public void HtmlParser_OnCommentStateTest() {
string text = "<!-- abcde -->";
var target = new HtmlParser();
target._cs = new HtmlCharStream(text);
target._tokenizer = new HtmlTokenizer(target._cs);
target.CommentFound +=
delegate (object sender, HtmlParserCommentEventArgs args) {
Assert.True(args.CommentToken is CommentToken);
CommentToken ct = args.CommentToken;
Assert.Equal(1, ct.Count);
Assert.Equal(0, ct.Start);
Assert.Equal(14, ct.End);
Assert.True(ct[0] is HtmlToken);
Assert.True(ct[0] is IExpandableTextRange);
Assert.Equal(0, ct[0].Start);
Assert.Equal(14, ct[0].End);
};
target.OnCommentState();
}
public void HtmlParser_OnEndTagStateTest() {
var target = new HtmlParser();
target._cs = new HtmlCharStream("</span \r\n foo='bar' nowrap>");
target._tokenizer = new HtmlTokenizer(target._cs);
target.EndTagOpen +=
delegate (object sender, HtmlParserOpenTagEventArgs args) {
Assert.True(args.NameToken is NameToken);
Assert.Equal(2, args.NameToken.Start);
Assert.Equal(6, args.NameToken.End);
endTagOpenCalls++;
};
target.EndTagClose +=
delegate (object sender, HtmlParserCloseTagEventArgs args) {
Assert.False(args.IsShorthand);
Assert.Equal(26, args.CloseAngleBracket.Start);
endTagCloseCalls++;
};
target.AttributeFound +=
delegate (object sender, HtmlParserAttributeEventArgs args) {
attributeCalls++;
};
target.OnEndTagState();
Assert.Equal(1, endTagOpenCalls);
Assert.Equal(1, endTagCloseCalls);
Assert.Equal(0, attributeCalls);
}
public void HtmlParser_FindEndOfBlockTest() {
var target = new HtmlParser();
string script = "<script>var x = \"boo\"; if(x < y) { }</script>";
target._cs = new HtmlCharStream(script);
target._tokenizer = new HtmlTokenizer(target._cs);
var r = target.FindEndOfBlock("</script>");
Assert.Equal(script.Length - 9, r.Length);
}
public void OnStartTagState_WellFormedTest() {
var target = new HtmlParser();
target._cs = new HtmlCharStream("<input\r\n type='button' nowrap/>");
int startTagOpenCalls = 0;
int startTagCloseCalls = 0;
int attributeCalls = 0;
target._tokenizer = new HtmlTokenizer(target._cs);
target.StartTagOpen +=
delegate (object sender, HtmlParserOpenTagEventArgs args) {
Assert.True(args.NameToken is NameToken);
Assert.Equal(1, args.NameToken.Start);
Assert.Equal(6, args.NameToken.End);
startTagOpenCalls++;
};
target.StartTagClose +=
delegate (object sender, HtmlParserCloseTagEventArgs args) {
Assert.Equal(29, args.CloseAngleBracket.Start);
Assert.Equal(31, args.CloseAngleBracket.End);
Assert.True(args.IsShorthand);
Assert.True(args.IsClosed);
startTagCloseCalls++;
};
target.AttributeFound +=
delegate (object sender, HtmlParserAttributeEventArgs args) {
Assert.True(args.AttributeToken is AttributeToken);
var at = args.AttributeToken;
switch (attributeCalls) {
case 0:
Assert.True(at.HasName());
Assert.True(at.HasValue());
Assert.Equal(9, at.NameToken.Start);
Assert.Equal(13, at.NameToken.End);
Assert.Equal(14, at.ValueToken.Start);
Assert.Equal(22, at.ValueToken.End);
break;
case 1:
Assert.True(at.HasName());
Assert.False(at.HasValue());
Assert.Equal(23, at.NameToken.Start);
Assert.Equal(29, at.NameToken.End);
break;
}
attributeCalls++;
};
target.OnStartTagState();
Assert.Equal(1, startTagOpenCalls);
Assert.Equal(1, startTagCloseCalls);
Assert.Equal(2, attributeCalls);
}
public void OnEntity_WellFormedTest1() {
HtmlParser parser = new HtmlParser();
parser.EntityFound += delegate (object sender, HtmlParserRangeEventArgs args) {
Assert.Equal(3, args.Range.Start);
Assert.Equal(9, args.Range.End);
};
parser.Parse("foo ");
}
public void HtmlParser_UnknownDocType() {
var parser = new HtmlParser();
parser.Parse("<!doctype randomname1><html></html>");
Assert.Equal(DocType.Unrecognized, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
parser.Parse("<html></html>");
Assert.Equal(DocType.Undefined, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
}
public void HtmlParser_DuplicateDocType() {
var parser = new HtmlParser();
parser.Parse("<!doctype randomname1>" + DocTypeSignatures.Html5 + "<html></html>");
Assert.Equal(DocType.Unrecognized, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Html5 + "<!doctype randomname1><html></html>");
Assert.Equal(DocType.Html5, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
}
public void OnEntity_BrokenTest() {
HtmlParser parser = new HtmlParser();
int count = 0;
parser.EntityFound += delegate (object sender, HtmlParserRangeEventArgs args) {
count++;
};
parser.Parse(">");
Assert.Equal(0, count);
}
public void OnEntity_WellFormedTest3() {
HtmlParser parser = new HtmlParser();
int count = 0;
