Make the tests portable:

This patch addresses two problems with tests
 - Various tests rely on Dir.glob ordering. This is not reliable; this patch
   introduces assert_directory_contains to avoid Dir.glob ordering
   inconsistencies.
 - test_ocr_extraction relies on exact text match from tesseract. However, this
   differs with each version of tesseract. This patch instead checks that all
   required txt files exist and that they have reasonable size.

Author: kremso

test/fixtures/corrosion/corrosion_1.txt

@@ -14,4 +14,13 @@ def teardown
     clear_output
   end
 
-end
+  def assert_directory_contains(dir, files)
+    files_in_directory = Dir["#{dir}/*"]
+    if files.kind_of?(Array)
+      assert files_in_directory.length == files.length, "Expected directory to contain exactly #{files.length} files"
+    else
+      files = [files]
+    end
+    files.each { |f| assert files_in_directory.include?(File.join(dir, f)), "Expected directory #{dir} to contain file #{f}, but it contains #{files_in_directory.inspect}" }
+  end
+end

test/fixtures/corrosion/corrosion_2.txt

@@ -4,7 +4,7 @@ class ExtractImagesTest < Test::Unit::TestCase
 
   def test_basic_image_extraction
     Docsplit.extract_images('test/fixtures/obama_arts.pdf', :format => :gif, :size => "250x", :output => OUTPUT)
-    assert Dir["#{OUTPUT}/*"] == ['test/output/obama_arts_1.gif', 'test/output/obama_arts_2.gif']
+    assert_directory_contains(OUTPUT, ['obama_arts_1.gif', 'obama_arts_2.gif'])
   end
 
   def test_image_formatting
@@ -36,14 +36,15 @@ def test_password_protected_extraction
 
   def test_repeated_extraction_in_the_same_directory
     Docsplit.extract_images('test/fixtures/obama_arts.pdf', :format => :gif, :size => "250x", :output => OUTPUT)
-    assert Dir["#{OUTPUT}/*"] == ['test/output/obama_arts_1.gif', 'test/output/obama_arts_2.gif']
+    assert_directory_contains(OUTPUT, ['obama_arts_1.gif', 'obama_arts_2.gif'])
     Docsplit.extract_images('test/fixtures/obama_arts.pdf', :format => :gif, :size => "250x", :output => OUTPUT)
-    assert Dir["#{OUTPUT}/*"] == ['test/output/obama_arts_1.gif', 'test/output/obama_arts_2.gif']
+    assert_directory_contains(OUTPUT, ['obama_arts_1.gif', 'obama_arts_2.gif'])
   end
 
   def test_name_escaping_while_extracting_images
     Docsplit.extract_images('test/fixtures/PDF file with spaces \'single\' and "double quotes".pdf', :format => :gif, :size => "250x", :output => OUTPUT)
-    assert Dir["#{OUTPUT}/*"] == ['test/output/PDF file with spaces \'single\' and "double quotes"_1.gif', 'test/output/PDF file with spaces \'single\' and "double quotes"_2.gif']
+    assert_directory_contains(OUTPUT, ['PDF file with spaces \'single\' and "double quotes"_1.gif',
+                                       'PDF file with spaces \'single\' and "double quotes"_1.gif'])
   end
 
 end

test/fixtures/corrosion/corrosion_3.txt

@@ -30,11 +30,10 @@ def test_unicode_extraction
 
   def test_ocr_extraction
     Docsplit.extract_text('test/fixtures/corrosion.pdf', :pages => 'all', :output => OUTPUT)
-    assert Dir["#{OUTPUT}/*.txt"].length == 4
     4.times do |i|
       file = "corrosion_#{i + 1}.txt"
-      # File.open("test/fixtures/corrosion/#{file}", "w+") {|f| f.write(File.read("#{OUTPUT}/#{file}")) }
-      assert File.read("#{OUTPUT}/#{file}") == File.read("test/fixtures/corrosion/#{file}")
+      assert_directory_contains(OUTPUT, file)
+      assert File.read(File.join(OUTPUT, file)).size > 1, "Expected that file with extracted text should have reasonable size"
     end
   end