Track traffic to GitHub repo longer than 14 days

[cirosantilli]

Possible? If not, +1 from me.

Sample reference URL: https://github.com/cirosantilli/test/graphs/traffic

On SE: http://webapps.stackexchange.com/questions/60915

[cirosantilli]

Ivan replied it is not currently possible.

[greckyogurt]

+1

[Danappelxx]

👍

[alexbudmsft]

+1

[saqib-razzaq-aspose]

+1

[xXFracXx]

+1

[sanyamc-msft]

+1 there are many people in my team who dearly want this data. Currently we manually parse the data into a csv and power our charts on a fortnightly basis. Would really appreciate api/ json data, even if its for 14 days.

[lolbot-iichan]

+1

[nakulpathak3]

+1

[wvangeit]

+1

[rljordan-zz]

+1

[ruhbehka]

+1

[KindKitty]

+1

[JarekToro]

+1

[ManoMarks]

+1

[johngronberg-okta]

+1

[ZahediAquino]

+1

[KrisSiegel]

This would be incredibly helpful especially for those maintaining open source libraries.

[timgrossmann]

@burnash Thank you for the answer and clarifying this.

That's pretty odd of github to ask for this.
I'm not really into allowing any platform to access my all my projects with owner right permissions...
Would've loved to use your tool.

Maybe at one point there will be a solution.

Thanks

[geniusdibya]

But doesn't have the Clone data. Can you add that too?

Maybe. It'd be a nice feature to have. And it's has been already requested (repotraffic/repotraffic-feedback#3 and repotraffic/repotraffic-feedback#7)
I've been tracking my own open source project gspread for two years, but haven't really needed to track the number of clones.

@burnash Actually, I want to track what is the real usage of my repo. And, clones/Fork data are got indicators.

Also is the code available on GIT to contribute?

Not yet. But I'm considering open sourcing it (repotraffic/repotraffic-feedback#6) if I have enough time.

@burnash Sure. Would love to contribute.

[glenn-jocher]

I made a little web app that stores and plots traffic data for longer than 14 days. I'll be glad if someone finds it useful.
@burnash This is cool. But doesn't have the Clone data. Can you add that too?
Also is the code available on GIT to contribute?

This works great!

[timqian]

Hi all, I made another open-source web app to record repo traffic, clones, referrers and more. Welcome to try it out

• Website: Repo Analytics
• Source code: GitHub

[timgrossmann]

@timqian Same problem as with the other tooling... It's a shame that GitHub doesn't change that...

[timqian]

@timgrossmann Yeah, unfortunately, this is required 😂 Only thing I can do is to take care of your token. Also when you don't need this service, you can delete the token from your setting anytime

[UbhiTS]

it really is a shame that github has to make users create solutions just to know the extended history and analytics of their repositories :( .. I would prefer not to give anyone write access to my profile and repositories just to know extended analytics. Github, hope you're listening :)

[BradleyA]

Prefer a solution that does NOT give someone else write access to your repositories
A simple shell script that uses crontab to schedule the download of GitHub traffic statistics in plain text from your GitHub repositories. Another script gives you the text lines to add to crontab and sets up the directories for GitHub traffic statistics. It takes minutes to setup but weeks to get enough data. At this point you can process the data with any tools you wish.

I created a third script that parses the GitHub traffic and creates two markdown tables (clone, views) like this clone table.

2019 GitHub clones

Date:	07-29	08-05	08-12	08-19	08-26	09-02	09-09	12-09	12-16	12-23	12-30
Clones:	34	31	2	2	4	2	2	1	2	9	4
Unique clones:	2	3	2	2	2	2	2	1	2	8	2

Total clones: 93

It solves my challenge. It may help you.

You can find the scripts: https://github.com/BradleyA/Linux-admin/tree/master/github-repository-traffic#github-repository-traffic

[MBurtsev]

Great and useful idea, I am sure that it should be implemented.

[UbhiTS]

this is seriously frustrating, in this age of cheap storage, GitHub would not even have a totals counter if not the entire history graph. At least give us Total Hits, Hits in last 1 year and hits in last 1 month. I'm sure having running counters would not add much to you load. For the love of god GitHub, please listen !

[sangonzal]

For anyone still looking for a solution for this, I made a GitHub action that you might find useful: https://github.com/marketplace/actions/repository-traffic

With a GitHub action, the access token that is needed for accessing the GitHub API stays in your control.

[kevinmgates]

For anyone still looking for a solution for this, I made a GitHub action that you might find useful: https://github.com/marketplace/actions/repository-traffic

With a GitHub action, the access token that is needed for accessing the GitHub API stays in your control.

AWESOME!

[matthansen0]

I wrote an automated solution for this also that uses Azure tech - https://github.com/matthansen0/logicapp-githubstats

[UbhiTS]

After my frustrations, I also wrote a solution for myself (and others) who do not wish to share any info with anyone else. No access token, no nothing. I basically trick github and found an easy way to maintain a running counter.

https://medium.com/@ubhits/how-i-beat-githubs-14-day-only-traffic-insights-using-aws-lambda-sqs-dynamodb-and-the-api-98f9216dbe4c

[antonkomarev]

I've made a pet project Ÿ HŸPE which helps me to collect and display GitHub traffic data longer than 14 days and even more. Just login and pick what repositories you want to track.

[InbarGazit]

does it work retroactively? or only from that point forward?

[antonkomarev]

It can only start collecting data after installation -14 days. GitHub doesn't provide data from the past periods... 😭

[glenn-jocher]

@antonkomarev @timqian do you know any way (paid or otherwise) that allows for more comprehensive github analytics over extended timespans (more similar to google analytics)?

One major limitation I see with accumulating the short 14 day numbers is that it's impossible to determine the number of unique visitors over an extended time period. For example if a repo has 1000 unique visitors over 2 weeks, this does not mean that there will be 2000 unique visitors over 4 weeks, or 4000 unique visitors over 8 weeks, etc, as many visitors over the following time period will be repeat visitors. Does my question make sense?

[antonkomarev]

@glenn-jocher in https://yhype.me we summarize daily counters of unique visitors and if there was 3 unique visitors in January and 2 in March we tend to think that it was 5 different users.

GitHub says that it is unique visitors, but there is no 100% proof that if someone will be counted as unique visitor in January will not be treated as unique visitor in March one more time.

[jgehrcke]

I did have a rather close look at all existing solutions to 'work around this', for custom aggregation. I've built https://github.com/marketplace/actions/github-repo-stats / https://github.com/jgehrcke/github-repo-stats and you might want to have a look! Every day, it generates a nice HTML and PDF report and from my point of view the most important part is that the data stays in GitHub. You don't need a cloud service to integrate this with. No S3 or so.

The time series data is simply stored in a "data repository" (where you run this action in), with transparent evolution of history.

The HTML report can be exposed via GitHub Pages. Example (demo): https://jgehrcke.github.io/ghrs-test/jgehrcke/covid-19-germany-gae/latest-report/report.html

Would appreciate feedback! Thanks!

And: storing this actually in GitHub is a funny way to reply to @UbhiTS:

this is seriously frustrating, in this age of cheap storage, GitHub would not even have a totals counter if not the entire history graph. ... For the love of god GitHub, please listen !

[ElliotB256]

To add something to this discussion, GitHub does store the access data for repositories for longer than the 14 day period. I recently filed a DMCA takedown on an infringing repository, and they were able to provide page views and clones going back a few months for the repo on request. I'm not sure why the data is not made easily available.