wraith 3.0.4 → 3.1.0

data/lib/wraith/spider.rb

@@ -1,9 +1,12 @@
 require "wraith"
+require "wraith/helpers/logger"
 require "anemone"
 require "nokogiri"
 require "uri"
 
 class Wraith::Spidering
+  include Logging
+
   def initialize(config)
     @wraith = Wraith::Wraith.new(config)
   end
@@ -11,10 +14,10 @@ class Wraith::Spidering
   def check_for_paths
     if @wraith.paths.nil?
       unless @wraith.sitemap.nil?
-        puts "no paths defined in config, loading paths from sitemap"
+        logger.info "no paths defined in config, loading paths from sitemap"
         spider = Wraith::Sitemap.new(@wraith)
       else
-        puts "no paths defined in config, crawling from site root"
+        logger.info "no paths defined in config, crawling from site root"
         spider = Wraith::Crawler.new(@wraith)
       end
       spider.determine_paths
@@ -48,6 +51,8 @@ class Wraith::Spider
 end
 
 class Wraith::Crawler < Wraith::Spider
+  include Logging
+
   EXT = %w(flv swf png jpg gif asx zip rar tar 7z \
            gz jar js css dtd xsd ico raw mp3 mp4 \
            wav wmv ape aac ac3 wma aiff mpg mpeg \
@@ -56,10 +61,10 @@ class Wraith::Crawler < Wraith::Spider
 
   def spider
     if File.exist?(@wraith.spider_file) && modified_since(@wraith.spider_file, @wraith.spider_days[0])
-      puts "using existing spider file"
+      logger.info "using existing spider file"
       @paths = eval(File.read(@wraith.spider_file))
     else
-      puts "creating new spider file"
+      logger.info "creating new spider file"
       spider_list = []
       Anemone.crawl(@wraith.base_domain) do |anemone|
         anemone.skip_links_like(/\.(#{EXT.join('|')})$/)
@@ -76,9 +81,11 @@ class Wraith::Crawler < Wraith::Spider
 end
 
 class Wraith::Sitemap < Wraith::Spider
+  include Logging
+
   def spider
     unless @wraith.sitemap.nil?
-      puts "reading sitemap.xml from #{@wraith.sitemap}"
+      logger.info "reading sitemap.xml from #{@wraith.sitemap}"
       if @wraith.sitemap =~ URI.regexp
         sitemap = Nokogiri::XML(open(@wraith.sitemap))
       else

data/lib/wraith/thumbnails.rb

@@ -11,8 +11,6 @@ class Wraith::Thumbnails
   end
 
   def generate_thumbnails
-    puts "Generating thumbnails"
-
     files = Dir.glob("#{wraith.directory}/*/*.png")
 
