lib/sunscraper.rb in sunscraper-1.0.0 vs lib/sunscraper.rb in sunscraper-1.1.0.beta1

@@ -1,50 +1,87 @@
-require 'sunscraper/library'
+if !defined?(RUBY_ENGINE) && RUBY_VERSION =~ /^1.8/
+  raise RuntimeError, "Sunscraper does not work on Ruby MRI 1.8.x."
+end
 
 # Sunscraper loads an HTML page in a headless browser and waits for `Sunscraper.finish()`
 # method to be called. It blocks the calling thread, but is threadsafe, does
 # not acquire GIL and thus can be called from multiple threads simultaneously.
 module Sunscraper
   # ScrapeTimeout error is raised when the page could not be loaded fast enough.
   class ScrapeTimeout < StandardError; end
 
   class << self
+    def os_x?
+      RUBY_PLATFORM =~ /darwin/i || RbConfig::CONFIG['target_os'] == 'darwin'
+    end
+
+    attr_reader :worker
+    def worker=(worker_type)
+      if [:embed, :standalone].include?(worker_type)
+        @worker = worker_type
+      else
+        raise RuntimeError, "Invalid Sunscraper worker type: #{worker_type.inspect}"
+      end
+    end
+
     # Scrape an inline HTML. The content is loaded without a particular base URL.
     # If your application depends on base URL being available, use {scrape_url}.
     #
     # @param [Integer] timeout timeout in milliseconds
     def scrape_html(html, timeout=5000)
-      scrape(timeout) do |context|
-        Library.load_html context, html
+      scrape(timeout) do |worker, context|
+        worker.load_html context, html
       end
     end
 
     # Scrape an URL.
     #
     # @param [Integer] timeout timeout in milliseconds
     def scrape_url(url, timeout=5000)
-      scrape(timeout) do |context|
-        Library.load_url context, url
+      scrape(timeout) do |worker, context|
+        worker.load_url context, url
       end
     end
 
     private
 
     def scrape(timeout)
-      context = Library.create
+      worker = load_worker
 
-      yield context
+      context = worker.create
+      yield worker, context
+      worker.wait(context, timeout)
 
-      Library.wait(context, timeout)
+      data = worker.fetch(context)
 
-      data = Library.fetch(context)
-
       if data == "!SUNSCRAPER_TIMEOUT"
         raise ScrapeTimeout, "Sunscraper has timed out waiting for the callback"
       else
         data
       end
     ensure
-      Library.discard(context) if context
+      worker.discard(context) if context
     end
+
+    def load_worker
+      case @worker
+      when :standalone
+        require 'sunscraper/standalone'
+
+        Sunscraper::Standalone
+
+      when :embed
+        require 'sunscraper/library'
+
+        Sunscraper::Library
+      end
+    end
   end
 end
+
+if Sunscraper.os_x?
+  # OS X is braindead
+  Sunscraper.worker = :standalone
+else
+  # ... even Win32 is better.
+  Sunscraper.worker = :embed
+end
\ No newline at end of file