libxml-ruby 4.1.2 → 6.0.0

data/test/test_encoding.rb

@@ -1,129 +1,126 @@
-# encoding: UTF-8
-
-require_relative './test_helper'
-
-# Code  UTF8        Latin1      Hex
-# m      109          109        6D
-# ö      195 182      246        C3 B6 / F6
-# t      116          116        74
-# l      108          108        6C
-# e      101          101        65
-# y      121          121        79
-# _       95           95        5F
-# c       99           99        63
-# r      114          114        72
-# ü      195 188      252        C3 BC / FC
-# e      101          101        65
-
-# See:
-#  http://en.wikipedia.org/wiki/ISO/IEC_8859-1
-#  http://en.wikipedia.org/wiki/List_of_Unicode_characters
-
-class TestEncoding < Minitest::Test
-  def setup
-    @original_encoding = Encoding.default_internal
-    Encoding.default_internal = nil
-  end
-
-  def teardown
-    Encoding.default_internal = @original_encoding
-  end
-
-  def file_for_encoding(encoding)
-    file_name = "model/bands.#{encoding.name.downcase}.xml"
-    File.join(File.dirname(__FILE__), file_name)
-  end
-
-  def load_encoding(encoding)
-    @encoding = encoding
-    file = file_for_encoding(encoding)
-
-    # Strip spaces to make testing easier
-    LibXML::XML.default_keep_blanks = false
-    @doc = LibXML::XML::Document.file(file)
-    LibXML::XML.default_keep_blanks = true
-  end
-
-  def test_encoding
-    doc = LibXML::XML::Document.new
-    assert_equal(LibXML::XML::Encoding::NONE, doc.encoding)
-    assert_equal(Encoding::ASCII_8BIT, doc.rb_encoding)
-
-    file = File.expand_path(File.join(File.dirname(__FILE__), 'model/bands.xml'))
-    doc = LibXML::XML::Document.file(file)
-    assert_equal(LibXML::XML::Encoding::UTF_8, doc.encoding)
-    assert_equal(Encoding::UTF_8, doc.rb_encoding)
-
-    doc.encoding = LibXML::XML::Encoding::ISO_8859_1
-    assert_equal(LibXML::XML::Encoding::ISO_8859_1, doc.encoding)
-    assert_equal(Encoding::ISO8859_1, doc.rb_encoding)
-  end
-
-  def test_no_internal_encoding_iso_8859_1
-    load_encoding(Encoding::ISO_8859_1)
-    node = @doc.root.children.first
-
-    name = node.name
-    assert_equal(Encoding::UTF_8, name.encoding)
-    assert_equal("m\u00F6tley_cr\u00FCe", name)
-    assert_equal("109 195 182 116 108 101 121 95 99 114 195 188 101",
-                 name.bytes.to_a.join(" "))
-    assert_equal("M\u00F6tley Cr\u00FCe is an American heavy metal band formed in Los Angeles, California in 1981.",
-                 node.content)
-
-    name = name.encode(Encoding::ISO_8859_1)
-    assert_equal(Encoding::ISO_8859_1, name.encoding)
-    assert_equal("m\xF6tley_cr\xFCe".force_encoding(Encoding::ISO_8859_1), name)
-    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
-                 name.bytes.to_a.join(" "))
-    assert_equal("M\xF6tley Cr\xFCe is an American heavy metal band formed in Los Angeles, California in 1981.".force_encoding(Encoding::ISO_8859_1),
-                node.content.encode(Encoding::ISO_8859_1))
-  end
-
-  def test_internal_encoding_iso_8859_1
-    Encoding.default_internal = Encoding::ISO_8859_1
-    load_encoding(Encoding::ISO_8859_1)
-    node = @doc.root.children.first
-
-    name = node.name
-    assert_equal(Encoding::ISO_8859_1, name.encoding)
-    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
-                 name.bytes.to_a.join(" "))
-    assert_equal("m\xF6tley_cr\xFCe".force_encoding(Encoding::ISO_8859_1), name)
-    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
-                 name.bytes.to_a.join(" "))
-    assert_equal("M\xF6tley Cr\xFCe is an American heavy metal band formed in Los Angeles, California in 1981.".force_encoding(Encoding::ISO_8859_1),
-                node.content.encode(Encoding::ISO_8859_1))
-  end
-
-  def test_no_internal_encoding_utf_8
-    load_encoding(Encoding::UTF_8)
-    node = @doc.root.children.first
