Refactor: Separate classes for configuration and node script invocation logic

lib/Dhalang.rb

@@ -5,7 +5,8 @@ module Dhalang
   require_relative 'Dhalang/url_utils'
   require_relative 'Dhalang/file_utils'
   require_relative 'Dhalang/error'
-  require_relative 'Dhalang/puppeteer'
+  require_relative 'Dhalang/configuration'
+  require_relative 'Dhalang/node_script_invoker'
   require 'uri'
   require 'tempfile'
   require 'shellwords'

lib/Dhalang/configuration.rb

@@ -0,0 +1,87 @@
+module Dhalang
+  # Groups Puppeteer and Dhalang configuration.
+  class Configuration
+    NODE_MODULES_PATH = Dir.pwd + '/node_modules/'.freeze
+    USER_OPTIONS = {
+      navigationTimeout: 10000,
+      printToPDFTimeout: 0, # unlimited
+      navigationWaitUntil: 'load',
+      navigationWaitForSelector: '',
+      navigationWaitForXPath: '',
+      userAgent: '',
+      isHeadless: true,
+      viewPort: '',
+      httpAuthenticationCredentials: '',
+      isAutoHeight: false,
+      chromeOptions: []
+    }.freeze
+    DEFAULT_PDF_OPTIONS = {
+      scale: 1,
+      displayHeaderFooter: false,
+      headerTemplate: '',
+      footerTemplate: '',
+      headerTemplateFile: '',
+      footerTemplateFile: '',
+      printBackground: true,
+      landscape: false,
+      pageRanges: '',
+      format: 'A4',
+      width: '',
+      height: '',
+      margin: { top: 36, right: 36, bottom: 20, left: 36 },
+      preferCSSPageSize: true,
+      omitBackground: false
+    }.freeze
+    DEFAULT_SCREENSHOT_OPTIONS = {
+      fullPage: true,
+      clip: nil,
+      omitBackground: false
+    }.freeze
+    DEFAULT_JPEG_OPTIONS = {
+      quality: 100
+    }.freeze
+
+    private_constant :NODE_MODULES_PATH
+    private_constant :USER_OPTIONS
+    private_constant :DEFAULT_PDF_OPTIONS
+    private_constant :DEFAULT_SCREENSHOT_OPTIONS
+    private_constant :DEFAULT_JPEG_OPTIONS
+
+    private attr_accessor :page_url
+    private attr_accessor :temp_file_path
+    private attr_accessor :temp_file_extension
+    private attr_accessor :user_options
+    private attr_accessor :pdf_options
+    private attr_accessor :screenshot_options
+    private attr_accessor :jpeg_options
+
+    # @param [Hash]   custom_options        Changes that override default.
+    # @param [String] page_url              Url for Puppeteer to visit.
+    # @param [String] temp_file_path        Absolute path of temp file to use for writing script results.
+    #                                       Can be nil for scripts using stdout.
+    # @param [String] temp_file_extension   Extension of temp file. Can be nil for scripts using stdout.
+    def initialize(custom_options, page_url, temp_file_path = nil, temp_file_extension = nil)
+      self.page_url = page_url
+      self.temp_file_path = temp_file_path
+      self.temp_file_extension = temp_file_extension
+      self.user_options = USER_OPTIONS.map { |key, default_value| [key, custom_options.fetch(key, default_value)] }
+      self.pdf_options = DEFAULT_PDF_OPTIONS.map { |key, default_value| [key, custom_options.fetch(key, default_value)] }
+      self.screenshot_options = DEFAULT_SCREENSHOT_OPTIONS.map { |key, default_value| [key, custom_options.fetch(key, default_value)] }
+      self.jpeg_options = DEFAULT_JPEG_OPTIONS.map { |key, default_value| [key, custom_options.fetch(key, default_value)] }
+    end
+
+    # Returns configuration as JSON string.
+    def json
+      return {
+        webPageUrl: page_url,
+        tempFilePath: temp_file_path,
+        puppeteerPath: NODE_MODULES_PATH,
+        imageType: temp_file_extension,
+        userOptions: user_options.to_h,
+        pdfOptions: pdf_options.to_h,
+        screenshotOptions: screenshot_options.to_h,
+        jpegOptions: jpeg_options.to_h
+      }.to_json
+    end
+  end
+end

