Merge branch 'NielsSteensma-master'

.gitignore

@@ -8,6 +8,7 @@
 /spec/examples.txt
 /test/tmp/
 /test/version_tmp/
+/node_modules/
 /tmp/
 .idea/*
 

Gemfile.lock

@@ -37,6 +37,7 @@ GEM
     ttfunk (1.6.2.1)
 
 PLATFORMS
+  ruby
   x64-mingw32
 
 DEPENDENCIES
@@ -48,4 +49,4 @@ DEPENDENCIES
   rspec (~> 3.0)
 
 BUNDLED WITH
-   1.16.1
+   1.17.2

README.md

@@ -49,15 +49,15 @@ end
 To return the PNG from a Rails controller you can do the following:  
 ```
 def example_controller_method  
-    binary_png = Dhalang::Screenshot.get_from_url_as_png("https://www.google.nl")
+    binary_png = Dhalang::Screenshot.get_from_url_as_png("https://www.google.com")
     send_data(binary_png, filename: 'screenshotofgoogle.png', type: 'image/png')   
 end
 ```
 
 To return the JPEG from a Rails controller you can do the following:  
 ```
 def example_controller_method  
-    binary_jpeg = Dhalang::Screenshot.get_from_url_as_jpeg("https://www.google.nl")
+    binary_jpeg = Dhalang::Screenshot.get_from_url_as_jpeg("https://www.google.com")
     send_data(binary_jpeg, filename: 'screenshotofgoogle.jpeg', type: 'image/jpeg')   
 end
 ```

lib/Dhalang.rb

@@ -1,4 +1,11 @@
 module Dhalang
-  require 'PDF'
-  require 'Screenshot'
+  require_relative 'PDF'
+  require_relative 'Screenshot'
+  require_relative 'Dhalang/version'
+  require_relative 'Dhalang/url_utils'
+  require_relative 'Dhalang/file_utils'
+  require_relative 'Dhalang/puppeteer'
+  require 'uri'
+  require 'tempfile'
+  require 'shellwords'
 end

lib/Dhalang/file_utils.rb

@@ -0,0 +1,37 @@
+module Dhalang
+    # Contains common logic for files. 
+    class FileUtils
+
+        # Reads the file under the given filepath as a binary.
+        #
+        # @param [String] file_path The absolute path of the file to read.
+        #
+        # @return [String]      The binary content under the file_path.
+        def self.read_binary(file_path)
+            IO.binread(file_path)
+        end
+
+        # Creates a new temp file.
+        #
+        # @param [String] extension The extension of the file.
+        # @param [String] content   The content of the file. (Optional)
+        #
+        # @return [Tempfile]    The created temp file.
+        def self.create_temp_file(extension, content = nil)
+            temp_file = Tempfile.new(["dhalang",".#{extension}"])
+            unless(content == nil)
+                temp_file.write(content)
+                temp_file.rewind
+            end
+            temp_file
+        end
+
+        # Deletes the given file.
+        #
+        # @param [File] file    The file to delete.
+        def self.delete(file)
+            file.close unless file.closed?
+            file.unlink
+        end
+    end
+end

lib/Dhalang/puppeteer.rb

@@ -0,0 +1,17 @@
+module Dhalang
+    # Contains common logic for interacting with Puppeteer.
+    class Puppeteer
+        NODE_MODULES_PATH = Dir.pwd + '/node_modules/'.freeze
+        private_constant :NODE_MODULES_PATH
+
+        # Launches a new Node process, executing the (Puppeteer) script under the given script_path.
+        #
+        # @param [String] page_url              The url to pass to the goTo method of Puppeteer.
+        # @param [String] script_path           The absolute path of the JS script to execute.
+        # @param [String] temp_file_path        The absolute path of the temp file to use to write any actions tom from Puppeteer.
+        # @param [String] temp_file_extension   The extension of the temp file.
+        def self.visit(page_url, script_path, temp_file_path, temp_file_extension)
+            system("node #{script_path} #{Shellwords.escape(NODE_MODULES_PATH)} #{page_url} #{Shellwords.escape(temp_file_path)} #{Shellwords.escape(temp_file_extension)}")
+        end
+    end
+end

lib/Dhalang/url_utils.rb

@@ -0,0 +1,14 @@
+module Dhalang
+    # Contains common logic for URL's. 
+    class UrlUtils
+
+        # Raises an error if the given URL cannot be used for navigation with Puppeteer.
+        #
+        # @param [String] url The url to validate
+        def self.validate(url)
+            if (url !~ URI::DEFAULT_PARSER.regexp[:ABS_URI])
+                raise URI::InvalidURIError, 'The given url was invalid, use format http://www.example.com'
+            end
+        end
+    end
+end

