Add AttachmentToHTML library

Extracts the attachment processing from FoiAttachment#body_to_html

AttachmentToHTML contains adapters which convert
  - text/plain
  - application/pdf
  - application/rtf

Results are returned as an AttachmentHTML::HTML instance which contains
the raw HTML and other metadata about the conversion.

7 files changed, 537 insertions, 0 deletions

diff --git a/spec/lib/attachment_to_html/adapters/could_not_convert_spec.rb b/spec/lib/attachment_to_html/adapters/could_not_convert_spec.rb
new file mode 100644
index 000000000..aee68e986
--- /dev/null
+++ b/spec/lib/attachment_to_html/adapters/could_not_convert_spec.rb
@@ -0,0 +1,67 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../../spec_helper')
+
+describe AttachmentToHTML::Adapters::CouldNotConvert do
+
+    let(:attachment) { FactoryGirl.build(:pdf_attachment) }
+    let(:adapter) { AttachmentToHTML::Adapters::CouldNotConvert.new(attachment) }
+
+    describe :wrapper do
+
+        it 'defaults to wrapper' do
+           adapter.wrapper.should == 'wrapper'
+        end
+
+        it 'accepts a wrapper option' do
+            adapter = AttachmentToHTML::Adapters::CouldNotConvert.new(attachment, :wrapper => 'wrap')
+            adapter.wrapper.should == 'wrap'
+        end
+ 
+    end
+
+    describe :to_html do
+
+        it 'should be a valid html document' do
+            parsed = Nokogiri::HTML.parse(adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.errors.any?.should be_false
+        end
+
+        it 'contains the attachment filename in the title tag' do
+            parsed = Nokogiri::HTML.parse(adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('title').inner_html.should == attachment.display_filename
+        end
+
+        it 'contains the wrapper div in the body tag' do
+            adapter = AttachmentToHTML::Adapters::CouldNotConvert.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('body div').first.attributes['id'].value.should == 'wrap'
+        end
+
+        it 'should contain text about the conversion failure' do
+            adapter = AttachmentToHTML::Adapters::CouldNotConvert.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(adapter.to_html) do |config|
+                config.strict
+            end
+
+            expected = "<p>Sorry, we were unable to convert this file to HTML. " \
+                       "Please use the download link at the top right.</p>"
+
+            parsed.css('div#wrap div#view-html-content').inner_html.should == expected
+        end
+
+    end
+
+    describe :success? do
+
+        it 'is always true' do
+            adapter.success?.should be_true
+        end
+
+    end
+
+end
diff --git a/spec/lib/attachment_to_html/adapters/google_docs_viewer_spec.rb b/spec/lib/attachment_to_html/adapters/google_docs_viewer_spec.rb
new file mode 100644
index 000000000..166ca3241
--- /dev/null
+++ b/spec/lib/attachment_to_html/adapters/google_docs_viewer_spec.rb
@@ -0,0 +1,81 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../../spec_helper')
+
+describe AttachmentToHTML::Adapters::GoogleDocsViewer do
+
+    let(:attachment) { FactoryGirl.build(:pdf_attachment) }
+    let(:google_adapter) do
+        AttachmentToHTML::Adapters::GoogleDocsViewer.new(attachment, :attachment_url => 'http://example.com/test.pdf')
+    end
+
+    describe :wrapper do
+
+        it 'defaults to wrapper_google_embed' do
+           google_adapter.wrapper.should == 'wrapper_google_embed'
+        end
+
+        it 'accepts a wrapper option' do
+            google_adapter = AttachmentToHTML::Adapters::GoogleDocsViewer.new(attachment, :wrapper => 'wrap')
+            google_adapter.wrapper.should == 'wrap'
+        end
+ 
+    end
+
+    describe :to_html do
+
+        it 'should be a valid html document' do
+            parsed = Nokogiri::HTML.parse(google_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.errors.any?.should be_false
+        end
+
+        it 'contains the attachment filename in the title tag' do
+            parsed = Nokogiri::HTML.parse(google_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('title').inner_html.should == attachment.display_filename
+        end
+
+        it 'contains the wrapper div in the body tag' do
