My Own Ruby Fibers Babystep

My adventures rereading the Pickaxe Book have reached the chapter on Fibers. Interesting stuff. Thought I would extrapolate from their initial example. My old static pages started from a MANIFEST file that looked something like this:

index.html
/babblings/index.html
/babblings/2013-05-30-javascript.html
/babblings/2013-04-05-perl-and-opensuse.html
/babblings/2013-03-big-updates.html
/babblings/seattle.html
/babblings/stalkingswfans.html
/babblings/bra.html
/brian/index.html

And so on. The path components create a topic heirarchy. There are only 89 files. This is not a lot to track, but it is enough that I can still be hazy about some high level details. For example, I have no idea how many pages are in each section.

sections = Fiber.new do
  File.foreach "MANIFEST" do |line|
    line.match %r{^/(?<path>\w+)/} do |section|
      Fiber.yield section[:path]
    end
  end

  nil
end

counts = Hash.new 0

while section = sections.resume
  counts[section] += 1
end

counts.keys.sort.each { |section| puts "#{section}: #{counts[section]}" }

Yes, this is just the example from the Pickaxe book with line.scan changed to line.match with a slightly altered regular expression.

$ ruby nom-manifest.rb
babblings: 7
brian: 2
geekery: 78

This isn’t that helpful though. I already knew that the majority of my pages were in /geekery/. Let’s adjust the regular expression so that the first two pieces of the entry count as a section.

sections = Fiber.new do
  File.foreach "MANIFEST" do |line|
    line.match %r{
      ^/(?<path>\w+ # main section: /geekery
      (?:/\w+)?)    # subsection:   /ruby
      /             # stop at path separator
    }x do |section|
        Fiber.yield section[:path]
    end
  end
  nil
end

Now I’m looking for possible subsections and lumping them with the top level section. Does this change get me more useful information (for varying definitions of useful)?

$ ruby nom-manifest.rb
babblings: 7
brian: 2
geekery: 2
geekery/editors: 3
geekery/js: 1
geekery/lisp: 1
geekery/osx: 1
geekery/parrot: 17
geekery/perl: 13
geekery/php: 2
geekery/python: 9
geekery/rakudo: 3
geekery/rebol: 10
geekery/ruby: 10
geekery/tools: 2
geekery/unix: 2
geekery/xml: 2

Yeah. It does. I can now see that the most of my static pages are about Parrot or Perl.

I recognize that all I’m doing in this example is shuffling complexity around. There’s nothing in the task that screams “OMG YOU NEED FIBERS TO DO THIS!” Still - I need to figure this stuff out somehow.

Anyways, back to work.