crabfarm 0.0.16 → 0.0.17

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 01f53897c32d27e6253c74660a0842b9cb7c3c97
-  data.tar.gz: 7993ac4152d6128ce8b69950458e48dfe042089a
+  metadata.gz: 62fc780d5c9db277ef5147d564c74cd684dbd39b
+  data.tar.gz: 9028aeb0ce71914ab549644948c0ceb83e1efdda
 SHA512:
-  metadata.gz: af73913311cb5404d95724470ce07bb37e3ed686abb02a39e20b6a85b0ad028a31ef4524c8ffd09381ce20157e107da7cbe20ddea22c08f3908997d81c1e37b1
-  data.tar.gz: 3345f29499e15089b3fd6feccf984f9121187500e9859a2016ccfaf14947d363ca335b0212025722b9eac273996171a6f133e2c37f9d0fa5f2bd649952333742
+  metadata.gz: f003f40fbb5727dee2f831fede26ff6bfd559028c5d339f56aa63aabe3cc0fe7f03519d73df03ebee5121a8478dc0a082f8124272b194dc2a5023402c5147cbd
+  data.tar.gz: a19ccb92b10a0b1d316992cbde562987ac1a3928b4ae32811e56e570f93073f839b4422ae2a0e06bdcbebff3557bf5e0f1ea197bdbff442052c7b878273ca297

data/lib/crabfarm/adapters/browser/capybara.rb

@@ -1,7 +1,7 @@
 module Crabfarm
   class CapybaraBrowserDsl
     def self.wrap(_bucket)
-      raise "Capybara adapter is incompleted"
+      raise NotImplementedError.new "Capybara adapter is incompleted"
     end
   end
 end

data/lib/crabfarm/adapters/browser/watir.rb

@@ -1,21 +1,18 @@
-require 'watir-webdriver'
-
 class Watir::Browser
-  def parse(_parser_class, _options={})
-    Crabfarm::ParserService.parse _parser_class, html, _options
+  def to_html
+    html
   end
 end
 
 class Watir::Element
-  def parse(_parser_class, _options={})
-    Crabfarm::ParserService.parse _parser_class, html, _options
+  def to_html
+    html
   end
 end
 
 class Watir::ElementCollection
-  def parse(_parser_class, _options={})
-    full_html = self.map(&:html).join
-    Crabfarm::ParserService.parse _parser_class, full_html, _options
+  def to_html
+    self.map(&:html).join
   end
 end
 

data/lib/crabfarm/adapters/output/jbuilder.rb

@@ -1,5 +1,3 @@
-require 'jbuilder'
-
 module Crabfarm
   class JbuilderOutputBuilder
     def self.prepare

data/lib/crabfarm/adapters/parser/nokogiri.rb

@@ -1,9 +1,13 @@
 require 'nokogiri'
 
 module Crabfarm
-  class NokogiriDsl
-    def self.parse(_html)
-      Nokogiri::HTML _html
+  class NokogiriAdapter
+    def self.parse(_element)
+      if _element.respond_to? :to_html
+        Nokogiri::HTML _element.to_html
+      else
+        Nokogiri::HTML _element
+      end
     end
   end
 end

data/lib/crabfarm/adapters/parser/pdf_reader.rb

@@ -0,0 +1,9 @@
+require "pdf-reader"
+
+module Crabfarm
+  class PdfReaderAdapter
+    def self.parse(_pdf_data)
+      PDF::Reader.new StringIO.new _pdf_data
+    end
+  end
+end

data/lib/crabfarm/base_parser.rb

@@ -1,18 +1,18 @@
 module Crabfarm
   class BaseParser < Delegator
 
-    attr_reader :params, :root
+    attr_reader :params, :document
 
-    def self.parser_dsl(_dsl)
-      @parser_dsl = _dsl
+    def self.engine(_engine)
+      @engine = _engine
     end
 
-    def initialize(_html, _params)
-      dsl_class = Strategies.load(:parser_dsl, class_parser_dsl || Crabfarm.config.parser_dsl)
-      @root = dsl_class.parse _html
+    def initialize(_target, _params)
+      engine_class = Strategies.load(:parser_engine, class_engine || Crabfarm.config.parser_engine)
+      @document = engine_class.parse _target
       @params = _params
 
