Merge pull request tcocca#25 from elmatou/master

rules the binary content of damn pdfs

spec/active_pdftk/wrapper_spec.rb

@@ -10,7 +10,7 @@ def get_input(input_type, file_name = 'spec.fields.pdf')
   when :hash
     {path_to_pdf(file_name) => nil}
   when :file
-    File.new(path_to_pdf(file_name))
+    File.new(path_to_pdf(file_name), 'rb')
   when :tempfile
     t = Tempfile.new('input.spec')
     t.write(File.read(path_to_pdf(file_name)))
@@ -25,7 +25,7 @@ def get_output(output_type)
   when :path
     path_to_pdf('output.spec')
   when :file
-    File.new(path_to_pdf('output.spec'), 'w+')
+    File.new(path_to_pdf('output.spec'), 'wb+')
   when :tempfile
     Tempfile.new('output.spec')
   when :stringio
@@ -63,14 +63,17 @@ def map_output_type(output_specified)
     end
   end
 
-  shared_examples "a working command" do
+  shared_examples "a command" do
     it "should return a #{@output.nil? ? StringIO : @output.class}" do
       @call_output.should be_kind_of(@output.nil? ? StringIO : @output.class)
     end
 
     it "should return expected data" do
       if example.metadata[:genesis] && @output.is_a?(String)
         FileUtils.copy_entry(@output, @example_expect.to_s, true, false, true)
+      elsif example.metadata[:cleanup]
+        #cleanup_file_content!(File.open(@output, 'r:binary').read).should == cleanup_file_content!(File.open(@example_expect, 'r:binary').read) if @output.is_a?(String) # lets keep this line for debugging purpose.
+        @call_output.should look_like_the_same_pdf_as(@example_expect)
       else
         @call_output.should have_the_content_of(@example_expect)
       end
@@ -79,22 +82,6 @@ def map_output_type(output_specified)
     after(:each) { remove_output(@call_output) }
   end
 
-  shared_examples "a combination command" do
-    it "should return a #{@output.nil? ? StringIO : @output.class}" do
-      @call_output.should be_kind_of(@output.nil? ? StringIO : @output.class)
-    end
-
-    it "should return expected data" do
-      if example.metadata[:genesis] && @output.is_a?(String)
-        FileUtils.copy_entry(@output, @example_expect.to_s, true, false, true)
-      else
-        @call_output.should look_like_the_same_pdf_as(@example_expect)
-      end
-    end
-
-    after(:each) { remove_output(@call_output) }
-  end
-
   inputs.each do |input_type|
     outputs.each do |output_type|
 
@@ -106,39 +93,39 @@ def map_output_type(output_specified)
         end
 
         describe "#dump_data_fields" do
-          it_behaves_like "a working command" do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('dump_data_fields/expect.data_fields') }
             before(:each) { @call_output = @pdftk.dump_data_fields(@input, :output => @output) }
           end
         end
 
         describe "#fill_form" do
-          it_behaves_like "a working command" do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('fill_form/expect.pdf') }
             before(:each) { @call_output = @pdftk.fill_form(@input, path_to_pdf('fill_form/spec.fdf'), :output => @output) }
           end
-          it_behaves_like "a working command" do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('fill_form/expect.pdf') }
             before(:each) { @call_output = @pdftk.fill_form(@input, path_to_pdf('fill_form/spec.xfdf'), :output => @output) }
           end
         end
 
         describe "#generate_fdf" do
-          it_behaves_like "a working command" do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('generate_fdf/expect.fdf') }
             before(:each) { @call_output = @pdftk.generate_fdf(@input,:output => @output) }
           end
         end
 
         describe "#dump_data" do
-          it_behaves_like "a working command" do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('dump_data/expect.data') }
             before(:each) { @call_output = @pdftk.dump_data(@input,:output => @output) }
           end
         end
 
         describe "#update_info" do
-          it_behaves_like "a working command" do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('update_info/expect.pdf') }
             before(:each) { @call_output = @pdftk.update_info(@input, path_to_pdf('update_info/spec.data'), :output => @output) }
           end
@@ -200,43 +187,43 @@ def map_output_type(output_specified)
           end
         end
 
