public override INode Clone(Boolean deep = true)
{
var node = new HtmlDocument(Context, new TextSource(Source.Text));
CopyProperties(this, node, deep);
CopyDocumentProperties(this, node, deep);
return node;
}
public override INode Clone(Boolean deep = true)
{
var source = new TextSource(Source.Text);
var node = new HtmlDocument(Context, source);
CloneDocument(node, deep);
return node;
}
public void CreateMap()
{
var document = new HtmlDocument();
a = new HtmlElement(document, "a");
a.SetAttribute("data-test1", "test");
a.SetAttribute("data-b", "b");
stringMap = new StringMap("data-", a);
}
/// <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>
internal HtmlDomBuilder(HtmlDocument document)
{
_tokenizer = new HtmlTokenizer(document.Source, document.Options.Events);
_document = document;
_openElements = new List<Element>();
_templateModes = new Stack<HtmlTreeMode>();
_formattingElements = new List<Element>();
_frameset = true;
_currentMode = HtmlTreeMode.Initial;
}
/// <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>
internal HtmlDomBuilder(HtmlDocument document)
{
var resolver = document.Options.GetService<IEntityService>() ?? HtmlEntityService.Resolver;
_tokenizer = new HtmlTokenizer(document.Source, document.Options.Events, resolver);
_document = document;
_openElements = new List<Element>();
_templateModes = new Stack<HtmlTreeMode>();
_formattingElements = new List<Element>();
_frameset = true;
_currentMode = HtmlTreeMode.Initial;
}
internal async static Task<IDocument> LoadAsync(IBrowsingContext context, CreateDocumentOptions options, CancellationToken cancelToken)
{
var scripting = context.Configuration.IsScripting();
var parserOptions = new HtmlParserOptions { IsScripting = scripting };
var document = new HtmlDocument(context, options.Source);
var parser = new HtmlDomBuilder(document);
document.Setup(options);
context.NavigateTo(document);
context.Fire(new HtmlParseEvent(document, completed: false));
await parser.ParseAsync(parserOptions, cancelToken).ConfigureAwait(false);
context.Fire(new HtmlParseEvent(document, completed: true));
return document;
}
/// <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>
public HtmlDomBuilder(HtmlDocument document)
{
var options = document.Options;
var context = document.Context;
var resolver = options.GetProvider<IEntityProvider>() ?? HtmlEntityService.Resolver;
_tokenizer = new HtmlTokenizer(document.Source, resolver);
_tokenizer.Error += (_, error) => context.Fire(error);
_document = document;
_openElements = new List<Element>();
_templateModes = new Stack<HtmlTreeMode>();
_formattingElements = new List<Element>();
_frameset = true;
_currentMode = HtmlTreeMode.Initial;
_htmlFactory = options.GetFactory<IElementFactory<HtmlElement>>();
_mathFactory = options.GetFactory<IElementFactory<MathElement>>();
_svgFactory = options.GetFactory<IElementFactory<SvgElement>>();
}
/// <summary>
/// Creates a new document fragment with the given nodelist as
/// children.
/// </summary>
/// <param name="context">The context for the fragment mode.</param>
/// <param name="html">The HTML source code to use.</param>
internal DocumentFragment(Element context, String html)
: this(context.Owner)
{
var source = new TextSource(html);
var document = new HtmlDocument(Owner.Context, source);
var parser = new HtmlDomBuilder(document);
var options = new HtmlParserOptions
{
IsEmbedded = false,
IsScripting = Owner.Options.IsScripting()
};
var root = parser.ParseFragment(options, context).DocumentElement;
while (root.HasChildNodes)
{
var child = root.FirstChild;
root.RemoveChild(child);
this.PreInsert(child, null);
}
}
internal async static Task<IDocument> LoadTextAsync(IBrowsingContext context, CreateDocumentOptions options, CancellationToken cancelToken)
{
var scripting = context.Configuration.IsScripting();
var parserOptions = new HtmlParserOptions { IsScripting = scripting };
var document = new HtmlDocument(context, options.Source);
document.Setup(options);
context.NavigateTo(document);
var root = document.CreateElement(TagNames.Html);
var head = document.CreateElement(TagNames.Head);
var body = document.CreateElement(TagNames.Body);
var pre = document.CreateElement(TagNames.Pre);
document.AppendChild(root);
root.AppendChild(head);
root.AppendChild(body);
body.AppendChild(pre);
pre.SetAttribute(AttributeNames.Style, "word-wrap: break-word; white-space: pre-wrap;");
await options.Source.PrefetchAllAsync(cancelToken).ConfigureAwait(false);
pre.TextContent = options.Source.Text;
return document;
}
/// <summary>
/// Loads the document in the provided context from the given response.
