public override BlogPost Parse(XmlElement entryNode, bool includeCategories, Uri documentUri)
{
BlogPost post = new BlogPost();
AtomEntry atomEntry = new AtomEntry(_atomVer, _atomNS, CategoryScheme, _nsMgr, documentUri, entryNode);
post.Title = atomEntry.Title;
post.Excerpt = atomEntry.Excerpt;
post.Id = PostUriToPostId(atomEntry.EditUri);
post.Permalink = atomEntry.Permalink;
string content = atomEntry.ContentHtml;
if (content.Trim() != string.Empty)
{
HtmlExtractor ex = new HtmlExtractor(content);
int start, length;
if (Options.SupportsExtendedEntries && ex.Seek("<a name=\"more\">").Success)
{
start = ex.Element.Offset;
length = ex.Element.Length;
if (ex.Seek("</a>").Success)
{
post.SetContents(content.Substring(0, start), content.Substring(ex.Element.Offset + ex.Element.Length));
}
else
{
post.SetContents(content.Substring(0, start), content.Substring(start + length));
}
}
else
{
post.Contents = content;
}
}
post.DatePublished = atomEntry.PublishDate;
if (Options.SupportsCategories && includeCategories)
post.Categories = atomEntry.Categories;
return post;
}
private BlogPost ParseBlogPost(XmlNode postNode, bool includeCategories)
{
// create blog post
BlogPost blogPost = new BlogPost();
// get node values
blogPost.Id = NodeText(postNode.SelectSingleNode("member[name='postid']/value"));
if (blogPost.Id == String.Empty)
blogPost.Id = NodeText(postNode.SelectSingleNode("member[name='page_id']/value"));
SetPostTitleFromXmlValue(blogPost, NodeText(postNode.SelectSingleNode("member[name='title']/value")));
// attempt to discover the permalink
blogPost.Permalink = NodeText(postNode.SelectSingleNode("member[name='permaLink']/value"));
if (blogPost.Permalink == String.Empty)
blogPost.Permalink = NodeText(postNode.SelectSingleNode("member[name='permalink']/value"));
if (blogPost.Permalink == String.Empty)
blogPost.Permalink = NodeText(postNode.SelectSingleNode("member[name='link']/value"));
// contents and extended contents
string mainContents = NodeText(postNode.SelectSingleNode("member[name='description']/value"));
string extendedContents = NodeText(postNode.SelectSingleNode("member[name='mt_text_more']/value"));
blogPost.SetContents(mainContents, extendedContents);
// date published
XmlNode dateCreatedNode = postNode.SelectSingleNode("member[name='date_created_gmt']/value");
if (dateCreatedNode == null)
dateCreatedNode = postNode.SelectSingleNode("member[name='dateCreated']/value");
blogPost.DatePublished = ParseBlogDate(dateCreatedNode);
// extract comment field
if (Options.SupportsCommentPolicy)
{
string commentPolicy = NodeText(postNode.SelectSingleNode("member[name='mt_allow_comments']/value"));
if (commentPolicy != String.Empty)
{
switch (commentPolicy)
{
case "0":
blogPost.CommentPolicy = BlogCommentPolicy.None;
break;
case "1":
blogPost.CommentPolicy = BlogCommentPolicy.Open;
break;
case "2":
blogPost.CommentPolicy = BlogCommentPolicy.Closed;
break;
default:
Trace.Fail("Unexpected value for mt_allow_comments: " + commentPolicy);
break;
}
}
}
// extract trackback field
if (Options.SupportsPingPolicy)
{
string allowTrackbacks = NodeText(postNode.SelectSingleNode("member[name='mt_allow_pings']/value"));
if (allowTrackbacks != String.Empty)
blogPost.TrackbackPolicy = (allowTrackbacks == "1") ? BlogTrackbackPolicy.Allow : BlogTrackbackPolicy.Deny;
}
// extract keywords field
if (Options.SupportsKeywords)
{
string keywords = NodeText(postNode.SelectSingleNode("member[name='mt_keywords']/value"));
blogPost.Keywords = HtmlUtils.UnEscapeEntities(keywords, HtmlUtils.UnEscapeMode.Default);
}
// extract excerpt field
if (Options.SupportsExcerpt)
{