+            google_adapter = AttachmentToHTML::Adapters::GoogleDocsViewer.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(google_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('body div').first.attributes['id'].value.should == 'wrap'
+        end
+
+        it 'contains the google docs viewer url in the wrapper div' do
+            options = { :wrapper => 'wrap', :attachment_url => 'http://example.com/test.pdf' }
+            google_adapter = AttachmentToHTML::Adapters::GoogleDocsViewer.new(attachment, options)
+            parsed = Nokogiri::HTML.parse(google_adapter.to_html) do |config|
+               config.strict
+            end
+            expected = %Q(<iframe src="http://docs.google.com/viewer?url=http://example.com/test.pdf&amp;embedded=true" width="100%" height="100%" style="border: none;"></iframe>)
+            parsed.css('div#wrap div#view-html-content').inner_html.should include(expected)
+        end
+
+        describe 'uses the confugured alaveteli protocol' do
+
+            it 'https if force_ssl is on' do
+                AlaveteliConfiguration.stub(:force_ssl).and_return(true)
+                google_adapter.to_html.should include('https://docs.google.com')
+            end
+
+            it 'http if force_ssl is off' do
+                AlaveteliConfiguration.stub(:force_ssl).and_return(false)
+                google_adapter.to_html.should include('http://docs.google.com')
+            end
+
+        end
+
+    end
+
+    describe :success? do
+
+        it 'is always true' do
+            google_adapter.success?.should be_true
+        end
+
+    end
+
+end
diff --git a/spec/lib/attachment_to_html/adapters/pdf_spec.rb b/spec/lib/attachment_to_html/adapters/pdf_spec.rb
new file mode 100644
index 000000000..65c376043
--- /dev/null
+++ b/spec/lib/attachment_to_html/adapters/pdf_spec.rb
@@ -0,0 +1,98 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../../spec_helper')
+
+describe AttachmentToHTML::Adapters::PDF do
+
+    let(:attachment) { FactoryGirl.build(:pdf_attachment) }
+    let(:pdf_adapter) { AttachmentToHTML::Adapters::PDF.new(attachment) }
+
+    describe :wrapper do
+
+        it 'defaults to wrapper' do
+           pdf_adapter.wrapper.should == 'wrapper'
+        end
+
+        it 'accepts a wrapper option' do
+            pdf_adapter = AttachmentToHTML::Adapters::PDF.new(attachment, :wrapper => 'wrap')
+            pdf_adapter.wrapper.should == 'wrap'
+        end
+ 
+    end
+
+    describe :tmpdir do
+
+        it 'defaults to the rails tmp directory' do
+           pdf_adapter.tmpdir.should == Rails.root.join('tmp') 
+        end
+
+        it 'allows a tmpdir to be specified to store the converted document' do
+            pdf_adapter = AttachmentToHTML::Adapters::PDF.new(attachment, :tmpdir => '/tmp')
+            pdf_adapter.tmpdir.should == '/tmp'
+        end
+  
+    end
+
+    describe :to_html do
+
+        it 'should be a valid html document' do
+            parsed = Nokogiri::HTML.parse(pdf_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.errors.any?.should be_false
+        end
+
+        it 'contains the attachment filename in the title tag' do
+            parsed = Nokogiri::HTML.parse(pdf_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('title').inner_html.should == attachment.display_filename
+        end
+
+        it 'contains the wrapper div in the body tag' do
+            pdf_adapter = AttachmentToHTML::Adapters::PDF.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(pdf_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('body div').first.attributes['id'].value.should == 'wrap'
+        end
+
+        it 'contains the attachment body in the wrapper div' do
+            pdf_adapter = AttachmentToHTML::Adapters::PDF.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(pdf_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('div#wrap div#view-html-content').inner_html.should include('thisisthebody')
+        end
+
+        it 'operates in the context of the supplied tmpdir' do
+            pdf_adapter = AttachmentToHTML::Adapters::PDF.new(attachment, :tmpdir => '/tmp')
+            Dir.should_receive(:chdir).with('/tmp').and_call_original
+            pdf_adapter.to_html
+        end
+
+    end
+
+    describe :success? do
+
+        it 'is successful if the body has content excluding the tags' do
