protected override async Task Use(Uri url, HTMLDocument document, CancellationToken cancel)
{
State = Brushes.LightGray;
this.document = document;
ChangeQuery();
await Task.Yield();
}
protected override async Task Use(Uri url, HTMLDocument document, CancellationToken cancel)
{
var elements = new Dictionary<String, Int32>();
var attributes = new Dictionary<String, Int32>();
var classes = new Dictionary<String, Int32>();
var words = new Dictionary<String, Int32>();
var various = new Dictionary<String, Int32>();
Status = "Gathering statistics ...";
various.Add("Links", document.Anchors.Length);
various.Add("Images", document.Images.Length);
various.Add("Scripts", document.Scripts.Length);
various.Add("Stylesheets", document.StyleSheets.Length);
various.Add("Embeds", document.Embeds.Length);
various.Add("Forms", document.Forms.Length);
await Task.Run(() => Inspect(document.DocumentElement, elements, classes, attributes));
cancel.ThrowIfCancellationRequested();
await Task.Run(() => Words(document.DocumentElement.TextContent.ToCharArray(), words));
cancel.ThrowIfCancellationRequested();
MostElements = CreatePieChart("Most elements", elements);
MostClasses = CreatePieChart("Most classes", classes);
MostWords = CreatePieChart("Most words", words);
MostAttributes = CreatePieChart("Most attributes", attributes);
}
/// <summary>
/// Creates a new builder with the specified source.
/// </summary>
/// <param name="source">The code manager.</param>
/// <param name="document">The document to fill.</param>
/// <param name="options">Options to use for the document generation.</param>
DocumentBuilder(SourceManager source, HTMLDocument document, DocumentOptions options)
{
document.Options = options;
parser = new HtmlParser(document, source);
parser.ErrorOccurred += ParseErrorOccurred;
if (options.OnError != null)
parser.ErrorOccurred += options.OnError;
}
protected override async Task Use(Uri url, HTMLDocument document, CancellationToken cancel)
{
this.url = url;
Status = "Rendering ...";
Reset();
root.Blocks.Add(RenderBox(document.Body));
await Task.Yield();
}
public Document CreateDocument(String namespaceURI, String qualifiedName, DocumentType doctype)
{
Document doc = null;
if (Namespaces.Html == namespaceURI)
doc = new HTMLDocument();
else
doc = new Document();
doc.AppendChild(doctype);
doc.NodeName = qualifiedName ?? doc.NodeName;
return doc;
}
protected override async Task Use(Uri url, HTMLDocument document, CancellationToken cancel)
{
nodes.Clear();
Status = "Constructing tree ...";
var elements = TreeNodeViewModel.SelectFrom(document.ChildNodes);
foreach (var element in elements)
{
element.Parent = nodes;
nodes.Add(element);
}
await Task.Yield();
}
/// <summary>
/// Loads a document from the given source.
/// </summary>
/// <param name="source">
/// A valid HTML document.
/// </param>
/// <returns>The current instance.</returns>
public CssSelectorTest LoadFromSource(String source)
{
_doc = DocumentBuilder.Html(source);
return this;
}
/// <summary>
/// Builds a list of nodes according with 8.4 Parsing HTML fragments.
/// </summary>
/// <param name="sourceCode">The string to use as source code.</param>
/// <param name="context">The context node to use.</param>
/// <returns>A list of parsed nodes.</returns>
public static NodeList HtmlFragment(String sourceCode, Node context = null)
{
var source = new SourceManager(sourceCode);
var doc = new HTMLDocument();
var db = new DocumentBuilder(source, doc);
if (context != null)
{
if (context.OwnerDocument != null && context.OwnerDocument.QuirksMode != QuirksMode.Off)
doc.QuirksMode = context.OwnerDocument.QuirksMode;
// Note: For performance reasons, an implementation that does not report errors and that uses
// the actual state machine described in this specification directly could use the
// PLAINTEXT state instead of the RAWTEXT and script data states where those are mentioned
// in the list above. Except for rules regarding parse errors, they are equivalent, since
// there is no appropriate end tag token in the fragment case, yet they involve far
// fewer state transitions.
