GitHub metadata mirror

A Hugo-based GitHub issues and pull-request mirror using backup data from github-metadata-backup.

Usage

This tool generates static HTML pages from a github-metadata-backup. There are two steps to this process.

1. Generate mirror content: First, the Python script generate-data.py is used to generate MarkDown files into the content directory and JSON files into the data directory. Pass the folder with the backup as first argument and the output directory as second argument.

python generate-data.py /path/to/backup /path/to/project/directory

2. Build static site: The second step is to use Hugo to build the static HTML site. Make sure to address the FIXME's in the config.toml file or to set the environment variables HUGO_TITLE, HUGO_PARAMS_OWNER, HUGO_PARAMS_REPOSITORY, and HUGO_PARAMS_FOOTER before building the site. With larger projects (a few thousand issues and pull requests), Hugo's RAM usage will be noticable. Consider enabling swapping if your system has limited RAM. It might also make sense to pass the --debug flag to observe the build process. Set a base URL with --baseURL if you want to host the site in a sub path. By default, the command below will generate the static HTML site into the public directory.

hugo --source /path/to/project/directory --debug --baseURL /mirror

• HUGO_TITLE: Sets the title of the page. This is shown in the top left corner.
• HUGO_PARAMS_OWNER: Sets GitHub repository owner. This is needed for rewriting link and for linking to the repository on GitHub.
• HUGO_PARAMS_REPOSITORY: Sets GitHub repository name. This is needed for rewriting link and for linking to the repository on GitHub.
• HUGO_PARAMS_FOOTER: Sets a footer that is shown on each page. This can be HTML syntax.

Example

To generate a mirror from the Bitcoin Core bitcoin/bitcoin metadata backup (assuming it's located in /var/backup/bitcoin-bitcoin/), run generate-data.py from the github-metadata-mirror project directory.

python generate-data.py /var/backup/bitcoin-bitcoin/ .

When testing with large repositoires, you can pass -s to pick a subset of 100 issues and pull-requests. This speeds up the processing and the generation.

Then, address the FIXME's in the config.toml file or set the above enviroment variables. In this case, we set the environment variables. As we plan to host our site in the subdirectory bitcoin-bitcoin, we set the base URL. We set the owner and repository to "bitcoin".

export HUGO_TITLE="Bitcoin Core GitHub mirror"
export HUGO_PARAMS_OWNER="bitcoin"
export HUGO_PARAMS_REPOSITORY="bitcoin"
export HUGO_PARAMS_FOOTER="<b>This mirror is hosted by 0xB10C</b>"
hugo --source . --debug --baseURL "/bitcoin-bitcoin/

Hugo will generate the static files to the public/ directory. This might take a few minutes. From there on you can use nginx, caddy, or any other webserver to host them.

Nix Package and module

A Nix package and module for the github-metadata-mirror tool are avaliable in my personal nix-package-collection. Note that the package does not include a binary that can be run. Rather, it includes the generate-data.py Python script and the Hugo files (layout, tests, ...).