| requests        |    141,811,313 |
| docutils        |    136,073,108 |
| pyyaml          |    127,183,654 |
| jmespath        |    126,997,657 |
| s3transfer      |    123,275,444 |
| futures         |    121,993,875 |
| awscli          |    119,512,669 |
| rsa             |    112,884,251 |
| colorama        |    107,995,099 |
| idna            |     79,363,400 |
| wheel           |     79,098,241 |
| selenium        |     72,291,821 |
| awscli-cwlogs   |     69,708,863 |

Percentage of Python 3 downloads of the top 100 projects in the past year

Let's use --test to only see the query instead of sending it.

$ pypinfo --test --days 365 --limit 100 "" project percent3
SELECT
  file.project as project,
  ROUND(100 * SUM(CASE WHEN REGEXP_EXTRACT(details.python, r"^([^\.]+)") = "3" THEN 1 ELSE 0 END) / COUNT(*), 1) as percent_3,
  COUNT(*) as download_count,
FROM
  TABLE_DATE_RANGE(
    [the-psf:pypi.downloads],
    DATE_ADD(CURRENT_TIMESTAMP(), -366, "day"),
    DATE_ADD(CURRENT_TIMESTAMP(), -1, "day")
  )
GROUP BY
  project,
ORDER BY
  download_count DESC
LIMIT 100

Credits

• Donald Stufft for maintaining PyPI all these years.
• Google for donating BigQuery capacity to PyPI.
• Paul Kehrer for his awesome blog post.

Changelog

Important changes are emphasized.

14.0.0

• Added new file field!

13.0.0

• Added last_update JSON key, which is a UTC timestamp.

12.0.0

• Breaking: JSON output is now a mapping with keys rows, which is all the data that was previously outputted, and query, which is relevant metadata.
• Increased the resolution of percentages.

11.0.0

• Fixed JSON output.

10.0.0

• Fixed custom field ordering.

9.0.0

• Added new BigQuery usage stats.
• Lowered the default number of results to 10 from 20.
• Updated examples.
• Fixed table formatting regression.

8.0.0

• Updated google-cloud-bigquery dependency.

7.0.0

• Output table is now in Markdown format for easy copying to GitHub issues and PRs.

6.0.0

• Updated google-cloud-bigquery dependency.

5.0.0

• Numeric output (non-json) is now prettier (thanks hugovk)
• You can now filter results for only pip installs with the --pip flag (thanks hugovk)

4.0.0

• --order now works with all fields (thanks Brian Skinn)
• Updated installation docs (thanks Brian Skinn)

3.0.1

• Fix: project names are now normalized to adhere to PEP 503.

3.0.0

• Breaking: --json option is now just a flag and prints output as prettified JSON.

2.0.0

• Added --json path option.

1.0.0

• Initial release