((HtmlParser)db.parser).SwitchToFragment(context);
return db.HtmlResult.DocumentElement.ChildNodes;
}
return db.HtmlResult.ChildNodes;
}
/// <summary>
/// Creates a new builder with the specified source.
/// </summary>
/// <param name="source">The code manager.</param>
/// <param name="document">The document to fill.</param>
DocumentBuilder(SourceManager source, HTMLDocument document)
{
parser = new HtmlParser(document, source);
parser.ErrorOccurred += ParseErrorOccurred;
}
/// <summary>
/// Builds a list of nodes according with 8.4 Parsing HTML fragments.
/// </summary>
/// <param name="sourceCode">The string to use as source code.</param>
/// <param name="context">[Optional] The context node to use.</param>
/// <param name="options">[Optional] Options to use for the document generation.</param>
/// <returns>A list of parsed nodes.</returns>
public static NodeList HtmlFragment(String sourceCode, Node context = null, DocumentOptions options = null)
{
var source = new SourceManager(sourceCode);
var doc = new HTMLDocument();
//Disable scripting for HTML fragments (security reasons)
options = options ?? new DocumentOptions(scripting: false);
var db = new DocumentBuilder(source, doc, options);
if (context != null)
{
if (context.OwnerDocument != null && context.OwnerDocument.QuirksMode != QuirksMode.Off)
doc.QuirksMode = context.OwnerDocument.QuirksMode;
var parser = (HtmlParser)db.parser;
parser.SwitchToFragment(context);
return parser.Result.DocumentElement.ChildNodes;
}
return db.HtmlResult.ChildNodes;
}
public void TreeNoScriptWithNoScriptCommentInside()
{
var doc = new HTMLDocument();
var parser = new HtmlParser(doc, @"<noscript><!--<noscript></noscript>--></noscript>");
doc.Options = new DocumentOptions(scripting: true);
parser.Parse();
var dochtml0 = doc.ChildNodes[0];
Assert.AreEqual(2, dochtml0.ChildNodes.Length);
Assert.AreEqual(0, dochtml0.Attributes.Length);
Assert.AreEqual("html", dochtml0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0.NodeType);
var dochtml0head0 = dochtml0.ChildNodes[0];
Assert.AreEqual(1, dochtml0head0.ChildNodes.Length);
Assert.AreEqual(0, dochtml0head0.Attributes.Length);
Assert.AreEqual("head", dochtml0head0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0head0.NodeType);
var dochtml0head0noscript0 = dochtml0head0.ChildNodes[0];
Assert.AreEqual(1, dochtml0head0noscript0.ChildNodes.Length);
Assert.AreEqual(0, dochtml0head0noscript0.Attributes.Length);
Assert.AreEqual("noscript", dochtml0head0noscript0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0head0noscript0.NodeType);
var dochtml0head0noscript0Text0 = dochtml0head0noscript0.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml0head0noscript0Text0.NodeType);
Assert.AreEqual("<!--<noscript>", dochtml0head0noscript0Text0.TextContent);
var dochtml0body1 = dochtml0.ChildNodes[1];
Assert.AreEqual(1, dochtml0body1.ChildNodes.Length);
Assert.AreEqual(0, dochtml0body1.Attributes.Length);
Assert.AreEqual("body", dochtml0body1.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0body1.NodeType);
var dochtml0body1Text0 = dochtml0body1.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml0body1Text0.NodeType);
Assert.AreEqual("-->", dochtml0body1Text0.TextContent);
}
public void Setup()
{
document = DocumentBuilder.Html(Assets.QuirksMode);
}
/// <summary>
/// Loads a HTML document from the given URL.