     Parallel.each(files, :in_processes => Parallel.processor_count) do |filename|

data/lib/wraith/validate.rb

@@ -0,0 +1,98 @@
+require "wraith/wraith"
+require "wraith/helpers/logger"
+require "wraith/helpers/utilities"
+
+class Wraith::Validate
+  include Logging
+
+  def initialize(config, yaml_passed = false)
+    @wraith = Wraith::Wraith.new(config, yaml_passed)
+  end
+
+  def validate(mode = false)
+    list_debug_information if @wraith.verbose
+    validate_basic_properties
+    validate_mode_properties(mode) if mode
+    # if we get this far, we've only had warnings at worst, not errors.
+    "Config validated. No serious issues found."
+  end
+
+  def validate_basic_properties
+    if @wraith.engine.nil?
+      raise MissingRequiredPropertyError, "You must specify a browser engine! #{docs_prompt}"
+    end
+    unless @wraith.domains
+      raise MissingRequiredPropertyError, "You must specify at least one domain for Wraith to do anything! #{docs_prompt}"
+    end
+    #@TODO validate fuzz is not nil, etc
+  end
+
+  def validate_mode_properties(mode)
+    case mode
+    when "capture"
+      validate_capture_mode
+    when "history"
+      validate_history_mode
+    when "latest"
+      validate_history_mode
+      validate_base_shots_exist
+    else
+      logger.warn "Wraith doesn't know how to validate mode '#{mode}'. Continuing..."
+    end
+  end
+
+  def validate_capture_mode
+    if @wraith.domains.length != 2
+      raise InvalidDomainsError, "`wraith capture` requires exactly two domains. #{docs_prompt}"
+    end
+    if @wraith.history_dir
+      logger.warn "You have specified a `history_dir` in your config, but this is used in `history` mode, NOT `capture` mode. #{docs_prompt}"
+    end
+  end
+
+  def validate_history_mode
+    unless @wraith.history_dir
+      raise MissingRequiredPropertyError, "You must specify a `history_dir` to run Wraith in history mode. #{docs_prompt}"
+    end
+    if @wraith.domains.length != 1
+      raise InvalidDomainsError, "History mode requires exactly one domain. #{docs_prompt}"
+    end
+  end
+
+  def validate_base_shots_exist
+    unless File.directory?(@wraith.history_dir)
+      logger.error "You need to run `wraith history` at least once before you can run `wraith latest`!"
+    end
+  end
+
+  def docs_prompt
+    "See the docs at http://bbc-news.github.io/wraith/"
+  end
+
+  def list_debug_information
+    wraith_version      = Wraith::VERSION
+    ruby_version        = run_command_safely('ruby -v')             || 'Ruby not installed'
+    phantomjs_version   = run_command_safely('phantomjs --version') || 'PhantomJS not installed'
+    casperjs_version    = run_command_safely('casperjs --version')  || 'CasperJS not installed'
+    imagemagick_version = run_command_safely('convert -version')    || 'ImageMagick not installed'
+
+    logger.debug "#################################################"
+    logger.debug "  Wraith version:     #{wraith_version}"
+    logger.debug "  Ruby version:       #{ruby_version}"
+    logger.debug "  ImageMagick:        #{imagemagick_version}"
+    logger.debug "  PhantomJS version:  #{phantomjs_version}"
+    logger.debug "  CasperJS version:   #{casperjs_version}"
+    # @TODO - add a SlimerJS equivalent
+    logger.debug "#################################################"
+    logger.debug ""
+  end
+
+  def run_command_safely(command)
+    begin
+      output = `#{command}`
+    rescue Exception => e
+      return false
+    end
+    output.lines.first
+  end
+end

data/lib/wraith/version.rb

@@ -1,3 +1,3 @@
 module Wraith
-  VERSION = "3.0.4"
+  VERSION = "3.1.0"
 end

data/lib/wraith/wraith.rb

@@ -1,14 +1,18 @@
 require "yaml"
-require "wraith/utilities"
+require "wraith/helpers/logger"
+require "wraith/helpers/utilities"
 
 class Wraith::Wraith
+  include Logging
   attr_accessor :config
 
   def initialize(config, yaml_passed = false)
-    @config = yaml_passed ? config : open_config_file(config)
-  rescue
-    puts "unable to find config at #{config}"
-    exit 1
+    begin
+      @config = yaml_passed ? config : open_config_file(config)
+      logger.level = verbose ? Logger::DEBUG : Logger::INFO
+    rescue
+      logger.error "unable to find config at #{config}"
+    end
   end
 
   def open_config_file(config_name)
@@ -26,15 +30,13 @@ class Wraith::Wraith
   end
 
   def history_dir
-    @config["history_dir"]
+    @config["history_dir"] || false
   end
 
   def engine
     engine = @config["browser"]
     # Legacy support for those using the old style "browser: \n phantomjs: 'casperjs'" configs
-    if engine.is_a? Hash
-      engine = engine.values.first
-    end
+    engine = engine.values.first if engine.is_a? Hash
     engine
   end
 
@@ -51,12 +53,12 @@ class Wraith::Wraith
       path_to_js_templates + "/casper.js"
     # @TODO - add a SlimerJS option
     else
-      abort "Wraith does not recognise the browser engine '#{engine}'"
+      logger.error "Wraith does not recognise the browser engine '#{engine}'"
     end
   end
 
   def before_capture
-    @config["before_capture"] ? convert_to_absolute(@config["before_capture"]) : "false"
+    @config["before_capture"] ? convert_to_absolute(@config["before_capture"]) : false
   end
 
   def widths
@@ -158,4 +160,9 @@ class Wraith::Wraith
   def phantomjs_options
     @config["phantomjs_options"]
   end
-end
+
+  def verbose
+    # @TODO - also add a `--verbose` CLI flag which overrides whatever you have set in the config
+    @config['verbose'] || false
+  end
+end

data/spec/_helpers.rb

@@ -2,8 +2,10 @@ require "rspec"
 require "./lib/wraith/cli"
 
 def create_diff_image
-  saving.capture_page_image(wraith.engine, test_url1, 320, test_image1, selector, 'false', 'false')
-  saving.capture_page_image(wraith.engine, test_url2, 320, test_image2, selector, 'false', 'false')
+  capture_image = saving.construct_command(320, test_url1, test_image1, selector, false, false)
+  `#{capture_image}`
+  capture_image = saving.construct_command(320, test_url2, test_image2, selector, false, false)
+  `#{capture_image}`
 end
 
 def crop_images
@@ -14,14 +16,6 @@ def compare_images
   Wraith::CompareImages.new(config_name).compare_task(test_image1, test_image2, diff_image, data_txt)
 end
 