-      super @root
+      super @document
     end
 
     def parse
@@ -20,17 +20,17 @@ module Crabfarm
     end
 
     def __getobj__
-      @root
+      @document
     end
 
     def __setobj__(obj)
-      @root = obj
+      @document = obj
     end
 
   private
 
-    def class_parser_dsl
-      self.class.instance_variable_get :@parser_dsl
+    def class_engine
+      self.class.instance_variable_get :@engine
     end
   end
 end

data/lib/crabfarm/base_state.rb

@@ -5,10 +5,13 @@ module Crabfarm
   class BaseState
     extend Forwardable
 
+    PARSE_METHOD_RX = /^parse_(.*)$/
+
     attr_reader :params, :output
 
-    def_delegators :@pool, :driver
-    def_delegators :@store, :get, :fetch
+    def_delegators '@context', :http
+    def_delegators '@context.pool', :driver
+    def_delegators '@context.store', :get, :fetch
 
     def self.browser_dsl(_dsl)
       @class_browser_dsl = _dsl
@@ -18,11 +21,9 @@ module Crabfarm
       @class_output_builder = _builder
     end
 
-    def initialize(_pool, _store, _params)
-      @pool = _pool
-      @store = _store
+    def initialize(_context, _params)
+      @context = _context
       @params = _params
-      @events = []
 
       @dsl = Strategies.load(:browser_dsl, class_browser_dsl || Crabfarm.config.browser_dsl)
       @builder = Strategies.load(:output_builder, class_output_builder || Crabfarm.config.output_builder)
@@ -32,6 +33,10 @@ module Crabfarm
       @dsl.wrap driver(_name)
     end
 
+    def download(_url)
+      @context.http.get(_url).body
+    end
+
     def output
       @output ||= @builder.prepare
     end
@@ -44,16 +49,16 @@ module Crabfarm
       raise NotImplementedError.new
     end
 
-    def event(_type, _msg)
-      @events << { created_at: Time.current, type: _type, msg: _msg }
-    end
+    def parse(_target=nil, _options={})
+      parser_class = _options.delete :using
 
-    def alert(_msg)
-      event(:alert, _msg)
-    end
+      if parser_class.nil?
+        parser_class = (self.class.name + 'Parser').constantize
+      end
 
-    def info(_msg)
-      event(:info, _msg)
+      parser = parser_class.new _target, @params.merge(_options)
+      parser.parse
+      return parser
     end
 
     def fork_each(_enumerator, &_block)
@@ -66,6 +71,20 @@ module Crabfarm
       ThreadsWait.all_waits(*ths)
     end
 
+    def method_missing(_method, *_args, &_block)
+      m = PARSE_METHOD_RX.match(_method)
+      if m
+        options = _args[1] || {}
+        options[:using] = (m[1].camelize + 'Parser').constantize
+        parse _args[0], options
+      else super end
+    end
+
+    def respond_to?(_method, _include_all=false)
+      return true if PARSE_METHOD_RX === _method
+      super
+    end
+
   private
 
     def class_browser_dsl

data/lib/crabfarm/configuration.rb

@@ -6,7 +6,7 @@ module Crabfarm
 
     OPTIONS = [
       [:browser_dsl, :string, 'Default browser dsl used by states'],
-      [:parser_dsl, :string, 'Default parser dsl used by parsers'],
+      [:parser_engine, :string, 'Default parser engine used by parsers'],
       [:output_builder, :string, 'Default json output builder used by states'],
       [:driver_factory, :mixed, 'Driver factory, disabled if phantom_mode is used'],
       [:log_path, :string, 'Path where logs should be stored'],
@@ -51,7 +51,7 @@ module Crabfarm
     def reset
       @values = {
         browser_dsl: :surfer,
-        parser_dsl: :nokogiri,
+        parser_engine: :nokogiri,
         output_builder: :hash,
         driver_factory: nil,
         log_path: nil,

data/lib/crabfarm/context.rb

@@ -4,7 +4,7 @@ module Crabfarm
   class Context
     extend Forwardable
 
-    def_delegators :@pool, :driver
+    attr_accessor :pool, :store, :http
 
     def initialize
       @store = StateStore.new
@@ -14,16 +14,10 @@ module Crabfarm
     def load
       init_phantom_if_required
       init_driver_pool
+      init_http_client
       @loaded = true
     end
 
