hssc_pdfkit 0.5.4

data/spec/middleware_spec.rb

@@ -0,0 +1,314 @@
+require 'spec_helper'
+
+def app; Rack::Lint.new(@app); end
+
+def mock_app(options = {}, conditions = {})
+  main_app = lambda { |env|
+    @env = env
+    headers = {'Content-Type' => "text/html"}
+    [200, headers, @body || ['Hello world!']]
+  }
+
+  builder = Rack::Builder.new
+  builder.use PDFKit::Middleware, options, conditions
+  builder.run main_app
+  @app = builder.to_app
+end
+
+describe PDFKit::Middleware do
+
+  describe "#call" do
+    describe "conditions" do
+      describe ":only" do
+
+        describe "regex" do
+          describe "one" do
+            before { mock_app({}, :only => %r[^/public]) }
+
+            context "matching" do
+              specify do
+                get 'http://www.example.org/public/test.pdf'
+                last_response.headers["Content-Type"].should == "application/pdf"
+                last_response.body.bytesize.should == PDFKit.new("Hello world!").to_pdf.bytesize
+              end
+            end
+
+            context "not matching" do
+              specify do
+                get 'http://www.example.org/secret/test.pdf'
+                last_response.headers["Content-Type"].should == "text/html"
+                last_response.body.should == "Hello world!"
+              end
+            end
+          end # one regex
+
+          describe "multiple" do
+            before { mock_app({}, :only => [%r[^/invoice], %r[^/public]]) }
+
+            context "matching" do
+              specify do
+                get 'http://www.example.org/public/test.pdf'
+                last_response.headers["Content-Type"].should == "application/pdf"
+                last_response.body.bytesize.should == PDFKit.new("Hello world!").to_pdf.bytesize
+              end
+            end
+
+            context "not matching" do
+              specify do
+                get 'http://www.example.org/secret/test.pdf'
+                last_response.headers["Content-Type"].should == "text/html"
+                last_response.body.should == "Hello world!"
+              end
+            end
+          end # multiple regex
+        end # regex
+
+        describe "string" do
+          describe "one" do
+            before { mock_app({}, :only => '/public') }
+
+            context "matching" do
+              specify do
+                get 'http://www.example.org/public/test.pdf'
+                last_response.headers["Content-Type"].should == "application/pdf"
+                last_response.body.bytesize.should == PDFKit.new("Hello world!").to_pdf.bytesize
+              end
+            end
+
+            context "not matching" do
+              specify do
+                get 'http://www.example.org/secret/test.pdf'
+                last_response.headers["Content-Type"].should == "text/html"
+                last_response.body.should == "Hello world!"
+              end
+            end
+          end # one string
+
+          describe "multiple" do
+            before { mock_app({}, :only => ['/invoice', '/public']) }
+
+            context "matching" do
+              specify do
+                get 'http://www.example.org/public/test.pdf'
+                last_response.headers["Content-Type"].should == "application/pdf"
+                last_response.body.bytesize.should == PDFKit.new("Hello world!").to_pdf.bytesize
+              end
+            end
+
+            context "not matching" do
+              specify do
+                get 'http://www.example.org/secret/test.pdf'
+                last_response.headers["Content-Type"].should == "text/html"
+                last_response.body.should == "Hello world!"
+              end
+            end
+          end # multiple string
+        end # string
+
+      end
+
+      describe ":except" do
+
+        describe "regex" do
+          describe "one" do
+            before { mock_app({}, :except => %r[^/secret]) }
+
+            context "matching" do
+              specify do
+                get 'http://www.example.org/public/test.pdf'
+                last_response.headers["Content-Type"].should == "application/pdf"
+                last_response.body.bytesize.should == PDFKit.new("Hello world!").to_pdf.bytesize
+              end
+            end
+
+            context "not matching" do
+              specify do
+                get 'http://www.example.org/secret/test.pdf'
+                last_response.headers["Content-Type"].should == "text/html"
+                last_response.body.should == "Hello world!"
