public void Format_PrefixWithoutNamespace_ReturnsPrefix()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("div")),
MarkupGrammar.TokenAttribute(new DataName("visible", "jbst", null)),
MarkupGrammar.TokenPrimitive(false),
MarkupGrammar.TokenElementEnd,
};
const string expected = @"<div jbst:visible=""false""></div>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_HtmlEntityEuroEncodeNonAscii_ReturnsMarkup()
{
var input = new[]
{
MarkupGrammar.TokenPrimitive("€")
};
const string expected = @"€";
var formatter = new HtmlFormatter(new DataWriterSettings())
{
EncodeNonAscii = true
};
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_SingleAttributeWhitespace_ReturnsMarkup()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("root")),
MarkupGrammar.TokenAttribute(new DataName("whitespace")),
MarkupGrammar.TokenPrimitive(" this contains whitespace "),
MarkupGrammar.TokenElementEnd
};
const string expected = @"<root whitespace="" this contains whitespace ""></root>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_XmlEntityHex_ReturnsMarkup()
{
var input = new[]
{
MarkupGrammar.TokenPrimitive("\uABCD")
};
const string expected = "\uABCD";
var formatter = new HtmlFormatter(new DataWriterSettings())
{
EncodeNonAscii = false
};
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_MathML_ReturnsMarkup()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("p")),
MarkupGrammar.TokenPrimitive(@"You can add a string to a number, but this stringifies the number:"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n"),
MarkupGrammar.TokenElementBegin(new DataName("math")),
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementBegin(new DataName("ms")),
MarkupGrammar.TokenPrimitive(@"x<y"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementBegin(new DataName("mo")),
MarkupGrammar.TokenPrimitive(@"+"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementBegin(new DataName("mn")),
MarkupGrammar.TokenPrimitive(@"3"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementBegin(new DataName("mo")),
MarkupGrammar.TokenPrimitive(@"="),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementBegin(new DataName("ms")),
MarkupGrammar.TokenPrimitive(@"x<y3"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n"),
MarkupGrammar.TokenElementEnd,
};
const string expected =
@"<p>You can add a string to a number, but this stringifies the number:</p>
<math>
<ms>x<y</ms>
<mo>+</mo>
<mn>3</mn>
<mo>=</mo>
<ms>x<y3</ms>
</math>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_ParentAndChildDifferentDefaultNamespaces_ReturnsMarkup()
{
var input = 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,
};
const string expected = @"<foo xmlns=""http://json.org""><child xmlns=""http://jsonfx.net"">text value</child></foo>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_NamespacedChildTag_ReturnsMarkup()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("foo")),
MarkupGrammar.TokenElementBegin(new DataName("child", String.Empty, "http://example.com/schema")),
MarkupGrammar.TokenPrimitive("value"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd
};
const string expected = @"<foo><child xmlns=""http://example.com/schema"">value</child></foo>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_ParentAndChildSharePrefixedNamespace_ReturnsMarkup()
{
var input = 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,
};
const string expected = @"<bar:foo xmlns:bar=""http://example.org""><bar:child>value</bar:child></bar:foo>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_OverlappingNamespacedTagsErrorRecovery_ReturnsAutoBalanced()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("odd", "a", "http://example.com/odd/a")),
MarkupGrammar.TokenElementBegin(new DataName("auto-closed", "b", "http://example.com/auto-closed/b")),
MarkupGrammar.TokenElementBegin(new DataName("even", "c", "http://example.com/even/c")),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd,
};