lib/Dhalang/node_script_invoker.rb

@@ -0,0 +1,28 @@
+module Dhalang
+  class NodeScriptInvoker
+
+    # Executes JS script under given script_path by launching a new Node process.
+    #
+    # @param [String] script_path           Absolute path of JS script to execute.
+    # @param [Configuration] configuration  Configuration to use.
+    def self.execute_script(script_path, configuration)
+      command = create_node_command(script_path, configuration)
+      Open3.popen2e(command) do |_stdin, stdouterr, wait|
+        return nil if wait.value.success?
+
+        output = stdouterr.read.strip
+        output = nil if output == ''
+        message = output || "Exited with status #{wait.value.exitstatus}"
+        raise DhalangError, message
+      end
+    end
+
+    # Returns a [String] with node command that invokes the provided script with the configuration.
+    #
+    # @param [String] script_path           Absolute path of JS script to invoke.
+    # @param [Configuration] configuration  Configuration to use.
+    private_class_method def self.create_node_command(script_path, configuration)
+      "node #{script_path} #{Shellwords.escape(configuration.json)}"
+    end
+  end
+end

lib/PDF.rb

@@ -1,8 +1,8 @@
 module Dhalang
   # Allows consumers of this library to create PDFs with Puppeteer. 
   class PDF
-    PUPPETEER_SCRIPT_PATH = File.expand_path('../js/pdf-generator.js', __FILE__).freeze
-    private_constant :PUPPETEER_SCRIPT_PATH
+    SCRIPT_PATH = File.expand_path('../js/pdf-generator.js', __FILE__).freeze
+    private_constant :SCRIPT_PATH
 
     # Captures the full webpage under the given url as PDF.
     #
@@ -43,7 +43,8 @@ def self.get_from_html(html, options = {})
     private_class_method def self.get(url, options)
       temp_file = FileUtils.create_temp_file("pdf")
       begin
-        Puppeteer.visit(url, PUPPETEER_SCRIPT_PATH, temp_file.path, "pdf", options)
+        configuration = Configuration.new(options, url, temp_file.path, "pdf")
+        NodeScriptInvoker.execute_script(SCRIPT_PATH, configuration)
         binary_pdf_content = FileUtils.read_binary(temp_file.path)
       ensure
         FileUtils.delete(temp_file)

lib/Screenshot.rb

@@ -1,9 +1,9 @@
 module Dhalang
   # Allows consumers of this library to take screenshots with Puppeteer. 
   class Screenshot
-    PUPPETEER_SCRIPT_PATH = File.expand_path('../js/screenshot-generator.js', __FILE__).freeze
+    SCRIPT_PATH = File.expand_path('../js/screenshot-generator.js', __FILE__).freeze
     IMAGE_TYPES = [:jpeg, :png, :webp].freeze
-    private_constant :PUPPETEER_SCRIPT_PATH
+    private_constant :SCRIPT_PATH
     private_constant :IMAGE_TYPES
 
     # <b>DEPRECATED:</b> Please use `get_from_url(url, :jpeg)` instead.
@@ -44,7 +44,8 @@ def self.get_from_url(url, image_type, options = {})
 
       temp_file = FileUtils.create_temp_file(image_type)
       begin
-        Puppeteer.visit(url, PUPPETEER_SCRIPT_PATH, temp_file.path, image_type, options)
+        configuration = Configuration.new(options, url, temp_file.path, image_type)
+        NodeScriptInvoker.execute_script(SCRIPT_PATH, configuration)
         binary_image_content = FileUtils.read_binary(temp_file.path)
       ensure
         FileUtils.delete(temp_file)