milkfarm-onix 0.7.7

data/spec/normaliser_spec.rb

@@ -0,0 +1,125 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::Normaliser", "with a simple short tag file" do
+
+  before(:each) do
+    @data_path = File.join(File.dirname(__FILE__),"..","data")
+    @filename  = File.join(@data_path, "short_tags.xml")
+    @outfile   = @filename + ".new"
+  end
+
+  after(:each) do
+    File.unlink(@outfile) if File.file?(@outfile)
+  end
+
+  specify "should correctly convert short tag file to reference tag" do
+    ONIX::Normaliser.process(@filename, @outfile)
+
+    File.file?(@outfile).should be_true
+    content = File.read(@outfile)
+    content.include?("<m174>").should be_false
+    content.include?("<FromCompany>").should be_true
+  end
+
+end
+
+context "ONIX::Normaliser", "with an ISO-8859-1 file" do
+
+  before(:each) do
+    @data_path = File.join(File.dirname(__FILE__),"..","data")
+    @filename  = File.join(@data_path, "iso_8859_1.xml")
+    @outfile   = @filename + ".new"
+  end
+
+  after(:each) do
+    File.unlink(@outfile) if File.file?(@outfile)
+  end
+
+  specify "should correctly convert an iso-8859-1 file to UTF-8" do
+    ONIX::Normaliser.process(@filename, @outfile)
+
+    File.file?(@outfile).should be_true
+    content = File.read(@outfile)
+
+    content.include?("ISO-8859-1").should be_false
+    content.include?("UTF-8").should be_true
+
+    `isutf8 #{File.expand_path(@outfile)}`.strip.should eql("")
+  end
+
+end
+
+context "ONIX::Normaliser", "with an file using entities" do
+
+  before(:each) do
+    @data_path = File.join(File.dirname(__FILE__),"..","data")
+    @filename  = File.join(@data_path, "entities.xml")
+    @outfile   = @filename + ".new"
+  end
+
+  after(:each) do
+    File.unlink(@outfile) if File.file?(@outfile)
+  end
+
+  specify "should correctly convert named entities to numeric entities" do
+    ONIX::Normaliser.process(@filename, @outfile)
+
+    File.file?(@outfile).should be_true
+    content = File.read(@outfile)
+
+    content.include?("&ndash;").should be_false
+    content.include?("&#x02013;").should be_true
+  end
+end
+
+context "ONIX::Normaliser", "with a utf8 file that has no declared encoding" do
+
+  before(:each) do
+    @data_path = File.join(File.dirname(__FILE__),"..","data")
+    @filename  = File.join(@data_path, "no_encoding.xml")
+    @outfile   = @filename + ".new"
+  end
+
+  after(:each) do
+    File.unlink(@outfile) if File.file?(@outfile)
+  end
+
+  # this is to test for a bug where an exception was raised on files that
+  # had no declared encoding
+  specify "should add a utf-8 marker to the file" do
+    ONIX::Normaliser.process(@filename, @outfile)
+
+    File.file?(@outfile).should be_true
+    content = File.read(@outfile)
+
+    content.include?("encoding=\"UTF-8\"").should be_true
+  end
+end
+
+context "ONIX::Normaliser", "with a utf8 file that has illegal control chars" do
+
+  before(:each) do
+    @data_path = File.join(File.dirname(__FILE__),"..","data")
+    @filename  = File.join(@data_path, "control_chars.xml")
+    @outfile   = @filename + ".new"
+  end
+
+  after(:each) do
+    File.unlink(@outfile) if File.file?(@outfile)
+  end
+
+  # this is to test for a bug where an exception was raised on files that
+  # had no declared encoding
+  specify "should remove all control chars except LF, CR and TAB" do
+    ONIX::Normaliser.process(@filename, @outfile)
+
+    File.file?(@outfile).should be_true
+    content = File.read(@outfile)
+
+    content.include?("<TitleText>OXFORDPICTURE DICTIONARY CHINESE</TitleText>").should be_true
+  end
+end

data/spec/other_text_spec.rb

@@ -0,0 +1,40 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::OtherText" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(data_path, "other_text.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @root = @doc.root
+  end
+
+  specify "should correctly convert to a string" do
+    ot = ONIX::OtherText.from_xml(@root.to_s)
+    ot.to_xml.to_s[0,11].should eql("<OtherText>")
+  end
+
+  specify "should provide read access to first level attibutes" do
+    ot = ONIX::OtherText.from_xml(@root.to_s)
+
+    ot.text_type_code.should eql(2)
+    ot.text[0,7].should eql("A woman")
+  end
+
+  specify "should provide write access to first level attibutes" do
+    ot = ONIX::OtherText.new
+
+    ot.text_type_code = 2
+    ot.to_xml.to_s.include?("<TextTypeCode>02</TextTypeCode>").should be_true
+
+    ot.text = "James Healy"
+    ot.to_xml.to_s.include?("<Text>James Healy</Text>").should be_true
+
+  end
+
+end
+