+              end
+            end
+          end # one regex
+
+          describe "multiple" do
+            before { mock_app({}, :except => [%r[^/prawn], %r[^/secret]]) }
+
+            context "matching" do
+              specify do
+                get 'http://www.example.org/public/test.pdf'
+                last_response.headers["Content-Type"].should == "application/pdf"
+                last_response.body.bytesize.should == PDFKit.new("Hello world!").to_pdf.bytesize
+              end
+            end
+
+            context "not matching" do
+              specify do
+                get 'http://www.example.org/secret/test.pdf'
+                last_response.headers["Content-Type"].should == "text/html"
+                last_response.body.should == "Hello world!"
+              end
+            end
+          end # multiple regex
+        end # regex
+
+        describe "string" do
+          describe "one" do
+            before { mock_app({}, :except => '/secret') }
+
+            context "matching" do
+              specify do
+                get 'http://www.example.org/public/test.pdf'
+                last_response.headers["Content-Type"].should == "application/pdf"
+                last_response.body.bytesize.should == PDFKit.new("Hello world!").to_pdf.bytesize
+              end
+            end
+
+            context "not matching" do
+              specify do
+                get 'http://www.example.org/secret/test.pdf'
+                last_response.headers["Content-Type"].should == "text/html"
+                last_response.body.should == "Hello world!"
+              end
+            end
+          end # one string
+
+          describe "multiple" do
+            before { mock_app({}, :except => ['/prawn', '/secret']) }
+
+            context "matching" do
+              specify do
+                get 'http://www.example.org/public/test.pdf'
+                last_response.headers["Content-Type"].should == "application/pdf"
+                last_response.body.bytesize.should == PDFKit.new("Hello world!").to_pdf.bytesize
+              end
+            end
+
+            context "not matching" do
+              specify do
+                get 'http://www.example.org/secret/test.pdf'
+                last_response.headers["Content-Type"].should == "text/html"
+                last_response.body.should == "Hello world!"
+              end
+            end
+          end # multiple string
+        end # string
+
+      end
+    end
+
+  describe "remove .pdf from PATH_INFO and REQUEST_URI" do
+    before { mock_app }
+
+      context "matching" do
+
+        specify do
+          get 'http://www.example.org/public/file.pdf'
+          @env["PATH_INFO"].should == "/public/file"
+          @env["REQUEST_URI"].should == "/public/file"
+          @env["SCRIPT_NAME"].should be_empty
+        end
+        specify do
+          get 'http://www.example.org/public/file.txt'
+          @env["PATH_INFO"].should == "/public/file.txt"
+          @env["REQUEST_URI"].should be_nil
+          @env["SCRIPT_NAME"].should be_empty
+        end
+      end
+
+      context "subdomain matching" do
+        before do
+          main_app = lambda { |env|
+            @env = env
+            @env['SCRIPT_NAME'] = '/example.org'
+            headers = {'Content-Type' => "text/html"}
+            [200, headers, @body || ['Hello world!']]