-def run_js_then_capture(config)
-  generated_image = 'shots/test/temporary_jsified_image.png'
-  saving.capture_page_image(config[:engine], test_url1, 320, generated_image, selector, config[:global_js], config[:path_js])
-  Wraith::CompareImages.new(config_name).compare_task(generated_image, config[:output_should_look_like], diff_image, data_txt)
-  diff = File.open('shots/test/test.txt', "rb").read
-  expect(diff).to eq '0.0'
-end
-
 def get_path_relative_to(current_file, file_to_find)
   File.expand_path(File.join(File.dirname(current_file), file_to_find))
 end

data/spec/before_capture_spec.rb

@@ -1,16 +1,27 @@
 require "_helpers"
 
+def run_js_then_capture(config)
+  saving     = Wraith::SaveImages.new(config_name)
+  generated_image = 'shots/test/temporary_jsified_image.png'
+  capture_image   = saving.construct_command(320, "http://www.bbc.com/afrique", generated_image, selector, config[:global_js], config[:path_js])
+  `#{capture_image}`
+  Wraith::CompareImages.new(config_name).compare_task(generated_image, config[:output_should_look_like], "shots/test/test_diff.png", "shots/test/test.txt")
+  diff = File.open('shots/test/test.txt', "rb").read
+  expect(diff).to eq '0.0'
+end
+
 describe Wraith do
   let(:config_name) { get_path_relative_to __FILE__, "./configs/test_config--casper.yaml" }
-  let(:test_url1) { "http://www.bbc.com/afrique" }
-  let(:diff_image) { "shots/test/test_diff.png" }
-  let(:data_txt) { "shots/test/test.txt" }
-  let(:saving) { Wraith::SaveImages.new(config_name) }
   let(:wraith) { Wraith::Wraith.new(config_name) }
   let(:selector) { "body" }
   let(:before_suite_js) { "spec/js/global.js" }
   let(:before_capture_js) { "spec/js/path.js" }
 
+  before(:each) do
+    Wraith::FolderManager.new(config_name).clear_shots_folder
+    Dir.mkdir("shots/test")
+  end
+
   describe "different ways of determining the before_capture file" do
     it "should allow users to specify the relative path to the before_capture file" do
       config = YAML.load '
@@ -19,7 +30,7 @@ describe Wraith do
       '
       wraith = Wraith::Wraith.new(config, true)
       # not sure about having code IN the test, but we want to get this right.
-      expect(wraith.before_capture).to eq (`pwd`.chomp! + '/javascript/do_something.js')
+      expect(wraith.before_capture).to eq (Dir.pwd + '/javascript/do_something.js')
     end
 
     it "should allow users to specify the absolute path to the before_capture file" do
@@ -35,11 +46,10 @@ describe Wraith do
   # @TODO - we need tests determining the path to "path-level before_capture hooks"
 
   describe "When hooking into beforeCapture (CasperJS)" do
-
     it "Executes the global JS before capturing" do
       run_js_then_capture(
         global_js: before_suite_js,
-        path_js:   'false',
+        path_js:   false,
         output_should_look_like: 'spec/base/global.png',
         engine:    'casperjs'
       )
@@ -47,7 +57,7 @@ describe Wraith do
 
     it "Executes the path-level JS before capturing" do
       run_js_then_capture(
-        global_js: 'false',
+        global_js: false,
         path_js: before_capture_js,
         output_should_look_like: 'spec/base/path.png',
         engine:    'casperjs'
@@ -100,5 +110,4 @@ describe Wraith do
   #     )
   #   end
   # end
-
-end
+end

data/spec/config_spec.rb

@@ -1,7 +1,6 @@
 require "_helpers"
 
 describe "wraith config" do
-
   let(:config_name) { get_path_relative_to __FILE__, "./configs/test_config--phantom.yaml" }
   let(:wraith) { Wraith::Wraith.new(config_name) }
 