lib/Dhalang/version.rb

@@ -1,3 +1,3 @@
 module Dhalang
-  VERSION = "0.2.0"
+  VERSION = "0.3.0"
 end

lib/PDF.rb

@@ -1,65 +1,51 @@
-require "Dhalang/version"
-require 'uri'
-require 'tempfile'
-
 module Dhalang
+  # Allows consumers of this library to create PDFs with Puppeteer. 
   class PDF
-    PDF_GENERATOR_JS_PATH = File.expand_path('../js/pdfgenerator.js', __FILE__)
-    PROJECT_PATH = Dir.pwd + '/node_modules/'
-
+    PUPPETEER_SCRIPT_PATH = File.expand_path('../js/pdf-generator.js', __FILE__).freeze
+    private_constant :PUPPETEER_SCRIPT_PATH
+
+    # Captures the full webpage under the given url as PDF.
+    #
+    # @param  [String] url    The url to get as PDF.
+    #
+    # @return [String] The PDF that was created as binary.
     def self.get_from_url(url)
-      validate_url(url)
-      temporary_pdf_save_file = create_temporary_pdf_file
-      begin
-        visit_page_with_puppeteer(url, temporary_pdf_save_file.path)
-        binary_pdf_content = get_file_content_as_binary_string(temporary_pdf_save_file)
-      ensure
-        temporary_pdf_save_file.close unless temporary_pdf_save_file.closed?
-        temporary_pdf_save_file.unlink
-      end
-      return binary_pdf_content
+      UrlUtils.validate(url)
+      get(url)
     end
 
+    # Captures the full HTML as PDF.
+    # Useful when creating dynamic content, for example invoices.
+    #
+    # @param  [String] html   The html to get as PDF.
+    #
+    # @return [String] The PDF that was created as binary.
     def self.get_from_html(html)
-      html_file = create_temporary_html_file(html)
-      temporary_pdf_save_file = create_temporary_pdf_file
+      html_file = FileUtils.create_temp_file("html", html)
+      url = "file://" + html_file.path
       begin
-        visit_page_with_puppeteer("file://" + html_file.path, temporary_pdf_save_file.path)
-        binary_pdf_content = get_file_content_as_binary_string(temporary_pdf_save_file)
+          binary_pdf_content = get(url)
       ensure
-        temporary_pdf_save_file.close unless temporary_pdf_save_file.closed?
-        html_file.close unless html_file.closed?
-        temporary_pdf_save_file.unlink
-        html_file.unlink
+        FileUtils.delete(html_file)
       end
       return binary_pdf_content
     end
 
-    private
-    def self.validate_url(url)
-      if (url !~ URI::DEFAULT_PARSER.regexp[:ABS_URI])
-        raise URI::InvalidURIError, 'The given url was invalid, use format http://www.example.com'
+
+    # Groups and executes the logic for creating a PDF of a webpage.
+    #
+    # @param  [String] url    The url to create a PDF for.
+    #
+    # @return [String] The PDF that was created as binary.
+    private_class_method def self.get(url)
+      temp_file = FileUtils.create_temp_file("pdf")
+      begin
+        Puppeteer.visit(url, PUPPETEER_SCRIPT_PATH, temp_file.path, "pdf")
+        binary_pdf_content = FileUtils.read_binary(temp_file.path)
+      ensure
+        FileUtils.delete(temp_file)
       end
-    end
-
-    def self.create_temporary_pdf_file
-      Tempfile.new("pdf")
-    end
-
-    ## Creates a temp .html file which can be browsed to by puppeteer for creating a pdf
-    def self.create_temporary_html_file(content)
-      html_file = Tempfile.new(['page', '.html'])
-      html_file.write(content)
-      html_file.rewind
-      return html_file
-    end
-
-    def self.visit_page_with_puppeteer(page_to_visit, path_to_save_pdf_to)
-      system("node #{PDF_GENERATOR_JS_PATH} #{page_to_visit} #{Shellwords.escape(path_to_save_pdf_to)} #{Shellwords.escape(PROJECT_PATH)}")
-    end
-
-    def self.get_file_content_as_binary_string(file)
-      IO.binread(file.path)
+      return binary_pdf_content
     end
   end
 end

lib/Screenshot.rb

@@ -1,51 +1,43 @@
-require "Dhalang/version"
-require 'uri'
-require 'tempfile'
-
 module Dhalang
+  # Allows consumers of this library to take screenshots with Puppeteer. 
   class Screenshot
-    SCREENSHOT_GENERATOR_JS_PATH = File.expand_path('../js/screenshotgenerator.js', __FILE__)
-    PROJECT_PATH = Dir.pwd + '/node_modules/'
-
+    PUPPETEER_SCRIPT_PATH = File.expand_path('../js/screenshot-generator.js', __FILE__).freeze
+    private_constant :PUPPETEER_SCRIPT_PATH
+
+    # Captures a full JPEG screenshot of the webpage under the given url.
+    #
+    # @param  [String] url        The url to take a screenshot of.
+    #
+    # @return [String] the screenshot that was taken as binary.
     def self.get_from_url_as_jpeg(url)
-      validate_url(url)
-      get_image(url, :jpeg)
+      get(url, "jpeg")
     end
 