-
-    name = node.name
-    assert_equal(@encoding, name.encoding)
-    assert_equal("109 195 182 116 108 101 121 95 99 114 195 188 101",
-                 name.bytes.to_a.join(" "))
-
-    name = name.encode(Encoding::ISO_8859_1)
-    assert_equal(Encoding::ISO_8859_1, name.encoding)
-    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
-                 name.bytes.to_a.join(" "))
-  end
-
-  def test_internal_encoding_utf_8
-    Encoding.default_internal = Encoding::ISO_8859_1
-    load_encoding(Encoding::UTF_8)
-    node = @doc.root.children.first
-
-    name = node.name
-    assert_equal(Encoding::ISO_8859_1, name.encoding)
-    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
-                 name.bytes.to_a.join(" "))
-  end
-
-  def test_encoding_conversions
-    assert_equal("UTF-8", LibXML::XML::Encoding.to_s(LibXML::XML::Encoding::UTF_8))
-    assert_equal(LibXML::XML::Encoding::UTF_8, LibXML::XML::Encoding.from_s("UTF-8"))
-  end
-end
+# encoding: UTF-8
+
+require_relative './test_helper'
+
+# Code  UTF8        Latin1      Hex
+# m      109          109        6D
+# ö      195 182      246        C3 B6 / F6
+# t      116          116        74
+# l      108          108        6C
+# e      101          101        65
+# y      121          121        79
+# _       95           95        5F
+# c       99           99        63
+# r      114          114        72
+# ü      195 188      252        C3 BC / FC
+# e      101          101        65
+
+# See:
+#  http://en.wikipedia.org/wiki/ISO/IEC_8859-1
+#  http://en.wikipedia.org/wiki/List_of_Unicode_characters
+
+class TestEncoding < Minitest::Test
+  def setup
+    @original_encoding = Encoding.default_internal
+    Encoding.default_internal = nil
+  end
+
+  def teardown
+    Encoding.default_internal = @original_encoding
+  end
+
+  def file_for_encoding(encoding)
+    file_name = "model/bands.#{encoding.name.downcase}.xml"
+    File.join(File.dirname(__FILE__), file_name)
+  end
+
+  def load_encoding(encoding)
+    @encoding = encoding
+    file = file_for_encoding(encoding)
+
+    @doc = LibXML::XML::Document.file(file, options: LibXML::XML::Parser::Options::NOBLANKS)
+  end
+
+  def test_encoding
+    doc = LibXML::XML::Document.new
+    assert_equal(LibXML::XML::Encoding::NONE, doc.encoding)
+    assert_equal(Encoding::ASCII_8BIT, doc.rb_encoding)
+
+    file = File.expand_path(File.join(File.dirname(__FILE__), 'model/bands.xml'))
+    doc = LibXML::XML::Document.file(file)
+    assert_equal(LibXML::XML::Encoding::UTF_8, doc.encoding)
+    assert_equal(Encoding::UTF_8, doc.rb_encoding)
+
+    doc.encoding = LibXML::XML::Encoding::ISO_8859_1
+    assert_equal(LibXML::XML::Encoding::ISO_8859_1, doc.encoding)
+    assert_equal(Encoding::ISO8859_1, doc.rb_encoding)
+  end
+
+  def test_no_internal_encoding_iso_8859_1
+    load_encoding(Encoding::ISO_8859_1)
+    node = @doc.root.children.first
+
+    name = node.name
+    assert_equal(Encoding::UTF_8, name.encoding)
+    assert_equal("m\u00F6tley_cr\u00FCe", name)
+    assert_equal("109 195 182 116 108 101 121 95 99 114 195 188 101",
+                 name.bytes.to_a.join(" "))
+    assert_equal("M\u00F6tley Cr\u00FCe is an American heavy metal band formed in Los Angeles, California in 1981.",
+                 node.content)
+
+    name = name.encode(Encoding::ISO_8859_1)
+    assert_equal(Encoding::ISO_8859_1, name.encoding)
+    assert_equal("m\xF6tley_cr\xFCe".force_encoding(Encoding::ISO_8859_1), name)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+    assert_equal("M\xF6tley Cr\xFCe is an American heavy metal band formed in Los Angeles, California in 1981.".force_encoding(Encoding::ISO_8859_1),
+                node.content.encode(Encoding::ISO_8859_1))
+  end
+
+  def test_internal_encoding_iso_8859_1
+    Encoding.default_internal = Encoding::ISO_8859_1