-        describe "#background" do
-          it_behaves_like "a working command" do
+        describe "#background", :cleanup => true do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('background/expect.pdf') }
             before(:each) { @call_output = @pdftk.background(get_input(input_type, 'multi.pdf'), path_to_pdf('poly.pdf'), :output => @output) }
           end
         end
 
-        describe "#multibackground" do
-          it_behaves_like "a working command" do
+        describe "#multibackground", :cleanup => true do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('multibackground/expect.pdf') }
             before(:each) { @call_output = @pdftk.multibackground(get_input(input_type, 'multi.pdf'), path_to_pdf('poly.pdf'), :output => @output) }
           end
         end
 
-        describe "#stamp" do
-          it_behaves_like "a working command" do
+        describe "#stamp", :cleanup => true do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('stamp/expect.pdf') }
             before(:each) { @call_output = @pdftk.stamp(get_input(input_type, 'multi.pdf'), path_to_pdf('poly.pdf'), :output => @output) }
           end
         end
 
-        describe "#multistamp" do
-          it_behaves_like "a working command" do
+        describe "#multistamp", :cleanup => true do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('multistamp/expect.pdf') }
             before(:each) { @call_output = @pdftk.multistamp(get_input(input_type, 'multi.pdf'), path_to_pdf('poly.pdf'), :output => @output) }
           end
         end
 
-        describe "#cat" do
-          it_behaves_like "a combination command" do
+        describe "#cat", :cleanup => true do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('cat/expect.pdf')}
             before(:each) { @call_output = @pdftk.cat([{:pdf => path_to_pdf('multi.pdf')}, {:pdf => path_to_pdf('poly.pdf'), :start => 1, :end => 'end', :orientation => 'N', :pages => 'even'}], :output => @output) }
           end
         end
 
-        describe "#shuffle" do
-          it_behaves_like "a combination command" do
+        describe "#shuffle", :cleanup => true do
+          it_behaves_like "a command" do
             before(:all) { @example_expect = fixtures_path('shuffle/expect.pdf')}
             before(:each) { @call_output = @pdftk.shuffle([{:pdf => path_to_pdf('multi.pdf')}, {:pdf => path_to_pdf('poly.pdf'), :start => 1, :end => 'end', :orientation => 'N', :pages => 'even'}], :output => @output) }
           end

spec/support/inputs_helper.rb

@@ -59,9 +59,21 @@ def open_or_rewind(target)
 end
 
 def cleanup_file_content!(text)
+
+  unless @filter
+    @filter = {
+        :date => /\(D\:.*\)/,                   # Remove dates ex: /CreationDate (D:20111106104455-05'00')
+        :ids => /\/ID \[<\w*><\w*>\]/,          # Remove ID values ex: /ID [<4ba02a4cf55b1fc842299e6f01eb838e><33bec7dc37839cadf7ab76f3be4d4306>]
+        :stream => /stream .* 9|10 0 obj /m,    # Remove some binary stream
+        :content => /\/Contents \[.*\]/,
+        :xref => /^\d{10} \d{5} n|f $/          # Remove Cross-references dictionnary
+    }
+    @filter.each {|k,reg| @filter[k] = Regexp.new(reg.source.encode('ASCII-8BIT'), reg.options) if reg.source.respond_to? :encode }
+  end
+
+
   text.force_encoding('ASCII-8BIT') if text.respond_to? :force_encoding   # PDF embed some binary data breaking gsub with ruby 1.9.2
-  text.gsub!(/\(D\:.*\)/, '')                                             # Remove dates ex: /CreationDate (D:20111106104455-05'00')
-  text.gsub!(/\/ID \[<\w*><\w*>\]/, '')                                   # Remove ID values ex: /ID [<4ba02a4cf55b1fc842299e6f01eb838e><33bec7dc37839cadf7ab76f3be4d4306>]
+  @filter.each {|k,reg| text.gsub!(reg, '')}
   text
 end
 

spec/support/matchers/content_matcher.rb

@@ -59,7 +59,7 @@ def sha256_hash_of_almost(entry)
       if entry.directory?
         sha256_hash_of_almost(Dir.new(entry))
       elsif entry.file?
-        sha256_hash_of_almost(File.read(entry))
+        sha256_hash_of_almost(File.open(entry, 'r:binary').read)
       end
   end
 end