+            pdf_adapter.to_html
+            pdf_adapter.success?.should be_true
+        end
+
+        it 'is successful if the body contains images' do
+            mocked_return = %Q(<!DOCTYPE html><html><head></head><body><img src="logo.png" /></body></html>)
+            pdf_adapter = AttachmentToHTML::Adapters::PDF.new(attachment)
+            pdf_adapter.stub(:to_html).and_return(mocked_return)
+            pdf_adapter.success?.should be_true
+        end
+
+        it 'is not successful if the body has no content other than tags' do
+            # TODO: Add and use spec/fixtures/files/empty.pdf
+            attachment = FactoryGirl.build(:body_text, :body => '')
+            pdf_adapter = AttachmentToHTML::Adapters::PDF.new(attachment)
+            pdf_adapter.to_html
+            pdf_adapter.success?.should be_false
+        end
+
+    end
+
+end
diff --git a/spec/lib/attachment_to_html/adapters/rtf_spec.rb b/spec/lib/attachment_to_html/adapters/rtf_spec.rb
new file mode 100644
index 000000000..f84073c51
--- /dev/null
+++ b/spec/lib/attachment_to_html/adapters/rtf_spec.rb
@@ -0,0 +1,98 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../../spec_helper')
+
+describe AttachmentToHTML::Adapters::RTF do
+
+    let(:attachment) { FactoryGirl.build(:rtf_attachment) }
+    let(:rtf_adapter) { AttachmentToHTML::Adapters::RTF.new(attachment) }
+
+    describe :wrapper do
+
+        it 'defaults to wrapper' do
+           rtf_adapter.wrapper.should == 'wrapper'
+        end
+
+        it 'accepts a wrapper option' do
+            rtf_adapter = AttachmentToHTML::Adapters::RTF.new(attachment, :wrapper => 'wrap')
+            rtf_adapter.wrapper.should == 'wrap'
+        end
+ 
+    end
+
+    describe :tmpdir do
+
+        it 'defaults to the rails tmp directory' do
+           rtf_adapter.tmpdir.should == Rails.root.join('tmp') 
+        end
+
+        it 'allows a tmpdir to be specified to store the converted document' do
+            rtf_adapter = AttachmentToHTML::Adapters::RTF.new(attachment, :tmpdir => '/tmp')
+            rtf_adapter.tmpdir.should == '/tmp'
+        end
+  
+    end
+
+    describe :to_html do
+
+        it 'should be a valid html document' do
+            parsed = Nokogiri::HTML.parse(rtf_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.errors.any?.should be_false
+        end
+
+        it 'contains the attachment filename in the title tag' do
+            parsed = Nokogiri::HTML.parse(rtf_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('head title').inner_html.should == attachment.display_filename
+        end
+
+        it 'contains the wrapper div in the body tag' do
+            rtf_adapter = AttachmentToHTML::Adapters::RTF.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(rtf_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('body div').first.attributes['id'].value.should == 'wrap'
+        end
+
+        it 'contains the attachment body in the wrapper div' do
+            rtf_adapter = AttachmentToHTML::Adapters::RTF.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(rtf_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.css('div#wrap div#view-html-content').inner_text.should include('thisisthebody')
+        end
+
+        it 'operates in the context of the supplied tmpdir' do
+            rtf_adapter = AttachmentToHTML::Adapters::RTF.new(attachment, :tmpdir => '/tmp')
+            Dir.should_receive(:chdir).with('/tmp').and_call_original
+            rtf_adapter.to_html
+        end
+
+    end
+
+    describe :success? do
+
+        it 'is successful if the body has content excluding the tags' do
+            rtf_adapter.to_html
+            rtf_adapter.success?.should be_true
+        end
+
+        it 'is successful if the body contains images' do
+            mocked_return = %Q(<!DOCTYPE html><html><head></head><body><img src="logo.png" /></body></html>)
+            rtf_adapter = AttachmentToHTML::Adapters::RTF.new(attachment)
+            rtf_adapter.stub(:to_html).and_return(mocked_return)
+            rtf_adapter.success?.should be_true
+        end
+
+        it 'is not successful if the body has no content other than tags' do
+            empty_rtf = load_file_fixture('empty.rtf')
+            attachment = FactoryGirl.build(:rtf_attachment, :body => empty_rtf)
