internal static string HtmlWordpressElement(this XElement element, WordpressNamespaces namespaces, string name)
{
var wElement = WordpressElement(element, namespaces, name);
if (wElement == null)
{
return(null);
}
var data = wElement.FirstNode as XCData;
if (data == null)
{
return(null);
}
var value = data.Value.Trim();
if (string.IsNullOrWhiteSpace(value))
{
return(null);
}
return(HttpUtility.HtmlDecode(value));
}
private void GetCategories(Blog blog, WordpressNamespaces namespaces, XElement channel) {
// Loop through the elements and build the category list.
IEnumerable<XElement> categories =
from cat in channel.Elements(namespaces.WpNamespace + "category")
select cat;
foreach (XElement categoryElement in categories) {
var createdCategory = InternalSchemaAssemblers.AssembleCategory(namespaces, categoryElement);
if (blog.Categories.CategoryList.All(o => o.ID != createdCategory.ID))
blog.Categories.CategoryList.Add(createdCategory);
}
// WordPress stores the parent category as the description, but BlogML wants the ID. Now that we have a
// complete list, we'll go back through and fix them.
IEnumerable<Category> children =
from a in blog.Categories.CategoryList
where a.ParentCategory != null
select a;
foreach (Category child in children) {
IEnumerable<Category> parent =
from a in blog.Categories.CategoryList
where a.Title == child.ParentCategory
select a;
if (parent.Any()) {
child.ParentCategory = parent.ElementAt(0).ID;
}
}
}
private Blog CreateTopLevelBlog(WordpressNamespaces namespaces, XElement channel)
{
Blog blog = new Blog(_clock);
blog.DateCreated = DateTime.Parse(Constants.ParseRssDate(channel.Element("pubDate").Value));
blog.Title = new Title();
blog.Title.Value = channel.Element("title").Value;
blog.SubTitle = new Title();
blog.SubTitle.Value = channel.Element("description").Value;
// This is the first element we use the WP namespace; make sure it works.
// (See issue #4 - thanks to stephenway for reporting it.)
var rootUrlElement = channel.WordpressElement(namespaces, "base_blog_url");
if (rootUrlElement == null)
{
throw new NotSupportedException("Unable to determine the blog base URL.");
}
blog.RootURL = rootUrlElement.Value;
blog.Authors = new Authors();
blog.Categories = new Categories();
blog.Tags = new Tags();
blog.Posts = new Posts();
return(blog);
}
private Blog CreateTopLevelBlog(WordpressNamespaces namespaces, XElement channel) {
Blog blog = new Blog(_clock);
blog.DateCreated = DateTime.Parse(Constants.ParseRssDate(channel.Element("pubDate").Value));
blog.Title = new Title();
blog.Title.Value = channel.Element("title").Value;
blog.SubTitle = new Title();
blog.SubTitle.Value = channel.Element("description").Value;
// This is the first element we use the WP namespace; make sure it works.
// (See issue #4 - thanks to stephenway for reporting it.)