-    def run_state(_name, _params={})
-      load
-      state = LoaderService.load_state(_name).new @pool, @store, _params
-      state.crawl
-      state
-    end
-
     def reset
       @store.reset
       @pool.reset unless @pool.nil?
@@ -63,6 +57,14 @@ module Crabfarm
       @phantom = nil
     end
 
+    def init_http_client
+      @http = build_http_client if @http.nil?
+    end
+
+    def release_http_client
+      @http = nil
+    end
+
     def build_driver_factory
       if @phantom
         PhantomDriverFactory.new @phantom, driver_config
@@ -72,6 +74,10 @@ module Crabfarm
       end
     end
 
+    def build_http_client
+      HttpClient.new config.proxy
+    end
+
     def config
       Crabfarm.config
     end

data/lib/crabfarm/crabtrap_context.rb

@@ -28,6 +28,10 @@ module Crabfarm
 
   private
 
+    def build_http_client
+      HttpClient.new proxy_address
+    end
+
     def restart_with_options(_options)
       stop_daemon
       @runner = CrabtrapRunner.new Crabfarm.config.crabtrap_config.merge(_options)

data/lib/crabfarm/crabtrap_runner.rb

@@ -1,5 +1,3 @@
-require 'net/http'
-
 module Crabfarm
   class CrabtrapRunner
 

data/lib/crabfarm/dsl/surfer/search_context.rb

@@ -91,10 +91,6 @@ module Crabfarm
           end
         end
 
-        def parse(_parser_class, _params={})
-          ParserService.parse _parser_class, to_html, _params
-        end
-
         def to_html
           elements.map { |e| e['outerHTML'] }.join
         end

data/lib/crabfarm/engines/safe_state_loop.rb

@@ -109,7 +109,7 @@ module Crabfarm
                 begin
                   ActiveSupport::Dependencies.clear
                   logger.info "StateLoop: loading state: #{@next_state_name}"
-                  @doc = context.run_state(@next_state_name, @next_state_params).output_as_json
+                  @doc = TransitionService.apply_state(context, @next_state_name, @next_state_params).output_as_json
                   logger.info "StateLoop: state loaded successfully: #{@next_state_name}"
                   @error = nil
                 rescue Exception => e

data/lib/crabfarm/forked_state.rb

@@ -1,13 +1,12 @@
 module Crabfarm
-  class ForkedState
-    extend Forwardable
-
-    def_delegators :@state, :params, :get, :fetch, :event, :alert, :info
+  class ForkedState < Delegator
 
     def initialize(_state, _name, _mutex)
       @state = _state
       @name = _name
       @mutex = _mutex
+
+      super @state
     end
 
     def driver
@@ -18,10 +17,22 @@ module Crabfarm
       @browser ||= @state.browser(@name)
     end
 
+    def output
+      raise ScriptError.new 'Use lock_output to access output in forked states'
+    end
+
     def lock_output
       @mutex.synchronize {
         yield @state.output
       }
     end
+
+    def __getobj__
+      @state
+    end
+
+    def __setobj__(obj)
+      @state = obj
+    end
   end
 end

