From d5f28162047750ef3271454d0984e163dc07b9ad Mon Sep 17 00:00:00 2001
From: Wesley Moore <wez@wezm.net>
Date: Wed, 18 Nov 2009 13:02:37 +1100
Subject: [PATCH] Implement post importing

---
 importer/import.rb    |  7 ++++-
 importer/wordpress.rb | 61 ++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 63 insertions(+), 5 deletions(-)

diff --git a/importer/import.rb b/importer/import.rb
index 121c5bd..2dbbbe0 100644
--- a/importer/import.rb
+++ b/importer/import.rb
@@ -2,6 +2,11 @@ require 'rubygems'
 
 require 'wordpress'
 
-i = Importer::Wordpress.new('wezm.net.2009-11-17.xml')
+if ARGV.size < 2
+  puts "Usage importer.rb worpress-export.xml /path/to/nanoc/site"
+  exit 3
+end
+
+i = Importer::Wordpress.new(ARGV[0], ARGV[1])
 
 i.run
diff --git a/importer/wordpress.rb b/importer/wordpress.rb
index abf8fe1..bfc7be5 100644
--- a/importer/wordpress.rb
+++ b/importer/wordpress.rb
@@ -2,12 +2,13 @@ require 'nokogiri'
 require 'nanoc3'
 
 module Importer
-  
+
   class Wordpress
 
-    def initialize(wordpress_export_path)
+    def initialize(wordpress_export_path, nanoc_site_path)
       @export_file = File.open(wordpress_export_path)
       @export = Nokogiri::XML(@export_file)
+      @site = Nanoc3::Site.new(nanoc_site_path)
     end
 
     def run
@@ -29,10 +30,48 @@ module Importer
 
     protected
 
+    def get(node, xpath)
+      elem = node.xpath(xpath).first
+      elem ? elem.content : nil
+    end
+
     def process_post(post)
       puts "Processing post: #{post.css('title').first.text}"
+      content = get(post, 'content:encoded')
+
+      tags = []
+      post.css('category[domain=tag]').each do |tag|
+        if tag['nicename']
+          tags << tag['nicename']
+        else
+          tags << tag.text.downcase
+        end
+      end
+
+      categories = []
+      post.css('category[domain=category]').each do |category|
+        categories << category.text
+      end
+
+      attributes = {
+        :tags => tags.uniq,
+        :categories => categories.uniq,
+        :permalink => get(post, 'link'),
+        :status => get(post, 'wp:status'),
+        :slug => get(post, 'wp:post_name'),
+        :post_id => get(post, 'wp:post_id').to_i,
+        :post_date => get(post, 'wp:post_date_gmt'),
+      }
+
+      unless attributes[:slug]
+        puts "Error post #{post_id} has no slug"
+        return
+      end
+
+      path = "/articles/#{slug}"
+      add_item(content, attributes, identifier)
     end
-    
+
     def process_page(page)
       puts "Processing page: #{page.css('title').first.text}"
     end
@@ -41,6 +80,20 @@ module Importer
       puts "Processing attachment"
     end
 
+    def add_item(content, attributes, identifier)
+      # content = row['post_content']
+      # attributes = {
+      #   :title        => row['post_title'],
+      #   :published_on => row['post_date_gmt'],
+      #   :modified_on  => row['post_modified_gmt'],
+      #   :status       => row['post_status'],
+      #   :excerpt      => row['post_excerpt']
+      # }
+      # identifier = '/posts/' + post_date.year.to_s + '/' + post_date.month.to_s + '/' + post_name + '/'
+      @site.data_sources.first.create_item(content, attributes, identifier)
+      puts "Added item at #{identifier}"
+    end
+
   end
-  
+
 end