var rootUrlElement = channel.WordpressElement(namespaces, "base_blog_url");
if (rootUrlElement == null)
throw new NotSupportedException("Unable to determine the blog base URL.");
blog.RootURL = rootUrlElement.Value;
blog.Authors = new Authors();
blog.Categories = new Categories();
blog.Tags = new Tags();
blog.Posts = new Posts();
return blog;
}
internal static T WordpressElement <T>(this XElement element, WordpressNamespaces namespaces, string name, Func <XElement, T> afterwork)
{
var wElement = WordpressElement(element, namespaces, name);
if (wElement != null)
{
return(afterwork(wElement));
}
return(default(T));
}
internal static Trackback AssembleTrackback(WordpressNamespaces namespaces, XElement trackbackElement)
{
Trackback trackback = new Trackback();
trackback.ID = trackbackElement.WordpressElement(namespaces, "comment_id").Value;
trackback.Title = ((XCData)trackbackElement.WordpressElement(namespaces, "comment_author").FirstNode).Value;
trackback.DateCreated =
DateTime.Parse(trackbackElement.WordpressElement(namespaces, "comment_date_gmt").Value);
trackback.Url = trackbackElement.WordpressElement(namespaces, "comment_author_url").Value;
return(trackback);
}
internal static Tag AssembleTag(WordpressNamespaces namespaces, XElement tagElement) {
var tag = new Tag();
tag.ID =
tagElement.WordpressElement(namespaces, "tag_slug", (e) => e.Value.Trim());
tag.Slug =
tagElement.WordpressElement(namespaces, "tag_slug", (e) => e.Value.Trim());
tag.Title =
tagElement.HtmlWordpressElement(namespaces, "tag_name");
return tag;
}
internal static Tag AssembleTag(WordpressNamespaces namespaces, XElement tagElement)
{
var tag = new Tag();
tag.ID =
tagElement.WordpressElement(namespaces, "tag_slug", (e) => e.Value.Trim());
tag.Slug =
tagElement.WordpressElement(namespaces, "tag_slug", (e) => e.Value.Trim());
tag.Title =
tagElement.HtmlWordpressElement(namespaces, "tag_name");
return(tag);
}
internal static Category AssembleCategory(WordpressNamespaces namespaces, XElement categoryElement) {
var category = new Category();
category.ID =
categoryElement.WordpressElement(namespaces, "category_nicename", (e) => e.Value.Trim());
category.Description =
categoryElement.WordpressElement(namespaces, "category_description", (e) =>((XCData)e.FirstNode).Value.Trim());
category.Title =
categoryElement.HtmlWordpressElement(namespaces, "cat_name");
category.ParentCategory =
categoryElement.WordpressElement(namespaces, "category_parent", (e) => e.Value.Trim());
return category;
}
internal static Category AssembleCategory(WordpressNamespaces namespaces, XElement categoryElement)
{
var category = new Category();
category.ID =
categoryElement.WordpressElement(namespaces, "category_nicename", (e) => e.Value.Trim());
category.Description =
categoryElement.WordpressElement(namespaces, "category_description", (e) => ((XCData)e.FirstNode).Value.Trim());
category.Title =
categoryElement.HtmlWordpressElement(namespaces, "cat_name");
category.ParentCategory =
categoryElement.WordpressElement(namespaces, "category_parent", (e) => e.Value.Trim());
return(category);
}
private void GetPosts(Blog blog, WordpressNamespaces namespaces, XElement channel)
{
IEnumerable <XElement> posts =
from item in channel.Elements("item")
where item.WordpressElement(namespaces, "status").Value == "publish"
select item;
// NGM Might want update urls within content to stop redirects?
foreach (XElement item in posts)
{
Post post = InternalSchemaAssemblers.AssemblePost(namespaces, item);
// We need to get the author reference separately, as we need the AuthorList from the blog.
AuthorReference author = new AuthorReference();
author.ID = GetAuthorReference(blog,
((XText)item.Element(namespaces.DcNamespace + "creator").FirstNode).Value);
post.Authors.AuthorReferenceList.Add(author);
blog.Posts.PostList.Add(post);
}
}
public Blog Assemble(Stream stream) {
var file = XElement.Load(stream);
// Get the RSS channel.