data/lib/crabfarm/http_client.rb

@@ -0,0 +1,97 @@
+require "uri"
+
+module Crabfarm
+  class HttpClient
+
+    class HttpRequestError < StandardError
+      extend Forwardable
+
+      def_delegators :@response, :code, :body
+
+      attr_reader :response
+
+      def initialize(_response)
+        @response = _response
+        super _response.message
+      end
+    end
+
+    class MaximumRedirectsError < StandardError
+      def initialize
+        super 'Redirection loop detected!'
+      end
+    end
+
+    attr_reader :proxy_addr, :proxy_port
+
+    def initialize(_proxy=nil)
+      if _proxy.nil?
+        @proxy_addr = nil
+        @proxy_port = nil
+      else
+        @proxy_addr, @proxy_port = _proxy.split ':'
+      end
+    end
+
+    def get(_url, _query={}, _headers={})
+      uri = URI _url
+      perform_request Net::HTTP::Get, uri, _headers
+    end
+
+    def post(_url, _data, _headers={})
+      perform_request Net::HTTP::Post, URI(_url), _headers do |req|
+        req.body = prepare_data(_data)
+      end
+    end
+
+    def put(_url, _data, _headers={})
+      perform_request Net::HTTP::Put, URI(_url), _headers do |req|
+        req.body = prepare_data(_data)
+      end
+    end
+
+    def delete(_url)
+      perform_request Net::HTTP::Delete, URI(_url), _headers
+    end
+
+  private
+
+    def perform_request(_req_type, _uri, _headers, _limit=10)
+
+      raise MaximumRedirectsError.new if _limit == 0
+
+      request = _req_type.new(_uri.path.empty? ? '/' : _uri.path)
+      _headers.keys.each { |k| request[k] = _headers[k] }
+      yield request if block_given?
+
+      response = build_client(_uri).request request
+
+      case response
+      when Net::HTTPSuccess then
+        response
+      when Net::HTTPRedirection then
+        location = response['location']
+        perform_request(_req_type, URI.parse(location), _headers, _limit - 1)
+      else
+        handle_error_response response
+      end
+    end
+
+    def build_client(uri)
+      client = Net::HTTP.new uri.host, uri.port || 80, proxy_addr, proxy_port
+      client.use_ssl = true if uri.scheme == 'https'
+      client.verify_mode = OpenSSL::SSL::VERIFY_NONE
+      client
+    end
+
+    def handle_error_response(_response)
+      raise HttpRequestError.new _response
+    end
+
+    def prepare_data(_data)
+      if _data.is_a? Hash
+        _data.keys.map { |k| "#{k}=#{_data[k]}" }.join '&'
+      else _data end
+    end
+  end
+end

data/lib/crabfarm/modes/console.rb

@@ -34,7 +34,7 @@ module Crabfarm
           begin
             elapsed = Benchmark.measure do
               puts "Transitioning to #{_name.to_s.camelize} state"
-              doc = @context.run_state(_name, _params).output_as_json
+              doc = TransitionService.apply_state(@context, _name, _params).output_as_json
 
               puts "State changed, generated document:"
               puts JSON.pretty_generate(doc).color(:green).gsub(/(^|\\n)/, '  ')

data/lib/crabfarm/modes/generator.rb

@@ -35,18 +35,20 @@ module Crabfarm
       end
 
       def generate_state(_name)
+        class_name = _name.camelize
         with_crawler_path do
-          binding = { state_class: _name.camelize }
-          path('app', 'states', _name.underscore + '.rb').render('state.rb', binding)
-          path('spec', 'states', _name.underscore + '_spec.rb').render('state_spec.rb', binding)
+          binding = { state_class: class_name.camelize }
+          path('app', 'states', class_name.underscore + '.rb').render('state.rb', binding)
+          path('spec', 'states', class_name.underscore + '_spec.rb').render('state_spec.rb', binding)
         end
       end
 
       def generate_parser(_name)
+        class_name = _name.camelize + 'Parser'
         with_crawler_path do
-          binding = { parser_class: _name.camelize }
-          path('app', 'parsers', _name.underscore + '.rb').render('parser.rb', binding)
-          path('spec', 'parsers', _name.underscore + '_spec.rb').render('parser_spec.rb', binding)
+          binding = { parser_class: class_name }
+          path('app', 'parsers', class_name.underscore + '.rb').render('parser.rb', binding)
+          path('spec', 'parsers', class_name.underscore + '_spec.rb').render('parser_spec.rb', binding)
         end
       end
 

data/lib/crabfarm/phantom_runner.rb

@@ -1,4 +1,3 @@
-require 'net/http'
 require 'timeout'
 
 module Crabfarm

data/lib/crabfarm/rspec.rb

@@ -3,7 +3,6 @@ require 'net/http'
 