/// </summary>
/// <param name="url">The URL that hosts the HTML content.</param>
/// <returns>The document with the parsed content.</returns>
public static HTMLDocument LoadFromURL(String url)
{
var doc = new HTMLDocument();
return doc.Load(url);
}
public void ScriptNoScriptAfterDoctypeWithIFrameContentAndTextAfter()
{
var doc = new HTMLDocument();
var parser = new HtmlParser(doc, "<!doctype html><noscript><iframe></noscript>X");
doc.Options = new DocumentOptions(scripting: true);
parser.Parse();
var docType0 = doc.ChildNodes[0] as DocumentType;
Assert.IsNotNull(docType0);
Assert.AreEqual(NodeType.DocumentType, docType0.NodeType);
Assert.AreEqual(@"html", docType0.Name);
var dochtml1 = doc.ChildNodes[1];
Assert.AreEqual(2, dochtml1.ChildNodes.Length);
Assert.AreEqual(0, dochtml1.Attributes.Length);
Assert.AreEqual("html", dochtml1.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1.NodeType);
var dochtml1head0 = dochtml1.ChildNodes[0];
Assert.AreEqual(1, dochtml1head0.ChildNodes.Length);
Assert.AreEqual(0, dochtml1head0.Attributes.Length);
Assert.AreEqual("head", dochtml1head0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1head0.NodeType);
var dochtml1head0noscript0 = dochtml1head0.ChildNodes[0];
Assert.AreEqual(1, dochtml1head0noscript0.ChildNodes.Length);
Assert.AreEqual(0, dochtml1head0noscript0.Attributes.Length);
Assert.AreEqual("noscript", dochtml1head0noscript0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1head0noscript0.NodeType);
var dochtml1head0noscript0Text0 = dochtml1head0noscript0.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml1head0noscript0Text0.NodeType);
Assert.AreEqual("<iframe>", dochtml1head0noscript0Text0.TextContent);
var dochtml1body1 = dochtml1.ChildNodes[1];
Assert.AreEqual(1, dochtml1body1.ChildNodes.Length);
Assert.AreEqual(0, dochtml1body1.Attributes.Length);
Assert.AreEqual("body", dochtml1body1.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1body1.NodeType);
var dochtml1body1Text0 = dochtml1body1.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml1body1Text0.NodeType);
Assert.AreEqual("X", dochtml1body1Text0.TextContent);
}
public void ScriptNoScriptWithCommentStartAndTextInsideBeforeClosing()
{
var doc = new HTMLDocument();
var parser = new HtmlParser(doc, "<!doctype html><noscript><!--</noscript>X<noscript>--></noscript>");
doc.Options = new DocumentOptions(scripting: true);
parser.Parse();
var docType0 = doc.ChildNodes[0] as DocumentType;
Assert.IsNotNull(docType0);
Assert.AreEqual(NodeType.DocumentType, docType0.NodeType);
Assert.AreEqual(@"html", docType0.Name);
var dochtml1 = doc.ChildNodes[1];
Assert.AreEqual(2, dochtml1.ChildNodes.Length);
Assert.AreEqual(0, dochtml1.Attributes.Length);
Assert.AreEqual("html", dochtml1.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1.NodeType);
var dochtml1head0 = dochtml1.ChildNodes[0];
Assert.AreEqual(1, dochtml1head0.ChildNodes.Length);
Assert.AreEqual(0, dochtml1head0.Attributes.Length);
Assert.AreEqual("head", dochtml1head0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1head0.NodeType);
var dochtml1head0noscript0 = dochtml1head0.ChildNodes[0];
Assert.AreEqual(1, dochtml1head0noscript0.ChildNodes.Length);
Assert.AreEqual(0, dochtml1head0noscript0.Attributes.Length);
Assert.AreEqual("noscript", dochtml1head0noscript0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1head0noscript0.NodeType);
var dochtml1head0noscript0Text0 = dochtml1head0noscript0.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml1head0noscript0Text0.NodeType);
Assert.AreEqual("<!--", dochtml1head0noscript0Text0.TextContent);
var dochtml1body1 = dochtml1.ChildNodes[1];
Assert.AreEqual(2, dochtml1body1.ChildNodes.Length);
Assert.AreEqual(0, dochtml1body1.Attributes.Length);
Assert.AreEqual("body", dochtml1body1.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1body1.NodeType);