+    load_encoding(Encoding::ISO_8859_1)
+    node = @doc.root.children.first
+
+    name = node.name
+    assert_equal(Encoding::ISO_8859_1, name.encoding)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+    assert_equal("m\xF6tley_cr\xFCe".force_encoding(Encoding::ISO_8859_1), name)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+    assert_equal("M\xF6tley Cr\xFCe is an American heavy metal band formed in Los Angeles, California in 1981.".force_encoding(Encoding::ISO_8859_1),
+                node.content.encode(Encoding::ISO_8859_1))
+  end
+
+  def test_no_internal_encoding_utf_8
+    load_encoding(Encoding::UTF_8)
+    node = @doc.root.children.first
+
+    name = node.name
+    assert_equal(@encoding, name.encoding)
+    assert_equal("109 195 182 116 108 101 121 95 99 114 195 188 101",
+                 name.bytes.to_a.join(" "))
+
+    name = name.encode(Encoding::ISO_8859_1)
+    assert_equal(Encoding::ISO_8859_1, name.encoding)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+  end
+
+  def test_internal_encoding_utf_8
+    Encoding.default_internal = Encoding::ISO_8859_1
+    load_encoding(Encoding::UTF_8)
+    node = @doc.root.children.first
+
+    name = node.name
+    assert_equal(Encoding::ISO_8859_1, name.encoding)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+  end
+
+  def test_encoding_conversions
+    assert_equal("UTF-8", LibXML::XML::Encoding.to_s(LibXML::XML::Encoding::UTF_8))
+    assert_equal(LibXML::XML::Encoding::UTF_8, LibXML::XML::Encoding.from_s("UTF-8"))
+  end
+end

data/test/test_encoding_sax.rb

@@ -1,115 +1,115 @@
-# encoding: UTF-8
-require_relative './test_helper'
-
-class SaxEncodingCallbacks
-  attr_reader :encoding
-  attr_accessor :assertions
-  include Minitest::Assertions
-
-  def initialize
-    @assertions = 0
-    @encoding = Encoding::UTF_8
-  end
-
-  # Called for a CDATA block event.
-  def on_cdata_block(cdata)
-    assert_equal(self.encoding, cdata.encoding)
-  end
-
-  # Called for a characters event.
-  def on_characters(chars)
-    assert_equal(self.encoding, chars.encoding)
-  end
-
-  # Called for a comment event.
-  def on_comment(msg)
-    assert_equal(self.encoding, msg.encoding)
-  end
-
-  # Called for a end document event.
-  def on_end_document
-  end
-
-  # Called for a end element event.
-  def on_end_element_ns(name, prefix, uri)
-    assert_equal(self.encoding, name.encoding)
-    assert_equal(self.encoding, prefix.encoding) if prefix
-    assert_equal(self.encoding, uri.encoding) if uri
-  end
-
-  # Called for parser errors.
-  def on_error(msg)
-    assert_equal(self.encoding, msg.encoding)
-  end
-
-  # Called for an external subset event.
-  def on_external_subset(name, external_id, system_id)
-    assert_equal(self.encoding, name.encoding)
-    assert_equal(self.encoding, external_id.encoding)
-    assert_equal(self.encoding, system_id.encoding)
-  end
-
-  # Called for an external subset notification event.
-  def on_has_external_subset
-  end
-
-  # Called for an internal subset notification event.
-  def on_has_internal_subset
-  end
-
-  # Called for an internal subset event.
-  def on_internal_subset(name, external_id, system_id)
-    assert_equal(self.encoding, name.encoding)
-    assert_equal(self.encoding, external_id.encoding)
-    assert_equal(self.encoding, system_id.encoding)
-  end
-
-  # Called for 'is standalone' event.
-  def on_is_standalone
-  end
-
-  # Called for an processing instruction event.
-  def on_processing_instruction(target, data)
-    assert_equal(self.encoding, target.encoding)
-    assert_equal(self.encoding, data.encoding)
-  end
-
-  # Called for a reference event.
-  def on_reference(name)
-    assert_equal(self.encoding, name.encoding)
-  end
-
-  # Called for a start document event.
-  def on_start_document
-  end
-
-  # Called for a start element event.