XElement channel = file.Element("channel");
WordpressNamespaces namespaces = new WordpressNamespaces {
ExcerptNamespace = file.GetNamespaceOfPrefix("excerpt"),
ContentNamespace = file.GetNamespaceOfPrefix("content"),
WfwNamespace = file.GetNamespaceOfPrefix("wfw"),
DcNamespace = file.GetNamespaceOfPrefix("dc"),
WpNamespace = file.GetNamespaceOfPrefix("wp")
};
Blog blog = CreateTopLevelBlog(namespaces, channel);
GetCategories(blog, namespaces, channel);
GetTags(blog, namespaces, channel);
GetPosts(blog, namespaces, channel);
return blog;
}
internal static Comment AssembleComment(WordpressNamespaces namespaces, XElement commentElement)
{
Comment comment = new Comment();
// Node (parent) properties.
comment.ID = commentElement.WordpressElement(namespaces, "comment_id").Value;
comment.Title = comment.ID;
comment.DateCreated = DateTime.Parse(commentElement.WordpressElement(namespaces, "comment_date_gmt").Value);
comment.Content = new Content();
comment.Content.Type = Content.TypeHTML;
comment.Content.Value = ((XCData)commentElement.WordpressElement(namespaces, "comment_content").FirstNode).Value;
comment.UserName = ((XCData)commentElement.WordpressElement(namespaces, "comment_author").FirstNode).Value;
string email = commentElement.WordpressElement(namespaces, "comment_author_email").Value;
if (!string.IsNullOrWhiteSpace(email))
{
comment.UserEmail = email;
}
string url = commentElement.WordpressElement(namespaces, "comment_author_url").Value;
if (!string.IsNullOrWhiteSpace(url))
{
comment.UserURL = url;
}
string approved = commentElement.WordpressElement(namespaces, "comment_approved").Value;
if (!string.IsNullOrWhiteSpace(approved))
{
comment.Approved = approved == "1";
}
return(comment);
}
public Blog Assemble(Stream stream)
{
var file = XElement.Load(stream);
// Get the RSS channel.
XElement channel = file.Element("channel");
WordpressNamespaces namespaces = new WordpressNamespaces {
ExcerptNamespace = file.GetNamespaceOfPrefix("excerpt"),
ContentNamespace = file.GetNamespaceOfPrefix("content"),
WfwNamespace = file.GetNamespaceOfPrefix("wfw"),
DcNamespace = file.GetNamespaceOfPrefix("dc"),
WpNamespace = file.GetNamespaceOfPrefix("wp")
};
Blog blog = CreateTopLevelBlog(namespaces, channel);
GetCategories(blog, namespaces, channel);
GetTags(blog, namespaces, channel);
GetPosts(blog, namespaces, channel);
return(blog);
}
private void GetTags(Blog blog, WordpressNamespaces namespaces, XElement channel)
{
// Loop through the elements and build the category list.
IEnumerable <XElement> tags =
from tag in channel.Elements(namespaces.WpNamespace + "tag")
select tag;
foreach (XElement tagElement in tags)
{
var createdTag = InternalSchemaAssemblers.AssembleTag(namespaces, tagElement);
if (blog.Tags.TagList.All(o => o.ID != createdTag.ID))
{
blog.Tags.TagList.Add(createdTag);
}
}
// WordPress stores the parent category as the description, but BlogML wants the ID. Now that we have a
// complete list, we'll go back through and fix them.
IEnumerable <Tag> children =
from a in blog.Tags.TagList
where a.Slug != null
select a;
foreach (Tag child in children)
{
IEnumerable <Tag> parent =
from a in blog.Tags.TagList
where a.Title == child.Slug
select a;
if (0 < parent.Count())
{
child.Slug = parent.ElementAt(0).ID;
}
}
}
private void GetCategories(Blog blog, WordpressNamespaces namespaces, XElement channel)
{
// Loop through the elements and build the category list.
IEnumerable <XElement> categories =
from cat in channel.Elements(namespaces.WpNamespace + "category")
select cat;
foreach (XElement categoryElement in categories)
{
var createdCategory = InternalSchemaAssemblers.AssembleCategory(namespaces, categoryElement);
if (blog.Categories.CategoryList.All(o => o.ID != createdCategory.ID))
{
blog.Categories.CategoryList.Add(createdCategory);
}
}
// WordPress stores the parent category as the description, but BlogML wants the ID. Now that we have a
// complete list, we'll go back through and fix them.
