Since Ruby is not Bash, you have to parse the file:
File.foreach("myenv") do |line|
line.chomp.scan(/^export (.+?)=(.*)$/) do |name, val|
ENV[name] = val
end
end
p ENV['FOO'] # => "bar"
That's pretty fragile -- it would severely limit what you can put in the file.
There are a few ways of doing this, depending on your goal. One would be to
run a bash (or better yet, sh) script as a wrapper around your Ruby script --
that is, if your command is installed as /usr/local/bin/foo, have it be a
shell script which does this:
#!/bin/sh
~/.myenv
exec /usr/lib/foo/foo.rb $@
Another hack I've used, when I wasn't sure whether or not I'd need those
variables, is to have bash execute the file, and then tell me. If you only need
one variable, this is easy:
ENV['FOO'] = `source ~/.myenv && echo -n $FOO`
If you wanted all of the variables, here's one (clumsy) way to do it without
having to (really) parse anything:
require 'yaml'
s = `source ~/.myenv && ruby -e 'require "yaml"; print ENV.to_hash.to_yaml`
Yaml.load(s).each_pair{|k,v| ENV[k]=v}
I realize I could've made that less ugly, but I'm not sure there's a good way
to make it pretty. It's essentially the same as extracting a single variable,
except that it invokes another Ruby interpreter to dump the entire environment
as a yaml stream. I did this because it's a lot easier to just throw the yaml
parser at the problem than try to parse the output of something like 'env' --
I couldn't even figure out which environment variables existed, otherwise.
About the only way I could "improve" it is by streaming that Yaml
generation/parsing, but if anyone has an environment big enough for it to
matter, they've already got problems.
One final way would be to add a 'source ~/.myfile' to any other programs you try
to run -- since you mentioned EDITOR, I'm assuming that's why.
