private BlogEditingTemplate ParseBlogPostIntoTemplate(BlogPostRegionLocatorStrategy regionLocator, IProgressHost progress, string postUrl)
{
progress.UpdateProgress(Res.Get(StringId.ProgressCreatingEditingTemplate));
BlogPostRegions regions = regionLocator.LocateRegionsOnUIThread(progress, postUrl);
IHTMLElement primaryTitleRegion = GetPrimaryEditableTitleElement(regions.BodyRegion, regions.Document, regions.TitleRegions);
// IF
// - primaryTitleRegion is not null (title found)
// - BodyRegion is null (no post body found)
// - AND primaryTitleRegion is a link
if (primaryTitleRegion != null && regions.BodyRegion == null && primaryTitleRegion.tagName.ToLower() == "a")
{
// Title region was detected, but body region was not.
// It is possible that only titles are shown on the homepage
// Try requesting the post itself, and loading regions from the post itself
// HACK Somewhere the 'about:' protocol replaces http/https, replace it again with the correct protocol
var pathMatch = new Regex("^about:(.*)$").Match((primaryTitleRegion as IHTMLAnchorElement).href);
Debug.Assert(pathMatch.Success); // Assert that this URL is to the format we expect
var newPostPath = pathMatch.Groups[1].Value; // Grab the path from the URL
var homepageUri = new Uri(_blogHomepageUrl);
var newPostUrl = $"{homepageUri.Scheme}://{homepageUri.Host}{newPostPath}"; // Recreate the full post URL
// Set the NextTryPostUrl flag in the region locater
// This will indicate to the other thread that another page should be parsed
_nextTryPostUrl = newPostUrl;
return(null);
}
BlogEditingTemplate template = GenerateBlogTemplate((IHTMLDocument3)regions.Document, primaryTitleRegion, regions.TitleRegions, regions.BodyRegion);
progress.UpdateProgress(100, 100);
return(template);
}
private BlogEditingTemplate ParseBlogPostIntoTemplate(BlogPostRegionLocatorStrategy regionLocator, IProgressHost progress)
{
progress.UpdateProgress(Res.Get(StringId.ProgressCreatingEditingTemplate));
BlogPostRegions regions = regionLocator.LocateRegionsOnUIThread(progress);
IHTMLElement primaryTitleRegion = GetPrimaryEditableTitleElement(regions.BodyRegion, regions.Document, regions.TitleRegions);
BlogEditingTemplate template = GenerateBlogTemplate((IHTMLDocument3)regions.Document, primaryTitleRegion, regions.TitleRegions, regions.BodyRegion);
progress.UpdateProgress(100, 100);
return(template);
}
public override BlogPostRegions LocateRegionsOnUIThread(IProgressHost progress)
{
blogHomepageContents.Seek(0, SeekOrigin.Begin);
IHTMLDocument2 doc2 = HTMLDocumentHelper.GetHTMLDocumentFromStream(blogHomepageContents, _blogHomepageUrl);
// Ensure that the document is fully loaded.
// If it is not fully loaded, then viewing its current style is non-deterministic.
DateTime startedDoingEvents = DateTime.Now;
while (!progress.CancelRequested && !HTMLDocumentHelper.IsReady(doc2))
{
if (DateTime.Now.Subtract(startedDoingEvents).TotalMilliseconds > 10000)
{
// Timing out here is not fatal.
Trace.WriteLine("Timed out while loading blog homepage for theme detection.");
break;
}
Application.DoEvents();
}
//The Google/Blogger dynamic templates load the pages dynmaically usig Ajax, so we dont have any template to use.
if (IsUsingDynamicTemplate(doc2))
{
throw new BlogClientAbortGettingTemplateException();
}
IHTMLElement[] titles = FindText(_titleText, doc2.body);
IHTMLElement[] bodies = FindText(_bodyText, doc2.body);
if (titles.Length == 0 || bodies.Length == 0)
{
throw new Exception("Unable to locate blog post elements using most recent post");
}
if (IsSmartContent(bodies[0]))
{
throw new Exception("Most recent post is smart content");
}
BlogPostRegions regions = new BlogPostRegions();
regions.TitleRegions = titles;
//scrub the post body element to avoid improperly including extraneous parent elements
regions.BodyRegion = ScrubPostBodyRegionParentElements(bodies[0]);
regions.Document = doc2;
progress.UpdateProgress(100, 100);
return(regions);
}
private BlogPostRegions ParseBlogPostIntoTemplate(Stream stream, string postSourceUrl, IProgressHost progress)
{
progress.UpdateProgress(Res.Get(StringId.ProgressCreatingEditingTemplate));
//parse the document to create the blog template
IHTMLDocument2 doc2 = HTMLDocumentHelper.GetHTMLDocumentFromStream(stream, postSourceUrl);
IHTMLDocument3 doc = (IHTMLDocument3)doc2;
IHTMLElement[] titleElements = HTMLDocumentHelper.FindElementsContainingText(doc2, TEMPORARY_POST_TITLE_GUID);
IHTMLElement bodyElement = HTMLDocumentHelper.FindElementContainingText(doc2, TEMPORARY_POST_BODY_GUID);
if (bodyElement != null && bodyElement.tagName == "P")
{
//the body element is the <p> we planted, so replace it with a DIV since that will be the safest
//element to have a as parent to all post content.