 CF_TEST_CONTEXT = Crabfarm::CrabtrapContext::new
 CF_TEST_CONTEXT.load
-CF_TEST_BUCKET = CF_TEST_CONTEXT.driver
 
 module Crabfarm
   module RSpec
@@ -16,7 +15,9 @@ module Crabfarm
         Net::HTTP.get(URI.parse _snap_or_url)
       end
 
-      ParserService.parse described_class, html, _options
+      parser = described_class.new html, _options
+      parser.parse
+      parser
     end
 
     def crawl(_state=nil, _params={})
@@ -27,9 +28,9 @@ module Crabfarm
 
       if _state.nil?
         return nil unless described_class < BaseState # TODO: maybe raise an error here.
-        @state = @last_state = CF_TEST_CONTEXT.run_state(described_class, _params)
+        @state = @last_state = TransitionService.apply_state CF_TEST_CONTEXT, described_class, _params
       else
-        @last_state = CF_TEST_CONTEXT.run_state(_state, _params)
+        @last_state = TransitionService.apply_state CF_TEST_CONTEXT, _state, _params
       end
     end
 

data/lib/crabfarm/strategies.rb

@@ -2,25 +2,42 @@ module Crabfarm
   module Strategies
 
     class Loader
-      def initialize(_klass, _pkg=nil)
+      def initialize(_name, _klass, _pkg, _deps)
+        @name = _name
         @klass = _klass
         @pkg = _pkg
+        @deps = _deps
       end
 
       def load
+        load_dependencies
         require @pkg if @pkg
         if @klass.is_a? String then Object.const_get @klass else @klass end
       end
+
+    private
+
+      def load_dependencies
+        @deps.each do |dep|
+          begin
+            require dep
+            # TODO: check dependency version!
+          rescue LoadError
+            raise ConfigurationError.new "Missing #{@name} dependency, please add `gem \"#{dep}\"` to the crawler's Gemfile"
+          end
+        end
+      end
     end
 
     @@register = {}
 
-    def self.register(_cat, _name, _klass, _pkg=nil)
-      @@register[_cat.to_s + _name.to_s] = Loader.new(_klass, _pkg)
+    def self.register(_cat, _name, _klass, _pkg=nil, _deps=[])
+      full_name = _cat.to_s + ':' + _name.to_s
+      @@register[full_name] = Loader.new(full_name, _klass, _pkg, _deps)
     end
 
     def self.load(_cat, _name)
-      full_name = _cat.to_s + _name.to_s
+      full_name = _cat.to_s + ':' + _name.to_s
       raise ConfigurationError.new "Invalid #{_cat} name #{_name}" unless @@register.has_key? full_name
       @@register[full_name].load
     end

data/lib/crabfarm/templates/Crabfile.erb

@@ -1,10 +1,14 @@
 
 # The default crawling dsl to use in states and parsers, can be overriden in each component using the `browser_dsl :dsl` modifier
-# Available options are :surfer, :watir and :capybara
+# Available options are :surfer, :watir and :capybara. Both watir and capybara require an additional gem to be added to Gemfile
 set_browser_dsl :surfer
 
+# The default parser engine for parsers that do not specify one.
+# Available options are :nokogiri and :pdf_parser. :pdf_parser requires an additional gem to be added to Gemfile
+# set_parser_engine :nokogiri
+
 # Change the defaut output builder used in a state to generate the output document.
-# Available options are :hash, :ostruct, :jbuilder
+# Available options are :hash, :ostruct, :jbuilder. :jbuilder requires an additional gem to be added to Gemfile
 # set_output_builder :hash
 
 # The path where every crawler log is stored.

data/lib/crabfarm/transition_service.rb

@@ -0,0 +1,23 @@
+module Crabfarm
+  class TransitionService
+
+    def self.apply_state(_context, _name, _params={})
+      state_class = if _name.is_a? String or _name.is_a? Symbol
+        load_by_name _name
+      else _name end
+
+      _context.load
+      state = state_class.new _context, _params
+      state.crawl
+      state
+    end
+
+  private
+
+    def self.load_by_name(_name)
+      name = _name.to_s.gsub(/[^A-Z0-9:]+/i, '_').camelize
+      name.constantize
+    end
+
+  end
+end