const string expected = @"<a:odd xmlns:a=""http://example.com/odd/a""><b:auto-closed xmlns:b=""http://example.com/auto-closed/b""><c:even xmlns:c=""http://example.com/even/c""></c:even></b:auto-closed></a:odd>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_OverlappingTagsAutoBalanced_ReturnsMarkupAsIs()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("odd")),
MarkupGrammar.TokenElementBegin(new DataName("auto-closed")),
MarkupGrammar.TokenElementBegin(new DataName("even")),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenElementEnd
};
const string expected = @"<odd><auto-closed><even></even></auto-closed></odd>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_XmlDocTypeExternal_ReturnsUnparsed()
{
var input = new[]
{
MarkupGrammar.TokenUnparsed("!", "",
@"DOCTYPE html PUBLIC
""-//W3C//DTD XHTML 1.1//EN""
""http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd""" )
};
const string expected =
@"<!DOCTYPE html PUBLIC
""-//W3C//DTD XHTML 1.1//EN""
""http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd"">" ;
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
private void WriteXmlns(TextWriter writer, string prefix, string namespaceUri)
{
// " xmlns"
writer.Write(MarkupGrammar.OperatorValueDelim);
this.WriteLocalName(writer, "xmlns");
if (!String.IsNullOrEmpty(prefix))
{
// ":prefix"
writer.Write(MarkupGrammar.OperatorPrefixDelim);
this.WriteLocalName(writer, prefix);
}
// ="value"
writer.Write(MarkupGrammar.OperatorPairDelim);
writer.Write(MarkupGrammar.OperatorStringDelim);
HtmlFormatter.HtmlAttributeEncode(writer, namespaceUri, this.encodeNonAscii, this.canonicalForm);
writer.Write(MarkupGrammar.OperatorStringDelim);
}
public void Format_SingleEmptyAttributeHtmlStyle_ReturnsMarkup()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("root")),
MarkupGrammar.TokenAttribute(new DataName("noValue")),
MarkupGrammar.TokenPrimitive(String.Empty),
MarkupGrammar.TokenElementEnd
};
const string expected = @"<root noValue></root>";
var formatter = new HtmlFormatter(new DataWriterSettings())
{
EmptyAttributes = HtmlFormatter.EmptyAttributeType.Html
};
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
//[Fact(Skip="Embedded DOCTYPE not supported")]
public void Format_XmlDocTypeLocal_ReturnsUnparsed()
{
var input = new[]
{
MarkupGrammar.TokenUnparsed("!", "",
@"DOCTYPE doc [
<!ATTLIST normId id ID #IMPLIED>
<!ATTLIST normNames attr NMTOKENS #IMPLIED>
]")
};
const string expected =
@"<!DOCTYPE doc [
<!ATTLIST normId id ID #IMPLIED>
<!ATTLIST normNames attr NMTOKENS #IMPLIED>
]>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_DifferentPrefixSameNamespace_ReturnsMarkup()
{
// Not sure if this is correct: http://stackoverflow.com/questions/3312390
// "The namespace name for an unprefixed attribute name always has no value"
// "The attribute value in a default namespace declaration MAY be empty.
// This has the same effect, within the scope of the declaration, of there being no default namespace."
// http://www.w3.org/TR/xml-names/#defaulting
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("foo", String.Empty, "http://example.org")),
MarkupGrammar.TokenAttribute(new DataName("key", "blah", "http://example.org")),
MarkupGrammar.TokenPrimitive("value"),
MarkupGrammar.TokenElementEnd,
};
const string expected = @"<foo xmlns=""http://example.org"" key=""value""></foo>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_NestedDefaultNamespaces_ReturnsMarkup()
{
var input = 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,
};
const string expected = @"<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 formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_MultipleAttributesCanonicalForm_ReturnsCanonicalMarkup()
{
var input = new[]
{
MarkupGrammar.TokenElementVoid(new DataName("root")),
MarkupGrammar.TokenAttribute(new DataName("whitespace")),
MarkupGrammar.TokenPrimitive(" this contains whitespace "),
MarkupGrammar.TokenAttribute(new DataName("no-value")),
MarkupGrammar.TokenPrimitive(String.Empty),
MarkupGrammar.TokenAttribute(new DataName("anyQuotedText")),
MarkupGrammar.TokenPrimitive("/\\\uCAFE\uBABE\uAB98\uFCDE\ubcda\uef4A\b\f\n\r\t`1~!@#$%^&*()_+-=[]{}|;:',./<>?")