IHTMLElement div = doc2.createElement("div");
(bodyElement.parentElement as IHTMLDOMNode).replaceChild(div as IHTMLDOMNode, bodyElement as IHTMLDOMNode);
bodyElement = div;
}
//locate the title element. Note that is there are more than 1 copies of the title text detected, we use the one
//that is anchored closest to the left or the body element.
if (titleElements.Length > 0)
{
BlogPostRegions regions = new BlogPostRegions();
regions.Document = (IHTMLDocument)doc;
regions.TitleRegions = titleElements;
regions.BodyRegion = bodyElement;
progress.UpdateProgress(100, 100);
return(regions);
}
else
{
throw new Exception("unable to access test post.");
}
}
public override BlogPostRegions LocateRegionsOnUIThread(IProgressHost progress)
{
blogHomepageContents.Seek(0, SeekOrigin.Begin);
IHTMLDocument2 doc2 = HTMLDocumentHelper.GetHTMLDocumentFromStream(blogHomepageContents, _blogHomepageUrl);
// Ensure that the document is fully loaded.
// If it is not fully loaded, then viewing its current style is non-deterministic.
DateTime startedDoingEvents = DateTime.Now;
while (!progress.CancelRequested && !HTMLDocumentHelper.IsReady(doc2))
{
if (DateTime.Now.Subtract(startedDoingEvents).TotalMilliseconds > 10000)
{
// Timing out here is not fatal.
Trace.WriteLine("Timed out while loading blog homepage for theme detection.");
break;
}
Application.DoEvents();
}
IHTMLElement[] titles = FindText(_titleText, doc2.body);
IHTMLElement[] bodies = FindText(_bodyText, doc2.body);
if (titles.Length == 0 || bodies.Length == 0)
throw new Exception("Unable to locate blog post elements using most recent post");
if (IsSmartContent(bodies[0]))
throw new Exception("Most recent post is smart content");
BlogPostRegions regions = new BlogPostRegions();
regions.TitleRegions = titles;
//scrub the post body element to avoid improperly including extraneous parent elements
regions.BodyRegion = ScrubPostBodyRegionParentElements(bodies[0]);
regions.Document = doc2;
progress.UpdateProgress(100, 100);
return regions;
}
private BlogPostRegions ParseBlogPostIntoTemplate(Stream stream, string postSourceUrl, IProgressHost progress)
{
progress.UpdateProgress(Res.Get(StringId.ProgressCreatingEditingTemplate));
//parse the document to create the blog template
IHTMLDocument2 doc2 = HTMLDocumentHelper.GetHTMLDocumentFromStream(stream, postSourceUrl);
IHTMLDocument3 doc = (IHTMLDocument3)doc2;
IHTMLElement[] titleElements = HTMLDocumentHelper.FindElementsContainingText(doc2, TEMPORARY_POST_TITLE_GUID);
IHTMLElement bodyElement = HTMLDocumentHelper.FindElementContainingText(doc2, TEMPORARY_POST_BODY_GUID);
if (bodyElement != null && bodyElement.tagName == "P")
{
//the body element is the <p> we planted, so replace it with a DIV since that will be the safest
//element to have a as parent to all post content.
IHTMLElement div = doc2.createElement("div");
(bodyElement.parentElement as IHTMLDOMNode).replaceChild(div as IHTMLDOMNode, bodyElement as IHTMLDOMNode);
bodyElement = div;
}
//locate the title element. Note that is there are more than 1 copies of the title text detected, we use the one
//that is anchored closest to the left or the body element.
if (titleElements.Length > 0)
{
BlogPostRegions regions = new BlogPostRegions();
regions.Document = (IHTMLDocument)doc;
regions.TitleRegions = titleElements;
regions.BodyRegion = bodyElement;
progress.UpdateProgress(100, 100);
return regions;
}
else
{
throw new Exception("unable to access test post.");
}
}