var dochtml1body1Text0 = dochtml1body1.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml1body1Text0.NodeType);
Assert.AreEqual("X", dochtml1body1Text0.TextContent);
var dochtml1body1noscript1 = dochtml1body1.ChildNodes[1];
Assert.AreEqual(1, dochtml1body1noscript1.ChildNodes.Length);
Assert.AreEqual(0, dochtml1body1noscript1.Attributes.Length);
Assert.AreEqual("noscript", dochtml1body1noscript1.NodeName);
Assert.AreEqual(NodeType.Element, dochtml1body1noscript1.NodeType);
var dochtml1body1noscript1Text0 = dochtml1body1noscript1.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml1body1noscript1Text0.NodeType);
Assert.AreEqual("-->", dochtml1body1noscript1Text0.TextContent);
}
protected abstract Task Use(Uri url, HTMLDocument document, CancellationToken cancel);
protected override async Task Use(Uri url, HTMLDocument document, CancellationToken cancel)
{
local = url;
Selected = null;
source.Clear();
Status = "Looking for stylesheets ...";
for (int i = 0; i < document.StyleSheets.Length; i++)
{
var s = document.StyleSheets[i];
source.Add(s);
}
await Task.Yield();
}
void UpdateRecent(HTMLDocument document, String url, Uri uri)
{
recentAddress = url;
SettingsViewModel.Instance.AddUrl(url);
recentDocument = document;
recentUrl = uri;
recentState++;
state = recentState;
}
public void Setup()
{
document = DocumentBuilder.Html(Assets.SelectorsWebpage);
}
public void TreeParagraphWithTightAttributesAndNoScriptTagScriptingEnabled()
{
var doc = new HTMLDocument();
var parser = new HtmlParser(doc, @"<p id=""status""><noscript><strong>A</strong></noscript><span>B</span></p>");
doc.Options = new DocumentOptions(scripting: true);
parser.Parse();
var dochtml0 = doc.ChildNodes[0];
Assert.AreEqual(2, dochtml0.ChildNodes.Length);
Assert.AreEqual(0, dochtml0.Attributes.Length);
Assert.AreEqual("html", dochtml0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0.NodeType);
var dochtml0head0 = dochtml0.ChildNodes[0];
Assert.AreEqual(0, dochtml0head0.ChildNodes.Length);
Assert.AreEqual(0, dochtml0head0.Attributes.Length);
Assert.AreEqual("head", dochtml0head0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0head0.NodeType);
var dochtml0body1 = dochtml0.ChildNodes[1];
Assert.AreEqual(1, dochtml0body1.ChildNodes.Length);
Assert.AreEqual(0, dochtml0body1.Attributes.Length);
Assert.AreEqual("body", dochtml0body1.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0body1.NodeType);
var dochtml0body1p0 = dochtml0body1.ChildNodes[0];
Assert.AreEqual(2, dochtml0body1p0.ChildNodes.Length);
Assert.AreEqual(1, dochtml0body1p0.Attributes.Length);
Assert.AreEqual("p", dochtml0body1p0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0body1p0.NodeType);
Assert.AreEqual("status", dochtml0body1p0.Attributes["id"].Value);
var dochtml0body1p0noscript0 = dochtml0body1p0.ChildNodes[0];
Assert.AreEqual(1, dochtml0body1p0noscript0.ChildNodes.Length);
Assert.AreEqual(0, dochtml0body1p0noscript0.Attributes.Length);
Assert.AreEqual("noscript", dochtml0body1p0noscript0.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0body1p0noscript0.NodeType);
var dochtml0body1p0noscript0Text0 = dochtml0body1p0noscript0.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml0body1p0noscript0Text0.NodeType);
Assert.AreEqual("<strong>A</strong>", dochtml0body1p0noscript0Text0.TextContent);
var dochtml0body1p0span1 = dochtml0body1p0.ChildNodes[1];
Assert.AreEqual(1, dochtml0body1p0span1.ChildNodes.Length);
Assert.AreEqual(0, dochtml0body1p0span1.Attributes.Length);
Assert.AreEqual("span", dochtml0body1p0span1.NodeName);
Assert.AreEqual(NodeType.Element, dochtml0body1p0span1.NodeType);
var dochtml0body1p0span1Text0 = dochtml0body1p0span1.ChildNodes[0];
Assert.AreEqual(NodeType.Text, dochtml0body1p0span1Text0.NodeType);
Assert.AreEqual("B", dochtml0body1p0span1Text0.TextContent);
}
/// <summary>
/// Loads a document from the given URL.