@@ -20,7 +19,6 @@ describe "wraith config" do
   end
 
   describe "When creating a wraith worker" do
-
     it "should have a browser engine defined" do
       expect(wraith.engine).to be_a String
     end
@@ -71,7 +69,6 @@ describe "wraith config" do
   end
 
   describe "different ways of initialising browser engine" do
-
     it "should let us directly specify the engine" do
       config = YAML.load 'browser: phantomjs'
       wraith = Wraith::Wraith.new(config, true)
@@ -90,7 +87,6 @@ describe "wraith config" do
   end
 
   describe "different ways of determining the snap file" do
-
     it "should calculate the snap file from the engine" do
       config = YAML.load 'browser: phantomjs'
       wraith = Wraith::Wraith.new(config, true)
@@ -117,7 +113,7 @@ describe "wraith config" do
       '
       wraith = Wraith::Wraith.new(config, true)
       # not sure about having code IN the test, but we want to get this right.
-      expect(wraith.snap_file).to eq (`pwd`.chomp! + '/path/to/snap.js')
+      expect(wraith.snap_file).to eq (Dir.pwd + '/path/to/snap.js')
     end
 
     it "should allow users to specify the absolute path to their own snap file" do
@@ -131,7 +127,6 @@ describe "wraith config" do
   end
 
   describe "different modes of efficiency (resize or reload)" do
-
     it "should trigger efficient mode if resize was specified" do
       config = YAML.load 'resize_or_reload: "resize"'
       wraith = Wraith::Wraith.new(config, true)
@@ -143,7 +138,5 @@ describe "wraith config" do
       wraith = Wraith::Wraith.new(config, true)
       expect(wraith.resize).to eq false
     end
-
   end
-
-end
+end

data/spec/configs/test_config--casper.yaml

@@ -4,11 +4,7 @@
 ##########
 
 #Headless browser option
-browser:
-  phantomjs: "casperjs"
-  # slimerjs: "slimerjs"
-
-#overriding the snap file
+browser: "casperjs"
 snap_file: "spec/js/custom_snap_file.js"
 
 # Type the name of the directory that shots will be stored in

data/spec/construct_command_spec.rb

@@ -0,0 +1,43 @@
+require "_helpers"
+
+describe "Wraith config to CLI argument mapping" do
+  describe "passing variables to construct_command" do
+    # set default variables we can override if necessary
+    let(:config_name) { get_path_relative_to __FILE__, "./configs/test_config--phantom.yaml" }
+    let(:saving)      { Wraith::SaveImages.new(config_name) }
+    let(:width)       { 320 }
+    let(:url)         { 'http://example.com/my-page' }
+    let(:file_name)   { 'wraith/my-page/320_phantomjs_latest.png' }
+    let(:selector)    { '.my_selector' }
+    let(:global_bc)   { 'javascript/before_capture.js' }
+    let(:path_bc)     { false }
+
+    it "should take a load of variables and construct a command" do
+      expected = "phantomjs  '#{Dir.pwd}/lib/wraith/javascript/phantom.js' 'http://example.com/my-page' '320' 'wraith/my-page/320_phantomjs_latest.png' '.my_selector' '#{Dir.pwd}/javascript/before_capture.js' 'false'"
+      actual   = saving.construct_command(width, url, file_name, selector, global_bc, path_bc)
+      expect(actual).to eq expected
+    end
+
+    it "should allow hashtags in selectors" do
+      selector = '#some-id'
+      expected = "phantomjs  '#{Dir.pwd}/lib/wraith/javascript/phantom.js' 'http://example.com/my-page' '320' 'wraith/my-page/320_phantomjs_latest.png' '\\#some-id' '#{Dir.pwd}/javascript/before_capture.js' 'false'"
+      actual   = saving.construct_command(width, url, file_name, selector, global_bc, path_bc)
+      expect(actual).to eq expected
+    end
+
+    it "should be able to pass multiple widths at once" do
+      width    = [320, 624, 976]
+      expected = "phantomjs  '#{Dir.pwd}/lib/wraith/javascript/phantom.js' 'http://example.com/my-page' '320,624,976' 'wraith/my-page/320_phantomjs_latest.png' '.my_selector' '#{Dir.pwd}/javascript/before_capture.js' 'false'"
+      actual   = saving.construct_command(width, url, file_name, selector, global_bc, path_bc)
+      expect(actual).to eq expected
+    end
+
+    it "should call casperjs when the config says so" do
+      config_name = get_path_relative_to(__FILE__, "./configs/test_config--casper.yaml")
+      saving      = Wraith::SaveImages.new(config_name)
+      expected    = "casperjs  '#{Dir.pwd}/spec/js/custom_snap_file.js' 'http://example.com/my-page' '320' 'wraith/my-page/320_phantomjs_latest.png' '.my_selector' '#{Dir.pwd}/javascript/before_capture.js' 'false'"
+      actual      = saving.construct_command(width, url, file_name, selector, global_bc, path_bc)
+      expect(actual).to eq expected
+    end
+  end
+end