/// </summary>
/// <param name="context">The browsing context.</param>
/// <param name="response">The response to consider.</param>
/// <param name="source">The source to use.</param>
/// <param name="cancelToken">Token for cancellation.</param>
/// <returns>The task that builds the document.</returns>
internal async static Task<HtmlDocument> LoadAsync(IBrowsingContext context, IResponse response, TextSource source, CancellationToken cancelToken)
{
var document = new HtmlDocument(context, source);
var evt = new HtmlParseStartEvent(document);
var config = context.Configuration;
var events = config.Events;
var parser = new HtmlDomBuilder(document);
document.ContentType = response.Headers.GetOrDefault(HeaderNames.ContentType, MimeTypes.Html);
document.Referrer = response.Headers.GetOrDefault(HeaderNames.Referer, String.Empty);
document.DocumentUri = response.Address.Href;
document.Cookie = response.Headers.GetOrDefault(HeaderNames.SetCookie, String.Empty);
document.ReadyState = DocumentReadyState.Loading;
if (events != null)
events.Publish(evt);
var options = new HtmlParserOptions { IsScripting = config.IsScripting() };
await parser.ParseAsync(options, cancelToken).ConfigureAwait(false);
evt.FireEnd();
return document;
}
public static INodeList ToHtmlFragment(this String sourceCode, IElement context = null, IConfiguration configuration = null)
{
var ctx = BrowsingContext.New(configuration);
var source = new TextSource(sourceCode);
var document = new HtmlDocument(ctx, source);
var parser = new HtmlDomBuilder(sourceCode, configuration);
var element = context as Element;
if (element != null)
{
var options = new HtmlParserOptions
{
IsEmbedded = false,
IsScripting = configuration.IsScripting()
};
return parser.ParseFragment(options, element).DocumentElement.ChildNodes;
}
else
{
return parser.Parse(default(HtmlParserOptions)).ChildNodes;
}
}
public void CSSStyleDeclarationBoundInboundDirection()
{
var context = BrowsingContext.New(new Configuration().WithCss());
var document = new HtmlDocument(context);
var element = new HtmlSpanElement(document);
var text = "background-color: rgb(255, 0, 0); color: rgb(0, 0, 0)";
element.Style.CssText = text;
Assert.AreEqual(text, element.GetAttribute("style"));
Assert.AreEqual(2, element.Style.Length);
}
public void DOMTokenListWritesBack()
{
var document = new HtmlDocument();
var testClass = "myclass";
var div = new HtmlDivElement(document);
div.ClassName = "";
div.ClassList.Add(testClass);
Assert.AreEqual(testClass, div.ClassName);
}
public void HtmlButtonToggleProperties()
{
var document = new HtmlDocument();
var form = document.CreateElement<IHtmlFormElement>();
var element = document.CreateElement<IHtmlButtonElement>();
document.AppendChild(form);
form.AppendChild(element);
Assert.IsFalse(element.IsDisabled);
Assert.IsFalse(element.IsDraggable);
Assert.AreEqual(String.Empty, element.FormMethod);
element.IsDisabled = true;
element.IsDraggable = true;
element.FormMethod = "get";
Assert.IsTrue(element.IsDisabled);
Assert.IsTrue(element.IsDraggable);
Assert.AreEqual("get", element.FormMethod);
}
public void CSSStyleDeclarationBoundOutboundDirectionDirect()
{
var context = BrowsingContext.New(new Configuration().WithCss());
var document = new HtmlDocument(context);
var element = document.CreateElement<IHtmlSpanElement>();
var text = "background-color: rgb(255, 0, 0); color: rgb(0, 0, 0)";
element.SetAttribute("style", String.Empty);
Assert.AreEqual(String.Empty, element.Style.CssText);
element.SetAttribute("style", text);
Assert.AreEqual(text, element.Style.CssText);
Assert.AreEqual(2, element.Style.Length);
}
public void DOMStringMapBindingSet()
{
var document = new HtmlDocument();
var value = "SomeUser";
var div = new HtmlDivElement(document);
div.Dataset["user"] = value;
Assert.AreEqual(div.GetAttribute("data-user"), value);
}
public void HtmlAnchorToggleProperties()
{
var document = new HtmlDocument();
var element = document.CreateElement<IHtmlAnchorElement>();
Assert.IsTrue(element.IsTranslated);
Assert.IsFalse(element.IsSpellChecked);
Assert.IsFalse(element.IsContentEditable);
Assert.IsFalse(element.IsDraggable);
Assert.IsFalse(element.IsHidden);
element.IsTranslated = false;
element.IsSpellChecked = true;
element.IsDraggable = true;
element.IsHidden = true;
Assert.IsFalse(element.IsTranslated);
Assert.IsTrue(element.IsSpellChecked);
Assert.IsTrue(element.IsDraggable);
Assert.IsTrue(element.IsHidden);
}
public void HtmlHasRightBodyElement()
{
var document = new HtmlDocument();
var root = new HtmlHtmlElement(document);
document.AppendChild(root);
var body = new HtmlBodyElement(document);
root.AppendChild(body);
Assert.AreEqual(body, document.Body);
}
IWindow IWindow.Open(String url, String name, String features, String replace)
{
//TODO Context ?