+          }
+
+          builder = Rack::Builder.new
+          builder.use PDFKit::Middleware
+          builder.run main_app
+          @app = builder.to_app
+        end
+        specify do
+          get 'http://example.org/sub/public/file.pdf'
+          @env["PATH_INFO"].should == "/sub/public/file"
+          @env["REQUEST_URI"].should == "/sub/public/file"
+          @env["SCRIPT_NAME"].should == "/example.org"
+        end
+        specify do
+          get 'http://example.org/sub/public/file.txt'
+          @env["PATH_INFO"].should == "/sub/public/file.txt"
+          @env["REQUEST_URI"].should be_nil
+          @env["SCRIPT_NAME"].should == "/example.org"
+        end
+      end
+
+    end
+  end
+
+  describe "#translate_paths" do
+    before do
+      @pdf = PDFKit::Middleware.new({})
+      @env = { 'REQUEST_URI' => 'http://example.com/document.pdf', 'rack.url_scheme' => 'http', 'HTTP_HOST' => 'example.com' }
+    end
+
+    it "should correctly parse relative url with single quotes" do
+      @body = %{<html><head><link href='/stylesheets/application.css' media='screen' rel='stylesheet' type='text/css' /></head><body><img alt='test' src="/test.png" /></body></html>}
+      body = @pdf.send :translate_paths, @body, @env
+      body.should == "<html><head><link href='http://example.com/stylesheets/application.css' media='screen' rel='stylesheet' type='text/css' /></head><body><img alt='test' src=\"http://example.com/test.png\" /></body></html>"
+    end
+
+    it "should correctly parse relative url with double quotes" do
+      @body = %{<link href="/stylesheets/application.css" media="screen" rel="stylesheet" type="text/css" />}
+      body = @pdf.send :translate_paths, @body, @env
+      body.should == "<link href=\"http://example.com/stylesheets/application.css\" media=\"screen\" rel=\"stylesheet\" type=\"text/css\" />"
+    end
+
+    it "should return the body even if there are no valid substitutions found" do
+      @body = "NO MATCH"
+      body = @pdf.send :translate_paths, @body, @env
+      body.should == "NO MATCH"
+    end
+  end
+
+  describe "#translate_paths with root_url configuration" do
+    before do
+      @pdf = PDFKit::Middleware.new({})
+      @env = { 'REQUEST_URI' => 'http://example.com/document.pdf', 'rack.url_scheme' => 'http', 'HTTP_HOST' => 'example.com' }
+      PDFKit.configure do |config|
+        config.root_url = "http://example.net/"
+      end
+    end
+
+    it "should add the root_url" do
+      @body = %{<html><head><link href='/stylesheets/application.css' media='screen' rel='stylesheet' type='text/css' /></head><body><img alt='test' src="/test.png" /></body></html>}
+      body = @pdf.send :translate_paths, @body, @env
+      body.should == "<html><head><link href='http://example.net/stylesheets/application.css' media='screen' rel='stylesheet' type='text/css' /></head><body><img alt='test' src=\"http://example.net/test.png\" /></body></html>"
+    end
+
+    after do
+      PDFKit.configure do |config|
+        config.root_url = nil
+      end
+    end
+  end
+
+  it "should not get stuck rendering each request as pdf" do
+    mock_app
+    # false by default. No requests.
+    @app.send(:rendering_pdf?).should be_false
+
+    # Remain false on a normal request
+    get 'http://www.example.org/public/file'
+    @app.send(:rendering_pdf?).should be_false
+
+    # Return true on a pdf request.
+    get 'http://www.example.org/public/file.pdf'
+    @app.send(:rendering_pdf?).should be_true
+
+    # Restore to false on any non-pdf request.
+    get 'http://www.example.org/public/file'
+    @app.send(:rendering_pdf?).should be_false
+  end
+
+end

data/spec/pdfkit_spec.rb

@@ -0,0 +1,255 @@
+#encoding: UTF-8
+require 'spec_helper'
+
+describe PDFKit do
+
+  context "initialization" do
+    it "should accept HTML as the source" do