data/spec/price_spec.rb

@@ -0,0 +1,40 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::Price" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(data_path, "price.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @root = @doc.root
+  end
+
+  specify "should correctly convert to a string" do
+    p = ONIX::Price.from_xml(@root.to_s)
+    p.to_xml.to_s[0,7].should eql("<Price>")
+  end
+
+  specify "should provide read access to first level attibutes" do
+    p = ONIX::Price.from_xml(@root.to_s)
+
+    p.price_type_code.should eql(2)
+    p.price_amount.should eql(BigDecimal.new("7.5"))
+  end
+
+  specify "should provide write access to first level attibutes" do
+    p = ONIX::Price.new
+
+    p.price_type_code = 1
+    p.to_xml.to_s.include?("<PriceTypeCode>01</PriceTypeCode>").should be_true
+
+    p.price_amount = BigDecimal.new("7.5")
+    p.to_xml.to_s.include?("<PriceAmount>7.5</PriceAmount>").should be_true
+
+  end
+
+end
+

data/spec/product_identifier_spec.rb

@@ -0,0 +1,40 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::ProductIdentifier" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(data_path, "product_identifier.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @root = @doc.root
+  end
+
+  specify "should correctly convert to a string" do
+    id = ONIX::ProductIdentifier.from_xml(@root.to_s)
+    id.to_xml.to_s[0,19].should eql("<ProductIdentifier>")
+  end
+
+  specify "should provide read access to first level attibutes" do
+    id = ONIX::ProductIdentifier.from_xml(@root.to_s)
+
+    id.product_id_type.should eql(2)
+    id.id_value.should eql("0858198363")
+  end
+
+  specify "should provide write access to first level attibutes" do
+    id = ONIX::ProductIdentifier.new
+
+    id.product_id_type = 2
+    id.to_xml.to_s.include?("<ProductIDType>02</ProductIDType>").should be_true
+
+    id.id_value = "James"
+    id.to_xml.to_s.include?("<IDValue>James</IDValue>").should be_true
+
+  end
+
+end
+

data/spec/product_spec.rb

@@ -0,0 +1,96 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+require 'date'
+
+context "ONIX::Product" do
+
+  before(:each) do
+    @data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(@data_path, "product.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @product_node = @doc.root
+  end
+
+  specify "should provide read access to first level attibutes" do
+    product = ONIX::Product.from_xml(@product_node.to_s)
+
+    product.record_reference.should eql("365-9780194351898")
+    product.notification_type.should eql(3)
+    product.product_form.should eql("BC")
+    product.edition_number.should eql(1)
+    product.number_of_pages.should eql(100)
+    product.bic_main_subject.should eql("EB")
+    product.publishing_status.should eql(4)
+    product.publication_date.should eql(Date.civil(1998,9,1))
+    product.year_first_published.should eql(1998)
+
+    # including ye olde, deprecated ones
+    product.height.should eql(100)
+    product.width.should eql(BigDecimal.new("200.5"))
+    product.weight.should eql(300)
+    product.thickness.should eql(300)
+    product.dimensions.should eql("100x200")
+  end
+
+  specify "should provide read access to product IDs" do
+    product = ONIX::Product.from_xml(@product_node.to_s)
+    product.product_identifiers.size.should eql(3)
+  end
+
+  specify "should provide read access to titles" do
+    product = ONIX::Product.from_xml(@product_node.to_s)
+    product.titles.size.should eql(1)
+  end
+
+  specify "should provide read access to subjects" do
+    product = ONIX::Product.from_xml(@product_node.to_s)
+    product.subjects.size.should eql(1)
+  end
+
+  specify "should provide read access to measurements" do
+    product = ONIX::Product.from_xml(@product_node.to_s)
+    product.measurements.size.should eql(1)
+  end
+
+  specify "should provide write access to first level attibutes" do
+    product = ONIX::Product.new
+
+    product.notification_type = 3
+    product.to_xml.to_s.include?("<NotificationType>03</NotificationType>").should be_true
+
+    product.record_reference = "365-9780194351898"
+    product.to_xml.to_s.include?("<RecordReference>365-9780194351898</RecordReference>").should be_true
+
+    product.product_form = "BC"
+    product.to_xml.to_s.include?("<ProductForm>BC</ProductForm>").should be_true
+
+    product.edition_number = 1
+    product.to_xml.to_s.include?("<EditionNumber>1</EditionNumber>").should be_true
+
+    product.number_of_pages = 100
+    product.to_xml.to_s.include?("<NumberOfPages>100</NumberOfPages>").should be_true
+
+    product.bic_main_subject = "EB"
+    product.to_xml.to_s.include?("<BICMainSubject>EB</BICMainSubject>").should be_true
+
+    product.publishing_status = 4
+    product.to_xml.to_s.include?("<PublishingStatus>04</PublishingStatus>").should be_true
+
+    product.publication_date = Date.civil(1998,9,1)
+    product.to_xml.to_s.include?("<PublicationDate>19980901</PublicationDate>").should be_true
+
+    product.year_first_published = 1998
+    product.to_xml.to_s.include?("<YearFirstPublished>1998</YearFirstPublished>").should be_true
+  end
+
+  specify "should correctly from_xml files that have an invalid publication date" do
+    file = File.join(@data_path, "product_invalid_pubdate.xml")
+    product = ONIX::Product.from_xml(File.read(file))
+
+    product.bic_main_subject.should eql("VXFC1")
+    product.publication_date.should be_nil
+  end
+end