/// </summary>
/// <param name="url">
/// A valid URL e.g. http://www.w3.org/TR/2001/CR-css3-selectors-20011113/.
/// </param>
/// <returns>The current instance.</returns>
public CssSelectorTest LoadFromUrl(string url)
{
Console.Write("Loading ");
Console.Write(url);
Console.Write(" . . . ");
_url = url;
HttpClient client = new HttpClient();
var content = client.GetAsync(url).Result.Content.ReadAsStreamAsync().Result;
_doc = DocumentBuilder.Html(content);
Console.WriteLine("loaded !");
return this;
}
/// <summary>
/// Creates a new instance of the HTML parser with the specified document
/// based on the given source manager.
/// </summary>
/// <param name="document">The document instance to be constructed.</param>
/// <param name="source">The source to use.</param>
internal HtmlParser(HTMLDocument document, SourceManager source)
{
tokenizer = new HtmlTokenizer(source);
tokenizer.ErrorOccurred += (s, ev) =>
{
if (ErrorOccurred != null)
ErrorOccurred(this, ev);
};
started = false;
doc = document;
open = new List<Element>();
formatting = new List<Element>();
frameset = true;
insert = HtmlTreeMode.Initial;
}
public void HtmlCustomTitleGeneration()
{
var doc = new HTMLDocument();
var title = "My Title";
doc.Title = title;
Assert.AreEqual(title, doc.Title);
}
public void HtmlHasRightHeadElement()
{
var doc = new HTMLDocument();
var root = new HTMLHtmlElement();
doc.AppendChild(root);
var head = new HTMLHeadElement();
root.AppendChild(head);
Assert.AreEqual(head, doc.Head);
}
protected override async Task Use(Uri url, HTMLDocument document, CancellationToken cancel)
{
Status = "Constructing the DOM ...";
Root = new DOMNodeViewModel(document);
await Task.Yield();
}
/// <summary>
/// Generates the unit tests from the given overview document.
/// </summary>
/// <param name="document">The document to inspect.</param>
public static XMLConfTest GenerateFromHtml(HTMLDocument document)
{
var body = document.Body;
var mode = Mode.None;
var test = new XMLConfTest();
for (int i = 0; i < body.ChildNodes.Length; i++)
{
var child = body.ChildNodes[i];
if (child is HTMLAnchorElement && child.Attributes["name"] != null)
{
switch (child.Attributes["name"].Value)
{
case "valid":
mode = Mode.Valid;
break;
case "invalid":
mode = Mode.Invalid;
break;
case "not-wf":
mode = Mode.NotWf;
break;
case "error":
mode = Mode.Error;
break;
default:
continue;
}
test.tests.Add(mode, new List<Entry>());
}
else if (mode != Mode.None && child is HTMLTableElement)
{
var list = test.tests[mode];
Inspect((HTMLTableElement)child, list);
}
}
return test;
}
public override Node CloneNode(Boolean deep = true)
{
var node = new HTMLDocument();
CopyProperties(this, node, deep);
CopyDocumentProperties(this, node, deep);
node.embedded = this.embedded;
node.scripting = this.scripting;
return node;
}
public void HtmlHasRightBodyElement()
{
var doc = new HTMLDocument();
var root = new HTMLHtmlElement();
doc.AppendChild(root);
var body = new HTMLBodyElement();
root.AppendChild(body);
Assert.AreEqual(body, doc.Body);
}
protected override async Task Use(Uri url, HTMLDocument document, CancellationToken cancel)
{
await Task.Yield();
}
public override Node CloneNode(Boolean deep = true)
{
var node = new HTMLDocument();
CopyProperties(this, node, deep);
CopyDocumentProperties(this, node, deep);
return node;
}