+      pdfkit = PDFKit.new('<h1>Oh Hai</h1>')
+      pdfkit.source.should be_html
+      pdfkit.source.to_s.should == '<h1>Oh Hai</h1>'
+    end
+
+    it "should accept a URL as the source" do
+      pdfkit = PDFKit.new('http://google.com')
+      pdfkit.source.should be_url
+      pdfkit.source.to_s.should == 'http://google.com'
+    end
+
+    it "should accept a File as the source" do
+      file_path = File.join(SPEC_ROOT,'fixtures','example.html')
+      pdfkit = PDFKit.new(File.new(file_path))
+      pdfkit.source.should be_file
+      pdfkit.source.to_s.should == file_path
+    end
+
+    it "should parse the options into a cmd line friedly format" do
+      pdfkit = PDFKit.new('html', :page_size => 'Letter')
+      pdfkit.options.should have_key('--page-size')
+    end
+
+    it "should parse complex options into a cmd line friedly format" do
+      pdfkit = PDFKit.new('html', :replace => {'value' => 'something else'} )
+      pdfkit.options.should have_key('--replace')
+    end
+
+    it "should provide default options" do
+      pdfkit = PDFKit.new('<h1>Oh Hai</h1>')
+      ['--margin-top', '--margin-right', '--margin-bottom', '--margin-left'].each do |option|
+        pdfkit.options.should have_key(option)
+      end
+    end
+
+    it "should default to 'UTF-8' encoding" do
+      pdfkit = PDFKit.new('Captación')
+      pdfkit.options['--encoding'].should == 'UTF-8'
+    end
+
+    it "should not have any stylesheedt by default" do
+      pdfkit = PDFKit.new('<h1>Oh Hai</h1>')
+      pdfkit.stylesheets.should be_empty
+    end
+  end
+
+  context "command" do
+    it "should contstruct the correct command" do
+      pdfkit = PDFKit.new('html', :page_size => 'Letter', :toc_l1_font_size => 12, :replace => {'foo' => 'bar'})
+      command = pdfkit.command
+      command.should include "wkhtmltopdf"
+      command.should include "--page-size Letter"
+      command.should include "--toc-l1-font-size 12"
+      command.should include "--replace foo bar"
+    end
+
+    it "will not include default options it is told to omit" do
+      PDFKit.configure do |config|
+        config.default_options[:disable_smart_shrinking] = true
+      end
+
+      pdfkit = PDFKit.new('html')
+      pdfkit.command.should include('--disable-smart-shrinking')
+      pdfkit = PDFKit.new('html', :disable_smart_shrinking => false)
+      pdfkit.command.should_not include('--disable-smart-shrinking')
+    end
+
+    it "should encapsulate string arguments in quotes" do
+      pdfkit = PDFKit.new('html', :header_center => "foo [page]")
+      pdfkit.command.should include "--header-center foo\\ \\[page\\]"
+    end
+
+    it "should sanitize string arguments" do
+      pdfkit = PDFKit.new('html', :header_center => "$(ls)")
+      pdfkit.command.should include "--header-center \\$\\(ls\\)"
+    end
+
+    it "read the source from stdin if it is html" do
+      pdfkit = PDFKit.new('html')
+      pdfkit.command.should match /- -$/
+    end
+
+    it "specify the URL to the source if it is a url" do
+      pdfkit = PDFKit.new('http://google.com')
+      pdfkit.command.should match /http:\/\/google.com -$/
+    end
+
+    it "should specify the path to the source if it is a file" do
+      file_path = File.join(SPEC_ROOT,'fixtures','example.html')
+      pdfkit = PDFKit.new(File.new(file_path))
+      pdfkit.command.should match /#{file_path} -$/
+    end
+
+    it "should specify the path for the ouput if a apth is given" do
+      file_path = "/path/to/output.pdf"
+      pdfkit = PDFKit.new("html")
+      pdfkit.command(file_path).should match /#{file_path}$/
+    end
+
+    it "should detect special pdfkit meta tags" do
+      body = %{
+        <html>
+          <head>
+            <meta name="pdfkit-page_size" content="Legal"/>
+            <meta name="pdfkit-orientation" content="Landscape"/>
+          </head>