};
const string expected = @"<root anyQuotedText=""/\" + "\uCAFE\uBABE\uAB98\uFCDE\uBCDA\uEF4A" + @"

	`1~!@#$%^&*()_+-=[]{}|;:',./<>?"" no-value whitespace="" this contains whitespace ""></root>";
var formatter = new HtmlFormatter(new DataWriterSettings())
{
CanonicalForm = true
};
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_PhpHelloWorld_ReturnsMarkup()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("html")),
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementBegin(new DataName("head")),
MarkupGrammar.TokenPrimitive("\r\n\t\t"),
MarkupGrammar.TokenElementBegin(new DataName("title")),
MarkupGrammar.TokenPrimitive("PHP Test"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementBegin(new DataName("body")),
MarkupGrammar.TokenPrimitive("\r\n\t\t"),
MarkupGrammar.TokenUnparsed("?", "?", @"php echo '<p>Hello World</p>'; "),
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n"),
MarkupGrammar.TokenElementEnd,
};
const string expected =
@"<html>
<head>
<title>PHP Test</title>
</head>
<body>
<?php echo '<p>Hello World</p>'; ?>
</body>
</html>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
public void Format_HtmlContentPrettyPrinted_ReturnsMarkup()
{
var input = new[]
{
MarkupGrammar.TokenElementBegin(new DataName("div")),
MarkupGrammar.TokenAttribute(new DataName("class")),
MarkupGrammar.TokenPrimitive("content"),
MarkupGrammar.TokenPrimitive("\r\n\t"),
MarkupGrammar.TokenElementBegin(new DataName("p")),
MarkupGrammar.TokenAttribute(new DataName("style")),
MarkupGrammar.TokenPrimitive("color:red"),
MarkupGrammar.TokenPrimitive("\r\n\t\t"),
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.\r\n\t"),
MarkupGrammar.TokenElementEnd,
MarkupGrammar.TokenPrimitive("\r\n"),
MarkupGrammar.TokenElementEnd,
};
const string expected =
@"<div class=""content"">
<p style=""color:red"">
<strong>Lorem ipsum</strong> dolor sit amet, <i>consectetur</i> adipiscing elit.
</p>
</div>";
var formatter = new HtmlFormatter(new DataWriterSettings());
var actual = formatter.Format(input);
Assert.Equal(expected, actual);
}
/// <summary>
/// Emits valid XML attribute character data
/// </summary>
/// <param name="writer"></param>
/// <param name="value"></param>
/// <param name="encodeNonAscii">encodes all non-ASCII chars</param>
public static void HtmlAttributeEncode(TextWriter writer, string value, bool encodeNonAscii)
{
HtmlFormatter.HtmlAttributeEncode(writer, value, encodeNonAscii, false);
}
private void WriteAttribute(TextWriter writer, string prefix, string localName, Token <MarkupTokenType> value)
{
// " "
writer.Write(MarkupGrammar.OperatorValueDelim);
if (!String.IsNullOrEmpty(prefix))
{
// "prefix:"
this.WriteLocalName(writer, prefix);
writer.Write(MarkupGrammar.OperatorPrefixDelim);
}
// local-name
this.WriteLocalName(writer, localName);
ITextFormattable <MarkupTokenType> formattable = value.Value as ITextFormattable <MarkupTokenType>;
string attrValue = (formattable == null) ? value.ValueAsString() : null;
if ((formattable == null) &&
String.IsNullOrEmpty(attrValue))
{
switch (this.EmptyAttributes)
{
case EmptyAttributeType.Html:
{
return;
}
case EmptyAttributeType.Xhtml:
{
attrValue = localName;
break;
}
case EmptyAttributeType.Xml:
{
break;
}
}
}
// ="value"
writer.Write(MarkupGrammar.OperatorPairDelim);
writer.Write(MarkupGrammar.OperatorStringDelim);
switch (value.TokenType)
{
case MarkupTokenType.Primitive:
{
if (formattable != null)
{
formattable.Format(this, writer);
}
else
{
HtmlFormatter.HtmlAttributeEncode(writer, attrValue, this.encodeNonAscii, this.canonicalForm);
}
break;
}
default:
{
throw new TokenException <MarkupTokenType>(
value,
String.Format(HtmlFormatter.ErrorUnexpectedToken, value));
}
}
writer.Write(MarkupGrammar.OperatorStringDelim);
}
/// <summary>
/// Emits valid XML attribute character data
/// </summary>
/// <param name="writer"></param>
/// <param name="value"></param>
/// <remarks>
/// From XML 1.0, 5th ed. http://www.w3.org/TR/xml/#syntax
/// CharData is defined as all chars except less-than ('<'), ampersand ('&'), the sequence "]]>", optionally encoding greater-than ('>').
