The contents of PyPI, in numbers

Total files

1.05 Billion

78,926,653 unique

Total lines of text

327.9 Billion

327,927,103,998 to be precise

Total uncompressed size

56.3 TiB

That is ~42,169,471.024 floppy disks

Lines of code added per second

3,067

In the month 2023-08-01

This page contains a breakdown of the contents of PyPI from parsing the contents of packages. You can download PyPI locally to do your own analysis or run SQL queries on the data in your browser

Language Features

This data only counts unique projects, not versions. e.g if a project has published 10 versions in a month, each containing an async function, it will only be counted once.

Breakdown

Name	Projects	Percent
list comp	226,072	49
fstring	155,226	34
annotations	131,143	29
generator expression	121,229	26
dict comp	89,232	19
async	29,329	6
dataclasses	26,935	6
set comp	21,746	5
walrus	9,728	2
match	2,676	1
async comp	1,064	0
try star	22	0

Project Contents

This data only counts unique projects, not versions. e.g if a project has published 10 versions in a month, each with a setup.py file, it will only be counted once.

Secrets Detected

PyPI contains a lot of secrets.

type	count
Google API Key	4,053
OpenAI API Key	3,568
Tencent Cloud Secret ID	1,901
Amazon AWS Secret Access Key	1,640
Amazon AWS Access Key ID	1,373
Google Cloud Private Key ID	1,098
Slack API Token	1,063
Telegram Bot Token	868
Slack Incoming Webhook URL	784
SendGrid API Key	748
Mailgun API Key	718
Mailchimp API Key	676
Stripe API Key	662
Twilio Account String Identifier	576
Alibaba Cloud AccessKey Secret	555
Total	25,826

Growth

8.92 million

60TB

1 billion

PyPI is growing fast. If this dangerous expansion not stopped, our advanced machine learning models predict that in only 8 years the number of packages will outnumber human beings.

Witness this inevitable future

Binary files

This shows a breakdown of the binary files on PyPI, by extension. Binary files are the vast majority of the content on PyPI, accounting for nearly 75% of the uncompressed size.

extension	total files	total size	unique files
.so	6,415,869	19.7 TiB	3,574,770
.pyd	1,713,498	4.0 TiB	1,573,713
.dylib	1,058,397	2.7 TiB	348,995
.dll	1,266,881	1.9 TiB	372,103
No extension	4,754,719	1.9 TiB	1,663,203
.2	149,114	1.4 TiB	17,297
.0	586,225	1.2 TiB	65,414
.jar	412,313	837.4 GiB	42,548
.png	24,858,234	491.2 GiB	765,834
.1	284,505	447.2 GiB	39,041
.lib	112,458	430.6 GiB	32,836
.exe	194,819	390.8 GiB	44,652
.gz	4,168,631	380.0 GiB	548,053
.tgz	335,368	357.9 GiB	154,603
.7	36,771	304.7 GiB	2,497
Total	46,347,802	36.4 TiB	9,245,559

Largest Projects by size

Tensorflow dominates this list with 8.9 TiB of uncompressed data, 16% of all data on PyPI.

project name	unique files	total files	total lines	total size
tf-nightly	85,750	20,242,423	8,325,796,068	2.3 TiB
tf-nightly-cpu	80,842	19,960,824	8,080,533,316	1.8 TiB
tf-nightly-gpu	70,457	11,657,756	4,814,002,048	1.4 TiB
lalsuite	1,719,153	9,969,505	4,542,135,454	1.1 TiB
tensorflow	98,007	7,661,737	2,882,658,484	864.7 GiB
paddlepaddle-gpu	30,859	2,000,670	433,161,447	856.7 GiB
tensorflow-io-nightly	14,020	927,623	116,109,192	742.5 GiB
tf-nightly-cpu-aws	49,655	8,372,198	3,236,007,762	683.7 GiB
tensorflow-gpu	83,932	4,276,504	1,575,491,698	638.0 GiB
catboost-dev	32,526	256,620	66,348,683	582.2 GiB
tf-nightly-intel	77,131	7,263,510	2,958,885,573	488.4 GiB
tensorflow-cpu	55,842	4,899,681	1,903,928,312	476.2 GiB
tf-nightly-macos	25,662	4,949,069	2,355,608,468	463.8 GiB
OpenVisus	60,484	3,708,820	730,470,294	437.1 GiB
graphscope-client	36,304	1,294,627	149,056,559	425.4 GiB
Total	2,520,624	107,441,567	42,170,193,358	13.1 TiB

Stats By Extensions

This only considers the last suffix of the file path as the extension

extension	total files	total lines	total size	unique files
.py	456,168,853	119,469,907,893	4.4 TiB	30,833,340
.h	80,898,960	24,382,766,597	963.8 GiB	693,320
No extension	57,772,234	7,667,850,460	2.3 TiB	15,549,473
.json	52,329,206	21,161,457,055	1.0 TiB	1,574,195
.hpp	38,587,034	7,946,850,198	320.6 GiB	278,247
.txt	37,176,488	17,224,921,091	635.3 GiB	3,289,949
.js	30,938,557	12,515,763,947	1.0 TiB	1,299,028
.png	24,893,739	899,161	491.3 GiB	768,446
.rst	21,516,495	1,350,186,975	51.8 GiB	1,229,223
.svg	16,958,243	1,361,240,312	171.2 GiB	314,635
.pyi	16,357,154	3,166,271,355	104.7 GiB	456,382
.html	15,165,631	2,830,097,499	201.3 GiB	1,586,120
.yaml	11,662,147	1,102,201,412	39.6 GiB	326,808
.pyc	10,661,980	276,458	67.2 GiB	4,970,135
.yml	9,967,343	717,085,606	21.3 GiB	297,940
Total	881,054,064	220,897,776,019	11.7 TiB	63,467,241

Files not committed to Github

Not all files can be committed to GitHub due to size limits. Some have a few very, very long lines whilst others are junk like mistakenly added virtualenvs or VCS directories. This table shows a breakdown of the reasons why files where skipped.

skip reason	count	unique files	max size	max lines	total size	total lines	total projects
empty	30,857,863	1	0 B	0	0 B	0	209,872
binary	82,649,485	15,947,755	1.9 GiB	1	41.8 TiB	127,683	107,435
virtualenv	9,458,976	433,396	2.1 MiB	71,126	89.5 GiB	2,705,495,750	5,660
too-large	6,575,134	751,114	1.4 GiB	20,010,001	7.5 TiB	136,015,624,311	31,208
text-long-lines	1,371,711	150,743	5.0 MiB	5	138.8 GiB	2,235,288	9,879
version-control	163,374	27,081	197.0 KiB	10,500	175.0 MiB	4,507,173	608
Total	131,076,543	17,310,090	3.3 GiB	20,091,633	49.5 TiB	138,727,990,205	364,662

The contents of PyPI, in numbers

Language Features

This data only counts unique projects, not versions. e.g if a project has published 10 versions in a month, each containing an async function, it will only be counted once.

Mature Features

New Features

Comprehensions

Breakdown

Project Contents

This data only counts unique projects, not versions. e.g if a project has published 10 versions in a month, each with a setup.py file, it will only be counted once.

Setup.py vs PyProject.toml

Markdown vs RST

Other Files

Typing?