+            rtf_adapter = AttachmentToHTML::Adapters::RTF.new(attachment)
+            rtf_adapter.to_html
+            rtf_adapter.success?.should be_false
+        end
+
+    end
+
+end
diff --git a/spec/lib/attachment_to_html/adapters/text_spec.rb b/spec/lib/attachment_to_html/adapters/text_spec.rb
new file mode 100644
index 000000000..599670603
--- /dev/null
+++ b/spec/lib/attachment_to_html/adapters/text_spec.rb
@@ -0,0 +1,121 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../../spec_helper')
+
+describe AttachmentToHTML::Adapters::Text do
+
+    let(:attachment) { FactoryGirl.build(:body_text) }
+    let(:text_adapter) { AttachmentToHTML::Adapters::Text.new(attachment) }
+
+    describe :wrapper do
+
+        it 'defaults to wrapper' do
+           text_adapter.wrapper.should == 'wrapper'
+        end
+
+        it 'accepts a wrapper option' do
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment, :wrapper => 'wrap')
+            text_adapter.wrapper.should == 'wrap'
+        end
+ 
+    end
+
+    describe :to_html do
+
+        it 'should be a valid html document' do
+            parsed = Nokogiri::HTML.parse(text_adapter.to_html) do |config|
+               config.strict
+            end
+            parsed.errors.any?.should be_false
+        end
+
+        it 'contains the attachment filename in the title tag' do
+            parsed = Nokogiri::HTML.parse(text_adapter.to_html) do |config|
+                config.strict
+            end
+            parsed.css('title').inner_html.should == attachment.display_filename
+        end
+
+        it 'contains the wrapper div in the body tag' do
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(text_adapter.to_html) do |config|
+                config.strict
+            end
+            parsed.css('body').children.first.attributes['id'].value.should == 'wrap'
+        end
+
+        it 'contains the attachment body in the wrapper div' do
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment, :wrapper => 'wrap')
+            parsed = Nokogiri::HTML.parse(text_adapter.to_html) do |config|
+                config.strict
+            end
+            parsed.css('div#wrap div#view-html-content').inner_html.should == attachment.body
+        end
+ 
+        it 'strips the body of trailing whitespace' do
+            attachment = FactoryGirl.build(:body_text, :body => ' Hello ')
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment)
+            parsed = Nokogiri::HTML.parse(text_adapter.to_html) do |config|
+                config.strict
+            end
+            parsed.css('div#wrapper div#view-html-content').inner_html.should == 'Hello'
+        end
+
+        # NOTE: Can't parse this spec with Nokogiri at the moment because even
+        # in strict mode Nokogiri tampers with the HTML returned:
+        #      Failure/Error: parsed.css('div#wrapper div#view-html-content').
+        #        inner_html.should == expected
+        #          expected: "Usage: foo &quot;bar&quot; &gt;baz&lt;"
+        #          got: "Usage: foo \"bar\" &gt;baz&lt;" (using ==)
+        it 'escapes special characters' do
+            attachment = FactoryGirl.build(:body_text, :body => 'Usage: foo "bar" >baz<')
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment)
+            expected = %Q(Usage: foo &quot;bar&quot; &gt;baz&lt;)
+            text_adapter.to_html.should include(expected)
+        end
+
+        it 'creates hyperlinks for text that looks like a url' do
+            attachment = FactoryGirl.build(:body_text, :body => 'http://www.whatdotheyknow.com')
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment)
+            parsed = Nokogiri::HTML.parse(text_adapter.to_html) do |config|
+                config.strict
+            end
+            parsed.css('div#wrapper div#view-html-content a').first.text.should == 'http://www.whatdotheyknow.com'
+            parsed.css('div#wrapper div#view-html-content a').first['href'].should == 'http://www.whatdotheyknow.com'
+        end
+
+        it 'substitutes newlines for br tags' do
+            attachment = FactoryGirl.build(:body_text, :body => "A\nNewline")
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment)
+            parsed = Nokogiri::HTML.parse(text_adapter.to_html) do |config|
+                config.strict
