Getting GitHub Metrics

[cansavvy]

Summary

Collecting all the metrics we might want from GitHub. Basically this summarizes to

authorize("github")
# You can collect all the metrics for all the repos for an organization
all_repos_metrics <- get_repos_metrics(owner = "fhdsl")

# Or you can provide a list of repos that you want metrics for
repo_names <- c("fhdsl/metricminer", "jhudsl/OTTR_Template")
some_repos_metrics <- get_repos_metrics(repo_names = repo_names)

There's also some creds handling improvements here.

Caveat

The metrics bit only reports the last two weeks. This is just a thing GitHub does that isn't great. So we will be able to have more than just two weeks of metrics but only after we start collecting and storing.

[cansavvy]

Looks like the metrics for this are limited to a 2 week span https://github.com/orgs/community/discussions/15189 womp womp. So we scraping here I come.

[cansavvy]

Looks like the metrics for this are limited to a 2 week span https://github.com/orgs/community/discussions/15189 womp womp. So we scraping here I come.

Webscraping will still happen. But it will not address this issue. ☹️

[howardbaik]

get_github_metrics() is buggy, as described in my inline comments. I can take a look at what's causing the error.

I created a sub-branch of cansavvy/github called howardbaek/github-fix. I'll be working on that branch, and request a PR that merges my changes into cansavvy/github

[howardbaik]

Now, I'm getting this error message from running metrics <- get_github_metrics(repo = "fhdsl/metricminer")

Error in lapply(repo_metric_list$contributors, function(contributor) { : 
  could not find function "%>%"

I'm getting the same error message when running all_repos_metrics <- get_repos_metrics(owner = "fhdsl") because get_github_metrics() is inside get_repos_metrics()

[cansavvy]

I'll need to DRY up some of this auth section in a future PR

[cansavvy]

I think this PR has gotten unwieldy.

So my general strategy at this point is to:

1. Merge this when we feel "decent enough" about it; meaning @howardbaek is able to get the data he needs to start working on the dashboard: This means, you'll need to run get_repos_metrics() but just for ITCR relevant repositories. Meaning OTTR ones, this one, and probably all course PRs. (But again, maybe focus on building the dashboard page for Google Analytics first if that data collecting is working for you.)
2. Then I will start working on unit testing UNIT TESTING #8 so that we can try to avoid unwieldy PRs like this in the future.
3. From there we can document bugs and file smaller PRs for the rest.

Sound good @howardbaek ?

[cansavvy]

I believe a majority of the bugs are fixed (until we find more when I build unit testing). @howardbaek Can you just test using:

devtools::load_all()
repo_names <- c("fhdsl/metricminer", "jhudsl/OTTR_Template")
some_repos_metrics <- get_repos_metrics(repo_names = repo_names)

And see if this works for you? if so then I think it would be easiest to merge this and go from there

[howardbaik]

some_repos_metrics <- get_repos_metrics(repo_names = repo_names)

That code chunk works on my end, but some_repos_metrics is a data frame with unnecessary rownames that are the same values as the repo_name column. What do you think of outputting a tibble using as_tibble()?

[cansavvy]

some_repos_metrics <- get_repos_metrics(repo_names = repo_names)

That code chunk works on my end, but some_repos_metrics is a data frame with unnecessary rownames that are the same values as the repo_name column. What do you think of outputting a tibble using as_tibble()?

Yeah I'm okay with tibbles or data frames. We can further discuss this data's format for future PRs but lets get this in for now