/// Attributes should additionally encode double-quote ('"') and single-quote ('\'')
/// Rather than detect "]]>", this simply encodes all '>'.
/// </remarks>
public static void HtmlAttributeEncode(TextWriter writer, string value)
{
HtmlFormatter.HtmlAttributeEncode(writer, value, false, false);
}
/// <summary>
/// Formats the token sequence to the writer
/// </summary>
/// <param name="writer"></param>
/// <param name="tokens"></param>
public void Format(IEnumerable <Token <MarkupTokenType> > tokens, TextWriter writer)
{
if (tokens == null)
{
throw new ArgumentNullException("tokens");
}
IStream <Token <MarkupTokenType> > stream = Stream <Token <MarkupTokenType> > .Create(tokens);
PrefixScopeChain.Scope scope = null;
while (!stream.IsCompleted)
{
Token <MarkupTokenType> token = stream.Peek();
switch (token.TokenType)
{
case MarkupTokenType.ElementBegin:
case MarkupTokenType.ElementVoid:
{
DataName tagName = token.Name;
MarkupTokenType tagType = token.TokenType;
stream.Pop();
token = stream.Peek();
scope = new PrefixScopeChain.Scope();
if (this.ScopeChain.ContainsNamespace(tagName.NamespaceUri) ||
(String.IsNullOrEmpty(tagName.NamespaceUri) && !this.ScopeChain.ContainsPrefix(String.Empty)))
{
string prefix = this.ScopeChain.GetPrefix(tagName.NamespaceUri, false);
scope.TagName = new DataName(tagName.LocalName, prefix, tagName.NamespaceUri);
}
else
{
scope[tagName.Prefix] = tagName.NamespaceUri;
scope.TagName = tagName;
}
this.ScopeChain.Push(scope);
IDictionary <DataName, Token <MarkupTokenType> > attributes = null;
while (!stream.IsCompleted && token.TokenType == MarkupTokenType.Attribute)
{
if (attributes == null)
{
attributes = this.canonicalForm ?
(IDictionary <DataName, Token <MarkupTokenType> >) new CanonicalList() :
(IDictionary <DataName, Token <MarkupTokenType> >) new Dictionary <DataName, Token <MarkupTokenType> >();
}
DataName attrName = token.Name;
string prefix = this.ScopeChain.EnsurePrefix(attrName.Prefix, attrName.NamespaceUri);
if (prefix != null)
{
if (prefix != attrName.Prefix)
{
attrName = new DataName(attrName.LocalName, prefix, attrName.NamespaceUri, true);
}
if (!this.ScopeChain.ContainsNamespace(attrName.NamespaceUri) &&
(!String.IsNullOrEmpty(attrName.NamespaceUri) || this.ScopeChain.ContainsPrefix(String.Empty)))
{
scope[prefix] = attrName.NamespaceUri;
}
}
stream.Pop();
token = stream.Peek();
attributes[attrName] = token ?? MarkupGrammar.TokenNone;
stream.Pop();
token = stream.Peek();
}
this.WriteTag(writer, tagType, tagName, attributes, scope);
scope = null;
break;
}
case MarkupTokenType.ElementEnd:
{
if (this.ScopeChain.HasScope)
{
this.WriteTag(writer, MarkupTokenType.ElementEnd, this.ScopeChain.Peek().TagName, null, null);
this.ScopeChain.Pop();
}
else
{
// TODO: decide if this is should throw an exception
}
stream.Pop();
token = stream.Peek();
break;
}
case MarkupTokenType.Primitive:
{
ITextFormattable <MarkupTokenType> formattable = token.Value as ITextFormattable <MarkupTokenType>;
if (formattable != null)
{
formattable.Format(this, writer);
}
else
{
HtmlFormatter.HtmlEncode(writer, token.ValueAsString(), this.encodeNonAscii, this.canonicalForm);
}
stream.Pop();
token = stream.Peek();
break;
}
default:
{
throw new TokenException <MarkupTokenType>(
token,
String.Format(ErrorUnexpectedToken, token.TokenType));
}
}
}
}
