reader_spec.rb in milkfarm-onix-0.8.12

- old
+ new

@@ -10,10 +10,11 @@
     @file2    = File.join(@data_path, "two_products.xml")
     @long_file   = File.join(@data_path, "Bookwise_July_2008.xml")
     @entity_file = File.join(@data_path, "entities.xml")
     @utf_16_file = File.join(@data_path, "utf_16.xml")
     @iso_8859_1_file = File.join(@data_path, "iso_8859_1.xml")
+    @no_encoding_decl_file = File.join(@data_path, "aau.xml")
   end
 
   it "should initialize with a filename" do
     reader = ONIX::Reader.new(@file1)
     reader.instance_variable_get("@reader").should be_a_kind_of(Nokogiri::XML::Reader)
@@ -88,23 +89,40 @@
     counter.should eql(346)
   end
 
   it "should transparently convert a iso-8859-1 file to utf-8" do
     reader = ONIX::Reader.new(@iso_8859_1_file)
-    product = nil
-    reader.each do |p|
-      product = p
-    end
+    reader.each do |product|
+      if RUBY_VERSION >= "1.9"
+        utf8 = Encoding.find("utf-8")
+        product.contributors[0].person_name_inverted.encoding.should eql(utf8)
+      end
 
-    # ROXML appears to munge the string encodings
-    if RUBY_VERSION >= "1.9"
-      utf8 = Encoding.find("utf-8")
-      product.contributors[0].person_name_inverted.encoding.should eql(utf8)
+      product.contributors[0].person_name_inverted.should eql("Küng, Hans")
     end
+  end
 
-    product.contributors[0].person_name_inverted.should eql("Küng, Hans")
+  # This isn't ideal behaviour, but i'm somewhat hamstrung by nokogiri API. It'd
+  # be nice to have the option to replace unrecognised bytes with a valid char.
+  it "should raise an exception when an iso-8859-1 file isn't declared as such" do
+    reader = ONIX::Reader.new(@no_encoding_decl_file)
+    lambda {
+      reader.each do |product|
+      end
+    }.should raise_error(Nokogiri::XML::SyntaxError)
+  end
 
+  it "should transparently convert an iso-8859-1 file to utf-8 when there's no declaration but the user manually specifies iso-8859-1" do
+    reader = ONIX::Reader.new(@no_encoding_decl_file, :encoding => "iso-8859-1")
+    reader.each do |product|
+      if RUBY_VERSION >= "1.9"
+        utf8 = Encoding.find("utf-8")
+        product.contributors[0].person_name_inverted.encoding.should eql(utf8)
+      end
+
+      product.contributors[0].person_name_inverted.should eql("Melo,Patr¡cia")
+    end
   end
 
   it "should transparently convert a utf-16 file to utf-8" do
     reader = ONIX::Reader.new(@utf_16_file)
     product = nil
@@ -117,8 +135,21 @@
       utf8 = Encoding.find("utf-8")
       product.contributors[0].person_name_inverted.encoding.should eql(utf8)
     end
 
     product.contributors[0].person_name_inverted.should eql("Küng, Hans")
+  end
 
+  it "should support returning an APAProduct using deprecated API" do
+    reader = ONIX::Reader.new(@file1, ONIX::APAProduct)
+    reader.each do |product|
+      product.should be_a_kind_of(ONIX::APAProduct)
+    end
+  end
+
+  it "should support returning an APAProduct using new API" do
+    reader = ONIX::Reader.new(@file1, :product_class => ONIX::APAProduct)
+    reader.each do |product|
+      product.should be_a_kind_of(ONIX::APAProduct)
+    end
   end
 end