IEnumerable <Category> children =
from a in blog.Categories.CategoryList
where a.ParentCategory != null
select a;
foreach (Category child in children)
{
IEnumerable <Category> parent =
from a in blog.Categories.CategoryList
where a.Title == child.ParentCategory
select a;
if (parent.Any())
{
child.ParentCategory = parent.ElementAt(0).ID;
}
}
}
internal static Post AssemblePost(WordpressNamespaces namespaces, XElement postElement)
{
Post post = new Post();
// Node (parent) properties.
post.ID = postElement.WordpressElement(namespaces, "post_id").Value;
post.Title = postElement.Element("title").Value;
post.DateCreated = DateTime.Parse(Constants.ParseRssDate(postElement.Element("pubDate").Value));
post.DateModified = DateTime.Parse(Constants.ParseRssDate(postElement.Element("pubDate").Value));
post.PostUrl = (new Uri(postElement.Element("link").Value).AbsolutePath).TrimStart('/'); // NGM
post.Approved = true;
// Object properties.
post.Content = new Content();
post.Content.Type = Content.TypeHTML;
post.Content.Value = ((XCData)postElement.ContentElement(namespaces, "encoded").FirstNode).Value;
post.PostName = new Title();
post.PostName.Type = Content.TypeHTML;
post.PostName.Value = postElement.Element("title").Value;
string excerpt = ((XCData)postElement.Element(namespaces.ExcerptNamespace + "encoded").FirstNode).Value;
if (String.Empty == excerpt)
{
post.HasExcerpt = false;
}
else
{
post.Excerpt = new Content();
post.Excerpt.Type = Content.TypeHTML;
post.Excerpt.Value = excerpt;
post.HasExcerpt = true;
}
// Category references.
IEnumerable <XElement> categories =
from cat in postElement.Elements("category")
where cat.Attribute("domain") != null && cat.Attribute("domain").Value == "category"
select cat;
if (categories.Any())
{
post.Categories = new CategoryReferences();
foreach (XElement reference in categories)
{
post.Categories.CategoryReferenceList.Add(AssembleCategoryReference(reference));
}
}
// Tag references.
IEnumerable <XElement> tags =
from tag in postElement.Elements("category")
where tag.Attribute("domain") != null && (tag.Attribute("domain").Value == "post_tag" || tag.Attribute("domain").Value == "tag") && tag.Attribute("nicename") != null
select tag;
if (tags.Any())
{
post.Tags = new TagReferences();
foreach (XElement reference in tags)
{
post.Tags.TagReferenceList.Add(new TagReference(reference));
}
}
// Comments on this post.
IEnumerable <XElement> comments =
from comment in postElement.Elements(namespaces.WpNamespace + "comment")
where comment.WordpressElement(namespaces, "comment_approved").Value == "1" &&
string.IsNullOrEmpty(comment.WordpressElement(namespaces, "comment_type").Value)
select comment;
post.Comments = new Comments();
if (comments.Any())
{
foreach (XElement comment in comments)
{
post.Comments.CommentList.Add(AssembleComment(namespaces, comment));
}
}
if (postElement.WordpressElement(namespaces, "comment_status").Value == "open")
{
post.Comments.Enabled = true;
}
else
{
post.Comments.Enabled = false;
}
// Trackbacks for this post.