+        </html>
+      }
+      pdfkit = PDFKit.new(body)
+      command = pdfkit.command
+      command.should include "--page-size Legal"
+      command.should include "--orientation Landscape"
+    end
+
+    it "should detect special pdfkit meta tags despite bad markup" do
+      body = %{
+        <html>
+          <head>
+            <meta name="pdfkit-page_size" content="Legal"/>
+            <meta name="pdfkit-orientation" content="Landscape"/>
+          </head>
+          <br>
+        </html>
+      }
+      pdfkit = PDFKit.new(body)
+      command = pdfkit.command
+      command.should include "--page-size Legal"
+      command.should include "--orientation Landscape"
+    end
+
+    it "should skip non-pdfkit meta tags" do
+      body = %{
+        <html>
+          <head>
+            <meta name="test-page_size" content="Legal"/>
+            <meta name="pdfkit-orientation" content="Landscape"/>
+          </head>
+          <br>
+        </html>
+      }
+      pdfkit = PDFKit.new(body)
+      command = pdfkit.command
+      command.should_not include "--page-size Legal"
+      command.should include "--orientation Landscape"
+    end
+
+  end
+
+  context "#to_pdf" do
+    it "should generate a PDF of the HTML" do
+      pdfkit = PDFKit.new('html', :page_size => 'Letter')
+      pdf = pdfkit.to_pdf
+      pdf[0...4].should == "%PDF" # PDF Signature at beginning of file
+    end
+
+    it "should generate a PDF with a numerical parameter" do
+      pdfkit = PDFKit.new('html', :header_spacing => 1)
+      pdf = pdfkit.to_pdf
+      pdf[0...4].should == "%PDF" # PDF Signature at beginning of file
+    end
+
+    it "should generate a PDF with a symbol parameter" do
+      pdfkit = PDFKit.new('html', :page_size => :Letter)
+      pdf = pdfkit.to_pdf
+      pdf[0...4].should == "%PDF" # PDF Signature at beginning of file
+    end
+
+    it "should have the stylesheet added to the head if it has one" do
+      pdfkit = PDFKit.new("<html><head></head><body>Hai!</body></html>")
+      css = File.join(SPEC_ROOT,'fixtures','example.css')
+      pdfkit.stylesheets << css
+      pdfkit.to_pdf
+      pdfkit.source.to_s.should include("<style>#{File.read(css)}</style>")
+    end
+
+    it "should prepend style tags if the HTML doesn't have a head tag" do
+      pdfkit = PDFKit.new("<html><body>Hai!</body></html>")
+      css = File.join(SPEC_ROOT,'fixtures','example.css')
+      pdfkit.stylesheets << css
+      pdfkit.to_pdf
+      pdfkit.source.to_s.should include("<style>#{File.read(css)}</style><html>")
+    end
+
+    it "should throw an error if the source is not html and stylesheets have been added" do
+      pdfkit = PDFKit.new('http://google.com')
+      css = File.join(SPEC_ROOT,'fixtures','example.css')
+      pdfkit.stylesheets << css
+      lambda { pdfkit.to_pdf }.should raise_error(PDFKit::ImproperSourceError)
+    end
+
+    it "should be able to deal with ActiveSupport::SafeBuffer" do
+      pdfkit = PDFKit.new(ActiveSupport::SafeBuffer.new "<html><head></head><body>Hai!</body></html>")
+      css = File.join(SPEC_ROOT,'fixtures','example.css')
+      pdfkit.stylesheets << css
+      pdfkit.to_pdf
+      pdfkit.source.to_s.should include("<style>#{File.read(css)}</style></head>")
+    end
+
+    it "should throw an error if it is unable to connect" do
+      pdfkit = PDFKit.new("http://google.com/this-should-not-be-found/404.html")
+      lambda { pdfkit.to_pdf }.should raise_error
+    end
+  end
+
+  context "#to_file" do
+    before do
+      @file_path = File.join(SPEC_ROOT,'fixtures','test.pdf')
+      File.delete(@file_path) if File.exist?(@file_path)
+    end
+
+    after do
+      File.delete(@file_path)
+    end
+
+    it "should create a file with the PDF as content" do
+      pdfkit = PDFKit.new('html', :page_size => 'Letter')
+      file = pdfkit.to_file(@file_path)
