Replace hpricot with nokogiri in wordpressdotcom

jekyll-import.gemspec

@@ -51,7 +51,6 @@ Gem::Specification.new do |s|
 
     # importer dependencies:
     # s.add_development_dependency("behance", "~> 0.3") # uses outdated dependencies
-    s.add_development_dependency("hpricot", "~> 0.8")
     s.add_development_dependency("htmlentities", "~> 4.3")
     s.add_development_dependency("mysql2", "~> 0.3")
     s.add_development_dependency("open_uri_redirections", "~> 0.2")

lib/jekyll-import/importers/wordpressdotcom.rb

@@ -8,7 +8,7 @@ def self.require_deps
           rubygems
           fileutils
           safe_yaml
-          hpricot
+          nokogiri
           time
           open-uri
           open_uri_redirections
@@ -22,8 +22,8 @@ def self.specify_options(c)
       end
 
       # Will modify post DOM tree
-      def self.download_images(title, post_hpricot, assets_folder)
-        images = (post_hpricot / "img")
+      def self.download_images(title, post_doc, assets_folder)
+        images = post_doc.css("img")
         return if images.empty?
 
         Jekyll.logger.info "Downloading images for ", title
@@ -58,11 +58,11 @@ def initialize(node)
         end
 
         def text_for(path)
-          @node.at(path).inner_text
+          @node.at(path).text
         end
 
         def title
-          @title ||= text_for(:title).strip
+          @title ||= text_for("title").strip
         end
 
         def permalink_title
@@ -76,11 +76,9 @@ def permalink_title
         end
 
         def permalink
-          # Hpricot thinks "link" is a self closing tag so it puts the text of the link after the tag
-          # but sometimes it works right! I think it's the xml declaration
           @permalink ||= begin
             uri = text_for("link")
-            uri = @node.at("link").following[0] if uri.empty?
+            uri = @node.at("link").next_sibling.text if uri.empty?
             URI(uri.to_s).path
           end
         end
@@ -127,12 +125,8 @@ def published?
 
         def excerpt
           @excerpt ||= begin
-            text = Hpricot(text_for("excerpt:encoded")).inner_text
-            if text.empty?
-              nil
-            else
-              text
-            end
+            text = Nokogiri::HTML(text_for("excerpt:encoded")).text
+            text.empty? ? nil : text
           end
         end
       end
@@ -144,33 +138,36 @@ def self.process(options)
         FileUtils.mkdir_p(assets_folder)
 
         import_count = Hash.new(0)
-        doc = Hpricot::XML(File.read(source))
+        doc = Nokogiri::XML(File.read(source))
         # Fetch authors data from header
         authors = Hash[
-          (doc / :channel / "wp:author").map do |author|
-            [author.at("wp:author_login").inner_text.strip, {
-              "login"        => author.at("wp:author_login").inner_text.strip,
-              "email"        => author.at("wp:author_email").inner_text,
-              "display_name" => author.at("wp:author_display_name").inner_text,
-              "first_name"   => author.at("wp:author_first_name").inner_text,
-              "last_name"    => author.at("wp:author_last_name").inner_text,
-            },]
+          doc.css("channel > wp|author").map do |author|
+            [
+              author.at("wp|author_login").text.strip,
+              {
+                "login"        => author.at("wp|author_login").text.strip,
+                "email"        => author.at("wp|author_email").text,
+                "display_name" => author.at("wp|author_display_name").text,
+                "first_name"   => author.at("wp|author_first_name").text,
+                "last_name"    => author.at("wp|author_last_name").text,
+              }
+            ]
           end
         ] rescue {}
 
-        (doc / :channel / :item).each do |node|
+        doc.css("channel > item").each do |node|
           item = Item.new(node)
-          categories = node.search('category[@domain="category"]').map(&:inner_text).reject { |c| c == "Uncategorized" }.uniq
-          tags = node.search('category[@domain="post_tag"]').map(&:inner_text).uniq
+          categories = node.css('category[domain="category"]').map(&:text).reject { |c| c == "Uncategorized" }.uniq
+          tags = node.css('category[domain="post_tag"]').map(&:text).uniq
 
           metas = {}
-          node.search("wp:postmeta").each do |meta|
-            key = meta.at("wp:meta_key").inner_text
-            value = meta.at("wp:meta_value").inner_text
+          node.css("wp|postmeta").each do |meta|
+            key = meta.at("wp|meta_key").text
+            value = meta.at("wp|meta_value").text
             metas[key] = value
           end
 
-          author_login = item.text_for("dc:creator").strip
+          author_login = item.text_for("dc|creator").strip
 
           header = {
             "layout"     => item.post_type,
@@ -189,7 +186,7 @@ def self.process(options)
           }
 
           begin
-            content = Hpricot(item.text_for("content:encoded"))
+            content = Nokogiri::HTML(item.text_for("content:encoded"))
             header["excerpt"] = item.excerpt if item.excerpt
 
             if fetch