var document = new HtmlDocument();
document.Location.Href = url;
return new Window(document) { Name = name };
}
public void HtmlCustomTitleGeneration()
{
var document = new HtmlDocument();
var title = "My Title";
document.Title = title;
Assert.AreEqual("", document.Title);
var html = document.CreateElement(TagNames.Html);
var head = document.CreateElement(TagNames.Head);
document.AppendChild(html);
html.AppendChild(head);
document.Title = title;
Assert.AreEqual(title, document.Title);
}
public void HtmlHasRightHeadElement()
{
var document = new HtmlDocument();
var root = new HtmlHtmlElement(document);
document.AppendChild(root);
var head = new HtmlHeadElement(document);
root.AppendChild(head);
Assert.AreEqual(head, document.Head);
}
public void DOMTokenListCorrectlyInitializedNoClass()
{
var document = new HtmlDocument();
var testClass = "myclass1";
var div = new HtmlDivElement(document);
div.ClassName = "myclass2 whatever anotherclass";
Assert.IsFalse(div.ClassList.Contains(testClass));
}
public void DOMStringMapHasAttributesButRequestedMissing()
{
var document = new HtmlDocument();
var div = new HtmlDivElement(document);
div.SetAttribute("data-some", "test");
div.SetAttribute("data-another", "");
div.SetAttribute("data-test", "third attribute");
Assert.IsTrue(div.Dataset["user"] == null);
}
public void DOMStringMapHasNoAttribute()
{
var document = new HtmlDocument();
var div = new HtmlDivElement(document);
Assert.IsTrue(div.Dataset["user"] == null);
}
/// <summary>
/// Loads the document in the provided context from the given response.
/// </summary>
/// <param name="context">The browsing context.</param>
/// <param name="options">The creation options to consider.</param>
/// <param name="cancelToken">Token for cancellation.</param>
/// <returns>The task that builds the document.</returns>
internal async static Task<IDocument> LoadAsync(IBrowsingContext context, CreateDocumentOptions options, CancellationToken cancelToken)
{
var document = new HtmlDocument(context, options.Source);
var evt = new HtmlParseStartEvent(document);
var config = context.Configuration;
var events = config.Events;
var parser = new HtmlDomBuilder(document);
var parserOptions = new HtmlParserOptions
{
IsScripting = config.IsScripting()
};
document.Setup(options);
context.NavigateTo(document);
if (events != null)
{
events.Publish(evt);
}
await parser.ParseAsync(parserOptions, cancelToken).ConfigureAwait(false);
evt.FireEnd();
return document;
}
public void NormalizeRemovesEmptyTextNodes()
{
var document = new HtmlDocument();
var div = document.CreateElement("div");
div.AppendChild(document.CreateElement("a"));
div.AppendChild(document.CreateTextNode(String.Empty));
div.AppendChild(document.CreateElement("div"));
div.AppendChild(document.CreateTextNode("Hi there!"));
div.AppendChild(document.CreateElement("img"));
div.Normalize();
Assert.AreEqual(div.ChildNodes.Length, 4);
}
HtmlDocument CreateDocument(TextSource textSource)
{
var document = new HtmlDocument(_context, textSource);
return document;
}
public void NormalizeMergeTextNodes()
{
var document = new HtmlDocument();
var div = document.CreateElement("div");
var a = document.CreateElement("a");
a.AppendChild(document.CreateTextNode(""));
a.AppendChild(document.CreateTextNode("Not empty."));
div.AppendChild(a);
div.AppendChild(document.CreateTextNode(""));
div.AppendChild(document.CreateElement("div"));
div.AppendChild(document.CreateTextNode("Certainly not empty!"));
div.AppendChild(document.CreateTextNode("Certainly not empty!"));
div.AppendChild(document.CreateTextNode("Certainly not empty!"));
div.AppendChild(document.CreateTextNode("Certainly not empty!"));
div.AppendChild(document.CreateElement("img"));
div.Normalize();
Assert.AreEqual(div.ChildNodes.Length, 4);
}
public void DOMStringMapIEnumerableWorking()
{
var document = new HtmlDocument();
var div = new HtmlDivElement(document);
div.SetAttribute("data-some", "test");
div.SetAttribute("data-another", "");
div.SetAttribute("data-test", "third attribute");
Assert.AreEqual(3, div.Dataset.Count());
Assert.AreEqual("some", div.Dataset.First().Key);
Assert.AreEqual("test", div.Dataset.First().Value);
Assert.AreEqual("test", div.Dataset.Last().Key);
Assert.AreEqual("third attribute", div.Dataset.Last().Value);
}
public void DOMTokenListToggleWorksTurnOn()
{
var document = new HtmlDocument();
var testClass = "myclass";
var otherClasses = "otherClass someOther more";
var div = new HtmlDivElement(document);
div.ClassName = otherClasses;
div.ClassList.Toggle(testClass);
Assert.AreEqual(div.ClassName, otherClasses + " " + testClass);
}
