public void GetTokens_HtmlContent_ReturnsSequence()
{
const string input = @"<div class=""content""><p style=""color:red""><strong>Lorem ipsum</strong> dolor sit amet, <i>consectetur</i> adipiscing elit.</p></div>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("div")),
MarkupGrammar.TokenAttribute(new DataName("class")),
MarkupGrammar.TokenPrimitive("content"),
MarkupGrammar.TokenElementBegin(new DataName("p")),
MarkupGrammar.TokenAttribute(new DataName("style")),
MarkupGrammar.TokenPrimitive("color:red"),
MarkupGrammar.TokenElementBegin(new DataName("strong")),
MarkupGrammar.TokenPrimitive("Lorem ipsum"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive(" dolor sit amet, "),
MarkupGrammar.TokenElementBegin(new DataName("i")),
MarkupGrammar.TokenPrimitive("consectetur"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive(" adipiscing elit."),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd,
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_MixedEntities_ReturnsSequence()
{
const string input = @"there should <b>e decoded chars & inside this text";
var expected = new[]
{
MarkupGrammar.TokenPrimitive(@"there should <b>e decoded chars & inside this text")
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_EntityWithTrailingText_ReturnsSequence()
{
const string input = @"&trailing";
var expected = new[]
{
MarkupGrammar.TokenPrimitive("&trailing")
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_HtmlEntityEuro_ReturnsSequence()
{
const string input = @"€";
var expected = new[]
{
MarkupGrammar.TokenPrimitive("€")
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_XmlEntityHexLowerCase_ReturnsSequence()
{
const string input = @"ꯍ";
var expected = new[]
{
MarkupGrammar.TokenPrimitive("\uabcd")
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_XmlDeclaration_ReturnsUnparsed()
{
const string input = @"<?xml version=""1.0""?>";
var expected = new[]
{
MarkupGrammar.TokenUnparsed("?", "?", @"xml version=""1.0""")
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_XmlComment_ReturnsUnparsed()
{
const string input = @"<!-- a quick note -->";
var expected = new[]
{
MarkupGrammar.TokenUnparsed("!--", "--", @" a quick note ")
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_XmlCData_ReturnsTextValue()
{
const string input = @"<![CDATA[value>""0"" && value<""10"" ?""valid"":""error""]]>";
var expected = new[]
{
MarkupGrammar.TokenPrimitive(@"value>""0"" && value<""10"" ?""valid"":""error""")
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_DifferentPrefixSameNamespace_ReturnsSequence()
{
const string input = @"<foo xmlns=""http://example.org"" xmlns:blah=""http://example.org"" blah:key=""value"" />";
var expected = new[]
{
MarkupGrammar.TokenElementVoid(new DataName("foo", String.Empty, "http://example.org")),
MarkupGrammar.TokenAttribute(new DataName("key", "blah", "http://example.org")),
MarkupGrammar.TokenPrimitive("value")
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_SingleAttributeWhitespace_ReturnsSequence()
{
const string input = @"<root whitespace="" this contains whitespace ""></root>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("root")),
MarkupGrammar.TokenAttribute(new DataName("whitespace")),
MarkupGrammar.TokenPrimitive(" this contains whitespace "),
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_AttributeWhitespaceQuotDelims_ReturnsSequence()
{
const string input = @"<root white = "" extra whitespace around quote delims "" ></root>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("root")),
MarkupGrammar.TokenAttribute(new DataName("white")),
MarkupGrammar.TokenPrimitive(" extra whitespace around quote delims "),
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_SingleAttributeEmptyValue_ReturnsSequence()
{
const string input = @"<root emptyValue=""""></root>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("root")),
MarkupGrammar.TokenAttribute(new DataName("emptyValue")),
MarkupGrammar.TokenPrimitive(String.Empty),
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_ParentAndChildSharePrefixedNamespace_ReturnsSequence()
{
const string input = @"<bar:foo xmlns:bar=""http://example.org""><bar:child>value</bar:child></bar:foo>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("foo", "bar", "http://example.org")),
MarkupGrammar.TokenElementBegin(new DataName("child", "bar", "http://example.org")),
MarkupGrammar.TokenPrimitive("value"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_ParentAndChildDifferentDefaultNamespaces_ReturnsSequence()
{
const string input = @"<foo xmlns=""http://json.org""><child xmlns=""http://jsonfx.net"">text value</child></foo>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("foo", String.Empty, "http://json.org")),
MarkupGrammar.TokenElementBegin(new DataName("child", String.Empty, "http://jsonfx.net")),
MarkupGrammar.TokenPrimitive("text value"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_NamespacedChildTag_ReturnsSequence()
{
const string input = @"<foo><child xmlns=""http://example.com/schema"">value</child></foo>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("foo")),
MarkupGrammar.TokenElementBegin(new DataName("child", String.Empty, "http://example.com/schema")),
MarkupGrammar.TokenPrimitive("value"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_MultipleAttributes_ReturnsSequence()
{
const string input = @"<root no-value="""" whitespace="" this contains whitespace "" anyQuotedText=""" + "/\\\uCAFE\uBABE\uAB98\uFCDE\ubcda\uef4A\n\r\t`1~!@#$%^&*()_+-=[]{}|;:',./<>?" + @"""></root>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("root")),
MarkupGrammar.TokenAttribute(new DataName("anyQuotedText")),
MarkupGrammar.TokenPrimitive("/\\\uCAFE\uBABE\uAB98\uFCDE\uBCDA\uEF4A `1~!@#$%^&*()_+-=[]{}|;:',./<>?"),
MarkupGrammar.TokenAttribute(new DataName("no-value")),
MarkupGrammar.TokenPrimitive(String.Empty),
MarkupGrammar.TokenAttribute(new DataName("whitespace")),
MarkupGrammar.TokenPrimitive(" this contains whitespace "),
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_NestedDefaultNamespaces_ReturnsSequence()
{
const string input = @"<outer xmlns=""http://example.org/outer""><middle-1 xmlns=""http://example.org/inner""><inner>this should be inner</inner></middle-1><middle-2>this should be outer</middle-2></outer>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("outer", String.Empty, "http://example.org/outer")),
MarkupGrammar.TokenElementBegin(new DataName("middle-1", String.Empty, "http://example.org/inner")),
MarkupGrammar.TokenElementBegin(new DataName("inner", String.Empty, "http://example.org/inner")),
MarkupGrammar.TokenPrimitive("this should be inner"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementBegin(new DataName("middle-2", String.Empty, "http://example.org/outer")),
MarkupGrammar.TokenPrimitive("this should be outer"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
var actual = tokenizer.GetTokens(input).ToArray();
Assert.Equal(expected, actual);
}
public void GetTokens_UndeclaredPrefixes_ThrowsDeserializationException()
{
const string input = @"<a:one><b:two><c:three></d:three></e:two></f:one>";
var expected = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("one")),
MarkupGrammar.TokenElementBegin(new DataName("two")),
MarkupGrammar.TokenElementBegin(new DataName("three")),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd
};
var tokenizer = new XmlReader.XmlTokenizer();
DeserializationException ex = Assert.Throws <DeserializationException>(
delegate()
{
var actual = tokenizer.GetTokens(input).ToArray();
});
Assert.Equal(2, ex.Index);
}