parser.EntityFound += delegate (object sender, HtmlParserRangeEventArgs args) {
count++;
};
parser.Parse(new TextStream("foo­bar ઼&abc;<html < dir=\">"));
Assert.Equal(2, count);
}
public void OnEntity_WellFormedTest2() {
HtmlParser parser = new HtmlParser();
int count = 0;
parser.EntityFound += delegate (object sender, HtmlParserRangeEventArgs args) {
count++;
};
parser.Parse("foo bar <<html < dir=\">\"> > "");
Assert.Equal(3, count);
}
public EventLogger(HtmlParser parser) {
parser.AttributeFound += OnAttribute;
parser.CommentFound += OnComment;
parser.EndTagClose += OnEndTagClose;
parser.EndTagOpen += OnEndTagOpen;
parser.EntityFound += OnEntity;
parser.ScriptBlockFound += OnScriptBlock;
parser.StartTagClose += OnStartTagClose;
parser.StartTagOpen += OnStartTagOpen;
parser.StyleBlockFound += OnStyleBlock;
Log = new StringBuilder();
_parser = parser;
}
public void HtmlParser_OnStyleStateTest() {
var target = new HtmlParser();
target._cs = new HtmlCharStream(" body { background-color: green }</style>");
target._tokenizer = new HtmlTokenizer(target._cs);
int count = 0;
target.StyleBlockFound +=
delegate (object sender, HtmlParserBlockRangeEventArgs args) {
Assert.Equal(0, args.Range.Start);
Assert.Equal(33, args.Range.End);
count++;
};
target.OnStyleState(NameToken.Create(0, 0));
Assert.Equal(1, count);
}
public HtmlTreeBuilder(HtmlParser parser, HtmlTree tree) {
Parser = parser;
_tree = tree;
parser.ParsingStarting += OnParseBegin;
parser.ParsingComplete += OnParseEnd;
parser.StartTagOpen += OnStartTagOpen;
parser.StartTagClose += OnStartTagClose;
parser.AttributeFound += OnAttribute;
parser.EndTagOpen += OnEndTagOpen;
parser.EndTagClose += OnEndTagClose;
parser.CommentFound += OnComment;
// For tree building purposes we don't care about comments or entities.
// Colorable item collection cares so it should be listening to those events.
}
public void HtmlParser_OnScriptStateTest() {
string text = " var x = 1; while(x < 5 && x > 0) { x++; }</script>";
var target = new HtmlParser();
target._cs = new HtmlCharStream(text);
target._tokenizer = new HtmlTokenizer(target._cs);
int count = 0;
target.ScriptBlockFound +=
delegate (object sender, HtmlParserBlockRangeEventArgs args) {
Assert.Equal(0, args.Range.Start);
Assert.Equal(text.Length - 9, args.Range.End);
count++;
};
target.OnScriptState(String.Empty, NameToken.Create(0, 0));
Assert.Equal(1, count);
}
static public void ParseText(HtmlTestFilesFixture fixture, string name) {
string testFile = fixture.GetDestinationPath(name);
string baselineFile = testFile + ".log";
string text = fixture.LoadDestinationFile(name);
EventLogger logger = null;
HtmlParser p = new HtmlParser();
logger = new EventLogger(p);
p.Parse(text);
var actual = logger.ToString();
if (_regenerateBaselineFiles) {
// Update this to your actual enlistment if you need to update baseline
baselineFile = Path.Combine(fixture.SourcePath, name) + ".log";
TestFiles.UpdateBaseline(baselineFile, actual);
} else {
TestFiles.CompareToBaseLine(baselineFile, actual);
}
}
public void HtmlParser_KnownDocTypes() {
var parser = new HtmlParser();
parser.Parse(DocTypeSignatures.Html32 + "<html></html>");
Assert.Equal(DocType.Html32, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Html401Frameset + "<html></html>");
Assert.Equal(DocType.Html401Frameset, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Html401Strict + "<html></html>");
Assert.Equal(DocType.Html401Strict, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Html401Transitional + "<html></html>");
Assert.Equal(DocType.Html401Transitional, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Html5 + "<html></html>");
Assert.Equal(DocType.Html5, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Xhtml10Frameset + "<html></html>");
Assert.Equal(DocType.Xhtml10Frameset, parser.DocType);
Assert.Equal(ParsingMode.Xhtml, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Xhtml10Strict + "<html></html>");
Assert.Equal(DocType.Xhtml10Strict, parser.DocType);
Assert.Equal(ParsingMode.Xhtml, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Xhtml10Transitional + "<html></html>");
Assert.Equal(DocType.Xhtml10Transitional, parser.DocType);
Assert.Equal(ParsingMode.Xhtml, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Xhtml11 + "<html></html>");
Assert.Equal(DocType.Xhtml11, parser.DocType);
Assert.Equal(ParsingMode.Xhtml, parser.ParsingMode);
parser.Parse(DocTypeSignatures.Xhtml20 + "<html></html>");
Assert.Equal(DocType.Xhtml20, parser.DocType);
Assert.Equal(ParsingMode.Xhtml, parser.ParsingMode);
}
/// <summary>
/// Main method to build a subtree from a text range.