+    # Captures a full PNG screenshot of the webpage under the given url.
+    #
+    # @param  [String] url        The url to take a screenshot of.
+    #
+    # @return [String] The screenshot that was taken as binary.
     def self.get_from_url_as_png(url)
-      validate_url(url)
-      get_image(url, :png)
-    end
-
-    private
-    def self.validate_url(url)
-      if (url !~ URI::DEFAULT_PARSER.regexp[:ABS_URI])
-        raise URI::InvalidURIError, 'The given url was invalid, use format http://www.example.com'
-      end
-    end
-
-    def self.create_temporary_screenshot_file
-      Tempfile.new("png")
+      get(url, "png")
     end
-
-    def self.get_image(url, type)
-      temporary_screenshot_save_file = create_temporary_screenshot_file
+
+    # Groups and executes the logic for taking a screenhot of a webpage.
+    #
+    # @param  [String] url        The url to take a screenshot of.
+    # @param  [String] image_type The image type to use for storing the screenshot.
+    #
+    # @return [String] The screenshot that was taken as binary.
+    private_class_method def self.get(url, image_type)
+      UrlUtils.validate(url)
+      temp_file = FileUtils.create_temp_file(image_type)
       begin
-        visit_page_with_puppeteer(url, temporary_screenshot_save_file.path, type)
-        binary_image_content = get_file_content_as_binary_string(temporary_screenshot_save_file)
+        Puppeteer.visit(url, PUPPETEER_SCRIPT_PATH, temp_file.path, image_type)
+        binary_image_content = FileUtils.read_binary(temp_file.path)
       ensure
-        temporary_screenshot_save_file.close unless temporary_screenshot_save_file.closed?
-        temporary_screenshot_save_file.unlink
+        FileUtils.delete(temp_file)
       end
       return binary_image_content
     end
-
-    def self.visit_page_with_puppeteer(page_to_visit, path_to_save_pdf_to, image_save_type)
-      system("node #{SCREENSHOT_GENERATOR_JS_PATH} #{page_to_visit} #{Shellwords.escape(path_to_save_pdf_to)} #{Shellwords.escape(PROJECT_PATH)} #{Shellwords.escape(image_save_type)}")
-    end
-
-    def self.get_file_content_as_binary_string(file)
-      IO.binread(file.path)
-    end
   end
 end

lib/js/dhalang.js

@@ -0,0 +1,54 @@
+/**
+ * @typedef {Object} Configuration
+ * @property {string} webPageUrl - The url of the webpage to visit.
+ * @property {string} tempFilePath - The path of the tempfile to write the screenshot/pdf to.
+ * @property {string} puppeteerModulePath - The path of the Puppeteer module.
+ * @property {string} imageType - The type of image to save ( undefined for pdfgenerator ).
+ */
+
+/**
+ * @typedef {Object} NavigationParameters
+ * @property {number} timeout - Maximum in milliseconds until navigation times out, we use a default of 10 seconds as timeout.
+ * @property {string} waituntil - Determines when the navigation was finished, we wait here until the Window.load event is fired ( meaning all images, stylesheet, etc was loaded ).
+ */
+
+/**
+ * Generates a configuration object based on the given process arguments from Ruby.
+ * @param {Boolean} isForScreenshotGenerator - Indicates if this configuration is for a screenshot generator.
+ * @returns {Configuration}
+ * The generated configuration object.
+ */
+exports.getConfiguration = function (isForScreenshotGenerator) {
+    return {
+        puppeteerPath: process.argv[2],
+        webPageUrl: process.argv[3],
+        tempFilePath: process.argv[4],
+        imageType: isForScreenshotGenerator ? process.argv[5] : undefined
+    }
+}
+
+/**
+ * Launches Puppeteer and returns its instance.
+ * @param {string} puppeteerModulePath - The path puppeteer is under.
+ * @returns {Promise<Object>}
+ * The launched instance of Puppeteer.
+ */
+exports.launchPuppeteer = async function (puppeteerModulePath) {
+    module.paths.push(puppeteerModulePath);
+    const puppeteer = require('puppeteer');
+    return await puppeteer.launch({
+        args: ['--no-sandbox', '--disable-setuid-sandbox']
+    });
+}
+
+/**
+ * Returns a new object containing the navigation parameters to use when navigating with Puppeteer to web pages.
+ * @returns {NavigationParameters}
+ * The navigation parameters to use.
+ */
+exports.getNavigationParameters = function () {
+    return {
+        timeout: 20000,
+        waitUntil: 'load'
+    }
+}