data/spec/publisher_spec.rb

@@ -0,0 +1,38 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::Publisher" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(data_path, "publisher.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @root = @doc.root
+  end
+
+  specify "should correctly convert to a string" do
+    pub = ONIX::Publisher.from_xml(@root.to_s)
+    pub.to_xml.to_s[0,11].should eql("<Publisher>")
+  end
+
+  specify "should provide read access to first level attibutes" do
+    pub = ONIX::Publisher.from_xml(@root.to_s)
+    pub.publishing_role.should eql(1)
+    pub.publisher_name.should eql("Desbooks Publishing")
+  end
+
+  specify "should provide write access to first level attibutes" do
+    pub = ONIX::Publisher.new
+
+    pub.publisher_name = "Paulist Press"
+    pub.to_xml.to_s.include?("<PublisherName>Paulist Press</PublisherName>").should be_true
+
+    pub.publishing_role = 2
+    pub.to_xml.to_s.include?("<PublishingRole>02</PublishingRole>").should be_true
+  end
+
+end
+

data/spec/reader_spec.rb

@@ -0,0 +1,131 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::Reader" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    @file1    = File.join(data_path, "9780194351898.xml")
+    @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")
+  end
+
+  specify "should initialize with a filename" do
+    reader = ONIX::Reader.new(@file1)
+    reader.instance_variable_get("@reader").should be_a_kind_of(LibXML::XML::Reader)
+  end
+
+  specify "should initialize with an IO object" do
+    File.open(@file1,"rb") do |f|
+      reader = ONIX::Reader.new(f)
+      reader.instance_variable_get("@reader").should be_a_kind_of(LibXML::XML::Reader)
+    end
+  end
+
+  # This is commented out as the code that I was using to read the ONIX version from the
+  # input was causing segfaults and other stability issues
+  specify "should provide access to various XML metadata from file"
+  #do
+  #  reader = ONIX::Reader.new(@file1)
+  #  reader.encoding.should eql("utf-8")
+  #  reader.xml_lang.should eql(nil)
+  #  reader.xml_version.should eql(1.0)
+  #  reader.version.should eql([2,1,0])
+  #end
+
+  specify "should provide access to the header in an ONIX file" do
+    reader = ONIX::Reader.new(@file1)
+    reader.header.should be_a_kind_of(ONIX::Header)
+  end
+
+  specify "should iterate over all product records in an ONIX file" do
+    reader = ONIX::Reader.new(@file1)
+    counter = 0
+    reader.each do |product|
+      product.should be_a_kind_of(ONIX::Product)
+      counter += 1
+    end
+
+    counter.should eql(1)
+  end
+
+  specify "should iterate over all product records in an ONIX file" do
+    reader = ONIX::Reader.new(@file2)
+    products = []
+    reader.each do |product|
+      products << product
+    end
+
+    products.size.should eql(2)
+    products[0].record_reference.should eql("365-9780194351898")
+    products[1].record_reference.should eql("9780754672326")
+  end
+
+  # libxml can handle the 3 standard entities fine (&amp; &lt; and ^gt;) but
+  # barfs when it encounters others. In theory other entities are defined in the
+  # ONIX DTD, but I can't work out how to get libxml to recognise them
+  specify "should correctly parse a file that has an entity in it" do
+    reader = ONIX::Reader.new(@entity_file)
+
+    products = []
+    reader.each do |product|
+      products << product
+    end
+
+    products.size.should eql(1)
+    products.first.titles.size.should eql(1)
+    products.first.titles.first.title_text.should eql("High Noon\342\200\223in Nimbin")
+    products.first.record_reference.should eql("9780732287573")
+  end
+
+  # for some reason I'm getting segfaults when I read a file with more than 7 records
+  specify "should correctly parse a file with more than 7 records in in" do
+    reader = ONIX::Reader.new(@long_file)
+    counter = 0
+    reader.each do |product|
+      counter += 1
+    end
+
+    counter.should eql(346)
+  end
+
+  specify "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
+
+    # 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)
+    end
+
+    product.contributors[0].person_name_inverted.should eql("Küng, Hans")
+
+  end
+
+  specify "should transparently convert a utf-16 file to utf-8" do
+    reader = ONIX::Reader.new(@utf_16_file)
+    product = nil
+    reader.each do |p|
+      product = p
+    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)
+    end
+
+    product.contributors[0].person_name_inverted.should eql("Küng, Hans")
+
+  end
+end