blogPost.Excerpt = NodeText(postNode.SelectSingleNode("member[name='mt_excerpt']/value"));
}
// extract ping url array
if (Options.SupportsTrackbacks)
{
XmlNode pingUrlsNode = postNode.SelectSingleNode("member[name='mt_tb_ping_urls']/value");
if (pingUrlsNode != null)
{
ArrayList sentPingUrls = new ArrayList();
if (Options.TrackbackDelimiter == TrackbackDelimiter.ArrayElement)
{
XmlNodeList pingUrlNodes = pingUrlsNode.SelectNodes("array/data/value");
foreach (XmlNode node in pingUrlNodes)
sentPingUrls.Add(node.InnerText.Trim());
}
else
{
string delimiter = null;
if (Options.TrackbackDelimiter == TrackbackDelimiter.Space)
delimiter = " ";
else if (Options.TrackbackDelimiter == TrackbackDelimiter.Comma)
delimiter = ",";
sentPingUrls.AddRange(StringHelper.Split(pingUrlsNode.InnerText, delimiter));
}
blogPost.PingUrlsSent = sentPingUrls.ToArray(typeof(string)) as string[];
}
}
// extract slug
if (Options.SupportsSlug)
{
blogPost.Slug = NodeText(postNode.SelectSingleNode("member[name='wp_slug']/value"));
if (blogPost.Slug == String.Empty)
blogPost.Slug = NodeText(postNode.SelectSingleNode("member[name='mt_basename']/value"));
}
// extract password
if (Options.SupportsPassword)
{
blogPost.Password = NodeText(postNode.SelectSingleNode("member[name='wp_password']/value"));
}
// extract author
if (Options.SupportsAuthor)
{
string authorId = NodeText(postNode.SelectSingleNode("member[name='wp_author_id']/value"));
// account for different display name syntax in various calls
string authorName = NodeText(postNode.SelectSingleNode("member[name='wp_author_display_name']/value"));
blogPost.Author = new PostIdAndNameField(authorId, authorName);
}
// extract page parent
if (Options.SupportsPageParent)
{
string pageParentId = NodeText(postNode.SelectSingleNode("member[name='wp_page_parent_id']/value"));
// convert 0 to empty string for parent-id
if (pageParentId == "0")
pageParentId = String.Empty;
string pageParentTitle = NodeText(postNode.SelectSingleNode("member[name='wp_page_parent_title']/value"));
blogPost.PageParent = new PostIdAndNameField(pageParentId, pageParentTitle);
}
// extract page order
if (Options.SupportsPageOrder)
{
blogPost.PageOrder = NodeText(postNode.SelectSingleNode("member[name='wp_page_order']/value"));
}
// extract categories
if (includeCategories && Options.SupportsCategories)
{
if (Options.SupportsCategoriesInline && !Options.SupportsHierarchicalCategories)
{
blogPost.Categories = MetaweblogExtractCategories(blogPost.Id, postNode);
}
else
{
blogPost.Categories = MovableTypeExtractCategories(blogPost.Id, postNode);
}
}
// allow post-processing of extracted blog post
BlogPostReadFilter(blogPost);
// return the post
return blogPost;
}
private void ParsePostContent(XmlNode xmlNode, BlogPost blogPost)
{
// get raw content (decode base64 if necessary)
string content;
XmlNode base64Node = xmlNode.SelectSingleNode("base64");
if (base64Node != null)
{
byte[] contentBytes = Convert.FromBase64String(base64Node.InnerText);
content = _utf8EncodingNoBOM.GetString(contentBytes);
}
else // no base64 encoding, just read text
{
content = xmlNode.InnerText;
}
// parse out the title and contents of the post
HtmlExtractor ex = new HtmlExtractor(content);
if (ex.Seek("<title>").Success)
{
SetPostTitleFromXmlValue(blogPost, ex.CollectTextUntil("title"));
content = content.Substring(ex.Parser.Position).TrimStart('\r', '\n');
}
if (content.Trim() != string.Empty)
{
HtmlExtractor ex2 = new HtmlExtractor(content);
if (Options.SupportsExtendedEntries && ex2.Seek("<lj-cut>").Success)
blogPost.SetContents(content.Substring(0, ex2.Element.Offset), content.Substring(ex2.Element.Offset + ex2.Element.Length));
else
blogPost.Contents = content;
}
}