-  def on_start_element_ns(name, attributes, prefix, uri, namespaces)
-    assert_equal(self.encoding, name.encoding) if name
-    assert_equal(self.encoding, prefix.encoding) if prefix
-    assert_equal(self.encoding, uri.encoding) if uri
-  end
-end
-
-class TestEncodingSax < Minitest::Test
-  def setup
-    Encoding.default_internal = nil
-  end
-
-  def file_for_encoding(encoding)
-    file_name = "model/bands.#{encoding.name.downcase}.xml"
-    File.join(File.dirname(__FILE__), file_name)
-  end
-
-  def test_encoding_iso_8859_1
-    parser = LibXML::XML::SaxParser.file(file_for_encoding(Encoding::ISO_8859_1))
-    parser.callbacks = SaxEncodingCallbacks.new
-    parser.parse
-  end
-
-  def test_encoding_utf8
-    parser = LibXML::XML::SaxParser.file(file_for_encoding(Encoding::UTF_8))
-    parser.callbacks = SaxEncodingCallbacks.new
-    parser.parse
-  end
+# encoding: UTF-8
+require_relative './test_helper'
+
+class SaxEncodingCallbacks
+  attr_reader :encoding
+  attr_accessor :assertions
+  include Minitest::Assertions
+
+  def initialize
+    @assertions = 0
+    @encoding = Encoding::UTF_8
+  end
+
+  # Called for a CDATA block event.
+  def on_cdata_block(cdata)
+    assert_equal(self.encoding, cdata.encoding)
+  end
+
+  # Called for a characters event.
+  def on_characters(chars)
+    assert_equal(self.encoding, chars.encoding)
+  end
+
+  # Called for a comment event.
+  def on_comment(msg)
+    assert_equal(self.encoding, msg.encoding)
+  end
+
+  # Called for a end document event.
+  def on_end_document
+  end
+
+  # Called for a end element event.
+  def on_end_element_ns(name, prefix, uri)
+    assert_equal(self.encoding, name.encoding)
+    assert_equal(self.encoding, prefix.encoding) if prefix
+    assert_equal(self.encoding, uri.encoding) if uri
+  end
+
+  # Called for parser errors.
+  def on_error(msg)
+    assert_equal(self.encoding, msg.encoding)
+  end
+
+  # Called for an external subset event.
+  def on_external_subset(name, external_id, system_id)
+    assert_equal(self.encoding, name.encoding)
+    assert_equal(self.encoding, external_id.encoding)
+    assert_equal(self.encoding, system_id.encoding)
+  end
+
+  # Called for an external subset notification event.
+  def on_has_external_subset
+  end
+
+  # Called for an internal subset notification event.
+  def on_has_internal_subset
+  end
+
+  # Called for an internal subset event.
+  def on_internal_subset(name, external_id, system_id)
+    assert_equal(self.encoding, name.encoding)
+    assert_equal(self.encoding, external_id.encoding)
+    assert_equal(self.encoding, system_id.encoding)
+  end
+
+  # Called for 'is standalone' event.
+  def on_is_standalone
+  end
+
+  # Called for an processing instruction event.
+  def on_processing_instruction(target, data)
+    assert_equal(self.encoding, target.encoding)
+    assert_equal(self.encoding, data.encoding)
+  end
+
+  # Called for a reference event.
+  def on_reference(name)
+    assert_equal(self.encoding, name.encoding)
+  end
+
+  # Called for a start document event.
+  def on_start_document
+  end
+
+  # Called for a start element event.
+  def on_start_element_ns(name, attributes, prefix, uri, namespaces)
+    assert_equal(self.encoding, name.encoding) if name
+    assert_equal(self.encoding, prefix.encoding) if prefix
+    assert_equal(self.encoding, uri.encoding) if uri
+  end
+end
+
+class TestEncodingSax < Minitest::Test
+  def setup
+    Encoding.default_internal = nil
+  end
+
+  def file_for_encoding(encoding)
+    file_name = "model/bands.#{encoding.name.downcase}.xml"
+    File.join(File.dirname(__FILE__), file_name)
+  end
+
+  def test_encoding_iso_8859_1
+    parser = LibXML::XML::SaxParser.file(file_for_encoding(Encoding::ISO_8859_1))
+    parser.callbacks = SaxEncodingCallbacks.new
+    parser.parse
+  end
+
+  def test_encoding_utf8
+    parser = LibXML::XML::SaxParser.file(file_for_encoding(Encoding::UTF_8))
+    parser.callbacks = SaxEncodingCallbacks.new
+    parser.parse
+  end
 end