IEnumerable <XElement> trackbax =
from tb in postElement.Elements(namespaces.WpNamespace + "comment")
where tb.WordpressElement(namespaces, "comment_approved").Value == "1" &&
((tb.WordpressElement(namespaces, "comment_type").Value == "trackback") ||
(tb.WordpressElement(namespaces, "comment_type").Value == "pingback"))
select tb;
if (trackbax.Any())
{
post.Trackbacks = new Trackbacks();
foreach (XElement trackback in trackbax)
{
post.Trackbacks.TrackbackList.Add(AssembleTrackback(namespaces, trackback));
}
}
post.Authors = new AuthorReferences();
post.Authors.AuthorReferenceList.Add(new AuthorReference {
ID = postElement.Element(namespaces.DcNamespace + "creator").Value
});
return(post);
}
internal static XElement WordpressElement(this XElement element, WordpressNamespaces namespaces, string name)
{
return(element.Element(namespaces.WpNamespace + name));
}
private void GetPosts(Blog blog, WordpressNamespaces namespaces, XElement channel) {
IEnumerable<XElement> posts =
from item in channel.Elements("item")
where item.WordpressElement(namespaces, "status").Value == "publish"
select item;
// NGM Might want update urls within content to stop redirects?
foreach (XElement item in posts) {
Post post = InternalSchemaAssemblers.AssemblePost(namespaces, item);
// We need to get the author reference separately, as we need the AuthorList from the blog.
AuthorReference author = new AuthorReference();
author.ID = GetAuthorReference(blog,
((XText)item.Element(namespaces.DcNamespace + "creator").FirstNode).Value);
post.Authors.AuthorReferenceList.Add(author);
blog.Posts.PostList.Add(post);
}
}
internal static Trackback AssembleTrackback(WordpressNamespaces namespaces, XElement trackbackElement) {
Trackback trackback = new Trackback();
trackback.ID = trackbackElement.WordpressElement(namespaces, "comment_id").Value;
trackback.Title = ((XCData)trackbackElement.WordpressElement(namespaces, "comment_author").FirstNode).Value;
trackback.DateCreated =
DateTime.Parse(trackbackElement.WordpressElement(namespaces, "comment_date_gmt").Value);
trackback.Url = trackbackElement.WordpressElement(namespaces, "comment_author_url").Value;
return trackback;
}
internal static Post AssemblePost(WordpressNamespaces namespaces, XElement postElement) {
Post post = new Post();
// Node (parent) properties.
post.ID = postElement.WordpressElement(namespaces, "post_id").Value;
post.Title = postElement.Element("title").Value;
post.DateCreated = DateTime.Parse(Constants.ParseRssDate(postElement.Element("pubDate").Value));
post.DateModified = DateTime.Parse(Constants.ParseRssDate(postElement.Element("pubDate").Value));
post.PostUrl = (new Uri(postElement.Element("link").Value).AbsolutePath).TrimStart('/'); // NGM
post.Approved = true;
// Object properties.
post.Content = new Content();
post.Content.Type = Content.TypeHTML;
post.Content.Value = ((XCData)postElement.ContentElement(namespaces, "encoded").FirstNode).Value;
post.PostName = new Title();
post.PostName.Type = Content.TypeHTML;
post.PostName.Value = postElement.Element("title").Value;
string excerpt = ((XCData)postElement.Element(namespaces.ExcerptNamespace + "encoded").FirstNode).Value;
if (String.Empty == excerpt) {
post.HasExcerpt = false;
}
else {
post.Excerpt = new Content();
post.Excerpt.Type = Content.TypeHTML;
post.Excerpt.Value = excerpt;
post.HasExcerpt = true;
}
// Category references.
IEnumerable<XElement> categories =
from cat in postElement.Elements("category")
where cat.Attribute("domain") != null && cat.Attribute("domain").Value == "category"
select cat;
if (categories.Any()) {
post.Categories = new CategoryReferences();
foreach(XElement reference in categories) {
post.Categories.CategoryReferenceList.Add(AssembleCategoryReference(reference));
}
}
// Tag references.
IEnumerable<XElement> tags =
from tag in postElement.Elements("category")
where tag.Attribute("domain") != null && (tag.Attribute("domain").Value == "post_tag" || tag.Attribute("domain").Value == "tag") && tag.Attribute("nicename") != null
select tag;
if (tags.Any()) {
post.Tags = new TagReferences();
foreach (XElement reference in tags) {
post.Tags.TagReferenceList.Add(new TagReference(reference));
}
}
// Comments on this post.