data/lib/crabfarm/version.rb

@@ -1,3 +1,3 @@
 module Crabfarm
-  VERSION = "0.0.16"
+  VERSION = "0.0.17"
 end

data/lib/crabfarm.rb

@@ -1,15 +1,16 @@
-require 'logger'
+require "logger"
 require "forwardable"
+require "net/http"
 require "active_support/inflector"
 require "selenium-webdriver"
 
 require "crabfarm/version"
 require "crabfarm/errors"
 require "crabfarm/configuration"
-require "crabfarm/loader_service"
-require "crabfarm/parser_service"
+require "crabfarm/transition_service"
 require "crabfarm/driver_bucket"
 require "crabfarm/driver_bucket_pool"
+require "crabfarm/http_client"
 require "crabfarm/default_driver_factory"
 require "crabfarm/phantom_driver_factory"
 require "crabfarm/phantom_runner"
@@ -43,15 +44,16 @@ module Crabfarm
   module Strategies
     # bundled browser dsl adapters
     register :browser_dsl, :surfer, 'Crabfarm::SurferBrowserDsl', 'crabfarm/adapters/browser/surfer'
-    register :browser_dsl, :watir, 'Crabfarm::WatirBrowserDsl', 'crabfarm/adapters/browser/watir'
-    register :browser_dsl, :capybara, 'Crabfarm::CapybaraBrowserDsl', 'crabfarm/adapters/browser/capybara'
+    register :browser_dsl, :watir, 'Crabfarm::WatirBrowserDsl', 'crabfarm/adapters/browser/watir', ['watir-webdriver']
+    register :browser_dsl, :capybara, 'Crabfarm::CapybaraBrowserDsl', 'crabfarm/adapters/browser/capybara', ['capybara']
 
     # bundled parsers dsl adapters
-    register :parser_dsl, :nokogiri, 'Crabfarm::NokogiriDsl', 'crabfarm/adapters/parser/nokogiri'
+    register :parser_engine, :nokogiri, 'Crabfarm::NokogiriAdapter', 'crabfarm/adapters/parser/nokogiri'
+    register :parser_engine, :pdf_reader, 'Crabfarm::PdfReaderAdapter', 'crabfarm/adapters/parser/pdf_reader', ['pdf-reader']
 
     # bundled state output builders
     register :output_builder, :hash, 'Crabfarm::HashOutputBuilder', 'crabfarm/adapters/output/hash'
     register :output_builder, :ostruct, 'Crabfarm::OStructOutputBuilder', 'crabfarm/adapters/output/ostruct'
-    register :output_builder, :jbuilder, 'Crabfarm::JbuilderOutputBuilder', 'crabfarm/adapters/output/jbuilder'
+    register :output_builder, :jbuilder, 'Crabfarm::JbuilderOutputBuilder', 'crabfarm/adapters/output/jbuilder', ['jbuilder']
   end
 end

