public static LightWeightHTMLDocument FromIHTMLDocument2(IHTMLDocument2 htmlDocument, string url, bool escapePaths, bool escapeEmptyString)
{
if (htmlDocument == null)
{
return(null);
}
return(LightWeightHTMLDocument.FromIHTMLDocument2(htmlDocument, url, null, escapePaths, escapeEmptyString));
}
public static LightWeightHTMLDocument FromIHTMLDocument2(IHTMLDocument2 htmlDocument, string url)
{
if (htmlDocument == null)
{
return(null);
}
return(LightWeightHTMLDocument.FromIHTMLDocument2(htmlDocument, url, true));
}
public static LightWeightHTMLDocument[] GetLightWeightDocumentForFrames(IHTMLDocument2 htmlDocument)
{
ArrayList frameLightWeightDocuments = new ArrayList();
// Get the IOleContainer for the for the html document (this requires that
// the document is the root document in the browser)
IOleContainer oleContainer = (IOleContainer)htmlDocument;
IEnumUnknown enumUnknown;
// Enumerate the controls in the browser
oleContainer.EnumObjects(OLECONTF.EMBEDDINGS, out enumUnknown);
// Iterate through the controls
object unknown;
for (int i = 0; HRESULT.S_OK == enumUnknown.Next(1, out unknown, IntPtr.Zero); i++)
{
// Only subframes should cast to IWebBrowser2
IWebBrowser2 webBrowser = unknown as IWebBrowser2;
// Since it is a subframe, we can also get the base frame implementation for it
IHTMLFrameBase frameBase = unknown as IHTMLFrameBase;
// It's a frame, add this to the list!
if (webBrowser != null)
{
try
{
IHTMLDocument2 frameDocument = webBrowser.Document as IHTMLDocument2;
if (frameDocument != null)
{
LightWeightHTMLDocument document = LightWeightHTMLDocument.FromIHTMLDocument2(frameDocument, frameDocument.url, frameBase.name);
if (document != null)
{
frameLightWeightDocuments.Add(document);
}
}
}
catch (InvalidCastException)
{
string html = "<HTML></HTML>";
LightWeightHTMLDocument document = LightWeightHTMLDocument.FromString(html, webBrowser.LocationURL, webBrowser.LocationURL, true);
if (document != null)
{
frameLightWeightDocuments.Add(document);
}
}
}
}
return((LightWeightHTMLDocument[])frameLightWeightDocuments.ToArray(typeof(LightWeightHTMLDocument)));
}
public string Capture(int timeoutMs)
{
// flag indicating whether we should continue with the capture
bool continueCapture = true;
// request the page
HttpWebResponse response = RequestPage(TargetUrl, timeoutMs);
OnHeadersReceived(response.Headers, ref continueCapture);
if (!continueCapture)
{
throw new OperationCancelledException();
}
// transfer it to a stream
MemoryStream pageStream = new MemoryStream();
using (Stream responseStream = response.GetResponseStream())
StreamHelper.Transfer(responseStream, pageStream);
pageStream.Seek(0, SeekOrigin.Begin);
// allow filter on content
OnContentReceived(new StreamReader(pageStream).ReadToEnd(), ref continueCapture);
if (!continueCapture)
{
throw new OperationCancelledException();
}
pageStream.Seek(0, SeekOrigin.Begin);
// Read the stream into a lightweight HTML doc. We use from LightWeightHTMLDocument.FromIHTMLDocument2
// instead of LightWeightHTMLDocument.FromStream because from stream improperly shoves a saveFrom declaration
// above the docType (bug 289357)
IHTMLDocument2 doc = HTMLDocumentHelper.StreamToHTMLDoc(pageStream, TargetUrl, false);
LightWeightHTMLDocument ldoc = LightWeightHTMLDocument.FromIHTMLDocument2(doc, TargetUrl, true);
// download references
FileBasedSiteStorage siteStorage = new FileBasedSiteStorage(DestinationPath, "index.htm");
PageToDownload page = new PageToDownload(ldoc, TargetUrl, siteStorage.RootFile);
PageAndReferenceDownloader downloader = new PageAndReferenceDownloader(new PageToDownload[] { page }, siteStorage);
downloader.Download(new TimeoutProgressHost(timeoutMs));
// return path to captured page
return(Path.Combine(DestinationPath, siteStorage.RootFile));
}
private PageToDownload DownloadUrl(string url, PageToDownload parent, IProgressHost progress)
{
PageToDownload thisPageToDownload = null;
// Download the current page
LightWeightHTMLDocument lightWeightDoc = null;
using (HTMLDocumentDownloader downloader = new HTMLDocumentDownloader(_parentControl, url, null, _context.CookieString, _context.TimeoutMS, true))
{
downloader.DownloadHTMLDocument(progress);
lightWeightDoc = LightWeightHTMLDocument.FromIHTMLDocument2(downloader.HtmlDocument, downloader.Url);
thisPageToDownload = new PageToDownload(lightWeightDoc, url, null, parent);
// Reset the url in the event that a redirect occurred
thisPageToDownload.AbsoluteUrl = downloader.Url;
}
foreach (HTMLDocumentHelper.ResourceUrlInfo styleUrl in lightWeightDoc.StyleResourcesUrls)
{
thisPageToDownload.AddReference(new ReferenceToDownload(styleUrl.ResourceUrl, thisPageToDownload, styleUrl.ResourceAbsoluteUrl));
}
return(thisPageToDownload);
}
private static LightWeightHTMLDocument FromIHTMLDocument2(IHTMLDocument2 htmlDocument, string url, string name, bool escapePaths)
{
return(LightWeightHTMLDocument.FromIHTMLDocument2(htmlDocument, url, name, escapePaths, true));
}
private static LightWeightHTMLDocument FromIHTMLDocument2(IHTMLDocument2 htmlDocument, string url, string name)
{
return(LightWeightHTMLDocument.FromIHTMLDocument2(htmlDocument, url, name, true));
}
