/github-analysis-scripts

Node scripts to assess a GitHub repo

Primary LanguageJavaScriptMIT LicenseMIT

You don't need to install any dependencies, it's just Node and npm.

Configuration options need to be set in config.mjs before running any of the scripts.

Scrapers

There are various scripts available, but issues-scraper.mjs must be run before the others:

npm run scrape-issues

Depending on what you're trying to do, that might be the only script you need.

If you want to know which files are changed by the PRs, you should run pr-files-scraper.mjs:

npm run scrape-pr-files

This will only download the files for open PRs, but it's a one-line change if you also want to include closed PRs.

To download the comments for issues (and PRs), run:

npm run scrape-comments

GitHub's rate limits prevent downloading the comments for all issues in a large repo, like vuejs/core. The script will only download comments for issues with a lot of comments or a lot of reactions on the original post. You can comment out that restriction to download them all, but you'll need to wait an hour once you hit the limit.

In addition, it will only download comments for 250 issues per run. Run the script repeatedly until you have what you need.

If you only want to download comments for open issues, there is a section in the code that can be uncommented to do that.

Analyzers

There are various scripts to analyze the downloaded data.

To generate monthly counts for the number of open, closed and merged PRs, run:

npm run analyze-pr-merges

This requires scrape-issues to be run first.

The script also dumps out the names of any bots it encounters, plus a summary of the users with the most PRs. These are useful for assessing whether some users should be excluded from the counts. Accounts marked as bots are automatically excluded.

To find PRs with small changes run:

npm run analyze-small-changes

This requires scrape-issues and scrape-pr-files to be run first.

To find issues/PRs with the most comments or the most reactions, run:

npm run analyze-comments

This requires scrape-issues and scrape-comments to be run first.