metadata

@@ -1,29 +1,15 @@
 --- !ruby/object:Gem::Specification
 name: crabfarm
 version: !ruby/object:Gem::Version
-  version: 0.0.16
+  version: 0.0.17
 platform: ruby
 authors:
 - Ignacio Baixas
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-03-11 00:00:00.000000000 Z
+date: 2015-03-12 00:00:00.000000000 Z
 dependencies:
-- !ruby/object:Gem::Dependency
-  name: jbuilder
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - ~>
-      - !ruby/object:Gem::Version
-        version: 2.2.0
-  type: :runtime
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - ~>
-      - !ruby/object:Gem::Version
-        version: 2.2.0
 - !ruby/object:Gem::Dependency
   name: selenium-webdriver
   requirement: !ruby/object:Gem::Requirement
@@ -38,34 +24,6 @@ dependencies:
     - - ~>
       - !ruby/object:Gem::Version
         version: '2.33'
-- !ruby/object:Gem::Dependency
-  name: capybara
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - '>='
-      - !ruby/object:Gem::Version
-        version: '0'
-  type: :runtime
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - '>='
-      - !ruby/object:Gem::Version
-        version: '0'
-- !ruby/object:Gem::Dependency
-  name: watir-webdriver
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - '>='
-      - !ruby/object:Gem::Version
-        version: '0'
-  type: :runtime
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - '>='
-      - !ruby/object:Gem::Version
-        version: '0'
 - !ruby/object:Gem::Dependency
   name: nokogiri
   requirement: !ruby/object:Gem::Requirement
@@ -170,20 +128,6 @@ dependencies:
     - - '>='
       - !ruby/object:Gem::Version
         version: '0'
-- !ruby/object:Gem::Dependency
-  name: multipart-post
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - '>='
-      - !ruby/object:Gem::Version
-        version: '0'
-  type: :runtime
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - '>='
-      - !ruby/object:Gem::Version
-        version: '0'
 - !ruby/object:Gem::Dependency
   name: bundler
   requirement: !ruby/object:Gem::Requirement
@@ -324,6 +268,62 @@ dependencies:
     - - '>='
       - !ruby/object:Gem::Version
         version: '0'
+- !ruby/object:Gem::Dependency
+  name: pdf-reader
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: 1.3.3
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: 1.3.3
+- !ruby/object:Gem::Dependency
+  name: capybara
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: watir-webdriver
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: jbuilder
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: 2.2.0
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: 2.2.0
 description: 
 email:
 - ignacio@platan.us
@@ -339,6 +339,7 @@ files:
 - lib/crabfarm/adapters/output/jbuilder.rb
 - lib/crabfarm/adapters/output/ostruct.rb
 - lib/crabfarm/adapters/parser/nokogiri.rb
+- lib/crabfarm/adapters/parser/pdf_reader.rb
 - lib/crabfarm/base_parser.rb
 - lib/crabfarm/base_state.rb
 - lib/crabfarm/cli.rb
@@ -355,14 +356,13 @@ files:
 - lib/crabfarm/engines/safe_state_loop.rb
 - lib/crabfarm/errors.rb
 - lib/crabfarm/forked_state.rb
-- lib/crabfarm/loader_service.rb
+- lib/crabfarm/http_client.rb
 - lib/crabfarm/mocks/noop_driver.rb
 - lib/crabfarm/modes/console.rb
 - lib/crabfarm/modes/generator.rb
 - lib/crabfarm/modes/publisher.rb
 - lib/crabfarm/modes/recorder.rb
 - lib/crabfarm/modes/server.rb
-- lib/crabfarm/parser_service.rb
 - lib/crabfarm/phantom_driver_factory.rb
 - lib/crabfarm/phantom_runner.rb
 - lib/crabfarm/rspec.rb
@@ -383,6 +383,7 @@ files:
 - lib/crabfarm/templates/spec_helper.rb.erb
 - lib/crabfarm/templates/state.rb.erb
 - lib/crabfarm/templates/state_spec.rb.erb
+- lib/crabfarm/transition_service.rb
 - lib/crabfarm/version.rb
 - lib/crabfarm.rb
 - bin/crabfarm

data/lib/crabfarm/loader_service.rb

@@ -1,29 +0,0 @@
-module Crabfarm
-  class LoaderService
-
-    def self.load_state(_name)
-      load_entity _name, 'state', BaseState
-    end
-
-    def self.load_parser(_name)
-      load_entity _name, 'parser', BaseParser
-    end
-
-  private
-
-    def self.load_entity(_name, _role, _type)
-
-      if _name.is_a? String or _name.is_a? Symbol
-        name = _name.to_s.gsub(/[^A-Z0-9:]+/i, '_').camelize
-        mod = name.constantize rescue nil
-      else
-        mod = _name
-      end
-
-      raise EntityNotFoundError.new _role, name if mod.nil?
-      raise EntityNotFoundError.new _role, name unless mod < _type
-      mod
-    end
-
-  end
-end

data/lib/crabfarm/parser_service.rb

@@ -1,12 +0,0 @@
-module Crabfarm
-  class ParserService
-
-    def self.parse(_parser_class, _html, _options={})
-      _parser_class = LoaderService.load_parser(_parser_class) if _parser_class.is_a? String or _parser_class.is_a? Symbol
-      parser = _parser_class.new _html, _options
-      parser.parse
-      parser
-    end
-
-  end
-end