/// </summary>
public virtual void Build(ITextRange range) {
CommentCollection = new CommentCollection();
HtmlParser parser = new HtmlParser(ParsingMode, ScriptTypeResolution, ScriptOrStyleTagNameService);
HtmlTreeBuilder builder = new HtmlTreeBuilder(parser, this);
parser.Parse(Text, range);
DocType = parser.DocType;
ParsingMode = parser.ParsingMode;
}
private HtmlTree ParseHtml(string html) {
HtmlParser parser = new HtmlParser();
HtmlTree tree = new HtmlTree(new TextStream(html));
tree.Build();
return tree;
}
public void AttributeParse_InlineStyle_Typo() {
var target = new HtmlParser();
target.AttributeFound +=
delegate (object sender, HtmlParserAttributeEventArgs args) {
Assert.True(args.AttributeToken is AttributeToken);
AttributeToken at = args.AttributeToken;
switch (attributeCalls) {
case 0:
Assert.True(at.HasName());
Assert.Equal(7, at.NameToken.Start);
Assert.Equal(5, at.NameToken.Length);
Assert.True(at.HasValue());
Assert.Equal(13, at.ValueToken.Start);
Assert.Equal(14, at.ValueToken.Length);
break;
case 1:
Assert.True(at.HasName());
Assert.Equal(27, at.NameToken.Start);
Assert.Equal(1, at.NameToken.Length);
Assert.False(at.HasValue());
break;
}
attributeCalls++;
};
attributeCalls = 0;
target.Parse("<input style=\"display:none\";>");
Assert.Equal(2, attributeCalls);
}
public void AttributeParse_RandomCharacters() {
var target = new HtmlParser();
target.AttributeFound +=
delegate (object sender, HtmlParserAttributeEventArgs args) {
Assert.True(args.AttributeToken is AttributeToken);
AttributeToken at = args.AttributeToken;
switch (attributeCalls) {
case 0:
Assert.True(at.HasName());
Assert.Equal(3, at.NameToken.Start);
Assert.Equal(2, at.NameToken.Length);
Assert.True(at.HasValue());
Assert.Equal(6, at.ValueToken.Start);
Assert.Equal(1, at.ValueToken.Length);
break;
case 1:
Assert.True(at.HasName());
Assert.Equal(8, at.NameToken.Start);
Assert.Equal(1, at.NameToken.Length);
Assert.True(at.HasValue());
Assert.Equal(10, at.ValueToken.Start);
Assert.Equal(1, at.ValueToken.Length);
break;
case 2:
Assert.True(at.HasName());
Assert.Equal(12, at.NameToken.Start);
Assert.Equal(2, at.NameToken.Length);
Assert.False(at.HasValue());
break;
}
attributeCalls++;
};
attributeCalls = 0;
target.Parse("<a id=# $=% ^^");
Assert.Equal(3, attributeCalls);
}
public static HtmlParser CreateParser(string text) {
var target = new HtmlParser();
target._cs = new HtmlCharStream(text);
target._tokenizer = new HtmlTokenizer(target._cs);
return target;
}
public void HtmlParser_XmlPi() {
var parser = new HtmlParser();
parser.Parse("<?xml charset='utf-8' ?><html></html>");
Assert.Equal(DocType.Undefined, parser.DocType);
Assert.Equal(ParsingMode.Xml, parser.ParsingMode);
}
public void HtmlParser_MalformedDocType() {
var parser = new HtmlParser();
parser.Parse("<!doctype randomname1<html></html>");
Assert.Equal(DocType.Undefined, parser.DocType);
Assert.Equal(ParsingMode.Html, parser.ParsingMode);
}