+            end
+            expected = %Q(A<br>Newline)
+            parsed.css('div#wrapper div#view-html-content').inner_html.should == expected
+        end
+
+    end
+
+    describe :success? do
+
+        it 'is successful if the body has content excluding the tags' do
+            text_adapter.to_html
+            text_adapter.success?.should be_true
+        end
+
+        it 'is successful if the body contains images' do
+            mocked_return = %Q(<!DOCTYPE html><html><head></head><body><img src="logo.png" /></body></html>)
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment)
+            text_adapter.stub(:to_html).and_return(mocked_return)
+            text_adapter.success?.should be_true
+        end
+
+        it 'is not successful if the body has no content other than tags' do
+            empty_txt = load_file_fixture('empty.txt')
+            attachment = FactoryGirl.build(:body_text, :body => empty_txt)
+            text_adapter = AttachmentToHTML::Adapters::Text.new(attachment)
+            text_adapter.to_html
+            text_adapter.success?.should be_false
+        end
+
+    end
+
+end
diff --git a/spec/lib/attachment_to_html/attachment_to_html_spec.rb b/spec/lib/attachment_to_html/attachment_to_html_spec.rb
new file mode 100644
index 000000000..f7df06f87
--- /dev/null
+++ b/spec/lib/attachment_to_html/attachment_to_html_spec.rb
@@ -0,0 +1,48 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+describe AttachmentToHTML do
+    include AttachmentToHTML
+
+    let(:attachment) { FactoryGirl.build(:body_text) }
+
+    describe :to_html do
+
+        it 'sends the attachment to the correct adapter for conversion' do
+            AttachmentToHTML::Adapters::Text.should_receive(:new).with(attachment, {}).and_call_original
+            to_html(attachment)
+        end
+
+        it 'returns the results in a HTML class' do
+            expected = AttachmentToHTML::Adapters::Text.new(attachment).to_html
+            to_html(attachment).should be_instance_of(AttachmentToHTML::HTML)
+        end
+ 
+        it 'accepts a hash of options to pass to the adapter' do
+            options = { :wrapper => 'wrap' }
+            AttachmentToHTML::Adapters::Text.should_receive(:new).with(attachment, options).and_call_original
+            to_html(attachment, options)
+        end
+
+        it 'converts an attachment that has an adapter, fails to convert, but has a google viewer' do
+            attachment = FactoryGirl.build(:pdf_attachment)
+            AttachmentToHTML::HTML.any_instance.stub(:success?).and_return(false)
+            AttachmentToHTML::Adapters::PDF.should_receive(:new).with(attachment, {}).and_call_original
+            AttachmentToHTML::Adapters::GoogleDocsViewer.should_receive(:new).with(attachment, {})
+            to_html(attachment)
+        end
+
+        it 'converts an attachment that doesnt have an adapter, but has a google viewer' do
+            attachment = FactoryGirl.build(:body_text, :content_type => 'application/vnd.ms-word')
+            AttachmentToHTML::Adapters::GoogleDocsViewer.should_receive(:new).with(attachment, {}).and_call_original
+            to_html(attachment)
+        end
+
+        it 'converts an attachment that has no adapter or google viewer' do
+            attachment = FactoryGirl.build(:body_text, :content_type => 'application/json')
+            AttachmentToHTML::Adapters::CouldNotConvert.should_receive(:new).with(attachment, {}).and_call_original
+            to_html(attachment)
+        end
+
+    end
+
+end
diff --git a/spec/lib/attachment_to_html/html_spec.rb b/spec/lib/attachment_to_html/html_spec.rb
new file mode 100644
index 000000000..65b63d383
--- /dev/null
+++ b/spec/lib/attachment_to_html/html_spec.rb
@@ -0,0 +1,24 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+describe AttachmentToHTML::HTML do
+
+    let(:adapter) { OpenStruct.new(:to_html => '<p>hello</p>', :success? => true) }
+    let(:html) { AttachmentToHTML::HTML.new(adapter) }
+
+    describe :to_s do
+
+        it 'returns the raw html' do
+           html.to_s.should == '<p>hello</p>'
+        end
+ 
+    end
+
+    describe :success? do
+
+        it 'returns whether the conversion succeeded' do
+           html.success?.should be_true 
+        end
+ 
+    end
+
+end