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 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;
}
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;
}
}
/// <summary>
/// Parses the given source asynchronously and creates the document.
/// </summary>
/// <param name="options">The options to use for parsing.</param>
/// <param name="cancelToken">The cancellation token to use.</param>
public async Task<HtmlDocument> ParseAsync(HtmlParserOptions options, CancellationToken cancelToken)
{
var source = _document.Source;
var token = default(HtmlToken);
_options = options;
do
{
if (source.Length - source.Index < 1024)
{
await source.Prefetch(8192, cancelToken).ConfigureAwait(false);
}
token = _tokenizer.Get();
Consume(token);
if (_waiting != null)
{
await _waiting.ConfigureAwait(false);
_waiting = null;
}
}
while (token.Type != HtmlTokenType.EndOfFile);
return _document;
}
/// <summary>
/// Creates a new parser with the custom options and the given context.
/// </summary>
/// <param name="options">The options to use.</param>
/// <param name="context">The context to use.</param>
public HtmlParser(HtmlParserOptions options, IBrowsingContext context)
{
_options = options;
_context = context;
}
/// <summary>
/// Creates a new parser with the custom options and configuration.
/// </summary>
/// <param name="options">The options to use.</param>
/// <param name="configuration">The configuration to use.</param>
public HtmlParser(HtmlParserOptions options, IConfiguration configuration)
: this(options, BrowsingContext.New(configuration))
{
}
/// <summary>
/// Creates a new parser with the custom options.
/// </summary>
/// <param name="options">The options to use.</param>
public HtmlParser(HtmlParserOptions options)
: this(options, Configuration.Default)
{
}
/// <summary>
/// Creates a new parser with the custom options and the given context.
/// </summary>
/// <param name="options">The options to use.</param>
/// <param name="context">The context to use.</param>
public HtmlParser(HtmlParserOptions options, IBrowsingContext context)
{
_options = options;
_context = context;
}
/// <summary>
/// Creates a new parser with the custom options and configuration.
/// </summary>
/// <param name="options">The options to use.</param>
/// <param name="configuration">The configuration to use.</param>
public HtmlParser(HtmlParserOptions options, IConfiguration configuration)
: this(options, BrowsingContext.New(configuration))
{
}
public void NormalModeShouldError()
{
var html = @"<!DOCTYPE html>
<title>Test</title>
<body>
<div myAttribute=""blabla>123</div>
</body>";
var errors = new List<HtmlErrorEvent>();
var options = new HtmlParserOptions { IsStrictMode = false };
var context = BrowsingContext.New(Configuration.Default);
context.ParseError += (s, ev) => errors.Add((HtmlErrorEvent)ev);
var parser = new HtmlParser(options, context);
parser.Parse(html);
Assert.AreEqual(1, errors.Count);
}
/// <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);
using (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);
}
return document;
}
/// <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;
}
/// <summary>
/// Parses the given source and creates the document.
/// </summary>
/// <param name="options">The options to use for parsing.</param>
public HtmlDocument Parse(HtmlParserOptions options)
{
var token = default(HtmlToken);
_tokenizer.IsStrictMode = options.IsStrictMode;
_options = options;
do
{
token = _tokenizer.Get();
Consume(token);
_waiting?.Wait();
_waiting = null;
}
while (token.Type != HtmlTokenType.EndOfFile);
return _document;
}
public void StrictModeShouldYieldException()
{
var html = @"<!DOCTYPE html>
<title>Test</title>
<body>
<div myAttribute=""blabla>123</div>
</body>";
var options = new HtmlParserOptions { IsStrictMode = true };
var parser = new HtmlParser(options);
Assert.Catch<HtmlParseException>(() => parser.Parse(html));
}
/// <summary>
/// Parses the given source and creates the document.
/// </summary>
/// <param name="options">The options to use for parsing.</param>
public HtmlDocument Parse(HtmlParserOptions options)
{
var token = default(HtmlToken);
_options = options;
do
{
token = _tokenizer.Get();
Consume(token);
if (_waiting != null)
{
_waiting.Wait();
_waiting = null;
}
}
while (token.Type != HtmlTokenType.EndOfFile);
return _document;
}
/// <summary>
/// Switches to the fragment algorithm with the specified context
/// element. Then parses the given source and creates the document.
/// </summary>
/// <param name="options">The options to use for parsing.</param>
/// <param name="context">
/// The context element where the algorithm is applied to.
/// </param>
public HtmlDocument ParseFragment(HtmlParserOptions options, Element context)
{
if (context == null)
{
throw new ArgumentNullException("context");
}
var tagName = context.LocalName;
if (tagName.IsOneOf(TagNames.Title, TagNames.Textarea))
{
_tokenizer.State = HtmlParseMode.RCData;
}
else if (tagName.IsOneOf(TagNames.Style, TagNames.Xmp, TagNames.Iframe, TagNames.NoEmbed, TagNames.NoFrames))
{
_tokenizer.State = HtmlParseMode.Rawtext;
}
else if (tagName.Is(TagNames.Script))
{
_tokenizer.State = HtmlParseMode.Script;
}
else if (tagName.Is(TagNames.Plaintext))
{
_tokenizer.State = HtmlParseMode.Plaintext;
}
else if (tagName.Is(TagNames.NoScript) && options.IsScripting)
{
_tokenizer.State = HtmlParseMode.Rawtext;
}
var root = new HtmlHtmlElement(_document);
_document.AddNode(root);
_openElements.Add(root);
if (context is HtmlTemplateElement)
{
_templateModes.Push(HtmlTreeMode.InTemplate);
}
Reset(context);
_fragmentContext = context;
_tokenizer.IsAcceptingCharacterData = !AdjustedCurrentNode.Flags.HasFlag(NodeFlags.HtmlMember);
do
{
if (context is HtmlFormElement)
{
_currentFormElement = (HtmlFormElement)context;
break;
}
context = context.ParentElement as Element;
}
while (context != null);
return Parse(options);
}
/// <summary>
/// Creates a new parser with the custom options.
/// </summary>
/// <param name="options">The options to use.</param>
public HtmlParser(HtmlParserOptions options)
: this(options, Configuration.Default)
{
}