IEnumerable<XElement> comments =
from comment in postElement.Elements(namespaces.WpNamespace + "comment")
where comment.WordpressElement(namespaces, "comment_approved").Value == "1"
&& string.IsNullOrEmpty(comment.WordpressElement(namespaces, "comment_type").Value)
select comment;
post.Comments = new Comments();
if (comments.Any()) {
foreach(XElement comment in comments) {
post.Comments.CommentList.Add(AssembleComment(namespaces, comment));
}
}
if (postElement.WordpressElement(namespaces, "comment_status").Value == "open")
post.Comments.Enabled = true;
else {
post.Comments.Enabled = false;
}
// Trackbacks for this post.
IEnumerable<XElement> trackbax =
from tb in postElement.Elements(namespaces.WpNamespace + "comment")
where tb.WordpressElement(namespaces, "comment_approved").Value == "1"
&& ((tb.WordpressElement(namespaces, "comment_type").Value == "trackback")
|| (tb.WordpressElement(namespaces, "comment_type").Value == "pingback"))
select tb;
if (trackbax.Any()) {
post.Trackbacks = new Trackbacks();
foreach(XElement trackback in trackbax) {
post.Trackbacks.TrackbackList.Add(AssembleTrackback(namespaces, trackback));
}
}
post.Authors = new AuthorReferences();
post.Authors.AuthorReferenceList.Add(new AuthorReference { ID = postElement.Element(namespaces.DcNamespace + "creator").Value });
return post;
}
internal static Comment AssembleComment(WordpressNamespaces namespaces, XElement commentElement) {
Comment comment = new Comment();
// Node (parent) properties.
comment.ID = commentElement.WordpressElement(namespaces, "comment_id").Value;
comment.Title = comment.ID;
comment.DateCreated = DateTime.Parse(commentElement.WordpressElement(namespaces, "comment_date_gmt").Value);
comment.Content = new Content();
comment.Content.Type = Content.TypeHTML;
comment.Content.Value = ((XCData)commentElement.WordpressElement(namespaces, "comment_content").FirstNode).Value;
comment.UserName = ((XCData)commentElement.WordpressElement(namespaces, "comment_author").FirstNode).Value;
string email = commentElement.WordpressElement(namespaces, "comment_author_email").Value;
if (!string.IsNullOrWhiteSpace(email)) {
comment.UserEmail = email;
}
string url = commentElement.WordpressElement(namespaces, "comment_author_url").Value;
if (!string.IsNullOrWhiteSpace(url)) {
comment.UserURL = url;
}
string approved = commentElement.WordpressElement(namespaces, "comment_approved").Value;
if (!string.IsNullOrWhiteSpace(approved)) {
comment.Approved = approved == "1";
}
return comment;
}
private void GetTags(Blog blog, WordpressNamespaces namespaces, XElement channel) {
// Loop through the elements and build the category list.
IEnumerable<XElement> tags =
from tag in channel.Elements(namespaces.WpNamespace + "tag")
select tag;
foreach (XElement tagElement in tags) {
var createdTag = InternalSchemaAssemblers.AssembleTag(namespaces, tagElement);
if (blog.Tags.TagList.All(o => o.ID != createdTag.ID))
blog.Tags.TagList.Add(createdTag);
}
// WordPress stores the parent category as the description, but BlogML wants the ID. Now that we have a
// complete list, we'll go back through and fix them.
IEnumerable<Tag> children =
from a in blog.Tags.TagList
where a.Slug != null
select a;
foreach (Tag child in children) {
IEnumerable<Tag> parent =
from a in blog.Tags.TagList
where a.Title == child.Slug
select a;
if (0 < parent.Count()) {
child.Slug = parent.ElementAt(0).ID;
}
}
}
