"Oneliner" to backup your public github repos.

backup-github.sh

#!/bin/sh

# Run from a directory in which you'd like to shove a _lot_ of bare repositories.
curl -s http://github.com/api/v2/yaml/repos/show/mikewest | grep ':url:' | awk '{ print $2 }' | sed -e 's#http://#git://#' -e 's#$#.git#' | xargs -L 1 git clone --bare

You can pass multiple -e commands to sed. Err, nvm.

I don't believe "git clone" will overwrite an existing local directory, so this won't work unless you run it in an empty directory or remove the local (*.git) directories from a previous run first.

Also, the grep and awk commands are unnecessary... it can be done with a single sed command:

cd [backup directory]
rm -rf _.git
curl -s [URL] | sed -n '/^ *:url:/ s/ *:url: *http:(._)/git:\1.git/p' | xargs -L 1 git clone --bare

In fact... it can be done without xargs as well.... just pipe to a shell, eg:

cd [backup directory]
rm -rf *.git
curl -s [URL] | sed -n '/^ *:url:/ s/ *:url: *http:\(.*\)/git clone --bare git:\1.git/p' | sh

aolcm: You're right. You can do everything with a single sed command. :)

I find it pretty hard to follow, though... using smaller steps makes the whole process a bit clearer to me; I know exactly what each command is doing and can follow the interactions. I think your solution, though surely functional, is simply less readable.