Ingest gitile
Summary:
A lister is needed for gitile (not a typo for gitiles #5045 (closed)), a simple web frontend for git repositories.
There is no unique string that could be used to find gitile sites in search engines.
gitile sites can be listed by downloading and parsing the front page. First there is a header
with id intro
, containing a h1
with content Every project
. The project listing is in a div
with id content
, inside that is a div
with id project-list
. Each project is in a div
with class project
, then inside a div
with class instructions
the clone URL is in a code
tag, prefixed with the git clone
command.
A naive shell lister looks like this:
curl -sL $site | pup 'code text{}' | sed -n 's/git clone //p'
So far I have found two gitile instances.
https://git.lepiller.eu/
https://git.elephly.net/
Unfortunately gitile has a memory problem that can kill instances so they are sometimes unstable. This has affected both the instances. I pinged both the folks running them, the second instance got fixed but the primary instance did not yet. I have done SCN for both of these instances too.
Plan:
-
Implement lister -
Implement loader -
docker: Run lister -
docker: Run loader -
Document lister -
Document loader -
Deploy on staging -
Call for public review -
Deploy on production