+      file.should be_instance_of(File)
+      File.read(file.path)[0...4].should == "%PDF" # PDF Signature at beginning of file
+    end
+
+    it "should not truncate data (in Ruby 1.8.6)" do
+      file_path = File.join(SPEC_ROOT,'fixtures','example.html')
+      pdfkit = PDFKit.new(File.new(file_path))
+      pdf_data = pdfkit.to_pdf
+      file = pdfkit.to_file(@file_path)
+      file_data = open(@file_path, 'rb') {|io| io.read }
+      pdf_data.size.should == file_data.size
+    end
+  end
+
+  context "security" do
+    before do
+      @test_path = File.join(SPEC_ROOT,'fixtures','security-oops')
+      File.delete(@test_path) if File.exist?(@test_path)
+    end
+
+    after do
+      File.delete(@test_path) if File.exist?(@test_path)
+    end
+
+    it "should not allow shell injection in options" do
+      pdfkit = PDFKit.new('html', :header_center => "a title\"; touch #{@test_path} #")
+      pdfkit.to_pdf
+      File.exist?(@test_path).should be_false
+    end
+  end
+end

data/spec/source_spec.rb

@@ -0,0 +1,78 @@
+require 'spec_helper'
+
+describe PDFKit::Source do
+  
+  describe "#url?" do
+    it "should return true if passed a url like string" do
+      source = PDFKit::Source.new('http://google.com')
+      source.should be_url
+    end
+    
+    it "should return false if passed a file" do
+      source = PDFKit::Source.new(File.new(__FILE__))
+      source.should_not be_url
+    end
+    
+    it "should return false if passed HTML" do
+      source = PDFKit::Source.new('<blink>Oh Hai!</blink>')
+      source.should_not be_url
+    end
+
+    it "should return false if passed HTML with embedded urls at the beginning of a line" do
+      source = PDFKit::Source.new("<blink>Oh Hai!</blink>\nhttp://www.google.com")
+      source.should_not be_url
+    end
+  end
+  
+  describe "#file?" do
+    it "should return true if passed a file" do
+      source = PDFKit::Source.new(File.new(__FILE__))
+      source.should be_file
+    end
+    
+    it "should return false if passed a url like string" do
+      source = PDFKit::Source.new('http://google.com')
+      source.should_not be_file
+    end
+    
+    it "should return false if passed HTML" do
+      source = PDFKit::Source.new('<blink>Oh Hai!</blink>')
+      source.should_not be_file
+    end
+  end
+  
+  describe "#html?" do
+    it "should return true if passed HTML" do
+      source = PDFKit::Source.new('<blink>Oh Hai!</blink>')
+      source.should be_html
+    end
+    
+    it "should return false if passed a file" do
+      source = PDFKit::Source.new(File.new(__FILE__))
+      source.should_not be_html
+    end
+    
+    it "should return false if passed a url like string" do
+      source = PDFKit::Source.new('http://google.com')
+      source.should_not be_html
+    end
+  end
+  
+  describe "#to_s" do
+    it "should return the HTML if passed HTML" do
+      source = PDFKit::Source.new('<blink>Oh Hai!</blink>')
+      source.to_s.should == '<blink>Oh Hai!</blink>'
+    end
+    
+    it "should return a path if passed a file" do
+      source = PDFKit::Source.new(File.new(__FILE__))
+      source.to_s.should == __FILE__
+    end
+    
+    it "should return the url if passed a url like string" do
+      source = PDFKit::Source.new('http://google.com')
+      source.to_s.should == 'http://google.com'
+    end
+  end
+  
+end

data/spec/spec_helper.rb

@@ -0,0 +1,15 @@
+SPEC_ROOT = File.dirname(__FILE__)
+$LOAD_PATH.unshift(SPEC_ROOT)
+$LOAD_PATH.unshift(File.join(SPEC_ROOT, '..', 'lib'))
+require 'pdfkit'
+require 'rspec'
+require 'rspec/autorun'
+require 'mocha'
+require 'rack'
+require 'rack/test'
+require 'active_support'
+require 'custom_wkhtmltopdf_path' if File.exists?(File.join(SPEC_ROOT, 'custom_wkhtmltopdf_path.rb'))
+
+RSpec.configure do |config|
+  include Rack::Test::Methods
+end