add-sivers-dot-org-books-to-couch-db.rb

#!/usr/bin/env ruby
# Add books from http://sivers.org/book to a couch db
# todo - parse author out of 'title' field

require 'rubygems'
require 'couchrest'
require 'nokogiri'
require 'open-uri'
require 'clip'

options = Clip do |p|
  p.optional 'd', 'database', :desc => 'URI of database', :default => 'http://localhost:5984/books'
  p.optional 's', 'source', :desc => 'Data source', :default => 'http://sivers.org/book'
  p.flag 'v', 'verbose', :desc => 'Make it chatty'
end

def get_content(noko_source, selector)
  noko_source.css(selector).first.inner_html
end

class Book
  def initialize(noko_source, source)
    @title = get_content(noko_source, "h3 a")
    @source = source
    @isbn = retrieve_isbn get_content(noko_source, "p.small")
    @derek_sivers_rating = retrieve_rating get_content(noko_source, "p.small")
    @derek_sivers_summary = get_content(noko_source, ":nth-child(4)")
  end

  # sample p.small string: ISBN: 1416541993 READ: 2009-04-03 RATING: 10/10
  def retrieve_isbn(source)
    source =~ /ISBN:\ (\w+)/
    raise source.inspect unless $1
    $1
  end

  def retrieve_rating(source)
    source =~ /RATING:\ (\w+)/
    raise source.inspect unless $1
    $1
  end

  def to_hash
    { "title" => @title,
      "source" => @source,
      "isbn" => @isbn,
      "derek_sivers_rating" => @derek_sivers_rating,
      "derek_sivers_summary" => @derek_sivers_summary }
  end
end

doc = Nokogiri::HTML(open(options.source))
db = CouchRest.database!(options.database)

books = doc.css('#content div.book').collect do |book|
  Book.new(book, options.source).to_hash
end

db.bulk_save books

STDERR.puts "#{books.length} saved."