data/spec/sales_restriction_spec.rb

@@ -0,0 +1,35 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::SalesRestriction" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(data_path, "sales_restriction.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @root = @doc.root
+  end
+
+  specify "should correctly convert to a string" do
+    sr = ONIX::SalesRestriction.from_xml(@root.to_s)
+    sr.to_xml.to_s[0,18].should eql("<SalesRestriction>")
+  end
+
+  specify "should provide read access to first level attibutes" do
+    sr = ONIX::SalesRestriction.from_xml(@root.to_s)
+
+    sr.sales_restriction_type.should eql(0)
+  end
+
+  specify "should provide write access to first level attibutes" do
+    sr = ONIX::SalesRestriction.new
+
+    sr.sales_restriction_type = 1
+    sr.to_xml.to_s.include?("<SalesRestrictionType>01</SalesRestrictionType>").should be_true
+  end
+
+end
+

data/spec/sender_identifier.rb

@@ -0,0 +1,40 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::SenderIdentifier" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(data_path, "sender_identifier.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @root = @doc.root
+  end
+
+  specify "should correctly convert to a string" do
+    id = ONIX::SenderIdentifier.from_xml(@root.to_s)
+    id.to_xml.to_s[0,18].should eql("<SenderIdentifier>")
+  end
+
+  specify "should provide read access to first level attibutes" do
+    id = ONIX::SenderIdentifier.from_xml(@root.to_s)
+
+    id.sender_id_type.should eql(1)
+    id.id_value.should eql("123456")
+  end
+
+  specify "should provide write access to first level attibutes" do
+    id = ONIX::SenderIdentifier.new
+
+    id.sender_id_type = 1
+    id.to_xml.to_s.include?("<SenderIDType>01</SenderIDType>").should be_true
+
+    id.id_value = "54321"
+    id.to_xml.to_s.include?("<IDValue>54321</IDValue>").should be_true
+
+  end
+
+end
+

data/spec/series_identifier_spec.rb

@@ -0,0 +1,38 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::SeriesIdentifier" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(data_path, "series_identifier.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @root = @doc.root
+  end
+
+  specify "should correctly convert to a string" do
+    series = ONIX::SeriesIdentifier.from_xml(@root.to_s)
+    series.to_xml.to_s[0,18].should eql("<SeriesIdentifier>")
+  end
+
+  specify "should provide read access to first level attibutes" do
+    series = ONIX::SeriesIdentifier.from_xml(@root.to_s)
+    
+    series.series_id_type.should eql(1)
+    series.id_value.should eql("10001")
+  end
+
+  specify "should provide write access to first level attibutes" do
+    series = ONIX::SeriesIdentifier.new
+    
+    series.series_id_type = 9
+    series.to_xml.to_s.include?("<SeriesIDType>09</SeriesIDType>").should be_true
+    
+    series.id_value = 999
+    series.to_xml.to_s.include?("<IDValue>999</IDValue>").should be_true
+  end
+
+end

data/spec/series_spec.rb

@@ -0,0 +1,34 @@
+# coding: utf-8
+
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+
+require 'onix'
+
+context "ONIX::Series" do
+
+  before(:each) do
+    data_path = File.join(File.dirname(__FILE__),"..","data")
+    file1    = File.join(data_path, "series.xml")
+    @doc = LibXML::XML::Document.file(file1)
+    @root = @doc.root
+  end
+
+  specify "should correctly convert to a string" do
+    series = ONIX::Series.from_xml(@root.to_s)
+    series.to_xml.to_s[0,8].should eql("<Series>")
+  end
+
+  specify "should provide read access to first level attibutes" do
+    series = ONIX::Series.from_xml(@root.to_s)
+    
+    series.title_of_series.should eql("Citizens and Their Governments")
+  end
+
+  specify "should provide write access to first level attibutes" do
+    series = ONIX::Series.new
+    
+    series.title_of_series = "Cool Science Careers"
+    series.to_xml.to_s.include?("<TitleOfSeries>Cool Science Careers</TitleOfSeries>").should be_true
+  end
+
+end